A segmented winding permanent magnet linear synchronous motor cross-section current prediction control method

Abstract

The application discloses a cross-section current prediction control method for a segmented winding permanent magnet linear synchronous motor, which unifies the main section and the slave section variables into a sum-difference reconstruction coordinate, and uses the real segmented output history voltage for remapping when the mode is switched, so that the model consistency of one beat prediction compensation at the cross-section boundary can be improved. Since the application can explicitly introduce the inter-section coupling parameters or matrices related to the double-section overlap state, the adaptability of the double-section overlap area prediction model to the coupling change can be improved. Compared with the post-compensation of the inter-section parameter change only at the residual or total disturbance level, the application introduces the inter-section coupling parameters as the model parameters into the prediction model, so that the compensation effect is moved forward to the current prediction and voltage back calculation process. Meanwhile, when the online identification and residual observation enhancement are enabled, the compatibility between the inter-section coupling parameter update and the disturbance compensation can be improved.

Description

Technical Field

[0001] This invention belongs to the field of motor control technology, specifically relating to a cross-segment current prediction control method for a segmented winding permanent magnet linear synchronous motor. Background Technology

[0002] Permanent magnet linear synchronous motors are characterized by high thrust density, fast response speed, fewer mechanical transmission links, and high positioning accuracy, and are widely used in long-stroke conveyors, precision direct drives, and high-speed motion platforms. For long-stroke applications, a segmented stator power supply structure is usually adopted, that is, multiple stator segments are arranged along the direction of motion, and power is supplied to only one or more segments near the mover, in order to reduce the capacity of power devices and system losses.

[0003] In a segmented winding permanent magnet linear synchronous motor, when the mover is located above a single segment, the electromagnetic thrust can be provided independently by that segment. When the mover spans two adjacent segments, both the master and slave segments need to contribute power simultaneously. In this case, the equivalent electromagnetic coupling between the two winding segments and the mover changes with the mover's position and overlap state. If a single-segment model is still used separately, or if the two segments are simply regarded as two independent single-segment models, the boundaries of single-segment operation, double-segment overlapping operation, and mode switching are difficult to describe continuously in the same current loop model, which can easily lead to increased prediction errors, current surges, or voltage spikes.

[0004] Predictive current control methods can inversely derive voltage from the motor voltage equation, offering advantages such as fast response and low computational complexity, as seen in references [Zhou Shijiong, Li Yaohua, Shi Liming. Model-free predictive current control of segmented dual three-phase permanent magnet linear synchronous motor. Journal of Electrical Engineering, 2024, 39(4):996-1009] and [Liu Gang, Zhang Jing, Zheng Shiqiang. Deadbeat-free predictive current control of high-speed permanent magnet motor based on online parameter identification. Journal of Electrical Machines and Control, 2023, 27(9):98-108]. However, in digital control systems, sampling, calculation, and pulse width modulation updates typically cause the voltage calculated in the current cycle to act on the motor in the next control cycle, resulting in a one-beat execution delay. During segment switching, the actual voltage output to the physical segment in the previous control cycle still affects the current change in the next control cycle, while the current control mode and control variable expression have changed. If the correspondence between the actual segment output voltage of the previous cycle and the historical voltage reconstructed after switching is not addressed, one-beat predictive compensation may use historical voltages that do not match the current mode.

[0005] Furthermore, the inter-segment coupling parameters within the double-segment overlap region may vary with the mover position, overlap degree, or magnetic circuit state, making it difficult to cover the entire cross-segment process using only offline fixed parameters. Some predictive current control methods can compensate for parameter changes, coupling changes, and external disturbances through disturbance observation or extended state observation. However, if the inter-segment coupling changes in the double-segment overlap region are not explicitly entered into the predictive model as model parameters, the predictive model itself still struggles to reflect the coupling structure changes during the cross-segment process. The controller often has to perform post-compensation at the residual or total disturbance level. Online identification of inter-segment coupling parameters helps improve model consistency, but when high-frequency injection identification is used, the injected signal may also enter the model residual or disturbance observation channel. If the residual observer directly learns the residual containing the artificially injected component, it may affect the compatibility of disturbance compensation and inter-segment coupling parameter identification.

[0006] Therefore, it is necessary to provide a unified predictive current control method for segmented winding permanent magnet linear synchronous motors operating across segments, so that single-segment operation, double-segment overlapping operation and mode switching boundaries can share the same sum-difference reconstructive model framework, and maintain the consistency between the real historical voltage and the variable expression of the predictive model under the one-step execution delay of digital control. Summary of the Invention

[0007] In view of the above, the present invention provides a cross-segment current prediction control method for a segmented winding permanent magnet linear synchronous motor. By reconstructing the sum and difference of d and q axis variables of the master and slave segments, modeling the inter-segment coupling parameters, and compensating for one-step execution delay, the motor exhibits outstanding performance in current tracking, switching boundary current and voltage impact, inter-segment coupling parameter change trends, and speed ripple.

[0008] A method for predictive control of segmented current in a segmented permanent magnet linear synchronous motor includes: Determine the master and slave segments and operating modes of the two adjacent stator segments currently participating in control, and construct the sum and difference reconstructed current vector and its reference vector, as well as the sum and difference reconstructed voltage vector of the previous control cycle. A sum-difference reconstructed voltage model is established using inter-segment coupling parameters and sum-difference reconstructed current state vectors. By constructing the target current change rate and reconstructing the voltage vector based on the sum and difference of the previous control cycle, the sum and difference reconstructed control voltage vector of the current control cycle is calculated using the sum and difference reconstructed voltage model. This vector is then transformed into the d-axis voltage reference and q-axis voltage reference of the main and slave segments of the current control cycle for segmented control.

