JP3173550B2 - Air conditioner operation control device and control method - Google Patents

Abstract

PURPOSE:To provide a method to search a most suitable combination of air- conditioning devices at a high speed for equipment which is equipped with a plurality of air-conditioning devices wherein the capabilities are variable, e.g. can be switched at every 1%. CONSTITUTION:A control device is constituted of a genetic neuro foreseen fuzzy air- conditioning control unit 3, display device 2 which displays a control/measuring condition, memory device 4 which stores the result of the control/measurement in a time series, measured value inputting device 5 which fetches the inputs from various types of measuring instruments being provided in a control objective room 1, equipment condition inputting device 6 which inputs the operation status of the air-conditioning equipment, and equipment control outputting device 7 which outputs control signals to a group of effect apparatus being provided in the control objective room 1. An objective function which shows items to be minimized or maximized in the control of an operating number of air-conditioning machines is prepared, and one operation plan wherein the evaluation value of the objective function which is determined by a genetic optimizing operation is a maximum or minimum is made an optimum operation candidate.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an air conditioner control system, and more particularly, to an air conditioner control suitable for determining an optimum operating condition of an air conditioner in an apparatus in which a plurality of air conditioners such as heaters and coolers are combined. Method and apparatus.

[0002]

2. Description of the Related Art Conventionally, the optimal operation of an air conditioner has been determined by searching for all possible combinations of equipment operable conditions based on foreseeable fuzzy logic. However, the required responsiveness cannot be ensured due to an increase in the number of operating combinations as the number of equipment increases. There could have been a case.

For example, there are one or more effectors such as a cooler, a heater, a humidifier, a ventilator, a fan, etc. as air conditioning equipment, and the temperature (T (t): temperature at time t), humidity W (T), the CO ₂ concentration C (t) is obtained, and it is assumed that the target values To, Wo, and Co are given to each. Here, a case is considered where there are two coolers, two heaters, one humidifier, and one ventilator of the ventilation effector. The cooler and the heater may have the same capacity or may have different capacities. Cooler, heater, humidifier, each rated capacity of the ventilator is, CL _1, CL _2, H
T ₁ , HT ₂ (the unit is kcal / h), K (m ³ / h)
(Volume of water vaporized in 1 hour), F (m ³ / h) (air exchange volume of ventilator) are known, and it is assumed that the capacity can be switched every 1% from 0 to 100%. .

In this case, the control combination of the effector is 101
⁶ (> 10 ⁷⁾ passes realistic, heater and because cooler not move at the same time, 101 ⁴ (> 10 ⁵⁾ becomes as, cooler, combinations each one by increasing the heater Number 10 ²
Multiply by two. Searching for all of these combinations is
It becomes impossible with the increase in the number.

[0005]

SUMMARY OF THE INVENTION The present invention has a capacity of 0 to 0.
It is an object of the present invention to provide a method for quickly searching for an optimal combination of air conditioners in a facility having a plurality of air conditioners that can be switched every 1% of 100%.

Another object of the present invention is to provide an air conditioner operation control method and apparatus capable of improving the accuracy of predicting the air conditioning load and reducing unnecessary starting and stopping of the air conditioner.

Another object of the present invention is to provide an air conditioner operation control method and apparatus capable of reducing the required power by reducing the air conditioner operation switching amount or the operation switching frequency.

[0008]

SUMMARY OF THE INVENTION In order to achieve the above object, the present invention provides a genetic algorithm and a mutual integration type neural network which is capable of switching between 0% and 100% capable of switching every 1% at high speed. Allows you to search for combinations.

That is, the air conditioner control device of the present invention comprises: a process input means for receiving information from a sensor for measuring process information necessary for controlling the operation of the air conditioner; Genetic neuro-fuzzy control means for defining an operation plan of an air conditioner control means for minimizing or maximizing the value of the created objective function, and generating an objective function representing an item to be maximized or maximized. Storage means for storing variables necessary for determination, and output means for outputting the determined operation plan to the air conditioner control means, the genetic neuro-fuzzy control means,
For a plurality of different driving plans having a predetermined chromosome expression, the values of a plurality of target items when the driving is performed are foreseen and calculated, and the calculation is performed by a fuzzy membership function determined in advance. Predictive fuzzy evaluation means for determining evaluation values of a plurality of target item values obtained, and a selection constant determined in a predetermined order with respect to another operation plan of the evaluation value of the relevant operation plan from among a plurality of predetermined operation plans. , A new operation plan is generated by performing a mating operation of exchanging the properties of the two parents with each other according to a predetermined exchange template, and a new operation plan is generated. A genetic optimization unit that rearranges the operation plans in a predetermined order according to the evaluation value of the generated operation plan when the number of the plans reaches a predetermined number, and the evaluation determined by the genetic optimization unit. Calculates one of the maximum or minimum operation plans as an initial operation plan, calculates an evaluation value of an operation plan obtained by changing a part of the operation plan by a predetermined constant, and compares it with the evaluation value of the previous operation plan. When the difference between the evaluation values is smaller than the value of a preset variable, a neuro-optimizing means is provided which replaces the previous operation plan with the operation plan generated this time and sets the operation plan as an optimal operation candidate.

