A robust current control method for permanent magnet synchronous linear motor

Abstract

The application provides a kind of robust current control method of permanent magnet synchronous linear motor, belongs to permanent magnet synchronous linear motor control technical field, method includes: S1: the establishment permanent magnet synchronous linear motor multi-coordinate system mathematical model;S2: construct optimization sliding mode observer estimates rotor position and speed;S3: design model predictive current control determines optimal voltage vector;S4: based on SVPWM generates inverter switching signal and realizes closed loop feedback.The application significantly enhances the anti-interference ability of permanent magnet synchronous linear motor to parameter variation, external disturbance and load fluctuation, improves the running stability of motor under complex logistics working condition, avoids the performance attenuation caused by environmental change.Optimize current control accuracy, reduce current ripple and torque ripple, make motor output thrust more stable, improve the positioning accuracy and running smoothness of logistics transportation system.

Description

Technical Field

[0001] This invention relates to the field of permanent magnet synchronous linear motor (PMSLM) control technology, and in particular to a robust current control method for permanent magnet synchronous linear motors. Background Technology

[0002] In traditional intelligent modular logistics transportation systems, the driving components mostly use a combination of "rotary motor + mechanical transmission device" to achieve linear motion. This solution has inherent defects such as insufficient control precision, large mechanical wear, high operating noise, and short equipment life, making it difficult to meet the modern logistics requirements for high efficiency, precision, and low consumption.

[0003] Permanent magnet synchronous linear motors (PMSLMs) are gradually becoming the preferred choice for logistics drive systems due to their advantages such as no transmission loss, high torque power density, high efficiency, and rapid response. However, they face many technical bottlenecks in practical applications: In complex logistics environments, motors are susceptible to parameter drift (such as resistance and inductance fluctuations caused by temperature changes), load fluctuations (such as changes in cargo weight), and external disturbances, leading to a decrease in current control stability. In traditional control algorithms, field-oriented control (FOC) has limited dynamic response performance, and direct torque control (DTC) suffers from torque and flux pulsation problems, making it difficult to balance steady-state accuracy and dynamic robustness. In terms of sensorless technology, conventional sliding mode observers (SMOs) have problems such as narrow speed adaptation range, severe low-speed chattering, and significant phase delay, which cannot provide accurate rotor position and speed information for the control algorithm, thus restricting the reliable application of PMSLMs in logistics systems. Summary of the Invention

[0004] This invention provides a robust current control method for a permanent magnet synchronous linear motor to solve one or more of the problems mentioned above. To achieve the above objective, this invention adopts the following technical solution:

[0005] A robust current control method for a permanent magnet synchronous linear motor includes:

[0006] S1: Obtain the three-phase stator phase voltage and three-phase stator phase current of the permanent magnet synchronous linear motor, establish the motor mathematical model in the three-phase stationary coordinate system, the two-phase stationary coordinate system and the two-phase synchronous rotating coordinate system, convert the three-phase voltage and current into voltage and current in the two-phase stationary coordinate system through Clark transformation, and then convert them into voltage and current in the two-phase synchronous rotating coordinate system through Park transformation to obtain the inherent parameters of the motor.

[0007] S2: Optimize the sliding mode observer with voltage and current inputs in the two-phase stationary coordinate system, suppress chattering through saturation function, and estimate the rotor electric angle and rotor electric angular velocity by combining phase-locked loop to compensate for phase delay.

[0008] S3: Input the inherent parameters of the motor, the current in the two-phase synchronous rotating coordinate system, and the rotor electrical angle and rotor electrical angular velocity into the model to predict the current control algorithm. Construct a current prediction model based on forward Euler discretization, compensate for the one-step delay of digital control, define the objective function to minimize the current tracking error, determine all voltage vectors of the three-phase two-level inverter, calculate the objective function value corresponding to each voltage vector, select the voltage vector with the smallest objective function value as the optimal voltage vector, and obtain the switching state corresponding to the optimal voltage vector.

[0009] S4: Input the rotor electrical angle and optimal voltage vector into the space vector pulse width modulation module. Convert the optimal voltage vector into voltage components in a two-phase stationary coordinate system through inverse Park transformation. Determine the sector where the voltage components are located. Calculate the action time of adjacent effective voltage vectors and zero vectors based on the volt-second balance principle. Generate the inverter's PWM drive signal according to the principle of minimizing the number of switching state transitions to drive the permanent magnet synchronous linear motor. Real-time acquisition of the actual three-phase stator current during the operation of the permanent magnet synchronous linear motor. Use the actual three-phase stator current as the input of step S1 to form a closed-loop control. Continuously execute steps S1 to S4 to achieve robust current control of the permanent magnet synchronous linear motor.

[0010] In this specification, in step S2, the saturation function adopts a linear output within the boundary layer and a constant output outside the boundary layer. The boundary layer thickness is determined through simulation and debugging to bring the current estimation error to a preset range, thereby suppressing the chattering problem inherent in sliding mode control.

[0011] In this specification, in step S2, the phase-locked loop includes a phase detector, a loop filter, and a voltage-controlled oscillator. The phase detector calculates the angle error between the initially estimated rotor electrical angle and the rotor electrical angle output by the phase-locked loop. After being filtered and amplified by the loop filter, the voltage-controlled oscillator outputs the corrected rotor electrical angle and rotor electrical angular velocity. The proportional coefficient and integral coefficient of the loop filter are designed using the pole placement method to balance the anti-interference capability and dynamic response speed.

