A switch reluctance motor position sensorless control method based on CKF

Abstract

The application discloses a kind of switch reluctance motor position sensorless control methods based on CKF, comprising: determining the flux linkage characteristic and torque characteristic of switch reluctance motor;Set switch reluctance motor state vector, observation vector and control vector, and construct the space state equation and measurement equation of switch reluctance motor model;The space state equation and measurement equation are discretized;Based on volume Kalman filter (CKF) algorithm, the speed and rotor position of motor are estimated: state vector and state error covariance matrix are initialized, then time update and measurement update are carried out;Finally, the estimated value of speed and rotor position is obtained.The application does not need too much parameter adjustment, guarantees the real-time of calculation while improving the position identification accuracy, increases the robustness of system, effectively avoids the deficiency that traditional method is influenced by switch reluctance motor speed range.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of switched reluctance motor control, and particularly relates to a switched reluctance motor rotor position identification method based on a Cubature Kalman Filter (CKF) algorithm. BACKGROUND

[0002] As one of important components of a driving system, a motor is a factor that is considered in terms of performance and processing and manufacturing cost in different environments and different working conditions; with strategic rare earth resources becoming increasingly scarce, the price is also greatly increased; a switched reluctance motor driving system has broad application prospects due to advantages such as no permanent magnet, simple structure, low cost, high reliability, and good fault tolerance performance. However, continuous and stable operation of the switched reluctance motor needs to obtain real-time rotor position information; in high-temperature high-speed, humid and dusty environments, and precise driving and other special applications, a mechanical sensor not only occupies installation space, but also greatly increases the failure rate of the position sensor due to dust abrasion, so that the motor system cannot work stably; therefore, it is necessary to study a position sensorless control scheme to improve the fault tolerance of the switched reluctance motor.

[0003] According to different application speed ranges, the position sensorless technology also has differences in the methods used; traditional position sensorless control methods of the switched reluctance motor include an inductance method, a flux linkage method, an observer method, and an intelligent algorithm. However, most algorithms can only work under specific speed and characteristic load, and the algorithm is complex and has high requirements for control frequency. SUMMARY

[0004] The application proposes a switched reluctance motor position sensorless control method based on a CKF to solve the problem that traditional position sensorless control methods depend on motor speed and have low estimation accuracy, so as to improve the dynamic performance of a switched reluctance motor speed regulation system and improve the robustness of the system.

[0005] In order to achieve the above task, the application adopts the following technical scheme:

[0006] A switched reluctance motor position sensorless control method based on a CKF includes:

[0007] Determine the flux linkage characteristic and torque characteristic of the switched reluctance motor;

[0008] Set motor state quantity and control quantity, construct a switched reluctance motor model based on the flux linkage characteristic and torque characteristic, and set a state vector, an observation vector, and a control vector;

[0009] The switched reluctance motor model is discretized, and the angular velocity and speed of the motor are estimated based on a cubature Kalman filter algorithm, wherein the cubature Kalman filter algorithm includes:

[0010] The initial estimate of the state vector and the initial estimate of the error state covariance are initialized, and the time update of the cubature Kalman filter algorithm is realized by calculating the volume points, using the state equation to perform nonlinear transformation propagation of the volume points, and calculating the state quantity prediction value and the error covariance prediction value; the measurement update of the cubature Kalman filter algorithm is realized by calculating the volume points, using the observation equation to perform nonlinear transformation propagation of the volume points, calculating the measurement estimate value, calculating the measurement error covariance and cross-covariance, calculating the Kalman gain, the state estimate value and the state error covariance matrix; finally, the state estimate value is used to determine the estimated values ​​of the angular velocity and velocity.