[0009] Furthermore, the sum and difference reconstructed current vector and its reference vector are constructed using the following expressions: , , , in: x To reconstruct the current vector through sum and difference, For the sum and difference reconstructed current reference vector, i d1 and i q1 These are the d-axis current and q-axis current of the main segment, respectively. i d2 and i q2 These are the d-axis current and q-axis current of the segment, respectively. and These are the d-axis current reference and q-axis current reference for the main segment, respectively. and These are the d-axis current reference and q-axis current reference for the segment, respectively. i dΣ and i qΣ These are the d-axis and q-axis components of the modulus reconfiguration current, respectively. i dΔ and i qΔ These are the d-axis and q-axis components of the differential-mode reconfiguration current, respectively. and These are the d-axis and q-axis components of the modulus reconfiguration current reference, respectively. and These are the d-axis and q-axis components of the differential-mode reconfiguration current reference, respectively. T This indicates transpose.

[0010] Furthermore, the sum and difference voltage vectors of the previous control cycle are constructed using the following expression: in: u ( k -1) represents the sum and difference reconstructed voltage vector from the previous control cycle. u d1 ( k -1) and u q1 ( k -1) These are the d-axis voltage and q-axis voltage actually output to the main segment in the previous control cycle. u d2 ( k -1) and u q2 (k -1) These are the actual d-axis voltage and q-axis voltage output to the slave segment in the previous control cycle, respectively. u dΣ ( k -1) and u qΣ ( k -1) These are the d-axis and q-axis components of the previous control cycle and the module reconfiguration voltage, respectively. u dΔ ( k -1) and u qΔ ( k -1) These are the d-axis and q-axis components of the differential mode reconfiguration voltage from the previous control cycle, respectively. T This indicates transpose.

[0011] Furthermore, the expression for the sum-difference reconstructed voltage model is as follows: , In the formula: To reconstruct the current vector for sum and difference x The first derivative, u To reconstruct the voltage vector by sum and difference, u dΣ and u qΣ These are the d-axis and q-axis components of the modulus reconstructed voltage, respectively. u dΔ and u qΔ These are the d-axis and q-axis components of the differential-mode reconfiguration voltage, respectively. M ( or () is the sum and difference reconstruction of the inductance matrix. A ( or , oh e (This includes stator resistance and electric angular velocity) oh e And the cross-coupling terms between the dq axis and the sum and difference channels, b ( or , oh e ) represents the back electromotive force term corresponding to the permanent magnet flux linkage; or Inter-segment coupling parameters describe the equivalent electromagnetic coupling strength between the master and slave segments during double-segment overlap operation, varying with the position of the mover, the degree of overlap, or the magnetic circuit state. When a single equivalent coupling strength is used to describe the double-segment overlap state, the inter-segment coupling parameters... or scalar r The corresponding sum and difference reconstructed inductance matrix M ( r )for: in: L This is the equivalent inductance in the current operating mode; in the dual-segment overlapping operating mode, r This characterizes the equivalent coupling strength between the master segment and the slave segment; in single-segment operation mode, it makes... r =0, causing the sum-difference reconstructed voltage model to degenerate into a single-segment equivalent model.

[0012] Furthermore, the calculation expression for the sum-difference reconstructed control voltage vector of the current control cycle is as follows: in: The control voltage vector is reconstructed using the sum and difference of the current control cycle. x p ( k +1| k () represents the sum and difference reconstructed predicted current vector for the next control cycle. T s The sampling period is g ( k () represents the target current change rate. x ( k () represents the sum and difference reconstructed current vector for the current control cycle. , , , These are the sum and difference reconstructed current reference vectors, respectively. The four element values ​​in x p,1 , x p,2 , x p,3 , x p,4 Reconstruct the predicted current vectors by sum and difference respectively. x p ( k +1| k The four element values ​​in ) T d Let be the closed-loop expected time constant of the d-axis correlation and difference reconstruction channel. T q Let be the closed-loop expected time constant of the q-axis correlation and difference reconstruction channel.

[0013] Furthermore, the sum-difference reconstruction control voltage vector is transformed into the d-axis voltage reference and q-axis voltage reference of the master and slave segments of the current control cycle using the following expression: , in: , , , These are the sum and difference reconstructed control voltage vectors. The four element values ​​in and These are the d-axis voltage reference and q-axis voltage reference for the main segment of the current control cycle, respectively. and These are the d-axis voltage reference and q-axis voltage reference for the current control cycle segment, respectively.

[0014] Furthermore, the inter-segment coupling parameters in the sum-difference reconstructed voltage model are updated through online identification. Specifically, in the dual-segment overlapping operation mode, a high-frequency injection signal is first superimposed on the d-axis component of the differential-mode reconstruction current reference. Then, the d-axis component of the sum-mode reconstruction voltage is synchronously demodulated and low-pass processed to extract the high-frequency component corresponding to the high-frequency injection signal. Finally, the inter-segment coupling parameters are identified based on the high-frequency component using the following expression, and the sum-difference reconstructed voltage model is updated using the identified inter-segment coupling parameters. , in: c ρ To identify the update gain, This is the instantaneous estimate of the inter-segment coupling parameters for the current control cycle. and These are the estimated inter-segment coupling parameters for the next control cycle and the current control cycle, respectively. For the high-frequency components of the current control cycle, I h The amplitude of the high-frequency injected signal. oh h The angular frequency of the high-frequency injected signal.