[0010]

According to the air conditioner operation control method of the present invention, a plurality of air conditioner combination plans are generated based on the control target data of the control target room and the operation state data of the air conditioner in the air conditioner room. A simulation is performed for each of the air conditioner operation combinations until a predetermined time has elapsed, and as a result of the simulation, the operation of the air conditioner is controlled by the air conditioner operation combination plan with the best evaluation. Captures information from sensors that measure the process information needed to control the operation of the machine,
To create an objective function representing an item to be minimized or maximized in the control of the number of operating air conditioners, to determine an operation plan of an air conditioner control unit that minimizes or maximizes the value of the created objective function, For a plurality of different driving plans having a predetermined chromosome expression, the values of a plurality of target items when the driving is performed are foreseen and calculated, and the calculation is performed by a fuzzy membership function determined in advance. An evaluation value of a plurality of target item values obtained is determined, and two types of predetermined operation plans are selected based on a selection constant determined in a predetermined order with respect to another operation plan of the evaluation value of the operation plan in advance. By selecting different parents and performing a mating operation for exchanging the properties of the two parents with each other by a predetermined exchange template, a new operation plan is generated, and the generated operation plan is generated. When a certain number is reached, a genetic optimization operation of rearranging the operation plans in a predetermined order according to the generated evaluation value of the operation plan is performed, and the evaluation value determined in the genetic optimization operation is the maximum or The minimum one operation plan is set as an initial operation plan, an evaluation value of the operation plan obtained by changing a part of the operation plan by a predetermined constant is calculated, compared with the evaluation value of the previous operation plan, and the evaluation value is calculated. If the difference value is smaller than the value of a preset variable, the previous operation plan is replaced with the operation plan generated this time, and the determined operation plan is output to the air conditioner control means, and is set as an optimal operation candidate. It is characterized by.

In order to find the air conditioner operation combination with the best evaluation, prediction calculations are performed by a membership function determined in advance based on a priori knowledge, and the prediction calculation amounts are evaluated as fuzzy quantities. A load average is calculated from load values determined in advance based on a priori knowledge with respect to the evaluation result, and an air conditioner operation combination plan that maximizes the load average is adopted as the plan with the best evaluation.

In the present invention, the relationship between the state and the evaluation is described as a membership function by applying fuzzy logic and stored, so that a general operator can make the same judgment as a skilled operator. I have to. In addition, a load as a control target is provided to a plurality of criteria, so that the air conditioner operation control can be performed according to the decision of the operator.

Here, it is considered that the present invention is applied to a case where there are two coolers, two heaters, one humidifier, and one ventilator of the ventilation effector.
That is, the cooler and the heater may have the same capacity or may have different capacities. Cooler, heater, humidifier,
The rated capacity of each ventilator is known and the capacity is 0 to 100%
It is assumed that the equipment can be switched every 1%. Since the control combination of effectors of cases, the 101 ⁶ (> 10 ⁷⁾ as practical, the heater and the cooler are not simultaneously moved, 101 ⁴
(> 10 ⁵ ), and the number of coolers and heaters increases, and the number of combinations increases by 10 ² times for each unit. Here, since the operation capability of the equipment can be expressed by 8 bits, in the above case, it can be expressed by 8 (bits) × 6 (equipment) = 48 bits. In particular, the exclusive control of the heater and cooler can be expressed by 33 bits.

[0014]

FIG. 1 shows the configuration of an embodiment of the present invention. The air-conditioning control device of the present invention includes an air-conditioning control unit 3 based on genetic neuro-fuzzy, a display device 2 for displaying control / measurement status, a storage device 4 for storing control / measurement results in time series, and a control target room 1. A measurement value input device 5 for inputting inputs from various installed measuring instruments, an equipment status input device 6 for inputting the operation status of the air conditioning equipment, and an equipment for outputting control signals to an effector group installed in the control target room 1 It comprises a control output device 7.

One or more coolers 11, heaters 12, humidifiers 13, and ventilators 1 are provided as air conditioning equipment in a space partitioned by walls, floors, ceilings, and the like as the room 1 to be air-conditioned.
4. There are effectors such as fans. The temperature (T (t): the temperature at time t) from the space by the thermometer 16;
The humidity W (t) is obtained by the hygrometer 15, and the CO ₂ concentration C (t) is obtained by the CO ₂ concentration meter 17. In addition, each temperature,
It is assumed that target values To, Wo, and Co of the humidity and the CO ₂ concentration are given.

The air-conditioning control unit 3 is a means for determining an optimal operation plan of the control effector, and includes a genetic optimization processing unit 3A, a neuro-optimization processing unit 3B, a preview fuzzy evaluation unit 3C, and an overall control processing unit 3D. Consists of

FIG. 2 shows a software configuration of the air-conditioning control unit 3. The genetic optimization processing unit 3A and the neuro-optimization processing unit 3B are both a mating operation processing unit 31, a control preview simulation. It has a unit 32, a fuzzy evaluation unit 33, a control output unit 34, and a memory 35.

The mating operation processing unit 31 generates all the operation combinations of the cooler and the heater 5 installed in the control target 1 and controls the combinations one by one.
Output to 2.

The feature of this embodiment is that real-time simulation and prediction of how the air conditioners 11 to 14 are operated at the control timing to obtain the best results are provided.
The point is to evaluate the goodness of each driving operation plan and output the best operation plan as a control command. Operation plan creation means 3
1 generates a combination of the operation of the air conditioner to be applied to the process of repeating the prediction and the evaluation. There are four types of control effectors in this process: a cooler 11, a heater 12, a humidifier 13, and a ventilator 14, each of which can be switched from 0 to 100% at 1% intervals.

In the present invention, the evaluation of the set values of temperature and humidity and the permissible range when all the operation plans are executed, and the evaluation of the amount of electric power used and the amount of electric power for switching are performed, and the best operation plan is determined. Control output. Operation plan generation means 3
1 generates a plan that can be operated in this way. Control is 1
It is performed periodically at intervals of up to 5 minutes while predicting the transition of the process after a certain time.