[0011] Furthermore, determining the flux linkage characteristics and torque characteristics of the switched reluctance motor includes:

[0012] (I) The Fourier series expression of the magnetic flux characteristics is:

[0013] ψ(i,θ)=Θ 1×5 ×λ 5×1 (1)

[0014] Where:

[0015]

[0016] where λ 5×1 are five undetermined coefficients related to the phase current i; λ0(i), λ1(i), λ2(i), λ3(i), λ4(i) are flux linkage coefficients, which are related to the current; Θ 1×5 is the basis function matrix of the rotor position angle θ, and the superscript T represents the matrix transpose operation;

[0017] (II) The flux linkage of the motor at five special positions, 0°, 7.5°, 11.5°, 5°, and 22.5°, can be expressed as:

[0018] ψ 5×1 =A 5×5 ×λ 5×1 (2)

[0019] in:

[0020]

[0021] In the above formula, A 5×5 represents the Fourier angle fitting coefficient matrix, ψ 0° ψ 7.5° ψ 11.25° ψ 15° ψ22.5 Flux linkage of the motor at 0°, 7.5°, 11.5°, 15° and 22.5°, respectively;

[0022] (III) When the angle is fixed, the relationship between the flux linkage and the current can be expressed by a six-order polynomial fitting without constant term:

[0023] ψ 5×1 = B 5×6 × I 6×1 (3)

[0024] Where, I 6×1 = [i 6 , i 5 , i 4 , i 3 , i 2 , i] T is the current matrix, and the upper index number represents the phase current power or the current order; B 5×6 is the fitting coefficient;

[0025] (IV) According to the equality of formula (2) and formula (3), we have:

[0026] λ 5×1 = A 5×5 -1 × B 5×6 × I 6×1 (4)

[0027] (V) Using formula (1) and (4), the torque formula of the SRM can be further obtained:

[0028]

[0029] In the formula, W c is the magnetic energy of the motor;

[0030] (VI) According to the voltage balance equation of the SRM, the calculation formula of the single-phase current is:

[0031]

[0032] In the formula, t is the time parameter, u ph is the phase voltage, i ph is the phase current, R s is the stator phase resistance, ω is the angular velocity, ψ ph is the phase flux linkage, is the partial derivative of the phase flux linkage with respect to the phase current, is the partial derivative of the phase flux linkage with respect to the phase electrical angle θ ph .

[0033] Further, the setting motor state quantity and control quantity, based on the flux linkage characteristic and torque characteristic, constructs a switched reluctance motor model, including:

[0034] According to the flux linkage ψ(i, θ) and torque T(i, θ) to establish a switched reluctance motor model, and with three-phase current i a ,i b ,i c , the rotating speed ω and rotor position angle θ are five parameters as the motor state, and three-phase voltage U a ,U b ,U c as the control quantity:

[0035]

[0036] In formula (7), t is a time parameter, R a ,R b ,R c is the three-phase winding resistance, ψ a ,ψ b ,ψ c is the magnetic resistance of the flux linkage ψ(i, θ) on the three-phase winding, θ a ,θ b ,θ c is the three-phase angle, f r is the damping coefficient, J is the rotational inertia, T e is the torque T(i, θ), T L is the load torque, represents the derivative with respect to time, respectively represent the partial derivative of the A, B, C three-phase flux linkage with respect to the three-phase electric angle.

[0037] Further, the setting state vector, observation vector and control vector, including:

[0038] The selected state vector is: x = [i a i b i c ωθ] T The observation vector is: y = [i a i b i c ] T = h(x, u), and the control vector is: u = [u a u b u c ] T , wherein:

[0039]

[0040] Further, the switched reluctance motor model is discretized, comprising:

[0041]

[0042] In the formula, x k and x k-1 are state vectors of the system at k and k-1 time, respectively, u k and u k-1 are observation vectors of the system at k and k-1 time, respectively, y k is an observation vector of the system at k time, f() is a state equation of the time-varying system, and h() is an observation equation of the time-varying system; process noise W k and observation noise V k are Gaussian white noises with mean values of 0 and variances of Q and R, respectively, and Q and R are symmetric matrices.

[0043] Further, the initial estimated value of the state vector and the initial estimated value of the error state covariance are initialized, comprising:

[0044]

[0045] In the formula, x is the initial estimated value of the state vector, x0 is the true value of the state vector, P0 is the initial estimated error state covariance, and E is a mathematical expectation operation.

