FINITE TIME SPEED CONTROL METHOD FOR PERMANENT MAGNET SYNCHRONOUS MOTOR BASED ON FAST INTEGRAL TERMINAL SLIDING MODE AND DISTURBANCE ESTIMATION

Abstract

A finite time speed control method for a permanent magnet synchronous motor (PMSM) based on a fast integral terminal sliding mode and disturbance estimation comprises: firstly, determining a mathematical model of a speed loop of the PMSM under the influence of system parameters uncertainty and unknown load torque; secondly, designing an improved fast integral terminal sliding surface on the basis of the idea of terminal sliding mode control; then, proposing a disturbance estimation method based on an adaptive fuzzy system with respect to the disturbance in a PMSM system; designing a PMSM speed controller on this basis; and finally, completing the concrete implementation of the whole technical solution. The present invention designs the fast integral terminal sliding surface and a sliding mode control law to ensure that a motor speed tracking error converges to zero within finite time and enhances the rapidity of a PMSM speed regulating system.

Claims

1. A finite time speed control method for a permanent magnet synchronous motor (PMSM) based on a fast integral terminal sliding mode and disturbance estimation, comprising the following steps: S1. determining a mathematical model of a speed loop of the PMSM under the influence of system parameters uncertainty and unknown load torque; in a d-q coordinate system, a mathematical model of a speed loop of a non-salient pole permanent magnet synchronous motor is: $\stackrel{.}{\omega }=\frac{K_t}{J}{i}_q-\frac{B}{J}\omega -\frac{T_L}{J}$ where ω is motor speed; i.sub.q represents stator current of q axis; K.sub.t is a torque constant; J represents the moment of inertia; B is a viscous friction coefficient; and T.sub.L represents a load torque; considering the influence of system parameters uncertainty, unknown load torque and current loop tracking error, the mathematical model of the speed loop of the PMSM is: $\stackrel{.}{\omega }=\frac{K_t}{J_n}{i}_q^{*}-\frac{B_o+\Delta B}{J_n+\Delta J}\omega +\left(\frac{K_t}{J}-\frac{K_t}{J_n}\right)i_q-\frac{T_L}{J}+\frac{K_t}{J_n}\left(i_q-i_q^{*}\right)$ in the formula, B.sub.o and J.sub.n represent the nominal values of the viscous friction coefficient and the moment of inertia respectively; ΔB=B−B.sub.o and ΔJ=J−J.sub.n represent deviations between true values and the nominal values of the viscous friction coefficient and the moment of inertia; i.sub.q* represents a reference value of the stator current of the q axis, i.e., a PMSM speed controller to be designed; after processing the mathematical model of the speed loop of the PMSM which considers the system disturbance, obtaining:
{dot over (ω)}=αi.sub.q*+d in the formula, d(t) represents a lumped disturbance term; α is a known constant coefficient; S2. constructing a fast integral terminal sliding surface: firstly, defining a speed tracking error: e=ω−ω.sub.d, where ω.sub.d represents motor target speed; then, designing the fast integral terminal sliding surface as: s=e+α∫.sub.0.sup.tedσ+β∫.sub.0.sup.te.sup.q/pdσ; where α,β>0, which are constant coefficients; 0<q/p<1; q and p are positive odd numbers; when the tracking error of the motor speed converges to the sliding surface, s=0, i.e., e=−α∫.sub.0.sup.tedσ−β∫.sub.0.sup.te.sup.q/pdσ; solving the above equation to obtain the time for the tracking error of the motor speed to converge to zero from reaching the sliding surface: $t_s=\frac{p}{\alpha \left(p-q\right)}\left(\ln \left[\alpha {e\left(0\right)}^{\left(p-q\right)/p}+\beta \right]-\ln \beta \right)$ S3. conducting disturbance estimation on the lumped disturbance term based on an adaptive fuzzy system: estimating the lumped disturbance term d(t) defined in step S1 by the fuzzy system, and according to the universal approximation theorem of the fuzzy system, an optimal fuzzy system Φ.sup.TH(x) exists, so that:
