CN111007716A - Alternating current servo motor variable discourse domain fuzzy PI control method based on prediction function - Google Patents

Abstract

The invention discloses a variable-theory-domain fuzzy PI control method of an alternating current servo motor based on a prediction function, which comprises a PI controller, a variable-theory-domain fuzzy controller and a prediction function controller; the PI controller is used as a master controller to adjust the rotating speed of the servo motor; the prediction function controller outputs and predicts the future state of the system according to the current input instruction, the control signal and the feedback of the system; the variable-discourse-domain fuzzy controller takes the prediction information as input to adjust the gain of the PI controller on line; meanwhile, a variable domain expansion factor is designed, the domain of the fuzzy controller is adjusted in advance on line by using the prediction information, and a fuzzy rule for adjusting the PI control gain is indirectly increased. Compared with the traditional PI control and fuzzy PI control, the method has the advantages of high precision of variable-discourse-domain fuzzy control, improved dynamic response performance and disturbance resistance, and more outstanding advantages for application occasions with uncertain disturbance and large nonlinearity.

Description

Variable universe fuzzy PI control method for AC servo motor based on prediction function

technical field

The invention relates to the technical field of servo motor control, in particular to a variable universe fuzzy PI control method of an AC servo motor based on a prediction function.

Background technique

An AC servo motor is a machine that converts electrical energy into mechanical energy, consisting of an electromagnet winding or distributed stator winding for generating a magnetic field and a rotating armature or rotor. AC servo motors have the advantages of stable operation, good speed controllability and fast response speed, and have been widely used in CNC machine tools, industrial robots, medical equipment and other equipment. With the continuous development of industrial production, higher and higher requirements are also put forward for the control of AC servo motors, and the study of high-performance servo control systems has become one of the key issues to improve the level of equipment manufacturing.

The main problems and deficiencies in the prior art include:

PI controller has the characteristics of simple structure, strong robustness and easy implementation, and is still the most commonly used control strategy for AC servo motors. However, the fixed-gain PI control parameters have a weak ability to deal with time-varying working states, and cannot effectively deal with various nonlinear problems such as sensor delays, time-varying key parameters, and unknown load disturbances during the operation of the servo motor. Therefore, in order to further improve the control performance, the PI controller needs to be combined with other control strategies to form a composite control strategy, so as to realize online self-adaptation of control parameters.

The fuzzy PI control strategy combines the nonlinear control ability of fuzzy control and the good adaptability of PI control, and has been applied in the control of AC servo system. It can use expert knowledge to build a fuzzy rule base, and adjust the PI control gain online based on fuzzy reasoning, avoiding the complex and time-consuming system model identification process. Its limitation is that the adaptive ability of the fuzzy PI controller depends heavily on the number of fuzzy rules, and the fuzzy rules lack theoretical design methods. To further improve the regulation accuracy of fuzzy control, rich experience in motor control is required to design complex and detailed fuzzy rules. When the application conditions of AC servo motors are complex, it is difficult to design a high-performance fuzzy PI controller, which limits the further promotion of fuzzy PI control in the field of motor control. Therefore, it is necessary to explore simple and effective control strategies to further improve the active disturbance rejection capability and dynamic response performance of fuzzy PI control.

SUMMARY OF THE INVENTION

In view of the above problems and deficiencies existing in the prior art, the present invention provides a variable universe fuzzy PI control method for an AC servo motor based on a prediction function, which adopts the variable universe strategy to dynamically adjust the control range of the fuzzy rules, and changes the The number of rules near the input of the current fuzzy controller overcomes the problem of insufficient control performance caused by the difficulty of fuzzy control rule design; at the same time, the prediction function control is used to predict the future state of the AC servo motor control system, providing input information for the variable universe fuzzy controller, The parameters of fuzzy controller and PI controller can be adjusted quickly.

