[go: up one dir, main page]

CN119064820B - Laser power supply component performance detection method and system - Google Patents

Laser power supply component performance detection method and system Download PDF

Info

Publication number
CN119064820B
CN119064820B CN202411561878.2A CN202411561878A CN119064820B CN 119064820 B CN119064820 B CN 119064820B CN 202411561878 A CN202411561878 A CN 202411561878A CN 119064820 B CN119064820 B CN 119064820B
Authority
CN
China
Prior art keywords
value
power supply
state
performance index
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202411561878.2A
Other languages
Chinese (zh)
Other versions
CN119064820A (en
Inventor
戴畅
陈延明
黄修江
张振伟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Lianming Power Supply Co ltd
Original Assignee
Shenzhen Lianming Power Supply Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Lianming Power Supply Co ltd filed Critical Shenzhen Lianming Power Supply Co ltd
Priority to CN202411561878.2A priority Critical patent/CN119064820B/en
Publication of CN119064820A publication Critical patent/CN119064820A/en
Application granted granted Critical
Publication of CN119064820B publication Critical patent/CN119064820B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/40—Testing power supplies
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M11/00—Testing of optical apparatus; Testing structures by optical methods not otherwise provided for
    • G01M11/02—Testing optical properties
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Molecular Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Lasers (AREA)
  • Photometry And Measurement Of Optical Pulse Characteristics (AREA)

Abstract

本发明公开了一种激光电源组件性能检测方法及系统，涉及激光电源检测领域。通过对待检测电源由荷电状态上限SOCmax持续放电至荷电状态下限SOCmin过程中的性能指标参数及调光器的亮度值、环境参数、荷电状态值进行采样，筛选并剔除性能指标参数采样值中的异常数据进行归一化处理，以归一化处理后的调光器亮度值、环境参数、荷电状态采样值为输入数据，以对应的归一化处理后的性能指标参数采样值为输出数据，构建BP神经网络模型并进行训练，利用构建的神经网络模型获取待检测电源性能指标参数的预测值，并根据获取的预测值计算各项性能指标参数的修正系数，通过修正系数对性能指标参数的采样值进行修正，获取各项性能指标参数的修正值。

The present invention discloses a method and system for detecting the performance of a laser power supply component, and relates to the field of laser power supply detection. The performance index parameters of the power supply to be detected during continuous discharge from the upper limit of the state of charge SOC max to the lower limit of the state of charge SOC min , as well as the brightness value, environmental parameters, and state of charge value of the dimmer are sampled, and abnormal data in the performance index parameter sampling values are screened and removed for normalization processing. The normalized dimmer brightness value, environmental parameters, and state of charge sampling values are used as input data, and the corresponding normalized performance index parameter sampling values are used as output data. A BP neural network model is constructed and trained, and the constructed neural network model is used to obtain the predicted values of the performance index parameters of the power supply to be detected, and the correction coefficients of various performance index parameters are calculated based on the obtained predicted values. , through the correction factor Correct the sampled values of the performance indicator parameters to obtain corrected values of each performance indicator parameter.