d=Φ.sup.TH(x)+ε in the formula, Φ=[φ.sub.1, φ.sub.2, . . . , φ.sub.r].sup.T is a weight vector at optimal approximation, and r is the number of fuzzy rules; ε represents an estimation error which satisfies |ε|<ρ; ρ>0 is a positive constant; the value of ε can be infinitely reduced by increasing the number of the fuzzy rules; x=[x.sub.1, x.sub.2, . . . , x.sub.n].sup.T is an input vector of the fuzzy system, and n is the number of fuzzy inputs; selecting the fast integral terminal sliding surface and the speed tracking error as the inputs of fuzzy system, i.e., x=[s,e].sup.T, with H(x)=[h.sub.1(x), h.sub.2(x), . . . , h.sub.r((x))].sup.T representing a fuzzy basis function vector, and: $h_i\left(x\right)=\frac{\underset{k=1}{\overset{n}{.Math.}}{\mu }_{A_k^i}\left(x_k\right)}{\underset{i=1}{\overset{r}{.Math.}}\left[\underset{k=1}{\overset{n}{.Math.}}{\mu }_{A_k^i}\left(x_k\right)\right]},i=1,.Math.,r$ where μ.sub.A.sub.k.sub.i(x.sub.k) represents a membership function value of a fuzzy variable; since the weight vector Φ in optimal approximation cannot be obtained directly, estimating Φ; making {circumflex over (Φ)} represent the estimated value of Φ; and based on an adaptive theory, designing the online adaptive weight adjustment rate of the fuzzy system as:
{circumflex over ({dot over (Φ)})}ΓH(x)s where Γ∈R.sup.r×r is a positive definite symmetric matrix; and s represents the fast integral terminal sliding surface constructed in step S2. after obtaining the estimated value {circumflex over (Φ)} of Φ according to the above adaptive weight adjustment rate, using {circumflex over (Φ)}.sup.TH(x) to estimate the lumped disturbance term d(t) online; S4. designing the PMSM speed controller: designing the following form of PMSM speed controller based on steps S2 and S3: $i_q^{*}=\frac{1}{a}\left[{\stackrel{.}{\omega }}_d-{\hat{\Phi }}^{T}H\left(x\right)-\alpha e-\beta e^{q/p}-k_{1}s-k_2\mathrm{sign}\left(s\right)\right]$ in the formula, k.sub.1 and k.sub.2 are positive adjustable control gains; k.sub.2>l+ρ; l>0 is the upper bound of {tilde over (Φ)}.sup.TH(x), i.e., |{tilde over (Φ)}.sup.TH(x)|<l; {tilde over (Φ)}=Φ−{circumflex over (Φ)} represents a weight estimation error vector; and sign(s) represents a signum function. based on the above PMSM speed controller, making the speed tracking error converge to the sliding surface in time t.sub.o, with t.sub.o satisfying the following relational expression: $t_o\le \frac{1}{\lambda }.Math.s\left(0\right).Math.$ in the formula, λ=k.sub.2−ρ−l, which represents the constant coefficient; and s(0) represents the value of the fast integral terminal sliding surface s constructed in step S2 at time of 0; S5. specifically realizing the technical solution: 5.1) firstly, measuring the motor speed ω in real time through a sensor installed in the PMSM, and after obtaining a motor speed signal ω, making a difference between the signal and the motor target speed ω.sub.d to obtain a speed tracking error e; after obtaining e, further obtaining the value of the fast integral terminal sliding surface s; and meanwhile, based on step S3, obtaining a disturbance estimation value {circumflex over (Φ)}.sup.TH(x) outputted by the adaptive fuzzy system; 5.2) secondly, substituting the speed tracking error e, the fast integral terminal sliding surface s and the disturbance estimation value {circumflex over (Φ)}.sup.TH(x) into the PMSM speed controller given in step S4, and using the controller as a speed loop controller under a PMSM vector control frame to generate a reference value i.sub.q* of stator current of the q axis; 5.3) In a current loop, obtaining the voltage in the d-q coordinate system according to the input stator current reference value using a classical PI controller, obtaining a voltage signal in a static coordinate system through inverse Park transformation, generating a corresponding duty cycle, and obtaining a switch signal of a three-phase inverter; outputting PMSM three-phase stator voltage by the three-phase inverter, and then controlling the motor speed to track the target speed to realize the whole motor speed regulation process.

