CN104155877A - Brushless DC motor fuzzy control system based on genetic algorithm and control method thereof - Google Patents

Abstract

The present invention discloses a brushless DC motor fuzzy control system based on a genetic algorithm and a control method thereof. The control system comprises a drive system used for driving a brushless DC motor, a speed observer used for collecting the rotation speed of the brushless DC motor, a rotation speed reference model which provides the reference data of a motor rotation speed and compares the reference data of a motor rotation speed with the collected motor rotation speed so as to obtain a rotation speed error and an error change rate, a parameter fuzzification module which receives the rotation speed error and the error change rate, carries out quantification on the rotation speed error and the error change rate, and maps the rotation speed error and the error change rate to a fuzzy set discourse domain, a genetic algorithm optimization module which uses the genetic algorithm to carry out online optimization on a module control rule, adjusts related parameters in the fuzzy set discourse domain and makes a fuzzy decision, and a defuzzification module which maps the output amount of the fuzzy decision to a basic discourse domain. According to the system and the method, the genetic algorithm is used to carry out online adjustment of the parameter of the fuzzy controller, and the controller can have good static and dynamic performance in different operation environments.

Description

Fuzzy Control System of Brushless DC Motor based on genetic algorithm and control method

Technical field

The invention belongs to control or the adjusting field of motor, especially a kind of Fuzzy Control System of Brushless DC Motor and control method based on genetic algorithm.

Background technology

Brshless DC motor (brushless DC motor) has obtained increasingly extensive application in industrial circle, and its control system is typical non-linear, Multivariable Coupling system.Traditional pid control algorithm is difficult for meeting the control requirement of high-precision servo control system, is difficult to realize the high precision operation of motor.

Nonlinear control method based on modern control theory and Intelligent Control Theory for realize controlled system high-quality dynamically and steady-state behaviour lay a good foundation, in brshless DC motor control, be fully utilized.The multiple Advanced Control Strategies such as fuzzy control, ANN (Artificial Neural Network) Control, change structure control, robust control, parameter adaptive control have been used to the control of brshless DC motor.

Although many adaptive control algorithms are widely used in the control of brshless DC motor, based on linear model, there is certain deficiency to the control of this nonlinear model of brshless DC motor for the estimation of motor status in these algorithms.Nerual network technique has obtained good effect to the control of this nonlinear system of brshless DC motor, but control algolithm relative complex.Fuzzy control technology is widely used in Electric Machine Control, is a kind of intelligent control technology that imitates human thinking.But in actual applications, for the nonlinear system of time-varying parameter, the control law of its Fuzzy control system is difficult to determine, even if obtained fuzzy control rule under certain condition, along with the variation of system, fuzzy control rule is originally probably not ideal enough.

There is certain deficiency for the Control of Nonlinear Systems of this time-varying parameter of brshless DC motor in the existing method that combines to control brshless DC motor that fuzzy control and PID are controlled: fuzzy is controlled the control law relying on and existed certain deficiency, control system parameter lacks self-adjusting ability, and the aspects such as the optimization of Fuzzy PID Control Rules and Fuzzy control system parameter on-line control are Shortcomings also.

Summary of the invention

Goal of the invention a: object is to provide a kind of Fuzzy Control System of Brushless DC Motor based on genetic algorithm, to solve the problems referred to above of prior art, realize the on-line control of Fuzzy control system parameter, overcome the weakness of traditional fuzzy control system on-line control poor effect, improve the performance of control system.Further object is to provide a kind of control method.

Technical scheme: a kind of Fuzzy Control System of Brushless DC Motor based on genetic algorithm, it comprises limiting the electric current loop of maximum current and in order to the speed closed loop of brshless DC motor rotating speed.

In a further embodiment, be further:

Drive system, for driving brshless DC motor;

Speed observer, for gathering brshless DC motor rotating speed;

Rotating speed reference model, provides the reference data of motor speed, and compares with the motor speed collecting

Obtain speed error and error rate;

Parameter fuzzy module, receives described speed error and error rate, and quantizes, and is mapped to fuzzy set domain,

Genetic algorithm optimization module, adopts genetic algorithm to be optimized module controls rule, and regulates fuzzy theory

Correlation parameter in territory, makes fuzzy decision;

Reverse gelatinization module, is mapped to basic domain by the output quantity of fuzzy decision;

And gather the electric current of brshless DC motor and regulate the electric current loop of power of motor.