Here, two coolers 11 for the ventilation effector are provided.
Two heaters 12, one humidifier 13 and one ventilator 14
Consider a case where there are tables. The cooler and the heater may have the same capacity or may have different capacities. Cooler,
The rated capacity of each of the heater, humidifier and ventilator is CL ₁ ,
CL ₂ , HT ₁ , HT ₂ (kcal / h), K (m ³ / h)
It is assumed that the equipment has a known volume (volume of water vaporized in one hour) and F (m ³ / h) (air exchange volume of the ventilator), and has a capacity of 0 to 100% and can be switched every 1%. The effector control combination in this case is actually 10 ¹⁶ (> 10 ⁷ ) as follows:
Since the heater and cooler are not moved at the same time, 101 ⁴ (>
10 ⁵ ), and the number of coolers and heaters increases, and the number of combinations increases by 10 ² times for each unit. Here, since the operating capacity of the equipment can be expressed by 8 bits, in the above case, 8 (bits) × 6 (equipment) as shown in FIG.
BR> = 48 bits.

In particular, when the heater and the cooler are exclusively controlled, they can be expressed by 33 bits as shown in FIG. 3B.

FIG. 7 shows the transition of the adaptive value obtained by calculating the fuzzy evaluation value in a three-dimensional graph.
Here, the adaptive value is defined as (adaptive value) = (temperature deviation fuzzy evaluation value) + (humidity deviation fuzzy evaluation value) + (power amount fuzzy evaluation value) + (switching power amount fuzzy evaluation value). As shown in FIG. 7, (adaptive value) is a multimodal function having several vertices (peaks). The driving plan that gives the highest value among these peaks is the optimal solution. In order to determine the optimum solution, it is sufficient to consider all driving combinations, but as described above, it is difficult to perform the calculation within real time (several seconds).

The present invention solves the above-mentioned problem as follows. FIG. 4 shows a processing procedure of the air-conditioning control unit 3 by the genetic neuro-fuzzy control. This process is a genetic optimization process (step A) shown in detail in FIG.
, A sub-optimal solution is obtained by a global search. Next, an optimal solution is obtained by a neuro-optimizing process (step B) shown in FIG. 10 in detail, and the obtained optimal solution is converted into control information. The output (step C) consists of three types of processing.

FIG. 5 shows the details of the procedure of the genetic optimization processing (step A). The main purpose of this process is to find a sub-optimal solution in a short time.
Therefore, a plurality of operation plans are prepared in advance, and these are optimized as individuals of the genetic algorithm.

FIG. 6 shows the concept of selecting an initial individual. In order to search a space as wide as possible, the operation range of each control effector is equally divided, and representative points (grid points in the figure) are determined.
In the example of FIG. 6, the intersection of the ventilation function power (a1) and the cooler capacity (a2) is set as the initial individual.

Using these as initial values, optimization processing is performed in the following steps shown in FIG.

(Step A1) Following (Step A2)
(Step A6) is repeated for the number of generations. (Step A2) The following (Step A3) to (Step A5) are repeated the maximum number of individuals. (Step A3) Two kinds of parents are selected, a mating operation is performed using a template, and a new driving plan is generated. (Step A4) When the generated operation plan is executed,
Foreseeing and calculating the control purpose item. (Step A5) In the above procedure, each control target item value is evaluated by a fuzzy amount, and an adaptive value is obtained. (Step A6) The generated operation plans are rearranged in the descending order of the adaptive values.

Hereinafter, each processing step of FIG. 5 will be described. Here, the maximum value of the individual as the driving plan is 1
It is assumed that there are 0 and these are x1 to x10. It is assumed that each adaptive value is expressed as a function f. Since the purpose of the control is to determine the highest adaptive value,
It is necessary to select a driving plan with a high adaptation value from the ten driving plans as the parent of the next-generation driving plan. Therefore, first, the driving plans are rearranged in descending order of the evaluation value, and the order is determined. When the rank is determined, the probability of selection as a parent can be determined in advance for each generation, and the parent can be determined from a parent selection table having a range of 1 to 1000 by a uniform random number. In the parent selection table generation 1 shown in FIG. 8, for example, since the first uniform random number is i = 182, x1 is selected as the parent. Similarly, when uniform random number i = 430, x2 is selected. In this way, each driving plan is selected as a parent with a probability corresponding to the rank of the adaptive value for each generation.

FIG. 8 shows the properties determined by the parent selection table for each generation in the order of the adaptation value of the driving plan, and is an explanatory diagram of the important properties in the above-mentioned selection in evolution and dominant inheritance. The sum total of the evaluation values in the descending order arrays a (1) to a (10) of the evaluation values corresponding to the driving plan of the generation = 1 is: Σa (i) (i = 1 to 10) = 303. FIG. 9 is a trend graph in which the result of calculating the ratio of each evaluation value to the total sum is set as ap (%) in FIG. This approximates an exponential function having an index P (gn) determined by generation = gn,

[0031]

(Equation 1)

## EQU2 ## Here, C (gn) is a constant determined by the generation gn.

The lower left diagram is a pie chart in which the degree of dominance of each individual is determined by each evaluation value ratio (%), which is called a roulette wheel. The mechanism of dominant inheritance is that the higher the adaptability to the environment to which the individual belongs, the higher the survival rate and the preservation rate of offspring. Therefore, the higher the evaluation value ratio (%) is, What is necessary is just to raise the probability of becoming. For example, if the integer random number i1 generated at the first time is i1 = 182, it indicates the range from 1 to 277, and X1 is selected as the parent individual. In the case of the second uniform random number i2 = 492, the scale is 475 to 62.
X3 is selected as the parent individual to fall within the range of 7.
In this way, uniform random numbers for the number of parent individuals required for the genetic operation are generated, and the roulette wheel is searched to determine the parent individuals.