2. The finite time speed control method for the PMSM based on the fast integral terminal sliding mode and disturbance estimation according to claim 1, wherein the finite time speed control method for the PMSM is adopted so that the time used in the whole motor speed regulation process is finite, and the finite time tracking of motor speed can be realized; combining with the time t.sub.s taken for the motor speed tracking error to converge to zero on the designed fast integral terminal sliding surface obtained in step S2 and the time t.sub.o taken for the motor speed tracking error to converge to the sliding surface obtained in step S4, obtaining the time for the motor speed tracking error converges to zero from an initial state, i.e., time t.sub.r taken for the motor to reach the target speed, which satisfies: $t_r=t_o+t_s\le \frac{1}{\lambda }.Math.s\left(0\right).Math.+\frac{p}{\alpha \left(p-q\right)}\left(\ln \left[\alpha {e\left(0\right)}^{\left(p-q\right)/p}+\beta \right]-\ln \beta \right).$

Description

DESCRIPTION OF DRAWINGS

[0037] FIG. 1 is a vector control frame of a PMSM speed regulating system;

[0038] FIG. 2 is a control frame of a PMSM speed regulating system proposed by the present invention;

[0039] FIG. 3 is a vector control frame of a PI algorithm-based PMSM speed regulating system commonly used in industry;

[0040] FIG. 4 is a comparison diagram of speed curves under ideal conditions by using the proposed method and PI algorithm;

[0041] FIG. 5 shows i.sub.q* response curves under ideal conditions by using the proposed method and PI algorithm;

[0042] FIG. 6 is a comparison diagram of speed curves under the influence of disturbance by using the proposed method and PI algorithm; and

[0043] FIG. 7 shows i.sub.q* response curves under the influence of disturbance by using the proposed method and PI algorithm.

DETAILED DESCRIPTION

[0044] The technical solution proposed by the present invention is further described below in detail in combination with the drawings and specific embodiments.

[0045] The present embodiment discloses a finite time speed control technology for a permanent magnet synchronous motor based on a fast integral terminal sliding mode and disturbance estimation. Specific implementation modes are as follows:

[0046] The present invention is designed for speed control of a non-salient pole permanent magnet synchronous motor, and is based on a vector control frame of i.sub.d=0 permanent magnet synchronous motor (PMSM) speed regulating system as shown in FIG. 1. For the convenience of the design of the controller, a cascade control structure of a speed loop and two current loops is adopted, i.e., the output of the speed loop controller is the reference current input of the current loop. The present invention mainly designs and improves the speed loop in the vector control frame shown in FIG. 1, and builds the control frame of the PMSM speed regulating system as shown in FIG. 2. The design process is introduced in detail below:

[0047] S1: Determining a mathematical model of a speed loop of the PMSM under the influence of system parameters uncertainty and unknown load torque:

[0048] with a rotor coordinate system (d-q coordinate system) as a reference coordinate system, a mathematical model of a speed loop of the non-salient pole permanent magnet synchronous motor is:

[00009]$\begin{array}{cc}\stackrel{.}{\omega }=\frac{K_t}{J}{i}_q-\frac{B}{J}\omega -\frac{T_L}{J}& \left(1\right)\end{array}$

where ω is motor speed; i.sub.q represents stator current of q axis; K.sub.t is a torque constant; J represents the moment of inertia; B is a viscous friction coefficient; and T.sub.L represents a load torque.

[0049] Further considering the influence of system parameters uncertainty, unknown load torque and current loop tracking error, and rewriting the mathematical model of the PMSM as:

[00010]$\begin{array}{cc}\stackrel{.}{\omega }=\frac{K_t}{J_n}{i}_q^{*}-\frac{B_o+\Delta B}{J_n+\Delta J}\omega +\left(\frac{K_t}{J}-\frac{K_t}{J_n}\right)i_q-\frac{T_L}{J}+\frac{K_t}{J_n}\left(i_q-i_q^{*}\right)& \left(2\right)\end{array}$

where B.sub.o and J.sub.n represent the nominal values of the viscous friction coefficient and the moment of inertia respectively; ΔB=B−B.sub.o and ΔJ=J−J.sub.n represent deviations between true values and the nominal values of the viscous friction coefficient and the moment of inertia; and i.sub.q* represents a reference value of the stator current of the q axis, i.e., a PMSM speed controller to be designed.