Described genetic algorithm optimization module is further used for:

Determine decision variable and constraint condition thereof and coding, coding/decoding method:

In parameter optimization, with reference value K _e0, K _ec0, K _u0and fine setting parameter K ₁, K ₂, K ₃for decision variable, the requirement of the constraint condition of decision variable and system stability performance index is reference, if require systematic steady state error to be less than δ, generally has:

$k_e\≥\frac{1}{2\mathrm{\δ}}$

Set up Optimized model and definite individual evaluation method:

With system maximum overshoot M _padjustment time t _swith steady-state error e _srfor basis, adopt weight coefficient combined method, structure fitness function is:

f＝αexp[-(M _p/M _p0) ²]+βexp[-(t _s/t _s0) ²]+γexp[-(e _sr/e _sr0) ²]

In formula, M _p0, t _s0with, e _sr0be respectively the response index expectation value of system; α, beta, gamma is weight coefficient, M _p0expectation value for system maximum overshoot; e _sr0expectation value for systematic steady state error; t _s0for the system expectation value of adjustment time; Alpha+beta+γ=1;

Carry out genetic manipulation: design 3 genetic operators, determine the operational factor in genetic algorithm:

$P_{\mathrm{si}}=f_i/\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{f}_i$

In formula, f _ibe i individual fitness, P _sibe i individual selecteed probability, n is Population Size;

Determining of operational factor, the definite parameter of genetic algorithm kind needs mainly contains group size M, stops algebraically T, crossover probability P _cwith variation probability P _m, computing formula is respectively:

$p_c=\left\{\begin{array}{cc}{p}_{c1}-\frac{(p_{c1}-p_{c2})(f^{\&prime;}-f_{\mathrm{avg}})}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{c1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.$

$p_m=\left\{\begin{array}{cc}{p}_{m1}-\frac{(p_{m1}-p_{m2})(f_{\max }-f)}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{m1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.$

In formula, f _maxfor fitness maximum in colony, f _avgthe average fitness of Er Weimeidai colony, f' is larger fitness in 2 individualities, f is the individual fitness of variation, p _m1for the variation probability of minimum fitness individuality, p _m2the variation probability of maximum adaptation degree individuality, p _c1for being the crossover probability of minimum fitness individuality, p _c2crossover probability for maximum adaptation degree individuality.

A brshless DC motor fuzzy control method based on genetic algorithm, is characterized in that, comprises the steps:

Step 1: the input of calculating fuzzy controller: comprise that speed error e and speed error change ec;

Step 2: use respectively quantizing factor k _eand k _ecvalue to speed error e and speed error variation ec quantizes, and is mapped to fuzzy set domain;

Step 3: select Triangle-Profile function as the member function of fuzzy controller;

Step 4: formulate fuzzy control decision table;

Step 5: utilize genetic algorithm to be optimized the fuzzy control rule in fuzzy control decision table, draw the fuzzy decision table of optimizing;

Step 6: the optimizing of the Fuzzy Controller Parameters based on genetic algorithm and online adjustment;

Step 7: the output quantity scale factor K of fuzzy decision _ufrom fuzzy domain, be mapped to basic domain, act on control system system and control motor speed.

In a further embodiment, comprise the steps:

Step 1: gather the rotating speed of motor, and with speed reference model in data comparison, obtain velocity deviation and error rate;

Step 2: judgement velocity deviation is in the first state or the second state; If in the first state, execution step 3, if in the second state, execution step 4;

Step 3: definition Dynamic Signal: the velocity deviation of Fuzzy control system and error change are respectively speed error e and error change ec in the speed of brshless DC motor is controlled;

$\left\{\begin{array}{c}e\left(k\right)=n^{*}\left(k\right)-n\left(k\right)\\ \mathrm{ec}\left(k\right)=e\left(k\right)-e(k-1)\end{array}\right.$

In formula, e (k) is relatively obtained by the tach signal of reference model and actual measurement; Ec (k) is the poor of continuous two sampling period e values; n ^*(k) be the k time sampling period reference model response; N (k) is the rotating speed response of the k time sampling period motor;

Use quantizing factor k _eand k _ecthe actual value that quantizes velocity deviation and error rate thereof, is mapped to fuzzy set domain X={-m ,-m+1 ..., 0 ..., m-1, m}, need to be scale factor k _uthe basic domain that carrys out the fuzzy domain of variable clump of output to be transformed into actual output, enters step 5;

Step 4: determine decision variable and constraint condition thereof and coding, coding/decoding method:

In parameter optimization, with reference value K _e0, K _ec0, K _u0and fine setting parameter K ₁, K ₂, K ₃for decision variable, the requirement of the constraint condition of decision variable and system stability performance index is reference, if require systematic steady state error to be less than δ, generally has:

$k_e\&GreaterEqual;\frac{1}{2\mathrm{\&delta;}}$

Set up Optimized model and definite individual evaluation method:

With system maximum overshoot M _padjustment time t _swith steady-state error e _srfor basis, adopt weight coefficient combined method, structure fitness function is:

f＝αexp[-(M _p/M _p0) ²]+βexp[-(t _s/t _s0) ²]+γexp[-(e _sr/e _sr0) ²]

In formula, M _p0, t _s0with, e _sr0be respectively the response index expectation value of system; α, beta, gamma is weight coefficient, M _p0expectation value for system maximum overshoot; e _sr0expectation value for systematic steady state error; t _s0for the system expectation value of adjustment time; Alpha+beta+γ=1;

Carry out genetic manipulation: design 3 genetic operators, determine the operational factor in genetic algorithm:

$P_{\mathrm{si}}=f_i/\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i$

In formula, f _ibe i individual fitness, P _sibe i individual selecteed probability, n is Population Size;

Determining of operational factor, the definite parameter of genetic algorithm kind needs mainly contains group size M, stops algebraically T, crossover probability P _cwith variation probability P _m, computing formula is respectively:

$p_c=\left\{\begin{array}{cc}{p}_{c1}-\frac{(p_{c1}-p_{c2})(f^{\&prime;}-f_{\mathrm{avg}})}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{c1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.$

$p_m=\left\{\begin{array}{cc}{p}_{m1}-\frac{(p_{m1}-p_{m2})(f_{\max }-f)}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{m1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.$

In formula, f _maxfor fitness maximum in colony, f _avgthe average fitness of Er Weimeidai colony, f' is larger fitness in 2 individualities, f is the individual fitness of variation, p _m1for the variation probability of minimum fitness individuality, p _m2the variation probability of maximum adaptation degree individuality, p _c1for being the crossover probability of minimum fitness individuality, p _c2crossover probability for maximum adaptation degree individuality.

Step 5: output control signals to drive system, adjust the state of motor.

Beneficial effect: the present invention can make motor speed response accelerate, overshoot diminishes and the fluctuation of speed reduces, and makes motor have good static properties; Control method of the present invention can make controller in the situation that of load changing, the genetic algorithm adopting can be carried out online adaptive adjusting to the parameter of fuzzy controller, the antijamming capability of raising system, make motor can both have good dynamic property under different running environment, there is stronger robustness.

Accompanying drawing explanation

Fig. 1 is systematic schematic diagram of the present invention.

Fig. 2 is the membership function table of obfuscation of the present invention and reverse gelatinization.

Fig. 3 is fuzzy algorithm process flow diagram of the present invention.

Embodiment

As shown in Figure 1, control system adopts two closed-loop controls.Interior ring is electric current loop, and its Main Function is restriction maximum current, makes system have enough large acceleration torque, and can guarantee the stable operation of system, outer shroud is speed closed loop, 101 are connected to drive direct current generator with brshless DC motor 107, brshless DC motor 107 is connected with speed observer 103, speed observer is surveyed 103 and is obtained brshless DC motor rotating speed, rotating speed reference model 101 is connected with speed observer 103 relatively, relatively afterwards the speed error e and the error change ec that obtain, speed observer 103 is connected with parameter fuzzy module 106, the speed error e that relatively, handle obtains and error change ec are as error originated from input e and the error change ec of fuzzy controller, actual value to error e and variation ec thereof is used respectively quantizing factor k _e109 and k _ec110 quantize, be mapped to fuzzy set domain, obfuscation module 106 is connected with genetic algorithm optimization 105, then by genetic algorithm, fuzzy control rule is optimized to 105 also by genetic algorithm on-line control controller correlation parameter, genetic algorithm optimization 105 is connected with reverse gelatinization module 104, the output quantity of fuzzy decision scale factor k _u108 are mapped to basic domain from fuzzy domain, finally act on control system, to obtain good stability and higher control accuracy.

Its Main Function of outer shroud speed closed loop is ability and the inhibition fluctuation of speed of enhancing system anti-disturbance, and can guarantee static system and the performance of dynamically following the tracks of, and is directly connected to the stable of servo-drive system and high-performance operation.The present invention adopts " bimodulus " fuzzy controller, when velocity deviation is larger, adopts " coarse adjustment " pattern to reach the requirement of servo-drive system rapidity, adopts " fine tuning " pattern to reach the requirement of accuracy when rotating speed deviation is less.Adopt genetic algorithm to carry out on-line control to the parameter of fuzzy controller simultaneously, make controller under different running environment, can both have good Static and dynamic performance.

