CN110033118A - Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm - Google Patents
Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm Download PDFInfo
- Publication number
- CN110033118A CN110033118A CN201910137515.9A CN201910137515A CN110033118A CN 110033118 A CN110033118 A CN 110033118A CN 201910137515 A CN201910137515 A CN 201910137515A CN 110033118 A CN110033118 A CN 110033118A
- Authority
- CN
- China
- Prior art keywords
- value
- genetic algorithm
- data
- variable
- blower
- 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.)
- Pending
Links
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
A kind of elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm, the following steps are included: step 1: obtaining performance variable, the a certain combination in fan operation efficiency and wind pressure or efficiency and air quantity both combinations is chosen, and enabling is target variable；Step 2: enabling performance variable is input variable, target variable is output variable, is trained to data sample, completes the foundation of elastomeric network model；Step 3: Genetic Algorithm Model is established；Step 4: being used for the predicted values of the performance parameters such as rated wind pressure, efficiency and air quantity for the elastomeric network model that second step is established, and predicted value is used for seeking for target function value in Genetic Algorithm Model, to obtain multiobjective optimization solution；The operating parameter that finally optimal solution renormalization is optimized, and the value is transferred to fan operation mechanism and carries out practical control.Precision of the present invention is higher, effect is preferable, time-consuming short.
Description
Technical field
The invention belongs to the analog simulation fields of the control technology of fan operation process and industrial process, are related to one kind and are based on
Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm.
Background technique
Blower is the mechanical energy by input, the fluid machinery that admission pressure is improved and is discharged.Mainly by impeller, leaf
Piece, motor, driving member composition.
Blower gives air and gas system in such as power plant to provide power, its operating condition is efficiently run with power plant and energy conservation has
Huge connection.If can effectively control blower, its high efficiency range is expanded to improve its efficiency, reduce machine
The consumption of tool energy, this will improve the working efficiency of entire industrial process.Energy will all be reduced to daily life region such as subway station etc.
Source consumption.The change of each control parameter of blower is a kind of comprehensive effect to fan performance.I.e. when each operating parameter changes, blower
The variation tendency of each target component is not consistent.It so we are not only optimization efficiency, while being also to the excellent of wind pressure and air quantity
Change.Because air quantity is bigger, the mobility of air molecule is better.Wind pressure is bigger, and air molecule local density is bigger, air molecule energy
Flow to farther place.But when increasing with flow velocity, resistance also increases, and increases the loss of part and flowing, efficiency reduces.Institute
To need through Multipurpose Optimal Method, in the case where meeting real work and must ask, one group of optimal control parameter combination is obtained.
The operational process of blower is the process fluid flow and complicated energy transfer process of a disorder.Traditional more mesh
Mark calculates the weighted calculation really calculated single goal, and the value and staff's experience of weight have very big association, it is difficult to real
Now most accurate, efficient control.If by CFD approach, that is, Fluid Mechanics Computation, using electronic computer as tool, using it is various from
The mathematical method of dispersion, obtains the approximate solution of governing equation, and process is not only complicated but also takes a long time, when controlling at the scene I
Need when providing result in seconds, and be not suitable for.And current intelligent optimization algorithm includes genetic algorithm, population
Algorithm etc. possesses quick global optimizing ability, is widely used in solution multi-objective optimization question.
Summary of the invention
In order to overcome the precision of existing blower control method is lower, effect is poor, take a long time the deficiencies of place, the present invention
There is provided a kind of precision is higher, effect preferably, it is time-consuming it is short with elastomeric network modeling and the blower multiple target based on genetic algorithm it is excellent
Change control method.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm, comprising the following steps:
Step 1: obtaining influences very big performance variable to fan operation efficiency and wind pressure or efficiency and air quantity, and gives
Determine wind pressure and air quantity variation range；It requires to choose fan operation efficiency and wind pressure or efficiency and air quantity further according to blower actual motion
A certain combination in both combinations, and enabling is target variable；Wherein data sample composed by performance variable and target variable
Originally it can be obtained by experiment.
Step 2: enabling performance variable is input variable, target variable is output variable, is trained to data sample, complete
At the foundation of elastomeric network model, wherein carrying out the update of weight and threshold value using elasticity BP method；
Step 3: establishing Genetic Algorithm Model, wherein using non-dominated ranking operator, crowding comparison operator and elite plan
Slightly design operator；
Step 4: the elastomeric network model that second step is established is used for the performance parameters such as rated wind pressure, efficiency and air quantity
Predicted value, and predicted value is used for seeking for target function value in Genetic Algorithm Model, to obtain multiobjective optimization solution；Finally will
The operating parameter that optimal solution renormalization is optimized, and the value is transferred to fan operation mechanism and carries out practical control.
Further, in the step 1, the variable is chosen as follows: movable vane established angle, revolving speed being selected to become as operation
Amount, and the input variable enabled as neural network model；A kind of combination of efficiency of selection and wind pressure or efficiency and air quantity is as target
Variable, and the output variable enabled as neural network model, wherein combined selection is determined by the actual requirement of fan operation.
The step 2 is as follows to the foundation, initialization, training process of elastic neural network model: first to sample data
It is handled；Then, calculate neural network hidden layer node by treated data and output layer node respectively correspond it is defeated
Enter value and output valve；Finally, updating the weight and threshold of neural network according to the more new formula of weight in elastomeric network and threshold value
Value；By recycling several times, if error is in desired error range or arrived maximum cycle, training stops.
The processing step of the step 2 is as follows:
2.1 data processing
Relevant parameter, that is, movable vane established angle, revolving speed, efficiency, air quantity or wind pressure in collection step one, and to movable vane established angle,
Revolving speed and air quantity or wind pressure are normalized, and make it between [0,1], formula is as follows:
Wherein, k is the data after normalization, and x is to be normalized data, and xmin is the minimum value being normalized in data,
Xmax is to be normalized maximum value in data；
The classification of 2.2 data
The data set obtained after processing is divided into two parts, wherein it randomly selects in data set 70% and is used as training set,
Again using remaining 30% data as test set；
The initialization of 2.3 networks
Assuming that the node number of input layer is n, the node number of hidden layer is l, and the node number of output layer is m, is enabled hidden
It is one layer containing number layer by layer, number of nodes rule of thumb formula:Wherein a is that constant takes 1~10, and cycle-index is set
N times are set to, the weight of input layer to hidden layer is wij, the weight of hidden layer to output layer is wjk, the threshold of input layer to hidden layer
Value is aj, the threshold value of hidden layer to output layer is bk, each layer weight initial value take the random number between [- 1,1], and learning rate is
η, value are 0.1~0.2, and excitation function is g (x), and wherein excitation function takes Sigmoid function.Form are as follows:
2.4 training neural networks；
2.5 data test
After the data training for completing all training sets, neural network is tested with the data of test set.If error exists
In prescribed limit, then elastic neural network completes modeling, i.e., outputting and inputting for neural network model meets mapping relations.
In described 2.4, the process of training neural network is as follows:
2.4.1 the forward-propagating of signal
2.4.1.1 the output of hidden layer:
xiFor the data of the input of input layer, HjFor the output of hidden layer node；
2.4.1.2 the output of output layer:
OkFor the output of output layer node.
2.4.1.3 the calculating of error:
ek=Ok-Yk
In above formula, i=1...n, j=1...l, k=1...m, YkFor reality output data；
By comparing the reality output and desired output of output layer, the error of the two is obtained, if error is not in requirement
In error range, then error back propagation is transferred to.
2.4.2 the backpropagation of signal (error)
The weight of resilient BP algorithm and the following formula of the update of threshold value:
W (t+1)=w (t)+Δ w (t)
Wherein t is the number of iterations, and Δ w (t) obtains correction amount for weight for t, and gradient isOnly to the direction of weighed value adjusting
It works.If the direction adjusted twice in succession is identical, it is increased by Δ t, general α is selected as 1.2 at this time.If the side adjusted twice in succession
To on the contrary, being reduced by Δ t, general β is selected as 0.8 at this time.In the case of other, without updating；
All steps in step 2.4.1-2.4.2 are repeated, if error is in desired error range or cycle-index is super
Until crossing the setting upper limit, then training stops.
In the step 3, the step of genetic algorithm, is as follows:
3.1 generating parent population Pt, Population Size N；
3.2 calculate non-dominant rank individual in parent population Pt, crowding distance；
3.3 are selected again, are intersected, being made a variation, and subgroup Qt primary is generated；
3.4 merge subgroup Qt primary and parent population Pt, generate new parent population Pt+1, Population Size 2N；
3.5 calculate non-dominant rank individual in new parent population Pt+1, crowding distance；
3.6 couples of new parent population Pt+1 are selected, are intersected, are made a variation, and new parent subgroup Qt+1, Population Size N are generated；
3.7 judge whether evolutionary generation is less than maximum algebra G, if it is not, then exporting Qt+1；If so, t=t+1, returns to
Four steps continue to recycle.
Further, in the first step, population coding mode uses real coding, and population scale is set as N, evolutionary generation is set
It is set as 20%~50% for t, the general fork probability of intersection, mutation probability is set as 1%~10%.
Further, in the second step and the 5th step, using quick non-dominated ranking operator and crowding comparison operator,
It must be calculated the predicted value of neural network model as target function value in genetic algorithm.
Wherein, the principle of quick non-dominated ranking are as follows: find out non-dominant disaggregation in population first, be denoted as the first non-dominant layer
F1, non-dominant ordinal number is irank=1, and F1 is removed to the non-dominant disaggregation for finding out remaining population again, is denoted as F2.Successively
So.The non-dominant higher individual of ordinal number is preferentially selected.
Crowding: 2 adjacent with i individuals the distance between i+1 and i-1 on object space.In same non-dominant layer F
(i) it in, wins standard using the two attributes as individual, a physical efficiency in the quasi- domain Paroet is made to expand to entire Pareto
Domain is uniformly distributed, and maintains the diversity of population.
Further, in the third step and the 6th step, selection refers to through quick non-dominated ranking operator, crowding ratio
It is selected compared with operator；Intersect the combined crosswise referred to through chromosome, generates new individual；Variation refers to from group optionally
An individual makes a variation to generate more excellent individual to certain section of coding in selected chromosome, using real number interior extrapolation method
Carry out crossover operation；Using j-th of gene a of i-th of individualijMutation operation is carried out, variation method is as follows:
aij=aij+(aij-amax) * f (g) r > 0.5
aij=aij+(amin-aij)*f(g)r≤0.5
In formula, random number of the r between [0,1], amaxFor the previous of gene, aminFor the next time of gene；F (g)=rand ×
(1-g/Gmax)2；Random number of the rand between [0,1]；G is current iteration number；GmaxFor maximum evolution number.
Again further, in the 4th step, operator, i.e., the son generated parent population with it are designed using elitism strategy
It combines for population, is selected by quick non-dominated ranking operator and crowding comparison operator, generate next-generation population；This has
Enter the next generation conducive to the defect individual kept in parent, guarantees that the optimum individual in population will not be lost.
In the step 4, process is as follows: first passing through the training to data sample, completes building for elastic neural network model
Vertical, which has gone out the good mapping relations for outputting and inputting variable；Then the predicted value of neural network is used to lose
Target function value seeks in propagation algorithm；Global optimizing is carried out finally by genetic algorithm, finds out fan operation efficiency and wind pressure
Or most ideal point, that is, pareto optimum point of efficiency and air quantity；Since the end value that the model obtains is the value after normalization,
Therefore need to carry out anti-normalization processing to performance variable value corresponding to pareto optimum point again, it is converted into true value, formula
It is as follows: x=k (xmax-xmin)+xmin。
Technical concept of the invention are as follows: not only need to choose suitable algorithm, it is also necessary to have the prediction mould to target variable
Type.Artificial neural network is abstracted to biological neural network and is established naive model, with the non-of height by bionics
Linear Mapping ability, self-learning capability etc. have been applied to many fields.Wherein, elastomeric network eliminates in BP neural network to accidentally
The complicated calculations of poor gradient amplitude make convergence rate faster.Genetic algorithm and neural network are combined, calculation can be effectively improved
The accuracy of method convergence rate and parameter, is quickly obtained optimal control variable value, is efficiently controlled blower.
Beneficial effects of the present invention are mainly manifested in: precision is higher, effect is preferable, time-consuming is short.
Detailed description of the invention
Fig. 1 is the process of elastomeric network modeling of the invention and the blower multiobjective optimization control method based on genetic algorithm
Figure.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.
Referring to Fig.1, a kind of elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm, including with
Lower step:
Step 1: obtaining influences very big performance variable to fan operation efficiency and wind pressure or efficiency and air quantity, and gives
Determine wind pressure and air quantity variation range；It requires to choose fan operation efficiency and wind pressure or efficiency and air quantity further according to blower actual motion
A certain combination in both combinations, and enabling is target variable；Wherein data sample composed by performance variable and target variable
Originally it can be obtained by experiment.
Step 2: enabling performance variable is input variable, target variable is output variable, is trained to data sample, complete
At the foundation of elastomeric network model, wherein carrying out the update of weight and threshold value using elasticity BP method；
Step 3: establishing Genetic Algorithm Model, wherein using non-dominated ranking operator, crowding comparison operator and elite plan
Slightly design operator；
Step 4: the elastomeric network model that second step is established is used for the performance parameters such as rated wind pressure, efficiency and air quantity
Predicted value, and predicted value is used for seeking for target function value in Genetic Algorithm Model, to obtain multiobjective optimization solution；Finally will
The operating parameter that optimal solution renormalization is optimized, and the value is transferred to fan operation mechanism and carries out practical control.
Further, in the step 1, performance variable: movable vane established angle, revolving speed, guide vane twist angle etc., the variable
It chooses as follows: selecting movable vane established angle, revolving speed as performance variable, and the input variable enabled as neural network model；Selection effect
A kind of combination of rate and wind pressure or efficiency and air quantity is as target variable, and the output variable enabled as neural network model, wherein
Combined selection is determined by the actual requirement of fan operation.
The step 2 is as follows to the foundation, initialization, training process of elastic neural network model: first to sample data
It is handled；Then, calculate neural network hidden layer node by treated data and output layer node respectively correspond it is defeated
Enter value and output valve；Finally, updating the weight and threshold of neural network according to the more new formula of weight in elastomeric network and threshold value
Value；By recycling several times, if error is in desired error range or arrived maximum cycle, training stops.
The processing step of the step 2 is as follows:
2.1 data processing
Relevant parameter, that is, movable vane established angle, revolving speed, efficiency, air quantity or wind pressure in collection step one, and to movable vane established angle,
Revolving speed and air quantity or wind pressure are normalized, and make it between [0,1], formula is as follows:
Wherein, k is the data after normalization, and x is to be normalized data, and xmin is the minimum value being normalized in data,
Xmax is to be normalized maximum value in data；
The classification of 2.2 data
The data set obtained after processing is divided into two parts, wherein it randomly selects in data set 70% and is used as training set,
Again using remaining 30% data as test set；
The initialization of 2.3 networks
Assuming that the node number of input layer is n, the node number of hidden layer is l, and the node number of output layer is m, is enabled hidden
It is one layer containing number layer by layer, number of nodes rule of thumb formula:Wherein a is that constant takes 1~10, and cycle-index is set
N times are set to, the weight of input layer to hidden layer is wij, the weight of hidden layer to output layer is wjk, the threshold of input layer to hidden layer
Value is aj, the threshold value of hidden layer to output layer is bk, each layer weight initial value take the random number between [- 1,1], and learning rate is
η, value are 0.1~0.2, and excitation function is g (x), and wherein excitation function takes Sigmoid function.Form are as follows:
2.4 training neural networks, process are as follows:
2.4.1 the forward-propagating of signal
2.4.1.1 the output of hidden layer:
xiFor the data of the input of input layer, HjFor the output of hidden layer node；
2.4.1.2 the output of output layer:
OkFor the output of output layer node.
2.4.1.3 the calculating of error:
ek=Ok-Yk
In above formula, i=1...n, j=1...l, k=1...m, YkFor reality output data；
By comparing the reality output and desired output of output layer, the error of the two is obtained, if error is not in requirement
In error range, then error back propagation is transferred to.
2.4.2 the backpropagation of signal (error)
The weight of resilient BP algorithm and the following formula of the update of threshold value:
W (t+1)=w (t)+Δ w (t)
Wherein t is the number of iterations, and Δ w (t) obtains correction amount for weight for t, and gradient isOnly to the direction of weighed value adjusting
It works.If the direction adjusted twice in succession is identical, it is increased by Δ t, general α is selected as 1.2 at this time.If the side adjusted twice in succession
To on the contrary, being reduced by Δ t, general β is selected as 0.8 at this time.In the case of other, without updating；
All steps in step 2.4.1-2.4.2 are repeated, if error is in desired error range or cycle-index is super
Until crossing the setting upper limit, then training stops；
2.5 data test
After the data training for completing all training sets, neural network is tested with the data of test set.If error exists
In prescribed limit, then elastic BP neural network completes modeling, i.e., outputting and inputting for neural network model meets mapping relations.
In the step 3, the step of genetic algorithm, is as follows:
3.1 generate parent population Pt, Population Size N；
3.2 calculate non-dominant rank individual in parent population Pt, crowding distance；
3.3 are selected again, are intersected, being made a variation, and subgroup Qt primary is generated；
3.4 merge subgroup Qt primary and parent population Pt, generate new parent population Pt+1, Population Size 2N；
3.5 calculate non-dominant rank individual in new parent population Pt+1, crowding distance；
3.6 couples of new parent population Pt+1 are selected, are intersected, are made a variation, and new parent subgroup Qt+1, Population Size N are generated；
3.7 judge whether evolutionary generation is less than maximum algebra G, if it is not, then exporting Qt+1；If so, t=t+1, returns to
Four steps continue to recycle.
Further, in the first step, population coding mode uses real coding, and population scale is set as N, evolutionary generation is set
It is set as 20%~50% for t, the general fork probability of intersection, mutation probability is set as 1%~10%.
Further, in the second step and the 5th step, using quick non-dominated ranking operator and crowding comparison operator,
It must be calculated the predicted value of neural network model as target function value in genetic algorithm.
Wherein, the principle of quick non-dominated ranking are as follows: find out non-dominant disaggregation in population first, be denoted as the first non-dominant layer
F1, non-dominant ordinal number is irank=1, and F1 is removed to the non-dominant disaggregation for finding out remaining population again, is denoted as F2.Successively
So.The non-dominant higher individual of ordinal number is preferentially selected.
Crowding: 2 adjacent with i individuals the distance between i+1 and i-1 on object space.In same non-dominant layer F
(i) it in, wins standard using the two attributes as individual, a physical efficiency in the quasi- domain Paroet is made to expand to entire Pareto
Domain is uniformly distributed, and maintains the diversity of population.
Further, in the third step and the 6th step, selection refers to through quick non-dominated ranking operator, crowding ratio
It is selected compared with operator；Intersect the combined crosswise referred to through chromosome, generates new individual；Variation refers to from group optionally
An individual makes a variation to generate more excellent individual to certain section of coding in selected chromosome, using real number interior extrapolation method
Carry out crossover operation；Using j-th of gene a of i-th of individualijMutation operation is carried out, variation method is as follows:
aij=aij+(aij-amax) * f (g) r > 0.5
aij=aij+(amin-aij)*f(g) r≤0.5
In formula, random number of the r between [0,1], amaxFor the previous of gene, aminFor the next time of gene；F (g)=rand ×
(1-g/Gmax)2；Random number of the rand between [0,1]；G is current iteration number；GmaxFor maximum evolution number.
Again further, in the 4th step, operator, i.e., the son generated parent population with it are designed using elitism strategy
It combines for population, is selected by quick non-dominated ranking operator and crowding comparison operator, generate next-generation population；This has
Enter the next generation conducive to the defect individual kept in parent, guarantees that the optimum individual in population will not be lost.
In the step 4, process is as follows: first passing through the training to data sample, completes building for elastic neural network model
Vertical, which has gone out the good mapping relations for outputting and inputting variable；Then the predicted value of neural network is used to lose
Target function value seeks in propagation algorithm；Global optimizing is carried out finally by genetic algorithm, finds out fan operation efficiency and wind pressure
Or most ideal point, that is, pareto optimum point of efficiency and air quantity；Since the end value that the model obtains is the value after normalization,
Therefore need to carry out anti-normalization processing to performance variable value corresponding to pareto optimum point again, it is converted into true value, formula
It is as follows: x=k (xmax-xmin)+xmin。
Claims (10)
1. a kind of elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm, which is characterized in that described
Method the following steps are included:
Step 1: obtaining influences very big performance variable, and given wind to fan operation efficiency and wind pressure or efficiency and air quantity
Pressure and air quantity variation range；Further according to blower actual motion require to choose fan operation efficiency and wind pressure or efficiency and air quantity this two
A certain combination in kind combination, and enable as target variable；Wherein data sample composed by performance variable and target variable can
It is obtained by experiment.
Step 2: enabling performance variable is input variable, target variable is output variable, is trained to data sample, completes bullet
Property network model foundation, wherein using elasticity BP method progress weight and threshold value update；
Step 3: establishing Genetic Algorithm Model, wherein being set using non-dominated ranking operator, crowding comparison operator and elitism strategy
Calculate son；
Step 4: the elastomeric network model that second step is established is used for the predicted value of rated wind pressure, efficiency and air quantity, and will prediction
Value is sought for target function value in Genetic Algorithm Model, to obtain multiobjective optimization solution；Finally by optimal solution renormalization
The operating parameter optimized, and the value is transferred to fan operation mechanism and carries out practical control.
2. elastomeric network modeling as described in claim 1 and the blower multiobjective optimization control method based on genetic algorithm,
It is characterized in that, in the step 1, the variable is chosen as follows: selecting movable vane established angle, revolving speed as performance variable, and enables
For the input variable of neural network model；A kind of combination of efficiency of selection and wind pressure or efficiency and air quantity as target variable, and
Enable the output variable for neural network model, wherein combined selection is determined by the actual requirement of fan operation.
3. elastomeric network modeling as claimed in claim 1 or 2 and the blower multiobjective optimization control method based on genetic algorithm
Method, which is characterized in that as follows to the foundation, initialization, training process of elastic neural network model in the step 2: first right
Sample data is handled；Then, neural network hidden layer node and output layer node point are calculated by treated data
Input value and output valve are not corresponded to；Finally, updating the power of neural network according to the more new formula of weight in elastomeric network and threshold value
Value and threshold value；By recycling several times, if error is in desired error range or arrived maximum cycle, training stops
Only.
4. elastomeric network modeling as claimed in claim 3 and the blower multiobjective optimization control method based on genetic algorithm,
It is characterized in that, the processing step of the step 2 is as follows:
2.1 data processing
Relevant parameter, that is, movable vane established angle, revolving speed, efficiency, air quantity or wind pressure in collection step one, and to movable vane established angle, revolving speed
It is normalized with air quantity or wind pressure, makes it between [0,1], formula is as follows:
Wherein, k is the data after normalization, and x is to be normalized data, and xmin is the minimum value being normalized in data, xmax
To be normalized maximum value in data；
The classification of 2.2 data
The data set obtained after processing is divided into two parts, wherein randomly select in data set 70% and be used as training set, then will
Remaining 30% data are as test set；
The initialization of 2.3 networks
Assuming that the node number of input layer is n, the node number of hidden layer is l, and the node number of output layer is m, enables hidden layer
The number of plies is one layer, number of nodes rule of thumb formula:Wherein a is that constant takes 1~10, and cycle-index is set as N
Secondary, the weight of input layer to hidden layer is wij, the weight of hidden layer to output layer is wjk, the threshold value of input layer to hidden layer is
aj, the threshold value of hidden layer to output layer is bk, each layer weight initial value take the random number between [- 1,1], and learning rate η takes
Value is 0.1~0.2, and excitation function is g (x), and wherein excitation function takes Sigmoid function, form are as follows:
2.4 training neural networks；
2.5 data test
After the data training for completing all training sets, neural network is tested with the data of test set, if error is providing
In range, then elastic BP neural network completes modeling, i.e., outputting and inputting for neural network model meets mapping relations.
5. elastomeric network modeling as claimed in claim 4 and the blower multiobjective optimization control method based on genetic algorithm,
It is characterized in that, in described 2.4, the process of training neural network is as follows:
2.4.1 the forward-propagating of signal
2.4.1.1 the output of hidden layer:
xiFor the data of the input of input layer, HjFor the output of hidden layer node；
2.4.1.2 the output of output layer:
OkFor the output of output layer node；
2.4.1.3 the calculating of error:
ek=Ok-Yk
In above formula, i=1...n, j=1...l, k=1...m, YkFor reality output data；
By comparing the reality output and desired output of output layer, the error of the two is obtained, if error is not in the error of requirement
In range, then error back propagation is transferred to；
2.4.2 the backpropagation of signal
The weight of resilient BP algorithm and the following formula of the update of threshold value:
W (t+1)=w (t)+Δ w (t)
Wherein t is the number of iterations, and Δ w (t) obtains correction amount for weight for t, and gradient isOnly the direction of weighed value adjusting is acted as
With, if the direction adjusted twice in succession is identical, it is increased by Δ t, α is selected as 1.2 at this time, if what is adjusted twice in succession is contrary,
It is reduced by Δ t, β is selected as 0.8 at this time, in the case of other, without updating；
All steps in step 2.4.1-2.4.2 are repeated, if error is in desired error range or cycle-index is more than to set
Until setting the upper limit, then training stops.
6. elastomeric network modeling as claimed in claim 1 or 2 and the blower multiobjective optimization control method based on genetic algorithm,
It is characterized in that, in the step 3, the step of genetic algorithm, is as follows:
3.1 generate parent population Pt, Population Size N；
3.2 calculate non-dominant rank individual in parent population Pt, crowding distance；
3.3 are selected again, are intersected, being made a variation, and subgroup Qt primary is generated；
3.4 merge subgroup Qt primary and parent population Pt, generate new parent population Pt+1, Population Size 2N；
3.5 calculate non-dominant rank individual in new parent population Pt+1, crowding distance；
3.6 couples of new parent population Pt+1 are selected, are intersected, are made a variation, and new parent subgroup Qt+1, Population Size N are generated；
3.7 judge whether evolutionary generation is less than maximum algebra G, if it is not, then exporting Qt+1；If so, t=t+1, returns to the 4th step
Continue to recycle.
7. elastomeric network modeling as claimed in claim 1 or 2 and the blower multiobjective optimization control method based on genetic algorithm,
It is characterized in that, population coding mode uses real coding in the first step, population scale is set as N, evolutionary generation is set as t,
Intersect general fork probability and be set as 20%~50%, mutation probability is set as 1%~10%.
8. elastomeric network modeling as claimed in claim 1 or 2 and the blower multiobjective optimization control method based on genetic algorithm,
It is characterized in that, in the second step and the 5th step, it, will be neural using quick non-dominated ranking operator and crowding comparison operator
The predicted value of network model must be calculated as target function value in genetic algorithm.
9. elastomeric network modeling as claimed in claim 8 and the blower multiobjective optimization control method based on genetic algorithm,
Be characterized in that, in the third step and the 6th step, selection refer to by quick non-dominated ranking operator, crowding comparison operator into
Row selection；Intersect the combined crosswise referred to through chromosome, generates new individual；Variation refers to from group optionally one by one
Body is made a variation to generate more excellent individual to certain section of coding in selected chromosome, is handed over using real number interior extrapolation method
Fork operation；Using j-th of gene a of i-th of individualijMutation operation is carried out, variation method is as follows:
aij=aij+(aij-amax) * f (g) r > 0.5
aij=aij+(amin-aij)*f(g) r≤0.5
In formula, random number of the r between [0,1], amaxFor the previous of gene, aminFor the next time of gene；F (g)=rand × (1-g/
Gmax)2；Random number of the rand between [0,1]；G is current iteration number；GmaxFor maximum evolution number.
10. elastomeric network modeling as claimed in claim 1 or 2 and the blower multiobjective optimal control side based on genetic algorithm
Method, which is characterized in that in the 4th step, operator, i.e., the filial generation kind generated parent population with it are designed using elitism strategy
Group closes, and is selected by quick non-dominated ranking operator and crowding comparison operator, and next-generation population is generated；This is conducive to
It keeps the defect individual in parent to enter the next generation, guarantees that the optimum individual in population will not be lost；
In the step 4, process is as follows: the training to data sample is first passed through, the foundation of elastic neural network model is completed,
The models fitting has gone out the good mapping relations for outputting and inputting variable；Then the predicted value of neural network is used for hereditary calculation
Target function value seeks in method；Global optimizing is carried out finally by genetic algorithm, finds out fan operation efficiency and wind pressure or effect
Most ideal point, that is, pareto optimum point of rate and air quantity；Since the end value that the model obtains is the value after normalization, therefore need
Anti-normalization processing is carried out to performance variable value corresponding to pareto optimum point again, be converted into true value, formula is as follows:
X=k (xmax-xmin)+xmin。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910137515.9A CN110033118A (en) | 2019-02-25 | 2019-02-25 | Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910137515.9A CN110033118A (en) | 2019-02-25 | 2019-02-25 | Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110033118A true CN110033118A (en) | 2019-07-19 |
Family
ID=67235724
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910137515.9A Pending CN110033118A (en) | 2019-02-25 | 2019-02-25 | Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110033118A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112181867A (en) * | 2020-09-29 | 2021-01-05 | 西安电子科技大学 | On-chip network memory controller layout method based on multi-target genetic algorithm |
| CN113705098A (en) * | 2021-08-30 | 2021-11-26 | 国网江苏省电力有限公司营销服务中心 | Air duct heater modeling method based on PCA and GA-BP network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106681146A (en) * | 2016-12-31 | 2017-05-17 | 浙江大学 | Blast furnace multi-target optimization control algorithm based on BP neural network and genetic algorithm |
| CN106951983A (en) * | 2017-02-27 | 2017-07-14 | 浙江工业大学 | Injector performance Forecasting Methodology based on the artificial neural network using many parent genetic algorithms |
| CN108846423A (en) * | 2018-05-29 | 2018-11-20 | 中国农业大学 | Water quality prediction method and system |
| CN109241630A (en) * | 2018-09-11 | 2019-01-18 | 国网河北能源技术服务有限公司 | The method for optimizing scheduling and device of electric system |
-
2019
- 2019-02-25 CN CN201910137515.9A patent/CN110033118A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106681146A (en) * | 2016-12-31 | 2017-05-17 | 浙江大学 | Blast furnace multi-target optimization control algorithm based on BP neural network and genetic algorithm |
| CN106951983A (en) * | 2017-02-27 | 2017-07-14 | 浙江工业大学 | Injector performance Forecasting Methodology based on the artificial neural network using many parent genetic algorithms |
| CN108846423A (en) * | 2018-05-29 | 2018-11-20 | 中国农业大学 | Water quality prediction method and system |
| CN109241630A (en) * | 2018-09-11 | 2019-01-18 | 国网河北能源技术服务有限公司 | The method for optimizing scheduling and device of electric system |
Non-Patent Citations (6)
| Title |
|---|
| 丛爽: "《智能控制系统及其应用》", 31 August 2013, 中国科学技术大学出版社 * |
| 姚锡凡等: "《制造物联网技术》", 31 December 2018, 华中科技大学出版社 * |
| 徐振东: "人工神经网络的数学模型建立及成矿预测BP网络的实现", 《中国优秀博硕士学位论文全文数据库（硕士） 信息科技辑》 * |
| 朱会霞: "《二进制遗传算法的改进研究》", 31 March 2018, 东北大学出版社 * |
| 李兵等: "《机械故障信号的数学形态学分析与智能分类》", 31 October 2011, 国防工业出版社 * |
| 王华等: "《Matlab在电信工程中的应用》", 30 April 2001, 中国水利水电出版社 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112181867A (en) * | 2020-09-29 | 2021-01-05 | 西安电子科技大学 | On-chip network memory controller layout method based on multi-target genetic algorithm |
| CN112181867B (en) * | 2020-09-29 | 2022-07-26 | 西安电子科技大学 | On-chip network memory controller layout method based on multi-target genetic algorithm |
| CN113705098A (en) * | 2021-08-30 | 2021-11-26 | 国网江苏省电力有限公司营销服务中心 | Air duct heater modeling method based on PCA and GA-BP network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109932903A (en) | The air-blower control Multipurpose Optimal Method of more parent optimization networks and genetic algorithm | |
| CN108133258B (en) | Hybrid global optimization method | |
| CN109634121A (en) | More parent genetic algorithm air source heat pump multiobjective optimization control methods based on radial basis function neural network | |
| CN109598092A (en) | Merge the air source heat pump multi-objective optimization design of power method of BP neural network and more parent genetic algorithms | |
| CN102629106B (en) | Water supply control method and water supply control system | |
| CN109084415A (en) | Central air-conditioning operating parameter optimization method based on artificial neural network and genetic algorithms | |
| CN102129259B (en) | Neural network proportion integration (PI)-based intelligent temperature control system and method for sand dust environment test wind tunnel | |
| CN109146121A (en) | The power predicating method stopped in the case of limited production based on PSO-BP model | |
| CN109858093A (en) | The air source heat pump multi-objective optimization design of power method of the non-dominated sorted genetic algorithm of SVR neural network aiding | |
| CN105447509A (en) | Short-term power prediction method for photovoltaic power generation system | |
| CN109886448A (en) | Using learning rate changing BP neural network and the heat pump multiobjective optimization control method of NSGA-II algorithm | |
| Li et al. | A hybrid assembly sequence planning approach based on discrete particle swarm optimization and evolutionary direction operation | |
| CN105608295B (en) | The multi-objective genetic algorithm of coking furnace pressure and RBF neural Optimization Modeling method | |
| CN105487496A (en) | Optimization method for heat-engine plant thermal on-line process identification and control algorithm based on dual-objective parallel ISLAND-HFC mixed model genetic programming algorithm | |
| CN107400935A (en) | Adjusting method based on the melt-spinning technology for improving ELM | |
| CN112069656A (en) | Durable concrete mix proportion multi-objective optimization method based on LSSVM-NSGAII | |
| CN110033118A (en) | Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm | |
| CN111105077B (en) | DG-containing power distribution network reconstruction method based on firefly mutation algorithm | |
| CN109829244A (en) | The blower optimum design method of algorithm optimization depth network and three generations's genetic algorithm | |
| Bekker | Applying the cross-entropy method in multi-objective optimisation of dynamic stochastic systems | |
| CN106940068A (en) | A kind of determination method of the water cool-storage technology energy saving of system Optimal Control Strategy based on GACA algorithm | |
| CN105975701A (en) | Parallel scheduling disassembly path forming method based on mixing fuzzy model | |
| CN113654143B (en) | Air-conditioning water system pump valve linkage control method based on neural network | |
| Zhang et al. | Optimization of neural network based on genetic algorithm and BP | |
| CN110348576A (en) | A kind of decision-making technique for improving multiple target bacterial chemotaxis algorithm based on index |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190719 |