[0015] Furthermore, in the dual-segment overlapping operation mode with high-frequency injection enabled, a residual observer is activated to estimate the disturbance compensation amount, including residual model mismatch, external load disturbances, and unmodeled dynamics. This disturbance compensation amount is then used to correct the target current change rate during the solution of the sum-difference reconstruction control voltage vector. Specifically: the residual observer calculates the nominal current derivative based on the sum-difference reconstruction voltage, sum-difference reconstruction current, and inter-segment coupling parameters of the previous control cycle. Simultaneously, the difference between the current control cycle and the sum-difference reconstruction current of the previous control cycle is used as the measured current derivative. The difference between the measured current derivative and the nominal current derivative is used as the original residual. Before the original residual enters the residual observer, the cosine and sine components synchronized with the high-frequency injection signal are estimated and subtracted to obtain a clean residual. Finally, the disturbance compensation amount is estimated according to the following expression: in: The clean residual for the current control cycle, c f For the gain of the residual observer, and These are the disturbance compensation amounts for the next control cycle and the current control cycle, respectively.

[0016] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method for predictive control of segmented current in a segmented winding permanent magnet linear synchronous motor.

[0017] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method for predictive control of segmented current in a segmented winding permanent magnet linear synchronous motor.

[0018] This invention acquires the mover position, segment enable state, or segment overlap state within each control cycle, determines the master and slave segments currently participating in control, and establishes the current control mode. Subsequently, it constructs the effective d-axis and q-axis currents, effective d-axis and q-axis current references for the master and slave segments, and the actual segment output voltage from the previous control cycle. Then, it performs sum-difference reconstruction to obtain the sum-difference reconstruction current state, sum-difference reconstruction current reference, and sum-difference reconstruction historical voltage. Next, it explicitly introduces inter-segment coupling parameters or matrices as model parameters into the sum-difference reconstruction voltage model and updates these parameters through online identification. Based on this explicit parameter model, it uses the actual voltage history to complete a single-cycle current prediction and uses the predicted current as a starting point to inversely calculate the sum-difference reconstruction control voltage for the current control cycle. Finally, it restores the sum-difference reconstruction control voltage to the d-axis and q-axis control voltages of the master and slave segments and outputs it to the effective segments. If a mode switch occurs during the current control cycle, this invention first remaps the actual segment output voltage from the previous cycle to the sum-difference reconstruction historical voltage corresponding to the switched control mode, and then uses this historical voltage for prediction in the next control cycle.

[0019] Based on the above technical solutions, the core technical problems mainly addressed by this invention are: When a segmented winding permanent magnet linear synchronous motor operates across segments, the boundaries of single-segment operation, double-segment overlapping operation, and mode switching are difficult to describe using a unified current loop model; the equivalent electromagnetic coupling between the master and slave segments in the double-segment overlapping region changes with the overlapping state; under the one-step execution delay of digital control, the actual output voltage of the previous control cycle still acts on the physical winding, while the current control mode and control variable expression have already switched, leading to inconsistencies in the historical voltage expression of the predicted current control. Furthermore, this invention further addresses issues such as online updating of inter-segment coupling parameters, mutual interference between high-frequency injected signals and residual disturbance observer inputs, and transients during engineering handover at cross-segment boundaries.

[0020] Compared with the prior art, the present invention has at least the following beneficial technical effects: 1. This invention uses a sum-difference reconstruction model to uniformly describe the switching states of single-segment, double-segment, and cross-segment, so that the current, voltage, reference of the master segment and slave segment, as well as the historical voltage of the actual segmented output of the previous cycle are processed in the same predictive control coordinates, thereby reducing the impact of abrupt changes in model structure on the current loop.

[0021] 2. In the sum-difference reconstructed voltage model, this invention explicitly introduces inter-segment coupling parameters or inter-segment coupling parameter matrices related to the double-segment overlap state, so that the equivalent coupling change of the master and slave segments in the cross-segment overlap area is used as a model parameter in the prediction and voltage inverse calculation process, rather than being incorporated into the total disturbance or residual compensation channel for post-processing.

[0022] 3. This invention uses the actual segmented output voltage of the previous control cycle to predict the current of the next control cycle, and remaps the actual segmented output voltage to the historical voltage after the switch and the difference reconstruction during mode switching, thus solving the boundary consistency problem where the physical voltage still acts on the winding while the expression of the control variable has changed under a one-beat execution delay.

[0023] 4. This invention identifies inter-segment coupling parameters by synchronous demodulating differential mode d-axis voltage with differential mode d-axis voltage, and updates model parameters using the differential / differential mode coupling relationship in the two-segment sum-difference model, thereby providing online support for explicit modeling of inter-segment parameters.

[0024] 5. This invention estimates residual model mismatch and external disturbances beyond the explicit inter-segment parameter model using a residual observer; when high-frequency injection is enabled, synchronous residual component subtraction avoids the observer from mislearning artificially injected signals, thereby improving the compatibility between disturbance compensation and inter-segment coupling parameter identification.

[0025] 6. The present invention can set closed-loop expected time constants for the d-axis related channel and the q-axis related channel respectively, so as to improve the dynamic coordination of channels with different axes.

[0026] 7. This invention can be implemented using fixed-size scalars and vector operations, without relying on an online optimization solver, and is suitable for implementation in digital signal processors, microcontrollers, real-time simulation controllers or other real-time control platforms.

[0027] 8. This invention can be combined with engineering measures such as amplitude limiting, smoothing, injection delay and observer reset to improve cross-segment boundary transients without changing the core predictive control model. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the overall process of the cross-segment current prediction and control method of the present invention.

[0029] Figure 2This is a schematic diagram of the cross-segment operation structure of a segmented winding permanent magnet linear synchronous motor.