Genetic-fuzzy adaptive algorithm of the present invention below:

Step 1 301: calculate the input of fuzzy controller through formula 201: speed error e and speed error change ec.

Step 2 302: the value to speed error e and speed error variation ec is used respectively quantizing factor k _eand k _ecquantize, and be mapped to fuzzy set domain.

Step 3 303: select the Triangle-Profile function of Fig. 2 as the member function of fuzzy controller.

Step 4 304: the control law of concluding and sum up fuzzy controller according to expertise draws fuzzy control decision table.

Step 5 305: utilize genetic algorithm to be optimized fuzzy control rule, draw the fuzzy decision table of optimizing.

Step 6 306: the optimizing of the Fuzzy Controller Parameters based on genetic algorithm and online adjustment.

Step 7 307: the output quantity scale factor K of fuzzy decision _ufrom fuzzy domain, be mapped to basic domain, act on control system system and control motor speed.

Step 1 301: calculate the input of fuzzy controller through formula 201: speed error e and speed error change ec.

According to the actual needs of brushless DC motor control system, chosen two-dimensional fuzzy controller herein, choose the speed error e of motor and rate of change ec thereof as the input of two-dimensional fuzzy controller simultaneously, output through obfuscation and fuzzy decision generation one dimension, by reverse gelatinization output control signal, the rotating speed of brshless DC motor is regulated, its design is as follows again:

Dynamic Signal is defined: the speed tracking error e of fuzzy controller and error change ec are respectively speed error e and error change ec in the speed of brshless DC motor is controlled.It is defined as

$\left\{\begin{array}{c}e\left(k\right)=n^{*}\left(k\right)-n\left(k\right)\\ \mathrm{ec}\left(k\right)=e\left(k\right)-e(k-1)\end{array}\right.---201$

In formula, e (k) is relatively obtained by the tach signal of reference model and actual measurement; Ec (k) is the poor of continuous two sampling period e values; n ^*(k) be the k time sampling period reference model response; N (k) is the rotating speed response of the k time sampling period motor.

Step 2 302: the value to speed error e and speed error variation ec is used respectively quantizing factor k _eand k _ecquantize, and be mapped to fuzzy set domain.

In order to increase the sensitivity of control and to be convenient to apply fuzzy rule, the actual value of error e and variation ec thereof is used respectively to quantizing factor k _eand k _ecquantize, be mapped to fuzzy set domain X={-m ,-m+1 ..., 0 ..., m-1, m}.Increase along with m, the control effect of system can be improved, when but m is excessive, to fuzzy control rule, bring difficulty again, generally speaking, it is proper that m is taken as 6 or 7 effects, and in the present invention, fuzzy domain is divided into 7 linguistic variable values such as negative large NB, negative middle NB, negative little NS, zero ZE, just little PS, center PM and honest PB.The output of fuzzy decision can not directly be applied in control system, need to be scale factor k _uthe basic domain that carrys out the fuzzy domain of variable clump of output to be transformed into actual output, finally acts on control system.

Step 3 303: select the Triangle-Profile function of Fig. 2 as the member function of fuzzy controller.

In the present invention, selecting Triangle-Profile function is the member function of fuzzy controller, as shown in Figure 2.

Step 4 304: the control law of concluding and sum up fuzzy controller according to expertise draws fuzzy control decision table.

According to expertise, draw the fuzzy control decision table of " IF-THEN " form, control law can be expressed by form below:

If e is NB and e _c is PM then U is PM，

If e is PM and e _c is NB then U is PS。

Step 5 305: utilize genetic algorithm to be optimized fuzzy control rule, draw the fuzzy decision table of optimizing.

According to the feature of brshless DC motor self, the fuzzy decision that traditional empirical method obtains is relatively difficult to ensure that card all reaches good control effect under different duties, so the present invention is optimized the control law under motor different conditions by genetic algorithm, to obtain controlling preferably effect.The present invention adopts genetic algorithm to be optimized fuzzy control rule, obtains good stability and higher control accuracy.Coded system during the present invention's application genetic algorithm optimization fuzzy control rule: by fuzzy decision rule of 10 binary code representations, first is zone bit, is used for representing whether this rule is used, there is this rule in 1 representative, and this rule is abandoned in O representative; 2 one 4 bit codes, 5 one 7 bit codes, 8 one 10 bit codes are Representative errors e, error change e respectively _c, decision content FD, 3 variablees are all to represent respectively NB, NM, NS, ZE, PS, PM, PB with 001,010,011,100,101,110,111.

