Speed control method based on composite observer and self-adaptive super-twisted sliding mode

Abstract

The invention provides a speed control method based on a composite observer and a self-adaptive super-twisted sliding mode, and the method comprises the following steps: firstly, introducing an extension variable related to speed measurement, and building a PMLSM system extension dynamic model; secondly, designing a composite observer composed of a model reference adaptive observer and a generalized proportional-integral observer based on a PMLSM expansion dynamic model according to unknown motor parameters, internal disturbance, external load change and other factors; thirdly, aiming at the PMLSM extended dynamic model after the lumped disturbance is compensated, designing an outer ring AST speed control law based on a composite observer so as to enhance the robustness of the system; and finally, adopting an inner ring d-q axis current vector control method based on PI to realize that the stator current of the motor tracks the given value of the stator current. According to the method, actual complex working conditions such as unknown rotor mass, internal and external disturbance and speed measurement noise can be comprehensively considered, so that the speed control of the PMLSM driving system has good dynamic performance and steady-state performance.

Description

Technical Field

[0001] The invention belongs to the field of permanent magnet synchronous linear motor control and relates to a speed control method based on a composite observer and an adaptive super-twisted sliding mode. Background Art

[0002] PMLSM has the advantages of fast response speed, high precision, large thrust, and low noise. It is widely used in precision CNC machine tools, lithography machines, high-speed logistics equipment and other fields. In order to achieve efficient precision processing and production, the control requirements for the PMLSM drive system are fast response, high precision, and strong robustness. When designing the control strategy of the PMLSM drive system, it is necessary to fully consider the various factors faced in its operation, such as unknown mover mass, unknown viscous friction coefficient, internal interference such as nonlinear friction, thrust fluctuations, external interference such as load changes, and speed measurement noise pollution. For this reason, there are currently a variety of observer-based robust control strategies. Although the above methods can overcome the influence of parameter uncertainty or resist internal and external interference, they still cannot meet the high-precision requirements of the PMLSM drive system. The specific defects are as follows:

[0003] (1) For the PMLSM drive system, the existing observer-based control strategy only considers parameter uncertainty or external disturbance, but does not fully consider various actual complex working conditions such as unknown rotor mass, internal disturbance and external disturbance. In addition, it does not consider the speed measurement noise problem in speed negative feedback control.

[0004] (2) At present, the control strategy of PMLSM drive system only designs the mover mass observer or disturbance observer, but rarely considers the design of observers that can simultaneously estimate the mover mass, the actual mover speed and the lumped disturbance;

[0005] (3) For the PMLSM drive system, the existing observer-based control strategy usually requires not only measuring the speed signal, but also measuring the current or voltage signal. Each signal measurement inevitably introduces measurement noise, which is bound to affect the estimation accuracy of the observer and thus the control accuracy of the system.

[0006] (4) At present, the commonly used observers in the control strategy of PMLSM drive system are sliding mode observer (SMO), Kalman filter observer (EKFO), extended state observer (ESO) and disturbance observer (DOB). However, SMO has jitter and overshoot, EKF has high operation cost and large calculation amount, ESO is only used to estimate constant or slowly varying physical quantities, and DOB needs to compromise between system bandwidth and noise suppression. Summary of the invention

[0007] The present invention proposes a speed control method based on a composite observer and an adaptive super-twisted sliding mode, which comprehensively considers a series of adverse effects in the operation of the motor and designs a corresponding method. It can not only ensure the high-precision speed control of the PMLSM drive system, but also has good dynamic and steady-state performance, and the stability is strictly proved mathematically, and it has strong practicality and feasibility.

[0008] The present invention comprises the following steps:

[0009] Step 1: Based on the dynamic equation of the hidden-pole PMLSM in the dq-axis rotating coordinate system and the speed measurement noise condition, the PMLSM extended dynamic model is established.

