A sensorless control method for a permanent magnet linear motor across the entire speed range
By combining the weighted function switching strategy of high-frequency voltage injection method and adaptive adjustment boundary layer sliding mode observer method, the detection error and chattering problems of sensorless control of permanent magnet linear motor in the full speed domain are solved, and stable control in the full speed domain is achieved.
Patent Information
- Application Number
- CN202311067678.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-08-23
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-08-23
AI Technical Summary
Existing sensorless control algorithms cannot operate reliably across the full speed range of permanent magnet linear motors, especially at low and medium-high speeds where detection errors and system chattering issues exist, making it difficult to achieve stable control across the entire speed range.
An improved high-frequency voltage injection method and an adaptive boundary layer sliding mode observer method, combined with a weighted function switching strategy, are used to estimate the position and velocity of the mover in the low-speed and medium-to-high-speed stages, respectively. The sensorless control of the entire speed domain is achieved by smoothly switching between the two algorithms.
It improves the control accuracy and robustness of permanent magnet linear motors across the entire speed range, effectively suppresses system chattering, adapts to the speed and position estimation of the mover under different working conditions, and achieves a smooth transition from low speed to medium and high speed.
Smart Images

Figure CN117134667B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of permanent magnet linear motor control, specifically relating to a sensorless control method for a permanent magnet linear motor across the entire speed range. Background Technology
[0002] Permanent magnet linear motors (PMLMs) eliminate the need for numerous transmission mechanisms, reducing system size and improving efficiency. They are widely used in various transmission fields, including national defense, high-end CNC machine tools, semiconductor processing equipment, and transportation. While using mechanical position sensors for speed and position detection in motor vector control systems increases system cost and makes them unsuitable for extremely harsh working environments, sensorless control technology addresses these issues, enhancing the reliability and adaptability of motor control systems.
[0003] Currently, no sensorless control algorithm can reliably operate across the entire speed domain. Although efforts are being made to extend the mover position estimation algorithm in the mid-to-high speed domain to the zero-low speed stage, the amplitude of the fundamental frequency signal decreases significantly due to the reduced speed. System noise and detection errors make it difficult for these methods to extract the fundamental frequency signal for position estimation. To achieve mover position estimation across the entire speed domain, sensorless control algorithms in the low-speed and mid-to-high speed domains are typically combined to achieve complementary advantages between the two algorithms; this approach is called composite control. There are three main composite control strategies in the speed transition zone: single-point switching, hysteresis switching, and weighted switching. The single-point switching method sets a speed threshold as the switching point, at which the low-speed and mid-to-high-speed sensorless control algorithms are switched. This method is simple and easy to implement, but determining the switching point requires repeated experiments based on actual conditions. An inappropriate switching point can lead to increased jitter during switching, affecting the performance of the entire motor control system. Hysteresis switching involves setting a hysteresis switching interval within which two sensorless control algorithms (low-speed and medium-to-high-speed) estimate the mover speed and position through hysteresis switching. Because the estimated mover speed and position differ between the two algorithms at the switching point, hysteresis switching may cause fluctuations in speed and position within the switching interval. Weighted switching works similarly to hysteresis switching, but the difference lies in that the speed and position estimates in the speed transition interval are provided by a linear combination of the estimates from the two control algorithms. This invention uses a third weighted switching strategy for composite control. Summary of the Invention
[0004] The purpose of this invention is to solve the problem that there is currently no sensorless control algorithm that can reliably operate in the full speed range of a permanent magnet linear motor, and to provide a sensorless control method for a permanent magnet linear motor in the full speed range.
