CN104852654A - Permanent magnet synchronous motor speed loop control parameter optimization method based on artificial bee colony algorithm - Google Patents

Abstract

The invention discloses a permanent magnet synchronous motor speed loop control parameter optimization method based on an artificial bee colony algorithm. A artificial bee colony PID controller is employed to replace a conventional PID controller, self-tuning of parameters is automatically realized and applied to a speed loop of a permanent magnet alternating current vector control system, and accurate tracking of the speed of an alternating current servo system is realized. The method specifically includes: to-be-tuned parameters are regarded as nectar sources, and the optimal parameter combination is searched by the adoption of a specific role changing mechanism by the bee colony. The invention designs a fitness function including an overshoot penalty term in order to determine whether the nectar sources are the optimal. Compared with the conventional PID controller, according to the method, the dynamic performance of the servo system can be effectively improved, and no-overshoot control of the speed of the permanent magnet synchronous motor is realized.

Description

Based on artificial bee colony algorithm permagnetic synchronous motor speed ring Optimization about control parameter method

Technical field

The present invention relates to a kind of high accuracy PMSM Servo System method for control speed, particularly a kind of artificial bee colony optimization method of permagnetic synchronous motor speed ring parameter, belongs to high-precision AC SERVO CONTROL field.

Background technology

Permagnetic synchronous motor is little owing to having volume, and torque pulsation is little, and power density advantages of higher, is widely used in field of industrial automation control, particularly the high accuracy SERVO CONTROL field such as Aero-Space, industrial robot, special process equipment.Therefore also more and more higher to the requirement of its speed adjusting performance.

AC servo generally adopts vector control strategy, and in vector control system, speed ring is in order to improve stability and the rapidity of speed, and its performance quality directly has influence on the performance of whole servo system.And industry spot the most often uses is still PID controller, it is simple that it has algorithm, and robustness is good, high reliability.Permagnetic synchronous motor is non-linear a, close coupling, multivariable complication system, and the quality of system speed adjusting performance depends primarily on adjusting of PID controller parameter.So-called speed ring parameter tuning, namely searches out the parameter of one group of optimum, makes AC servo obtain good speed adjusting performance in numerous possible parameter combinations.

Artificial bee colony algorithm is the bionic intelligence evolution algorithm that a kind of simulation bee colony of latest development finds excellent nectar source.This algorithm dexterously honeybee populations is divided into lead honeybee, follow honeybee, the different role such as investigation honeybee, honeybee links together with concrete nectar source, likely separating in nectar source and solution space.Bee colony to the search procedure in nectar source is: 1, lead honeybee and concrete nectar source to link together, it finds nectar source, and shares nectar source information by swing; 2, follow honeybee according to the Information Selection leading honeybee to share nectar source, and carry out gathering honey in its vicinity, find new nectar source, if search the nectar source of higher fitness, then follow honeybee and change into and lead honeybee; 3, honeybee is led repeatedly to search near its previous nectar source neighborhood searched, when repeatedly searching for the nectar source finding higher fitness not yet, then abandoning this nectar source, changing investigation honeybee into, the new nectar source of random search, changes into again when it searches the nectar source of high fitness and leads honeybee.This algorithm outstanding advantages all carries out the overall situation and Local Search in each iterative process, is not easily absorbed in local optimum, and easily restrains.

Summary of the invention

The object of the invention is the dynamic responding speed in order to accelerate permagnetic synchronous motor, improving the problems such as existing velocity control accuracy is limited, designing a kind of new controller based on artificial bee colony algorithm, realize the high precision tracking to permagnetic synchronous motor given speed.