[0046] Further, the time update of the cubature Kalman filter algorithm is implemented by calculating a cubature point, performing nonlinear transformation and propagation of the cubature point by using a state equation, and calculating a state quantity prediction value and an error covariance prediction value, comprising:

[0047] (I) calculating a cubature point

[0048] Suppose that the current time is k, and the state estimation value at k-1 time is known and the k-1 state error covariance matrix P k-1k-1 is decomposed by Cholesky: k-1k-1

[0049]

[0050] In the formula, S k-1|k-1 is the Cholesky decomposition matrix of P k-1|k-1 at k-1 time;

[0051] According to the decomposed state error covariance matrix, the cubature point is obtained according to a third-order spherical radial cubature criterion:

[0052]

[0053] ​wherein, is the i th cubature point at time k-1, is the i th standard normal cubature point, [1] i denotes a vector with the i th element being 1, i = 1, 2, 3…m, m = 2n, n is the dimension of state vector;

[0054] (II) Propagating the above obtained cubature points through the state equation f() by cubature point nonlinear transformation:

[0055]

[0056] wherein, is the predicted value of the i th cubature point at time k through nonlinear transformation f(·);

[0057] (III) Computing the state quantity predicted value and error covariance predicted value:

[0058]

[0059] wherein, is the predicted value of the i th cubature point at time k, P k|k-1 is the covariance matrix of the prediction error at time k, Q k-1 is the noise matrix at time k-1, ∑(·) is summation operation.

[0060] Further, the measurement update of the cubature Kalman filter algorithm is realized by computing cubature points, propagating the cubature points through nonlinear transformation of the observation equation, computing measurement estimate, computing measurement error covariance and cross-covariance, computing Kalman gain, state estimate and state error covariance matrix, comprising:

[0061] (I) Computing cubature points

[0062] Cholesky decomposition is performed on the state prediction error covariance matrix obtained from equation (14) and a new cubature point is calculated through third-order spherical radial cubature criterion:

[0063]

[0064] wherein, S k|k-1 is the Cholesky decomposition matrix of P k|k-1 at time k;

[0065] (II) According to the system described in equation (9), the above obtained cubature points are propagated through nonlinear transformation of cubature points by observation equation h():

[0066]

[0067] wherein, is the measurement prediction value of the i th cubature point at k moment;

[0068] (III) the measurement estimation value is obtained by weighted summation of the propagated points:

[0069]

[0070] wherein, is the measurement prediction value at k moment;

[0071] (IV) the measurement error covariance and cross-covariance are calculated:

[0072]

[0073] wherein, is the measurement error covariance matrix at k moment, is the measurement error cross-covariance matrix at k moment, and R is the measurement noise matrix;

[0074] (V) the Kalman gain K at k moment is calculated k , the state estimation value and the state error covariance matrix P k|k :

[0075]

[0076] Further, the estimation value of the angular velocity and the speed is determined by using the state estimation value, and the method comprises the steps that:

[0077]

[0078] wherein, the fourth element in the state vector is the estimated angular velocity, and the fifth element is the estimated angle.

[0079] Compared with the prior art, the application has the following technical features:

[0080] The application cites the cubature Kalman filter algorithm, selects the mean value and variance of the cubature point as parameters based on the spherical radial cubature criterion, so that the application has high precision in the system of more than three dimensions, does not need too much parameter adjustment, improves the position recognition precision while ensuring the calculation real-time, increases the system robustness, and is suitable for full-speed range operation, effectively avoids the shortage that the traditional method is affected by the motor speed range, and widens the application of the cubature Kalman filter algorithm. BRIEF DESCRIPTION OF DRAWINGS

[0081] Figure 1 is the principle block diagram of the switched reluctance motor cubature Kalman filter position estimation algorithm of the embodiment of the application;

[0082] Figure 2This is a flow chart of position estimation of a switched reluctance motor with a volumetric Kalman filter algorithm according to an embodiment of the present invention;

[0083] Figure 3 This is an overall diagram and a partial magnified comparison diagram of the estimated motor speed and actual speed of the switched reluctance motor position sensorless speed control system based on the volumetric Kalman filter algorithm according to an embodiment of the present invention;

[0084] Figure 4 This is an overall diagram and a partial magnified comparison diagram of the estimated motor angle and the actual angle of the switched reluctance motor position sensorless speed control system based on the volumetric Kalman filter algorithm according to an embodiment of the present invention; DETAILED DESCRIPTION

[0085] To address the issues of the switched reluctance motor position-free control algorithm being dependent on the motor speed and having limited adaptability, and to broaden the application of the Kalman filter algorithm in the switched reluctance motor speed control system, the present invention proposes a CKF-based switched reluctance motor position-free sensor control method. The present invention is further described with reference to a three-phase 12 / 8 switched reluctance motor as an example and the accompanying drawings. The specific implementation is as follows:

[0086] Step 1: Determine the flux characteristics and torque characteristics of the switched reluctance motor.