[0050] Further processing the mathematical model of the speed loop of the PMSM which considers the system disturbance to obtain:

{dot over (ω)}=αi.sub.q*+d  (3)

In the formula,

[00011]$a=\frac{K_t}{J_n};$

since the torque constant K.sub.t and the nominal value J.sub.n of the moment of inertia are known, a is a known constant coefficient; d(t) represents a lumped disturbance term of which the expression is

[00012]$\begin{array}{cc}d=-\frac{B_o+\Delta B}{J_n+\Delta J}\omega +\left(\frac{K_t}{J}-\frac{K_t}{J_n}\right)i_q-\frac{T_L}{J}+\frac{K_t}{J_n}\left(i_q-i_q^{*}\right)& \left(4\right)\end{array}$

[0051] S2: Constructing a fast integral terminal sliding surface

[0052] defining a speed tracking error:

e=ω−ω.sub.d  (5)

where ω.sub.d represents motor target speed.

[0053] Sliding mode control is widely used in the field of PMSM speed control because of its advantages of strong robustness and anti-disturbance capability. However, traditional sliding mode control can only realize the control effect of asymptotic convergence, and cannot ensure that the motor speed can track the target value in finite time. In terminal sliding mode control, the nonlinear item is introduced into the sliding surface and a nonlinear hyperplane is used as the sliding surface, so that the system state on the sliding surface can converge to an equilibrium point in finite time. At the same time, in order to ensure that the signals used in the proposed technical solution can be directly obtained and improve the steady-state tracking performance of the PMSM speed regulating system, the present invention selects the integral terminal sliding surface as the sliding mode. The expression of the traditional integral terminal sliding surface is as follows:

s=e+β.sub.1∫.sub.0.sup.te.sup.q.sup.1.sup./p.sup.1dσ  (6)

where β.sub.1>0 is a constant coefficient; 0<q.sub.1/p.sub.1<1; and q.sub.1 and p.sub.1 are positive odd numbers.

[0054] The convergence of the above sliding surface is analyzed below. When the speed tracking error converges to the sliding surface, i.e., s=0, the following formula holds:

e=−β.sub.1∫.sub.0.sup.te.sup.q.sup.1.sup./p.sup.1dσ  (7)

[0055] The derivative of formula (7) is taken to obtain

ė=−β.sub.1e.sup.q.sup.1.sup./p.sup.1dσ  (8)

[0056] 0<q.sub.1/p.sub.1<1. Thus, when the speed tracking error is far from the equilibrium point, i.e. |e|>>1, the value of |ė will be greatly reduced. Namely, at this moment, the convergence rate of the speed tracking error e will be obviously reduced, and is lower than that of the traditional linear sliding surface, which is the disadvantage of the traditional integral terminal sliding surface.

[0057] To further increase the convergence rate of the traditional integral terminal sliding surface, the present invention proposes the following form of fast integral terminal sliding surface:

s=e+α∫.sub.0.sup.tedσ+β∫.sub.0.sup.te.sup.q/pdσ  (9)

where α,β>0, which are constant coefficients; 0<q/p<1; and q and p are positive odd numbers.

[0058] Next, the convergence rate of the designed sliding surface is analyzed. When the speed tracking error converges to the sliding surface, i.e., s=0, then

e=−α∫.sub.0.sup.tedσ−β∫.sub.0.sup.te.sup.q/pdσ  (10)

[0059] The derivative of the above formula is taken to obtain

ė=−αe−βe.sup.q/p  (11)

[0060] Due to the existence of −αe term which provides the convergence speed that can ensure a positive correlation relation with the distance between the motor speed tracking error and the equilibrium point, when the speed tracking error is far from the equilibrium point, i.e. |e|>>1, although the convergence rate provided by −βe.sup.q/p term is small, −αe term will provide a large convergence rate. When the speed tracking error is close to the equilibrium point, i.e. |e|<<1, then the convergence rate provided by −αe term is small, but due to 0<q/p<1, −βe.sup.q/p term will provide a large convergence rate. In conclusion, the fast integral terminal sliding surface designed by the present invention can ensure high convergence rate as long as the motor speed tracking error is located on the sliding surface.

