阅读说明：本技术 内置式永磁同步电机单神经元自适应pid弱磁控制方法 (Internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method ) 是由 莫金海 潘海波 陶辉 王红 于 2019-09-12 设计创作，主要内容包括：本发明公开一种内置式永磁同步电机单神经元自适应PID弱磁控制方法,在弱磁控制的基础上,通过单神经元自适应PID控制器对转速环进行调节,兼收了神经网络和PID的各自优势,使控制系统具有更好的动态响应能力和鲁棒性,使控制精度更高,转矩脉动更小,有效提高直流母线电压的利用率；另外,通过利用单神经元自适应PID控制器代替常规积分器,使得控制器各项参数可以在线实时自整定,对于电机运行和环境变化,更具适应性；此外,结合电流解耦前馈补偿方式,使得控制系统完全解耦,有效提高了控制系统性能和控制精度。(The present invention discloses a kind of internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method, on the basis of weak magnetic control, der Geschwindigkeitkreis is adjusted by single-neuron adaptive PID controller, neural network and the respective advantage of PID are taken in concurrently, make control system that there is better dynamic response capability and robustness, keep control precision higher, torque pulsation is smaller, effectively improves the utilization rate of DC bus-bar voltage；In addition, allowing controller parameters real-time Self-tuning System online, for motor operation and environmental change, more adaptability by replacing regular integral device using single-neuron adaptive PID controller；In addition, so that control system is full decoupled, effectively increasing control system performance and control precision in conjunction with Current Decoupling feedforward compensation mode.)

1. internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method, characterized in that comprise the following steps that

Step 1, building single-neuron adaptive PID controller, and the parameter of single-neuron adaptive PID controller is carried out just Beginningization；

Step 2, the mechanical angular velocity omega for detecting internal permanent magnet synchronous motor_m, and after being sampled to it, obtain revolving speed sampling Value ω；

Step 3, by the rotary speed setting value ω of setting^*Its input given value as single-neuron adaptive PID controller, simultaneously Using step 2 gained revolving speed sampled value ω as the actual sample value of single-neuron adaptive PID controller, it is sent into single nerve together First self-adaptive PID controller, then single-neuron adaptive PID controller exports to obtain torque reference value T^*；

Step 4, by torque reference value T^*Maximum torque per ampere control is carried out, quadrature axis current given value is obtainedAnd direct-axis current Given value

Step 5, the three-phase current for detecting internal permanent magnet synchronous motor, and successively sat by static coordinate transformation and synchronous rotary After mark transformation, the quadrature axis current sampled value i under synchronous rotating frame is obtained_qWith direct-axis current sampled value i_d；

Step 6, in conjunction with the resulting quadrature axis current given value of step 4With direct-axis current given valueAnd the resulting friendship of step 5 Shaft current sampled value i_qWith direct-axis current sampled value i_d, calculate quadrature-axis voltage given value u_qWith direct-axis voltage given value u_d:

In formula, L_dFor motor stator d-axis inductance, L_qFor motor stator axis inductor amount, p_nFor motor number of pole-pairs, ω_mFor motor Mechanical angular speed, ψ_fFor motor permanent magnet magnetic linkage, R is motor stator resistance, i '_dFor the benefit for the direct-axis current that weak magnetic module generates Value is repaid, t is the time；

Step 7, the DC voltage u in conjunction with internal permanent magnet synchronous motor_dc, quadrature-axis voltage given value u_qIt is given with direct-axis voltage Value u_dJudge whether weak magnetic module starts, it may be assumed that

WhenWhen, weak magnetic module does not start, at this time i '_d=0；

WhenWhen, weak magnetic module starting, at this time by u_dcAs the defeated of single-neuron adaptive PID controller Enter given value, simultaneously willAs the actual sample value of single-neuron adaptive PID controller, it is sent into single mind together Through first self-adaptive PID controller, then single-neuron adaptive PID controller exports to obtain the offset i ' of direct-axis current_d；

Step 8, by quadrature-axis voltage given value u_qWith direct-axis voltage given value u_dRotational coordinates inverse transformation is synchronized, is obtained static The phase voltage u of α axis under coordinate system_αWith the phase voltage u of β axis_β；

Step 9, the phase voltage u by α axis_αWith the phase voltage u of β axis_βBy space vector pulse width modulation, three-phase inverter is controlled, and After carrying out IGBT rectification, it is conveyed to internal permanent magnet synchronous motor, is controlled it；

The process of step 10, repetitive cycling step 2-9, to realize the real-time control to internal permanent magnet synchronous motor.

2. internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method according to claim 1, special Sign is, in step 1, constructed single-neuron adaptive PID controller are as follows:

Wherein:

w_i(k) i-th of the weight coefficient calculated for kth time:

w₁(k)=w₁(k-1)+η_P[y^*-y(k)]u(k)x₁(k)

w₂(k)=w₂(k-1)+η_I[y^*-y(k)]u(k)x₂(k)

w₃(k)=w₃(k-1)+η_D[y^*-y(k)]u(k)x₃(k)

x_i(k) i-th of error originated from input amount of controller calculated for kth time:

x₁(k)=y^*-y(k)

x₂(k)=[y^*-y(k)]-[y^*-y(k-1)]

x₃(k)=[y^*-y(k)]-2[y^*-y(k-1)]+[y^*-y(k-2)]

In the above formulas, u (k) is the controller output quantity that kth time calculates, and u (k-1) is that the controller of (k-1) secondary calculating is defeated Output；w₁(k) the 1st weight coefficient calculated for kth time, w₂(k) the 2nd weight coefficient calculated for kth time, w₃It (k) is kth time The 3rd weight coefficient calculated；w₁It (k-1) is the 1st weight coefficient of (k-1) secondary calculating, w₂It (k-1) is (k-1) secondary calculating 2nd weight coefficient, w₃It (k-1) is the 3rd weight coefficient of (k-1) secondary calculating；x₁(k) controller the 1st calculated for kth time Error originated from input amount, x₂(k) the 2nd error originated from input amount of controller calculated for kth time, x₃(k) controller the 3rd calculated for kth time A error originated from input amount；y^*For the input given value of controller, y (k) is the actual sample value for the controller that kth time calculates, y (k- It 1) is the actual sample value of the controller of (k-1) secondary calculating, y (k-2) is the actual samples of the controller of (k-2) secondary calculating Value；η_PFor ratio learning rate, η_ITo integrate learning rate, η_DFor differential learning rate；K is proportionality coefficient, K > 0.

3. internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method according to claim 2, special Sign is, in step 1, the parameter for initializing single-neuron adaptive PID controller includes: given Proportional coefficient K, ratio study speed Rate η_P, integral learning rate η_PWith differential learning rate η_D；Simultaneously when first calculating is k=1, u (k-1)=0, w is enabled₁(k-1) =0, w₂(k-1)=0, w₃(k-1)=0, y (k-1)=0, y (k-2)=0 and i '_d=0.

4. internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method according to claim 1, special Sign is, in step 2, the sampled value ω of revolving speed are as follows:

In formula, ω_mFor mechanical angular speed.

Technical field

The present invention relates to internal permanent magnet synchronous motor technical fields, and in particular to a kind of internal permanent magnet synchronous motor list Single nerve PID field weakening control method.

Background technique

Internal permanent magnet synchronous motor (IPMSM), energy density is high, small in size, and response is fast, runs in wide speed regulating range High-efficient, driveability is ideal, at present in new-energy automobile using relatively broad.Turn is played in order to improve IPMSM Square and speed adjustable range generally use unitary current maximum torque control (MTPA) mode in permanent torque area, and adopt in invariable power area With weak magnetic control mode.