Description

Laser power supply component performance detection method and system
Technical Field
The invention relates to the field of laser power supply detection, in particular to a method and a system for detecting performance of a laser power supply component.
Background
The laser power supply is a device capable of converting electric energy into light energy and is widely applied to the fields of laser communication, laser display, laser cutting, laser welding and the like. In order to ensure the performance and reliability of the laser power supply, a series of performance tests including static power test, dynamic power test, efficiency test, stability test, noise test, etc. are required.
The dynamic power test is a method for evaluating dynamic power performance of a laser power supply, currently, a dimmer is generally used for measurement, the laser power supply is connected to the dimmer and a power meter in the test process, the brightness of the dimmer is dynamically regulated, brightness readings and power meter numbers are recorded, and rated power, actual power, efficiency and other performance indexes of the power supply are evaluated according to the obtained brightness values and power values.
However, the existing testing method does not consider external influencing factors, such as environmental interference, battery state and the like, so that deviation occurs between a measurement result and a real situation, and the dynamic power performance of the detected laser power supply cannot be reflected truly. Therefore, we propose a method and system for detecting the performance of a laser power supply assembly.
Disclosure of Invention
The invention mainly aims to provide a method and a system for detecting the performance of a laser power supply assembly, which can effectively solve the problems in the background technology.
In order to achieve the above purpose, the invention adopts the technical proposal that,
A method for detecting performance of a laser power supply assembly, comprising:
The method comprises the steps of connecting a power supply to be detected to a dimmer and a power meter, dynamically adjusting the brightness of the dimmer, and sampling performance index parameters, brightness values, environment parameters and state-of-charge values of the dimmer in the process that the power supply to be detected is continuously discharged from an upper limit SOC max to a lower limit SOC min of the state of charge, wherein the performance index parameters comprise rated power, actual power and conversion efficiency, the environment parameters comprise the running temperature of the power supply to be detected and the humidity of the environment, the value range of the upper limit SOC max of the state of charge is [0.8,0.9], and the value range of the lower limit SOC min of the state of charge is [0.15,0.2];
Calculating local outlier factors in the performance index parameters by using an LOF algorithm, judging whether a j sampling value x ij of an i-th performance index is abnormal data or not by using the obtained local outlier factors, screening and removing the abnormal data in the sampling values of the performance index parameters, wherein the judging process comprises the following steps:
The distance value D i(j-q) between the sampling value x ij and the rest sampling value x iq is calculated respectively, and the calculation formula is as follows: wherein j and q are N, q is not equal to j, and N is the total amount of sampling data;
Sequentially sorting the obtained distance values D i(j-q) from small to large, setting the kth element closest to the sampling value x ij as x ip, p epsilon N, and p not equal to j, obtaining a kth distance value D k(xij of the sampling value x ij and the sampling value x ip), wherein,
A kth distance neighborhood N k(xij) of the sampling value x ij is determined from the kth distance value d k(xij);
Obtaining the kth reachable distance of the sampling value x ij, wherein ,dk(xij,xip)＝max[dk(xij),d(xij,xip)],dk(xij,xip) is the kth reachable distance of the sampling value x ij, max [ d k(xij),d(xij,xip) ] is the maximum value in d k(xij)、d(xij,xip), and d (x ij,xip) is the distance value set between the sampling value x ij and all elements in the kth distance neighborhood N k(xij);
Obtaining a local reachable density of the sampling value x ij, wherein the local reachable density is defined as:
and calculating a local outlier factor of a sampling value x ij according to the acquired local reachable density, wherein a calculation formula is as follows:
Where LOF k(xij) is the local outlier of sample value x ij and ρ k(xip) is the local reachable density of sample value x ip;
And judging the abnormal condition of the sampling value x ij by using the acquired local outlier factor, wherein the judging principle is as follows:
When LOF k(xij) >1, it means that the sampling value x ij is likely to be abnormal data, and when LOF k(xij) is larger, the data abnormality degree is more likely;
When LOF k(xij) is close to 1, it means that the sampling value x ij may belong to the same cluster as the neighborhood point, and is normal data;
When LOF k(xij) <1, it means that the sample value x ij is a dense data point;
Normalizing the screened performance index parameters, the brightness value of the dimmer, the environment parameters and the state of charge value to a [0,1] interval, wherein a normalization processing formula is as follows:
Wherein x' is represented as a data value after normalization processing, x is represented as a data value before non-normalization processing, x max、xmin is respectively represented as a maximum value and a minimum value in the data before non-normalization processing, a brightness value, an environment parameter and a charge state sampling value of a light modulator after normalization processing are taken as input data, a corresponding performance index parameter sampling value after normalization processing is taken as output data, a BP neural network model is constructed, the established neural network model is trained, a target error and the maximum circulation number of the network model are set, when the number of hidden layers of the model is adjusted to be not lower than a predicted accuracy according to the error of a training result, a neural network model for predicting the performance index parameter is obtained, and a calculation formula of a predicted accuracy expected value is that,
Wherein E (Y) is expressed as a predicted accuracy expected value, Q is expressed as an input sample size of the neural network, f (X k) is expressed as an output function of the neural network, X k is expressed as a kth output sample of the neural network;
Obtaining a predicted value of the performance index parameter of the power supply to be detected by using the constructed neural network model, and calculating a correction coefficient k i of each performance index parameter according to the obtained predicted value, wherein the calculation formula is as follows: Wherein, Q is represented as the input sample size of the neural network, x' ij is represented as the predicted value of the performance index parameter obtained by using the neural network model, the sampling value of the performance index parameter is corrected by a correction coefficient k i, the correction value of each performance index parameter is obtained, and the calculation formula is as follows:
A performance detection system of a laser power supply assembly comprises a dimmer, a power meter, a dynamic adjustment module, an environmental parameter acquisition module, a state of charge acquisition module, a data processing module, a neural network construction module and a data correction module;
The light modulator is connected with a power supply to be detected and is used for acquiring a light modulator brightness sampling value in the process that the power supply to be detected is continuously discharged from the state of charge upper limit SOC max to the state of charge lower limit SOC min;
The power meter is connected with the power supply to be detected and is used for acquiring a performance index parameter sampling value comprising rated power, actual power and conversion efficiency in the process that the power supply to be detected is continuously discharged from the state of charge upper limit SOC max to the state of charge lower limit SOC min;
the dynamic adjustment module is used for dynamically adjusting the brightness value of the dimmer in the process that the power supply to be detected is continuously discharged from the upper limit SOC max to the lower limit SOC min;
The environment parameter acquisition module is used for acquiring an environment parameter sampling value comprising the running temperature of the power supply to be detected and the humidity of the environment in the process of continuously discharging the power supply to be detected from the upper limit SOC max to the lower limit SOC min;
The charge state acquisition module is used for acquiring a charge state sampling value of the power supply to be detected;
The data processing module is used for calculating local outlier factors in the performance index parameters by adopting an LOF algorithm, judging whether a j sampling value x ij of an i-th performance index is abnormal data or not by utilizing the obtained local outlier factors, and screening and eliminating the abnormal data in the sampling values of the performance index parameters;
The neural network construction module is used for constructing a BP neural network model which takes a normalized light modulator brightness value, an environment parameter and a charge state sampling value as input data and a corresponding normalized performance index parameter sampling value as output data, training the established neural network model, setting a target error and a maximum circulation number of the network model, and obtaining the neural network model for predicting the performance index parameter when the number of hidden layers of the model is adjusted to be not lower than an expected value according to the error of a training result;
The data correction module is used for obtaining predicted values of the performance index parameters of the power supply to be detected by using the constructed neural network model, and calculating correction coefficients k i of the performance index parameters according to the obtained predicted values, wherein the calculation formula is as follows: Wherein, Q is represented as the input sample size of the neural network, x' ij is represented as the predicted value of the performance index parameter obtained by using the neural network model, the sampling value of the performance index parameter is corrected by a correction coefficient k i, the correction value of each performance index parameter is obtained, and the calculation formula is as follows:
The system includes a memory, a processor, and a computer program stored on the memory and executable on the processor.
The invention has the following advantages that,
Compared with the prior art, the method has the advantages that the performance index parameters and the brightness value, the environment parameter and the state of charge value of the dimmer in the process of continuously discharging the power supply to be detected from the state of charge upper limit SOC max to the state of charge lower limit SOC min are sampled, abnormal data in the performance index parameter sampling values are screened and removed for normalization processing, the normalized dimmer brightness value, the environment parameter and the state of charge sampling value are taken as input data, the corresponding normalized performance index parameter sampling values are taken as output data, a BP neural network model is constructed and trained, the constructed neural network model is utilized to obtain the predicted value of the performance index parameters of the power supply to be detected, the correction coefficient k i of each performance index parameter is calculated according to the obtained predicted value, the sampling value of each performance index parameter is corrected through the correction coefficient k i, the influence of external influence factors such as environment interference and battery state on the dynamic power performance measurement process of the laser power supply is fully considered, the accuracy of the measurement result is improved by correcting the measurement result, and the dynamic power performance of the power supply is reflected.
Drawings
FIG. 1 is a flow chart of a method for detecting performance of a laser power supply assembly according to the present invention;
FIG. 2 is a block diagram of a laser power module performance detection system according to the present invention;
Fig. 3 is a block diagram of a neural network constructed in the scheme of the present invention.
Detailed Description
The present invention will be further described with reference to the following detailed description, wherein the drawings are for illustrative purposes only and are presented as schematic drawings, rather than physical drawings, and are not to be construed as limiting the invention, and wherein certain components of the drawings are omitted, enlarged or reduced in order to better illustrate the detailed description of the present invention, and are not representative of the actual product dimensions.
The specific implementation flow of the technical scheme of the invention comprises the following steps:
Step 1, connecting a power supply to be detected into a dimmer and a power meter, dynamically adjusting the brightness of the dimmer, and sampling performance index parameters, brightness values, environment parameters and state of charge values of the dimmer in the process that the power supply to be detected is continuously discharged from an upper limit of state of charge SOC max to a lower limit of state of charge SOC min;
The performance index parameters comprise rated power, actual power and conversion efficiency, the environment parameters comprise the running temperature of a power supply to be detected and the humidity of the environment, the value range of the SOC max is [0.8,0.9], and the value range of the SOC min is [0.15,0.2];
The specific steps of measuring rated power, actual power and efficiency of a power supply by using a power meter are as follows:
Checking whether the power meter is intact, and ensuring no damage, no rust, no looseness and the like. The stable power supply voltage and the firm connection of the power supply wires are ensured, and the power meter is prevented from being installed in an environment influenced by vibration, moisture and high temperature so as to ensure the stable operation of equipment;
According to the characteristics of the circuit and the requirements of the power meter, a proper connection mode, such as serial connection or parallel connection, is selected, and the circuit to be tested is correctly connected to the input end of the power meter, so that firm connection and good contact are ensured;
