Model-free control method for synchronous reluctance motor

Abstract

The invention discloses a model-free control method for a synchronous reluctance motor, and the method comprises the steps: carrying out the modeling of a mathematical model of the synchronous reluctance motor based on a two-phase rotating coordinate system, and obtaining a mathematical model of the synchronous reluctance motor; then, according to the obtained mathematical model of the synchronous reluctance motor, the rotating speed of the motor is used as an input variable, the torque is used as an output variable, a hyper-local model is constructed, and a hyper-local variable power logarithmic reaching law sliding mode controller is constructed by adopting a variable power logarithmic reaching law; and finally, designing an improved extended Kalman filter observer to perform estimation and feed-forward compensation on unknown items of the super-local model to realize real-time updating of the super-local model, and then introducing a time-varying fading factor to improve the adaptive ability of the system to dynamic change. According to the method, real-time updating of the hyperlocal model is realized, and the adaptive capacity of the system to dynamic change is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-performance synchronous reluctance motor control, and in particular relates to a model-free control method for a synchronous reluctance motor. Background Art

[0002] Synchronous reluctance motors (SynRMs) offer significant advantages in a variety of applications due to their unique design and performance characteristics. Their simple structure and lack of permanent magnets not only reduce costs but also minimize maintenance requirements. In terms of efficiency, SRMs maintain high efficiency across a wide speed range, which is crucial for energy conservation and performance optimization. Therefore, developing high-performance control strategies is crucial to improving the performance and efficiency of SRMs.

[0003] Maximum Torque-to-Amp (MTPA) control for synchronous reluctance motors provides an efficient and precise motor control strategy that optimizes motor performance by independently regulating the motor's magnetic flux and torque current. This control method allows the motor to deliver high efficiency and dynamic response across its entire operating range, including low and zero speeds. MTPA control specifically optimizes current distribution, ensuring the motor produces maximum torque output for a given voltage, which is critical for applications requiring rapid acceleration.

[0004] However, torque in synchronous reluctance motor MTPA control is achieved by adjusting the parameters of a proportional-integral (PI) controller. While this approach is widely adopted in many applications for its simplicity and reliability, it also has drawbacks, particularly in applications requiring high performance and high precision. PI controllers lack fast response speed, struggle to completely eliminate steady-state errors, and suffer from poor interference immunity. PI controller performance is highly dependent on the adjustment of the proportional and integral gains, requiring complex debugging and optimization. The nonlinear characteristics of synchronous reluctance motors, such as magnetic saturation and cross-coupling effects, also make it difficult to achieve optimal performance with PI controllers. Under load fluctuations, the PI controller may need to be retuned to maintain performance, which is difficult to achieve in practical applications. PI controllers do not provide optimal dynamic performance, particularly in systems requiring fast and precise response. They lack adaptive capabilities and cannot automatically adjust to changes in system parameters or external conditions. Summary of the Invention

[0005] The purpose of the present invention is to provide a model-free control method for a synchronous reluctance motor, thereby realizing real-time updating of a super-local model and improving the adaptive capability of the system to dynamic changes.

[0006] The technical solution adopted by the present invention is a model-free control method for a synchronous reluctance motor, which is specifically implemented according to the following steps:

[0007] Step 1: Modeling a mathematical model of a synchronous reluctance motor based on a two-phase rotating coordinate system to obtain a mathematical model of the synchronous reluctance motor;

[0008] Step 2: Based on the mathematical model of the synchronous reluctance motor obtained in step 1, a hyperlocal model is constructed with the motor speed as the input variable and the torque as the output variable, and a hyperlocal variable power logarithmic reaching law sliding mode controller is constructed using the variable power logarithmic reaching law;

[0009] Step 3: Design an improved extended Kalman filter observer to estimate the unknown terms of the hyperlocal model and perform feedforward compensation to achieve real-time update of the hyperlocal model. Then, introduce a time-varying fading factor to improve the system's adaptability to dynamic changes.

[0010] The present invention is also characterized in that:

[0011] Step 1 is implemented as follows:

[0012] In the two-phase rotating coordinate system, the mathematical model of the synchronous reluctance motor is:

[0013]

[0014] T e =1.5n p (ψ d i q -ψ q i d ) (3)

[0015]

[0016] Where R s represents the stator resistance, u d 、u q represents the d and q axis stator voltage components, i d 、i q represents the d and q axis stator current components, ψ d , ψ q represents the d and q axis stator flux components, ω e Indicates the electrical angular velocity, L d , L q represents the d and q axis stator inductance components, T e Represents electromagnetic torque, n p represents the number of pole pairs, T l represents the load torque, B represents the friction coefficient, and J represents the moment of inertia.

[0017] Step 2 is implemented as follows:

[0018] Step 201: Establish a hyperlocal model of a single-input single-output nonlinear system:

[0019]

[0020] Where, is the time derivative of the state variable x, x is the system state variable, u is the system input, κ is the system state variable gain, γ is the non-physical constant gain, and f is the unknown part of the system, including all structural parameters;

[0021] Step 202: Based on the input and output data of the synchronous reluctance motor speed loop control, a speed loop super local model is constructed as follows:

[0022]

[0023] Where, ω e represents the electrical angular velocity, T e represents the electromagnetic torque, and f is the unknown part of the system;

[0024] Step 203: Combine the hyperlocal model control with the sliding mode control. According to equation (6), a speed sliding mode feedback controller based on the hyperlocal model is designed as follows:

[0025]

[0026] Where, T e * is the torque reference value, is the reference rotor electrical angular velocity differential, is the output of the super-local model sliding mode controller to be designed, is the unknown part of the new hyperlocal model to be observed;

[0027] Step 204: Define the speed error as:

[0028]

[0029] In order to reduce the steady-state error, a sliding surface with e as the independent variable is designed:

[0030]

[0031] Derivative of the above formula:

[0032]

[0033] Substituting (6), (7), (8) into (10), we obtain:

[0034]

[0035] In order to ensure the effectiveness of the observer and the high efficiency of the speed loop tracking, have to:

[0036]

[0037] Step 205: Control the nonlinearity of the reaching law by adjusting α.

[0038] Step 205 is specifically implemented according to the following steps:

[0039] The specific formula is as follows:

[0040]

[0041] Substituting formula (13) into formula (12), we get:

[0042]

[0043] According to formula (14), the input of the sliding mode controller is:

[0044]

[0045] Proof of stability:

[0046] Define the Lyapunov function as:

[0047]

[0048] Derivative of formula (16) yields:

[0049]

[0050] According to formula (13), ssgn(s)≥0, k1>0, k2>0, we can get Therefore, the selected reaching law satisfies the sliding mode reachability condition, that is, the sliding mode controller is asymptotically stable. The proof is complete.