[0005] Considering the saturation effect of the magnetic circuit, cross-coupled inductance, phase lag caused by the band-pass filter (BPF), and errors in mover position estimation during low-speed operation, this invention proposes an improved high-frequency voltage injection method. First, the concept of dynamic inductance is introduced, and a mathematical model of the permanent magnet linear motor considering the nonlinear changes in inductance is established. Then, the BPF is replaced with a two-stage cascaded time-delayed filter (CTDF), which significantly increases the amplitude of the extracted high-frequency current response. Furthermore, a synchronous shaft high-pass filter is used to extract the useful clockwise rotation component of the high-frequency current response. Finally, a mover position estimation error compensation strategy is proposed to calculate and compensate for errors in real time, improving the sensorless control and detection accuracy of the permanent magnet linear motor. This invention improves upon existing low-speed sensorless control algorithms, further enhancing control speed and position detection performance.
[0006] During medium-to-high-speed operation, traditional sliding mode observers suffer from severe system chattering and phase lag. To suppress system chattering and improve the accuracy of mover velocity and position estimation, an improved sliding mode observer is proposed. This invention replaces the switching function with a designed adaptively adjustable boundary layer inverse hyperbolic tangent function, replaces the low-pass filter with an adaptive filter, and obtains the mover's position information through a phase-locked loop. The method of this invention can effectively suppress system chattering, exhibits good estimation accuracy and robustness, and can adapt to mover velocity and position estimation under different operating conditions.
[0007] Both low-speed and medium-to-high-speed position detection methods have their limitations: the high-frequency signal injection method becomes difficult to extract high-frequency current response signals when the motor is running at medium-to-high speeds due to the increase in the fundamental frequency, thus rendering it unsuitable for medium-to-high speed stages; the sliding mode observer method is unsuitable for low-speed operation because the back electromotive force is very small and difficult to observe at low speeds. Therefore, this invention combines the two algorithms and designs a weighted function switching strategy. The high-frequency voltage injection method is used at low speeds, while the sliding mode observer method is used at medium-to-high speeds. In the transition zone from low speed to medium-to-high speed, a linear combination of the two algorithms provides the velocity and position information of the mover, thereby achieving sensorless composite control across the entire speed range.
[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:
[0009] A sensorless control method for a permanent magnet linear motor across its entire speed range, the method comprising the following steps:
[0010] Step 1: When the permanent magnet linear motor is running at low speed, the position and speed of the permanent magnet linear motor mover are estimated using the high-frequency signal injection method;
[0011] Step 2: When the permanent magnet linear motor is running at medium and high speed, the position and speed of the permanent magnet linear motor mover are estimated by using the adaptive adjustment boundary layer sliding mode observer method;
[0012] Step 3: Employ a weighted function switching strategy to achieve a smooth switch between the two algorithms in Step 1 and Step 2.
[0013] Furthermore, the specific process of step one is as follows:
[0014] First, a dynamic inductance is defined. Then, considering the nonlinear change of the inductance in a permanent magnet linear motor, a high-frequency sinusoidal voltage signal is injected into the α and β axes respectively, and the high-frequency current response under the injected high-frequency sinusoidal voltage vector is obtained.
[0015] The definition of dynamic inductance is:
[0016]
[0017] Among them, f d (i d i q ) and f q (i d i q ) represent the dynamic self-inductance of the d and q axes, respectively; f dq (i d i q ) and f qd (i d i q ) represent the dynamic mutual inductance of the dq axes, respectively; ψ d and ψ q i represents the components of the stator flux linkage in the dq coordinate system; d and i q These are the components of the stator current in the dq coordinate system;
[0018] High-frequency current response is extracted by replacing the band-pass filter (BPF) with a two-stage cascaded pure time-delay filter.
[0019] The clockwise rotating high-frequency current component is extracted using a synchronous shaft high-pass filter and used to estimate the velocity and position of the mover;
[0020] The final velocity and position estimates are made using an Extended State Observer (ESO).
[0021] Finally, considering the nonlinear change in inductance, the final estimated mover position will have a position estimation error Δθ, which is expressed as:
[0022]
[0023] Where: θ M L is the angle between the clockwise rotating high-frequency current vector A and the clockwise rotating composite vector; d It is the d-axis inductance;
[0024] To address this error, a mover position estimation error compensation strategy is proposed to achieve real-time calculation and compensation of the error.