Before using the power meter, calibration must be performed to ensure accuracy, and measurement parameters such as power range, sampling rate, display format, etc. are set according to actual requirements;
The stable circuit operates, avoids interference and fluctuation, and selects a measurement function;
Reading a measurement result, observing whether the display is stable or not, and avoiding measurement errors caused by jitter or instability;
The monitored data is recorded, and the rated and actual power, and efficiency of the power supply are calculated.
The calculation formula of the power efficiency is η=P out/Pin ×100%, wherein P out is the output power of the power supply, P in is the input power of the power supply, and in the measurement process, a power meter can be used for directly measuring the output voltage and current to obtain the output power, and meanwhile, the input voltage and current are measured to obtain the input power;
Step 2, calculating local outlier factors in the performance index parameters by using an LOF algorithm, judging whether a j sampling value x ij of the i-th performance index is abnormal data or not by using the obtained local outlier factors, screening and removing the abnormal data in the sampling values of the performance index parameters, wherein the judging process comprises the following steps:
Step 21, respectively calculating a distance value D i(j-q) between the sampling value x ij and the rest sampling values x iq, wherein the calculation formula is as follows: wherein j and q are N, q is not equal to j, and N is the total amount of sampling data;
Step 22, sequentially sorting the obtained distance values D i(j-q) from small to large, setting the kth element closest to the sampling value x ij as x ip, p epsilon N, and p not equal to j, obtaining a kth distance value D k(xij of the sampling value x ij and the sampling value x ip),
Step 23, determining a kth distance neighborhood N k(xij of the sampling value x ij according to the kth distance value d k(xij);
Step 24, obtaining the kth reachable distance of the sampling value x ij, wherein ,dk(xij,xip)＝max[dk(xij),d(xij,xip)],dk(xij,xip) is the kth reachable distance of the sampling value x ij, max [ d k(xij),d(xij,xip) ] is the maximum value operation in d k(xij)、d(xij,xip), and d (x ij,xip) is the distance value set between the sampling value x ij and all elements in the kth distance neighborhood N k(xij);
Step 25, obtaining a local reachable density of the sampling value x ij, wherein the local reachable density is defined as:
Step 26, calculating a local outlier factor of the sampling value x ij according to the acquired local reachable density, wherein the calculation formula is as follows:
Where LOF k(xij) is the local outlier of sample value x ij and ρ k(xip) is the local reachable density of sample value x ip;
step 27, judging the abnormal condition of the sampling value x ij by using the obtained local outlier factor, wherein the judging principle is as follows:
When LOF k(xij) >1, it means that the sampling value x ij is likely to be abnormal data, and when LOF k(xij) is larger, the data abnormality degree is more likely;
When LOF k(xij) is close to 1, it means that the sampling value x ij may belong to the same cluster as the neighborhood point, and is normal data;
When LOF k(xij) <1, it means that the sample value x ij is a dense data point;
step 3, normalizing the screened performance index parameters, the brightness value of the dimmer, the environment parameters and the state of charge value to a [0,1] interval, wherein the normalization processing formula is as follows:
Wherein x' is represented as a numerical value after normalization processing, x is represented as a data value before non-normalization processing, x max、xmin is respectively represented as a maximum value and a minimum value in the data before non-normalization processing, a brightness value, an environment parameter and a charge state sampling value of a light modulator after normalization processing are taken as input data, a corresponding performance index parameter sampling value after normalization processing is taken as output data, a BP neural network model is built, the built neural network model is trained, a target error and the maximum circulation number of the network model are set, when the number of hidden layers of the model is adjusted to be not lower than a desired value according to the error of a training result, a neural network model for predicting the performance index parameter is obtained, and a calculation formula of the desired value of the prediction accuracy is that,
Wherein E (Y) is expressed as a predicted accuracy expected value, Q is expressed as an input sample size of the neural network, f (X k) is expressed as an output function of the neural network, X k is expressed as a kth output sample of the neural network;
For the BP neural network model in the scheme, the input layer comprises a dimmer brightness value, the running temperature of a power supply to be detected, the humidity of the environment and the state of charge, the number of neuron nodes of the input layer is 4, the output layer comprises rated power, actual power and conversion efficiency, the number of neuron nodes of the output layer is 3, the number of neuron nodes of the middle layer can be determined according to an empirical formula, and the method specifically comprises the following steps: wherein a is a constant of 1 to 9, Represented as a pair ofThe calculation results of (2) are rounded upwards, so that the structure of the neural network can be determined to be the input layer of 4 neuron nodes, the middle layer of s neuron nodes and the output layer structure of 3 neuron nodes, as shown in fig. 3;
In this embodiment, taking a neural network model construction by using MATLAB software as an example, training a network by using a standard BP neural network algorithm, setting a target error and a maximum cycle number, selecting an S-type tangent function tansig from a transfer function of a hidden layer neuron, selecting an S-type logarithmic function logsig from a transfer function of an output layer neuron, selecting a traingdx function as a training function, and continuously adjusting the number of intermediate layer neuron nodes to enable the prediction accuracy to be not lower than an expected value, thereby obtaining a neural network model for predicting performance index parameters;
And 4, acquiring predicted values of the performance index parameters of the power supply to be detected by using the constructed neural network model, and calculating correction coefficients k i of the performance index parameters according to the acquired predicted values, wherein the calculation formula is as follows: Wherein, Q is represented as the input sample size of the neural network, x' ij is represented as the predicted value of the performance index parameter obtained by using the neural network model, the sampling value of the performance index parameter is corrected by a correction coefficient k i, the correction value of each performance index parameter is obtained, and the calculation formula is as follows:
The foregoing has shown and described the basic principles and main features of the present invention and the advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (7)