[0051] Substituting Equation (17) into Equation (7), the speed sliding mode controller expression based on the super-local model is obtained as follows:

[0052]

[0053] T e * is the torque reference value.

[0054] Step 3 is implemented as follows:

[0055] Step 301: Based on the state space equation of the synchronous reluctance motor, the mathematical form of the extended Kalman filter EKF for nonlinear system state estimation is:

[0056]

[0057] Where A, B, and H represent the system matrix, input matrix, and output matrix respectively; x, u, and y represent the state variable, input variable, and output variable respectively; W and V represent the input noise and output noise respectively;

[0058] Discretize (19) and we get:

[0059]

[0060] Where,

[0061] x(k)=[ω e (k),f(k)] T (twenty one)

[0062] u(k)=ω e (k) (22)

[0063] Step 302: Linearize the nonlinear system at each moment by calculating the Jacobian matrix of the system model:

[0064]

[0065] Where, F k and H k are the Jacobian matrices of the system model and observation model, respectively;

[0066] Step 303: State and error covariance prediction:

[0067] Calculate the predicted value of the state variable

[0068]

[0069] Calculate the error covariance matrix predicted value

[0070]

[0071] Where χ(k) is the time-varying fading factor, and its value is ≥1;

[0072] Step 304: Calculate the time-varying fading factor;

[0073] Step 305: Calculate the gain matrix

[0074]

[0075] Step 306: Correction of predicted value:

[0076] Correction of predicted values of state variables:

[0077]

[0078] Error covariance matrix prediction value correction:

[0079]

[0080] In the above formula, “~” represents the predicted value, and “-” represents the corrected value.

[0081] Step 304 is specifically implemented according to the following steps:

[0082] 304a. Calculate the adaptive forgetting factor ρ k ;

[0083] 304b. Update the residual covariance matrix V0(k);

[0084] 304c. Dynamically adjust the fading factor λ0;

[0085] 304d. Solve the fading factor matrix by updating matrices N and M.

[0086] Step 304a is specifically implemented according to the following steps:

[0087]

[0088] Where, ρ base is the initial forgetting factor, which is 0.9. α is a positive constant that controls the sensitivity of the adjustment. k || 2 is the square norm of the residual, σ 2 is a positive constant used to adjust the scale of the residual effect.

[0089] Step 304b is specifically implemented according to the following steps:

[0090] Update the residual covariance matrix V0(k)

[0091]

[0092] Step 304c is specifically implemented according to the following steps:

[0093] Define the matrix N k and M k , by calculating N k and M k , dynamically adjust the fading factor λ0 so that the filter can adaptively adjust its performance according to the historical performance and current state of the system. The relationship is as follows:

[0094]

[0095] Where β is the weakening factor, R k is the measurement noise covariance matrix, C k is the observation matrix, Q k-1is the process noise covariance matrix.

[0096] Step 304d is specifically implemented according to the following steps:

[0097] By updating matrices N and M, solve the fading factor matrix:

[0098] A dynamic adjustment mechanism is introduced to adjust the calculation method of λ0 based on the system's historical performance and current status. A historical performance indicator p is introduced to reflect the performance stability of the system over a period of time:

[0099]

[0100] Where ε is an adjustment coefficient used to adjust the value of λ0 according to the historical performance p, which is estimated by calculating the stability of the residual over a period of time;

[0101] In determining λ k When λ is smoothed, a smoothing factor d is introduced to smooth λ k changes, to avoid the λ caused by short-term noise fluctuations k Drastic changes:

[0102]

[0103] Where d is a smoothing factor with a value between 0 and 1. The smoothing factor d helps the filter adjust λ when facing changes in noise characteristics. k value.

[0104] The beneficial effect of the present invention is that the model-free control method for synchronous reluctance motors provides a series of beneficial effects for the technical challenges existing in the control of synchronous reluctance motors. This method effectively reduces the dependence on the precise model of the motor by introducing a hyperlocal model and an improved EKF observer, making the control strategy more flexible and robust. In traditional synchronous reluctance motor control methods, accurate knowledge of motor parameters is essential, which may be difficult to obtain or maintain in actual applications. The present invention reduces the dependence on precise models by designing a hyperlocal model and only utilizing system input and output data without the need for other parameters. This model allows the controller to achieve effective control even when there is uncertainty in the motor parameters or when they are difficult to measure accurately.

[0105] The improved EKF observer can estimate and compensate for unknown disturbances in the system in real time, improving the control accuracy and system stability of the synchronous reluctance motor. This real-time update and compensation mechanism facilitates rapid response to system changes, reduces steady-state errors, and enhances the system's resilience to external disturbances. Furthermore, by introducing an adaptive forgetting factor, the EKF observer can more effectively handle system noise and model uncertainty, thereby enhancing the system's robustness.

[0106] In summary, this paper introduces a hyperlocal model and an improved EKF observer to provide a new approach for controlling synchronous reluctance motors. This approach demonstrates significant advantages in improving control accuracy, enhancing robustness, improving dynamic response, and reducing reliance on precise models. These technical improvements enable synchronous reluctance motors to achieve superior performance in a wider range of applications, particularly in cost-sensitive and performance-critical scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0107] Figure 1 This is a block diagram of the model-free control method for a synchronous reluctance motor in the present invention;

[0108] Figure 2 It is a recursive algorithm block diagram of the extended Kalman filter in the present invention;

[0109] Figure 3 This is a basic voltage vector block diagram of the two-level voltage source inverter in the present invention;

[0110] Figure 4 This is a block diagram of a two-level voltage source inverter in the present invention. DETAILED DESCRIPTION

[0111] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0112] The proposed synchronous reluctance motor control method achieves high-performance control of the motor by combining an improved extended Kalman filter (EKF) observer with super-local model-free sliding film control technology. This method first establishes a mathematical model based on the synchronous reluctance motor's two-phase rotating coordinate system, encompassing both the motor's electrical and mechanical characteristics. A variable power logarithmic reaching law is then introduced to achieve rapid current response and accurate tracking.

[0113] The improved EKF observer, the core of the present invention, improves the system's adaptability to dynamic changes by estimating the unknowns in the hyperlocal model in real time. These estimates are used to construct the hyperlocal model, which does not rely on precise motor parameters.

[0114] The MTPA module calculates the motor's electromagnetic torque, while the PI controller adjusts the current reference for precise control. Park and inverse Park transforms, as well as the Clark transform, are used to convert the motor's current and voltage signals between different coordinate systems. These transforms are key to achieving current and voltage control.