[0010] The dynamic equation of the hidden-pole PMLSM in the dq-axis rotating coordinate system is:

[0011]

[0012] In the formula, v m is the actual speed of the PMLSM mover without noise; i q is the q-axis stator current; M is the mover mass; B is the viscous friction coefficient; F e is the electromagnetic thrust; F f is the nonlinear friction force; F r is thrust fluctuation; F d is the external load; n p is the number of magnetic pole pairs; ψ f is the permanent magnet flux of constant value; τ is the pole pitch; t is the current time;

[0013] Introducing the extended variable v related to speed measurement, the PMLSM extended model is:

[0014]

[0015] Where v is the value of integrating the speed measurement signal containing noise; η is the speed measurement noise; k f is the electromagnetic thrust coefficient, d is the system lumped disturbance, k f The expressions of and d are:

[0016]

[0017] In the formula, is the q-axis current reference value;

[0018] Because B and F f 、F r 、F d If , M are unknown, then d is also unknown.

[0019] Step 2: Design a composite observer based on the PMLSM extended dynamic model in step 1, wherein the composite observer consists of a model reference adaptive observer and a generalized proportional integral observer;

[0020] Design of a model-referenced adaptive observer for obtaining PMLSM mover mass estimates

[0021] The estimated mass of the mover for:

[0022]

[0023] Where, T s is the sampling period; k represents the kth sampling moment; β is the positive real gain; is the estimated mass of the mover at time k; is the difference between the q-axis current reference values ​​at the two sampling moments (k-2) and (k-1), that is, e 0 (k) is the difference between the actual value of v at time k and the prior estimate, that is, Among them, the prior estimate of v(k) Expressed as

[0024] Step 3: Design a generalized proportional integral observer to estimate the PMLSM lumped disturbance and the actual speed of the mover;

[0025] The generalized proportional integral observer is:

[0026]

[0027] In the formula, is the estimated value of v; v m An estimated value of is the estimated value of d; for The estimated value of i >0 (i=0,1,2,3) is the generalized proportional-integral observer gain; Estimate the mass of the mover.

[0028] Step 4: Based on the model reference adaptive observer in step 2 and the generalized proportional integral observer in step 3, design the outer loop AST speed control law based on the composite observer;

[0029] The q-axis current reference value is:

[0030]

[0031] Where, u is the outer loop AST speed control law; is the estimated value of the lumped disturbance d, that is,

[0032] The total disturbance to the system If compensation is performed, the extended system model of PMLSM can be expressed as:

[0033]

[0034] In the formula, is the estimated value of v; v m The estimated value of

[0035] The outer loop AST speed control law based on the composite observer is:

[0036]

[0037] In the formula, is the PMLSM reference speed; s is the sliding surface; u 1 is the reference speed change rate; u 2 is the super-twisted sliding mode control law; sign(s) is the sign function;

[0038] l is the super-twisted sliding mode gain, and its adaptive law is:

[0039]

[0040] In the formula, is a positive real number; the initial value of l(0) is a positive real number; the sliding surface s is

[0041] Step 5: Design the inner loop dq axis current control law using PI-based vector control;

[0042] The inner loop dq axis current control law is:

[0043]

[0044] In the formula, u d * is the d-axis reference voltage; u q * is the q-axis reference voltage; i d * is the d-axis reference current, and its value is 0; k pd is the d-axis proportional gain; k pq is the q-axis proportional gain; k id is the d-axis integral gain; k iq is the q-axis integral gain; k pd , k pq , kid , k iq are all positive real numbers.