Existing IPMSM weak magnetic control mode mostly uses conventional PID controller to realize.However, due to PID controller itself It is strong to parameter of electric machine dependence, it is weak to the dynamic change adaptability of motor, therefore energy is adjusted for non-linear, uncertain system Power is weak, and control effect is poor.Conventional PID controller needs comparative example, integral and differential to be adjusted to there is preferable control effect It is whole, and there is the relationship for cooperating and mutually restricting between three, it is extremely difficult to find out optimal parameter combination, and by There are complicated non-linear and time-varying characteristics in IPMSM, along with the variable element as caused by influence of the motor with environment itself, become The uncertainties such as structure are saved, so that conventional PID controller parameter tuning is difficult, online self-tuning is especially unable to, it is made to be difficult to meet The requirement of closed optimized control.In addition, meter of the conventional voltage outer loop feedback mode frequently with integrator as d axis compensation offset Module is calculated, but the effect is unsatisfactory for weak magnetic, especially with the raising of motor speed, integral constant is a non-constant value.

Summary of the invention

To be solved by this invention is the bad problem of control effect of existing internal permanent magnet synchronous motor, provides one kind Internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method.

To solve the above problems, the present invention is achieved by the following technical solutions:

Internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method, comprises the following steps that

Step 1, building single-neuron adaptive PID controller, and to the parameter of single-neuron adaptive PID controller into Row initialization；

Step 2, the mechanical angular velocity omega for detecting internal permanent magnet synchronous motor_m, and after being sampled to it, obtain revolving speed Sampled value ω；

Step 3, by the rotary speed setting value ω of setting^*Its input given value as single-neuron adaptive PID controller, Simultaneously using step 2 gained revolving speed sampled value ω as the actual sample value of single-neuron adaptive PID controller, it is sent into together single Neuron adaptive PID control device, then single-neuron adaptive PID controller exports to obtain torque reference value T^*；

Step 4, by torque reference value T^*Maximum torque per ampere control is carried out, quadrature axis current given value is obtainedAnd d-axis Given value of current value

Step 5, the three-phase current for detecting internal permanent magnet synchronous motor, and successively by static coordinate transformation and synchronous rotation After turning coordinate transform, the quadrature axis current sampled value i under synchronous rotating frame is obtained_qWith direct-axis current sampled value i_d；

Step 6, in conjunction with the resulting quadrature axis current given value of step 4With direct-axis current given valueAnd obtained by step 5 Quadrature axis current sampled value i_qWith direct-axis current sampled value i_d, calculate quadrature-axis voltage given value u_qWith direct-axis voltage given value u_d:

In formula, L_dFor motor stator d-axis inductance, L_qFor motor stator axis inductor amount, p_nFor motor number of pole-pairs, ω_m For electromechanics angular speed, ψ_fFor motor permanent magnet magnetic linkage, R is motor stator resistance, i '_dThe d-axis electricity generated for weak magnetic module The offset of stream, t are the time；

Step 7, the DC voltage u in conjunction with internal permanent magnet synchronous motor_dc, quadrature-axis voltage given value u_qAnd direct-axis voltage Given value u_dJudge whether weak magnetic module starts, it may be assumed that

WhenWhen, weak magnetic module does not start, at this time i '_d=0；

WhenWhen, weak magnetic module starting, at this time by u_dcAs single neuron self-adaptation PID control The input given value of device simultaneously willAs the actual sample value of single-neuron adaptive PID controller, send together Enter single-neuron adaptive PID controller, then single-neuron adaptive PID controller exports to obtain the offset of direct-axis current i′_d；

Step 8, by quadrature-axis voltage given value u_qWith direct-axis voltage given value u_dRotational coordinates inverse transformation is synchronized, is obtained The phase voltage u of α axis under rest frame_dWith the phase voltage u of β axis_β；

Step 9, the phase voltage u by α axis_αWith the phase voltage u of β axis_βBy space vector pulse width modulation, three-phase inversion is controlled Device, and after carrying out IGBT rectification, it is conveyed to internal permanent magnet synchronous motor, is controlled it；

The process of step 10, repetitive cycling step 2-9, to realize the real-time control to internal permanent magnet synchronous motor.