FIG. 8 shows that the above processing is performed from generation 1 to generation 20.
2 shows the transition of the evaluation value of a living individual when a certain genetic operation is performed. According to the lower left figure, in the generations at the initial stage, the evaluation value has a large variation, but becomes smaller as the generation proceeds, and consequently converges to the vicinity of the optimal value when the generation reaches 20. Understand. The lower right figure is a plot of the same data as an evaluation value ratio (%). It can be seen that the above (Equation 1) is obtained. In the control problem to be optimized in the present embodiment, the position and the number of control effectors were changed, and statistics and analysis were performed. As a result, it was found that P (gn) and C (gn) are affected by the initial parent individual and the genetic operation, but become the same when the parameters and the genetic operation procedure of the initial individual are fixed. That is, if the optimization target and the genetic operation are determined, it is indicated that the parent dominantly selected for each generation is determined by the evaluation value rank. That is, the roulette wheel creation processing and the generation of uniform random numbers do not need to be performed each time during the planning process, and can be determined deterministically in advance.

As described above, the genetic optimization means 3A
Thus, a sub-optimal solution can be obtained by examining a very small number of operation plans.

Next, the processing of step A3 in FIG. 5 will be described in detail with reference to FIG. In FIG. 9, by the above operation, X1 = (a1, a2, a3, a4, a5) = (1, 72, 0, 100, 0) X2 = (0, 0, 50, 60, 20) is selected, and a 33 (bit) unsigned random number PTN
(Referred to as a template). Here, each bit of the parent corresponding to 1 in the PTN is exchanged with each other, and new child operation plans Y1 and Y2 are generated. Y1 = (1,64,2,92,0) Y2 = (0,8,48,36,20) Y1 and Y2 inherit the properties of parents X1 and X2, respectively.

Next, step B in FIG. 4, that is, the processing of the neuro optimizing means 3B will be described with reference to FIG. The processing of this means comprises the following six steps. (Step B1) The following (Step B2) to (Step B6) are repeated for the allowable number of control cycles (time). (Step B2) A part of the previous operation plan is changed to generate a new operation plan. (Step B3) Foreseeing calculation of the control purpose item value when the new operation plan is executed. (Step B4) Each control target item is evaluated with a fuzzy quantity to determine an adaptive value. (Step B5) The difference value between the adaptation value of the previous operation plan and the adaptation value of the new operation plan is calculated and compared with a predetermined value.
If the new step change plan is excellent, the following (step B6)
Proceed to. Otherwise, return to (Step 1). (Step B6) Return to (Step B1) the new driving plan as the optimal solution candidate.

Hereinafter, each step will be described in detail. FIG. 11 shows an example of the processing operation of step B2 in FIG. In the present example, the sub-optimal operation plan X determined in the genetic optimization processing is X = (1,100,50,60,0). A part of this operation plan is changed by the method shown in FIG. That is, when all situations where each control effector can be operated are arranged, 2 + 101 + 101 + 101 + 101 = 406
There are different driving situations. Here, in order to give the minimum change to this operation plan, a uniform integer random number m in the range of 0 to 405 is generated. For example, if m = 104, it is understood that this is 1% of the cooler (a3) that is the control effector. Therefore, as shown in FIG. 11, the capacity of the cooler (a3) is changed from 50% to 1% to obtain a new operation Y. Y = (1,100,1,60,0) Next, in step B3, a control target item value when the new operation plan is executed is foreseen and calculated, and in step B4, the fuzzy adaptive value is calculated as described above. We ask by procedure. At this time,
The adaptation value of the pre-rotation operation plan is f (X), and the adaptation value of the new operation plan is f
(Y).

Next, step B5 and step B6 in FIG.
Will be described. FIG. 13 shows the unique properties of the transition of the objective function value found in the present invention. FIG. 13A defines a planning order vector X (i) determined for each planning number i on the horizontal axis and its objective function value on the vertical axis.

The arrangement of the planning order vectors is changed little by little, and the objective function values at that time are compared and properly replaced with a planning order vector close to the optimal solution.

In this case, as shown in FIG.
Even in the case of the extreme point (minimum point in this embodiment) as shown in (i), it can be seen that the process shifts to the next vector and reaches the optimal solution vector Xop after planning a sufficient number of times.

Therefore, a new operation plan Xi + 1 is created each time, and its adaptive value f (Xi + 1) is calculated,
What is necessary is just to perform processing of making smaller or larger ones as the optimum driving plan candidates compared with the adaptation value f (Xi) of the previous driving plan Xi.

In FIG. 13B, the horizontal axis represents the planned order vector Xi, and the vertical axis represents the differential value Δf (Xi) of the adaptive value f (Xi) with respect to Xi. That is, Δf (Xi) = dF (Xi) / dXi = f (Xi + 1) −f (Xi).

In this example, Δf (Xi) = (Y) −f (X).

This represents the adaptation value difference value for each planning count, and becomes negative when the adaptation value of the new operation plan “Xi + 1” is smaller than the adaptation value in the previous plan,
Otherwise, it is zero or positive.