[0115] The SVPWM module generates switching signals for the inverter, converting DC power into AC power to drive the synchronous reluctance motor. The three-phase inverter receives control signals from the SVPWM module and supplies the appropriate voltage to the motor. A speed sensor measures the actual motor speed and converts it to the required control format using Clark transform.

[0116] The method of the present invention improves the control accuracy and stability of synchronous reluctance motors while reducing the accuracy requirements for the motor model, thus having broad application prospects. This method is particularly suitable for cost-sensitive and performance-critical scenarios. By combining a hyperlocal model with sliding film control, the present invention provides a new approach to improve the control performance of synchronous reluctance motors, enabling them to achieve better performance in a wider range of applications.

[0117] Example 1

[0118] The present invention provides a model-free control method for a synchronous reluctance motor. Figure 1 , specifically follow the steps below:

[0119] Step 1: Modeling a mathematical model of a synchronous reluctance motor based on a two-phase rotating coordinate system to obtain a mathematical model of the synchronous reluctance motor;

[0120] Step 1 is implemented as follows:

[0121] In the two-phase rotating coordinate system, the mathematical model of the synchronous reluctance motor is:

[0122]

[0123] T e =1.5n p (ψ d i q -ψ q i d ) (3)

[0124]

[0125] Where R s represents the stator resistance, u d 、u q represents the d and q axis stator voltage components, i d 、i q represents the d and q axis stator current components, ψ d , ψ q represents the d and q axis stator flux components, ω e Indicates the electrical angular velocity, L d , L q represents the d and q axis stator inductance components, T e Represents electromagnetic torque, n prepresents the number of pole pairs, T l represents the load torque, B represents the friction coefficient, and J represents the moment of inertia.

[0126] Combine Figure 2 、 Figure 3 、 Figure 4 , step 2, based on the mathematical model of the synchronous reluctance motor obtained in step 1, with the motor speed as the input variable and the torque as the output variable, a hyperlocal model is constructed, and a hyperlocal variable power logarithmic reaching law sliding mode controller is constructed using the variable power logarithmic reaching law;

[0127] Step 2 is implemented as follows:

[0128] Step 201: Establish a hyperlocal model of a single-input single-output nonlinear system:

[0129]

[0130] Where, is the time derivative of the state variable x, x is the system state variable, u is the system input, κ is the system state variable gain, γ is the non-physical constant gain, and f is the unknown part of the system, including all structural parameters such as uncertainty, disturbance and their derivatives.

[0131] Step 202: Using the concept of model-free control and based on the input and output data of the synchronous reluctance motor speed loop control, a speed loop super local model is constructed as follows:

[0132]

[0133] Where, ω e represents the electrical angular velocity, T e represents the electromagnetic torque, and f is the unknown part of the system;

[0134] Step 203: Combine the hyperlocal model control with the sliding mode control. According to equation (6), a speed sliding mode feedback controller based on the hyperlocal model is designed as follows:

[0135]

[0136] Where, T e * is the torque reference value, is the reference rotor electrical angular velocity differential, is the output of the super-local model sliding mode controller to be designed, is the unknown part of the new hyperlocal model to be observed;

[0137] Step 204: Define the speed error as:

[0138]

[0139] In order to effectively reduce the steady-state error, a sliding surface with e as the independent variable is designed:

[0140]

[0141] Derivative of the above formula:

[0142]

[0143] Substituting (6), (7), (8) into (10), we obtain:

[0144]

[0145] In order to ensure the effectiveness of the observer and the high efficiency of the speed loop tracking, have to:

[0146]

[0147] Step 205. In order to reduce the chattering problem that occurs during the control process, the present invention proposes a new convergence law of a variable power logarithmic function. The logarithmic function changes smoothly near the origin. When the system state approaches the sliding surface, the correction force of the controller will not be too large, thereby avoiding overcorrection and possible oscillations. When s increases, the value of the logarithmic function increases, making the controller respond more quickly to larger s and accelerating the convergence of the system state to the sliding surface. Near the origin, the curve of the logarithmic function is smooth, which helps to reduce the sensitivity of the controller to small errors, thereby reducing the chattering of the system near the sliding surface. The log in the variable power logarithmic function b (cs α The term (+1) introduces nonlinearity, which helps provide stronger correction force when s is large and weaker correction force when s is small, further suppressing chatter. Gain coefficients k1 and k2 adjust the controller's response speed and chatter suppression capabilities based on specific system requirements. The logarithmic base b and coefficient c influence the shape of the logarithmic function, thereby affecting controller performance. A positive even value α ensures smooth convergence, while adjusting α can control the degree of nonlinearity in the convergence law.

[0148] Step 205 is specifically implemented according to the following steps:

[0149] The specific formula is as follows:

[0150]

[0151] Substituting formula (13) into formula (12), we get:

[0152]

[0153] According to formula (14), the input of the sliding mode controller is:

[0154]

[0155] Proof of stability:

[0156] Define the Lyapunov function as:

[0157]

[0158] Derivative of formula (16) yields:

[0159]

[0160] According to formula (13), ssgn(s)≥0, k1>0, k2>0, we can get Therefore, the selected reaching law satisfies the sliding mode reachability condition, that is, the sliding mode controller is asymptotically stable. The proof is complete.

[0161] Substituting Equation (17) into Equation (7), the speed sliding mode controller expression based on the super-local model is obtained as follows:

[0162]

[0163] T e * is the torque reference value.

[0164] Step 3: Design an improved extended Kalman filter observer to estimate the unknown terms of the hyperlocal model and perform feedforward compensation to achieve real-time update of the hyperlocal model. Then, introduce a time-varying fading factor to improve the system's adaptability to dynamic changes.

