KR102775723B1 - System and method for predicting operating safety of power plant - Google Patents

Abstract

The embodiments relate to a system and method for predicting operation health of a power plant, including obtaining sensing data from each of a plurality of sensors in the power plant, receiving a target power generation value, operating the power plant to produce power according to the target power generation value, and obtaining an actual power generation value currently being produced, inputting at least a portion of each of the sensing data of the plurality of sensors into a pre-learned power generation prediction model to calculate a current power generation prediction value and at least one future power generation prediction value, and calculating an operation health score based on at least a portion of the target power generation value, the actual power generation value, the current power generation prediction value, and the at least one future power generation prediction value.

Description

{SYSTEM AND METHOD FOR PREDICTING OPERATING SAFETY OF POWER PLANT}

Embodiments of the present application relate to a technology for predicting the operational health of a power plant, and more specifically, to a system and method for predicting the operational health of a power plant based on the relationship between the actual power generation value and the target power generation value of the power plant.

Power plants are generally composed of multiple systems, and each system includes multiple power generation facilities. Each system or power generation facility is organically connected to each other. Because of this, various types of failures can occur during the power generation process. Since failures in the machinery and equipment of a specific system can spread to other systems and cause major failures that cause enormous losses, it is important to detect the location of the system where the failure first started early.

The environment inside the power plant is constantly monitored in real time through a number of sensors installed on or around the power plant itself. The central system installed in the power plant monitors risks such as failures of the power plant equipment through sensor data.

However, conventional systems typically notify when a measured value representing the current status of a power plant or power plant reaches or exceeds a preset risk threshold.

Therefore, conventional systems have limitations in preventing failures because they detect failures after they occur in the power plant.

Patent Registration No. 10-1713985 (2017.03.02.)

According to one aspect of the present application, a system and method for predicting operational soundness at present and/or future points in time can be provided based on a relationship between an actual power generation value and a target power generation value of a power plant.

In addition, a computer-readable recording medium recording a program for performing a method for predicting the operational soundness of a power plant can be provided.

A power plant operation soundness prediction system according to one aspect of the present application may include: a plurality of sensors installed in a plurality of power generation facilities to measure the operation states or surrounding environments of the plurality of power generation facilities; a power plant operation unit that receives a target power generation value, operates the power plant to produce power according to the target power generation value, and obtains an actual power generation value currently produced; a power generation prediction unit that calculates a current power generation prediction value and at least one future power generation prediction value from at least a portion of each of the sensing data of the plurality of sensors using a pre-learned power generation prediction model; and an operation evaluation unit that calculates at least one of a current operation soundness score and a future operation soundness score based on an operation soundness model based on at least a portion of the target power generation value, the actual power generation value, the current power generation prediction value, and the at least one future power generation prediction value.

In one embodiment, the power generation prediction model may be configured to use a plurality of training samples to, when an input data set including measurement values of an input sensor is input, produce a current power generation prediction value and at least one future power generation prediction value. Each of the plurality of training samples includes measurement values for a portion or all of a period from the past to an input time point including an input time point, an actual power generation value during the period, and a target power generation value.

In one embodiment, the power generation prediction model may be configured to produce power generation prediction values for n future time periods, where n is a hyperparameter.

In one embodiment, for each of at least one future point in time, the driving health model may be based on one or more of the following relationships: a difference between a target generation value and an actual generation value, a difference between a current actual generation value and a current predicted generation value, and a difference between predicted generation values at the same future point in time.

In one embodiment, for each of at least one future, the future power generation prediction value may be calculated as one or more values. The driving health model is configured to calculate a driving health score (S) based on one or more of components S ₁ , S ₂ , and S ₃ . The component S ₁ is a variable based on a difference between a target power generation value and an actual power generation value, the component S ₂ is a variable based on a difference between a current actual power generation value and a current power generation prediction value, and the component S ₃ is a variable based on a difference between power generation prediction values at the same future time point, for each of at least one future time point.

In one embodiment, the component S ₁ is calculated via the following mathematical formula:

[Mathematical formula]

|Target power generation value - Actual power generation| < Th ₁ ,

s1 = C _1,

|Target power generation value - Actual power generation| If ≥ Th ₁ ,

Here, Th ₁ is a preset first threshold value, and C ₁ is a first constant specified for component S ₁ .

In one embodiment, the component S ₂ is calculated via the following mathematical formula:

[Mathematical formula]

If actual power generation - predicted power generation < Th ₂ ,

s ₂ = C ₂ ,

If actual power generation - predicted power generation ≥ Th ₂ ,

Here, Th ₂ is a preset second threshold value, and C ₂ is a second constant specified for component S ₂ .

In one embodiment, the component S ₃ is calculated via the following mathematical formula:

[Mathematical formula]

If max(future power generation prediction value of 1, future power generation prediction value of 2, …future power generation prediction value of n) - min(future power generation prediction value of 1, future power generation prediction value of 2, …future power generation prediction value of n) < Th ₃ , then S ₃ =C ₃ ; and

If max(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) - min(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) ≥ Th ₃ ,

Here, Th ₃ is a preset third threshold value, and C ₃ is a third constant specified for component S ₃ .

In one embodiment, the driving soundness model may be modeled by the following mathematical equation.

[Mathematical formula]

S = λ ₁ *S ₁ +λ ₂ *S ₂ +λ ₃ *S ₃

Here, λ ₁ + λ ₂ + λ ₃ = 1, 0≤S≤1

λ ₁ , λ ₂ , and λ ₃ are weights for components S ₁ , S ₂ , and S ₃ , respectively.

In one embodiment, the driving evaluation unit may be further configured to generate a driving health screen including power generation history and a driving health evaluation result. The power generation history includes actual power generation values and target values from the past to the present at the time of the driving health evaluation. The driving health evaluation result includes a current driving health score and at least one driving health score in the future.

In one embodiment, the driving soundness screen may include a power generation graph with an x-axis and a y-axis depicting power generation history from the past to the present, and a score graph with an x-axis and a y-axis depicting driving soundness evaluation results from the present to the future. The x-axis of the power generation graph and the score graph represents time. The y-axis of the power generation graph represents power generation units. The y-axis of the score graph represents a driving soundness score.