The transition is, as shown in the figure, f (Xi)
However, the waveform is similar to a damped oscillation centered at zero (that is, a waveform whose amplitude is reduced and whose period is also changed).

According to rigorous experiments, a general control problem has such a shape regardless of the content of the problem.
It was found that the degree of attenuation of the amplitude when Xi was considered as a time was, for example, similar to C / log (i + 2) (C = 1.0 in this embodiment).

This is because when ΔF becomes positive, that is,
Even when the adaptation value of the new plan is worse than the previous plan, if the adaptation value of the new plan is smaller than C / log (i + 2), the forward vector corresponding to sufficiently large i is obtained by replacing the new operation plan with the optimal solution candidate. Xi indicates that the optimal solution is always reached.

By repeating this process sufficiently within the permissible control time, the optimum solution can be surely approached.

FIG. 14 shows an example of constants stored in the storage device 4 in advance. As described above, the table M contains the number of situations in which each control effector can operate, that is, 0.
４０ 405 are stored as positive random numbers uniformly distributed. C is a constant for comparison with the adaptive value difference (f (Y) −f (X)). For example, for the number of repetitions i, the following value:
It is set on a table provided in the storage means 5. C (i) = (C ₁ · a (i)) / log (i + 2) where log is a natural logarithm, and a (i) is from “0.0” to “1.
It is a uniform real random number distributed to "0".

That is, any value may be used as long as the range of distribution becomes smaller with an increase in the number i of examination repetitions. C
₁ can be determined by the adaptive value range. In this example, C ₁ =
The transition of the distribution area of C (i) when 1.0 is set is shown. C (i) will be present in the shaded area.

FIG. 15 is a diagram showing the concept of the processing in the neuro-optimizing means 3B described above. The horizontal axis indicates the change in the capacity of the cooler (a2), and the vertical axis indicates the adaptation value.
This is a situation where the optimum value of the adaptation value is reached. Other control effectors are omitted in the drawing. It can be seen that, as an initial value of this process, the optimal solution is reached through some extreme values, with the suboptimal operation plan determined by the genetic optimization process as an initial value. When the initial operation plan is determined by a simple random method, the time to reach the optimal solution (the number of studies) varies greatly. However, in the present invention, a sufficiently good operation plan is initially given by the genetic optimization method. Therefore, the time to reach the optimal solution was set to be sufficiently short and stable.

As described above, the neuro-optimizing means 3B
Uses the sub-optimal operation plan obtained by the genetic optimization means 3A as an initial operation plan and reliably determines an operation plan close to the optimal plan within an allowable time.

FIG. 16 shows an example of the result of controlling the capacity of an air conditioner such as a heater or a cooler by the control device according to the present invention. In the conventional method, the value becomes less than the reference value several times, but in the present invention, the value is more than the reference value. The operation of the heater is also shown in FIG.
It can be seen that since the number of startups and the abrupt change in the amount of operation are small, the amount of power is also greatly reduced as compared with the conventional method.

FIG. 17 shows an experimental result of the processing performance of determining the optimum operation plan according to the present invention. As a conventional method, a standard genetic algorithm (GA) using a mating operation and a mutation operation as a genetic operation, and a Boltzmann machine having a neural network structure are used.

In the conventional GA, 0.3 (second) from the start of processing
The adaptation value increases well until later, but the increase rate after 0.3 seconds is small, and almost no improvement is seen. On the other hand, in the neuro (Boltzmann machine), the adaptation value increases gradually from the start of the processing, but is steadily performed, and is better than the best value by GA 2.3 (seconds) after the start of the processing. You can see that.

On the other hand, according to the present invention, from the start, it is set to be 0.
After about 3 (seconds), there was a sharp improvement in the adaptive value, and
Even after 3 (seconds), the improvement is steadily achieved, and after 2.3 (seconds), the solution almost reaches the optimum.

Returning to FIG. 2, the control preview simulation unit 32 has the configuration shown in FIG. This means comprises a calorific value calculating means 32 connected to the operation plan generating means 31.
1 and an externally affected calorie computing means 3 connected to the output side of the generated calorie computing means 321 and to which the room temperature and the outside air temperature are inputted.
22, an air conditioner calorie computing means 323 connected to the output side of the external influence calorie computing means 322, and a temperature predictive computing means 324 connected to the output side of the air conditioner calorie computing means 323 and outputting a predicted temperature value. , A latent heat load calculating means 325 connected to the operation plan generating means 31, and an air conditioner latent heat calculating means 32 connected to the output side of the latent heat load calculating means 325.
6, a humidity prediction calculating means 327 connected to the output side of the air conditioner latent heat calculating means 326 and outputting a predicted humidity value, and an electric energy prediction calculating means connected to the operation plan generating means 31 and outputting a predicted electric energy value 328 and the operation plan generating means 31
And a switching power amount calculating means 329 for inputting each air conditioner state and outputting a switching power amount prediction value, wherein the temperature, the humidity, the power amount, and the switching when each of the operation plans are executed. The predicted value of the electric energy is output.