[0061] The specific convergence time is provided below. The equation (11) is solved to calculate the time for the motor speed tracking error to converge to zero from reaching the sliding surface:

[00013]$\begin{array}{cc}{t}_s=\frac{p}{\alpha \left(p-q\right)}\left(\ln \left[\alpha {e\left(0\right)}^{\left(p-q\right)/p}+\beta \right]-\ln \beta \right)& \left(12\right)\end{array}$

[0062] S3: Disturbance estimation method based on an adaptive fuzzy system

[0063] Through product inference, weighted average and singleton fuzzifier, the output of the fuzzy system can be expressed as

y(x)=Φ.sup.TH(x)  (13);

in the formula, y represents the output of the fuzzy system; x−[x.sub.1, x.sub.2, . . . , x.sub.n].sup.T is an input vector of the fuzzy system, and n is the number of fuzzy inputs. In the technical solution proposed by the present invention, the fast integral terminal sliding surface and the speed tracking error are selected as the inputs of the fuzzy system, i.e., x=[s,e].sup.T; Φ.sup.T=[Φ.sub.1, Φ.sub.2, . . . , Φ.sub.r].sup.T is an adjustable weight vector, and r is the number of fuzzy rules; H(x)=[h.sub.1(x), h.sub.2(x), . . . , h.sub.r((x))].sup.T represents a fuzzy basis function vector, and:

[00014]$\begin{array}{cc}{h}_i\left(x\right)=\frac{\underset{k=1}{\overset{n}{.Math.}}{\mu }_{A_k^i}\left(x_k\right)}{\underset{i=1}{\overset{r}{.Math.}}\left[\underset{k=1}{\overset{n}{.Math.}}{\mu }_{A_k^i}\left(x_k\right)\right]},i=1,.Math.,r& \left(14\right)\end{array}$

where μ.sub.A.sub.k.sub.i(x.sub.k) represents a membership function value of a fuzzy variable.

[0064] In the present invention, the lumped disturbance term d(t) defined in step S1 is estimated by the fuzzy system, and according to the universal approximation theorem of the fuzzy system, an optimal fuzzy system Φ.sup.TH(x) exists, so that:

d=Φ.sup.TH(x)+ε  (15)

In the formula, Φ=[φ.sub.1, φ.sub.2, . . . , φ.sub.r].sup.T is a weight vector at optimal approximation; ε represents an estimation error which satisfies |ε|<ρ; ρ>0 is a positive constant; and the value of εcan be infinitely reduced by increasing the number of the fuzzy rules.

[0065] Since the weight vector Φ in optimal approximation cannot be obtained directly, estimating Φ; making {circumflex over (Φ)} represent the estimated value of Φ; and based on an adaptive theory, designing the online adaptive weight adjustment rate of the fuzzy system as:

{circumflex over ({dot over (Φ)})}ΓH(x)s  (16)

where Γ∈R.sup.r×r is a positive definite symmetric matrix; and s represents the fast integral terminal sliding surface constructed in step S2. After obtaining the estimated value {circumflex over (Φ)} of Φ according to the above adaptive weight adjustment rate, {circumflex over (Φ)}.sup.TH(x) can be used to estimate the lumped disturbance term d(t) online.

[0066] S4: Designing the PMSM speed controller

[0067] designing the following form of PMSM speed controller based on steps S2 and S3:

[00015]$\begin{array}{cc}{i}_q^{*}=\frac{1}{a}\left[{\stackrel{.}{\omega }}_d-{\hat{\Phi }}^{T}H\left(x\right)-\alpha e-\beta e^{q/p}-k_{1}s-k_2\mathrm{sign}\left(s\right)\right]& \left(17\right)\end{array}$

where k.sub.1 and k.sub.2 are positive adjustable control gains; k.sub.2>l+ρ; l>0 is the upper bound of {tilde over (Φ)}.sup.TH(x), i.e., |{tilde over (Φ)}.sup.TH(x)|<l; {tilde over (Φ)}=Φ−{circumflex over (Φ)} represents a weight estimation error vector; the boundedness of the signal will be proved later; and sign(s) represents a signum function.