[0030] Figure 3 This is a schematic diagram of the response curves between the experimental current reference value and the d-axis and q-axis current feedback values ​​in the cross-segment dual-segment region in an embodiment of the present invention.

[0031] Figure 4 This is a schematic diagram of the experimental mover velocity response curve for a cross-segment process in an embodiment of the present invention.

[0032] Figure 5 This is a schematic diagram of the identification curves of the experimental mover position and inter-segment coupling parameters in an embodiment of the present invention. Detailed Implementation

[0033] To describe the present invention in more detail, the technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0034] This embodiment provides a method for predictive control of segmented current in a segmented winding permanent magnet linear synchronous motor, the specific implementation of which is as follows: Figure 1 As shown: First, the position of the mover, the segment enable state, or the segment overlap state are obtained in each control cycle. The master segment and slave segment currently participating in the control are determined to determine the current control mode.

[0035] A segmented winding permanent magnet linear synchronous motor includes multiple stator segments arranged along the direction of motion. The controller determines the master and slave segments based on the mover position, segment overlap state, or segment enable state. Figure 2 As shown; when the master segment is active and the slave segment is inactive, the controller enters the first single-segment operation mode; when both the master and slave segments are active, the controller enters the double-segment overlapping operation mode; when the master segment is inactive and the slave segment is active, the controller enters the second single-segment operation mode; when both the master and slave segments are inactive or the operation enable is off, the controller enters the idle mode and outputs zero voltage. A typical cross-segment operation process may include the first single-segment operation mode—double-segment overlapping operation mode—second single-segment operation mode—double-segment overlapping operation mode—first single-segment operation mode, but this sequence is only a typical operating condition.

[0036] Subsequently, under the current control mode, the effective d-axis and q-axis currents, effective d-axis and q-axis current references, and the segmented output voltages actually output to the master and slave segments in the previous control cycle are constructed. In the dual-segment overlapping operation mode, the controller uses the d-axis and q-axis currents and d-axis and q-axis references of the master and slave segments; in the single-segment operation mode, the controller retains the d-axis and q-axis currents and references of the effective segments and sets the corresponding variables of the invalid segments to zero.

[0037] The effective d-axis and q-axis currents are reconstructed into sum-mode current components and differential-mode current components to obtain the sum-difference reconstructed current state; the effective d-axis and q-axis current references are reconstructed into sum-difference reconstructed current references; and when no control mode switching occurs, the segmented d-axis and q-axis voltages actually output to the master and slave segments in the previous control cycle are reconstructed into sum-difference reconstructed historical voltages according to the current control mode.

[0038] Controller construction and analog current components i dΣ , i qΣ and differential mode current component i dΔ , i qΔ : , Correspondingly, the construction and mode current reference components , and differential mode current reference component , : , in: i d1 and i q1 These are the d-axis current and q-axis current of the main segment, respectively. i d2 and i q2 These are the d-axis current and q-axis current, respectively. and These are the d-axis current reference and q-axis current reference for the main segment, respectively. and These are the d-axis current reference and the q-axis current reference, respectively. x The sum and difference are used to reconstruct the current state vector.

[0039] In this embodiment, in the dual-segment overlapping operation mode, the master segment and the slave segment receive the same d-axis and q-axis current references respectively. At this time, the sum and difference reconstruction controller does not simply divide the total reference into two segments according to the geometric ratio, but constrains the current state of the two segments in the sum and difference reconstruction coordinates through the sum mode and difference mode channels, so that the two segments jointly participate in current tracking and parameter identification in the overlapping area.

[0040] The sum-difference reconstructed historical voltage is constructed according to the same summation and difference relationship as the sum-difference reconstructed current state, where the sum-modulus reconstructed historical voltage component... u dΣ ( k -1) u qΣ (k -1) and differential mode reconstruction historical voltage components u dΔ ( k -1) and u qΔ ( k -1): in: u d1 ( k -1) u q1 ( k -1) represents the segmented d-axis and q-axis voltages actually output to the main segment in the previous control cycle. u d2 ( k -1) u q2 ( k -1) is the segmented d-axis and q-axis voltages that were actually output to the slave segment in the previous control cycle.

[0041] In the above sum-difference reconstruction system, the controller determines the inter-segment coupling parameters related to the double-segment overlap state and explicitly introduces these parameters as model parameters into the model matrix construction or model update. In this embodiment, or This represents a generalized form of the inter-segment coupling parameters (which can be a scalar, vector, or matrix), used to describe the equivalent electromagnetic coupling strength of the master and slave segments during double-segment overlap operation, varying with the position of the mover, the degree of overlap, or the magnetic circuit state. It can be determined by the overlap state, offline calibration relationships, or online identification results. A sum-difference reconstructed current state vector is defined. x Reconstructing the voltage vector with sum and difference u for: , The sum-difference reconstructed voltage model can be written as: in: M ( or () is the sum and difference reconstruction of the inductance matrix. A ( or , oh e This includes stator resistance, electrical angular velocity, and cross-coupling terms between the d-axis, q-axis, and the sum and difference channels. b ( or , oh e The term ) contains the back electromotive force corresponding to the permanent magnet flux linkage; this model is derived by combining the mathematical model of a common PMSM with the characteristics of a segmented linear motor.