In above-mentioned steps 1, constructed single-neuron adaptive PID controller are as follows:

Wherein:

w_i(k) i-th of the weight coefficient calculated for kth time:

w₁(k)=w₁(k-1)+η_P[y^*-y(k)]u(k)x₁(k)

w₂(k)=w₂(k-1)+η_I[y^*-y(k)]u(k)x₂(k)

w₃(k)=w₃(k-1)+η_D[y^*-y(k)]u(k)x₃(k)

x_i(k) i-th of error originated from input amount of controller calculated for kth time:

x₁(k)=y^*-y(k)

x₂(k)=[y^*-y(k)]-[y^*-y(k-1)]

x₃(k)=[y^*-y(k)]-2[y^*-y(k-1)]+[y^*-y(k-2)]

In the above formulas, u (k) is the controller output quantity that kth time calculates, and u (k-1) is the control of (k-1) secondary calculating Device output quantity；w₁(k) the 1st weight coefficient calculated for kth time, w₂(k) the 2nd weight coefficient calculated for kth time, w₃It (k) is the 3rd weight coefficient of k calculating；w₁It (k-1) is the 1st weight coefficient of (k-1) secondary calculating, w₂It (k-1) is (k-1) secondary meter The 2nd weight coefficient calculated, w₃It (k-1) is the 3rd weight coefficient of (k-1) secondary calculating；x₁(k) controller calculated for kth time the 1 error originated from input amount, x₂(k) the 2nd error originated from input amount of controller calculated for kth time, x₃(k) controller calculated for kth time 3rd error originated from input amount；y^*For the input given value of controller, y (k) is the actual sample value for the controller that kth time calculates, y It (k-1) is the actual sample value of the controller of (k-1) secondary calculating, y (k-2) is the reality of the controller of (k-2) secondary calculating Sampled value；η_PFor ratio learning rate, η_ITo integrate learning rate, η_DFor differential learning rate；K is proportionality coefficient, K > 0.

In above-mentioned steps 1, the parameter for initializing single-neuron adaptive PID controller includes: given Proportional coefficient K, ratio Example learning rate η_P, integral learning rate η_PWith differential learning rate η_D；Simultaneously it is first calculate be k=1 when, enable u (k-1)= 0, w₁(k-1)=0, w₂(k-1)=0, w₃(k-1)=0, y (k-1)=0, y (k-2)=0 and i '_d=0.

In above-mentioned steps 2, the sampled value ω of revolving speed are as follows:

In formula, ω_mFor mechanical angular speed.

Compared with prior art, the present invention has a characteristic that

1, on the basis of weak magnetic controls, der Geschwindigkeitkreis is adjusted by single-neuron adaptive PID controller, is taken in concurrently The respective advantage of neural network and PID makes control system have better dynamic response capability and robustness, makes to control precision Higher, torque pulsation is smaller, effectively improves the utilization rate of DC bus-bar voltage；

2, by replacing regular integral device using single-neuron adaptive PID controller, so that controller parameters can With online real-time Self-tuning System, for motor operation and environmental change, more adaptability；In addition, in conjunction with Current Decoupling feedforward compensation Mode effectively increases control system performance and control precision so that control system is full decoupled.

Detailed description of the invention

Fig. 1 is the schematic diagram of internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method.

Fig. 2 is single-neuron adaptive PID controller schematic diagram.

Fig. 3 is Current Decoupling feedforward compensation schematic diagram.

Fig. 4 is the Voltage Feedback control principle drawing based on single-neuron adaptive PID controller.

Specific embodiment

To make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with specific example, to the present invention It is further described.

A kind of internal permanent magnet synchronous motor single neuron self-adaptive PID controller field weakening control method, as shown in Figure 1, it is specific It comprises the following steps that

Step 1, single-neuron adaptive PID controller is constructed, as shown in Figure 2:

y^*Expression system inputs given value, and y (k) indicates the actual sample value that system kth time calculates, the two error e_y(k) it passes through State converter is crossed, is exported as three intermediate state variable x₁、x₂、x₃, then using these three intermediate state variables as neuron 3 input quantities, value is respectively as follows:

e_y(k)=y^*-y(k)

x₁(k)=e_y(k)

x₂(k)=e_y(k)-e_y(k-1)

x₃(k)=e_y(k)-2e_y(k-1)+e_y(k-2)

Wherein, x_i(k) indicate i-th of x that kth time calculates value (i=1,2,3；K=1,2,3...), e_y(k) kth is indicated The e of secondary calculating_yIt is worth (k=1,2,3...).