[0165] Step 3 is implemented as follows:

[0166] Step 301: Based on the state space equation of the synchronous reluctance motor, the mathematical form of the extended Kalman filter EKF for nonlinear system state estimation is:

[0167]

[0168] Where A, B, and H represent the system matrix, input matrix, and output matrix respectively; x, u, and y represent the state variable, input variable, and output variable respectively; W and V represent the input noise and output noise respectively;

[0169] Discretize (19) and we get:

[0170]

[0171] Where,

[0172] x(k)=[ωe (k),f(k)] T (twenty one)

[0173] u(k)=ω e (k) (22)

[0174] Step 302: Since the EKF is based on linear system theory, it is necessary to linearize the nonlinear system at each moment, which is achieved by calculating the Jacobian matrix of the system model:

[0175]

[0176] Where, F k and H k are the Jacobian matrices of the system model and observation model, respectively;

[0177] Step 303: State and error covariance prediction:

[0178] Calculate the predicted value of the state variable

[0179]

[0180] Calculate the error covariance matrix predicted value

[0181]

[0182] Where χ(k) is the time-varying fading factor, and its value is ≥1;

[0183] Step 304: Calculate the time-varying fading factor;

[0184] Step 304 is specifically implemented according to the following steps:

[0185] 304a. Calculate the adaptive forgetting factor ρ k ;

[0186] Step 304a is specifically implemented according to the following steps:

[0187] Traditional extended Kalman filters usually use a fixed forgetting factor, which is insensitive to changes in noise characteristics. This method works well when the noise characteristics are stable or change slowly, but the performance will degrade in environments where the noise characteristics change rapidly or the uncertainty is high. The adaptive forgetting factor ρ proposed in this paper is k According to the current residual e k The residual is the difference between the observed and predicted values, reflecting the system's error level at the current moment. By analyzing the size of the residual, we can infer whether the characteristics of the noise have changed. If the residual is large, the noise level has increased or the system model error has increased. In this case, it is necessary to reduce the forgetting factor to increase the filter's sensitivity to new observations.

[0188]

[0189] Where, ρ base is the initial forgetting factor, which is 0.9. α is a positive constant that controls the sensitivity of the adjustment. k || 2 is the square norm of the residual, σ 2 is a positive constant used to adjust the scale of the residual effect.

[0190] 304b. Update the residual covariance matrix V0(k);

[0191] Step 304b is specifically implemented according to the following steps:

[0192] Update the residual covariance matrix V0(k)

[0193]

[0194] 304c. Dynamically adjust the fading factor λ0;

[0195] Step 304c is specifically implemented according to the following steps:

[0196] Define the matrix N k and M k , the role of these two matrices in the Kalman filter is to help estimate and adjust the performance of the filter. By calculating N k and M k , dynamically adjust the fading factor λ0 so that the filter can adaptively adjust its performance according to the historical performance and current state of the system. The relationship is as follows:

[0197]

[0198] Where β is the weakening factor, R k is the measurement noise covariance matrix, C k is the observation matrix, Q k-1 is the process noise covariance matrix.

[0199] 304d. Solve the fading factor matrix by updating matrices N and M.

[0200] Step 304d is specifically implemented according to the following steps:

[0201] By updating matrices N and M, solve the fading factor matrix:

[0202] In the original method, λ0 is obtained by N k The trace of and the matrix M from time step 1 to k rTo increase its robustness, the present invention introduces a dynamic adjustment mechanism that adjusts the calculation method of λ0 based on the system's historical performance and current state. This mechanism is implemented by introducing a historical performance index p, which reflects the performance stability of the system over a period of time:

[0203]

[0204] Where ε is an adjustment coefficient used to adjust the value of λ0 according to the historical performance p, which is estimated by calculating the stability of the residual over a period of time;

[0205] In determining λ k When λ is λ, a smoothing factor d is introduced to smooth λ k changes, to avoid the λ caused by short-term noise fluctuations k Drastic changes:

[0206]

[0207] Where d is a smoothing factor with a value between 0 and 1. The smoothing factor d helps the filter to adjust λ more smoothly when facing changes in noise characteristics. k value.

[0208] Step 305: Calculate the gain matrix

[0209]

[0210] Step 306: Correction of predicted value:

[0211] Correction of predicted values of state variables:

[0212]

[0213] Error covariance matrix prediction value correction:

[0214]

[0215] In the above formula, “~” represents the predicted value, and “-” represents the corrected value.

[0216] Example 2

[0217] The present invention provides a model-free control method for a synchronous reluctance motor. Figure 1 , specifically follow the steps below:

[0218] Step 1: Modeling a mathematical model of a synchronous reluctance motor based on a two-phase rotating coordinate system to obtain a mathematical model of the synchronous reluctance motor;

[0219] Step 2: Based on the mathematical model of the synchronous reluctance motor obtained in step 1, a hyperlocal model is constructed with the motor speed as the input variable and the torque as the output variable, and a hyperlocal variable power logarithmic reaching law sliding mode controller is constructed using the variable power logarithmic reaching law;

[0220] Step 3: Design an improved extended Kalman filter observer to estimate the unknown terms of the hyperlocal model and perform feedforward compensation to achieve real-time update of the hyperlocal model. Then, introduce a time-varying fading factor to improve the system's adaptability to dynamic changes.

[0221] Example 3

[0222] The present invention provides a model-free control method for a synchronous reluctance motor. Figure 1 , specifically follow the steps below:

[0223] Step 1: Modeling a mathematical model of a synchronous reluctance motor based on a two-phase rotating coordinate system to obtain a mathematical model of the synchronous reluctance motor;

[0224] Step 1 is implemented as follows:

[0225] In the two-phase rotating coordinate system, the mathematical model of the synchronous reluctance motor is:

[0226]

[0227] T e =1.5n p (ψ d i q -ψ q i d ) (3)

[0228]

[0229] Where R s represents the stator resistance, u d 、u q represents the d and q axis stator voltage components, i d 、i q represents the d and q axis stator current components, ψ d , ψ q represents the d and q axis stator flux components, ω e Indicates the electrical angular velocity, L d , L q represents the d and q axis stator inductance components, T e Represents electromagnetic torque, n p represents the number of pole pairs, T l represents the load torque, B represents the friction coefficient, and J represents the moment of inertia.

[0230] Step 2: Based on the mathematical model of the synchronous reluctance motor obtained in step 1, a hyperlocal model is constructed with the motor speed as the input variable and the torque as the output variable, and a hyperlocal variable power logarithmic reaching law sliding mode controller is constructed using the variable power logarithmic reaching law;

[0231] Step 3: Design an improved extended Kalman filter observer to estimate the unknown terms of the hyperlocal model and perform feedforward compensation to achieve real-time update of the hyperlocal model. Then, introduce a time-varying fading factor to improve the system's adaptability to dynamic changes.

[0232] Example 4

[0233] The present invention provides a model-free control method for a synchronous reluctance motor. Figure 1 , specifically follow the steps below:

[0234] Step 1: Modeling a mathematical model of a synchronous reluctance motor based on a two-phase rotating coordinate system to obtain a mathematical model of the synchronous reluctance motor;

[0235] Step 1 is implemented as follows:

[0236] In the two-phase rotating coordinate system, the mathematical model of the synchronous reluctance motor is:

[0237]

[0238] T e =1.5n p (ψ d i q -ψ q i d ) (3)

[0239]

[0240] Where R s represents the stator resistance, u d 、u q represents the d and q axis stator voltage components, i d 、i q represents the d and q axis stator current components, ψ d , ψ q represents the d and q axis stator flux components, ω e Indicates the electrical angular velocity, L d , L q represents the d and q axis stator inductance components, T e Represents electromagnetic torque, n p represents the number of pole pairs, T l represents the load torque, B represents the friction coefficient, and J represents the moment of inertia.