[0087] The three-phase 12 / 8 switched reluctance motor is modeled, and the Fourier series model of the switched reluctance motor is established through the characteristic position to realize the online calculation of the motor characteristics.

[0088] To avoid storing a large number of motor characteristic parameters and occupying processor memory, this solution uses a method to obtain the flux curve at an unknown position based on a known curve. By establishing a flux Fourier series model, the motor flux ψ(i,θ) and torque T(i,θ) are calculated online. The details are as follows:

[0089] (I) The Fourier series expression of the magnetic flux characteristics is:

[0090] ψ(i,θ)=Θ 1×5 ×λ 5×1 (1)

[0091] Where:

[0092]

[0093] where λ 5×1 are the five undetermined coefficients related to the phase current i. The coefficient matrix can be obtained through the flux characteristics of the five special positions. λ0(i),λ1(i),λ2(i),λ3(i),λ4(i) are the coefficients of the flux at the five special positions, which are related to the current; Θ 1×5is the rotor position angle, and the superscript T denotes the matrix transpose operation.

[0094] (II) The flux linkage of the motor at five special positions of 0°, 7.5°, 11.5°, 5° and 22.5° can be expressed as:

[0095] ψ 5×1 = A 5×5 × λ 5×1 (2)

[0096] wherein:

[0097]

[0098] In the above formula, A 5×5 is the Fourier angle fitting coefficient matrix, ψ 0° ψ 7.5° ψ 11.25° ψ 15° ψ 22.5 are the flux linkages of the motor at 0°, 7.5°, 11.5°, 5° and 22.5° respectively.

[0099] (III) When the angle is fixed, the relationship between the flux linkage and the current can be expressed by a six-order polynomial fitting without constant term:

[0100] ψ 5×1 = B 5×6 × I 6×1 (3)

[0101] wherein, I 6×1 = [i 6 , i 5 , i 4 , i 3 , i 2 , i] T is the current matrix, and the superscript number indicates the phase current power or current order; B 5×6 is the fitting coefficient.

[0102] (IV) According to the equality of formula (2) and formula (3), we have:

[0103] λ 5×1 = A 5×5 -1 × B 5×6 × I 6×1 (4)

[0104] (V) Using formula (1) and (4), the torque formula of the SRM can be further derived:

[0105]

[0106] wherein, Wc is the magnetic total energy of the motor.

[0107] (VI) The single-phase current calculation formula can be derived from the voltage balance equation of the SRM:

[0108]

[0109] Where t is the time parameter, u ph is the phase voltage, i ph is the phase current, R s is the stator phase resistance, ω is the angular velocity, ψ ph is the phase flux linkage, is the partial derivative of the phase flux with respect to the phase current, The phase flux is the electrical angle θ of the phase ph The partial derivative of is expressed as:

[0110]

[0111] Step 2: Set the motor state and control variables, and build a switched reluctance motor model based on the flux linkage characteristics and torque characteristics.

[0112] According to the flux ψ(i,θ) and torque T(i,θ) obtained in step 1, a switched reluctance motor model is established, and the three-phase current i a ,i b ,i c , speed ω, rotor position angle θ are five parameters as the state of the motor, and the three-phase voltage U a ,U b ,U c As a control quantity:

[0113]

[0114] In formula (7), t is the time parameter, R a ,R b ,R c is the three-phase winding resistance, ψ a ,ψ b ,ψ c is the magnetic resistance of the flux linkage ψ(i,θ) on the three-phase winding, θ a ,θ b ,θ c is the three-phase angle, f r is the damping coefficient, J is the moment of inertia, T e is the torque T(i,θ), T L is the load torque, represents the time derivative, They represent the partial derivatives of the three-phase flux linkage A, B, and C with respect to the three-phase electrical angle.