[0025] Furthermore, the specific process of step two is as follows:
[0026] First, an adaptively adjustable boundary layer inverse hyperbolic tangent function F(s) was designed, whose expression is:
[0027]
[0028]
[0029] Where, l is the set initial boundary layer thickness; S α S β These are the error values of the stator current along the αβ axis, respectively;
[0030] Next, the adaptive filter is designed, and its expression is as follows:
[0031]
[0032] Among them, E α E β The back potentials along the α and β axes are respectively; k f This is for the adaptive filter gain;
[0033] ω e This is the equivalent electrical angular velocity of the permanent magnet linear motor;
[0034] Finally, a phase-locked loop (PLL) method is used to obtain the position information of the mover.
[0035] Furthermore, the specific process of step three is as follows:
[0036] When the permanent magnet linear motor is running at low speed, the speed and position estimates of the mover are provided by the high-frequency signal injection method; when it is running at medium and high speed, they are provided by the sliding mode observer method; and in the transition zone between low speed and medium and high speed, they are provided by a linear combination of the output estimates of the high-frequency signal injection method and the sliding mode observer method.
[0037] The expression for the weighting function is as follows:
[0038]
[0039] Where, θ HFI θ SMOThese are the mover position estimates output by the high-frequency signal injection method and the sliding mode observer method, respectively; v HFI v SMO ζ represents the estimated mover velocity outputs from the high-frequency signal injection method and the sliding mode observer method, respectively; ζ is the weighting coefficient; θ and v are the mover position angle and mover velocity, respectively.
[0040] Compared with the prior art, the beneficial effects of the present invention are:
[0041] 1) This invention takes into account that the ideal mathematical model of PMLM is established on the premise of neglecting core saturation, and it is generally believed that L d and L q All parameters are constants, and the issue of cross-coupling inductance is rarely considered. Therefore, a mathematical model for PMLM considering the nonlinear variation of inductance was established. This solves the problems arising from the nonlinear characteristics of the magnetic circuit and ferromagnetic materials, the influence of stator current on the magnetic circuit during PMLM operation, and the inductance parameter L under saturation effects. d and L q The variation with stator current has a significant impact on the operating performance of the motor.
[0042] 2) In the low-speed operation of the permanent magnet linear motor, the present invention replaces the BPF with a two-stage series pure time delay filter (CTDF). This method can increase the amplitude of the extracted high-frequency current response by several times. Compared with the amplitude attenuation and phase lag caused by the BPF, which affect the motion speed and position estimation results, the present method has obvious advantages.
[0043] 3) The present invention proposes a mover position estimation error compensation strategy for low-speed operation of permanent magnet linear motor, which can perform real-time calculation and compensation for errors caused by nonlinear changes in inductance, thereby improving the estimation accuracy of the control algorithm.
[0044] 4) The adaptive adjustment boundary layer inverse hyperbolic tangent function designed in this invention for high-speed operation of permanent magnet linear motor can automatically adjust the thickness of the boundary layer according to the error of the sliding surface. When the error of the sliding surface is small, it will not cause the boundary layer thickness to be too large, thus affecting the approach speed. Moreover, the error of the sliding surface is affected by speed, thrust and other parameters, and can be adjusted under different working conditions.
[0045] 5) In the high-speed operation of the permanent magnet linear motor, the present invention uses an adaptive filter to replace the LPF, which effectively avoids the phase lag phenomenon; and uses a phase-locked loop to extract position information, which effectively avoids the drawback of chattering in the system amplified by the arctangent function.
[0046] 6) This invention combines low-speed and medium-high-speed control algorithms to complement each other's advantages and invents a transition switching method between low speed and medium-high speed (expressed through a weighted function), thereby realizing sensorless composite control of permanent magnet linear motor across the entire speed range. Attached Figure Description
[0047] Figure 1 This is a structural diagram of the sensorless control principle of a permanent magnet linear motor across the entire speed range;
[0048] Figure 2 This is a schematic diagram of the high-frequency current vector during low-speed operation;
[0049] Figure 3 This is a block diagram of the CTDF principle structure during low-speed operation;
[0050] Figure 4 This is a block diagram illustrating the principle of extracting high-frequency current signals during clockwise rotation at low speeds.