[0068] The derivative of the fast integral terminal sliding surface (9) with respect to time t is taken:

{dot over (s)}=ė+αe+βe.sup.q/p={dot over (ω)}−{dot over (ω)}.sub.d+αe+βe.sup.q/p  (18)

[0069] Further, in combination with the system mathematical model (3), the following formula can be obtained:

{dot over (s)}−αi.sub.q*+d−{dot over (ω)}.sub.d+αe+βe.sup.q/p  (19)

[0070] The controller expression (17) is substituted into the above equation, and according to formula (15):

{dot over (s)}=Φ.sup.TH(x)−{circumflex over (Φ)}.sup.TH(x)−k.sub.1s−k.sub.2sign(s)+ε  (20)

[0071] The stability of a closed-loop system is analyzed below according to Lyapunov method, which proves that the technical solution proposed by the present invention can control the motor speed to reach a given value in finite time and can effectively overcome the influence of system disturbance. Proof: constructing a Lyapunov function of the following form:

[00016]$\begin{array}{cc}V=\frac{1}{2}{s}^2+\frac{1}{2}{\tilde{\Phi }}^T{\Gamma }^{-1}\tilde{\Phi }& \left(21\right)\end{array}$

[0072] taking the derivative of the Lyapunov function, and combining with formulas (16) and (20) to obtain:

[00017]$\begin{array}{cc}\begin{array}{c}\stackrel{.}{V}=s\stackrel{.}{s}+{\tilde{\Phi }}^T{\Gamma }^{-1}\stackrel{.}{\stackrel{~}{\Phi }}\\ =s\left({\stackrel{\_}{\Phi }}^{T}H\left(x\right)-{\hat{\Phi }}^{T}H\left(x\right)-k_{1}s-k_2\mathrm{sign}\left(s\right)+\epsilon \right)-{\tilde{\Phi }}^T{\Gamma }^{-1}\left(\Gamma H\left(x\right)s\right)\\ =-k_1{s}^2-k_2.Math.s.Math.+s\epsilon \le k_1{s}^2-\left(k_2-\rho \right).Math.s.Math.\le 0\end{array}& \left(22\right)\end{array}$

[0073] according to formulas (21) and (22), obtaining that V(t) is bounded, i.e.,

V∈ζ.sub.∞  (23)

[0074] From the form of V(t), it can be seen that s,{tilde over (Φ)}∈ζ.sub.∞, and further combining with formulas (5), (6) and (16), obtaining:

e,ω,{circumflex over (Φ)},{dot over (s)},i.sub.q*∈ζ.sub.∞  (24)

Namely, all the signals in the closed-loop system are bounded.

[0075] In addition, h.sub.i(x)i=1, . . . , r represents a fuzzy basis function which is a bounded function, and then H(x)=[h.sub.1(x),h.sub.2(x), . . . ,h.sub.r((x))].sup.T∈ζ.sub.∞;

due to {tilde over (Φ)}∈ζ.sub.∞, {tilde over (Φ)}.sup.TH(x)∈ζ.sub.∞ holds. Assuming that:

|{tilde over (Φ)}.sup.TH(x)|<1  (25)

where l>0 is a positive constant.

[0076] Next, it is proved that the speed tracking error can converge to the sliding surface in finite time, that is, the time for s to converge to zero is finite.