In order to achieve the above object, technical scheme of the present invention there is provided a kind of permagnetic synchronous motor speed ring Optimization about control parameter method based on artificial bee colony algorithm, it is characterized in that, comprises the following steps:

Step one: the three-phase current signal i utilizing current sensor motor in real time _a, i _b, i _c, as the feedback quantity of the electric current loop of vector control system after Clarke transform, park transforms; Photoelectric encoder is utilized to detect rotor-position and the tach signal of motor, and using the feedback quantity of tachometer value as the speed ring of vector control system;

Step 2: the value of feedback of speed ring and set-point are compared, comparative result is sent into artificial bee colony PID controller, and using the given value of current of the output variable of artificial bee colony PID controller as the q axle of motor, meanwhile, the given value of current of the d axle of motor adopts i _d ^*=0, wherein, being designed to of artificial bee colony PID controller:

Regard the parameter Kp to be optimized of the speed ring of vector control system, a certain class value of Ki and Kd as a nectar source, the span of these three parameters is regarded as the search volume of bee colony, remember that its Spatial Dimension is D, honeybee populations algebraically is n, and honeybee adds up to N _a, wherein, lead honeybee scale to be N _g, following honeybee scale is N _f, in the search incipient stage, stochastic generation N _aindividual nectar source, namely obtains initial solution and have:

$X_i^j={\mathrm{Lb}}^j+\mathrm{rand}\left(\mathrm{0,1}\right)({\mathrm{Ub}}^j-L{b}^j)---\left(1\right)$

In formula (1), Ub and Lb is respectively the bound of search volume, j ∈ 1,2 ..., D} is certain component of the D dimension solution vector of optimised parameter composition;

Calculate the fitness that each is separated again, by N before rank _gsolution lead honeybee population X (0) as initial generation, honeybee X is led for the n-th generation _i(n), i ∈ 1,2 ..., N _g, carry out neighborhood search in i position, nectar source, namely have:

In formula (2), l is random integers in [1, D], represents and leads honeybee Stochastic choice one dimension to search for, k ∈ 1,2 ..., N _g, stochastic generation and k ≠ l, for the random number between [-1,1], and must ensure that the V generated is in solution space;

Adopt greedy back-and-forth method at nectar source V _iand X _iin choose more excellent fitness nectar source remain to population of future generation, to lead nectar source information sharing after the search of honeybee perfect (2) when all, then, respectively follow honeybee and select one to lead honeybee according to leading honeybee Population adaptation angle value, and equally its neighborhood is searched for, select probability is:

$P_i=\frac{f\left(X_i\right)}{\underset{t=1}{\overset{N_G}{\mathrm{\Σ}}}f\left(X_t\right)}---\left(3\right)$

In formula (3), f (X _i) be the fitness value in i-th nectar source;

If certain only leads honeybee searching times times around it to reach certain threshold value Limit and does not find more excellent solution yet, then abandon this nectar source, the corresponding honeybee that leads changes investigation honeybee into, reinitializes the nectar source that this leads honeybee corresponding, that is: simultaneously

$X_i(n+1)=\left\{\begin{array}{cc}\mathrm{Lb}+\mathrm{rand}\left(\mathrm{0,1}\right)(\mathrm{Ub}-\mathrm{Lb}),& \mathrm{times}\&GreaterEqual;\mathrm{Limit}\\ X_i\left(n\right),& \mathrm{times}<\mathrm{Limit}\end{array}\right.;$

Step 3: the given value of current in step 2 is sent into PI controller more afterwards with the feedback quantity of the electric current loop in step one, obtains the d axle component u of stator voltage _dand q axle component u _q, run through Parker's inverse transformation and SVPWM module rear drive motor.

Preferably, whether optimumly fitness function f (t) comprising overshoot penalty term is adopted to carry out pricing vector control system medium velocity ring controling parameters, this fitness function f (t) comprises the time and is multiplied by Error Absolute Value integral performance index ITAE and system overshoot Mp, that is:

$f\left(t\right)=\frac{1}{\mathrm{\&alpha;ITAE}+\mathrm{\&beta;Mp}}=\frac{1}{\mathrm{\&alpha;}{\&Integral;}_0^{\&infin;}t\left|e\left(t\right)\right|\mathrm{dt}+\mathrm{\&beta;Mp}},$ In formula, t is the time, and e (t) is velocity error, and α, β are respectively punishment weights.