In one embodiment, the driving evaluation unit may be further configured to determine the location of a facility in which a specific sensor having a relatively large amount of change in a measurement value among a plurality of sensors that detected a measurement value used to calculate a driving health score is installed as the location of an abnormal facility.

In one embodiment, the driving evaluation unit may cluster the variation amounts of the plurality of sensors into at least two groups based on the distribution of variation amounts of the measurement values of each of the plurality of sensors, designate a representative value of the variation amount for each group, and determine some groups among the entire groups as an abnormal equipment group based on the representative value of the variation amount for each group.

In one embodiment, the driving evaluation unit may be further configured to generate a screen that visually displays the location of the determined abnormal equipment in the entire power plant system so as to be distinguished from normal equipment.

In one embodiment, the location of the abnormal equipment on the abnormal equipment notification screen may be displayed in a color corresponding to the operational health score for the abnormal equipment.

According to another aspect of the present application, a method for predicting the operational soundness of a power plant is performed by a processor and a plurality of sensors. The method for predicting the operational soundness of a power plant may include: a step of obtaining sensing data from each of a plurality of sensors in the power plant; a step of receiving a target power generation value, operating the power plant to produce power according to the target power generation value, and obtaining an actual power generation value currently produced; a step of inputting at least a portion of each of the sensing data of the plurality of sensors into a pre-learned power generation prediction model to calculate a current power generation prediction value and at least one future power generation prediction value; and a step of calculating an operational soundness score based on at least a portion of the target power generation value, the actual power generation value, the current power generation prediction value, and the at least one future power generation prediction value.

In one embodiment, the step of calculating the driving soundness score may include: when a current power generation target value, a current power generation actual value, a current power generation prediction value, and at least one future power generation prediction value are input into a driving soundness model, calculating one or more of components S ₁ , S ₂ , and S ₃ , and calculating a driving soundness score (S) based on the calculated one or more of components S ₁ , S ₂ , and S ₃ . The driving soundness model is configured to calculate the driving soundness score (S) based on one or more of components S ₁ , S ₂ , and S ₃ .

In one embodiment, the method may further include the step of generating a driving health screen including power generation history and driving health evaluation results.

In one embodiment, the method may further include the step of determining a location of a facility having a particular sensor installed among a plurality of sensors that detected a measurement value used to calculate a driving health score, wherein the sensor has a relatively large amount of change in the measurement value, as the location of the abnormal facility.

In one embodiment, the method may further include the step of generating a screen that visually displays the location of the determined abnormal equipment in the entire power plant system so as to be distinct from normal equipment.

According to another aspect of the present application, a computer-readable recording medium may record a program for performing a method for predicting power plant operation soundness according to the above-described embodiments.

A power plant operation soundness prediction system according to one aspect of the present application can prevent equipment failures in a power plant by predicting operation soundness at a future point in time before an accident occurs.

In particular, the system may also produce a driving health score that can evaluate driving health at multiple future points in time.

In addition, the system can display the predicted driving soundness and power generation on the same screen to provide the generated prediction results to the user in a screen with greater convenience.

As a result, it is possible to prevent enormous financial losses and loss of life that could occur due to actual failures.

The effects of the present application are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

In order to more clearly explain the technical solution of the present application or the prior art embodiment, the drawings required in the description of the embodiment are briefly introduced below. It should be understood that the drawings below are only for the purpose of explaining the embodiment of the present specification and are not for the purpose of limitation. In addition, some elements may be depicted with various modifications such as exaggeration or omission for the sake of clarity of explanation in the drawings below.
FIG. 1 is a schematic diagram of a system for predicting the operational soundness of a power plant according to one aspect of the present application.
FIG. 2 is a diagram illustrating a driving soundness score screen according to one embodiment of the present application.
FIG. 3 is a drawing illustrating an abnormal equipment notification screen according to one embodiment of the present application.
FIG. 4 is a flowchart of a method for predicting operational soundness based on the relationship between the actual power generation amount and the target power generation amount of a power plant according to another aspect of the present application.

In this specification, the expressions “have,” “may have,” “include,” or “may include” indicate the presence of a feature (e.g., a component such as a number, function, operation, step, part, element, and/or component), but do not exclude the presence or addition of additional features.

When it is said that a component is "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but that there may be other components in between. On the other hand, when it is said that a component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

The expressions “first,” “second,” “first,” or “second” used in various embodiments can describe various components, regardless of order and/or importance, and do not limit the components. The expressions can be used to distinguish one component from another. For example, the first component and the second component can represent different components, regardless of order or importance.

The expression “configured to” as used herein can be used interchangeably with, for example, “suitable for,” “having the capacity to,” “designed to,” “adapted to,” “made to,” or “capable of.” The term “configured to” does not necessarily mean something is “specifically designed to” in terms of hardware. Instead, in some contexts, the expression “a device configured to” can mean that the device is “capable of” doing something together with other devices or components. For example, the phrase “a processor configured (or set) to perform A, B, and C” can mean a dedicated processor (e.g., an embedded processor) to perform those operations, or a generic-purpose processor (e.g., a CPU or application processor) that can perform those operations by executing one or more software programs stored in a memory device.

The system (100) according to the embodiments of the present application calculates the current predicted power generation and the future predicted power generation based on the current sensing data acquired using the pre-learned power generation prediction model, and calculates the driving soundness score by applying the acquired current power generation actual value, the current predicted power generation and the future predicted power generation to the pre-learned driving soundness model. As a result, the system can evaluate the driving soundness of the power plant at a future point in time using the driving soundness score.

Hereinafter, embodiments of the present application will be described in detail with reference to the drawings.

FIG. 1 is a schematic diagram of a system for predicting the operational soundness of a power plant according to one aspect of the present application.

Referring to FIG. 1, the system (100) may include a plurality of sensors (110); a power plant operation unit (120); a power generation prediction unit (140); and an operation evaluation unit (150).