[0045] The beneficial effects of the present invention are:

[0046] (1) The present invention comprehensively considers actual complex working conditions such as unknown rotor mass, internal disturbance, external disturbance and speed measurement noise, integrates relevant uncertain parameters, various internal and external disturbances and q-axis current into lumped disturbance, establishes an extended dynamic model of PMLSM, and provides convenience for the design of composite observer and AST control law;

[0047] (2) The present invention designs a composite observer composed of two estimators connected in series, which can simultaneously estimate the mover mass, the actual mover speed and the lumped disturbance, so that the control strategy based on the composite observer can resist the adverse effects of various parameter uncertainties and internal and external disturbances faced by the PMLSM drive system;

[0048] (3) The composite observer of the present invention only uses the speed measurement signal and does not need to measure the current or voltage signal, which not only reduces the system hardware installation cost, but also avoids the adverse effects of current measurement noise, thereby improving the estimation accuracy of the composite observer and further improving the control accuracy of the PMLSM drive system;

[0049] (4) The model reference adaptive observer designed by the present invention, due to the use of a model reference adaptive mechanism, can estimate the constant or time-varying quantum mass more quickly and accurately compared with other existing estimation methods such as the Coulomb-viscous friction model and the extended Kalman filter, and can completely eliminate the adverse effects of lumped disturbances;

[0050] (5) The generalized proportional integral observer designed by the present invention, due to the introduction of the speed measurement integral variable polluted by noise, realizes the complete decoupling of the generalized proportional integral observer gain selection and the speed measurement noise, thereby eliminating the influence of the speed measurement noise on the estimation accuracy of the generalized proportional integral observer, and by reasonably selecting the generalized proportional integral observer gain, it is possible to reduce the estimation error and improve the estimation accuracy; the generalized proportional integral observer can simultaneously estimate the PMLSM time-varying lumped disturbance and the actual speed of the mover. Compared with the sliding mode observer, the disturbance observer and the extended state observer, the present invention does not have estimation jitter, so there is no need to compromise between system bandwidth and noise suppression;

[0051] (6) The present invention compensates for the lumped disturbance estimated by the composite observer. The designed AST control law can not only completely resist the lumped disturbance, but also enhance the robustness against the change of the mover mass. In addition, the AST control strategy based on the composite observer designed by the present invention can also weaken the system jitter and effectively improve the speed control accuracy of the PMLSM drive system.

[0052] Figure Description

[0053] Figure 1 is a control block diagram of the PMLSM drive system of this embodiment;

[0054] Figure 2 is a block diagram of an equivalent nonlinear feedback system of this embodiment;

[0055] Figure 3 is a system control flow chart of this embodiment;

[0056] Figure 4 This is a schematic diagram comparing the speed and speed error under the three outer loop speed control strategies in Experiment 1 (reference speed change);

[0057] Figure 5 This is a schematic diagram of the motor load changes in Experiment 2;

[0058] Figure 6 This is a schematic diagram comparing the speed and speed error under the three outer loop speed control strategies in Experiment 2 (under load variation);

[0059] Figure 7 This is a schematic diagram of the electromagnetic thrust comparison under three outer loop speed control strategies in Experiment 2 (under load variation);

[0060] Figure 8 This is a schematic diagram of speed error comparison under three outer loop speed control strategies in Experiment 3 (when the mover mass changes);

[0061] Fig. 9 This is a schematic diagram of the comparison of q-axis current under three outer loop speed control strategies in Experiment 3 (when the mover mass changes);

[0062] Fig.10 This is a schematic diagram of speed error comparison under three outer loop speed control strategies in Experiment 4 (when the viscous friction coefficient changes);

[0063] Fig.11 This is a schematic diagram comparing the q-axis current under the three outer loop speed control strategies in Experiment 4 (when the viscous friction coefficient changes). DETAILED DESCRIPTION

[0064] The invention is described clearly and completely below in conjunction with the figures and implementation methods.