For this reason, the present invention adopts the following technical solutions:

A variable universe fuzzy PI control method for an AC servo motor based on a prediction function, comprising a PI controller, a variable universe fuzzy controller and a prediction function control module. The PI controller acts as a main controller and is used to adjust the AC servo motor rotation speed; the prediction function control module predicts the future state of the system according to the current input command, control signal and feedback output of the system; the variable universe fuzzy controller uses the prediction information as input to adjust the gain of the PI controller online; design variable theory The domain scaling factor uses the prediction information to adjust the universe of variable universe fuzzy controller ahead of time, and indirectly increases the fuzzy rules used to adjust the gain of the PI controller.

Further, the following steps are included:

Step 1: According to the characteristics of the AC servo motor control system, the prediction function control equation is improved to predict the future feedback speed ω _p (t+p), feedback error e(t+p) and feedback error change rate Δe of the system (t+p);

Step 2: Select the prediction feedback error e(t+p) and the prediction feedback error change rate Δe(t+p) after p steps as the input variables of the variable universe fuzzy controller, and the corresponding fuzzy variables are E and ΔE;

Step 3, analyze the control process of the AC servo motor, and design the fuzzy rules based on the variation law of the PI control parameters in different control stages; at the same time, determine the membership function and defuzzification strategy of the fuzzy controller, so as to design the Mamdani fuzzy controller;

Step 4: Design the scaling factor and adjust the fuzzy input universe online;

Step 5, perform fuzzy inference according to the Mamdani fuzzy inference rule to obtain the fuzzy output;

Step 6, perform a clearing operation on the fuzzy output, and convert the fuzzy output into a clear value;

Step 7: Modify the PI control parameters and output them to the controlled object, so as to adjust the motor speed according to the current and future system states.

Preferably, the specific process of step 1 is as follows:

The prediction function control uses the first-order autoregressive model as the prediction model of the AC servo motor control system: ω(t)=aω(t-1)+bi _q (t-1), where a and b are the models to be identified Parameters, i _q (t-1) and ω(t-1) are the q-axis input current and feedback output speed of the AC servo motor at time t-1, respectively, and ω(t) is the output speed of the AC servo motor at time t ; In the AC servo motor control system, when the prediction time does not exceed the closed-loop response time of the servo system, the input current of the q-axis will remain unchanged, that is, i _q (t+i)=i _q (t), i=1, 2,3...,p, where p is the prediction step size;

According to the prediction model, the output of the AC servo motor after p steps can be obtained as:

Using the deviation information before time t, predict the deviation of the output of the prediction model at time t+p:

In the formula, H={h(tj)=ω(tj) _-ωm (tj)|j=1,2,...,p}, h(tj) is the deviation information at time tj; W =[w ₀ , w ₁ ,...w _j ...w _p-1 ] ^T is the bias weight vector;

Based on the above prediction deviation, the predicted output of the AC servo motor after p steps is corrected as:

ω _p (t+p)=ω _m (t+p)+h(t+p);

The feedback speed error and error rate of change of the AC servo motor after p steps are predicted as:

e(t+p)= _ωr (t+ _p )-ωp(t+p),

Δe(t+p)=e(t+p)−e(t+p−1).

Preferably, the specific process of step 2 is as follows:

映射到模糊论域

其量化因子为η_e和η_Δe，相应的计算公式为：remove input variables from the natural universe

Mapping to Fuzzy Universe

Its quantization factors are η _e and η _Δe , and the corresponding calculation formulas are:

和

分别为自然论域和模糊论域的最大值，x_i代表输入变量e(t+p)或Δe(t+p)，u_i代表模糊变量E或ΔE，k代表输入量化因子η_e或η_Δe且可以通过公式

来确定。In the formula,

and

are the maximum values of the natural universe and the fuzzy universe, respectively, x _i represents the input variable e(t+p) or Δe(t+p), _ui represents the fuzzy variable E or ΔE, and k represents the input quantization factor η _e or η _Δe and can be obtained by the formula

to make sure.

Preferably, the fuzzy linguistic variables used in step 3 are {N, Z, P}, representing {negative, zero, positive} respectively; before fuzzy inference, fuzzy numerical variables are fuzzified by membership function and converted into fuzzy Linguistic variables; triangular and Gaussian membership functions near and on both sides of the fuzzy domain zero.