[0051] Figure 5 It is the inverse hyperbolic tangent function curve with a fixed boundary layer thickness during medium- and high-speed operation;
[0052] Figure 6 This is a diagram of the phase-locked loop structure during medium- and high-speed operation;
[0053] Figure 7 This is a schematic diagram of the weighted function switching principle;
[0054] Figure 8 This is a comparison chart of the estimated position and the actual position of a permanent magnet linear motor under a given ramp signal; because the estimation is relatively accurate, the estimated curve and the actual curve basically coincide.
[0055] Figure 9 This is the position error curve of a permanent magnet linear motor under a given ramp signal;
[0056] Figure 10 This is the speed error curve of a permanent magnet linear motor under a given ramp signal;
[0057] Figure 11 This is a comparison chart of the estimated position and the actual position of a permanent magnet linear motor under a given sinusoidal signal; because the estimation is relatively accurate, the estimated curve and the actual curve basically coincide.
[0058] Figure 12 This is the position error curve of a permanent magnet linear motor under a given sinusoidal signal;
[0059] Figure 13 It is the speed error curve of a permanent magnet linear motor under a given sinusoidal signal. Detailed Implementation
[0060] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the invention, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0061] Specific Implementation Method 1: This implementation method describes a sensorless control method for a permanent magnet linear motor across the entire speed range, the method comprising the following steps:
[0062] Step 1: When the permanent magnet linear motor is running at low speed, the position and speed of the permanent magnet linear motor mover are estimated using the high-frequency signal injection method;
[0063] Step 2: When the permanent magnet linear motor is running at medium and high speed, the position and speed of the permanent magnet linear motor mover are estimated by using the adaptive adjustment boundary layer sliding mode observer method;
[0064] Step 3: Employ a weighted function switching strategy to achieve a smooth switch between the two algorithms in Step 1 and Step 2.
[0065] Furthermore, the specific process of step one (which is a concrete manifestation of the high-frequency signal injection method) is as follows:
[0066] First, a dynamic inductance is defined. Then, considering the nonlinear change of the inductance in a permanent magnet linear motor, a high-frequency sinusoidal voltage signal is injected into the α and β axes respectively, and the high-frequency current response under the injected high-frequency sinusoidal voltage vector is obtained.
[0067] The definition of dynamic inductance is:
[0068]
[0069] Among them, f d (i d i q ) and f q (i d i q ) represent the dynamic self-inductance of the d and q axes, respectively; f dq (i d i q ) and f qd (i d i q The dynamic mutual inductance of the d and q axes are respectively (usually f can be used as the dynamic mutual inductance of the d and q axes). dq (i d i q ) and f qd (i d i q (considered as equal); ψ d and ψq i represents the components of the stator flux linkage in the dq coordinate system; d and i q These are the components of the stator current in the dq coordinate system;
[0070] High-frequency current response is extracted by replacing the band-pass filter (BPF) with a two-stage cascaded pure time-delay filter.
[0071] The clockwise rotating high-frequency current component is extracted using a synchronous shaft high-pass filter and used to estimate the velocity and position of the mover;
[0072] The final velocity and position estimates are made using an Extended State Observer (ESO).
[0073] Finally, considering the nonlinear change in inductance, the final estimated mover position will have a position estimation error Δθ, which is expressed as:
[0074]
[0075] Where: θ M L is the angle between the clockwise rotating high-frequency current vector A and the clockwise rotating composite vector; d It is the d-axis inductance;
[0076] To address this error, a mover position estimation error compensation strategy is proposed to achieve real-time calculation and compensation of the error (implemented as described above).
