JPH06266407A - Parameter adjusting device of combustion simulation system - Google Patents

Abstract

PURPOSE:To provide the device for adjusting efficiently a parameter in the combustion simulation system in a short time. CONSTITUTION:The device is provided with a recording part 1 for recording an individual for showing a parameter corresponding to plural digits in a binary number, a simulation executing part 2 for executing a simulation, on the basis of the individual, an adaptive degree calculating part 3 for calculating an adaptive degree of each individual according to a result of simulation, a selecting part 4 for selecting the recorded individual, an intersection part 5 for recombining series of two selected individuals, a sudden varying part 6 for varying a part of the series of the individuals, and an evaluating aprt 7 for finishing a calculation processing, when the adaptive degree goes into a certain reference value.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a parameter adjusting device for a combustion simulation system for adjusting parameters of a system for performing a combustion simulation of an electric power plant or the like.

[0002]

2. Description of the Related Art For example, in a coal-fired power plant, the number of coal feeders that supply coal to a boiler is changed in accordance with changes in power demand to control the amount of coal supply and adjust the power generation output. Is being carried out.

At the time of adjusting the power generation output as described above, NOx, which is a nitrogen oxide, may become high, which may cause a problem. Therefore, the NOx dynamic characteristic simulation of the plant is performed using the NOx analysis model, measures are taken to reduce the generation of NOx, and the result is reflected in the control circuit.
It displays driving assistance guidance.

Conventionally, the adjustment of the parameters of this simulation system has been performed by a human being by trial and error while judging the result.

[0005]

As described above, the parameters of the conventional combustion simulation system have been adjusted by manually changing the parameter values. There are many parameters in the simulation system, and their combination is enormous. Therefore, it takes a long time to adjust the parameters.

An object of the present invention is to provide a combustion simulation system parameter adjusting device capable of adjusting parameters efficiently in a short time.

[0007]

Means adopted by the present invention for solving the above-mentioned problems will be described. First, the principle will be described before describing the means adopted by the present invention. A genetic algorithm is used to adjust the parameters of the present invention. That is, it applies the principle of generation selection by selection and mutation in biological evolution.

Generational updating in the evolution of living organisms is carried out by mating and mutation of parental chromosomes, and those compatible with the natural world are selected and selected to form the next generation. What kind of genetic information is described at each position of the chromosome (gene) is determined, and this gene is made to correspond to each parameter for adjusting the present invention, and the numerical value of the parameter is represented by 0 and 1. It is represented by a binary number.

As described above, selective mating is performed on individuals represented by binary numbers, individuals with high fitness leave more progeny, and genes that form good individuals are spread throughout the population. In addition, a sudden change that changes the value of the chromosome with a certain probability is added. When such an operation is sequentially performed and an individual with high fitness is obtained, each parameter value represented by the individual is output as the optimum parameter of the combustion simulation system.

Next, the means adopted by the present invention will be described.
A parameter adjustment device for a combustion simulation system, comprising: a recording unit that records each numerical value of each of the parameters as an individual represented by a plurality of digits in a binary number; and an individual read out from the recording unit. A simulation executing unit that executes a simulation, a fitness calculating unit that calculates the fitness of each individual recorded in the recording unit based on the result executed by the simulation executing unit, and the fitness calculated adaptation A selection unit for selecting an individual recorded in the recording unit based on frequency, a crossover unit for recombining the sequence of two individuals selected by the selection unit, and a sequence of the individual numbers selected by the selection unit. A mutation part for changing the part and an evaluation part for ending the calculation process when the fitness calculated as the fitness falls within a certain standard value.

[0011]

In the recording unit, an individual is represented in which each numerical value of the parameters of the simulation system is represented by a plurality of digits in the binary number. In the simulation execution part,
The simulation is executed based on the individual read out from the recording unit.

The fitness calculating section calculates the fitness of each individual recorded in the recording section based on the result executed by the simulation executing section. The selection unit selects the individual recorded in the recording unit based on the fitness. In the crossover section, a sequence of two individuals selected in the selection section is recombined to create a new individual and recorded in the recording section.

In the mutation part, a part of the sequence of the individuals selected by the selection part is changed to create a new individual and the new part is recorded in the recording part. The evaluation unit ends the process when the fitness calculated by the fitness calculation unit falls within a certain standard value.

As described above, each parameter is represented by an individual corresponding to a plurality of digits in a binary number, a recombination of the sequence of two individuals and a part of the numerical value of the individual are changed to create a new individual. , The population with more highly adaptable individuals is formed, so the optimal solution is efficiently derived, and if the convergence of the solution is poor, the solution area is expanded to achieve the optimal solution in a short time. Can be obtained.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described with reference to FIGS. 1 is a block diagram of an embodiment of the present invention, FIG. 2 is an operation flowchart of the embodiment, FIG. 3 is a specific example of a simulation system, FIG. 4 is a specific example of an individual, and FIG. 5 is a specific example of crossover.

In FIG. 1, reference numeral 1 denotes a recording unit, which records individuals and fitness levels, as will be described later in detail. Two
Is a simulation execution unit, which executes the simulation according to the NOx analysis model. Reference numeral 3 denotes a fitness calculation unit that calculates the fitness. Reference numeral 4 denotes a selection unit, which selects an individual recorded in the recording unit 1. Reference numeral 5 denotes a crossover portion, which recombines the sequence of individuals selected by the selection portion 4 to create a new individual. 6 is a mutation part, which changes a part of the sequence of individuals to make a new individual. An evaluation unit 7 evaluates whether or not the fitness falls within a predetermined standard value. 8 is an input unit, 9 is a processor (CP
U), and executes all processing.

The combustion simulation system simulates the combustion system by expressing it with a mathematical model. FIG. 3 is a concrete example of the simulation system represented by a mathematical model. Four coal feeders of A, B, C and D stages are installed in the boiler, and all coal feeders are used during normal operation. However, due to changes in power demand, 3
The figure shows a case where the NOx generation amount, the O ₂ generation amount, the A-stage coal feed amount, and the D-stage coal feed amount are mathematically modeled when the power generation output is adjusted by switching the number of units to two or two stages.

When the mathematical model is used in this way, the parameters included in the mathematical model are as follows: (1) TIME2: dead time from setting of D-stage coal supply amount (2) T _GCALT : A to C-stage coal supply delay time ( 3) T _GCALD : D-stage feed delay time (4) T _NOx : NOx analyzer delay time (5) T _Q2 : O ₂ analysis delay time (6) T _NOxD : NOx dead time (7) T _O2D : O ₂ There are seven dead times, and these parameters are determined.

Therefore, the individual is represented by a binary number, and the above seven parameters are associated with specific plural digits of the individual represented by the binary number. That is, in the case of the specific example shown in FIG. 4, an individual is represented by a binary number of 6 digits, where the lower 2 digits are the parameter TIME2 and the middle 2 digits are the parameter T _GCALT.
The upper two digits indicate the case corresponding to the parameter T _GCALD .

If the last two digits are 00, then TIME
When the value of 2 is 0.0,01, it is 0.4, and when it is 10,
In the case of 0.8 and 11, it corresponds to 1.2.
Data T _GCALTAnd T_GCALDSimilarly, for each para
Corresponds to the meter value. Next, referring to FIG.
The operation will be described.

Process S1 In process S1, an initial population of individuals is input from the input unit 8 and recorded in the recording unit 1. Process S2 In process S2, the simulation executing unit 2 executes process S1.
Each individual of the initial population input in (3) is read out and the simulation is executed according to the set mathematical model.

Process S3 In process S3, the fitness calculating section 3 calculates the fitness for the simulation result executed in process S2. That is, when the fitness f _i for the individual i to the error between the simulation result executed in step S2 and the test data by actual equipment and _{E i (t), f i} = {(t = t 1, t 2) E i _(t) } (1) where {(i = 1,3) x _i } represents x ₁ + x ₂ + x ₃ t ₁ : start time of comparison section t ₂ : end time of comparison section It is calculated as the reciprocal of the sum of the represented errors.