Preferably, the calculation formula of the scaling factor in step 4 is:

是对称型模糊论域的最大值，ε是任意的非零最小正值且τ∈[0.5,1)。where _ui is the input variable after fuzzification,

is the maximum value of the symmetric fuzzy universe, ε is any non-zero minimum positive value and τ∈[0.5,1).

和

Preferably, the fuzzy output in step 5 is y _Δk (e(t+p), Δe(t+p)), where y _Δk represents the blurring of the PI adjustment value

and

Further, according to the Mamdani-type fuzzy inference criterion and formula, the firing rate of each rule of the variable universe fuzzy controller can be calculated as:

In the formula, η( ) is the membership degree of fuzzy variables E _i and ΔE _i , ∧ is the operator that takes the larger value, and the index Δk represents Δk _p or Δk _i ;

According to the defuzzification strategy of the center of gravity method, the fuzzy output of the fuzzy controller is

和

C_j代表模糊输出论域的离散值。In the formula, y _Δk represents

and

C _j represents the discrete value of the fuzzy output universe.

Preferably, the specific process of step 6 is as follows:

According to the output quantization factors α ₀ and β ₀ , the fuzzy output is sharpened, and the fuzzy output is converted into sharp values (Δk _p , Δk _i ); the output quantization factor is adjusted online by designing the optimization coefficient ζ _k , and the calculation formula is: :

ζ _Δk =γ+y _Δk (e(t+p),Δe(t+p)),

In the formula, γ is the maximum value of the fuzzy universe whose midpoint is 0, and the index Δk represents the parameter used to generate _{Δk p} or the parameter used to generate Δki;

The online update formula of the output quantization factor is:

The PI control parameter adjustment amount (Δk _p , Δk _i ) is updated according to the following formula:

Preferably, the calculation formula of the control parameter in step 7 is as follows:

Among them, k _p0 and k _i0 respectively represent the initial parameters of the PI controller.

In the present invention, firstly, based on the prediction function control theory, a prediction model for AC servo motor control is designed to predict the motor speed and feedback speed error after p steps; then, a variable universe fuzzy controller is constructed, and the input fuzzy universe of the fuzzy controller is Self-adaptation is realized by scaling factor, and the output quantization factor is adjusted online by the optimization coefficient designed according to the fuzzy output.

Compared with the prior art, the beneficial effects of the present invention are:

(1) The present invention improves the prediction function control strategy according to the motion characteristics of the AC servo motor, and the designed prediction model has the advantage of small online calculation amount on the basis of maintaining good prediction accuracy.

(2) The variable universe fuzzy controller designed in the present invention can increase the number of fuzzy rules under the current input through self-adaptive adjustment of the input fuzzy universe, so as to ensure that the fuzzy rules still have practical control accuracy; at the same time, the output quantization factor Through the online adjustment of the optimization coefficient, the effect of fuzzy control is strengthened.

(3) The present invention uses prediction information as the input to optimize the fuzzy control parameters and PI control parameters, which effectively improves the dynamic response performance of the system, especially when the system is disturbed, the control input can be adjusted in time to make the system return to a stable state.

(4) The present invention inherits the advantages of high precision of variable universe fuzzy control, and improves dynamic response performance and anti-disturbance capability, and has more prominent advantages for applications with uncertain disturbance and large nonlinearity.

Description of drawings

FIG. 1 is a schematic diagram of the control structure of a variable universe fuzzy PI control method for an AC servo motor based on a prediction function provided by the present invention.

2 is a schematic structural composition diagram of an AC servo motor speed regulation system in a variable universe fuzzy PI control method for an AC servo motor based on a prediction function provided by an embodiment of the present invention.

3 is a schematic diagram of a control structure of a variable universe fuzzy PI control method for an AC servo motor based on a prediction function provided by an embodiment of the present invention.

FIG. 4 is a schematic diagram of the composition of fuzzy rules provided by an embodiment of the present invention.