The system (100) according to the embodiments may have aspects that are entirely hardware, entirely software, or partially hardware and partially software. For example, a system may collectively refer to hardware having data processing capabilities and operating software for driving the same. In this specification, terms such as “unit,” “module,” “device,” or “system” are intended to refer to a combination of hardware and software driven by the hardware. For example, the hardware may be a data processing device including a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or another processor. In addition, the software may refer to a running process, an object, an executable, a thread of execution, a program, etc.

A plurality of sensors (110) are devices that detect the operating status of a plurality of power generation facilities installed in a power plant and changes in the surrounding environment, and generate sensing data including measurement values of the operating status and the surrounding environment.

The above-described plurality of sensors (110) include analog sensors and/or digital sensors. The analog sensor expresses a change in state as an analog signal and generates analog sensing data that has been converted into data. The digital sensor expresses a change in state as a digital signal and generates digital sensing data that has been converted into data.

The above-described plurality of sensors (110) obtain information related to power generation facilities and power plants from various aspects. The above-described plurality of sensors (110) may include, for example, an energy sensor, a temperature sensor, a pressure sensor, a flow sensor, a vibration sensor, a moisture sensor, a gravity sensor, a geomagnetic sensor, a motion sensor, a gyro sensor, an acceleration sensor, a tilt sensor, a brightness sensor, an opening sensor, an olfactory sensor, a depth sensor, a banding sensor, an audio sensor, an image sensor, or a combination thereof.

The above plurality of sensors (110) may obtain information related to power generation facilities or power plants within the detection range and generate sensing data. For example, the plurality of sensors (110) may generate sensing data including information related to power generation facilities (e.g., operation information of power generation facilities, etc.).

The above sensing data includes physical information of the measurement target, measurement time, time unit, measurement value (or signal), measurement unit, etc. In addition, the sensing data may include information related to the corresponding sensor (110). The information related to the corresponding sensor (110) may include identification information of the sensor (110), installed system information, and/or location information (e.g., location within a power plant or installation location in a power generation facility), etc.

Sensing data indicating the operating status of the power generation facility is generated by a plurality of sensors (110) and supplied to other components (e.g., power plant operation unit (120)) through wired/wireless electric communication.

The power plant operation unit (120) is a component that controls the power generation operation of the power plant.

The power plant operation unit (120) receives a target value of the power generation amount to be produced at the power plant. The power plant operation unit (120) may also receive a power generation request including a target value of the power generation amount to be produced at the power plant from the power exchange. In one embodiment, the power plant operation unit (120) may also receive a power generation request in real time.

The power plant operation department (120) operates the power plant to produce electricity equivalent to the target power generation amount. The power plant operation department (120) specifies the input amount of raw materials (e.g., coal input amount) and the operating values for each power generation facility, such as a boiler, turbine, and generator, according to the target power generation amount.

When the power plant operation department (120) receives a power generation request in real time, it controls the coal injection device to adjust the amount of coal input, thereby operating the power plant in accordance with the target power generation amount of the power generation request.

In addition, the power plant operation unit (120) obtains information on the current actual power generation value for the target power generation value. The current actual power generation value is the amount of power actually produced at the current power plant.

A power plant that is operating normally without equipment failure operates the power generation facility at a preset control value and inputs raw materials to produce the received target power generation value, thereby producing an actual power generation value that is in accordance with the target power generation value. For example, if the target power generation value is 860 MW, the actual power generation value may be approximately 855 MW to 865 MW. The power plant operation unit (120) may also supply information on the target power generation value and the actual power generation value to the power plant prediction unit (140).

The power generation prediction unit (140) is configured to predict power plant failures, including failures of power generation facilities. The power plant prediction unit (140) predicts power plant failures by calculating an operational soundness score based on the relationship between the received power generation target value of the power plant and the actual value of the power generation currently produced.

In certain embodiments, the power plant prediction unit (140) calculates a current power generation prediction value and a future power generation prediction value from at least a portion of each of a plurality of sensing data obtained from a plurality of sensors (110) using a pre-learned power generation prediction model.

If the power generation facility does not break down, the power generation facility operates in normal operation and achieves the workload required to produce the target power generation amount.

On the other hand, in the event of a failure, each power generation facility has a different workload from that under normal operation. If a failure occurs in a specific power generation facility, the failed facility cannot achieve the preset workload. For example, if a failure occurs in a single facility, the normal power generation amount corresponding to the target power generation amount will be insufficient to produce the amount of power that the facility cannot operate due to the failure. If a failed facility In case of a duplicated facility, the other duplicated facility performs the maximum work to operate in order to produce as much power as the target power generation amount. However, if the duplication capacity of the other duplicated facility is insufficient, the electricity as much as the target power generation amount cannot be produced. The difference in the amount of work due to a power plant failure, such as a failure of the power generation facility, is expressed as a change in the measurement values of a large number of analog sensors and the measurement values of digital sensors obtained by a plurality of sensors (110) installed in the power generation facility. The power generation prediction unit (140) uses the sensing data to predict the failure of the power plant.

The power generation prediction unit (140) obtains sensing data by a plurality of sensors (110). The power generation prediction unit (140) may also obtain current sensing data to calculate a health score.

The power generation prediction unit (140) includes a power generation prediction model that is learned in advance to produce a current power generation prediction value and at least one future power generation prediction value.

The predicted power generation value is the amount of power generation expected to be generated by the system (100) at a specific point in time.

The above specific point in time may be the present point in time or one or more future points in time after the present point in time.

The current point in time is the target point in time at which the user wants to predict the operational health of the power plant through the system (100). The current point in time may also be the point in time at which the operation of calculating the operational health score begins.

The above future point in time is a point in time that is in the future from the present point in time. In some embodiments, the future point in time may be expressed based on the present point in time. For example, the future point in time at which the power generation prediction value is to be calculated may be a point in time that is several minutes (e.g., 5 minutes) or several tens of minutes (e.g., 10 minutes, 30 minutes, 60 minutes) from the present point in time.

The above power generation prediction model is a machine learning model configured to produce a current power generation prediction value and at least one future power generation prediction value from input value(s).