[0241] Step 2: Based on the mathematical model of the synchronous reluctance motor obtained in step 1, a hyperlocal model is constructed with the motor speed as the input variable and the torque as the output variable, and a hyperlocal variable power logarithmic reaching law sliding mode controller is constructed using the variable power logarithmic reaching law;

[0242] Step 2 is implemented as follows:

[0243] Step 201: Establish a hyperlocal model of a single-input single-output nonlinear system:

[0244]

[0245] Where, is the time derivative of the state variable x, x is the system state variable, u is the system input, κ is the system state variable gain, γ is the non-physical constant gain, and f is the unknown part of the system, including all structural parameters such as uncertainty, disturbance and their derivatives.

[0246] Step 202: Using the concept of model-free control and based on the input and output data of the synchronous reluctance motor speed loop control, a speed loop super local model is constructed as follows:

[0247]

[0248] Where, ω e represents the electrical angular velocity, T e represents the electromagnetic torque, and f is the unknown part of the system;

[0249] Step 203: Combine the hyperlocal model control with the sliding mode control. According to equation (6), a speed sliding mode feedback controller based on the hyperlocal model is designed as follows:

[0250]

[0251] Where, T e * is the torque reference value, is the reference rotor electrical angular velocity differential, is the output of the super-local model sliding mode controller to be designed, is the unknown part of the new hyperlocal model to be observed;

[0252] Step 204: Define the speed error as:

[0253]

[0254] In order to effectively reduce the steady-state error, a sliding surface with e as the independent variable is designed:

[0255]

[0256] Derivative of the above formula:

[0257]

[0258] Substituting (6), (7), (8) into (10), we obtain:

[0259]

[0260] In order to ensure the effectiveness of the observer and the high efficiency of the speed loop tracking, have to:

[0261]

[0262] Step 205. In order to reduce the chattering problem that occurs during the control process, the present invention proposes a new convergence law of a variable power logarithmic function. The logarithmic function changes smoothly near the origin. When the system state approaches the sliding surface, the correction force of the controller will not be too large, thereby avoiding overcorrection and possible oscillations. When s increases, the value of the logarithmic function increases, making the controller respond more quickly to larger s and accelerating the convergence of the system state to the sliding surface. Near the origin, the curve of the logarithmic function is smooth, which helps to reduce the sensitivity of the controller to small errors, thereby reducing the chattering of the system near the sliding surface. The log in the variable power logarithmic function b (cs α The term (+1) introduces nonlinearity, which helps provide stronger correction force when s is large and weaker correction force when s is small, further suppressing chatter. Gain coefficients k1 and k2 adjust the controller's response speed and chatter suppression capabilities based on specific system requirements. The logarithmic base b and coefficient c influence the shape of the logarithmic function, thereby affecting controller performance. A positive even value α ensures smooth convergence, while adjusting α can control the degree of nonlinearity in the convergence law.

[0263] Step 205 is specifically implemented according to the following steps:

[0264] The specific formula is as follows:

[0265]

[0266] Substituting formula (13) into formula (12), we get:

[0267]

[0268] According to formula (14), the input of the sliding mode controller is:

[0269]

[0270] Proof of stability:

[0271] Define the Lyapunov function as:

[0272]

[0273] Derivative of formula (16) yields:

[0274]

[0275] According to formula (13), ssgn(s)≥0, k1>0, k2>0, we can get Therefore, the selected reaching law satisfies the sliding mode reachability condition, that is, the sliding mode controller is asymptotically stable. The proof is complete.

[0276] Substituting Equation (17) into Equation (7), the speed sliding mode controller expression based on the super-local model is obtained as follows:

[0277]

[0278] T e * is the torque reference value.

[0279] Step 3: Design an improved extended Kalman filter observer to estimate the unknown terms of the hyperlocal model and perform feedforward compensation to achieve real-time update of the hyperlocal model. Then, introduce a time-varying fading factor to improve the system's adaptability to dynamic changes.

[0280] Example 5

[0281] The present invention provides a model-free control method for a synchronous reluctance motor. Figure 1 , specifically follow the steps below:

[0282] Step 1: Modeling a mathematical model of a synchronous reluctance motor based on a two-phase rotating coordinate system to obtain a mathematical model of the synchronous reluctance motor;

[0283] Step 1 is implemented as follows:

[0284] In the two-phase rotating coordinate system, the mathematical model of the synchronous reluctance motor is:

[0285]

[0286] T e =1.5n p (ψ d i q -ψ q i d ) (3)

[0287]

[0288] Where R s represents the stator resistance, ud 、u q represents the d and q axis stator voltage components, i d 、i q represents the d and q axis stator current components, ψ d , ψ q represents the d and q axis stator flux components, ω e Indicates the electrical angular velocity, L d 、L q represents the d and q axis stator inductance components, T e Represents electromagnetic torque, n p represents the number of pole pairs, T l represents the load torque, B represents the friction coefficient, and J represents the moment of inertia.

[0289] Step 2: Based on the mathematical model of the synchronous reluctance motor obtained in step 1, a hyperlocal model is constructed with the motor speed as the input variable and the torque as the output variable, and a hyperlocal variable power logarithmic reaching law sliding mode controller is constructed using the variable power logarithmic reaching law;

[0290] Step 2 is implemented as follows:

[0291] Step 201: Establish a hyperlocal model of a single-input single-output nonlinear system:

[0292]

[0293] Where, is the time derivative of the state variable x, x is the system state variable, u is the system input, κ is the system state variable gain, γ is the non-physical constant gain, and f is the unknown part of the system, including all structural parameters such as uncertainty, disturbance and their derivatives.