FIG. 5 is a schematic diagram of a membership function provided by an embodiment of the present invention.

Detailed ways

The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments, wherein the specific embodiments and descriptions are only used to explain the present invention, but are not intended to limit the present invention.

As shown in FIG. 1 , the present invention discloses a variable universe fuzzy PI control method for an AC servo motor based on a prediction function, including a PI controller, a variable universe fuzzy controller and a prediction function control module. The PI controller acts as a The main controller is used to adjust the rotational speed of the AC servo motor; the prediction function control module predicts the future state of the system according to the current input command, control signal and feedback output of the system; the variable universe fuzzy controller takes the prediction information as input online Adjust the gain of the PI controller; design the variable universe scaling factor, use the prediction information to adjust the universe of the variable universe fuzzy controller ahead of time, and indirectly increase the fuzzy rules for adjusting the gain of the PI controller.

Specifically include the following steps:

(1) According to the characteristics of the AC servo motor control system, the prediction function control equation is improved to better predict the future feedback speed ω _p (t+p), feedback error e(t+p) and feedback error rate of change of the system Δe(t+p).

映射到模糊论域

其量化因子为η_e和η_Δe，相应的计算公式为：(2) Select the prediction feedback error e(t+p) after p steps and the prediction feedback error change rate Δe(t+p) as the input variables of the fuzzy controller, and the corresponding fuzzy variables are E and ΔE. Because the input quantities in the natural universe are clear values and the input quantities in the fuzzy universe are fuzzy values, it is necessary to change the input variables from the natural universe

Mapping to Fuzzy Universe

Its quantization factors are η _e and η _Δe , and the corresponding calculation formulas are:

和

分别为自然论域和模糊论域的最大值，x_i代表输入变量e(t+p)或Δe(t+p)，u_i代表模糊变量E或ΔE，k代表输入量化因子η_e或η_Δe且可以通过公式

来确定。In formula (1),

and

are the maximum values of the natural universe and the fuzzy universe, respectively, x _i represents the input variable e(t+p) or Δe(t+p), _ui represents the fuzzy variable E or ΔE, and k represents the input quantization factor η _e or η _Δe and can be obtained by the formula

to make sure.

(3) Analyze the control process of AC servo motor, and design fuzzy rules based on the variation law of PI control parameters in different control stages; at the same time, determine the membership function and defuzzification strategy of the fuzzy controller, so as to design the Mamdani fuzzy controller.

(4) Design the scaling factor and adjust the fuzzy input universe online. The calculation formula is:

是对称型模糊论域的最大值，ε是任意的非零最小正值且τ∈[0.5,1)。In formula (2), _ui is the input variable after fuzzification,

is the maximum value of the symmetric fuzzy universe, ε is any non-zero minimum positive value and τ∈[0.5,1).

和

(5) Perform fuzzy inference according to the Mamdani fuzzy inference rule to obtain the fuzzy output y _Δk (e(t+p), Δe(t+p)), where y _Δk represents the fuzzy quantity of the PI adjustment value

and

(6) Perform a sharpening operation on the fuzzy output according to the output quantization factors α ₀ and β ₀ , and convert the fuzzy output into sharp values (Δk _p , _Δki ). The invention designs the optimization coefficient ζ _k to adjust the output quantization factor online, and its calculation formula is:

ζ _Δk =γ+y _Δk (e(t+p),Δe(t+p)) (3)

In formula (3), γ is the maximum value of the fuzzy universe whose midpoint is 0, and the index Δk represents the parameter used to generate _{Δk p} or the parameter used to generate Δki.

Therefore, the online update formula of the output quantization factor of the present invention is:

The PI control parameter adjustment amount (Δk _p , Δk _i ) of the present invention is updated according to the following formula:

(7) Modify the PI control parameters and output to the controlled object, so as to adjust the motor speed according to the current and future system states.