[Temperature Prediction] The temperature in the room is determined mainly by the respiratory heat of a person, the thermal influence from the outside, and the heat generated by the operation of the air conditioner, as shown in the calorific value differential equation with respect to time (Equation 2 below). Is done. Where ΣCd: indoor heat capacity = (human heat capacity) + (equipment heat capacity) Ca: indoor air heat capacity Qout: heat from outside air = (heat transfer from wall) + (heat transfer from room entrance) Σ Qms: air conditioner Sensible heat load = (sensible heat load by heater operation) + (sensible heat load by cooler operation) + (sensible heat load by humidifier operation) + (sensible heat load by ventilator operation) Qbs: Human breathing and generated heat (Equation 2) The predicted value of the temperature after Δt time with respect to the current time t is expressed by Equation 1.
It is determined by the temperature prediction formula shown in the following Expression 3 derived from the above equation. Here, T (t) is the current temperature T (t + Δt) is the predicted temperature value after Δt time (3) In this equation, the heat quantity Qout from the outside air and the sensible heat of the air conditioner The load ΣQms and the like can be obtained by calculation based on the sensor information, the operating state of the air conditioner, and the operation plan. On the other hand, although there is an individual difference, the respiratory heat Qbs of a person can be obtained from a priori statistical information.

The respiratory heat calculation means 321 calculates respiratory heat Qbs. The externally-affected heat amount calculating means 322 calculates the externally-affected heat amount Qout based on the outside air temperature from the outside air temperature sensor, the temperature in the processing chamber, and the heat transmittance of a wall or door at an entrance or the like which is known in advance. The air conditioner calorie calculating means 323 calculates the sensible heat of the air conditioner for each operation plan. The temperature prediction calculation means 324 executes the calculation of Expression 2 based on the above result,
The temperature prediction value Tr (t + Δt) after Δt time is calculated.

[Humidity Prediction] The humidity in a room is mainly determined by the latent heat load of a person and the latent heat load due to the operation of an air conditioner, as shown in a moisture balance differential equation with respect to time (Equation 4 below). Where γ: latent heat of air evaporation ρ: air density Qbl: latent heat load of human ΣQml: latent heat load of air conditioner = (latent heat load by cooler operation) + (latent heat load by ventilator operation) + (latent heat load by humidifier operation) ... (Equation 4) The predicted humidity value H (t + Δt) after time t is determined by the humidity prediction formula shown in Equation 5 below (Equation 5). H (t) is the current humidity, and H (t + Δt) is the predicted humidity value after Δt time. (Equation 5) The latent heat load calculating means 325 calculates the latent heat load Qbl of the person. I do. The air conditioner latent heat calculation means 326 calculates a latent heat load ΣQml of the air conditioner determined for each operation plan. The humidity prediction calculation means 327 executes the calculation of Expression 4 based on the above result,
The humidity H (t + Δt) after Δt time is calculated.

The amount of power and the amount of switching power are determined by the operating state of the air conditioner and the operation plan, and are calculated by the power amount prediction calculating means 328 and the switching power amount calculating means 329 from the characteristics of each air conditioner.

In this way, the control preview simulation unit outputs the predicted values of the temperature, humidity, power amount, and switching power amount after a certain period of time when each operation plan is executed.

FIG. 19 shows the configuration of the qualitative evaluation means 33. This means comprises a temperature estimating means 331 connected to the temperature estimating means 324 and a humidity estimating means 32
7, a power amount estimating unit 333 connected to the power amount estimating / calculating unit 328, a switching power amount estimating unit 334 connected to the switching amount estimating unit 329, and a temperature estimating unit 331. , A humidity evaluating means 332, a power amount evaluating means 333, a comprehensive evaluating means 335 connected to each output side of the switching power amount evaluating means 334, a temperature evaluating means 331, a humidity evaluating means 332, and a power amount evaluating means 333.
And an evaluation function memory 336 connected to the switching power amount evaluation means 334. The prediction value output from the control preview simulation unit is used as an input, and for each item predetermined by a priori information. The input is evaluated using a membership function, and the degree of satisfaction as an overall evaluation value is calculated for each operation plan.

FIG. 20 shows an example of a qualitative evaluation membership function for each evaluation item stored in the evaluation function memory 336. The fitness is defined on the vertical axis in the range of 0.0 to 1.0, and the evaluation target item value is defined on the horizontal axis. For example, the horizontal axis defines the temperature deviation ΔT, which is the difference between the temperature measured by the sensor Ts and the temperature set value Tr with respect to the elapsed time stored in the processing schedule memory. In this example, ΔT is ± 0.3
If it is within the (° C.) range, the evaluation value is 1.0, which indicates that it is the best. Similarly, the horizontal axis defines the humidity deviation ΔH, which is the difference between the measured humidity Hs and the set humidity Ht. The values of the power amount and the switching power amount are defined as they are on the horizontal axis, and the smaller the value, the higher the evaluation.

In FIG. 19, the temperature evaluation means 331
The qualitative evaluation μt for the temperature set value is calculated by the humidity
Numeral 2 calculates and outputs a qualitative evaluation μh with respect to the set value of the humidity, the electric energy evaluating means 333 calculates a qualitative evaluation μp of the electric energy, and the switching electric energy evaluating means 334 calculates and outputs a qualitative evaluation μc of the switching electric energy.

The comprehensive evaluation means 335 calculates the qualitative evaluation value μ.
The satisfaction degree Si for the operation plan i is determined by the following equation using t, μh, μp, and μc as inputs. Si = μt × Gt + μh × Gh + μp × Gp + μc × Gc where Gt, Gh, Gp, and Gc are gains for qualitative evaluation of temperature, humidity, power, and switching power. That is, the weight of which item is emphasized.

In this way, the comprehensive evaluation means 335 calculates the satisfaction level Si and stores it in the evaluation result memory 35.

After the prediction and the qualitative evaluation for all the operation plans are completed, the control output means 34 determines the operation plan having the maximum satisfaction from the evaluation results stored in the evaluation result memory 35, and executes the process. On the other hand, it outputs ON / OFF commands for the cooler, heater, humidifier, and ventilator.