Table 1 has provided the fuzzy control rule after genetic algorithm optimization.Through Genetic Algorithm Evolution, draw the optimum code under this state, bad control law, through Genetic Algorithm Evolution, is optimized to good rule.Process relative complex due to Genetic algorithms optimization based fuzzy logic controller control law, in this control system running up of brshless DC motor, even if use digital signal processor (DSP) to be at a high speed also difficult to meet the requirement of system rapidity and real-time, therefore will the optimization of fuzzy control rule be carried out under off-line state, fuzzy control rule after optimizing is written in DSP, makes system there is higher feasibility.

Table 1

Step 6 306: the optimizing of the Fuzzy Controller Parameters based on genetic algorithm and online adjustment.

1, the performance of fuzzy controller determines the performance of Fuzzy control system, and the performance of fuzzy controller depends on fuzzy linguistic rules and fuzzy reasoning.After system performance changes, for making Fuzzy control system there is good dynamic and static character, should constantly adjust the operational factor of fuzzy controller; The situation that is difficult to be applied to different control objects for this fuzzy controller, the present invention has designed the adjustable fuzzy controller of weight coefficient.

The present invention is transformed into corresponding universe of fuzzy sets by input variable from basic domain, and input variable is multiplied by the corresponding factor, thereby draws error quantization factor K _ewith error rate quantizing factor K _ec.In addition, the controlled quentity controlled variable that each sampling provides through fuzzy controller, directly control object, also must be transformed in the receptible basic domain of control object institute, thereby be drawn scale factor K _u.

About operational factor K _e, K _ec, K _uimpact on system responses is as follows:

(1) K _elarger, response curve ascending velocity is faster; K _eexcessive, system produces larger overshoot, and the adjusting time is increased, and there will be oscillatory occurences when serious; K _etoo small, speed of convergence is excessively slow; K _eincrease, the static difference of system reduces.

(2) K _eclarger, the response of system is more blunt, K _ecless, system responses is sensitiveer, and climbing speed is faster, but can make system produce vibration when serious; K _ecincrease, the static difference of system will reduce.

(3) K _ube equivalent to the proportional gain in classical control system, generally, K _ularger, the speed of response is faster; K _uexcessive, there is the serious situation of vibrating of response; K _utoo small, rate of convergence is too slow.In these three factors, K _uvariation having the greatest impact to system responses.

2, basis analysis above, is subject to the restriction of system stability, and the quantizing factor of fuzzy controller can not be too large, so max quantization error e=e _maxtime maximum, the steady-state deviation being caused by quantization error has limited the raising of system progress, so the present invention has designed the variable element bimodulus fuzzy controller based on experience, it is controlled change-over switch and exists place, its control thought shows following 3 points:

(1) exist time, the requirement that focuses on meeting rapidity of control, adopts " coarse adjustment " pattern; Meeting under the prerequisite of system stability, increase as far as possible k _e, k _uto meet the requirement of rapidity, k _eccan select littlely.

(2) exist time, the emphasis of control is transferred to the requirement that meets accuracy, adopts " fine tuning " pattern; The ratio of the fuzzy controller of " fine tuning " pattern and quantizing factor k _e1, k _ec1, k _u1represent, tentatively select k _e1=m*k _e, k _ec1=m*k _ec,, make " coarse adjustment " identical with the universe of fuzzy sets of " fine tuning ", quantization error will reduce like this.

(3) according to the analysis in step 6 306, when fuzzy controller is determined at control law, quantization parameter K _e, K _ec, K _uthe control performance of decision systems, in the fuzzy controller of " coarse adjustment " and " fine tuning ", adopts band to adjust the fuzzy control rule of the factor, by two groups of different quantization parameters, meets requirement separately.

These three parameters are regulated, can accelerate response speed, reduce overshoot, and improve the Static and dynamic performance of fuzzy controller.In system operational process, adopt the fuzzy controller of preset parameter can not obtain good dynamic property and stable performance, therefore, according to the control process mode of system, these parameters are carried out to on-line automatic adjustment.Adjustment to controller parameter, the system dynamic error e of take is variable, to k _e, k _ecand k _uonline adjustment, its self-adjusting formula is:

$K_e=\left\{\begin{array}{cc}{K}_{e0}+K_1*e& \left|e\right|\&le;\frac{e_{\max }}{2}\\ K_{e0}+K_1*\frac{e_{\max }}{2}& \left|e\right|>\frac{e_{\max }}{2}\end{array}\right.---202$

$K_{\mathrm{ec}}=\left\{\begin{array}{cc}{K}_{\mathrm{ec}0}+K_2*e& \left|e\right|\&le;\frac{e_{\max }}{2}\\ K_{\mathrm{ec}0}+K_2*\frac{e_{\max }}{2}& \left|e\right|>\frac{e_{\max }}{2}\end{array}\right.---203$

$K_u=\left\{\begin{array}{cc}{K}_{u0}+K_3*e& \left|e\right|\&le;\frac{e_{\max }}{2}\\ K_{u0}+K_3*\frac{e_{\max }}{2}& \left|e\right|>\frac{e_{\max }}{2}\end{array}\right.---204$

In formula, K _e0, K _ec0, K _u0for reference value, K ₁, K ₂, K ₃for fine setting parameter, e _maxpositive maximal value for error domain.