[0077] A new Lyapunov function is designed as follows:

[00018]$\begin{array}{cc}{V}_1=\frac{1}{2}{s}^2& \left(26\right)\end{array}$

[0078] By taking the derivative of the Lyapunov function, in combination with formulas (20) and (25), it can be obtained that

[00019]$\begin{array}{cc}\begin{array}{c}{\stackrel{.}{V}}_1=s\stackrel{.}{s}\\ =s\left({\stackrel{\_}{\Phi }}^{T}H\left(x\right)-{\hat{\Phi }}^{T}H\left(x\right)-k_{1}s-k_2\mathrm{sign}\left(s\right)+\epsilon \right)\\ =-k_1{s}^2-k_2.Math.s.Math.+s{\tilde{\Phi }}^{T}H\left(x\right)+s\epsilon \le -k_1{s}^2-\left(k_2-\rho -l\right).Math.s.Math.\le 0-\lambda .Math.s.Math.\end{array}& \left(27\right)\end{array}$

In the formula, λ=k.sub.2−ρ−l. Further, according to formulas (26) and (27), it can be obtained that

{dot over (V)}.sub.1≤−λ√{square root over (2V.sub.1(s))}  (28)

[0079] It is assumed that the system state will reach the sliding surface at t=t.sub.o, i.e., V.sub.1(t.sub.o)=0; then the definite integral of formula (28) at time 0−t.sub.o can be calculated:

[00020]$\begin{array}{cc}\sqrt{V_1\left(t_o\right)}-\sqrt{V_1\left(0\right)}\le -\frac{\sqrt{2}}{2}{t}_o.Math.t_o\le \frac{1}{\lambda }\sqrt{2V_1\left(0\right)}& \left(29\right)\end{array}$

That is:

[0080] [00021]$\begin{array}{cc}{t}_o\le \frac{1}{\lambda }.Math.s\left(0\right).Math.& \left(30\right)\end{array}$

where s(0) represents the value of the fast integral terminal sliding surface s constructed in step S2 at time of 0.

[0081] Further, in combination with the time t.sub.s taken for the motor speed tracking error to converge to zero on the designed fast integral terminal sliding surface obtained in step S2, it is obtained the time that the motor speed tracking error converges to zero from an initial state, i.e., the time t.sub.r taken for the motor to reach the target speed, which satisfies:

[00022]$\begin{array}{cc}{t}_r=t_o+t_s\le \frac{1}{\lambda }.Math.s\left(0\right).Math.+\frac{p}{\alpha \left(p-q\right)}\left(\ln \left[\alpha {e\left(0\right)}^{\left(p-q\right)/p}+\beta \right]-\ln \beta \right)& \left(31\right)\end{array}$

Then, through the above rigorous theoretical analysis, it proves that the technical solution proposed by the present invention can enable the motor speed to reach a target value in finite time and can also effectively overcome the influence of system disturbance.

[0082] S5: Specifically realizing the technical solution

[0083] Through steps S1-S4, the control frame of the PMSM speed regulating system shown in FIG. 2 can be built. The specific realization is introduced in detail here. Firstly, measuring the motor speed ω in real time through a sensor installed in the PMSM, and after obtaining a motor speed signal ω, making a difference between the signal and the motor target speed ω.sub.d to obtain a speed tracking error e; after obtaining e, further obtaining the value of the fast integral terminal sliding surface s; meanwhile, based on step S3, obtaining a disturbance estimation value {circumflex over (Φ)}.sup.TH(x) outputted by the adaptive fuzzy system; substituting the speed tracking error e, the fast integral terminal sliding surface s and the disturbance estimation value {circumflex over (Φ)}.sup.TH(x) into the PMSM speed controller (17) given in step S4, and using the controller as a speed loop controller under a PMSM vector control frame to generate a reference value i.sub.q* of stator current of the q axis. In a current loop, obtaining the voltage in the d-q coordinate system according to the input stator current reference value using a classical PI controller, obtaining a voltage signal in a static coordinate system through inverse Park transformation, then generating a corresponding duty cycle through a SVPWM algorithm , and obtaining a switch signal of a three-phase inverter; outputting PMSM three-phase stator voltage by the three-phase inverter, and then controlling the motor speed to track the target speed to realize the whole motor speed regulation process.

[0084] In order to further verify the effectiveness and advancement of the proposed technology, the present invention compares the control performance with that of the “vector control frame of the PI algorithm-based PMSM speed regulating system” as the most commonly used industrial control frame through simulation. FIG. 3 shows the vector control frame of the PI algorithm-based PMSM speed regulating system, which is referred to as “PI algorithm” in later illustration. The PI algorithm is used in the speed loop and the current loop. This control solution has the advantages of convenient adjustment and easy realization. Thus, the PI algorithm is most widely used in the industrial field.