1.一种激光电源组件性能检测方法，其特征在于，包括：1. A method for detecting performance of a laser power supply assembly, comprising: 将待检测电源接入调光器和功率计，动态调节调光器的亮度，对待检测电源由荷电状态上限SOCmax持续放电至荷电状态下限SOCmin过程中的性能指标参数及调光器的亮度值、环境参数、荷电状态值进行采样；The power supply to be tested is connected to the dimmer and the power meter, the brightness of the dimmer is dynamically adjusted, and the performance index parameters of the power supply to be tested during continuous discharge from the upper limit of the state of charge SOC max to the lower limit of the state of charge SOC min , as well as the brightness value, environmental parameters, and state of charge value of the dimmer are sampled; 采用LOF算法计算所述性能指标参数中的局部离群因子，利用获取的局部离群因子对第i项性能指标的第j个采样值xij是否为异常数据进行判定，筛选并剔除性能指标参数采样值中的异常数据；The LOF algorithm is used to calculate the local outlier factor in the performance indicator parameter, and the obtained local outlier factor is used to determine whether the j-th sampling value x ij of the i-th performance indicator is abnormal data, and the abnormal data in the performance indicator parameter sampling value is screened and eliminated; 将经过筛选的所述性能指标参数及调光器亮度值、环境参数、荷电状态值归一化处理至[0，1]区间，以归一化处理后的调光器亮度值、环境参数、荷电状态采样值为输入数据，以对应的归一化处理后的性能指标参数采样值为输出数据，构建BP神经网络模型，对建立的神经网络模型进行训练，设定网络模型的目标误差及最大循环次数，根据训练结果的误差调整模型的隐含层数量至预测准确率不低于期望值时，得到用于预测所述性能指标参数的神经网络模型；The screened performance index parameters and dimmer brightness value, environmental parameters, and state of charge values are normalized to the interval [0, 1], the normalized dimmer brightness value, environmental parameters, and state of charge sampling values are used as input data, and the corresponding normalized performance index parameter sampling values are used as output data, a BP neural network model is constructed, the established neural network model is trained, the target error and maximum number of cycles of the network model are set, and the number of hidden layers of the model is adjusted according to the error of the training result until the prediction accuracy is not lower than the expected value, so as to obtain a neural network model for predicting the performance index parameters; 利用构建的神经网络模型获取待检测电源性能指标参数的预测值，并根据获取的预测值计算各项性能指标参数的修正系数ki，计算公式为：式中，Q表示为神经网络的输入样本量；x'ij表示为利用神经网络模型获取的性能指标参数的预测值，通过修正系数ki对性能指标参数的采样值进行修正，获取各项性能指标参数的修正值，计算公式为： The constructed neural network model is used to obtain the predicted values of the power supply performance index parameters to be detected, and the correction coefficients k i of various performance index parameters are calculated based on the obtained predicted values. The calculation formula is: In the formula, Q represents the input sample size of the neural network; x'ij represents the predicted value of the performance index parameter obtained by using the neural network model. The sampled value of the performance index parameter is corrected by the correction coefficient k i to obtain the corrected value of each performance index parameter. The calculation formula is: 性能指标参数采样值中的异常数据判定流程包括以下步骤：The abnormal data determination process in the performance indicator parameter sampling value includes the following steps: 分别计算采样值xij与其余采样值xiq间的距离值Di(j-q)，计算公式为：其中，j、q∈N，且q≠j，N为采样数据总量；The distance values D i(jq) between the sampled value x ij and the remaining sampled values x iq are calculated respectively, and the calculation formula is: Among them, j, q∈N, and q≠j, N is the total amount of sampled data; 将获取的距离值Di(j-q)按小到大依次排序,设与采样值xij距离最近的第k个元素为xip，p∈N，且p≠j，获取采样值xij与采样值xip的第k距离值dk(xij)，其中， Sort the obtained distance values D i(jq) from small to large, assume that the kth element closest to the sample value x ij is x ip , p∈N, and p≠j, and obtain the kth distance value d k (x ij ) between the sample value x ij and the sample value x ip , where, 根据第k距离值dk(xij)确定采样值xij的第k距离邻域Nk(xij)；Determine the k-th distance neighborhood N k ( xij ) of the sample value xij according to the k-th distance value d k ( xij ); 获取采样值xij的第k可达距离，其中，dk(xij，xip)＝max[dk(xij),d(xij,xip)]，dk(xij，xip)为采样值xij的第k可达距离；max[dk(xij),d(xij,xip)]为取dk(xij)、d(xij,xip)中的最大值操作；d(xij，xip)为采样值xij与第k距离邻域Nk(xij)中所有元素间的距离值集合；Obtain the k-th reachable distance of the sampled value x ij , where d k (x ij , x ip ) = max [ d k (x ij ), d (x ij , x ip )], d k (x ij , x ip ) is the k-th reachable distance of the sampled value x ij ; max [ d k (x ij ), d (x ij , x ip )] is the maximum value operation of d k (x ij ) and d (x ij , x ip ); d (x ij , x ip ) is the distance value set between the sampled value x ij and all elements in the k-th distance neighborhood N k (x ij ); 获取采样值xij的局部可达密度，其中，局部可达密度的定义为：Get the local reachable density of the sample value x ij , where the local reachable density is defined as: 根据获取的局部可达密度计算采样值xij的局部离群因子，计算公式为：The local outlier factor of the sample value x ij is calculated based on the obtained local reachable density. The calculation formula is: 式中，LOFk(xij)为采样值xij的局部离群因子；ρk(xip)为采样值xip的局部可达密度；Where LOF k ( xij ) is the local outlier factor of the sample value xij ; ρk ( xip ) is the local reachability density of the sample value xip ; 利用获取的局部离群因子对采样值xij的异常情况进行判定，判定原则为：The obtained local outlier factor is used to determine the abnormality of the sampling value xij . The determination principle is: 当LOFk(xij)＞1，表示采样值xij可能是异常数据，且当LOFk(xij)值越大时，数据异常程度可能性越大；When LOF k (x ij )>1, it means that the sample value x ij may be abnormal data, and the larger the LOF k (x ij ) value is, the greater the possibility of data abnormality; 当LOFk(xij)接近1，表示采样值xij可能和邻域点属于同一簇，是正常数据；When LOF k (x ij ) is close to 1, it means that the sample value x ij may belong to the same cluster as the neighboring point and is normal data; 当LOFk(xij)＜1，表示采样值xij是密集数据点。