[0294] Step 202: Using the concept of model-free control and based on the input and output data of the synchronous reluctance motor speed loop control, a speed loop super local model is constructed as follows:

[0295]

[0296] Where, ω e represents the electrical angular velocity, T e represents the electromagnetic torque, and f is the unknown part of the system;

[0297] Step 203: Combine the hyperlocal model control with the sliding mode control. According to equation (6), a speed sliding mode feedback controller based on the hyperlocal model is designed as follows:

[0298]

[0299] Where, T e *is the torque reference value, is the reference rotor electrical angular velocity differential, is the output of the super-local model sliding mode controller to be designed, is the unknown part of the new hyperlocal model to be observed;

[0300] Step 204: Define the speed error as:

[0301]

[0302] In order to effectively reduce the steady-state error, a sliding surface with e as the independent variable is designed:

[0303]

[0304] Derivative of the above formula:

[0305]

[0306] Substituting (6), (7), (8) into (10), we obtain:

[0307]

[0308] In order to ensure the effectiveness of the observer and the high efficiency of the speed loop tracking, have to:

[0309]

[0310] Step 205. In order to reduce the chattering problem that occurs during the control process, the present invention proposes a new convergence law of a variable power logarithmic function. The logarithmic function changes smoothly near the origin. When the system state approaches the sliding surface, the correction force of the controller will not be too large, thereby avoiding overcorrection and possible oscillations. When s increases, the value of the logarithmic function increases, making the controller respond more quickly to larger s and accelerating the convergence of the system state to the sliding surface. Near the origin, the curve of the logarithmic function is smooth, which helps to reduce the sensitivity of the controller to small errors, thereby reducing the chattering of the system near the sliding surface. The log in the variable power logarithmic function b (cs α The term (+1) introduces nonlinearity, which helps provide stronger correction force when s is large and weaker correction force when s is small, further suppressing chatter. Gain coefficients k1 and k2 adjust the controller's response speed and chatter suppression capabilities based on specific system requirements. The logarithmic base b and coefficient c influence the shape of the logarithmic function, thereby affecting controller performance. A positive even value α ensures smooth convergence, while adjusting α can control the degree of nonlinearity in the convergence law.

[0311] Step 205 is specifically implemented according to the following steps:

[0312] The specific formula is as follows:

[0313]

[0314] Substituting formula (13) into formula (12), we get:

[0315]

[0316] According to formula (14), the input of the sliding mode controller is:

[0317]

[0318] Proof of stability:

[0319] Define the Lyapunov function as:

[0320]

[0321] Derivative of formula (16) yields:

[0322]

[0323] According to formula (13), ssgn(s)≥0, k1>0, k2>0, we can get Therefore, the selected reaching law satisfies the sliding mode reachability condition, that is, the sliding mode controller is asymptotically stable. The proof is complete.

[0324] Substituting Equation (17) into Equation (7), the speed sliding mode controller expression based on the super-local model is obtained as follows:

[0325]

[0326] T e * is the torque reference value.

[0327] Step 3: Design an improved extended Kalman filter observer to estimate the unknown terms of the hyperlocal model and perform feedforward compensation to achieve real-time update of the hyperlocal model. Then, introduce a time-varying fading factor to improve the system's adaptability to dynamic changes.

[0328] Step 3 is implemented as follows:

[0329] Step 301: Based on the state space equation of the synchronous reluctance motor, the mathematical form of the extended Kalman filter EKF for nonlinear system state estimation is:

[0330]

[0331] Where A, B, and H represent the system matrix, input matrix, and output matrix respectively; x, u, and y represent the state variable, input variable, and output variable respectively; W and V represent the input noise and output noise respectively;

[0332] Discretize (19) and we get:

[0333]

[0334] Where,

[0335] x(k)=[ω e (k),f(k)] T (twenty one)

[0336] u(k)=ω e (k) (22)

[0337] Step 302: Since the EKF is based on linear system theory, it is necessary to linearize the nonlinear system at each moment, which is achieved by calculating the Jacobian matrix of the system model:

[0338]

[0339] Where, F k and H k are the Jacobian matrices of the system model and observation model, respectively;

[0340] Step 303: State and error covariance prediction:

[0341] Calculate the predicted value of the state variable

[0342]

[0343] Calculate the error covariance matrix predicted value

[0344]

[0345] Where χ(k) is the time-varying fading factor, and its value is ≥1;

[0346] Step 304: Calculate the time-varying fading factor;

[0347] Step 304 is specifically implemented according to the following steps:

[0348] 304a. Calculate the adaptive forgetting factor ρ k ;

[0349] 304b. Update the residual covariance matrix V0(k);

[0350] 304c. Dynamically adjust the fading factor λ0;

[0351] 304d. Solve the fading factor matrix by updating matrices N and M.

[0352] Step 305: Calculate the gain matrix

[0353]

[0354] Step 306: Correction of predicted value:

[0355] Correction of predicted values of state variables:

[0356]

[0357] Error covariance matrix prediction value correction:

[0358]

[0359] In the above formula, “~” represents the predicted value, and “-” represents the corrected value.

[0360] Example 6

[0361] The present invention provides a model-free control method for a synchronous reluctance motor. Figure 1 , specifically follow the steps below:

[0362] Step 1: Modeling a mathematical model of a synchronous reluctance motor based on a two-phase rotating coordinate system to obtain a mathematical model of the synchronous reluctance motor;

[0363] Step 1 is implemented as follows:

[0364] In the two-phase rotating coordinate system, the mathematical model of the synchronous reluctance motor is:

[0365]

[0366] T e =1.5n p (ψ d i q -ψ q i d ) (3)

[0367]

[0368] Where R s represents the stator resistance, u d 、u q represents the d and q axis stator voltage components, i d 、i q represents the d and q axis stator current components, ψ d , ψ q represents the d and q axis stator flux components, ω e Indicates the electrical angular velocity, Ld , L q represents the d and q axis stator inductance components, T e Represents electromagnetic torque, n p represents the number of pole pairs, T l represents the load torque, B represents the friction coefficient, and J represents the moment of inertia.

[0369] Step 2: Based on the mathematical model of the synchronous reluctance motor obtained in step 1, a hyperlocal model is constructed with the motor speed as the input variable and the torque as the output variable, and a hyperlocal variable power logarithmic reaching law sliding mode controller is constructed using the variable power logarithmic reaching law;

[0370] Step 2 is implemented as follows:

[0371] Step 201: Establish a hyperlocal model of a single-input single-output nonlinear system:

[0372]

[0373] Where, is the time derivative of the state variable x, x is the system state variable, u is the system input, κ is the system state variable gain, γ is the non-physical constant gain, and f is the unknown part of the system, including all structural parameters such as uncertainty, disturbance and their derivatives.