[0077] Furthermore, the specific process of step two is as follows:
[0078] First, an adaptively adjustable boundary layer inverse hyperbolic tangent function F(s) was designed, whose expression is:
[0079]
[0080]
[0081] Where, l is the set initial boundary layer thickness; S α S β These are the error values of the stator current along the αβ axis, respectively;
[0082] Next, the adaptive filter is designed, and its expression is as follows:
[0083]
[0084] Among them, E α E β The back potentials along the α and β axes are respectively; k fThis is for the adaptive filter gain;
[0085] ω e This is the equivalent electrical angular velocity of the permanent magnet linear motor;
[0086] Finally, a phase-locked loop (PLL) method is used to obtain the position information of the mover.
[0087] Furthermore, the specific process of step three is as follows:
[0088] When the permanent magnet linear motor is running at low speed, the speed and position estimates of the mover are provided by the high-frequency signal injection method (i.e., step one). When it is running at medium and high speed, they are provided by the sliding mode observer method (i.e., step two). In the transition zone between low speed and medium and high speed, the estimated values of the output values of the high-frequency signal injection method and the sliding mode observer method (the two control algorithms mentioned above) are provided by a linear combination.
[0089] The expression for the weighting function is as follows:
[0090]
[0091] Where, θ HFI θ SMO These are the mover position estimates output by the high-frequency signal injection method and the sliding mode observer method, respectively; v HFI v SMO ζ represents the estimated mover velocity outputs from the high-frequency signal injection method and the sliding mode observer method, respectively; ζ is the weighting coefficient; θ and v are the mover position angle and mover velocity, respectively.
[0092] Example 1:
[0093] This embodiment discloses a sensorless control method for a permanent magnet linear motor across the entire speed range, the method comprising the following steps:
[0094] Step 1: When the permanent magnet linear motor is running at low speed, the improved high-frequency signal injection method is used to estimate the position and speed of the linear motor mover.
[0095] Step 2: When the permanent magnet linear motor is running at medium and high speed, the position and speed of the linear motor mover are estimated by using the adaptive adjustment boundary layer sliding mode observer method.
[0096] Step 3: Employ a weighted function switching strategy to achieve a smooth switch between the two algorithms in Step 1 and Step 2.
[0097] Furthermore, the specific process of step one is as follows:
[0098] First, after defining the concept of dynamic inductance, considering the nonlinear change of inductance in a permanent magnet linear motor, a high-frequency rotating sinusoidal voltage signal is injected into the α and β axes respectively, and the high-frequency current response under this injected high-frequency rotating sinusoidal voltage vector is obtained. The high-frequency current response obtained by injecting the high-frequency rotating sinusoidal voltage vector contains two components: one is a counterclockwise rotating high-frequency current vector; the other is a clockwise rotating high-frequency current vector. Only the clockwise rotating high-frequency current vector contains the position signal of the mover. The high-frequency current vector in the αβ stationary coordinate system (the amplitude I of the clockwise rotating high-frequency current) r The amplitude I of the counterclockwise rotating high-frequency current f ),like Figure 2 As shown.
[0099] The definition of dynamic inductance is:
[0100]
[0101] Among them, f d (i d i q ) and f q (i d i q ) are the dynamic self-inductances of the d and q axes, respectively; f dq (i d i q ) and f qd (i d i q The dynamic mutual inductances of the d and q axes are respectively, and f can usually be considered as the dynamic mutual inductance of the d and q axes. dq (i d i q ) and f qd (i d i q ) are considered equal; ψ d and ψ q i represents the components of the stator flux linkage in the dq coordinate system; d and i q These are the components of the stator current in the dq coordinate system;
[0102] The PMLM simplified high-frequency voltage model considering the nonlinear variation of inductance can be expressed as:
[0103]
[0104] Among them, u αh and u βh These are the high-frequency voltages along the α and β axes, respectively; i αh and i βh α and β are the high-frequency currents along the α and β axes, respectively; p is the differential operator.