When LOF k ( xij ) < 1, it indicates that the sampled value xij is a dense data point. 2.根据权利要求1所述的一种激光电源组件性能检测方法，其特征在于，所述性能指标参数包括额定功率、实际功率、转换效率；2. A laser power supply component performance detection method according to claim 1, characterized in that the performance index parameters include rated power, actual power, and conversion efficiency; 所述环境参数包括待检测电源的运行温度和所处环境的湿度。The environmental parameters include the operating temperature of the power supply to be detected and the humidity of the environment in which it is located. 3.根据权利要求1所述的一种激光电源组件性能检测方法，其特征在于，数据归一化处理公式为：3. A laser power supply component performance detection method according to claim 1, characterized in that the data normalization processing formula is: 式中，x'表示为归一化处理后的数据值；x表示为未归一化处理前的数据值；xmax、xmin分别表示为未归一化处理前的数据中的最大值和最小值。In the formula, x' represents the data value after normalization; x represents the data value before normalization; x max and x min represent the maximum and minimum values of the data before normalization, respectively. 4.根据权利要求1所述的一种激光电源组件性能检测方法，其特征在于，预测准确率期望值的计算公式为，4. A method for detecting performance of a laser power supply assembly according to claim 1, characterized in that the calculation formula for the expected value of the prediction accuracy is: 其中，E(Y)表示为预测准确率期望值；Q表示为神经网络的输入样本量；f(Xk)表示为神经网络的输出函数；Xk表示为神经网络的第k个输出样本。Wherein, E(Y) represents the expected value of prediction accuracy; Q represents the input sample size of the neural network; f(X k ) represents the output function of the neural network; and X k represents the kth output sample of the neural network. 5.根据权利要求1所述的一种激光电源组件性能检测方法，其特征在于，荷电状态上限SOCmax的取值范围为[0.8,0.9]，荷电状态下限SOCmin的取值范围为[0.15,0.2]。5. A laser power supply component performance detection method according to claim 1, characterized in that the upper limit of the state of charge SOC max is in the range of [0.8, 0.9], and the lower limit of the state of charge SOC min is in the range of [0.15, 0.2]. 6.一种激光电源组件性能检测系统，其特征在于，包括调光器、功率计、动态调节模块、环境参数获取模块、荷电状态获取模块、数据处理模块、神经网络构建模块、数据修正模块；6. A laser power supply component performance detection system, characterized by comprising a dimmer, a power meter, a dynamic adjustment module, an environmental parameter acquisition module, a charge state acquisition module, a data processing module, a neural network construction module, and a data correction module; 所述调光器与待检测电源连接，用于获取待检测电源由荷电状态上限SOCmax持续放电至荷电状态下限SOCmin过程中的调光器亮度采样值；The dimmer is connected to the power supply to be detected, and is used to obtain the dimmer brightness sampling value during the process of the power supply to be detected continuously discharging from the upper limit of the state of charge SOC max to the lower limit of the state of charge SOC min ; 所述功率计与所述待检测电源连接，用于获取待检测电源由荷电状态上限SOCmax持续放电至荷电状态下限SOCmin过程中包括额定功率、实际功率、转换效率的性能指标参数采样值；The power meter is connected to the power supply to be detected, and is used to obtain performance indicator parameter sampling values including rated power, actual power, and conversion efficiency during the process of the power supply to be detected continuously discharging from the upper limit of the state of charge SOC max to the lower limit of the state of charge SOC min ; 所述动态调节模块用于在待检测电源由荷电状态上限SOCmax持续放电至荷电状态下限SOCmin过程中，对调光器的亮度值进行动态调节；The dynamic adjustment module is used to dynamically adjust the brightness value of the dimmer during the process of the power supply to be detected continuously discharging from the upper limit of the state of charge SOC max to the lower limit of the state of charge SOC min ; 所述环境参数获取模块用于获取待检测电源由荷电状态上限SOCmax持续放电至荷电状态下限SOCmin过程中包括待检测电源的运行温度和所处环境的湿度的环境参数采样值，其中，荷电状态上限SOCmax的取值范围为[0.8,0.9]，荷电状态下限SOCmin的取值范围为[0.15,0.2]；The environmental parameter acquisition module is used to obtain environmental parameter sampling values including the operating temperature of the power supply to be detected and the humidity of the environment in which the power supply to be detected is located during the process of continuous discharge from the upper limit of the state of charge SOC max to the lower limit of the state of charge SOC min , wherein the value range of the upper limit of the state of charge SOC max is [0.8, 0.9], and the value range of the lower limit of the state of charge SOC min is [0.15, 0.2]; 所述荷电状态获取模块用于获取待检测电源的荷电状态采样值；The state of charge acquisition module is used to acquire a sample value of the state of charge of the power supply to be detected; 所述数据处理模块用于采用LOF算法计算所述性能指标参数中的局部离群因子，利用获取的局部离群因子对第i项性能指标的第j个采样值xij是否为异常数据进行判定，筛选并剔除性能指标参数采样值中的异常数据，判定流程包括以下步骤：The data processing module is used to calculate the local outlier factor in the performance indicator parameter using the LOF algorithm, and use the obtained local outlier factor to determine whether the j-th sampling value x ij of the i-th performance indicator is abnormal data, and screen and eliminate the abnormal data in the performance