[0042] When a single equivalent coupling strength is used to describe the overlapping state of two segments, the inter-segment coupling parameter can be denoted as a scalar. r This can be understood as the coupled inductance term or equivalent coupling strength entering the sum-difference reconstructed inductance matrix, and the sum-difference reconstructed inductance matrix. M ( r )for: in: L This is the equivalent inductance under the current control mode. In the dual-segment overlapping operation mode, r It characterizes the equivalent coupling strength between the master segment and the slave segment, and participates in matrix construction as a structural model parameter of the sum-difference reconstructed voltage model; in single-segment operation mode, it can be set... r =0 or mask the corresponding inter-segment coupling terms, causing the model to degenerate into a single-segment equivalent model. In this way, the inter-segment coupling of the double-segment overlap region, which varies with position or overlap state, no longer merely manifests as prediction error or total perturbation, but directly enters the prediction model. The above matrix is ​​a scalarized equivalent form for ease of explanation; in other implementations... M ( or It can be in the form of a matrix containing different inductances along the d and q axes, different inductances in the master and slave segments, or position-related inductance terms.

[0043] Therefore, during the double-segment overlap process, the inter-segment parameters that change with position or overlap state are no longer merely compensated for post-event by the disturbance observer as part of the total disturbance, but are preferentially incorporated into the prediction model itself to participate in current prediction and voltage inverse calculation. In single-segment operation mode, the inter-segment coupling terms are set to zero or masked, and the controller still uses the same sum-difference reconstruction model framework to complete single-segment control.

[0044] To compensate for the one-cycle execution delay of digital control, the controller does not directly inversely calculate the output voltage from the current sampled current. Instead, it first uses the sum-difference reconstructed current state vector obtained from the sampling and the sum-difference reconstructed voltage of the previous control cycle to predict the sum-difference reconstructed current of the next control cycle using the sum-difference reconstructed voltage model. in: x p ( k +1| k () represents the sum and difference reconstructed predicted current vector for the next control cycle. T s The sampling period is u ( k -1) It does not simply use the internal control variables of the previous mode, but is reconstructed and remapped from the actual segmented output voltage of the previous cycle according to the current control mode or the control mode after switching.

[0045] Subsequently, the controller constructs the target current change rate based on the predicted current and the reference current. g ( k And inversely calculate the sum and difference of the current control cycle to reconstruct the control voltage vector: Target current change rate g ( k The desired time constants of the d- and q-axis closed-loop circuits can be used to construct the circuit. in: T d The closed-loop expected time constant for the d-axis correlation and difference reconstruction channel. T q The closed-loop expected time constants used for the q-axis correlated sum-difference reconstruction channel can be the same or different. If the residual observer is enabled, the residual model mismatch and external disturbance estimates outside the explicit parameter model can be used as preferred compensation terms in the target rate of change or voltage inverse calculation process, but without changing the basic structure of the sum-difference reconstruction model and the one-beat prediction compensation.

[0046] After obtaining the sum and difference reconstructed control voltage vector, the controller can recover the d-axis and q-axis reference voltages of the master and slave segments according to the following relationship: , If the current control mode switches, the segmented d-axis and q-axis voltages actually output to the master and slave segments in the previous control cycle are remapped to the sum and difference reconstructed historical voltages corresponding to the new control mode. These remapped sum and difference reconstructed historical voltages are then used for current prediction in the next control cycle. This step ensures that the historical voltages used for one-cycle prediction compensation originate from the actual physical output and conform to the control variable expression after the switch.

[0047] This embodiment also provides an online identification method for inter-segment coupling parameters, which supports the explicit input of inter-segment coupling parameters as model parameters into the sum-difference reconstructed voltage model. This method can be enabled in the dual-segment overlapping operation mode to update the scalar inter-segment coupling parameters. r This may provide estimation information for the inter-segment coupling parameter matrix.

[0048] In dual-segment overlapping operation mode, the controller superimposes a high-frequency signal into the differential-mode d-axis current reference: in: As a differential mode d-axis fundamental current reference, I h For the amplitude of the high-frequency injected signal, oh h The angular frequency of the injected signal.

[0049] Under the condition that the differential-mode d-axis high-frequency current can be effectively tracked, the inter-segment coupling parameters r This will cause the differential-mode high-frequency current to generate a high-frequency component in the sum-mode d-axis voltage that is synchronized with the injected signal. The controller will then synchronize the sum-mode d-axis voltage with 2cos( oh h t Multiply by, and pass through a low-pass filter operator Extracting the DC component: This allows us to obtain an instantaneous estimate of the inter-segment coupling parameters: In other embodiments that consider filtering phase, calibration ratio, or voltage normalization, the instantaneous estimation formula described above may include a preset correction coefficient, but this correction does not change the fundamental relationship between differential mode d-axis injection and sum mode d-axis synchronous demodulation used to identify inter-segment coupling parameters.

[0050] To reduce the impact of noise and filter residue on the control model, the controller obtains a smooth estimate through a first-order update: in: c ρ To identify the updated gain, the controller can... A physical safety boundary is set to maintain the non-singularity of the sum-difference reconstructed inductance matrix. This safety boundary is used to guarantee the invertibility of the model and the implementability of the controller, and is not used as an empirical theoretical correction.