[0374] Step 202: Using the concept of model-free control and based on the input and output data of the synchronous reluctance motor speed loop control, a speed loop super local model is constructed as follows:

[0375]

[0376] Where, ω e represents the electrical angular velocity, T e represents the electromagnetic torque, and f is the unknown part of the system;

[0377] Step 203: Combine the hyperlocal model control with the sliding mode control. According to equation (6), a speed sliding mode feedback controller based on the hyperlocal model is designed as follows:

[0378]

[0379] Where, T e * is the torque reference value, is the reference rotor electrical angular velocity differential, is the output of the super-local model sliding mode controller to be designed, is the unknown part of the new hyperlocal model to be observed;

[0380] Step 204: Define the speed error as:

[0381]

[0382] In order to effectively reduce the steady-state error, a sliding surface with e as the independent variable is designed:

[0383]

[0384] Derivative of the above formula:

[0385]

[0386] Substituting (6), (7), (8) into (10), we obtain:

[0387]

[0388] In order to ensure the effectiveness of the observer and the high efficiency of the speed loop tracking, have to:

[0389]

[0390] Step 205. In order to reduce the chattering problem that occurs during the control process, the present invention proposes a new convergence law of a variable power logarithmic function. The logarithmic function changes smoothly near the origin. When the system state approaches the sliding surface, the correction force of the controller will not be too large, thereby avoiding overcorrection and possible oscillations. When s increases, the value of the logarithmic function increases, making the controller respond more quickly to larger s and accelerating the convergence of the system state to the sliding surface. Near the origin, the curve of the logarithmic function is smooth, which helps to reduce the sensitivity of the controller to small errors, thereby reducing the chattering of the system near the sliding surface. The log in the variable power logarithmic function b (cs α The term (+1) introduces nonlinearity, which helps provide stronger correction force when s is large and weaker correction force when s is small, further suppressing chatter. Gain coefficients k1 and k2 adjust the controller's response speed and chatter suppression capabilities based on specific system requirements. The logarithmic base b and coefficient c influence the shape of the logarithmic function, thereby affecting controller performance. A positive even value α ensures smooth convergence, while adjusting α can control the degree of nonlinearity in the convergence law.

[0391] Step 205 is specifically implemented according to the following steps:

[0392] The specific formula is as follows:

[0393]

[0394] Substituting formula (13) into formula (12), we get:

[0395]

[0396] According to formula (14), the input of the sliding mode controller is:

[0397]

[0398] Proof of stability:

[0399] Define the Lyapunov function as:

[0400]

[0401] Derivative of formula (16) yields:

[0402]

[0403] According to formula (13), ssgn(s)≥0, k1>0, k2>0, we can get Therefore, the selected reaching law satisfies the sliding mode reachability condition, that is, the sliding mode controller is asymptotically stable. The proof is complete.

[0404] Substituting Equation (17) into Equation (7), the speed sliding mode controller expression based on the super-local model is obtained as follows:

[0405]

[0406] T e * is the torque reference value.

[0407] Step 3: Design an improved extended Kalman filter observer to estimate the unknown terms of the hyperlocal model and perform feedforward compensation to achieve real-time update of the hyperlocal model. Then, introduce a time-varying fading factor to improve the system's adaptability to dynamic changes.

[0408] Step 3 is implemented as follows:

[0409] Step 301: Based on the state space equation of the synchronous reluctance motor, the mathematical form of the extended Kalman filter EKF for nonlinear system state estimation is:

[0410]

[0411] Where A, B, and H represent the system matrix, input matrix, and output matrix respectively; x, u, and y represent the state variable, input variable, and output variable respectively; W and V represent the input noise and output noise respectively;

[0412] Discretize (19) and we get:

[0413]

[0414] Where,

[0415] x(k)=[ω e (k),f(k)] T (twenty one)

[0416] u(k)=ω e (k) (22)

[0417] Step 302: Since the EKF is based on linear system theory, it is necessary to linearize the nonlinear system at each moment, which is achieved by calculating the Jacobian matrix of the system model:

[0418]

[0419] Where, F k and H k are the Jacobian matrices of the system model and observation model, respectively;

[0420] Step 303: State and error covariance prediction:

[0421] Calculate the predicted value of the state variable

[0422]

[0423] Calculate the error covariance matrix predicted value

[0424]

[0425] Where χ(k) is the time-varying fading factor, and its value is ≥1;

[0426] Step 304: Calculate the time-varying fading factor;

[0427] Step 304 is specifically implemented according to the following steps:

[0428] 304a. Calculate the adaptive forgetting factor ρ k ;

[0429] Step 304a is specifically implemented according to the following steps:

[0430] Traditional extended Kalman filters usually use a fixed forgetting factor, which is insensitive to changes in noise characteristics. This method works well when the noise characteristics are stable or change slowly, but the performance will degrade in environments where the noise characteristics change rapidly or the uncertainty is high. The adaptive forgetting factor ρ proposed in this paper is k According to the current residual e kThe residual is the difference between the observed and predicted values, reflecting the system's error level at the current moment. By analyzing the size of the residual, we can infer whether the characteristics of the noise have changed. If the residual is large, the noise level has increased or the system model error has increased. In this case, it is necessary to reduce the forgetting factor to increase the filter's sensitivity to new observations.

[0431]

[0432] Where, ρ base is the initial forgetting factor, which is 0.9. α is a positive constant that controls the sensitivity of the adjustment. k || 2 is the square norm of the residual, σ 2 is a positive constant used to adjust the scale of the residual effect.

[0433] 304b. Update the residual covariance matrix V0(k);

[0434] 304c. Dynamically adjust the fading factor λ0;

[0435] 304d. Solve the fading factor matrix by updating matrices N and M.

[0436] Step 305: Calculate the gain matrix

[0437]

[0438] Step 306: Correction of predicted value:

[0439] Correction of predicted values of state variables:

[0440]

[0441] Error covariance matrix prediction value correction:

[0442]

[0443] In the above formula, “~” represents the predicted value, and “-” represents the corrected value.

[0444] A synchronous reluctance motor is modeled, obtaining its mathematical model in a two-phase rotating coordinate system. Based on this mathematical model, a hyperlocal model is established, utilizing only system input and output data, without requiring other parameters, to construct a speed loop. A hyperlocal variable-power logarithmic reaching law sliding mode controller is constructed using a variable-power logarithmic reaching law. The stability of the controller is demonstrated using Lyapunov stability theory. Considering the unknowns in the hyperlocal model, an improved extended Kalman filter observer is designed to estimate them and perform feedforward compensation, enabling real-time updates of the hyperlocal model. This invention improves the control accuracy and stability of synchronous reluctance motors, effectively reducing reliance on precise motor models and possessing broad application prospects.