Increase K _ebe equivalent to dwindle the basic domain of error, increased the control action of error variance.From formula 202, when time, along with the increase of error, K _eincrease, the control action of error is increased.For K _ec, when error reduces gradually, for reducing overshoot, should increase the control action of error change, require exactly K _ecincrease gradually.For K _u, from formula 204, K _ualong with the increase of error, increase, strengthen the control action to error, rate of convergence is accelerated.

3, use genetic algorithm to be optimized Fuzzy Controller Parameters, mainly carry out following steps:

(1) determine decision variable and constraint condition thereof and coding, coding/decoding method

In parameter optimization, with reference value K _e0, K _ec0, K _u0and fine setting parameter K ₁, K ₂, K ₃for decision variable.The requirement of the constraint condition of decision variable and system stability performance index is reference, if require systematic steady state error to be less than δ, generally has:

$k_e\&GreaterEqual;\frac{1}{2\mathrm{\&delta;}}---205$

(2) set up Optimized model and definite individual evaluation method

A feature of genetic algorithm is that it uses the objective function of required problem to obtain next step search information, and using by evaluating individual fitness of objective function embodied.Therefore, fitness function is the Key Functions of genetic algorithm, and fitness function is generally converted by objective function.With system maximum overshoot M _padjustment time t _swith steady-state error e _srfor basis, adopt weight coefficient combined method, structure fitness function is:

f＝αexp[-(M _p/M _p0) ²]+βexp[-(t _s/t _s0) ²]+γexp[-(e _sr/e _sr0) ²] 206

In formula, M _p0, t _s0with, e _sr0be respectively the response index expectation value of system; α, beta, gamma is weight coefficient, reflects the weight of each index in the overall performance of control system, requires alpha+beta+γ=1.This fitness function value is larger, and the performance of illustrative system is better.

(3) carry out genetic manipulation

Genetic manipulation is the control of simulation biological gene heredity, comprises the operational factor in 3 genetic operators of design (selection, crossover and mutation operator) and definite genetic algorithm.Select operator adoption rate to select operator here, each individual selecteed probability becomes ratio with its fitness, selects formula to be:

$P_{\mathrm{si}}=f_i/\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i---207$

In formula 207, f _ibe i individual fitness, P _sibe i individual selecteed probability, n is Population Size.

Crossover operator is that genetic algorithm kind produces new individual main method, is used as main operators because having ability of searching optimum, adopts single-point crossover operator here; And mutation operator just produces new individual householder method, because only there is local search ability, be used as auxiliary operator, adopt basic bit mutation operator here.

About determining of operational factor, the definite parameter of genetic algorithm kind needs mainly contains group size M, stops algebraically T, crossover probability P _cwith variation probability P _m.The present invention gets M=60, T=160 here.The self-adapted genetic algorithm simultaneously proposing according to Srinvivas, these two parameters can change automatically, and its computing formula is respectively:

$p_c=\left\{\begin{array}{cc}{p}_{c1}-\frac{(p_{c1}-p_{c2})(f^{\&prime;}-f_{\mathrm{avg}})}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{c1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.---208$

$p_m=\left\{\begin{array}{cc}{p}_{m1}-\frac{(p_{m1}-p_{m2})(f_{\max }-f)}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{m1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.---209$

In formula, f _maxfor fitness maximum in colony, f _avgthe average fitness of Er Weimeidai colony, f' is larger fitness in 2 individualities, f is the individual fitness of variation, p _m1for the variation probability of minimum fitness individuality, p _m2the variation probability of maximum adaptation degree individuality, p _c1for being the crossover probability of minimum fitness individuality, p _c2crossover probability for maximum adaptation degree individuality.

Crossover probability P _cwith variation probability P _mselection be the key that affects genetic algorithm behavior and performance, directly affect convergence;

P _clarger, the speed of new individual difference is just faster, but cross conference, makes the structure of excellent individual destroyed very soon; P _ctoo small, search procedure is slow, so that comes to a halt;

P _mtoo small, be difficult for producing new individual configurations, P _mexcessive, become pure random search.