Beneficial effect of the present invention is: the present invention makes full use of the advantage that artificial bee colony algorithm all carries out the overall situation and Local Search in each iterative process, effectively can avoid the local optimum of speed ring controling parameters, fitness function simultaneously comprises overshoot penalty term, effectively can suppress the overshoot of speed, be suitable for the optimal control of permanent magnet ac servo system speed ring.

Accompanying drawing explanation

Fig. 1 is permanent magnet synchronous motor vector control system structured flowchart;

Fig. 2 is artificial bee colony PID controller structured flowchart;

Fig. 3 is the speed preset value of permagnetic synchronous motor;

Fig. 4 is the speed tracing effect adopting artificial bee colony PID controller;

Fig. 5 is the speed tracing effect adopting conventional PID controllers.

Embodiment

For making the present invention become apparent, be hereby described in detail below with preferred embodiment.

Composition graphs 1-5, adopts artificial bee colony algorithm to optimize permagnetic synchronous motor speed ring controling parameters.

Choosing permagnetic synchronous motor parameter is: rated power P=1.5KW; Nominal torque Tn=10Nm; Stator resistance R=0.4 Ω; Stator inductance L=0.85mH; Number of pole-pairs p=4; Moment of inertia J=9kgcm ².

Step one: the three-phase current signal i utilizing current sensor motor in real time _a, i _b, i _c, as the feedback quantity of electric current loop after Clarke transform, park transforms, utilize photoelectric encoder to detect motor rotor position and tach signal, and using the feedback quantity of tachometer value as speed ring.

Step 2: the feedback quantity of step one medium velocity ring and set-point are compared, comparative result is sent into artificial bee colony PID controller, and using the given value of current of the output variable of controller as q axle, d shaft current given employing i simultaneously _d ^*=0.

As shown in Figure 2, its execution mode is artificial bee colony PID controller structured flowchart:

Regard the parameter Kp to be optimized of the speed ring of vector control system, a certain class value of Ki and Kd as a nectar source, the span of these three parameters is regarded as the search volume of bee colony, remember that its Spatial Dimension is D, corresponding three parameters to be optimized in this example, therefore D=3, honeybee populations algebraically is n, and the maximum iteration time that in this example, iterations n is corresponding is set as Cycle=50, and honeybee adds up to N _a, in this example, get N _a=20, wherein, lead honeybee scale to be N _g, following honeybee scale is N _f, in the search incipient stage, stochastic generation N _aindividual nectar source, namely obtains initial solution and have:

$X_i^j={\mathrm{Lb}}^j+\mathrm{rand}\left(\mathrm{0,1}\right)({\mathrm{Ub}}^j-{\mathrm{Lb}}^j)---\left(4\right)$

In formula (4), Ub and Lb is respectively the bound of search volume, j ∈ 1,2 ..., D} is certain component of the D dimension solution vector of optimised parameter composition;

Calculate the fitness that each is separated again, by N before rank _gsolution lead honeybee population X (0) as initial generation, honeybee X is led for the n-th generation _i(n), i ∈ 1,2 ..., N _g, carry out neighborhood search in i position, nectar source, namely have:

In formula (5), l is random integers in [1, D], represents and leads honeybee Stochastic choice one dimension to search for, k ∈ 1,2 ..., N _g, stochastic generation and k ≠ l, for the random number between [-1,1], and must ensure that the V generated is in solution space;

Adopt greedy back-and-forth method at nectar source V _iand X _iin choose more excellent fitness nectar source remain to population of future generation, to lead nectar source information sharing after the search of honeybee perfect (5) when all, then, respectively follow honeybee and select one to lead honeybee according to leading honeybee Population adaptation angle value, and equally its neighborhood is searched for, select probability is:

$P_i=\frac{f\left(X_i\right)}{\underset{t=1}{\overset{N_G}{\mathrm{\&Sigma;}}}f\left(X_t\right)}---\left(6\right)$