[0065] Example

[0066] Taking the hidden-pole PMLSM as the research object, this embodiment proposes a speed control method based on a composite observer and an adaptive super-twisted sliding mode to ensure that the PMLSM speed can quickly and accurately track a given value, which specifically includes the following steps:

[0067] Step 1: Based on the PMLSM dynamic equation, considering the speed measurement noise condition, establish its extended dynamic model;

[0068] The motion equation and thrust equation of the hidden pole PMLSM in the dq rotating coordinate system are:

[0069]

[0070] Formula (1) can be rewritten as

[0071]

[0072] In the formula,

[0073]

[0074] Considering the existence of speed measurement noise in the PMLSM feedback control system, a new extended variable is introduced and defined as:

[0075]

[0076] Combining equation (2) and equation (3), the PMLSM extended system model is as follows:

[0077]

[0078] Considering that the actual value of M is unknown, and considering B and F f 、F r With F d If both are unknown, then d is also unknown. Therefore, the present invention designs a composite observer, whose block diagram is as follows: Figure 1 As shown in the dashed box; the composite observer consists of a model reference adaptive observer and a generalized proportional integral observer, where the model reference adaptive observer is used to estimate the mover mass M, and the generalized proportional integral observer is used to estimate the lumped disturbance d and the actual mover velocity v m .

[0079] Step 2: Based on the PMLSM extended dynamic model described in step 1, a model reference adaptive observer in the composite observer is designed, which is used to estimate the PMLSM mover mass.

[0080] The model reference adaptive observer design process is as follows:

[0081] Discretize equation (4) to get:

[0082]

[0083] Formula (5) can also be expressed as:

[0084]

[0085] Substituting formula (6) into formula (5) yields:

[0086]

[0087] Since the system sampling frequency is very high, it can be considered that the following equations hold true at adjacent sampling times:

[0088]

[0089] Substituting equation (8) into equation (7), we obtain:

[0090]

[0091] In the formula,

[0092]

[0093] According to formula (9), the reference model of the model reference adaptive observer is established as:

[0094]

[0095] According to formula (11), the priori and a posteriori adjustable models of the model reference adaptive observer are established as:

[0096]

[0097] In the formula, is the estimated value of a; and are the prior and posterior estimates of v(k), respectively.

[0098] Accordingly, the prior estimation error e of the model reference adaptive observer is 0 (k) and the posterior estimation error e(k) are respectively defined as:

[0099]

[0100] design The adaptive law is:

[0101]

[0102] Thus, by The estimated mass of the mover is:

[0103]

[0104] From formula (17), we can see that the valuation Convergence depends on valuation The model reference adaptive observer in this embodiment is asymptotically stable, which can ensure the estimation Converge to the true value a, thereby ensuring the valuation Converges to the true value M, and the proof process of the former is as follows:

[0105] According to the A-type algorithm of Landau discrete parameter recursive mechanism, The adaptive laws are:

[0106]

[0107] In the formula, It's about e 0 (k) or the correction function of e(k).

[0108] To avoid inaccurate estimation due to delayed estimation, we first determine φ[e(k)] and then determine φ[e 0 (k)], and finally we get

[0109] Therefore, substituting (19) into (15) yields

[0110]

[0111] This embodiment constructs a nonlinear feedback system that is completely equivalent to equation (20). The system block diagram is as follows: Figure 2 As shown, G e (z)=1, the correction function is β is a positive real gain.

[0112] Because the linear link G of the forward path e (z)=1 is strictly positive and real, and the nonlinear link of the feedback channel satisfies Popov integral inequality, that is:

[0113]

[0114] Therefore, according to the hyperstability theory, Figure 2 The nonlinear feedback system shown is asymptotically stable. The designed φ[e(k)] can ensure that the system output e(k) converges to 0, that is, the estimation error of the model reference adaptive observer converges to 0, thereby ensuring Converges to the true value a.

[0115] In practical applications, we should use formula (18) to estimate To do this, we need to obtain φ[e 0 (k)]. After derivation, it can be obtained as follows:

[0116]

[0117] Substituting equation (22) into equation (18) yields the estimate shown in equation (16): Reconsider The estimated value shown in formula (17) is The above analysis applies the superstability theory to ensure the valuation The asymptotic convergence of .

[0118] Step 3: Based on the PMLSM extended dynamic model described in step 1, a generalized proportional integral observer in the composite observer is designed to estimate the PMLSM lumped disturbance and the actual speed of the mover.