[0105] First, define the two expressions L0 and L1 as follows:
[0106]
[0107] L αβ The inductance matrix in the αβ stationary coordinate system is expressed as follows:
[0108]
[0109] Where, θ e Let M be the angle between the d-axis and the α-axis; M is the mutual inductance along the dq-axis, M = f dq (i d i q );
[0110] The high-frequency current response under injected high-frequency rotating sinusoidal voltage vector is as follows:
[0111]
[0112] Among them, I rA I rB These are the amplitudes of the clockwise rotating high-frequency current components A and B, respectively; I f ω represents the amplitude of the high-frequency current rotating counterclockwise; h t represents the frequency of the injected high-frequency voltage signal; t represents time.
[0113] I rA I rB and I f The specific expression can be represented as:
[0114]
[0115] Among them, U h The amplitude of the injected high-frequency voltage signal.
[0116] Simultaneously, a two-stage series pure time-delay filter replaces the bandpass filter (BPF) to extract the high-frequency current response. The stator current of a PMLM contains high-frequency current components, fundamental current components, and harmonic currents at the switching frequency. While a BPF can extract the high-frequency current components, it introduces amplitude attenuation and phase lag, affecting the motioner velocity and position estimation results and introducing a certain degree of error. Therefore, this invention replaces the BPF with a two-stage series pure time-delay filter (CTDF), the principle structure of which is shown in the block diagram below. Figure 3 As shown. The method of the present invention can both increase the amplitude of the extracted high-frequency current response by a factor of two and avoid the phase lag problem caused by BPF, which is a huge advantage over BPF.
[0117] Assume the stator current expression along the αβ axis is:
[0118]
[0119] In the formula, i α i β These are the components of the stator current in the αβ coordinate system; I1 is the amplitude of the fundamental frequency current. θ represents the initial phase of the current at the fundamental frequency. M The angle between the clockwise rotating high-frequency current vector A and the clockwise rotating composite vector; The initial phase of the current at the fundamental frequency;
[0120] The high-frequency current response i obtained after CTDF filtering αh i βh It can be represented as:
[0121]
[0122] Subsequently, a synchronous shaft high-pass filter is used to extract the clockwise rotating high-frequency current component, which is then used to estimate the velocity and position of the mover. The high-frequency current response after CTDF filtering contains two components; the counterclockwise rotating high-frequency current component, which has no practical effect, needs to be filtered out to obtain the clockwise rotating component. This invention utilizes a synchronous shaft high-pass filter, such as... Figure 4 The diagram shows the principle block diagram of the clockwise rotation high-frequency current signal extraction process. The stator current in the αβ axis system is first processed by CTDF pure delay filtering to obtain the high-frequency current response signal, and then filtered by the synchronous axis high-pass filter to finally obtain the clockwise rotation high-frequency current signal.
[0123] The high-frequency current response in the mn coordinate system is expressed as follows:
[0124]
[0125] Among them, i mh i nh These are the high-frequency current components in the mn coordinate system, respectively;
[0126] A clockwise rotating high-frequency current signal in the mn coordinate system can be represented as:
[0127]
[0128] Among them, i mhr and i nhr These are the clockwise rotating high-frequency current components in the mn coordinate system;
[0129] Transforming the clockwise rotation component in the mn coordinate system to the αβ stationary coordinate system yields the clockwise high-frequency current component i. αhr iβhr Its expression is:
[0130]
[0131] Secondly, the extended state observer (ESO) is used for final velocity and position estimation. When the dynamic mutual inductance along the dq axis is f(d,q)=(i d i q When θ is not zero, i.e., when considering the nonlinear change of inductance, M The value of ε is not 0, and the error signal ε can be expressed as:
[0132]
[0133] Finally, considering the nonlinear variation of the inductance, the final estimated mover position will have a position angle error. To address this error, a mover position estimation error compensation strategy is proposed, enabling real-time calculation and compensation of the error. The current control strategy employed in this invention is i... d With the *=0 control method, the d-axis inductance can be considered an approximately constant. Based on the definitions of apparent inductance and dynamic inductance, the d-axis dynamic inductance can be considered equal to the d-axis inductance. Then, by solving a system of equations, the q-axis dynamic inductance f can be obtained. q (i d i q ) and dq axis dynamic mutual inductance f dq (i d i q Substituting the obtained values of the q-axis dynamic inductance and dq-axis dynamic mutual inductance into the following formula yields the position estimation error Δθ, thus completing real-time compensation for the final position estimate. When considering the nonlinear variation of the inductance, the final estimated mover position will have a position estimation error Δθ, expressed as:
[0134]
[0135] Among them, L d It is the d-axis inductance.