Claims

1. A model-free control method for a synchronous reluctance motor, characterized in that: Please follow the steps below to implement it: Step 1: Modeling a mathematical model of a synchronous reluctance motor based on a two-phase rotating coordinate system to obtain a mathematical model of the synchronous reluctance motor; Step 2: Based on the mathematical model of the synchronous reluctance motor obtained in step 1, a hyperlocal model is constructed with the motor speed as the input variable and the torque as the output variable, and a hyperlocal variable power logarithmic reaching law sliding mode controller is constructed using the variable power logarithmic reaching law; Step 3: Design an improved extended Kalman filter observer to estimate the unknown terms of the hyperlocal model and perform feedforward compensation to achieve real-time update of the hyperlocal model. Then, introduce a time-varying fading factor to improve the system's adaptability to dynamic changes.

2. The model-free control method for a synchronous reluctance motor according to claim 1, wherein: The step 1 is specifically implemented according to the following steps: In the two-phase rotating coordinate system, the mathematical model of the synchronous reluctance motor is: T e =1.5n p (ψ d I q -ψ q I d ) (3) Where R s represents the stator resistance, u d 、u q represents the d and q axis stator voltage components, i d 、i q represents the d and q axis stator current components, ψ d , ψ q represents the d and q axis stator flux components, ω e Indicates the electrical angular velocity, L d 、L q represents the d and q axis stator inductance components, T e Represents electromagnetic torque, n p represents the number of pole pairs, T l represents the load torque, B represents the friction coefficient, and J represents the moment of inertia.

3. The model-free control method for a synchronous reluctance motor according to claim 2, wherein: The step 2 is specifically implemented according to the following steps: Step 201: Establish a hyperlocal model of a single-input single-output nonlinear system: Where, is the time derivative of the state variable x, x is the system state variable, u is the system input, κ is the system state variable gain, γ is the non-physical constant gain, and f is the unknown part of the system, including all structural parameters; Step 202: Based on the input and output data of the synchronous reluctance motor speed loop control, a speed loop super local model is constructed as follows: Where, ω e represents the electrical angular velocity, T e represents the electromagnetic torque, and f is the unknown part of the system; Step 203: Combine the hyperlocal model control with the sliding mode control. According to equation (6), a speed sliding mode feedback controller based on the hyperlocal model is designed as follows: Where, is the torque reference value, is the reference rotor electrical angular velocity differential, is the output of the super-local model sliding mode controller to be designed, is the unknown part of the new hyperlocal model to be observed; Step 204: Define the speed error as: In order to reduce the steady-state error, a sliding surface with e as the independent variable is designed: Derivative of the above formula: Substituting (6), (7), (8) into (10), we obtain: In order to ensure the effectiveness of the observer and the high efficiency of the speed loop tracking, have to: Step 205: Control the nonlinearity of the reaching law by adjusting α.

4. The model-free control method for a synchronous reluctance motor according to claim 3, wherein: The step 205 is specifically implemented according to the following steps: The specific formula is as follows: Substituting formula (13) into formula (12), we get: According to formula (14), the input of the sliding mode controller is: Proof of stability: Define the Lyapunov function as: Derivative of formula (16) yields: According to formula (13), ssgn(s)≥0, k1>0, k2>0, we can get Therefore, the selected reaching law satisfies the sliding mode reachability condition, that is, the sliding mode controller is asymptotically stable. The proof is complete. Substituting Equation (17) into Equation (7), the speed sliding mode controller expression based on the super-local model is obtained as follows: is the torque reference value.

5. The model-free control method for a synchronous reluctance motor according to claim 4, characterized in that: The step 3 is specifically implemented according to the following steps: Step 301: Based on the state space equation of the synchronous reluctance motor, the mathematical form of the extended Kalman filter EKF for nonlinear system state estimation is: Where A, B, and H represent the system matrix, input matrix, and output matrix respectively; x, u, and y represent the state variable, input variable, and output variable respectively; W and V represent the input noise and output noise respectively; Discretize (19) and we get: Where, x(k)=[ω e (k),f(k)] T (21) u(k)=ω e (k) (22) Step 302: Linearize the nonlinear system at each moment by calculating the Jacobian matrix of the system model: Where, F k and H k are the Jacobian matrices of the system model and observation model, respectively; Step 303: State and error covariance prediction: Calculate the predicted value of the state variable Calculate the error covariance matrix predicted value Where χ(k) is the time-varying fading factor, and its value is ≥1; Step 304: Calculate the time-varying fading factor; Step 305: Calculate the gain matrix Step 306: Correction of predicted value: Correction of predicted values of state variables: Error covariance matrix prediction value correction: In the above formula, "~" represents the predicted value, and "-" represents the corrected value.

6. The model-free control method for a synchronous reluctance motor according to claim 5, characterized in that: Step 304 is specifically implemented according to the following steps: 304a. Calculate the adaptive forgetting factor ρ k ; 304b. Update the residual covariance matrix V0(k); 304c. Dynamically adjust the fading factor λ0; 304d. Solve the fading factor matrix by updating matrices N and M.

7. The model-free control method for a synchronous reluctance motor according to claim 6, wherein: Step 304a is specifically implemented according to the following steps: Where, ρ base is the initial forgetting factor, which is 0.

9. α is a positive constant that controls the sensitivity of the adjustment. k || 2 is the square norm of the residual, σ 2 is a positive constant used to adjust the scale of the residual effect.

8. The model-free control method for a synchronous reluctance motor according to claim 7, characterized in that: Step 304b is specifically implemented according to the following steps: Update the residual covariance matrix V0(k) 9. The model-free control method for a synchronous reluctance motor according to claim 8, characterized in that: Step 304c is specifically implemented according to the following steps: Define the matrix N k and M k , by calculating N k and M k , dynamically adjust the fading factor λ0 so that the filter can adaptively adjust its performance according to the historical performance and current state of the system. The relationship is as follows: Where β is the weakening factor, R k is the measurement noise covariance matrix, C k is the observation matrix, Q k-1 is the process noise covariance matrix.

10. The model-free control method for a synchronous reluctance motor according to claim 9, wherein: The step 304d is specifically implemented according to the following steps: By updating matrices N and M, solve the fading factor matrix: A dynamic adjustment mechanism is introduced to adjust the calculation method of λ0 based on the system's historical performance and current status. A historical performance indicator p is introduced to reflect the performance stability of the system over a period of time: Where ε is an adjustment coefficient used to adjust the value of λ0 according to the historical performance p, which is estimated by calculating the stability of the residual over a period of time; In determining λ k When λ is λ, a smoothing factor d is introduced to smooth λ k changes, to avoid the λ caused by short-term noise fluctuations k Drastic changes: Where d is a smoothing factor with a value between 0 and 1. The smoothing factor d helps the filter adjust λ when facing changes in noise characteristics. k value.

Images