[0119] Consider B and F f 、F r With F d If both are unknown, then d is also unknown. Considering the existence of speed measurement noise in the PMLSM feedback control system, the generalized proportional integral observer used to estimate the lumped disturbance and the actual speed of the mover is:

[0120]

[0121] Choose an appropriate gain parameter a i (i=0,1,2,3) can guarantee the stability and convergence of the generalized proportional-integral observer, and the proof process is as follows.

[0122] The estimation error is defined as:

[0123]

[0124] Then the estimated error equation of the generalized proportional integral observer is:

[0125]

[0126] In the formula,

[0127]

[0128] Therefore, the characteristic polynomial of the generalized proportional integral observer estimation error equation is:

[0129] △(λ)=λ 4 +a 3 λ 3 +a 2 λ 2 +a 1 λ+a 0 (27)

[0130] Select gain a i (i=0,1,2,3) can make A a Hurwitz matrix, thus ensuring the stability of the generalized proportional integral observer. At the same time, the designed generalized proportional integral observer has convergence, and its analysis process is as follows:

[0131] Since the system shown in equation (25) is stable, therefore, for any given symmetric positive definite matrix Q 1 , there must be a unique symmetric positive definite matrix P 1 The following Lyapunov equation is satisfied:

[0132] A T P 1 +P 1 A=-Q 1 (28)

[0133] Choose the Lyapunov function as:

[0134] V(E)=E T P 1 E / 2 (29)

[0135] The derivative of formula (29) is:

[0136]

[0137] In the formula, γ≥sup t ||η d (t)||, Represents the smallest eigenvalue of a matrix.

[0138] From formula (30), we can see that if the estimation error of the generalized proportional integral observer satisfies the following inequality:

[0139]

[0140] but According to Lyapunov stability theory, ||E|| will decrease until Then the generalized proportional integral observer estimation error trajectory will converge to a bounded domain:

[0141]

[0142] It can be seen from formula (32) that under the premise of ensuring that A is a Hurwitz matrix, the gain a is selected i (i=0,1,2,3) so that λ min (AA T ) increases, the error norm ||E|| decreases, and thus the estimation accuracy of the generalized proportional-integral observer is improved.

[0143] Step 4: Design the outer loop AST speed control law based on the composite observer and analyze its stability through Lyapunov stability theory.

[0144] The design idea of ​​the outer loop AST speed control law is as follows: first, the PMLSM extended dynamic model is compensated to completely eliminate the adverse effects of the lumped disturbance term; then, the AST control law is designed for the compensated PMLSM extended dynamic model; finally, the Lyapunov stability theory is applied to prove the stability of the control law; specifically:

[0145] First, consider the composite observer estimated and Design the q-axis current reference value as:

[0146]

[0147] Substitute equation (33) into equation (4) and use and Replace M and v respectively m and the true value of d, then the lumped disturbance The compensated PMLSM extended system model is rewritten as:

[0148]

[0149] Next, for the PMLSM extended dynamic model shown in equation (34), its AST speed control law is designed as follows:

[0150] Select the sliding surface s as:

[0151]

[0152] The AST speed control law is:

[0153]

[0154] The adaptive law of l is:

[0155]

[0156] Finally, the AST stability analysis based on Lyapunov stability theory is as follows:

[0157] The following coordinate transformation is introduced:

[0158]

[0159] The derivative of formula (38) is:

[0160]

[0161] Let ξ=[ξ 1 ξ 2 ] T , where ξ 2 =y2 / l.

[0162] Define the following Lyapunov function:

[0163]

[0164] By taking the derivative of formula (40), we can get:

[0165]

[0166] In the formula,

[0167]

[0168] Substituting the adaptive law of formula (37) into formula (41), we can obtain:

[0169]

[0170] Because λ min (Q 2 )||ξ|| 2 ≤ξ T Q 2 ξ 1 , so formula (43) satisfies the following inequality:

[0171]

[0172] And because λ min (Q 2 )>0, so when l>0, When l≤0, it can be known from formula (37) l will increase to So that In summary Therefore, according to the Lyapunov stability theory, the designed AST speed control law can make the PMLSM drive system run stably.