In the present invention, get P _c1=0.8, P _c1=0.6, P _m1=0.1, P _m2=0.001.

Step 7: the output quantity scale factor K of fuzzy decision _ufrom fuzzy domain, be mapped to basic domain, act on control system system and control motor speed.

More than describe the preferred embodiment of the present invention in detail; but the present invention is not limited to the detail in above-mentioned embodiment, within the scope of technical conceive of the present invention; can carry out multiple equivalents to technical scheme of the present invention, these equivalents all belong to protection scope of the present invention.

It should be noted that in addition each the concrete technical characterictic described in above-mentioned embodiment, in reconcilable situation, can combine by any suitable mode.For fear of unnecessary repetition, the present invention is to the explanation no longer separately of various possible array modes.

In addition, between various embodiment of the present invention, also can carry out combination in any, as long as it is without prejudice to thought of the present invention, it should be considered as content disclosed in this invention equally.

Claims (5)

1. the Fuzzy Control System of Brushless DC Motor based on genetic algorithm, is characterized in that, it comprises limiting the electric current loop of maximum current and in order to the speed closed loop of brshless DC motor rotating speed.

2. the Fuzzy Control System of Brushless DC Motor based on genetic algorithm as claimed in claim 1, is characterized in that: be further:

Drive system (102), for driving brshless DC motor (107);

Speed observer (103), for gathering brshless DC motor rotating speed;

Rotating speed reference model (101), provides the reference data of motor speed, and compare with the motor speed collecting acquisition speed error and error rate;

Parameter fuzzy module (106), receives described speed error and error rate, and quantizes, and is mapped to fuzzy set domain,

Genetic algorithm optimization module (105), adopts genetic algorithm to be optimized module controls rule, and regulates the correlation parameter in fuzzy domain, makes fuzzy decision;

Reverse gelatinization module (104), is mapped to basic domain by the output quantity of fuzzy decision;

And gather the electric current of brshless DC motor and regulate the electric current loop of power of motor.

3. the Fuzzy Control System of Brushless DC Motor based on genetic algorithm as claimed in claim 2, is characterized in that: described genetic algorithm optimization module (105) is further used for:

Determine decision variable and constraint condition thereof and coding, coding/decoding method:

In parameter optimization, with reference value K _e0, K _ec0, K _u0and fine setting parameter K ₁, K ₂, K ₃for decision variable, the requirement of the constraint condition of decision variable and system stability performance index is reference, if require systematic steady state error to be less than δ, generally has:

$k_e\&GreaterEqual;\frac{1}{2\mathrm{\&delta;}}$

Set up Optimized model and definite individual evaluation method:

With system maximum overshoot M _padjustment time t _swith steady-state error e _srfor basis, adopt weight coefficient combined method, structure fitness function is:

f＝αexp[-(M _p/M _p0) ²]+βexp[-(t _s/t _s0) ²]+γexp[-(e _sr/e _sr0) ²]

In formula, M _p0, t _s0with, e _sr0be respectively the response index expectation value of system; α, beta, gamma is weight coefficient, M _p0expectation value for system maximum overshoot; e _sr0expectation value for systematic steady state error; t _s0for the system expectation value of adjustment time; Alpha+beta+γ=1;

Carry out genetic manipulation: design 3 genetic operators, determine the operational factor in genetic algorithm:

$P_{\mathrm{si}}=f_i/\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i$

In formula, f _ibe i individual fitness, P _sibe i individual selecteed probability, n is Population Size;

Determining of operational factor, the definite parameter of genetic algorithm kind needs mainly contains group size M, stops algebraically T, crossover probability P _cwith variation probability P _m, computing formula is respectively:

$p_c=\left\{\begin{array}{cc}{p}_{c1}-\frac{(p_{c1}-p_{c2})(f^{\&prime;}-f_{\mathrm{avg}})}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{c1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.$

$p_m=\left\{\begin{array}{cc}{p}_{m1}-\frac{(p_{m1}-p_{m2})(f_{\max }-f)}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{m1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.$

In formula, f _maxfor fitness maximum in colony, f _avgthe average fitness of Er Weimeidai colony, f' is larger fitness in 2 individualities, f is the individual fitness of variation, p _m1for the variation probability of minimum fitness individuality, p _m2the variation probability of maximum adaptation degree individuality, p _c1for being the crossover probability of minimum fitness individuality, p _c2crossover probability for maximum adaptation degree individuality.