[0051] This embodiment also provides an auxiliary residual observation enhancement method. This method is based on the fact that the inter-segment coupling parameters have been explicitly incorporated as model parameters into the sum-difference reconstruction prediction model. It is used to estimate the remaining model mismatch, load disturbances, and unmodeled dynamics outside of this explicit parameter model. The controller uses a first-order residual observer based on derivative residuals to estimate the residual disturbances. For the previous sampling interval, the controller first calculates the nominal current derivative based on the previous cycle's sum-difference reconstruction historical voltage, the previous cycle's sum-difference reconstruction state, and the previous cycle's model parameters. The measured derivative is then calculated from the sum and difference currents of the current and the previous sampling period: The original residuals are: If the current operation is in a dual-segment overlapping mode and high-frequency injection is enabled, the raw residuals may contain artificial components synchronized with the injection frequency. To prevent the residual observer from learning this artificial component as a disturbance, the controller performs synchronization component estimation for each residual channel: Then construct the clean residual: Clean residual input first-order residual observer: in: For clean residuals to enter the residual observer, c f For residual observer gain, This is the disturbance compensation amount. This step only changes the residual observer input and does not change the physical scale of the high-frequency injection to identify inter-segment coupling parameters. Therefore, it can reduce the impact of high-frequency injection on disturbance estimation and voltage inverse compensation while maintaining the online identification capability of inter-segment coupling parameters.

[0052] In the voltage inverse calculation process, the compensation amount is estimated based on the residual model mismatch and external disturbances obtained from the residual observer. Corrected target current change rate: in: T i For the first i The expected closed-loop time constant of the channel. For the first i The remaining model mismatch and external disturbance estimate compensation values ​​for the channel.

[0053] When switching control modes, the voltage actually output to the master and slave segments in the previous control cycle will still affect the current change in the next control cycle. If the internal sum and difference reconstruction of the historical voltage of the old mode is continued, the one-step prediction compensation will use voltage information that does not match the current mode.

[0054] Therefore, the controller saves the actual segmented output historical voltage of the previous control cycle: When switching to the dual-segment overlapping operation mode, it is remapped as follows: When switching to the first single-segment operation mode, the controller maps the actual output voltage of the previous cycle of the effective master segment to the sum and difference reconstructed historical voltage in the single-segment mode, and sets or masks the relevant quantities of the invalid slave segment; when switching to the second single-segment operation mode, the controller maps the actual output voltage of the previous cycle of the effective slave segment to the sum and difference reconstructed historical voltage in the single-segment mode, and processes it according to the differential mode sign relationship.

[0055] The above remapping method addresses the issue of variable expression consistency under a one-cycle execution delay: the voltage of the previous cycle actually acts on the physical winding, while the current control mode and the sum-difference reconstruction variable expression have changed; by remapping with the actual segmented output voltage as the source, the controller can maintain consistency between the predictive model input and the physical historical voltage at the switching boundary.

[0056] In the dual-segment overlapping operation mode, the master segment and the slave segment receive the same or coordinated d-axis and q-axis current references respectively, and the sum-difference reconstruction controller converts the dual-segment current references into sum-mode current references and differential-mode current references. The sum-mode channel is used to constrain the common components of the currents of the master segment and the slave segment, and the differential-mode channel is used to constrain the current differences between the master segment and the slave segment and the parameter identification channel.

[0057] As a supplement to the engineering implementation, the controller can also be combined with measures such as observer reset, high-frequency injection delay, output voltage increment limit, DC bus limit and short window handover smoothing during mode switching to improve cross-segment boundary transients. These engineering measures do not change the core control framework of sum and difference reconstruction, one-beat prediction and real segmented output voltage history remapping.

[0058] In one implementation, the controller resets the residual observer state and the residual synchronization component estimation state during mode switching to avoid voltage recalculation when the residual estimation from the old mode enters the new mode. In another implementation, the controller suspends high-frequency injection, synchronous demodulation, or residual high-frequency synchronization component deduction within a preset sampling period after entering the dual-segment overlap mode, allowing predictive current control to complete boundary takeover first.

[0059] At the output voltage level, the controller can apply per-step increment limits to the d-axis and q-axis voltages of each segment: Where: lim represents the limiting operator. u target The target segmented voltage is obtained by reconstructing the control voltage based on the sum and difference in the current control cycle. u prev Δ is the actual output voltage of the previous control cycle. u max The maximum allowable voltage increment per control cycle, Δ u This is the voltage increment after limiting, and it can be made u limited = u prev +Δ u This represents the target voltage after the increment limit per beat.

[0060] Within the handover window, the controller can also gradually transition the voltage from the actual output of the previous cycle to the target voltage after limiting, according to a smoothing coefficient: in: u limited The target voltage after each increment limit, α h This is the handover smoothing coefficient, and its value can be 0 ≤ α h ≤1; when α h When =0, the output of the previous cycle is maintained. α h When =1, the target voltage after limiting is directly used, which can be achieved within the transfer window. The sampling period is gradually increased to achieve a smooth transition.

[0061] In addition, the controller can perform DC bus circular limiting on the d and q axis voltages of each segment: in: V dc This refers to the inverter's DC bus voltage. u d and u q These are the d-axis and q-axis output voltage components for the corresponding segments.

[0062] Figure 3 to Figure 5 The continuous effective operating intervals in the corresponding experiments are selected for explanation. The display intervals of the current response curve and the velocity response curve both end at the end of the fourth single-segment operating interval, which successively includes the first single-segment interval, the first overlapping interval, the second single-segment interval, the second overlapping interval, the third single-segment interval, the third overlapping interval, and the fourth single-segment interval.

[0063] like Figure 3 As shown in the figure, the black dashed line on the d-axis represents the zero-fundamental current reference without high-frequency identification components. In the dual-segment overlapping operation region, the controller superimposes a high-frequency injection signal onto the differential-mode d-axis current reference to identify the inter-segment coupling parameters; this injection, after sum-difference inverse transformation, enters the main segment and the slave segment respectively, causing the two d-axis currents to exhibit high-frequency fluctuations in opposite directions. Therefore, Figure 3The periodic fluctuations in the d-axis within the double-segment overlap region are the active excitation required for parameter identification, rather than d-axis current runaway or persistent steady-state error. During the same continuous span, the q-axis current can complete the engagement, interaction, and disengagement processes between the two winding segments as the mover region is located, and follows the changes in the corresponding current reference value. After single-sampling delay alignment of the first to third overlap regions, the correlation coefficient between the reference value and the feedback value of the q-axis and current (the sum of the q-axis currents in the two segments) is approximately 0.982, and the root mean square tracking error is approximately 0.227 A; the root mean square tracking error of the q-axis current in the second to fourth single-segment regions is approximately 0.034 A. These results demonstrate that the control method of this invention can maintain the continuity of current response during the inter-segment working current redistribution process.