The above-described development prediction model includes a deep learning-based network composed of multiple neuron nodes. The above-described development prediction model may have an artificial neural network structure including a RNN (Recurrent Neural Network)-based network such as LSTM, GRU, etc., a CNN-based network such as DenseNet, ResNet, TCN (Temporal convolutional network), or other networks.

In certain embodiments, the power generation prediction model may be configured to output a predicted power generation amount from measured values in the sensing data.

The measurement values represent values measured by each sensor (110), such as a temperature value for a temperature sensor or a current value for a current sensor.

When using sensing data acquired from multiple sensors (110), an input data set consisting of measurement values of each sensing data acquired from each of the multiple sensors (110) may be input into the power generation prediction model.

In one embodiment, the commercial power generation prediction unit (140) may form an input data set including measurement values of input sensors and input this into the power generation prediction model.

The above development prediction model is configured to use the sensing data of the present time and the sensing data of the past as input data. The measurement time of the input data may be a part or all of the operating period including the present from the time the sensor (110) starts operating to the present at the time of the health evaluation.

The above power generation prediction model may be configured to produce power generation prediction values at a single future point in time or multiple future points in time. When configured to produce power generation prediction values at n future points in time, the power generation prediction model may produce an output data set consisting of a current power generation prediction value, a first future power generation prediction value, a second future power generation prediction value, and an n-th future power generation prediction value.

The power generation prediction model is trained to produce a power generation prediction value at an input point in time and a future power generation prediction value from input value(s) using multiple training samples. Here, each of the multiple training samples includes a measurement value from the past to a part or all of a period including the input point in time, an actual power generation value during the period, and a target power generation value. The input point in time may be the current time at the time of training when the training measurement value is detected by the input sensor.

In the above development prediction model, the n value at multiple future points in time can also be set as a hyper parameter of the model.

The above system (100) may use a power generation prediction model learned by another component (e.g., a learning module (not shown)) within the system (100), or may receive and use a power generation prediction model learned in advance by an external device located externally. The power generation prediction unit (140) forms an input data set consisting of at least a portion of each sensing data (e.g., a measurement value) based on sensing data acquired from a plurality of sensors (110). The formed input data set is input into the power generation prediction model to produce a current power generation prediction value and at least one future power generation prediction value.

In one embodiment, the future power generation prediction value may be calculated as one or more values. The power generation prediction model may output a prediction value at a specific point in time, or may calculate multiple prediction values at the same specific point in time.

In some embodiments, the power generation prediction model may output as future power generation prediction values some or all of the maximum predicted value, the minimum predicted value, and any of the predicted values in between at the same specific point in time.

The above power generation prediction unit (140) supplies the current power generation prediction value and at least one future power generation prediction value to the driving evaluation unit (150).

The driving evaluation unit (150) calculates an driving soundness score that indicates the normal operation level of the power plant by using the values predicted by the power generation prediction unit (140) and the values obtained from the power plant operation unit (120).

In certain embodiments, the driving evaluation unit (150) may use a driving health model to calculate a driving health score from a target power generation value, an actual power generation value, a current predicted power generation value, and a future predicted power generation value. The target power generation value and the actual power generation value are the current target power generation value and the current actual power generation value at the time when the driving health is evaluated.

The driving evaluation unit (150) may also calculate one or more driving soundness scores from among the current driving soundness score and at least one future driving soundness score.

In one embodiment, the driving soundness model may be based on at least two of a current generation target value, a current generation actual value, a current generation forecast value, and at least one future generation forecast value. The driving soundness model is configured to designate at least one combination of two or more values from the current generation target value, the current generation actual value, the current generation forecast value, and the at least one future generation forecast value, and to calculate a driving soundness score through a relationship between two values within the designated combination.

Here, the future power generation forecast value may be a number of values that the power plant is expected to produce at the same future point in time. Then, a combination may be formed between the power generation forecast values at the same future point in time, and an operating soundness score may be calculated through the relationship between them.

In one embodiment, the driving health model may be based on one or more of the following relationships: a difference between a target generation value and an actual generation value, a difference between a current actual generation value and a current predicted generation value, and, for each of at least one future point in time, a difference between predicted generation values for the same future point in time.

The difference between the target power generation value and the actual power generation value indicates the degree to which the current power plant operation status matches the normal operation status. Then, the degree to which the current power plant operation status is close to the normal operation status is reflected in the operation health score.

The difference between the current actual power generation value and the current actual power generation prediction value indicates the power generation prediction performance of the system (100) at a specific point in time. Then, the accuracy of the power generation prediction result is reflected in the driving soundness score.

The difference between the predicted power generation values at the same future point in time indicates the accuracy of the predicted value at that future point in time. Then, the accuracy of the power plant's future power generation prediction results is reflected in the operating soundness score. The smaller the deviation between the candidate values predicted to produce power at the future point in time, especially the deviation between the maximum candidate value and the minimum candidate value, the more accurate the evaluation is, and this principle is reflected in calculating the operating soundness score.

In one embodiment, the driving health model may be configured to produce a driving health score (S) based on one or more of components S ₁ , S ₂ , and S ₃ .

The above component S ₁ is a variable based on the difference between the target power generation value and the actual power generation value.

In one embodiment, the component S ₁ may be calculated by the following mathematical expression 1.

[Mathematical Formula 1]

|Target power generation value - Actual power generation| < Th ₁ ,

S ₁ = C ₁

Additionally, the above component S ₁ can also be calculated by the following mathematical formula 2.

[Mathematical formula 2]

|Target power generation value - Actual power generation| If ≥ Th ₁ ,

Here, Th ₁ is a preset threshold value, which may be a value indicating that the current power plant operation state is acceptable as a normal operation state. C ₁ is a constant specified for component S ₁ , which indicates that the operation state of the power plant is regarded as a normal operation state. C ₁ may be, for example, 1.

The above component S ₂ is a variable based on the difference between the current actual power generation value and the current predicted power generation value.

In one embodiment, the component S ₂ may be calculated by the following mathematical expression 3.