In addition, as a further improvement of the present invention, the details of step (1) are as follows:

The prediction function control of the present invention uses the first-order autoregressive model as the prediction model of the AC servo motor control system: ω(t)=aω(t-1)+bi _q (t-1), where a and b are the models The parameters to be identified, i _q (t-1) and ω (t-1) are the q-axis input current and feedback output speed of the AC servo motor at time t-1, respectively. In the AC servo motor control system, when the prediction time does not exceed the closed-loop response time of the servo system, the input current of the q-axis will remain unchanged, that is, i _q (t+i)=i _q (t), i=1,2 ,3...,p, where p is the prediction step size.

Therefore, in the present invention, according to the prediction model, the output of the AC servo motor after p steps can be obtained as:

Due to the existence of model identification errors and the influence of external disturbances, the above predicted output inevitably deviates from the actual control output. Therefore, the deviation information before time t can be used to predict the deviation of the output of the prediction model at time t+p:

In formula (8), H={h(tj)=ω(tj) _-ωm (tj)|j=1,2,...,p}, h(tj) is the deviation at time tj Information; W = [w ₀ , w ₁ , ... w _j ... w _p-1 ] ^T is the bias weight vector.

Based on the above prediction deviation, the predicted output of the AC servo motor after p steps is corrected as

ω _p (t+p)=ω _m (t+p)+h(t+p) (9)

At the same time, the feedback speed error and error change rate of the AC servo motor after p steps can be predicted as:

e(t+p)=ω _r (t+p)-ω _p (t+p) (10)

Δe(t+p)=e(t+p)-e(t+p-1) (11)

Example

图中ω_r是输入指令速度，ω是反馈速度，i_q是q轴指令电流，K_f是力矩系数，J是电机转动惯量，B是粘性摩擦系数，T_l是包括负载力矩、摩擦转矩以及齿槽转矩等的负载扰动。虽然交流伺服电机调速系统是典型的非线性复杂系统，但是在不影响控制效果的前提下可以进行线性化处理。本实施例中，交流伺服电机调速系统采用一阶自回归模型(ARX)来描述：A variable universe fuzzy PI control method of AC servo motor based on prediction function, in which, the AC servo motor speed regulation system based on PI control strategy adopts the control structure shown in Figure 2, and the motor is equivalent to a first-order transfer function

In the figure, ω _r is the input command speed, ω is the feedback speed, i _q is the q-axis command current, K _f is the torque coefficient, J is the motor moment of inertia, B is the viscous friction coefficient, and T _l is the load torque and friction torque. and load disturbances such as cogging torque. Although the AC servo motor speed control system is a typical nonlinear complex system, it can be linearized without affecting the control effect. In this embodiment, the AC servo motor speed control system adopts a first-order autoregressive model (ARX) to describe:

ω(t)=aω(t-1)+bi _q (t-1) (12)

In formula (12), a and b are the model parameters to be identified. The influence of various disturbances on the system can be equivalent to the change of model parameters.

The variable universe fuzzy PI control strategy based on predictive function control in this embodiment is shown in FIG. 3 , and a predictive function control module and a variable universe fuzzy control module are introduced on the basis of the traditional PI control strategy. The feedback speed ω(t) is compared with the input speed command ω _r (t) to obtain the error e(t), which is converted into the q-axis current that controls the motor speed through the PI control module. In order to improve the response speed and accuracy of this control process, it is necessary to adjust the PI control parameters online according to the state of the system. Specific steps are as follows:

Step 1: First, use the prediction function control module to predict the future state of the system, so as to provide input reference information for the variable universe fuzzy control. In the AC servo motor control system, the input current command i _q within the closed-loop response time will remain unchanged. The prediction step p is within the closed-loop response time, then

i _q (t+i)=i _q (t),i=1,2,3...,p (13)

According to equations (12) and (13), using mathematical induction, the predicted speed of the AC servo motor can be obtained as

In formula (14), a ^p+1 represents a raised to the p+1 power.