[0064] like Figure 4 As shown, after the mover starts, its speed gradually reaches 0.3 m / s and remains near the given speed in the first to third overlap regions and adjacent single-segment regions. The root mean square speed error in the first to third overlap regions is approximately 0.0048 m / s, the peak-to-peak speed is approximately 0.0284 m / s, and the maximum absolute speed error is approximately 0.0176 m / s. No sustained speed deviation or divergence was observed. This result indicates that the inter-segment current switching did not disrupt the continuous and stable operation of the mover speed.

[0065] like Figure 5 As shown, the blue curve The red dashed line represents the estimated inter-segment coupling equivalent inductance obtained by the controller's online identification. r geom This represents a reference curve constructed and calibrated based on the mover position and the geometric overlap relationship between the master and slave stator segments. The S-curve below illustrates the correspondence between changes in coupling parameters and the mover position. First, a nominal reference value is constructed based on the geometric overlap ratio between the mover and the master and slave stator segments. r geom,nom Then, its amplitude is calibrated using single-parameter least squares calibration through the origin; in the single-segment operating range, there is no effective overlap between the master and slave stator segments. It remains essentially zero; after entering the double-segment overlap region, As the position of the follower changes, its rising, falling, and sign changes follow the trend of... r geom They are basically consistent. The calibration coefficients obtained from the first to third overlapping regions are 0.2395, that is... r geom =0.2395 r geom,nom The corresponding effective inductance difference parameter is 0.287mH; after calibration r geom With online identification values The shape correlation coefficient was approximately 0.760, the direct determination coefficient was approximately 0.549, and the root mean square error was approximately 0.073 mH. This result indicates that... It can reflect the overall trend of the change in the equivalent inductance of inter-segment coupling with the overlapping position; among which r geom This is a geometric reference curve that has been calibrated for amplitude and is not used as an independent measurement true value.

[0066] Based on the above experimental results, the present invention can achieve coordinated distribution of current in two winding segments during the continuous crossing of multiple single-segment and overlapping regions, maintain current and speed tracking stability, and obtain inter-segment coupling parameters that vary with the overlapping position, thereby reducing the prediction model deviation caused by changes in motor structure and overlapping state when using a fixed parameter model.

[0067] The above description of the embodiments is provided to enable those skilled in the art to understand and apply the present invention. Those skilled in the art can readily make various modifications to the above embodiments and apply the general principles described herein to other embodiments without creative effort. Therefore, the present invention is not limited to the above embodiments, and any improvements and modifications made to the present invention by those skilled in the art based on the disclosure thereof should be within the scope of protection of the present invention.

Claims

1. A method for predictive control of segmented current in a segmented permanent magnet linear synchronous motor, characterized in that: Determine the master and slave segments and operating modes of the two adjacent stator segments currently participating in control, and construct the sum and difference reconstructed current vector and its reference vector, as well as the sum and difference reconstructed voltage vector of the previous control cycle. The sum-difference reconstructed voltage model is established using inter-segment coupling parameters and the sum-difference reconstructed current state vector, as shown in the following expression: ， ， In the formula: x To reconstruct the current vector through sum and difference, i dΣ and i qΣ These are the d-axis and q-axis components of the modulus reconfiguration current, respectively. i dΔ and i qΔ These are the d-axis and q-axis components of the differential-mode reconfiguration current, respectively. T Indicates transpose. To reconstruct the current vector for sum and difference x The first derivative, u To reconstruct the voltage vector by sum and difference, u dΣ and u qΣ These are the d-axis and q-axis components of the modulus reconstructed voltage, respectively. u dΔ and u qΔ These are the d-axis and q-axis components of the differential-mode reconfiguration voltage, respectively. M ( η () is the sum and difference reconstruction of the inductance matrix. A ( η , ω e (This includes stator resistance and electric angular velocity) ω e And the cross-coupling terms between the dq axis and the sum and difference channels, b ( η , ω e ) represents the back electromotive force term corresponding to the permanent magnet flux linkage; η Inter-segment coupling parameters describe the equivalent electromagnetic coupling strength between the master and slave segments during double-segment overlap operation, varying with the position of the mover, the degree of overlap, or the magnetic circuit state. When a single equivalent coupling strength is used to describe the double-segment overlap state, the inter-segment coupling parameters... η scalar ρ The corresponding sum and difference reconstructed inductance matrix M ( ρ )for: in: L This is the equivalent inductance in the current operating mode; in the dual-segment overlapping operating mode, ρ This characterizes the equivalent coupling strength between the master segment and the slave segment; in single-segment operation mode, it makes... ρ =0, causing the sum-difference reconstructed voltage model to degenerate into a single-segment equivalent model; By constructing the target current change rate and reconstructing the voltage vector based on the sum and difference of the previous control cycle, the sum and difference reconstructed control voltage vector of the current control cycle is calculated using the sum and difference reconstructed voltage model. This vector is then transformed into the d-axis voltage reference and q-axis voltage reference of the main and slave segments of the current control cycle for segmented control.