In formula (6), f (X _i) be the fitness value in i-th nectar source;

If certain only leads honeybee searching times times around it to reach certain threshold value Limit and does not find more excellent solution yet, setting search threshold value Limit=80 in this example, then abandon this nectar source, and the corresponding honeybee that leads changes investigation honeybee into, reinitialize the nectar source that this leads honeybee corresponding, that is: simultaneously

$X_i(n+1)=\left\{\begin{array}{cc}\mathrm{Lb}+\mathrm{rand}\left(\mathrm{0,1}\right)(\mathrm{Ub}-\mathrm{Lb}),& \mathrm{times}\&GreaterEqual;\mathrm{Limit}\\ X_i\left(n\right),& \mathrm{times}<\mathrm{Limit}\end{array}\right.---\left(7\right).$

Step 3: the given value of current in step 2 is sent into PI controller more afterwards with the feedback quantity of the electric current loop in step one, obtains the d axle component u of stator voltage _dand q axle component u _q, run through Parker's inverse transformation and SVPWM module rear drive motor.

In order to whether pricing vector control system medium velocity ring controling parameters is optimum, need design one accurately can reflect the fitness function of governing system performance requirement, the present invention is in order to suppress the over control of permanent magnet synchronous electric motor speed, accelerate the dynamic effect of its speed tracing simultaneously, devise fitness function f (t) comprising overshoot penalty term, it, primarily of two part compositions, is respectively: the time is multiplied by Error Absolute Value integral performance index ITAE and system overshoot Mp, that is:

$f\left(t\right)=\frac{1}{\mathrm{\&alpha;ITAE}+\mathrm{\&beta;Mp}}=\frac{1}{\mathrm{\&alpha;}{\&Integral;}_0^{\&infin;}t\left|e\left(t\right)\right|\mathrm{dt}+\mathrm{\&beta;Mp}}---\left(8\right)$

In formula (8), t is the time, and e (t) is velocity error, and α, β are respectively punishment weights.

Thus, nectar source quality is higher, and the fitness value of its correspondence is larger, and mean that the time is multiplied by Error Absolute Value integral performance index ITAE less, simultaneity factor overshoot Mp is also less.Thus guide honeybee populations search best parameter group, make whole governing system obtain good dynamic and static state performance.

As shown in Figure 3, its value is the square-wave signal suddenlyd change between-1000r/min and+1000r/min to the speed preset value of permagnetic synchronous motor, and its cycle is 0.1s.

Fig. 4 and Fig. 5 is respectively the permagnetic synchronous motor speed tracing effect adopting artificial bee colony PID controller and adopt conventional PID controllers, as can be seen from waveform in figure, during given rotating speed sudden change, artificial bee colony PID controller still can tracing preset signal rapidly, almost non-overshoot, and conventional PID controllers just can will return to given through the long period, and overshoot is comparatively large, cannot follow Setting signal rapidly and accurately.

Punish weights α in the present invention, choosing of β can change neatly according to the tracking effect of speed, thus guides honeybee populations to search high-quality nectar source, makes permagnetic synchronous motor obtain preferably speed adjusting performance.

Above-described embodiment is only for explaining inventive concept of the present invention, but not the restriction to rights protection of the present invention, any change utilizing this design to carry out unsubstantiality, all falls into protection scope of the present invention.