The motor speed predicted based on the above model will inevitably have errors with the actual speed. In order to further improve the prediction accuracy, the deviation information H between the previous predicted value and the actual value can be used to predict ω _m and the future actual output ω(t+p ) existing deviation h(t+p), so as to correct ω _m (t+p) based on this deviation, the formula is:

ω _p (t+p)=ω _m (t+p)+h(t+p) (15)

in,

In formula (16), H={h(tj)=ω(tj) _-ωm (tj)|j=1,2,...,p}, W is the set of weight coefficients w _j , W=[w ₁ , w ₂ , ... w _j ... w _p ] ^T .

At the same time, the input command ω _r (t+p) after p steps can be predicted according to the input command trajectory. Therefore, the difference between the actual speed of the motor and the commanded speed after p steps can be predicted by the following formula:

e(t+p)= _ωr (t+ _p )-ωp(t+p) (17)

The corresponding error rate of change is:

Δe(t+p)=e(t+p)-e(t+p-1) (18)

映射到模糊论域

计算公式为：Step 2: Design a variable universe fuzzy controller, and use e(t+p) and Δe(t+p) as controller inputs to perform fuzzy reasoning to adjust PI control parameters online. Because both e(t+p) and Δe(t+p) are clear values in the natural universe, and fuzzy inference uses fuzzy values, it needs to be converted from the natural universe using the input quantization factor η _e or η _Δe

Mapping to Fuzzy Universe

The calculation formula is:

和

分别为自然论域和模糊论域的最大值，x_i代表输入变量e(t+p)或Δe(t+p)，u_i代表模糊变量E或ΔE，k代表输入量化因子η_e或η_Δe且可以通过公式

来确定。In formula (19),

and

are the maximum values of the natural universe and the fuzzy universe, respectively, x _i represents the input variable e(t+p) or Δe(t+p), _ui represents the fuzzy variable E or ΔE, and k represents the input quantization factor η _e or η _Δe and can be obtained by the formula

to make sure.

The fuzzy rules used by the fuzzy controller in this embodiment are shown in FIG. 4 , and only include 9 rules. According to the Mamdani-type fuzzy reasoning principle, the basic structure of these rules is: "If the error E is A and the error rate of change ΔE is B, then the proportional gain adjustment of the PI controller is C, and the integral gain adjustment is D", where A , B, C, and D represent fuzzy linguistic variables. The fuzzy linguistic variables used in this embodiment are {N, Z, P}, which represent {negative, zero, positive} respectively.

Before fuzzy inference, fuzzy numerical variables need to be fuzzified by membership function and converted into fuzzy linguistic variables. The membership functions used in this embodiment are shown in FIG. 5 , which are triangular and Gaussian membership functions near and on both sides of the zero point of the fuzzy universe. It can be seen that there are fewer fuzzy rules involved in control near the zero point. In order to dynamically increase the number of fuzzy rules, the fuzzy domain of discourse can be adjusted by the scaling factor:

In formula (20), ε is any non-zero minimum positive value and τ∈[0.5,1).

Define a scaling function g(x _i ) with respect to the input x _i to calculate the fuzzy variable after the variable universe:

According to the Mamdani-type fuzzy inference criterion and formula, the firing rate of each rule of the variable universe fuzzy controller can be calculated as:

In formula (22), η(·) is the membership degree of the fuzzy variables E _i and ΔE _i , ∧ is the operator that takes the larger value, and the superscript Δk represents Δk _p or Δk _i .

According to the defuzzification strategy of the center of gravity method, the fuzzy output of the fuzzy controller is

和

C_j代表模糊输出论域的离散值。In formula (23), y _Δk represents

and

C _j represents the discrete value of the fuzzy output universe.

进行清晰化操作，将其转为优化PI控制器的清晰值(Δk_p，Δk_i)。为强化模糊控制器的输出效果，本实施例设计了自适应输出量化因子：Through the output quantization factor α ₀ and β ₀ to the fuzzy output

A sharpening operation is performed to convert it into sharpening values (Δk _p , _Δki ) for the optimized PI controller. In order to strengthen the output effect of the fuzzy controller, this embodiment designs an adaptive output quantization factor:

in,

In formula (25), γ is the maximum value of the fuzzy universe whose midpoint is 0.