4. the brshless DC motor fuzzy control method based on genetic algorithm, is characterized in that, comprises the steps:

Step 1: the input of calculating fuzzy controller: comprise that speed error e and speed error change ec;

Step 2: use respectively quantizing factor k _eand k _ecvalue to speed error e and speed error variation ec quantizes, and is mapped to fuzzy set domain;

Step 3: select Triangle-Profile function as the member function of fuzzy controller;

Step 4: formulate fuzzy control decision table;

Step 5: utilize genetic algorithm to be optimized the fuzzy control rule in fuzzy control decision table, draw the fuzzy decision table of optimizing;

Step 6: the optimizing of the Fuzzy Controller Parameters based on genetic algorithm and online adjustment;

Step 7: the output quantity scale factor K of fuzzy decision _ufrom fuzzy domain, be mapped to basic domain, act on control system system and control motor speed.

5. the brshless DC motor fuzzy control method based on genetic algorithm as claimed in claim 4, is characterized in that, is further:

Step 1: gather the rotating speed of motor, and with speed reference model in data comparison, obtain velocity deviation and error rate;

Step 2: judgement velocity deviation is in the first state or the second state; If in the first state, execution step 3, if in the second state, execution step 4;

Step 3: definition Dynamic Signal: the velocity deviation of Fuzzy control system and error change are respectively speed error e and error change ec in the speed of brshless DC motor is controlled;

$\left\{\begin{array}{c}e\left(k\right)=n^{*}\left(k\right)-n\left(k\right)\\ \mathrm{ec}\left(k\right)=e\left(k\right)-e(k-1)\end{array}\right.$

In formula, e (k) is relatively obtained by the tach signal of reference model and actual measurement; Ec (k) is the poor of continuous two sampling period e values; n ^*(k) be the k time sampling period reference model response; N (k) is the rotating speed response of the k time sampling period motor;

Use quantizing factor k _eand k _ecthe actual value that quantizes velocity deviation and error rate thereof, is mapped to fuzzy set domain X={-m ,-m+1 ..., 0 ..., m-1, m}, need to be scale factor k _uthe basic domain that carrys out the fuzzy domain of variable clump of output to be transformed into actual output, enters step 5;

Step 4: determine decision variable and constraint condition thereof and coding, coding/decoding method:

In parameter optimization, with reference value K _e0, K _ec0, K _u0and fine setting parameter K ₁, K ₂, K ₃for decision variable, the requirement of the constraint condition of decision variable and system stability performance index is reference, if require systematic steady state error to be less than δ, generally has:

$k_e\&GreaterEqual;\frac{1}{2\mathrm{\&delta;}}$

Set up Optimized model and definite individual evaluation method:

With system maximum overshoot M _padjustment time t _swith steady-state error e _srfor basis, adopt weight coefficient combined method, structure fitness function is:

f＝αexp[-(M _p/M _p0) ²]+βexp[-(t _s/t _s0) ²]+γexp[-(e _sr/e _sr0) ²]

In formula, M _p0, t _s0with, e _sr0be respectively the response index expectation value of system; α, beta, gamma is weight coefficient, alpha+beta+γ=1;

Carry out genetic manipulation: design 3 genetic operators, determine the operational factor in genetic algorithm:

$P_{\mathrm{si}}=f_i/\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{f}_i$

In formula, f _ibe i individual fitness, P _sibe i individual selecteed probability, n is Population Size;

Determining of operational factor, the definite parameter of genetic algorithm kind needs mainly contains group size M, stops algebraically T, crossover probability P _cwith variation probability P _m, computing formula is respectively:

$p_c=\left\{\begin{array}{cc}{p}_{c1}-\frac{(p_{c1}-p_{c2})(f^{\&prime;}-f_{\mathrm{avg}})}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{c1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.$

$p_m=\left\{\begin{array}{cc}{p}_{m1}-\frac{(p_{m1}-p_{m2})(f_{\max }-f)}{f_{\max }-f_{\mathrm{avg}}}& f^{\&prime;}\&GreaterEqual;f_{\mathrm{avg}}\\ p_{m1}& f^{\&prime;}<f_{\mathrm{avg}}\end{array}\right.$

In formula, f _maxfor fitness maximum in colony, f _avgthe average fitness of Er Weimeidai colony, f' is larger fitness in 2 individualities, f is the individual fitness of variation, p _m1for the variation probability of minimum fitness individuality, p _m2the variation probability of maximum adaptation degree individuality, p _c1for being the crossover probability of minimum fitness individuality, p _c2crossover probability for maximum adaptation degree individuality.

Step 5: output control signals to drive system, adjust the state of motor.