[0115]

[0116] According to the state equation of the switched reluctance motor, the selected state vector is: x=[i a i b i c ωθ] T The observation vector is: y=[i a i b i c ] T =h(x,u), the control vector is: u=[u a u b u c ] T .

[0117] Step 3, the switched reluctance motor model is discretized.

[0118] After the switched reluctance motor model is discretized by Euler method and the system process noise and observation noise are considered, the following is obtained:

[0119]

[0120] In the formula, x k and x k-1 are the state vectors of the system at time k and k-1 respectively, u k and u k-1 are the system at time k and k-1 respectively, y k is the observation vector at time k, f() is the state equation of the time-varying system, h() is the observation equation of the time-varying system, and since the embodiment is a three-phase 12 / 8 switched reluctance motor, f() is formula (7) and h() is formula (8). Process noise W k and observation noise V k are Gaussian white noise with mean 0 and variance Q and R respectively, and Q and R are both symmetric matrices.

[0121] As can be seen from formula (8), the state transition equation and the observation equation are both nonlinear, so the position observer can be constructed based on the cubature Kalman filter algorithm.

[0122] Step 4, the cubature Kalman filter algorithm is initialized:

[0123]

[0124] In the formula, is the initial estimated value of the state vector, x0 is the true value of the state vector, P0 is the initial estimated error state covariance, and E is the mathematical expectation operation.

[0125] Step 5, time update of the cubature Kalman filter algorithm:

[0126] (I) Calculate volume points

[0127] Assume now at k time, known state estimation value at k-1 time and k-1 state error covariance matrix P k-1k-1 , Cholesky decomposition is performed on state error covariance matrix P k-1k-1 :

[0128]

[0129] In the formula, S k-1|k-1 is the Cholesky decomposition matrix of P k-1|k-1 at k-1 time.

[0130] According to the decomposed state error covariance matrix, the volume points are obtained according to the third-order spherical radial volume criterion:

[0131]

[0132] In the formula, is the i th volume point at k-1 time, is the i th standard normal volume point, [1] i indicates a vector with the i th element being 1, i = 1, 2, 3…m, m = 2n, n is the dimension of the state vector, in this embodiment, it is 5 state vectors in the motor state equation; 2n volume points are obtained by using the third-order spherical radial volume rule.

[0133] (II) According to the system described in formula (9), the volume points obtained above are propagated by volume point nonlinear transformation through the state equation f():

[0134]

[0135] In the formula, is the predicted value of the i th volume point at k time through nonlinear transformation f(·).

[0136] (III) Calculate the state quantity predicted value and error covariance predicted value:

[0137]

[0138] In the formula, is the predicted value of the i th volume point at k time, P k|k-1 is the covariance matrix of the prediction error at k time, Q k-1 is the noise matrix at k-1 time, and ∑(·) is the summation operation.

[0139] Step 6, measurement update of volume Kalman filtering algorithm

[0140] (I) Calculate the volume point

[0141] Cholesky decomposition is performed on the state prediction error covariance matrix obtained by formula (14), and a new volume point is calculated by a third-order spherical radial volume criterion:

[0142]

[0143] In the formula, S k|k-1 is the Cholesky decomposition matrix of P k|k-1 at time k.

[0144] (II) According to the system described by formula (9), the volume point obtained above is propagated by the nonlinear transformation of the volume point through the observation equation h():

[0145]

[0146] In the formula, is the measurement prediction value of the i-th volume point at time k.

[0147] (III) The propagated points are weighted and summed to obtain the measurement estimate:

[0148]

[0149] In the formula, is the measurement prediction value at time k.

[0150] (IV) Calculate the measurement error covariance and cross-covariance:

[0151]

[0152] In the formula, is the measurement error covariance matrix at time k, is the measurement error cross-covariance matrix at time k, and R is the measurement noise matrix.