[0173] Step 5: Design the inner loop dq axis current control law using PI-based vector control;

[0174] The inner loop dq axis PI current control law is:

[0175]

[0176] In order to verify the effectiveness and feasibility of the method proposed in this embodiment, a PMLSM drive system simulation model under three outer-loop control strategies was built on the StarSim semi-physical simulation platform for verification. The three outer-loop control strategies are: ordinary adaptive super-warp sliding mode (AST), adaptive super-warp sliding mode combined with generalized proportional integral observer (GPIO-AST) and this embodiment (CO-AST).

[0177] For fair comparison, the inner loop currents of the three PMLSM drive systems adopt PI-based vector control with exactly the same parameters, and the parameters of the generalized proportional-integral observer in GPIO-AST are exactly the same as those in this embodiment, except that the nominal value of M is used instead of the estimated value of the model reference adaptive observer.

[0178] Set the nonlinear friction and thrust fluctuation to F f +F r =10sin(4πt), and a white noise with a mean of 0 and a standard deviation of 10-5 is used to simulate the speed measurement noise η.

[0179] Other parameters involved in StarSim semi-physical simulation are shown in Table 1.

[0180] Table 1. Hardware-in-the-loop simulation parameter settings

[0181]

[0182] Experiment 1: The load is constant at 0N, the reference speed is initially set to 1m / s, and then suddenly drops to -1m / s at 0.5s.

[0183] Figure 4 (a) The speed comparison of the PMLSM drive system under three outer loop control strategies is given when the reference speed changes; Figure 4 (b) gives the speed error comparison of the PMLSM drive system under three outer loop control strategies when the reference speed changes; Figure 4 It can be seen from (a) and 4(b) that, compared with AST and GPIO-AST, this embodiment can significantly speed up the system response speed, reduce overshoot and weaken speed fluctuation. This embodiment can improve the speed control accuracy and enhance the robustness of the system to internal and external disturbances and interference factors such as measurement noise.

[0184] Experiment 2: Load changes such as Figure 5 As shown, the reference speed is always 1m / s.

[0185] Figure 6 (a) shows the speed comparison of the PMLSM drive system under three outer loop control strategies under load changes. Figure 6(b) The speed error comparison of the PMLSM drive system under three outer loop control strategies under load variation is given; Figure 7 A comparison of the electromagnetic thrust of the PMLSM drive system under three outer loop control strategies under load changes is given.

[0186] Depend on Figure 6 (a) Figure 6 (b) and Figure 7 It can be seen that compared with AST and GPIO-AST, this embodiment can significantly weaken the speed fluctuation and electromagnetic thrust fluctuation during load mutation, and enhance the robustness of the system to interference factors such as load mutation, internal and external disturbances, and measurement noise.

[0187] Experiment 3: The load is constant at 500N, the reference speed is constant at 1m / s, and M suddenly drops to 0.5M at time 0s.

[0188] Figure 8 The speed error comparison of the PMLSM drive system under three outer loop control strategies is given when the mover mass changes. Fig. 9 A comparison of the q-axis current of the PMLSM drive system under three outer loop control strategies when the mover mass changes is given.

[0189] Experiment 4: The load is constant at 500N, the reference speed is constant at 1m / s, and at time 0s, B suddenly increases to 2B.

[0190] Fig.10 The speed error comparison of the PMLSM drive system under three outer loop control strategies is given when the viscous friction coefficient changes. Fig.11 A comparison of the q-axis current of the PMLSM drive system under three outer loop control strategies when the viscous friction coefficient changes is given.

[0191] Depend on Figure 8-11 It can be seen that under the condition of sudden changes in motor parameters, compared with AST and GPIO-AST, the speed fluctuation of this embodiment is smaller and the q-axis current is smoother, which shows that this embodiment can enhance the robustness of the system to interference factors such as sudden changes in motor parameters, internal and external disturbances, and measurement noise.