According to the adaptive output quantization factor, the adjustment value of the PI parameter can be obtained:

Through the predictive function control module and the variable universe fuzzy controller, the parameters of the PI controller are adaptively adjusted online:

In formula (27), k _p0 and k _i0 are the initial parameters of the PI controller.

Aiming at nonlinear factors such as time delay, uncertain disturbance or time-varying parameters in the AC servo motor speed regulation system, the present invention provides a variable universe fuzzy PI control strategy based on predictive function control, which makes up for the lack of automatic control of traditional PI controllers. Adaptive deficiencies. Compared with the traditional fuzzy PI control strategy, the present invention has better response performance, precision and practicability.

The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and scope of the present invention shall be included in the protection of the present invention. within the range.

Claims (10)

1. A variable-discourse-domain fuzzy PI control method of an alternating current servo motor based on a prediction function comprises a PI controller, a variable-discourse-domain fuzzy controller and a prediction function control module, and is characterized in that: the PI controller is used as a master controller and used for adjusting the rotating speed of the alternating current servo motor; the prediction function control module predicts the future state of the system according to the current input instruction, the control signal and the feedback output of the system; the variable discourse domain fuzzy controller takes the prediction information as input to adjust the gain of the PI controller on line; designing a variable domain expansion factor, using the prediction information to adjust the domain of the variable domain fuzzy controller in advance and on line, and indirectly increasing the fuzzy rule for adjusting the gain of the PI controller.

2. The alternating current servo motor variable-discourse domain fuzzy PI control method based on the prediction function according to claim 1, characterized in that: the method comprises the following steps:

step one, aiming at the characteristics of an alternating current servo motor control system, a prediction function control equation is improved and used for predicting the future feedback speed omega of the system_p(t + p), a feedback error e (t + p), and a feedback error change rate Δ e (t + p);

selecting a prediction feedback error E (t + p) and a prediction feedback error change rate delta E (t + p) after the step p as input variables of the variable universe fuzzy controller, wherein the corresponding fuzzy variables are E and delta E;

analyzing the control process of the alternating current servo motor, and designing a fuzzy rule based on the change rule of the PI control parameter at different control stages; simultaneously determining a membership function and a defuzzification strategy of the fuzzy controller, thereby designing the Mamdani type fuzzy controller;

designing a scaling factor, and adjusting a fuzzy input discourse domain on line;

step five, carrying out fuzzy reasoning according to a Mamdani type fuzzy reasoning rule to obtain fuzzy output quantity;

step six, performing clarification operation on the fuzzy output quantity, and converting the fuzzy output into a clearness value;

and step seven, modifying the PI control parameters, and outputting the PI control parameters to a controlled object, so as to adjust the rotating speed of the motor according to the current and future system states.

3. The alternating current servo motor variable-discourse domain fuzzy PI control method based on the prediction function as claimed in claim 2, characterized in that: the specific process of the step one is as follows:

the prediction function control takes a first-order autoregressive model as a prediction model of an alternating current servo motor control system: ω (t) ═ a ω (t-1) + bi_q(t-1) wherein a and b are parameters to be identified for the model, i_q(t-1) and omega (t-1) are respectively the q-axis input current and the feedback output rotating speed of the alternating current servo motor at the time of t-1, and omega (t) is the output rotating speed of the alternating current servo motor at the time of t; in an alternating current servo motor control system, when the predicted time length does not exceed the closed loop response time of a servo system, the q-axis input current is kept unchanged, namely i_q(t+i)＝i_q(t), i 1,2,3.. times, p, wherein p is a prediction step length;

the output of the alternating current servo motor after p steps can be obtained according to the prediction model as follows:

and (3) predicting the deviation existing in the output of the prediction model at the t + p moment by using the deviation information before the t moment:

wherein H ═ H(t-j)＝ω(t-j)-ω_m(t-j) | j ═ 1,2, a. W ═ W₀,w₁,...w_j...w_p-1]^TIs a bias weight vector;

based on the above prediction deviation, the predicted output of the ac servo motor after p steps is corrected to:

ω_p(t+p)＝ω_m(t+p)+h(t+p)；

the feedback rotating speed error and the error change rate of the alternating current servo motor after the p steps are predicted as follows:

e(t+p)＝ω_r(t+p)-ω_p(t+p)，

Δe(t+p)＝e(t+p)-e(t+p-1)。

4. the alternating current servo motor variable domain fuzzy PI control method based on the prediction function according to claim 2 or 3, characterized in that: the specific process of the second step is as follows:

transforming input variables from natural discourse

Mapping to a universe of ambiguity

With a quantization factor of η_eAnd η_ΔeThe corresponding calculation formula is:

in the formula,

and

maximum, x, of the natural discourse domain and the fuzzy discourse domain, respectively_iRepresents the input variable e (t + p) or Δ e (t + p), u_iRepresenting the fuzzy variable E or Δ E, k representing the input quantisation factor η_eOr η_ΔeAnd can be represented by formulas

To be determined.

5. The alternating current servo motor variable-discourse domain fuzzy PI control method based on the prediction function according to claim 4, characterized in that: the fuzzy linguistic variables used in the third step are { N, Z, P }, which respectively represent { negative, zero, positive }; before fuzzy reasoning, fuzzification processing is carried out on fuzzy numerical variables through a membership function, and the fuzzy numerical variables are converted into fuzzy linguistic variables; and triangular and Gaussian membership functions are respectively arranged near and at two sides of the zero point of the ambiguity domain.

6. The alternating current servo motor variable-discourse domain fuzzy PI control method based on the prediction function according to claim 5, characterized in that: the formula for calculating the scaling factor in the fourth step is as follows:

in the formula u_iIs the input variable after the fuzzification,

is the maximum value of the symmetric modulo domain, ε is any nonzero minimum positive value and τ ∈ [0.5,1 ].

7. The alternating current servo motor variable domain fuzzy PI control method based on the prediction function according to claim 2 or 3, characterized in that: the output of the paste in the fifth step is y_Δk(e (t + p), Δ e (t + p)), where y_ΔkFuzzy quantity representing PI regulation value

And

8. the alternating current servo motor variable-discourse domain fuzzy PI control method based on the prediction function according to claim 7, characterized in that: according to the Mamdani type fuzzy inference rule and formula, the firing degree of each rule of the variable domain fuzzy controller can be calculated as follows:

wherein η (·) is the fuzzy variable E_iAnd Δ E_iIs a large operator, and the corner mark delta k represents delta k_pOr Δ k_i；

According to the defuzzification strategy of the gravity center method, the fuzzy output of the fuzzy controller is

In the formula, y_ΔkRepresents

And

C_jdiscrete values representing the universe of fuzzy outputs.

9. The alternating current servo motor variable-discourse domain fuzzy PI control method based on the prediction function according to claim 8, characterized in that: the concrete process of the step six is as follows:

according to output quantization factor α₀And β₀Performing sharpening operation on fuzzy output quantity, and converting fuzzy output into a sharpening value (delta k)_p，Δk_i) (ii) a Design optimization coefficient ζ_kAnd performing online adjustment on the output quantization factor, wherein the calculation formula is as follows:

ζ_Δk＝Υ+y_Δk(e(t+p),Δe(t+p))，

wherein γ is the maximum of the ambiguity domain with a midpoint of 0, and the corner mark Δ k represents the maximum for generating Δ k_pOr for generating Δ k_iThe parameters of (1);

the online updating formula of the output quantization factor is as follows:

PI control parameter adjustment quantity (delta k)_p，Δk_i) Updated according to the following formula:

10. the alternating current servo motor variable-discourse domain fuzzy PI control method based on the prediction function according to claim 9, characterized in that: the calculation formula of the control parameters in the step seven is as follows:

wherein k is_p0、k_i0Respectively, the initial parameters of the PI controller.

Images