According to there is supervision Hebb learning algorithm, the output of single neuron self-adaptation PID control can be obtained are as follows:

Wherein, K is proportionality coefficient (K > 0), η_P、η_I、η_DRespectively ratio, integral, differential learning rate.w_i(k)(i =1,2,3；K=1,2,3...) indicate weight coefficient,

Step 2, single-neuron adaptive PID controller parameters are initialized, and setting ratio COEFFICIENT K, ratio learn Rate η_P, integral learning rate η_IWith differential learning rate η_D；

When carrying out the initialization of single-neuron adaptive PID controller parameters, because of k=1,2,3..., so working as k When=1, u (k-1)=0, w₁(k-1)=0, w₂(k-1)=0, w₃(k-1)=0, e_y(k-1)=0, e_y(k-2)=0.In addition, just I ' when secondary calculating_d=0.These parameters are set in controller software programming phases, are inputted not as the parameter of controller itself.

Step 3, electromechanics angular velocity omega is detected by sensor_mObtain motor speed sampled value ω；

Step 4, motor speed given value ω is given^*, input given value y as single-neuron adaptive PID controller^*, Using each moment motor speed sampled value ω as the actual sample value y of single-neuron adaptive PID controller, by single nerve First PID controller, then the controller output quantity u (k) that the kth time of single-neuron adaptive PID controller output calculates, i.e. conduct The torque reference value T at each moment^*；

Step 5, by torque reference value T^*It is controlled by MTPA (torque capacity electric current ratio), using equation, passes through T^*First ask Quadrature axis current given value outThen it utilizesFind out direct-axis current given value

MTPA calculation formula are as follows:

Wherein, p_nIndicate motor number of pole-pairs, ψ_fIndicate permanent magnet flux linkage, L_dIndicate stator d-axis (d axis) inductance, L_qTable Show stator quadrature axis (q axis) inductance；Wherein d axis and q axis belong to the coordinate variable of synchronous rotating frame；

Step 6, as shown in figure 3, by direct-axis current given valueWith direct-axis current sampled value i_dIt compares, and combines weak magnetic Offset i ' of the module to direct-axis current_d, by PI current regulator, obtain the voltage u generated by resistance and differential part₁；

Wherein, i "_dIndicate that direct-axis current calculated value, R indicate stator resistance；The offset i ' of direct-axis current_dBy weak magnetic module It generates, but weak magnetic module does not start every time, needs to judge whether weak magnetic voltage reaches weak magnetic condition, that is, works asRight, weak magnetic module generates the offset i ' of direct-axis current_d, otherwise weak magnetic module does not generate direct-axis current Offset i '_d, i.e. i '_d=0；It is not reach requirement weak magnetic condition, i.e. i ' certainly when recycling first time_d=0.

Step 7, by quadrature axis current sampled value i_qWith electromechanics angular velocity omega_m, obtain in the d shaft voltage of feedforward compensation by q Component of voltage u caused by shaft current₂；

ω_e=p_nω_m

u₂=-ω_eL_qi_q

Wherein, ω_eIndicate angular rate；

Step 8, in conjunction with two component voltage u₁And u₂, obtain d shaft voltage given value u_d；

Step 9, as shown in figure 3, by quadrature axis current given valueWith quadrature axis current sampled value i_qComparison, pass through PI electric current Adjuster obtains the voltage u generated by resistance and differential part₁′；

Wherein, i "_qIndicate quadrature axis current calculated value；

Step 10, by direct-axis current sampled value i_dWith electromechanics angular velocity omega_m, obtain in the q shaft voltage of feedforward compensation by Component of voltage u caused by d shaft current₂′；

ω_e=p_nω_m

u₂'=ω_e(L_di_d+ψ_f)