[Mathematical Formula 3]

If actual power generation - predicted power generation < Th ₂ ,

S ₂ = C ₂

Additionally, the above component S2 can also be calculated by the following mathematical formula 4.

[Mathematical formula 4]

If actual power generation - predicted power generation ≥ Th ₂ ,

Here, Th ₂ may be a preset threshold value, different from Th ₁ here, which may be a value indicating that it is acceptable to utilize the power generation prediction performance at a specific point in time for the driving soundness evaluation. C ₂ is a constant specified for component S ₂ , which indicates that the power generation value at a specific point in time is considered to have been accurately predicted. C ₂ may be, for example, 1.

The above component S ₃ is a variable based on the difference between the predicted power generation values at the same future point in time, for each of at least one future point in time.

In one embodiment, the component S ₃ may be produced by the following mathematical expression 5.

[Mathematical Formula 5]

If max(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) - min(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) < Th ₃ ,

S ₃ =C ₃

Additionally, the above component S ₁ can also be calculated by the following mathematical formula 2.

[Mathematical Formula 6]

If max(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) - min(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) ≥ Th ₃ ,

Here, Th ₃ may be a threshold value different from Th ₁ and Th ₂ that is preset and may be a value indicating that it is acceptable to utilize the accuracy performance of the predicted value at a future point in time for the driving soundness evaluation. C ₃ is a constant specified for component S ₃ and indicates that the power generation value at a future point in time is considered to be accurately predicted. The above C ₃ may be, for example, 1.

When the maximum predicted value of power generation in the first future and the minimum predicted value of power generation in the first future are applied to the above mathematical expressions 5 and 6, the driving soundness score in the first future is calculated. When the maximum predicted value of power generation in the second future and the minimum predicted value of power generation in the second future are applied to the above mathematical expressions 5 and 6, the driving soundness score in the second future is calculated. When the maximum predicted value of power generation in the n-th future and the minimum predicted value of power generation in the n-th future are applied to the above mathematical expressions 5 and 6, the driving soundness score in the n-th future is calculated.

In one embodiment, the driving health model may be expressed by the following mathematical formula. The driving health model may also calculate the driving health score (S) of the power plant through this mathematical formula.

[Mathematical formula 7]

S = λ ₁ *S ₁ +λ ₂ *S ₂ + λ ₃ *S ₃

Here, λ ₁ + λ ₂ + λ ₃ = 1, 0≤S≤1.

Here, λ ₁ , λ ₂ , and λ ₃ are weights for components S ₁ , S ₂ , and S ₃ , respectively. In some embodiments, each weight may be calculated based on the history of power plant failure occurrences, the target power generation value for each history, the actual power generation value, and the predicted power generation value. Machine learning or various statistical methods may be used to specify the weight values.

When the maximum predicted value of power generation in the first future and the minimum predicted value of power generation in the first future are applied to the above mathematical expression 7, the driving soundness score in the first future is calculated. When the maximum predicted value of power generation in the second future and the minimum predicted value of power generation in the second future are applied to the above mathematical expression 7, the driving soundness score in the second future is calculated. When the maximum predicted value of power generation in the n-th future and the minimum predicted value of power generation in the n-th future are applied to the above mathematical expression 7, the driving soundness score in the n-th future is calculated.

The above driving evaluation unit (150) may calculate a current driving soundness score and a driving soundness score at one or more future points in time, and may also provide the user with the results of calculating the driving soundness score.

In one embodiment, the driving soundness score may be calculated as a specific scale range value. The closer it is to one of the upper and lower limits of the scale range, the more it indicates a normal driving state, and the closer it is to the other value, the more it indicates an abnormal driving state. The variables of the mathematical expression 7 and the like may have values that satisfy the specific scale range.

For example, the driving evaluation unit (150) may calculate the driving soundness score S so that a driving soundness score closer to 100 indicates a normal driving state, and a score closer to 0 indicates an abnormal driving state.

FIG. 2 is a diagram illustrating a driving soundness score screen according to one embodiment of the present application.

Referring to FIG. 2, the driving evaluation unit (150) may also generate a driving soundness screen including power generation history and driving soundness evaluation results. The power generation history includes actual power generation values and target values from the past to the present at the time of driving soundness evaluation. The driving soundness evaluation results include a current driving soundness score and at least one driving soundness score in the future.

Part of the above Driving Health screen illustrates the history of power generation from past to present. Another part of the above Driving Health screen illustrates the Driving Health score from present to future.

The section describing the power generation history from the past to the present may include a power generation graph consisting of an x-axis and a y-axis. The y-axis of the power generation graph represents the power generation unit (e.g., MW), and the x-axis represents time. The power generation graph depicts the actual power generation value and the power generation target value over time. The range from the past to the present may be specified by the user's screen scale setting.

In some embodiments, the location of the power generation in the power generation graph may be expressed more continuously within a time range. For example, it may be expressed as a line graph as in FIG. 2.

The part describing the driving soundness evaluation results from the present to the future may include a score graph consisting of an x-axis and a y-axis. The y-axis of the score graph represents the driving soundness score, and the x-axis represents time. The driving evaluation unit (150) may depict the driving soundness score (S) value itself or a converted value for the convenience of the user on the score graph. For example, the driving evaluation unit (150) may depict a score graph having a y-axis of 0 to 1, or a score graph having a y-axis of 0 to 100 on the driving soundness screen.

The location of the score in the above score graph may be expressed more discretely within the time range. For example, it may be expressed as a point corresponding to each specific point in time (present or future), as in Fig. 2.

In one embodiment, the driving soundness score screen may include two y-axes at the same time. The two y-axes include the y-axis of the power generation graph and the y-axis of the score graph. As illustrated in FIG. 2, the power generation history from the past and the driving soundness prediction results from the present to the future may be depicted on one screen at the same time.

In addition, the driving evaluation unit (150) may be further configured to provide an item list including some or all of the items forming the driving health screen as selectable items for display on the driving health screen according to the user's selection. This item list may be depicted in a sub-area of the driving health screen.