[0153] (V) Calculate the Kalman gain K k at time k, the state estimate and the state error covariance matrix P k|k :

[0154]

[0155] Step 7, using the motor model established in steps 1 and 2, the estimated value of the motor state is obtained by the volume Kalman filtering algorithm of steps 3, 4, 5 and 6, and the angular velocity and angle information is extracted by formula (20):

[0156]

[0157] The fourth element in the state vector is the estimated angular velocity, with the unit of rad / s, and the fifth element is the estimated angle, with the unit of rad.

[0158] The rotor position estimated by the cubature Kalman filter algorithm is compared with the actual rotor position, and the result is shown in FIG. Figure 3 It can be seen that the cubature Kalman filter algorithm has good position estimation accuracy at low and high speeds, and has good dynamic performance when the motor speed suddenly changes. Therefore, the method proposed in the application is effective.

[0159] The application adopts a cubature Kalman filter algorithm, selects the mean and variance of the cubature points as parameters based on the spherical radial cubature criterion, so that it has high accuracy in systems of more than three dimensions, and does not have the non-positive definite state error covariance matrix, thereby improving the system stability. The position identification accuracy is improved while ensuring the calculation real-time, the system robustness is increased, and the algorithm is suitable for operation in the full speed range, effectively avoiding the shortcomings of the traditional method affected by the motor speed range.

[0160] The above examples are only used to illustrate the technical solutions of the application, but not to limit them; although the application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can still be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the application, and should be included in the protection scope of the application.

Claims

1. A position sensorless control method for a switched reluctance motor based on CKF, characterized in that: include: Determine the flux linkage and torque characteristics of the switched reluctance motor; Setting the motor state and control variables, building a switched reluctance motor model based on the flux linkage characteristics and torque characteristics, and setting the state vector, observation vector, and control vector; The switched reluctance motor model is discretized, and the angular velocity and speed of the motor are estimated based on a cubature Kalman filter algorithm, wherein the cubature Kalman filter algorithm includes: Initial estimates of the state vector and the error state covariance are initialized, and the time update of the cubature Kalman filter algorithm is achieved by calculating volume points, performing nonlinear transformation propagation of the volume points using the state equation, and calculating the state quantity prediction value and the error covariance prediction value; measurement update of the cubature Kalman filter algorithm is achieved by calculating volume points, performing nonlinear transformation propagation of the volume points using the observation equation, calculating the measurement estimate value, calculating the measurement error covariance and cross-covariance, calculating the Kalman gain, the state estimate value, and the state error covariance matrix; finally, the state estimate value is used to determine the estimated values ​​of the angular velocity and velocity; Determining the flux linkage characteristics and torque characteristics of the switched reluctance motor includes: (I) The Fourier series expression of the magnetic flux characteristics is: (1) Where: in is the phase current i 5 undetermined coefficients related to size; is the flux linkage coefficient, which is related to the current; is the rotor position angle θ Basis function matrix, superscript T Represents the matrix transpose operation; (II) The flux linkage at five special positions of the motor at 0°, 7.5°, 11.5°, 5° and 22.5° can be expressed as: (2) in: In the above formula, represents the Fourier angle fitting coefficient matrix, These are the magnetic flux linkages of the motor at 0°, 7.5°, 11.5°, 5°, and 22.5° respectively; (III) When the angle is fixed, the relationship between flux linkage and current can be expressed by a sixth-order polynomial fit without a constant term: (3) in, is the current matrix, and the superscript numbers represent the phase current power or current order; is the fitting coefficient; (IV) According to the equality of formula (2) and formula (3), we can get: (4) (V) Using formulas (1) and (4), the torque formula of SRM can be further derived: (5) Where, is the magnetic total energy of the motor; (VI) The single-phase current calculation formula can be obtained from the voltage balance equation of the SRM: (6) Where, t is the time parameter, is the phase voltage, is the phase current, R s is the stator phase resistance, is the angular velocity, is the phase flux linkage, is the partial derivative of the phase flux with respect to the phase current, The phase flux linkage is the phase electrical angle The partial derivative of The setting of the motor state quantity and the control quantity, and the construction of a switched reluctance motor model based on the flux linkage characteristics and the torque characteristics, include: According to magnetic linkage and torque Establish a switched reluctance motor model and use three-phase current , speed , rotor position angle The five parameters are used as the state of the motor, with three-phase voltage As a control quantity: (7) In formula (7), t is the time parameter, is the three-phase winding resistance, Magnetic Link The reluctance on the three-phase windings, 、 、 is the three-phase angle, is the damping coefficient, is the moment of inertia, is the torque , is the load torque, represents the time derivative, , , , respectively represent the partial derivatives of the three-phase flux linkage of A, B, and C with respect to the three-phase electrical angle.