indicator parameter sampling value. The determination process includes the following steps: 分别计算采样值xij与其余采样值xiq间的距离值Di(j-q)，计算公式为：其中，j、q∈N，且q≠j，N为采样数据总量；The distance values D i(jq) between the sampled value x ij and the remaining sampled values x iq are calculated respectively, and the calculation formula is: Among them, j, q∈N, and q≠j, N is the total amount of sampled data; 将获取的距离值Di(j-q)按小到大依次排序,设与采样值xij距离最近的第k个元素为xip，p∈N，且p≠j，获取采样值xij与采样值xip的第k距离值dk(xij)，其中， Sort the obtained distance values D i(jq) from small to large, assume that the kth element closest to the sample value x ij is x ip , p∈N, and p≠j, and obtain the kth distance value d k (x ij ) between the sample value x ij and the sample value x ip , where, 根据第k距离值dk(xij)确定采样值xiq的第k距离邻域Nk(xij)；Determine the k-th distance neighborhood N k ( xij ) of the sample value x iq according to the k-th distance value d k ( xij ); 获取采样值xij的第k可达距离，其中，dk(xij，xip)＝max[dk(xij),d(xij,xip)]，dk(xij，xip)为采样值xij的第k可达距离；max[dk(xij),d(xij,xip)]为取dk(xij)、d(xij,xip)中的最大值操作；d(xij，xip)为采样值xij与第k距离邻域Nk(xij)中所有元素间的距离值集合；Obtain the k-th reachable distance of the sampled value x ij , where d k (x ij , x ip ) = max [ d k (x ij ), d (x ij , x ip )], d k (x ij , x ip ) is the k-th reachable distance of the sampled value x ij ; max [ d k (x ij ), d (x ij , x ip )] is the maximum value operation of d k (x ij ) and d (x ij , x ip ); d (x ij , x ip ) is the distance value set between the sampled value x ij and all elements in the k-th distance neighborhood N k (x ij ); 获取采样值xij的局部可达密度，其中，局部可达密度的定义为：Get the local reachable density of the sample value x ij , where the local reachable density is defined as: 根据获取的局部可达密度计算采样值xij的局部离群因子，计算公式为：The local outlier factor of the sample value x ij is calculated based on the obtained local reachable density. The calculation formula is: 式中，LOFk(xij)为采样值xij的局部离群因子；ρk(xip)为采样值xip的局部可达密度；Where LOF k ( xij ) is the local outlier factor of the sample value xij ; ρk ( xip ) is the local reachability density of the sample value xip ; 利用获取的局部离群因子对采样值xij的异常情况进行判定，判定原则为：The obtained local outlier factor is used to determine the abnormality of the sampling value xij . The determination principle is: 当LOFk(xij)＞1，表示采样值xij可能是异常数据，且当LOFk(xij)值越大时，数据异常程度可能性越大；When LOF k (x ij )>1, it means that the sample value x ij may be abnormal data, and the larger the LOF k (x ij ) value is, the greater the possibility of data abnormality; 当LOFk(xij)接近1，表示采样值xij可能和邻域点属于同一簇，是正常数据；When LOF k (x ij ) is close to 1, it means that the sample value x ij may belong to the same cluster as the neighboring point and is normal data; 当LOFk(xij)＜1，表示采样值xij是密集数据点；When LOF k (x ij ) < 1, it means that the sample value x ij is a dense data point; 所述神经网络构建模块用于构建以归一化处理后的调光器亮度值、环境参数、荷电状态采样值为输入数据，以对应的归一化处理后的性能指标参数采样值为输出数据的BP神经网络模型，并对建立的神经网络模型进行训练，设定网络模型的目标误差及最大循环次数，根据训练结果的误差调整模型的隐含层数量至预测准确率不低于期望值时，得到用于预测所述性能指标参数的神经网络模型；The neural network construction module is used to construct a BP neural network model with the normalized dimmer brightness value, environmental parameters, and charge state sampling values as input data, and the corresponding normalized performance index parameter sampling values as output data, and train the established neural network model, set the target error and maximum number of cycles of the network model, and adjust the number of hidden layers of the model according to the error of the training result until the prediction accuracy is not lower than the expected value, so as to obtain a neural network model for predicting the performance index parameter; 所述数据修正模块用于利用构建的神经网络模型获取待检测电源性能指标参数的预测值，并根据获取的预测值计算各项性能指标参数的修正系数ki，计算公式为：式中，Q表示为神经网络的输入样本量；x'ij表示为利用神经网络模型获取的性能指标参数的预测值，通过修正系数ki对性能指标参数的采样值进行修正，获取各项性能指标参数的修正值，计算公式为： The data correction module is used to obtain the predicted value of the power performance index parameter to be detected by using the constructed neural network model, and calculate the correction coefficient k i of each performance index parameter according to the obtained predicted value. The calculation formula is: In the formula, Q represents the input sample size of the neural network; x'ij represents the predicted value of the performance index parameter obtained by using the neural network model. The sampled value of the performance index parameter is corrected by the correction coefficient k i to obtain the corrected value of each performance index parameter. The calculation formula is: 7.根据权利要求6所述的一种激光电源组件性能检测系统，其特征在于，所述系统包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，其中，处理器执行程序时实现权利要求1-5中任一项所述方法的步骤。7. A laser power supply component performance detection system according to claim 6, characterized in that the system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in any one of claims 1 to 5 when executing the program.
CN202411561878.2A 2024-11-05 2024-11-05 Laser power supply component performance detection method and system Active CN119064820B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202411561878.2A CN119064820B (en) 2024-11-05 2024-11-05 Laser power supply component performance detection method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202411561878.2A CN119064820B (en) 2024-11-05 2024-11-05 Laser power supply component performance detection method and system