2. The method for predictive control of segmented current in a segmented winding permanent magnet linear synchronous motor according to claim 1, characterized in that, The sum-difference reconstructed current reference vector is constructed using the following expression: ， ， in: For the sum and difference reconstructed current reference vector, i d1 and i q1 These are the d-axis current and q-axis current of the main segment, respectively. i d2 and i q2 These are the d-axis current and q-axis current of the segment, respectively. and These are the d-axis current reference and q-axis current reference for the main segment, respectively. and These are the d-axis current reference and q-axis current reference for the segment, respectively. and These are the d-axis and q-axis components of the modulus reconfiguration current reference, respectively. and These are the d-axis and q-axis components of the differential-mode reconfiguration current reference, respectively.

3. The method for predictive control of segmented current in a segmented winding permanent magnet linear synchronous motor according to claim 1, characterized in that, The sum and difference reconstructed voltage vector of the previous control cycle is constructed using the following expression: in: u ( k -1) represents the sum and difference reconstructed voltage vector from the previous control cycle. u d1 ( k -1) and u q1 ( k -1) These are the d-axis voltage and q-axis voltage actually output to the main segment in the previous control cycle. u d2 ( k -1) and u q2 ( k -1) These are the actual d-axis voltage and q-axis voltage output to the slave segment in the previous control cycle, respectively. u dΣ ( k -1) and u qΣ ( k -1) These are the d-axis and q-axis components of the previous control cycle and the module reconfiguration voltage, respectively. u dΔ ( k -1) and u qΔ ( k -1) are the d-axis and q-axis components of the differential mode reconfiguration voltage of the previous control cycle, respectively.

4. The method for predictive control of segmented current in a segmented winding permanent magnet linear synchronous motor according to claim 1, characterized in that, The calculation expression for the sum-difference reconstructed control voltage vector of the current control cycle is as follows: in: The control voltage vector is reconstructed using the sum and difference of the current control cycle. x p ( k +1| k () represents the sum and difference reconstructed predicted current vector for the next control cycle. T s The sampling period is g ( k () represents the target current change rate. x ( k () represents the sum and difference reconstructed current vector for the current control cycle. , , , These are the sum and difference reconstructed current reference vectors, respectively. The four element values ​​in x p,1 , x p,2 , x p,3 , x p,4 Reconstruct the predicted current vectors by sum and difference respectively. x p ( k +1| k The four element values ​​in ) T d Let be the closed-loop expected time constant of the d-axis correlation and difference reconstruction channel. T q Let be the closed-loop expected time constant of the q-axis correlation and difference reconstruction channel.

5. The method for predictive control of segmented current in a segmented winding permanent magnet linear synchronous motor according to claim 4, characterized in that, The sum-difference reconstruction control voltage vector is transformed into the d-axis voltage reference and q-axis voltage reference for the master and slave segments of the current control cycle using the following expression: ， in: , , , These are the sum and difference reconstructed control voltage vectors. The four element values ​​in and These are the d-axis voltage reference and q-axis voltage reference for the main segment of the current control cycle, respectively. and These are the d-axis voltage reference and q-axis voltage reference for the current control cycle segment, respectively.

6. The method for predictive control of segmented current in a segmented winding permanent magnet linear synchronous motor according to claim 1, characterized in that, The inter-segment coupling parameters in the sum-difference reconstructed voltage model are updated through online identification. Specifically, in the dual-segment overlapping operation mode, a high-frequency injection signal is first superimposed on the d-axis component of the differential-mode reconstructed current reference, and then the d-axis component of the sum-mode reconstructed voltage is synchronously demodulated and low-pass processed to extract the high-frequency component corresponding to the high-frequency injection signal. Finally, based on the high-frequency component, the inter-segment coupling parameters are identified using the following expression, and the identified inter-segment coupling parameters are used to update the sum and difference reconstruct the voltage model. ， in: γ ρ To identify the update gain, This is the instantaneous estimate of the inter-segment coupling parameters for the current control cycle. and These are the estimated inter-segment coupling parameters for the next control cycle and the current control cycle, respectively. For the high-frequency components of the current control cycle, I h The amplitude of the high-frequency injected signal. ω h The angular frequency of the high-frequency injected signal.

7. The method for predictive control of segmented current in a segmented winding permanent magnet linear synchronous motor according to claim 1, characterized in that: In the dual-segment overlapping operation mode with high-frequency injection enabled, a residual observer is activated to estimate the disturbance compensation amount, including residual model mismatch, external load disturbances, and unmodeled dynamics. This disturbance compensation amount is then used to correct the target current rate of change during the solution of the sum-difference reconstruction control voltage vector. Specifically: the residual observer calculates the nominal current derivative based on the sum-difference reconstruction voltage, sum-difference reconstruction current, and inter-segment coupling parameters of the previous control cycle. Simultaneously, the difference between the current control cycle and the sum-difference reconstruction current of the previous control cycle is used as the measured current derivative. The difference between the measured current derivative and the nominal current derivative is used as the original residual. Before the original residual enters the residual observer, the cosine and sine components synchronized with the high-frequency injection signal are estimated and subtracted to obtain a clean residual. Finally, the disturbance compensation amount is estimated according to the following expression: in: The clean residual for the current control cycle, γ f For the gain of the residual observer, and These are the disturbance compensation amounts for the next control cycle and the current control cycle, respectively.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: The processor is used to execute the computer program to implement the segment current prediction control method for the segmented winding permanent magnet linear synchronous motor as described in any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, it implements the segment current prediction control method for the segmented winding permanent magnet linear synchronous motor as described in any one of claims 1 to 7.

Images