For example, among the items forming the driving soundness screen, an item list consisting of actual power generation value, target power generation value, predicted value after 5 minutes, predicted value after 10 minutes, predicted value after 30 minutes, and predicted value after 60 minutes may be provided to induce a user's selection through the driving soundness screen of Fig. 2. When all items in this item list are selected, the driving soundness screen of Fig. 2 is provided to the user.

In addition, if the driving evaluation unit (150) determines that the system status is an abnormal driving state rather than a normal driving state based on the driving soundness score, it can also search for the location of the abnormal equipment that caused the abnormal driving state.

In one embodiment, the driving evaluation unit (150) determines the location of equipment where a specific sensor(s) (110) having a relatively large amount of change in the measurement value among a plurality of sensors (110) that detected the measurement value used to calculate the driving soundness score is installed as the location of the abnormal equipment.

In one embodiment, the driving evaluation unit (150) may cluster the amount of change of the plurality of sensors (110) into at least two groups based on the distribution of the amount of change of the measurement values of each of the plurality of sensors (110). Each group may be assigned a representative value of the amount of change based on the component values included in the group.

Then, the driving evaluation unit (150) may determine some of the entire groups as abnormal equipment groups based on the representative value of the amount of change for each group. In addition, the driving evaluation unit (150) may determine the location of equipment in which some or all of the sensors (110) belonging to the abnormal equipment group are installed as the location of the abnormal equipment. The abnormal equipment group is an equipment group in which the representative amount of change value is relatively large.

For example, if the driving evaluation unit (150) divides a plurality of sensors (110) into two groups, it determines the location of the abnormal equipment in the group with a larger representative change value.

In one embodiment, the above-described abnormal facility group may be a group having the largest representative value of variation among the plurality of groups.

The above driving evaluation unit (150) may also provide the user with the location of an abnormal facility in which a large change in the measurement value of the sensor (110) is detected. Here, the location of the abnormal facility includes the location of the abnormal facility itself or the location of the system including the abnormal facility.

FIG. 3 is a drawing illustrating an abnormal equipment notification screen according to one embodiment of the present application.

Referring to Fig. 3, the driving evaluation unit (150) may also generate an abnormal equipment notification screen that displays the location of the abnormal equipment. The driving evaluation unit (150) supplies data of the abnormal equipment notification screen to a display device (not shown) to display the abnormal equipment notification screen.

The above abnormal facility alarm screen may include a diagram of the entire power plant system. The location of the abnormal facility in the entire power plant system is displayed so that it is visually distinct from the location of the normal facility. For example, in a diagram of the entire power plant system in which shapes indicating each system are displayed, the shape of the system including the abnormal facility may be displayed so that it is distinct from the shape of the system not including the abnormal facility.

In one embodiment, the location of the abnormal equipment on the abnormal equipment notification screen may be displayed in a color corresponding to the operational health score for the abnormal equipment.

The above driving evaluation unit (150) may determine the driving condition grade of the equipment based on the driving soundness score calculated for each equipment and may display the location of the abnormal equipment with a color corresponding to the determined driving condition grade. To this end, the driving evaluation unit (150) may store in advance a grade table that records the driving condition grade according to the driving soundness score. The grade table records a plurality of grades set to a certain score range.

The above plurality of grades includes a first grade (e.g., “normal”) indicating a normal driving condition. The above plurality of grades may also include one or more grades indicating an abnormal driving condition.

In some embodiments, the plurality of grades may include a second grade (e.g., “dangerous”) indicating a state in which a failure is likely to occur and a third grade (e.g., “failure”) indicating a state in which a failure driving state has been reached as abnormal driving states. The second grade has a driving health score closer to a normal driving state than the third grade. In the above example, if the driving health score is calculated such that a driving health score closer to 100 indicates a normal driving state and a score closer to 0 indicates a state in which the driving health score is closer to an abnormal driving state, the second grade corresponds to a higher driving health score than the third grade.

The colors corresponding to these first to third grades are different from each other. For example, as shown in Fig. 3, the first grade may correspond to green, the second grade to yellow, and the third grade to red.

In the system (100) according to the embodiments, each unit (120, 130, 150) may not necessarily be implemented as a separate block within the system. That is, the power plant operation unit (120), the power generation prediction unit (140), and the operation evaluation unit (150) in FIG. 1 may be implemented by being integrated within a single device according to the embodiments.

It will be apparent to one of ordinary skill in the art that the power generation system (100) may include other components not described herein. For example, the power generation system (100) may include other hardware elements necessary for the operations described herein, including a network interface, an input device for data entry, and an output device for display, printing or other display of data.

The method for predicting operational health based on the relationship between the actual power generation value and the target power generation amount of a power plant according to another aspect of the present application may be performed by a plurality of sensors and processors. For example, the method for predicting operational health based on the relationship between the actual power generation value and the target power generation amount of a power plant may be performed by a processor included in the system (100) of FIG. 1. Hereinafter, for the clarity of explanation, the method will be described in more detail with examples performed by the system (100) of FIG. 1.

FIG. 4 is a flowchart of a method for predicting operational soundness based on the relationship between the actual power generation amount and the target power generation amount of a power plant according to another aspect of the present application.

Referring to Fig. 4, the driving soundness prediction method (hereinafter, “driving soundness prediction method”) based on the relationship between the actual power generation amount of the power plant and the target power generation amount includes a step (S110) of acquiring sensing data from a plurality of sensors (110). The sensing data includes values measured from the operating status of equipment installed in the power plant and surrounding environment information.

In addition, the above driving soundness prediction method includes: a step (S120) of receiving a target power generation value and operating a power plant to produce the target power generation value; and a step (S130) of obtaining an actual power generation value. The target power generation value may be included in a power request received from a power exchange. The power plant produces power to produce the target power generation value. Then, an actual power generation value, which is a value at which power is actually produced by the power plant, may be obtained.

In addition, the driving soundness prediction method includes: a step (S140) of calculating a current power generation prediction value and at least one future power generation prediction value from some or all of the sensing data acquired in step (S110) using a power generation prediction model. At least some of each of the sensing data of the plurality of sensors may be input into a pre-learned power generation prediction model to calculate a current power generation prediction value and at least one future power generation prediction value (S140).