FIG. 21 illustrates the operation of the fuzzy control described above using an actual operation example. In this example, the temperature set value Tr is 13.0 (° C.), the current measured temperature Ts is 13.2 (° C.), and the humidity set value Ht is 90.0 (° C.).
(%), The measured humidity Hs is 92.0 (%), and the air conditioners are all off. Among all the operation plans, the operation plan that maintains the current state, that is, the operation plan in which all the air conditioners are off is indicated by a symbol ○, and the operation plan in which only the cooler is on is illustrated by a triangle.

In the operation plan of ○, the predicted temperature value rises to 13.4 (° C.) at time i + 1 after Δt time, and the deviation ΔT from the set temperature value is 0.4 (° C.). The humidity is 90.
Since the value becomes the same as the set value at 0 (%), the deviation ΔH is 0.0
(%). The qualitative evaluation value by each membership function is 0.68 for temperature and 1.00 for humidity.
Since all the air conditioners remain off, the evaluation of the electric energy and the switching electric energy are both 1.00. When the gain is Gt = 0.5,
Since Gh = 0.2, Gp = 0.2 and Gc = 0.2, the overall satisfaction is 0.84. Although this operation plan is slightly distant from the set value with respect to the temperature, it can be said that it is favorable when heated to the humidity, the power amount, and the switching power amount.

On the other hand, in the operation plan (1), since the cooler is turned on, the temperature drops and approaches the set value. However, the humidity has become too low on the contrary, and the evaluation has dropped. Further, since the cooler is turned on from the state where all the air conditioners are off, the evaluation of the electric energy and the switching electric energy is lower than the operation plan ○. As a result, the overall satisfaction of the operation plan I is 0.80.
After the degree of satisfaction with all the operation plans is determined in this way, if the degree of satisfaction of the operation plan ○ is maximum, the operation plan ○ is selected and all the air conditioners are kept off. Command is output.

FIGS. 22 and 23 show how the conventional problems are improved by the present embodiment. As shown in FIG. 22, the overshoot B and the undershoot B 'are well improved, and the disturbance C is restored in a short time. Further, the change in humidity, which has greatly fluctuated in the conventional control, is also stable. Also, the number of times of starting and stopping the air conditioner has been reduced by half, and the effects of reducing the power energy and prolonging the life of the device have been obtained. Further, as shown in FIG. 23, in the control result according to the present embodiment, the allowable temperature range is sufficiently used, and the temperature is maintained within the allowable range. Also, the number of heater operations has been reduced by half, and unnecessary cooler operation has not been observed.

[0074]

As described above, according to the present invention,
In addition to accurately predicting the air-conditioning load, the air-conditioning control unit considers a plurality of air-conditioner operation plans based on the prediction result, and based on the operation combination plan that is determined to be the best in the control result among the air-conditioner operation plans Control the operation of Therefore, the control can be stabilized within the allowable range and the operation of the air conditioner can be leveled, and the number of unnecessary start and stop of the air conditioner can be reduced, and the power can be further saved.

Further, since the knowledge about the operator's decision on the driving plan is taken as a priori knowledge, the mental burden on the operator can be reduced.

[Brief description of the drawings]

FIG. 1 is a schematic diagram of an air conditioner control system according to an embodiment of the present invention.

FIG. 2 is a configuration diagram of an air conditioner control unit of FIG. 1;

FIG. 3 is a diagram illustrating an example of gene expression in an air conditioner operating state.

FIG. 4 is a diagram illustrating a processing procedure of an air conditioner control unit based on genetic neuro-fuzzy control.

FIG. 5 is a diagram showing a procedure of a genetic optimization process.

FIG. 6 is a diagram showing an example of arrangement of initial individuals (driving plans).

FIG. 7 is an adaptive value distribution graph of humidification function power and cooler capacity.

FIG. 8 is an explanatory diagram of the characteristics of selection and dominant inheritance.

FIG. 9 is a diagram illustrating a process of a genetic operation.

FIG. 10 is a diagram showing an example of a neuro-optimizing processing unit.

FIG. 11 is a diagram illustrating an example of perturbation of an operation plan.

FIG. 12 is an explanatory diagram of a perturbation table of an operation plan.

FIG. 13 is an explanatory diagram of a property of an optimum value transition.

FIG. 14 is an example of an optimization constant.

FIG. 15 is a conceptual diagram of optimization by neuro-optimizing means.

FIG. 16 is a control result comparison graph.

FIG. 17 is a processing performance evaluation graph.

FIG. 18 is a configuration diagram of a control preview simulation unit.

FIG. 19 is a configuration diagram of a fuzzy evaluation unit.

FIG. 20 is a diagram showing a qualitative evaluation membership function.

FIG. 21 is a diagram illustrating an operation example of fuzzy evaluation control.

FIG. 22 is a diagram showing an operation state of air conditioning control according to the present invention.