Publications (2)

Publication Number Publication Date
CN119064820A CN119064820A (en) 2024-12-03
CN119064820B true CN119064820B (en) 2025-02-25

Family

ID=93635568

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202411561878.2A Active CN119064820B (en) 2024-11-05 2024-11-05 Laser power supply component performance detection method and system

Country Status (1)

Country Link
CN (1) CN119064820B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109031158A (en) * 2018-08-15 2018-12-18 大族激光科技产业集团股份有限公司 The test macro and method of laser power supply
CN116882303A (en) * 2023-09-06 2023-10-13 深圳市联明电源有限公司 Laser power supply life prediction method, system and storage medium

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118011265B (en) * 2024-04-10 2024-06-21 深圳市联明电源股份有限公司 Laser power supply aging test method and system

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109031158A (en) * 2018-08-15 2018-12-18 大族激光科技产业集团股份有限公司 The test macro and method of laser power supply
CN116882303A (en) * 2023-09-06 2023-10-13 深圳市联明电源有限公司 Laser power supply life prediction method, system and storage medium

Also Published As

Publication number Publication date
CN119064820A (en) 2024-12-03

Similar Documents

Publication Publication Date Title
CN116879662B (en) Transformer fault detection method based on data analysis
KR102106775B1 (en) Diagnosis method of battery using Deep Learning
CN111190113B (en) A method for detecting abnormal performance degradation of spacecraft battery
CN115902647B (en) A battery state intelligent monitoring method
TWI759027B (en) Battery performance evaluation method and battery performance evaluation device
CN112285587A (en) An online vehicle-mounted hydrogen fuel cell impedance spectrum measurement device and measurement method
CN116819377B (en) Battery pack abnormality identification diagnosis method and system
CN117936955A (en) Energy storage system and battery internal resistance detection method
CN118551334A (en) Performance test system for aluminum electrolytic capacitor
CN118914866B (en) Battery state evaluation system for energy storage power station
CN119064820B (en) Laser power supply component performance detection method and system
CN118981730A (en) A method for intelligently detecting residual energy of batteries for safe utilization
CN117233629B (en) Method, system, equipment and medium for testing electrical performance of lithium ion battery
CN114859249B (en) Method and device for detecting battery pack capacity
CN113191075B (en) Photovoltaic array fault diagnosis method based on improved goblet sea squirt group algorithm
CN118914865B (en) Battery state evaluation system
CN117907845B (en) Electrochemical energy storage system insulation detection method based on electrical parameter analysis
CN118501791B (en) Metering detection data precision evaluation method and system
KR102745775B1 (en) Battery Management Apparatus and Method Based on Impedance Data Type
CN117406158B (en) Calibration method and device for lithium ion battery short circuit tester
CN119147882B (en) Photovoltaic inverter electrical performance detection method and system
Gu et al. Physics-based battery SOC estimation: A joint Data-driven and window-varying adaptive extended Kalman filter approach
CN119247177A (en) A method for measuring battery internal resistance and self-discharge rate through battery impedance chip
CN119104917A (en) A fast online prediction method for proton exchange membrane fuel cell health status and electrochemical performance
CN118311489A (en) Online monitoring system and method of DC bus capacitor bank based on digital twin

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant