CN112152523B - NN/GA-based energy-saving speed regulation method for direct current motor - Google Patents

NN/GA-based energy-saving speed regulation method for direct current motor Download PDF

Info

Publication number
CN112152523B
CN112152523B CN202010994002.2A CN202010994002A CN112152523B CN 112152523 B CN112152523 B CN 112152523B CN 202010994002 A CN202010994002 A CN 202010994002A CN 112152523 B CN112152523 B CN 112152523B
Authority
CN
China
Prior art keywords
motor
pulse frequency
neural network
power supply
output
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202010994002.2A
Other languages
Chinese (zh)
Other versions
CN112152523A (en
Inventor
专祥涛
刘文豪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan University WHU
Original Assignee
Wuhan University WHU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Wuhan University WHU filed Critical Wuhan University WHU
Priority to CN202010994002.2A priority Critical patent/CN112152523B/en
Publication of CN112152523A publication Critical patent/CN112152523A/en
Application granted granted Critical
Publication of CN112152523B publication Critical patent/CN112152523B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
    • H02P7/00—Arrangements for regulating or controlling the speed or torque of electric DC motors
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/10—Text processing
    • G06F40/166—Editing, e.g. inserting or deleting
    • G06F40/177—Editing, e.g. inserting or deleting of tables; using ruled lines
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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

Abstract

The invention discloses an NN/GA-based energy-saving speed regulation method for a direct current motor, which comprises the following steps: step 1, carrying out multiple experiments on a direct current motor, acquiring a plurality of groups of data about pulse frequency, motor rotating speed, power voltage, power current and motor output torque, calculating power and motor output power, and further calculating motor efficiency; step 2, training by taking the pulse frequency, the power voltage and the motor rotating speed as input and the motor efficiency as output to obtain a neural network model; step 3, utilizing a genetic algorithm, taking the output of the neural network model as a fitness function, knowing the rotating speed of the motor, and obtaining the optimal pulse frequency which enables the motor efficiency to be highest for different power supply voltages; and 4, obtaining a plurality of groups of target rotating speeds, power supply voltages and optimal pulse frequency data through off-line, making a table, and obtaining the optimal pulse frequency on line by using a table look-up method in practical application. The invention can ensure higher energy utilization rate in the speed regulation process of the direct current motor, and obviously improves the efficiency of the speed regulation system.

Description

NN/GA-based energy-saving speed regulation method for direct current motor
Technical Field
The invention relates to the field of power electronic technology, optimization method and control technology, in particular to an NN/GA-based energy-saving speed regulation method for a direct current motor.
Background
At present, a dc motor is still widely used as a main power source of a drive system. However, almost all dc motors operate at a fixed pulse frequency, which is simple, but under the same conditions of pulse width, load torque, target rotation speed, etc., the pulse frequency is different, which results in different efficiency of the governor system, and an optimal pulse frequency exists, which maximizes the efficiency of the governor system. The optimal pulse frequency corresponding to different target rotating speeds and power supply voltages can be obtained, and therefore the purpose of energy saving and speed regulation is achieved. Here we will solve this optimum pulse frequency by means of neural network methods and genetic algorithms.
Genetic algorithms are widely used in a variety of optimization scenarios. Genetic algorithms typically include a series of operations including coding, selection, crossover, and mutation. Two common encoding methods are binary encoding and real number encoding. The genetic algorithm uses a selection operator to carry out the operation of winning or rejecting the population individuals based on the fitness function size and the constraint condition, and in order to protect the optimal individuals from being rejected, the optimal individuals of the population directly enter the sub-population. The crossing process is a process of crossing the parent population pairwise to obtain the offspring population. The mutation process is to randomly mutate part of individuals in the population, so as to increase the diversity of the individuals in the population. Among them, the crossover and mutation process is the most important operator in genetic algorithm, and improper selection can easily cause 'premature' phenomenon.
In order to use a genetic algorithm, an objective function is required to be used as a fitness function, and the motor has more involved parameters and is difficult to measure, so that the difficulty of representing the relation between the energy consumption and the pulse frequency and the power supply voltage by simply using a mathematical model is higher. The neural network method has strong nonlinear fitting capability, can map any complex nonlinear relation, has simple learning rule and is convenient for computer realization. The method has strong robustness, memory capability, nonlinear mapping capability and strong self-learning capability, and therefore, the method is widely applied to various fields in recent years. In the invention, a certain amount of measured data is used for training, and a model with good fitting property can be obtained, so that the relation between the efficiency of the speed regulating system and the pulse frequency, the power supply voltage and the target rotating speed is obtained and is applied to a genetic algorithm as a fitness function. By combining the two, the optimal pulse frequency corresponding to different power supply voltages at a certain target rotating speed can be calculated.
Disclosure of Invention
The technical problem to be solved by the invention is to provide an NN/GA-based direct current motor energy-saving speed regulation method, which aims at the defects in the prior art, takes a neural network method and a genetic algorithm as a framework, takes the energy efficiency of a direct current motor speed regulation system as an evaluation index, and can calculate to obtain a pulse frequency which enables the efficiency of the speed regulation system to be higher when the direct current motor speed regulation method faces different target rotating speeds and different power supply voltages. The method can obtain the optimal pulse frequency by combining a neural network method and a genetic algorithm aiming at different power supply voltages under a certain target rotating speed, so that the efficiency of the speed regulating system is the highest, and the utilization efficiency of energy is effectively improved under the condition of not investing extra cost.
The technical scheme adopted by the invention for solving the technical problems is as follows:
the invention provides an NN/GA-based energy-saving speed regulation method for a direct current motor, which comprises the following steps:
step 1, carrying out multiple experiments on a given direct current motor, and obtaining a plurality of groups of pulse frequency f, motor rotating speed n, power supply voltage U and power supply current I through measurementaAnd the output torque Te data of the motor, and calculating to obtain the power PinMotor output power PoutAnd further calculating to obtain the motor efficiency eta;
step 2, establishing a neural network model, taking the pulse frequency f, the power supply voltage U and the sample data of the motor rotating speed n obtained in the step 1 as input, and taking the sample data of the motor efficiency eta as output for training to obtain the neural network model;
step 3, utilizing a genetic algorithm, taking the neural network model output obtained in the step 2 as a fitness function, and obtaining the optimal pulse frequency which enables the motor efficiency to be highest aiming at different power supply voltages under the condition that the rotating speed of the motor is known;
and 4, obtaining a plurality of groups of target rotating speeds, power supply voltages and optimal pulse frequency data through off-line and making a table, so that the optimal pulse frequency can be quickly obtained by using a table look-up method in practical application.
Further, the specific method of step 1 of the present invention is:
carrying out a plurality of speed regulation experiments by using a direct current motor, and changing the power supply voltage U and the pulse frequency f by adjusting the PWM pulse width and the pulse frequency of a power supply; simultaneously measuring basic parameters of the circuit, and obtaining a plurality of groups of pulse frequency f, motor rotating speed n, power supply voltage U and power supply current IaAnd motor output torque Te data;
calculating the power of the power supply according to the measured data as follows:
Pin＝UIa (1)
the output power of the motor is as follows:
Figure BDA0002691864120000031
the motor efficiency is:
Figure BDA0002691864120000032
further, the specific method of step 2 of the present invention is:
the BP neural network has a three-layer structure: an input layer, a hidden layer, and an output layer. In the invention, an input layer is set to be 3 nodes, a hidden layer is set to be 1 layer and comprises 50 nodes, and an output layer is set to be 1 node; the input layer of the neural network is X, and the expression is as follows:
X＝[f,U,n]T (4)
in the formula, f is pulse frequency, U is power voltage, and n is motor rotating speed;
the input and output of the neural network hidden layer are respectively:
I(2)＝W1X+b0 (5)
O(2)＝g(I(2)) (6)
in the formula, W1As a weight matrix between the input layer and the hidden layer, b0For the bias of the neuron, superscript (2) represents the hidden layer of the neural network, and g (x) is the excitation function of the hidden layer, the expression is as follows:
Figure BDA0002691864120000033
the input and output of the neural network output layer are respectively:
I(3)＝W2O(2)+b1 (8)
O(3)＝h(I(3)) (9)
in the formula, W2As a weight matrix between the output layer and the hidden layer, b1For the bias of the neurons, superscript (3) represents the output layer of the neural network, and h (x) is the excitation function of the output layer due to the outputThe motor efficiency is shown, and the value range of the motor efficiency is 0-1, so that a sigmoid function is used as a stimulus function, and the expression is as follows:
Figure BDA0002691864120000041
in the above process, W1、W2、b0、b1All the variables are the self variables of the neural network, and the variables are updated by using a back propagation algorithm. The error function is set as follows:
Figure BDA0002691864120000042
in the formula, η is the motor efficiency sample data obtained in step 1. Will be the above formula to W1Obtaining W by calculating the partial derivative1The update formula of (2) is:
Figure BDA0002691864120000043
in the formula, r is a learning rate. In the same way, W can be obtained2、b0、b1The update formula of (2).
And (5) performing loop iteration by utilizing the steps until the value of the error function is smaller than the set target error value, and recording the self parameters of the neural network at the moment to finish model training.
Further, the specific method of step 3 of the present invention is:
setting the target rotating speed to be 400r/min and the power supply voltage to be 24V, and taking the pulse frequency as a variable and using 8-bit binary codes to express the pulse frequencies with different sizes as the individual genotypes of the population in order to find the optimal pulse frequency. The population is first initialized, giving a large amount of initial population data representing the starting search points, and the genotype of each individual in the population is randomly generated. In each generation of reproduction, these individuals have PcThe probability of (2) is mated two by two, namely, genes at certain positions are exchanged; has PmAm (a)The gene mutation is that the gene at some position changes from 1 to 0 or from 0 to 1. In this way new genotypes can be generated from the population. And the fitness function for natural selection of the population is:
fitnessfun＝Net(f,24,400)
wherein Net (f, U, n) is a trained neural network model, and U and n are set well and directly substituted; f is the pulse frequency, i.e., the result of conversion of each individual genotype to decimal; according to the fitness function, individuals with low fitness are eliminated in the replacement of each generation, and retained individuals and new individuals generated through mating and mutation are used as next generation individuals to continue heredity. Through a plurality of generations of heredity, the individual genotype with the highest fitness is obtained, namely the pulse frequency with the highest motor efficiency.
Further, the specific method for obtaining multiple groups of target rotation speed, power supply voltage and optimal pulse frequency data in an off-line manner and making a table in step 4 of the invention comprises the following steps:
by setting multiple sets of target rotation speed n and power supply voltage U, the corresponding optimal pulse frequency can be obtained by the above genetic method, that is, each set (n, U) corresponds to an optimal pulse frequency, and then the data can be made into a table, that is:
f＝Table(n,U)
in practical application, the optimal pulse frequency f is quickly obtained on line through a table look-up method.
The invention has the following beneficial effects: 1. a relatively accurate system efficiency model can be obtained based on the measured data and the neural network training, and is used for designing an energy optimization algorithm; 2. by combining a genetic algorithm, under the condition of knowing a target rotating speed, aiming at different power supply voltages, the pulse frequency which enables the efficiency of the speed regulating system to be highest can be obtained, and the power supply utilization efficiency is obviously improved.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a schematic design flow diagram of an embodiment of the present invention;
FIG. 2 is a graph of mean square error trend of a neural network training model process according to an embodiment of the present invention;
FIG. 3 is a regression graph of a neural network training model process according to an embodiment of the present invention;
FIG. 4 illustrates the comparison of predicted data and measured data using a neural network training model in accordance with an embodiment of the present invention;
FIG. 5 is a graph of the optimal pulse frequency versus supply voltage obtained by the genetic algorithm of an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As shown in fig. 1, the NN/GA-based energy-saving speed regulation method for a dc motor according to the embodiment of the present invention includes the following steps:
step 1, carrying out actual measurement on a DC motor speed regulation circuit to obtain a plurality of groups of pulse frequency f, motor rotating speed n, power supply voltage U and power supply current IaAnd the output torque Te and other data of the motor are calculated to obtain the power P of the power supplyinMotor output power PoutAnd further calculating to obtain the motor efficiency eta.
Carrying out a plurality of speed regulation experiments by using a direct current motor, and changing the power supply voltage U and the pulse frequency f by adjusting the PWM pulse width and the pulse frequency of a power supply; simultaneously measuring basic parameters of the circuit, and obtaining a plurality of groups of pulse frequency f, motor rotating speed n, power supply voltage U and power supply current IaAnd motor output torque Te data;
calculating the power of the power supply according to the measured data as follows:
Pin＝UIa (1)
the output power of the motor is as follows:
Figure BDA0002691864120000061
the motor efficiency is:
Figure BDA0002691864120000062
and 2, dividing the data obtained in the step 1 into a training set, a verification set and a test set which respectively account for 70%, 15% and 15% of all the data, setting an input layer to be 3 nodes, a hidden layer to be 1 layer and containing 50 nodes, and an output layer to be 1 node, establishing a neural network, starting training, obtaining a training model meeting error requirements, and storing the training model.
The BP neural network adopted by the embodiment of the invention has a three-layer structure: an input layer, a hidden layer, and an output layer. In the invention, an input layer is set to be 3 nodes, a hidden layer is set to be 1 layer and comprises 50 nodes, and an output layer is set to be 1 node; the input layer of the neural network is X, and the expression is as follows:
X＝[f,U,n]T (4)
in the formula, f is pulse frequency, U is power voltage, and n is motor rotating speed;
the input and output of the neural network hidden layer are respectively:
I(2)＝W1X+b0 (5)
O(2)＝g(I(2)) (6)
in the formula, W1As a weight matrix between the input layer and the hidden layer, b0For the bias of the neuron, superscript (2) represents the hidden layer of the neural network, and g (x) is the excitation function of the hidden layer, the expression is as follows:
Figure BDA0002691864120000063
the input and output of the neural network output layer are respectively:
I(3)＝W2O(2)+b1 (8)
O(3)＝h(I(3)) (9)
in the formula, W2Is an output layer andweight matrix between hidden layers, b1For the bias of the neuron, superscript (3) represents the output layer of the neural network, and h (x) represents the excitation function of the output layer, since the output represents the motor efficiency and has a value in the range of 0-1, we use the sigmoid function as the excitation function here, and the expression is as follows:
Figure BDA0002691864120000071
in the above process, W1、W2、b0、b1All the variables are the self variables of the neural network, and the variables are updated by using a back propagation algorithm. The error function is set as follows:
Figure BDA0002691864120000072
in the formula, η is the motor efficiency sample data obtained in step 1. Will be the above formula to W1Obtaining W by calculating the partial derivative1The update formula of (2) is:
Figure BDA0002691864120000073
in the formula, r is a learning rate. In the same way, W can be obtained2、b0、b1The update formula of (2).
And 3, using the training model obtained in the step 2 as a fitness function as a genetic algorithm, and under the condition that the target rotating speed and the power supply voltage are determined, coding by taking the pulse frequency as a population to complete optimization.
Setting the target rotating speed to be 400r/min and the power supply voltage to be 24V, and taking the pulse frequency as a variable and using 8-bit binary codes to express the pulse frequencies with different sizes as the individual genotypes of the population in order to find the optimal pulse frequency. The population is first initialized, giving a large amount of initial population data representing the starting search points, and the genotype of each individual in the population is randomly generated. In each generation of the multiplication, theseThe individual has PcThe probability of (2) is mated two by two, namely, genes at certain positions are exchanged; has PmI.e., the gene at a certain position changes from 1 to 0 or from 0 to 1. In this way new genotypes can be generated from the population. And the fitness function for natural selection of the population is:
fitnessfun＝Net(f,24,400)
wherein Net (f, U, n) is a trained neural network model, and U and n are set well and directly substituted; f is the pulse frequency, i.e., the result of conversion of each individual genotype to decimal; according to the fitness function, individuals with low fitness are eliminated in the replacement of each generation, and retained individuals and new individuals generated through mating and mutation are used as next generation individuals to continue heredity. Through a plurality of generations of heredity, the individual genotype with the highest fitness is obtained, namely the pulse frequency with the highest motor efficiency.
And 4, obtaining a plurality of groups of target rotating speeds, power supply voltages and optimal pulse frequency data through off-line, making a table, and obtaining the optimal pulse frequency on line by using a table look-up method in practical application.
By setting multiple sets of target rotation speed n and power supply voltage U, the corresponding optimal pulse frequency can be obtained by the above genetic method, that is, each set (n, U) corresponds to an optimal pulse frequency, and then the data can be made into a table, that is:
f＝Table(n,U)
in practical application, the optimal pulse frequency f is quickly obtained on line through a table look-up method.
FIG. 2 is a graph of mean square error trend during neural network training model. Training errors, validation errors, and test errors are shown, and it can be seen that the final mean square error is small, and the test set errors and validation set errors have similarities, with no significant overfitting occurring until iteration 76 (where the best validation performance occurs).
FIG. 3 is a regression graph of a neural network training model process. The network outputs associated with the training set target, the validation set target, and the test set target are shown. It can be seen from the figure that the data falls almost along the 45 degree line, with the network output equal to the target and the R value in each case reaching above 0.999. The phenomena show that the network fitting effect is good, and the network fitting effect can be used for subsequent use.
FIG. 4 is a comparison graph of system efficiency and actual efficiency calculated by using a neural network training model for different pulse frequencies under the conditions that the target rotating speed is 400r/min and the power supply voltage is 24V. As can be seen from the figure, the error between the result calculated by using the neural network model and the actually measured data is not more than 0.1 percent, and the neural network prediction model is proved to be basically consistent with the actual data. Therefore, the neural network can accurately simulate the relationship between the efficiency of the speed regulating system and the pulse frequency and the power supply voltage, and can be applied to subsequent genetic algorithms.
FIG. 5 is a graph of the relationship between the optimal pulse frequency and the power supply voltage obtained by calculation using a genetic algorithm for different power supply voltages under the condition that the target rotation speed is 400 r/min. As can be seen from the figure, overall, as the power supply voltage increases, the optimal pulse frequency for making the efficiency of the governor system higher also increases, and when the power supply voltage is smaller, the optimal pulse frequency increases slowly, and when the power supply voltage is larger, the optimal pulse frequency increases significantly.
It will be understood that modifications and variations can be made by persons skilled in the art in light of the above teachings and all such modifications and variations are intended to be included within the scope of the invention as defined in the appended claims.

Claims (5)

1. An NN/GA-based energy-saving speed regulation method for a direct current motor is characterized by comprising the following steps:
step 1, carrying out multiple experiments on a given direct current motor, and obtaining a plurality of groups of pulse frequency f, motor rotating speed n, power supply voltage U and power supply current I through measurementaAnd the output torque Te data of the motor, and calculating to obtain the power PinMotor output power PoutAnd further calculating to obtain the motor efficiency eta;
step 2, establishing a neural network model, taking the pulse frequency f, the power supply voltage U and the sample data of the motor rotating speed n obtained in the step 1 as input, and taking the sample data of the motor efficiency eta as output for training to obtain the neural network model;
step 3, utilizing a genetic algorithm, taking the neural network model output obtained in the step 2 as a fitness function, and obtaining the optimal pulse frequency which enables the motor efficiency to be highest aiming at different power supply voltages under the condition that the rotating speed of the motor is known;
step 4, obtaining a plurality of groups of target rotating speeds, power supply voltages and optimal pulse frequency data through off-line, making a table, and obtaining the optimal pulse frequency on line by using a table look-up method in practical application;
the specific method of the step 2 comprises the following steps:
the neural network has a three-layer structure: the system comprises an input layer, a hidden layer and an output layer, wherein the input layer is set to be 3 nodes, the hidden layer is 1 layer and comprises 50 nodes, and the output layer is 1 node; the input layer of the neural network is X, and the expression is as follows:
X＝[f,U,n]T
in the formula, f is pulse frequency, U is power voltage, and n is motor rotating speed;
the input and output of the neural network hidden layer are respectively:
I(2)＝W1X+b0
O(2)＝g(I(2))
in the formula, W1As a weight matrix between the input layer and the hidden layer, b0For the bias of the neuron, superscript (2) represents the hidden layer of the neural network, and g (x) is the excitation function of the hidden layer, the expression is as follows:
Figure FDA0003388264100000011
the input and output of the neural network output layer are respectively:
I(3)＝W2O(2)+b1
O(3)＝h(I(3))
in the formula, W2As a weight matrix between the output layer and the hidden layer, b1For the bias of the neuron, the superscript (3) represents the output layer of the neural network, and h (x) represents the excitation function of the output layer, since the output represents the motor efficiency and has a value in the range of 0-1, the sigmoid function is used as the excitation function, and the expression is as follows:
Figure FDA0003388264100000021
in the above process, W1、W2、b0、b1All the variables are the self variables of the neural network, and the variables are updated by using a back propagation algorithm; the error function is set as follows:
Figure FDA0003388264100000022
in the formula, eta is the motor efficiency sample data obtained in the step 1, and the formula is matched with W1Obtaining W by calculating the partial derivative1The update formula of (2) is:
Figure FDA0003388264100000023
wherein r is the learning rate, and W can be obtained by the same method2、b0、b1The update formula of (2);
and (3) performing loop iteration by utilizing the steps until the value of the error function is smaller than the set target error value, recording the parameters of the neural network at the moment, completing model training, and obtaining a fitted neural network model eta which is Net (f, U, n).
2. The NN/GA-based direct current motor energy-saving speed regulation method of claim 1, wherein the specific method in step 1 is as follows:
using a DC motor to carry out a plurality of speed regulation experiments, and regulating the PWM pulse width and the pulse frequency of a power supplyTo vary the supply voltage U and the pulse frequency f; simultaneously measuring basic parameters of the circuit, and obtaining a plurality of groups of pulse frequency f, motor rotating speed n, power supply voltage U and power supply current IaAnd motor output torque Te data;
calculating the power of the power supply according to the measured data as follows:
Pin＝UIa
the output power of the motor is as follows:
Figure FDA0003388264100000031
the motor efficiency is:
Figure FDA0003388264100000032
3. the NN/GA-based direct current motor energy-saving speed regulation method of claim 1, wherein the genetic algorithm in step 3 comprises the specific steps of:
setting a target rotating speed to be 400r/min and a power supply voltage to be 24V, taking the pulse frequency as a variable and using 8-bit binary codes to express pulse frequencies with different sizes as population individual genotypes in order to find an optimal pulse frequency; firstly, initializing a population, namely giving a large amount of initial population data representing initial search points, wherein the genotype of each individual in the population is randomly generated; in each generation of reproduction, these individuals have PcThe probability of (2) is mated two by two, namely, genes at certain positions are exchanged; has PmI.e. the genes at certain positions change from 1 to 0 or from 0 to 1, by which method the population generates new genotypes.
4. The NN/GA-based energy-saving speed regulation method for a direct current motor according to claim 1, wherein the fitness function in step 3 specifically is:
fitnessfun＝Net(f,24,400)
wherein Net (f, U, n) is a trained neural network model, and U and n are set well and directly substituted; f is the pulse frequency, i.e., the result of conversion of each individual genotype to decimal; according to the fitness function, individuals with low fitness are eliminated in the replacement of each generation, the retained individuals and new individuals generated through mating and mutation are used as next generation individuals, and inheritance is continued; through a plurality of generations of heredity, the individual genotype with the highest fitness is obtained, namely the pulse frequency with the highest motor efficiency.
5. The NN/GA-based direct current motor energy-saving speed regulation method of claim 4, wherein the specific method for obtaining multiple sets of target rotation speed, power supply voltage and optimal pulse frequency data offline and making a table in step 4 is as follows:
by setting multiple groups of target rotating speeds n and power supply voltages U, the corresponding optimal pulse frequency can be obtained through the genetic algorithm, namely, each group (n, U) corresponds to an optimal pulse frequency, and the data are made into a table, namely:
f＝Table(n,U)
in practical application, the optimal pulse frequency f is quickly obtained on line through a table look-up method.
CN202010994002.2A 2020-09-21 2020-09-21 NN/GA-based energy-saving speed regulation method for direct current motor Active CN112152523B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010994002.2A CN112152523B (en) 2020-09-21 2020-09-21 NN/GA-based energy-saving speed regulation method for direct current motor

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010994002.2A CN112152523B (en) 2020-09-21 2020-09-21 NN/GA-based energy-saving speed regulation method for direct current motor

Publications (2)

Publication Number Publication Date
CN112152523A CN112152523A (en) 2020-12-29
CN112152523B true CN112152523B (en) 2022-04-01

Family

ID=73893488

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010994002.2A Active CN112152523B (en) 2020-09-21 2020-09-21 NN/GA-based energy-saving speed regulation method for direct current motor

Country Status (1)

Country Link
CN (1) CN112152523B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103835878A (en) * 2013-04-07 2014-06-04 南京理工大学 Maximum power point tracing control method based on neural network optimization starting rotating speed
CN110518847A (en) * 2019-08-30 2019-11-29 长安大学 Surface permanent magnetic Synchronous Machine Models forecast Control Algorithm based on BP neural network

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102692289B (en) * 2012-06-25 2014-03-19 安徽安凯汽车股份有限公司 Test method for measuring efficiency of electric driving system of novel-energy automobile
CN106357178B (en) * 2016-09-30 2018-09-18 中车株洲电力机车研究所有限公司 A kind of low-speed direct driving Wind turbines electric drive system efficiency-optimized control method
CN108416103A (en) * 2018-02-05 2018-08-17 武汉大学 A kind of method for diagnosing faults of electric automobile of series hybrid powder AC/DC convertor
CN110429893B (en) * 2019-08-29 2021-10-08 上海电机学院 Motor controller carrier frequency dynamic optimization method and motor controller
CN111398812B (en) * 2020-03-27 2022-05-27 重庆金康动力新能源有限公司 Motor carrier frequency calibration system, method and device

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103835878A (en) * 2013-04-07 2014-06-04 南京理工大学 Maximum power point tracing control method based on neural network optimization starting rotating speed
CN110518847A (en) * 2019-08-30 2019-11-29 长安大学 Surface permanent magnetic Synchronous Machine Models forecast Control Algorithm based on BP neural network

Also Published As

Publication number Publication date
CN112152523A (en) 2020-12-29

Similar Documents

Publication Publication Date Title
CN111047085B (en) Hybrid vehicle working condition prediction method based on meta-learning
CN109084415A (en) Central air-conditioning operating parameter optimization method based on artificial neural network and genetic algorithms
CN110059887B (en) BP neural network risk identification method and system based on adaptive genetic algorithm
CN110837915B (en) Low-voltage load point prediction and probability prediction method for power system based on hybrid integrated deep learning
CN111119282A (en) Pressure monitoring point optimal arrangement method for water supply pipe network
Akpinar et al. Forecasting natural gas consumption with hybrid neural networks—Artificial bee colony
CN115186803A (en) Data center computing power load demand combination prediction method and system considering PUE
CN112396301A (en) Power consumer demand response characteristic control method based on energy big data driving
Pelusi Improving settling and rise times of controllers via intelligent algorithms
CN112152523B (en) NN/GA-based energy-saving speed regulation method for direct current motor
CN113762370A (en) Depth network set generation method combined with Gaussian random field
CN111563614A (en) Load prediction method based on adaptive neural network and TLBO algorithm
Song et al. Study on GA-based training algorithm for extreme learning machine
Wu et al. Echo state network prediction based on backtracking search optimization algorithm
CN110033118A (en) Elastomeric network modeling and the blower multiobjective optimization control method based on genetic algorithm
CN113762591A (en) Short-term electric quantity prediction method and system based on GRU and multi-core SVM counterstudy
CN113705098A (en) Air duct heater modeling method based on PCA and GA-BP network
CN113095489A (en) Inductive multi-level graph network representation learning method
CN111798037A (en) Data-driven optimal power flow calculation method based on stacked extreme learning machine frame
CN110288159A (en) A kind of smart city construction Measurement Method
Qu et al. Two-stage coevolution method for deep CNN: A case study in smart manufacturing
CN114841472B (en) GWO optimization Elman power load prediction method based on DNA hairpin variation
Kumar et al. Prediction of Rice Production in India Using Artificial Neural Network with Genetic Algorithm
Chen et al. Hybrid FCM learning algorithm based on particle swarm optimization and gradient descent algorithm
Yu Network complexity analysis of multilayer feedforward artificial neural networks

Legal Events

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