[0136] Furthermore, the specific process of step two is as follows:
[0137] First, this invention uses a function from the quasi-sliding mode to replace the switching function, thereby improving the control performance of the sliding mode observer (SMO). An adaptive adjustment function F(s) for the boundary layer inverse hyperbolic tangent is designed, as follows: Figure 5 As shown, its expression is:
[0138]
[0139]
[0140] Where l is the set initial boundary layer thickness, S α S β These are the error values of the stator current along the αβ axis, respectively;
[0141] Subsequently, to address the phase lag phenomenon caused by the low-pass filter in the traditional sliding mode observer, this invention employs an adaptive filter to replace the low-pass filter. The design expression for the adaptive filter is:
[0142]
[0143] Among them, E α E β The back potentials along the α and β axes are respectively; k f This is for the adaptive filter gain;
[0144] ω e This is the equivalent electrical angular velocity of the permanent magnet linear motor;
[0145] Finally, a phase-locked loop (PLL) method (an existing method) is used to obtain the position information of the mover. The principle and structure are as follows: Figure 6 As shown.
[0146] Furthermore, the specific process of step three is as follows:
[0147] This invention employs a weighted function switching strategy to achieve smooth switching between two algorithms. The principle diagram of the weighted function switching is shown below. Figure 7 As shown. When the PMLM is operating at low speed, the velocity and position estimates of the mover are provided by the high-frequency signal injection method; when operating at medium to high speed, they are provided by the sliding mode observer method; and in the transition zone between low speed and medium to high speed, they are provided by a linear combination of the output estimates of the two control algorithms.
[0148] The expression for the weighting function is as follows:
[0149]
[0150] θ HFI θ SMO These are the mover position estimates output by the high-frequency signal injection method and the sliding mode observer method, respectively; v HFI v SMO ζ represents the mover velocity estimates output by the high-frequency signal injection method and the sliding mode observer method, respectively; ζ is the weighting coefficient.
[0151] The expression for the weighting coefficients is as follows:
[0152]
[0153] Where v1 is the lower limit of the transition zone switching speed; v2 is the upper limit of the transition zone switching speed; This is the speed estimate for PMLM.
[0154] Example 2:
[0155] according to Figure 1 The structure shown is used to build a full-speed-domain sensorless control simulation model of a permanent magnet linear motor for system simulation analysis, verifying the feasibility of vector control of the permanent magnet linear motor under various working conditions.
[0156] Operating Condition 1: Given a ramp speed signal, the speed accelerates uniformly from 0 to 1.8 m / s, then decelerates uniformly to 0. The permanent magnet linear motor then reverses direction, accelerating uniformly back to 1.8 m / s, before decelerating uniformly until it stops. The lower limit of the transition zone speed is set to 0.7 m / s, and the upper limit is set to 0.8 m / s.
[0157] according to Figure 8 , Figure 9 and Figure 10 It can be seen that the permanent magnet linear motor has good overall speed tracking performance, accurately tracking the given speed signal with a smooth speed curve and no obvious jitter. Throughout the speed change process, the speed estimation error remains within 0.004 m / s, and the position estimation error remains within 1.05°. The speed change has virtually no impact on the estimation error, and the speed and position estimation accuracy is high. The maximum total displacement is 1.08 m, which does not exceed the travel range.
[0158] Operating Condition 2: Given a sinusoidal velocity signal with an amplitude of 1.4 m / s. The lower limit of the transition zone switching velocity value is set to 0.7 m / s, and the upper limit of the velocity value is set to 0.8 m / s.