Step 11, in conjunction with two component voltage u₁' and u₂', obtain q shaft voltage given value u_q

Step 12, pass through weak magnetic conditionJudge whether weak magnetic module starts.IfThen electric machine phase current and phase current are not up to maximum value, and weak magnetic module does not work, i '_d=0, it passes It is delivered to the step 6 of next iteration, wherein u_dcIndicate DC voltage；

Step 13, ifThen electric machine phase current and phase current reach maximum value, and weak magnetic module opens It is dynamic.It, will after the starting of weak magnetic moduleActual sample value y, u as input single-neuron adaptive PID controller_dc Given value y as input single-neuron adaptive PID controller^*, by single-neuron adaptive PID controller, mended Repay the current component i ' of d axis_d, it is transmitted to the step 6 of next iteration；As shown in Figure 4；

Step 14, by d shaft voltage given value u_dWith q shaft voltage given value u_q, (synchronously rotating reference frame is converted by anti-Park Transformation), obtain 2 voltage signal u under rest frame alpha-beta_αAnd u_β

Step 15, voltage signal u_αAnd u_β(space vector pulse width modulation) is modulated by SVPWM, controls three-phase inverter, and It is rectified in conjunction with IGBT (insulated gate bipolar transistor), is conveyed to IPMSM (internal permanent magnet synchronous motor)；

Step 16, sensor detects electromechanics angular velocity omega_m, be transmitted to next iteration step 3,7,10；

Step 17, sensor detects motor three-phase current i_a、i_b、i_c, first passing through Clark coordinate transform, (static coordinate becomes Change) obtain the current value i under rest frame_αAnd i_β, synchronized using Park coordinate transform (synchronous rotating angle) D shaft current sampled value i under rotating coordinate system_dWith q shaft current sampled value i_q, and it is transmitted to the step of next iteration respectively In rapid 6 and 9；

Repetitive cycling above-mentioned steps 3-17, is achieved in the real-time control to internal permanent magnet synchronous motor.

The present invention is using non-linear mapping capability possessed by neural network, learning ability and summarizes Generalization Ability, in conjunction with Regulatory PID control is theoretical, i.e. Neural network PID, by taking in the advantage of the two concurrently, so that controlled system is had adaptivity, can oneself Dynamic on-line control controller parameter, adapts to the variation of controlled process, so that control performance and reliability are improved, particularly with time-varying Object and nonlinear system have preferable control effect.The self-adaptive PID based on single neuron used herein is as nerve One of the method that network is combined with PID control, with simple, the easy to accomplish feature of control mode.

It can be not only restricted to the variation of the parameter of electric machine, automatic on-line adjusts control parameter, overshoot and torque is effectively reduced Pulsation, the dynamic response capability and stability of Lifting Control System.On the basis of Realization of pulse width modulation based on voltage space vector (SVPWM) On, speed error, and automatic on-line controller parameters setting are controlled using single-neuron adaptive PID controller.It is based on simultaneously The weak magnetic of Voltage Feedback mode controls, and in conjunction with Current Decoupling feedforward compensation mode, replaces conventional product with single neuron PID controller Divide device, effectively control torque pulsation and overshoot, optimizes weak magnetic control effect.

By building der Geschwindigkeitkreis, three closed loop internal permanent magnet synchronous motor Controlling models of electric current loop and Voltage loop, in electricity On the basis of pressing space vector pulse width modulation (SVPWM), using single-neuron adaptive PID controller, and torque capacity is combined Voltage Feedback weak magnetic control of the electric current than control (MTPA) and based on Current Decoupling feedforward compensation, is not only restricted to the parameter of electric machine Dynamic change, can on-line tuning controller parameters, effectively reduce overshoot and torque pulsation, dynamic response is fast, robust Property is strong.

It should be noted that although the above embodiment of the present invention be it is illustrative, this be not be to the present invention Limitation, therefore the invention is not limited in above-mentioned specific embodiment.Without departing from the principles of the present invention, all The other embodiment that those skilled in the art obtain under the inspiration of the present invention is accordingly to be regarded as within protection of the invention.