The fitness f _i of the individual i thus calculated is recorded in a memory (not shown) of the recording unit 1 in association with the individual i. Processing S4 In processing S4, the evaluation unit 7 ends the processing when the fitness calculated in the processing S3 is equal to or higher than a certain standard value, and moves to the processing S5 when the fitness is within the standard value.

Process S5 In process S5, selection is made to create a new individual.
Individuals are selected based on the principle that offspring with good fitness are generated from parents with good fitness, and more individuals with good fitness are selected in proportion to fitness. That is, when the Pselecti the probability of being selected for an individual _{i, Pselecti = f i / Σf} i ... (2) comprising selection probability is given, when the selection probabilities given 0
A random number is generated in the interval of 1 to determine which individual and which individual to select.

Process S6 In process S6, the crossover unit 5 performs crossover of the two individuals selected in process S5, that is, recombination of several sequences. A specific example of the crossover will be described with reference to FIG. In FIG. 5, the parent-
1 and parent-2 are the two individuals selected in process S5. On the other hand, a mask is prepared as shown in FIG.
When the bit of the mask is 0, the value of the parent-1 is transferred to the child-1, and when the bit of the mask is 1, the parent-2 is transferred to the child-1.
Transfer the value of. For child-2, do the opposite of child-1.

In this way, new individuals of child-1 and child-2 are newly created and recorded in the recording unit 1. When the recording is completed, the process proceeds to step S2 to execute the simulation,
In step S, the fitness is calculated, and in step S6, the evaluation is repeated.

If the fitness does not fall within a certain standard value even after the number of repetitions reaches a certain number, the mutation in the step S7 is executed instead of the crossover in the step S6. Process S7 In process S7, the mutation unit 6 changes a part of the number sequence of the individuals read from the recording unit 1 and records it in the recording unit 1. When mutation is performed, the selection of the individual by the selection unit 4 is excluded from the selection target so that the elite individual is preserved. That is, excluding individuals with good fitness,
Selected from other individuals.

When the mutation process is performed and recorded in the recording unit 1, the process proceeds to the process S2, the processes S2 to S7 are repeated, and the process is performed when the evaluation in the process S4 falls within the specified value. finish. Although one embodiment of the present invention has been described above, the present invention is not limited to this embodiment, and various modifications can be made according to the gist of the invention.

[0029]

As described above, according to the present invention, the following effects can be obtained. Each parameter is represented by an individual corresponding to a plurality of digits in a binary number, a recombination of a sequence of two individuals and a part of the numerical value of the individual are changed to create a new individual, and an individual with high fitness is more Since many groups are formed, the optimal solution can be efficiently derived, and when the solution convergence is poor, the optimal solution can be efficiently obtained in a short time by expanding the solution region.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an embodiment of the present invention.

FIG. 2 is an operation flowchart of the same embodiment.

FIG. 3 is a specific example of a simulation system.

FIG. 4 is a specific example of an individual.

FIG. 5 is a specific example of crossover.

[Explanation of symbols]

 1 recording unit 2 simulation execution unit 3 fitness calculation unit 4 selection unit 5 crossover unit 6 mutation unit 7 evaluation unit 8 input unit 9 processor (CPU)

Claims (2)

[Claims] 1. A parameter adjusting device for a combustion simulation system, comprising: a recording unit for recording an individual body in which each numerical value of the parameter is represented by a plurality of digits in a binary number; and a reading unit read from the recording unit. A simulation executing unit that executes a simulation based on the individual; an fitness calculating unit that calculates the fitness of each individual recorded in the recording unit based on the result executed by the simulation executing unit; A selection unit for selecting an individual recorded in the recording unit based on the calculated fitness, a crossover unit for recombining a sequence of two individuals selected by the selection unit, and a selection unit selected by the selection unit. A mutation part that changes a part of the sequence of individuals, and an evaluation that terminates the calculation process when the fitness calculated the fitness falls within a certain standard value. Parameter adjustment device of a combustion simulation system characterized by comprising: a part, the. 2. The parameter adjusting device for a combustion simulation system according to claim 1, wherein the change in the sequence of numbers at the mutation portion is excluded for the individuals having high fitness.