[0085] Simulation 1: comparison of speed regulation performance under ideal conditions

[0086] Firstly, the speed regulation performance of the proposed technical solution and the PI algorithm is compared under ideal conditions, that is, without parameters uncertainty and load torque disturbance. System parameters are set as follows: J=3.78×10.sup.−4 kg.Math.m.sup.2, B=1.74×10.sup.−5 N.Math.m.Math.s/rad, K.sub.t=1.4N.Math.m/A; and at the same time, the motor target speed is set as ω.sub.d=1200 r/min. Simulation results are shown in FIGS. 4-5. FIG. 4 is a comparison diagram of speed curves by using the proposed method and the PI algorithm; and FIG. 5 shows i.sub.q* response curves by using two different solutions. In combination with FIG. 4 and FIG. 5, under ideal conditions, the proposed method only needs 0.0075 s to make the motor reach the given speed, while the PI algorithm takes about 0.015 s which is about twice as long as the proposed method. At the same time, it can be seen that the steady-state error of the PI algorithm is significantly greater than that of the proposed method. The results of simulation 1 show that compared with the PI algorithm, the technical solution of the present invention has higher response speed and higher control accuracy.

[0087] Simulation 2: comparison of speed regulation performance under the influence of disturbance

[0088] Further, in order to verify and compare the anti-disturbance capability of the proposed method and the PI algorithm, the influence of internal and external disturbances including system parameters uncertainty and load torque change is comprehensively considered in simulation 2, the motor moment of inertia is adjusted to twice of that in simulation 1, and the viscous friction coefficient is adjusted to 5 times of the original value, i.e., J=2×3.78×10.sup.−4 kg.Math.m.sup.2, B=5×1.74×10.sup.−5 N.Math.m.Math.s/rad; and other system parameters are unchanged. Meanwhile, the controller parameters of the proposed method in the present invention and the PI algorithm are also kept the same as those in simulation 1. In addition, in order to simulate the influence of the unknown load torque, at 0.03 s during simulation, 0.5 N.Math.m load torque is applied, and the load torque is removed at 0.035 s. Simulation results are shown in FIGS. 6-7. FIG. 6 is a comparison diagram of speed curves by using the proposed method and the PI algorithm; and FIG. 7 shows i.sub.q* response curves by using two different solutions. In combination with FIG. 6 and FIG. 7, under the influence of disturbance, the proposed method can still ensure good dynamic performance and steady-state performance, and the motor accurately reaches the given speed at about 0.012 s Accordingly, the PI algorithm takes about 0.03 s to adjust the motor to the target speed, and the time used is nearly 3 times that of the proposed method in the present invention. More importantly, when the load torque suddenly changes, the motor speed under the control of the PI algorithm obviously fluctuates, and the decline range is far greater than the decline range of the speed of the method proposed by the present invention. At the same time, compared with the PI algorithm, the proposed method in the present invention can eliminate the influence of disturbance more quickly and make the motor speed return to the target value smoothly in a very short time. The results of simulation 2 show that compared with the PI algorithm, the proposed method of the present invention has obvious anti-disturbance performance advantage, can effectively overcome the influence of internal and external system disturbance, and can also ensure the control accuracy and rapidity of the motor speed regulating system.

[0089] In conclusion, the results of simulation 1 and simulation 2 show that compared with the PI algorithm control solution commonly used in industry, the technical solution proposed by the present invention has higher response speed and higher control accuracy, and can accurately adjust the motor speed to the given value within a shorter time. At the same time, the proposed technical solution has advantages in anti-disturbance capability and can effectively overcome the influence of disturbance including system parameters uncertainty and unknown load torque, which means that the present invention is more practical and suitable for application in actual systems.

[0090] The above embodiments only express the implementation of the present invention, and shall not be interpreted as a limitation to the scope of the patent for the present invention. It should be noted that, for those skilled in the art, several variations and improvements can also be made without departing from the concept of the present invention, all of which belong to the protection scope of the present invention.