The input data set input to the power generation prediction model may be composed of data of part or all of the sensing data obtained from each of a plurality of sensors (110). For example, an input data set composed of measurement values within each sensing data may be input to the power generation prediction model.

For each of at least one future point in time, one or more prediction values may be calculated as the future power generation prediction value for the same future point in time (S140). Such power generation prediction model has been described above with reference to the power generation prediction unit (140), and a detailed description thereof is omitted.

In addition, the driving soundness prediction method includes: a step (S150) of calculating a driving soundness score from the actual power generation value, the target power generation value, and the current/future power generation prediction value of the step (S140) using a driving soundness model.

In one embodiment, the driving health model may be based on at least two of a current power generation target value, a current power generation actual value, a current power generation forecast value, and at least one future power generation forecast value.

In one embodiment, the driving soundness model may be configured to calculate a driving soundness score (S) based on one or more of components S ₁ , S ₂ , and S ₃ . When a current power generation target value, a current power generation actual value, a current power generation forecast value, and at least one future power generation forecast value are input into the driving soundness model, one or more of components S ₁ , S ₂ , and S ₃ may be calculated, and a driving soundness score (S) may be calculated based on the calculated one or more of components S ₁ , S ₂ , and S ₃ .

The driving soundness model, components S ₁ , S ₂ , S ₃ , and driving soundness score (S) are described above with reference to mathematical equations 1 to 7, so a detailed description is omitted.

In addition, the driving soundness prediction method includes: a step (S151) of displaying a driving soundness screen including power generation details and driving soundness evaluation results. In step (S151), data of the driving soundness screen is supplied to a display device (not shown) to display the driving soundness screen.

The above driving soundness screen can simultaneously depict power generation history from the past and driving soundness prediction results from the present to the future on one screen.

The above driving soundness score calculation process and driving soundness screen are described above with reference to Fig. 2, so a detailed description is omitted.

In addition, the driving soundness prediction method further includes a step (S160) of determining the location of the abnormal equipment based on the calculated driving soundness score.

Among the multiple sensors (110) that detected the measurement values used to calculate the driving soundness score in step (S160), the location of the equipment in which a specific sensor(s) (110) with a relatively large amount of change in the measurement value is installed is determined as the location of the abnormal equipment.

In addition, the above-described driving soundness prediction method may further include: a step (S161) of displaying an abnormal facility notification screen indicating the location of the determined abnormal facility. The abnormal facility notification screen displays the location of the abnormal facility that is visually distinct from the locations of other normal facilities in the entire power plant system.

The process for determining the location of the above-mentioned abnormal equipment and the abnormal equipment notification screen have been described above with reference to Fig. 3, so a detailed description is omitted.

The operation of the method and system (100) for predicting operational soundness based on the relationship between the actual power generation value and the target power generation value of the power plant according to the embodiments described above may be implemented at least partially as a computer program and recorded on a computer-readable recording medium. For example, it may be implemented together with a program product consisting of a computer-readable medium including program codes, which may be executed by a processor to perform any or all of the steps, operations, or processes described.

The computer may be any computing device, such as a desktop computer, a laptop computer, a notebook, a smart phone, or the like, or may be integrated into any such device. The computer is a device having one or more alternative and special purpose processors, memory, storage, and networking components (either wireless or wired). The computer may run an operating system, such as, for example, a Windows compatible operating system from Microsoft, Apple OS X or iOS, a Linux distribution, or Google's Android OS.

The computer-readable recording medium includes all kinds of recording identification devices that store data that can be read by a computer. Examples of the computer-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage identification devices, etc. In addition, the computer-readable recording medium may be distributed over network-connected computer systems, so that computer-readable codes can be stored and executed in a distributed manner. In addition, functional programs, codes, and code segments for implementing the present embodiment may be easily understood by a person skilled in the art to which the present embodiment belongs.

The present application discussed above has been described with reference to the embodiments illustrated in the drawings, but these are merely exemplary, and those skilled in the art will understand that various modifications and variations of embodiments are possible from this. However, such modifications should be considered to be within the technical protection scope of the present application. Therefore, the true technical protection scope of the present application should be determined by the technical idea of the appended patent claims.

Claims (21)

In the power plant operation soundness prediction system,
A plurality of sensors installed in a plurality of power generation facilities to measure the operating status or surrounding environment of the plurality of power generation facilities;
A power plant operation unit that receives a power generation target value, operates the power plant to produce power according to the power generation target value, and obtains the actual value of the power generation currently produced;
A power generation prediction unit that calculates a current power generation prediction value and at least one future power generation prediction value from at least a portion of each of the sensing data of the plurality of sensors using a pre-learned power generation prediction model; and
A driving evaluation unit that calculates at least one driving health score among a current driving health score and at least one driving health score in each of the futures through a driving health model based on at least some of the above power generation target value, the above actual power generation value, the current power generation prediction value, and at least one future power generation prediction value;
For each of at least one future point in time, the driving soundness model is based on one or more of the relationships among the difference between the target generation value and the actual generation value, the difference between the current actual generation value and the current predicted generation value, and the difference between the predicted generation values at the same future point in time.
For each of at least one future, the future power generation forecast value is calculated as one or more values,
The above driving soundness model is configured to calculate a driving soundness score (S) based on one or more of components S ₁ , S ₂ , and S ₃ .
The above component S ₁ is a variable based on the difference between the target power generation value and the actual power generation value,
The above component S ₂ is a variable based on the difference between the current actual power generation value and the current predicted power generation value.
A power plant operation soundness prediction system, characterized in that the above component S ₃ is a variable based on the difference between power generation prediction values at the same future point in time for each of at least one future point in time.
In the first paragraph,
The above power generation prediction model is configured to use multiple training samples to, when an input data set including measurement values of an input sensor is input, produce a current power generation prediction value and at least one future power generation prediction value,
A power plant operation soundness prediction system, characterized in that each of the plurality of training samples includes a measurement value for a part or all of a period including an input point in the past up to an input point in time, an actual power generation value, and a target power generation value during the period.
In the first paragraph,
The above power generation prediction model is configured to produce power generation prediction values for n future times,
A power plant operation soundness prediction system, characterized in that the above n is designated as a hyperparameter.
delete delete In the first paragraph,
The above component S ₁ is calculated through the following mathematical formula:
[Mathematical formula]
|Target power generation value - Actual power generation| < Th ₁ ,
S ₁ = C _1,
|Target power generation value - Actual power generation| If ≥ Th ₁ ,