2. The position sensorless control method of a switched reluctance motor based on CKF according to claim 1, characterized in that: The setting of the state vector, observation vector and control vector includes: The selected state vector is: , the observation vector is: , the control vector is: ,in: (8)。 3. The position sensorless control method of a switched reluctance motor based on CKF according to claim 1, characterized in that: Discretizing the switched reluctance motor model includes: (9) In the formula and System k and k -1 moment state vector, and System k and k -1 moment, —the observation vector of the system at time k, — the state equation of a time-varying system, —Observation equation of time-varying system; process noise W k and observation noise V k are Gaussian white noises with mean 0 and variances Q and R respectively, and both Q and R are symmetric matrices.

4. The position sensorless control method of a switched reluctance motor based on CKF according to claim 1, characterized in that: Initializing the initial estimated value of the state vector and the initial estimated value of the error state covariance includes: (10) Where, is the initial estimate of the state vector, is the true value of the state vector, P 0 is the initial estimated error state covariance, and E is the mathematical expectation operation.

5. The position sensorless control method of a switched reluctance motor based on CKF according to claim 1, characterized in that: The time update of the volumetric Kalman filter algorithm is realized by calculating volume points, performing nonlinear transformation propagation of the volume points using state equations, and calculating state quantity prediction values ​​and error covariance prediction values, including: (I) Calculation of volume points Assume that you are now Moment, known The estimated state at time and k -1 state error covariance matrix , for the state error covariance matrix Perform Cholesky decomposition: (11) Where, for k -1 moment The Cholesky decomposition matrix of According to the decomposed state error covariance matrix, the volume point is obtained according to the third-order spherical radial volume criterion: (12) Where, for k -1 moment i Volume points, For the i standard normal volume points, [1] i Indicates the i A vector with elements of 1, i =1,2,3…m, m=2n, n is the dimension of the state vector; (II) The volume point obtained above is passed through the state equation Perform volume point nonlinear transformation propagation: (13) Where, for k Moment i Volume points are transformed nonlinearly The predicted value obtained; (III) Calculate the predicted value of the state quantity and the predicted value of the error covariance: (14) Where, is the kth moment i The predicted value of each volume point, for k The covariance matrix of the forecast error at time , for k The noise matrix at time -1, ∑(·) is the summation operation.

6. The position sensorless control method of a switched reluctance motor based on CKF according to claim 1, characterized in that: The method of calculating volume points, performing nonlinear transformation propagation of volume points using observation equations, calculating measurement estimates, calculating measurement error covariance and cross-covariance, and calculating Kalman gains, state estimates, and state error covariance matrices to implement measurement updates using a cubature Kalman filter algorithm includes: (I) Calculation of volume points The state prediction error covariance matrix obtained by equation (14) is decomposed by Cholesky and the new volume point is calculated by the third-order spherical radial volume criterion: (15) Where, for k time The Cholesky decomposition matrix of (II) According to the system described by equation (9), the volume point obtained above is transformed into Perform volume point nonlinear transformation propagation: (16) Where, for k Moment i Volume point measurement prediction value; (III) Perform weighted summation on the propagated points to obtain the measurement estimate: (17) Where, for k Time measurement prediction value; (IV) Calculate the measurement error covariance and cross-covariance: (18) Where, for k The moment measurement error covariance matrix, for k The cross-covariance matrix of the moment measurement error, R is the measurement noise matrix; (V) Calculate the k Momentary Kalman gain K k , state estimate and the state error covariance matrix : (19)。 7. The position sensorless control method of a switched reluctance motor based on CKF according to claim 1, characterized in that: Determining the estimated values ​​of the angular velocity and the velocity using the estimated state value includes: (20) in, The fourth element in the state vector is the estimated angular velocity, and the fifth element is the estimated angle.

Images