Claims (2)

1., based on a permagnetic synchronous motor speed ring Optimization about control parameter method for artificial bee colony algorithm, it is characterized in that, comprise the following steps:

Step one: the three-phase current signal i utilizing current sensor motor in real time _a, i _b, i _c, as the feedback quantity of the electric current loop of vector control system after Clarke transform, park transforms; Photoelectric encoder is utilized to detect rotor-position and the tach signal of motor, and using the feedback quantity of tachometer value as the speed ring of vector control system;

Step 2: the value of feedback of speed ring and set-point are compared, comparative result is sent into artificial bee colony PID controller, and using the given value of current of the output variable of artificial bee colony PID controller as the q axle of motor, meanwhile, the given value of current of the d axle of motor adopts i _d ^*=0, wherein, being designed to of artificial bee colony PID controller:

Regard the parameter Kp to be optimized of the speed ring of vector control system, a certain class value of Ki and Kd as a nectar source, the span of these three parameters is regarded as the search volume of bee colony, remember that its Spatial Dimension is D, honeybee populations algebraically is n, and honeybee adds up to N _a, wherein, lead honeybee scale to be N _g, following honeybee scale is N _f, in the search incipient stage, stochastic generation N _aindividual nectar source, namely obtains initial solution and have:

$X_i^j={\mathrm{Lb}}^j+\mathrm{rand}\left(\mathrm{0,1}\right)({\mathrm{Ub}}^j-{\mathrm{Lb}}^j)---\left(1\right)$

In formula (1), Ub and Lb is respectively the bound of search volume, j ∈ 1,2 ..., D} is certain component of the D dimension solution vector of optimised parameter composition;

Calculate the fitness that each is separated again, by N before rank _gsolution lead honeybee population X (0) as initial generation, honeybee X is led for the n-th generation _i(n), i ∈ 1,2 ..., N _g, carry out neighborhood search in i position, nectar source, namely have:

In formula (2), l is random integers in [1, D], represents and leads honeybee Stochastic choice one dimension to search for, k ∈ 1,2 ..., N _g, stochastic generation and k ≠ l, for the random number between [-1,1], and must ensure that the V generated is in solution space;

Adopt greedy back-and-forth method at nectar source V _iand X _iin choose more excellent fitness nectar source remain to population of future generation, to lead nectar source information sharing after the search of honeybee perfect (2) when all, then, respectively follow honeybee and select one to lead honeybee according to leading honeybee Population adaptation angle value, and equally its neighborhood is searched for, select probability is:

$P_i=\frac{f\left(X_i\right)}{\underset{t=1}{\overset{N_G}{\mathrm{\&Sigma;}}}f\left(X_t\right)}---\left(3\right)$

In formula (3), f (X _i) be the fitness value in i-th nectar source;

If certain only leads honeybee searching times times around it to reach certain threshold value Limit and does not find more excellent solution yet, then abandon this nectar source, the corresponding honeybee that leads changes investigation honeybee into, reinitializes the nectar source that this leads honeybee corresponding, that is: simultaneously

$X_i(n+1)=\left\{\begin{array}{cc}\mathrm{Lb}+\mathrm{rand}\left(\mathrm{0,1}\right)(\mathrm{Ub}-\mathrm{Lb}),& \mathrm{times}\&GreaterEqual;\mathrm{Limit}\\ X_i\left(n\right),& \mathrm{times}<\mathrm{Limit}\end{array}\right.;$

Step 3: the given value of current in step 2 is sent into PI controller more afterwards with the feedback quantity of the electric current loop in step one, obtains the d axle component u of stator voltage _dand q axle component u _q, run through Parker's inverse transformation and SVPWM module rear drive motor.

2. a kind of permagnetic synchronous motor speed ring Optimization about control parameter method based on artificial bee colony algorithm as claimed in claim 1, it is characterized in that, whether optimumly fitness function f (t) comprising overshoot penalty term is adopted to carry out pricing vector control system medium velocity ring controling parameters, this fitness function f (t) comprises the time and is multiplied by Error Absolute Value integral performance index ITAE and system overshoot Mp, that is:

$f\left(t\right)=\frac{1}{\mathrm{\&alpha;ITAE}+\mathrm{\&beta;Mp}}=\frac{1}{\mathrm{\&alpha;}{\&Integral;}_0^{\&infin;}t\left|e\left(t\right)\right|\mathrm{dt}+\mathrm{\&beta;Mp}},$ In formula, t is the time, and e (t) is velocity error, and α, β are respectively punishment weights.