Claims

1. A speed control method based on a composite observer and an adaptive super-twisted sliding mode, characterized in that: The following steps are involved: Step 1: Based on the PMLSM dynamic equations and speed measurement noise conditions in the dq axis rotating coordinate system, the PMLSM extended dynamic model is established; The dynamic equation of the hidden-pole PMLSM in the dq-axis rotating coordinate system is: In the formula, v m is the actual speed of the PMLSM mover without noise; i q is the q-axis stator current; M is the mover mass; B is the viscous friction coefficient; F e is the electromagnetic thrust; F f is the nonlinear friction force; F r is thrust fluctuation; F d is the external load; n p is the number of magnetic pole pairs; ψ f is the permanent magnet flux of constant value; τ is the pole pitch; t is the current time; After introducing the extended variable v related to speed measurement, the PMLSM extended model is: Where v is the value of integrating the speed measurement signal containing noise; η is the speed measurement noise; k f is the electromagnetic thrust coefficient; d is the system lumped disturbance; k f The expressions of and d are: In the formula, is the q-axis current reference value; Because B and F f 、F r 、F d If , M are unknown, then d is also unknown; Step 2: Based on the PMLSM extended dynamic model in step 1, a composite observer is designed. The composite observer consists of a model reference adaptive observer and a generalized proportional integral observer. The model reference adaptive observer is designed to obtain the PMLSM mover mass estimation. Step 3: Design a generalized proportional integral observer to estimate the PMLSM lumped disturbance and the actual speed of the mover; Step 4: Based on the model reference adaptive observer in step 2 and the generalized proportional integral observer in step 3, design the outer loop AST speed control law based on the composite observer; Step 5: Design the inner loop dq axis current control law using PI-based vector control.

2. The speed control method based on composite observer and adaptive super-twisted sliding mode according to claim 1 is characterized in that: The estimation of the mass of the mover in step 2 for: Where, T s is the sampling period; k represents the kth sampling moment; β is the positive real gain; is the estimated mass of the mover at time k; is the difference between the q-axis current reference values ​​at the two sampling moments (k-2) and (k-1), that is, e 0 (k) is the difference between the actual value of v at time k and the prior estimate, that is, Among them, the prior estimate of v(k) Expressed as 3. The speed control method based on composite observer and adaptive super-twisted sliding mode according to claim 1 is characterized in that: The generalized proportional integral observer in step 3 is: In the formula, is the estimated value of v; v m An estimated value of is the estimated value of d; for The estimated value of i >0 (i=0,1,2,3) is the generalized proportional-integral observer gain; Estimate the mass of the mover.

4. The speed control method based on composite observer and adaptive super-twisted sliding mode according to claim 1 is characterized in that: The q-axis current reference value designed in step 4 is: Where, u is the outer loop AST speed control law; is the estimated value of the lumped disturbance d, that is, The total disturbance to the system If compensation is performed, the extended system model of PMLSM is: In the formula, is the estimated value of v; v m The estimated value of The outer loop AST speed control law based on the composite observer is: In the formula, is the PMLSM reference speed; s is the sliding surface; u1 is the reference speed change rate; u2 is the super-twisted sliding mode control law; sign(s) is the sign function; l is the super-twisted sliding mode gain, and its adaptive law is: In the formula, is a positive real number; the initial value of l(0) is a positive real number; the sliding surface s is 5. The speed control method based on composite observer and adaptive super-twisted sliding mode according to claim 1 is characterized in that: In step 5, the inner loop dq axis current control law is designed as: In the formula, u d * is the d-axis reference voltage; u q * is the q-axis reference voltage; i d * is the d-axis reference current, and its value is 0; k pd is the d-axis proportional gain; k pq is the q-axis proportional gain; k id is the d-axis integral gain; k iq is the q-axis integral gain; k pd , k pq , k id , k iq are all positive real numbers.

Images