FIG. 23 is a diagram showing the effect of the present invention in comparison with a conventional example.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Control target room, 2 ... Display device, 3 ... Genetic neuro fuzzy air-conditioning control part, 3A ... Genetic optimization processing part, 3B ...
Neuro-optimization processing unit, 3C ... preview fuzzy evaluation means,
3D: overall control processing means, 3E: process input means, 3
F: process output means, 3G: storage means, 11: cooler, 12: heater, 13: humidifier, 14: ventilator, 15
... hygrometer, 16 ... thermometer, 17 ... CO ₂ densitometer 17

────────────────────────────────────────────────── ─── Continuing on the front page (72) Kenichi Nakamura 3-2-1, Sachimachi, Hitachi, Ibaraki Prefecture Within Hitachi Engineering Co., Ltd. (72) Mayumi Mizutani 3-2-1, Sachimachi, Hitachi, Ibaraki No. Hitachi Engineering Co., Ltd. (58) Field surveyed (Int.Cl. ⁷ , DB name) F24F 11/02

Claims (5)

(57) [Claims] 1. A combination plan of a plurality of air conditioners is generated based on data of a room to be controlled and operating state data of a plurality of air conditioners in the room to be controlled. For a certain period of time, and as a result of the simulation,
In an air conditioner operation control device that controls the operation of the air conditioner according to the best air conditioner operation combination plan, an input unit that captures information from a sensor that measures process information necessary for controlling the air conditioner. Creating an objective function representing an item to be minimized or maximized in the operation control of the air conditioner, and a genetic neurofuzzy control means for determining an air conditioner operation plan for minimizing or maximizing the value of the objective function; Storage means for storing variables required to determine the operation plan; and output means for outputting the determined operation plan to an air conditioner control means, wherein the genetic neuro-fuzzy control means Values of a plurality of target items when the operation is performed for a plurality of different operation plans having chromosome expressions preset for a combination of air conditioners Foreseeing calculation, foreseeing fuzzy evaluation means for determining the evaluation value of the plurality of target item values calculated by the fuzzy membership function determined in advance, and a predetermined fuzzy evaluation means among the plurality of operation plans The two different parents are selected based on a selection constant determined in a predetermined order with respect to another driving plan of the evaluation value of the corresponding driving plan, and the properties of the two parents are mutually determined by a predetermined exchange template. Generating a new driving plan by performing a mating operation for exchange and rearranging the driving plans in a predetermined order according to the evaluation value of the generated driving plan when the generated driving plan reaches a predetermined number. And an operation in which one operation plan having the maximum or minimum evaluation value determined by the genetic optimization means is set as an initial operation plan, and a part of the operation plan can be changed by a predetermined constant. In the case of the evaluation value is calculated, and compared with the evaluation value of the previous operation plan, smaller than the value of the variable difference value of the evaluation values is preset,
A neuro-optimizing unit that replaces a previous operation plan with an operation plan generated this time and sets the operation plan as an optimal operation candidate. 2. The air conditioner control device according to claim 1, wherein the preview fuzzy evaluation means determines in advance the air conditioner operation combination plan with the best evaluation based on a priori knowledge about the operation of the air conditioner. A driving combination plan is predicted and calculated by the determined membership function, the predicted calculation amount is evaluated as a fuzzy amount, and the evaluation result is loaded by a load value determined in advance based on a priori knowledge. An air conditioner operation control device, wherein an average is determined, and an air conditioner operation combination plan that maximizes the load average is adopted as the plan with the best evaluation. 3. The air conditioner control device according to claim 1, wherein said air conditioner is capable of adjusting its capacity in a range of 0 to 100%.
An air conditioner operation control device comprising a cooler, a heater, a humidifier, and a ventilator. 4. A plurality of air conditioner combination plans are generated based on the data of the controlled room and the operating state data of the air conditioners in the controlled room, and one of the air conditioner operation combinations is generated.
A method for controlling the operation of the air conditioner by performing a simulation for each of the air conditioners until a predetermined time has elapsed, and as a result of the simulation, as a result of the simulation, is performed. It takes in information from a sensor that measures the process information necessary to control the operation of the air conditioner, and creates an objective function representing an item to be minimized or maximized in the operation control of the air conditioner. In order to determine the operation plan of the air conditioner that minimizes or maximizes the value of the objective function, a plurality of operation plans having a plurality of different operation plans having a chromosome expression set in advance with respect to the combination of the air conditioners are performed. Predictive calculation of the value of the target item of the above, and the plurality calculated by the fuzzy membership function predetermined in advance. An evaluation value of the target item value is determined, and two types of different parents are determined from a plurality of predetermined driving plans based on a selection constant determined in advance in a predetermined order with respect to the other driving plans of the evaluation value of the corresponding driving plan. A new operation plan is generated by performing a mating operation of exchanging the properties of the two types of parents with each other according to a predetermined exchange template, and when the generated operation plan reaches a predetermined number. Performing a genetic optimization operation of rearranging the operation plans in a predetermined order in accordance with the evaluation value of the generated operation plan, and selecting one operation plan whose evaluation value determined by the genetic optimization operation is the maximum or the minimum. As an initial operation plan, an evaluation value of the operation plan that can be changed by a predetermined constant with a part of the operation plan is calculated, compared with the evaluation value of the previous operation plan, and a difference value of the evaluation value is set in advance. Is smaller than the value of the variable
An operation control method for an air conditioner, comprising: replacing a previous operation plan with an operation plan generated this time; outputting the determined operation plan to an air conditioner control unit; 5. The air conditioner operation control method according to claim 4, wherein said preview fuzzy evaluation means uses a priori knowledge on the operation of said air conditioner when said air conditioner operation combination plan with the best evaluation is determined. A driving combination plan is predicted and calculated by a membership function determined in advance, the predicted calculation amount is evaluated as a fuzzy amount, and the evaluation result is calculated based on a load value previously determined based on a priori knowledge. An air conditioner operation control method, wherein a load average is determined, and an air conditioner operation combination plan that maximizes the load average is adopted as the plan with the best evaluation.