[0159] Depend on Figure 11 , Figure 12 and Figure 13 It can be seen that the permanent magnet linear motor has good overall speed tracking performance, accurately tracking the given speed signal with a smooth speed curve and no obvious jitter. Throughout the speed change process, the speed estimation error remains within 0.005 m / s, and the position estimation error remains within 0.7°. The speed change has virtually no impact on the estimation error, and the speed and position estimation accuracy is high. The maximum total displacement is 0.8 m, which does not exceed the travel range.
[0160] In summary, through simulation analysis, this invention demonstrates that when switching from a low-speed control algorithm to a medium-to-high-speed control algorithm, the speed error and position error do not diverge and always converge within a small value. The switching algorithm based on the weighted function can effectively meet the sensorless composite control requirements of PMLM in the full speed domain, thus verifying the effectiveness and feasibility of this invention.
[0161] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered illustrative and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.
[0162] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.
Claims
1. A sensorless control method for a permanent magnet linear motor across its entire speed range, characterized in that: The method includes the following steps: Step 1: When the permanent magnet linear motor is running at low speed, the position and speed of the permanent magnet linear motor mover are estimated using the high-frequency signal injection method; Step 2: During high-speed operation of the permanent magnet linear motor, the position and speed of the motor mover are estimated using the adaptive boundary layer sliding mode observer method; the specific process is as follows: First, an adaptive adjustment function for the boundary layer inverse hyperbolic tangent was designed. Its expression is: in, The initial boundary layer thickness is set. , The stator currents are respectively The error value of the shaft; Next, the adaptive filter is designed, and its expression is as follows: in, , They are respectively The back electromotive force of the shaft; For adaptive filter gain; This is the equivalent electrical angular velocity of the permanent magnet linear motor; Finally, the position information of the mover is obtained using a phase-locked loop method; Step 3: Employ a weighted function switching strategy to achieve a smooth switch between the two algorithms in Step 1 and Step 2.
2. The sensorless control method for a permanent magnet linear motor across the entire speed range according to claim 1, characterized in that: The specific process of step one is as follows: First, define the dynamic inductance. Then, considering the nonlinear change of the inductance in a permanent magnet linear motor, respectively... , A high-frequency rotating sinusoidal voltage signal is injected into the shaft, and the high-frequency current response under this injected high-frequency rotating sinusoidal voltage vector is obtained; The definition of dynamic inductance is: in, and They are respectively The dynamic self-inductance of the shaft; and They are respectively Dynamic mutual inductance of shafts; and For stator flux linkage Components in the coordinate system; and The stator currents are respectively Components in the coordinate system; The high-frequency current response is extracted by replacing the bandpass filter with a two-stage cascaded pure time-delay filter; The clockwise rotating high-frequency current component is extracted using a synchronous shaft high-pass filter and used to estimate the velocity and position of the mover; The final velocity and position estimates are made using an extended state observer. Finally, when considering the nonlinear variation of the inductance, the final estimated mover position will have a position estimation error. Its expression is: in: The angle between the clockwise rotating high-frequency current vector A and the clockwise rotating composite vector; for Shaft inductance; To address this error, a mover position estimation error compensation strategy is proposed to achieve real-time calculation and compensation of the error.
3. The sensorless control method for a permanent magnet linear motor across the entire speed range according to claim 1, characterized in that: The specific process of step three is as follows: When the permanent magnet linear motor is running at low speed, the speed and position estimates of the mover are provided by the high-frequency signal injection method; when it is running at medium and high speed, they are provided by the sliding mode observer method; and in the transition zone between low speed and medium and high speed, they are provided by a linear combination of the output estimates of the high-frequency signal injection method and the sliding mode observer method. The expression for the weighting function is as follows: in, , These are the mover position estimates output by the high-frequency signal injection method and the sliding mode observer method, respectively. , These are the mover velocity estimates output by the high-frequency signal injection method and the sliding mode observer method, respectively. These are weighting coefficients; , These are the mover position angle and the mover velocity, respectively.