A power plant operation soundness prediction system, characterized in that, here, Th ₁ is a preset first threshold value, and C ₁ is a first constant specified for component S ₁ .
In the first paragraph,
The above component S ₂ is calculated using the following mathematical formula:
[Mathematical formula]
If actual power generation - predicted power generation < Th ₂ ,
S ₂ = C ₂ ,
If actual power generation - predicted power generation ≥ Th ₂ ,

A power plant operation soundness prediction system, characterized in that, here, Th ₂ is a preset second threshold value, and C ₂ is a second constant specified for component S ₂ .
In the first paragraph,
The above component S ₃ is calculated using the following mathematical formula:
[Mathematical formula]
If max(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) - min(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) < Th ₃ ,
S ₃ =C ₃ ; and
If max(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) - min(predicted power generation value of future 1, predicted power generation value of future 2, …predicted power generation value of future n) ≥ Th ₃ ,

A power plant operation soundness prediction system, characterized in that, here, Th ₃ is a preset third threshold value, and C ₃ is a third constant specified for component S ₃ .
In the first paragraph,
The above driving soundness model is expressed by the following mathematical formula:
[Mathematical formula]
S = λ ₁ *S ₁ +λ ₂ *S ₂ + λ ₃ *S ₃
Here, λ ₁ + λ ₂ + λ ₃ = 1, 0≤S≤1
A power plant operation soundness prediction system, characterized in that λ ₁ , λ ₂ , and λ ₃ are weights for components S ₁ , S ₂ , and S _{3 ,} respectively.
In the first paragraph, the driving evaluation unit,
It is further configured to generate a driving health screen including power generation history and driving health assessment results.
The above power generation history includes the actual power generation value and target value from the past to the present at the time of the operation soundness evaluation.
A power plant operation health prediction system, characterized in that the above operation health evaluation result includes a current operation health score and at least one future operation health score.
In Article 10,
The above driving soundness screen is a power generation graph consisting of an x-axis and a y-axis, which depicts the power generation history from the past to the present, and
The description of the driving soundness assessment results from the present to the future includes a score graph consisting of an x-axis and a y-axis.
The x-axis of the above power generation graph and score graph represents time,
The y-axis of the above power generation graph represents the power generation unit.
A power plant operation soundness prediction system, characterized in that the y-axis of the above score graph represents the operation soundness score.
In the first paragraph, the driving evaluation unit,
A power plant operation health prediction system further characterized in that the location of equipment in which a specific sensor having a relatively large amount of change in measurement value is installed among a plurality of sensors that detect measurement values used to calculate an operation health score is determined as the location of an abnormal equipment.
In the 12th paragraph, the driving evaluation unit,
Based on the distribution of the amount of change in the measurement values of each of the plurality of sensors, the amount of change of the plurality of sensors is clustered into at least two groups, and a representative value of the amount of change for each group is designated,
A power plant operation soundness prediction system characterized in that some groups among the entire groups are determined as abnormal equipment groups based on the representative value of the amount of change for each group.
In the 12th paragraph, the driving evaluation unit,
A power plant operation health prediction system further characterized by being configured to generate a screen that visually displays the location of the determined abnormal equipment in the entire power plant system so as to be distinguished from normal equipment.
In Article 14,
A power plant operation health prediction system characterized in that the location of abnormal equipment on the above abnormal equipment notification screen is displayed in a color corresponding to the operation health score for the above abnormal equipment.
In a method for predicting the operational health of a power plant performed by a processor and multiple sensors,
A step of acquiring sensing data from each of multiple sensors within a power plant;
A step of receiving a target power generation value, operating a power plant to produce power according to the target power generation value, and obtaining the actual value of the currently produced power generation;
A step of inputting at least a portion of each of the sensing data of the plurality of sensors into a pre-learned power generation prediction model to produce a current power generation prediction value and at least one future power generation prediction value; and
A step of calculating an operating soundness score based on at least some of the above power generation target value, the above power generation actual value, the present power generation prediction value and at least one future power generation prediction value,
The steps for calculating the above driving soundness score are:
When the current power generation target value, the current power generation actual value, the current power generation forecast value, and at least one future power generation forecast value are input into the driving soundness model, one or more components among the components S ₁ , S ₂ , and S ₃ are calculated, and an driving soundness score (S) is calculated based on the one or more components among the calculated components S ₁ , S ₂ , and S ₃ .
The above driving soundness model is configured to calculate a driving soundness score (S) based on one or more of components S ₁ , S ₂ , and S ₃ .
The above component S ₁ is a variable based on the difference between the target power generation value and the actual power generation value,
The above component S ₂ is a variable based on the difference between the current actual power generation value and the current predicted power generation value.
A method for predicting the operational soundness of a power plant, characterized in that the above component S ₃ is a variable based on the difference between predicted power generation values at the same future point in time for each of at least one future point in time.
delete In Article 16,
A method for predicting the operational health of a power plant further comprising the step of generating an operational health screen including power generation history and operational health evaluation results.
In Article 16,
A method for predicting the operational health of a power plant, further comprising a step of determining the location of a facility in which a specific sensor having a relatively large amount of variation in a measured value among a plurality of sensors that detect a measured value used to calculate an operational health score is installed, as the location of an abnormal facility.
In Article 19,
A method for predicting the operational health of a power plant further comprising the step of generating a screen that visually displays the location of the determined abnormal equipment in the entire power plant system so as to be distinguished from normal equipment.
A computer-readable recording medium having recorded thereon a program for performing a method for predicting the operational soundness of a power plant according to any one of claims 16, 18, to 20.

Images