[0012] In this specification, in step S3, the current prediction model first predicts the current at time k+1 by discretizing the voltage equation in the two-phase synchronous rotating coordinate system using forward Euler, and then predicts the current at time k+2 based on the current at time k+1, where k is the current discrete time step. This two-step prediction compensates for one-step delay in the digital control system, ensuring that the timing of the control quantity action is consistent with the timing of the current prediction.

[0013] In this specification, in step S3, the weight coefficients of the objective function are all set to 1. The reference current of the magnetic flux component is fixed to zero, and the reference current of the torque component is calculated by the speed outer loop PI controller. The input of the speed outer loop PI controller is the difference between the rotor reference angular velocity and the rotor electric angular velocity output in step S2. The rotor reference angular velocity is issued by the system host computer according to the logistics transportation task.

[0014] In this specification, in step S3, all voltage vectors of the three-phase two-level inverter include 6 effective voltage vectors and 2 zero vectors, for a total of 8 voltage vectors. The switching state corresponding to each voltage vector is determined, and the current and objective function value at time k+2 are calculated one by one under each switching state.

[0015] In this specification, in step S4, the sector determination is based on the sign and amplitude relationship of the voltage components in the two-phase stationary coordinate system. The two-phase stationary coordinate system is divided into 6 sectors, and each sector corresponds to two adjacent effective voltage vectors. The sector to which the optimal voltage vector belongs and the effective voltage vector synthesized by the cooperative synthesis are determined by the numerical characteristics of the voltage components.

[0016] In this specification, in step S4, the duration of action of adjacent effective voltage vectors is solved by solving a series of equations based on the volt-second balance principle. The total duration of action of the zero vector is the sampling period minus the duration of action of the two adjacent effective voltage vectors. The duration of action of the zero vector is evenly distributed to the two zero vectors and placed at the beginning and end of the sampling period, respectively, to reduce the number of switching states.

[0017] In this specification, the objective function of the model prediction current control algorithm in step S3 adopts a multi-objective optimization design. In addition to the current tracking error, a switching loss penalty term is introduced. The penalty coefficient is dynamically adjusted to balance the current control accuracy and the inverter switching loss. The penalty coefficient is adaptively adjusted according to the rotor electric angular velocity. Under high-speed conditions, the penalty coefficient is increased to reduce the switching frequency, and under low-speed conditions, the penalty coefficient is decreased to improve control accuracy. The voltage vector enumeration adopts a pruning strategy. Based on the current deviation direction and the rotor electric angular velocity, the three adjacent voltage vectors that are likely to be the optimal solution are first selected. Then, the objective function is calculated on the selected voltage vectors to reduce the enumeration calculation and improve the real-time performance of the control algorithm. The selection criterion is the matching degree between the direction of the voltage vector and the current deviation compensation requirement. The current prediction model introduces a parameter adaptive correction term. Based on the deviation between the actual current and the predicted current, the inductance and resistance values ​​in the inherent parameters of the motor are adjusted in real time to compensate for the prediction error caused by parameter drift. The weight of the correction term increases with the increase of the deviation.

[0018] In this specification, the space vector pulse width modulation module in step S4 employs a fast sector determination algorithm. By comparing the sign combination and amplitude of the voltage components in the two-phase stationary coordinate system, the sector to which the optimal voltage vector belongs is directly determined, thereby shortening the sector determination time. The sign combination reflects the positive and negative directions of the voltage components, and the amplitude comparison reflects the relative magnitude of the voltage components. The duty cycle of the PWM drive signal adopts a dynamic adjustment mechanism, which collects the inverter DC bus voltage in real time. If the bus voltage fluctuates, the action time of adjacent effective voltage vectors and zero vectors is corrected based on the fluctuation amplitude to ensure that the average value of the synthesized voltage always tracks the optimal voltage vector and offsets the impact of bus voltage fluctuations on control accuracy.

[0019] In summary, the present invention has at least the following beneficial effects:

[0020] It significantly enhances the permanent magnet synchronous linear motor's ability to resist interference from parameter changes, external disturbances and load fluctuations, improves the motor's operational stability under complex logistics conditions, and avoids performance degradation caused by environmental changes.

[0021] Optimize current control accuracy, reduce current ripple and torque pulsation, make the motor output thrust more stable, and improve the positioning accuracy and smooth operation of the logistics transportation system.

[0022] Reduce the risk of equipment failure caused by fluctuations in motor performance, reduce the frequency and cost of maintenance in the logistics system, and extend the overall service life of the equipment.

[0023] Improving the energy conversion efficiency of motors and reducing operating energy consumption aligns with the needs of green logistics development.

[0024] It adapts to the flexible configuration requirements of intelligent modular logistics transportation systems, enhances the compatibility of control algorithms with different logistics modules, and improves the overall scalability and adaptability of the system. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the robust current control method for a permanent magnet synchronous linear motor involved in this invention.

[0026] Figure 2 This is a schematic diagram of the inverter topology involved in this invention.

[0027] Figure 3 This is a schematic diagram of the spatial distribution of the voltage vector involved in this invention.

[0028] Figure 4 This is a schematic diagram of the control block diagram of the permanent magnet synchronous linear motor involved in this invention.

[0029] Figure 5 This is a schematic diagram of the sliding mode observer involved in the present invention. Detailed Implementation

[0030] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0031] like Figure 1 As shown, this embodiment provides a robust current control method for a permanent magnet synchronous linear motor. This solution achieves robust current control of the permanent magnet synchronous linear motor through a collaborative design of the entire process of "multi-coordinate system modeling - sensorless state estimation - model predictive current control - SVPWM drive and closed-loop feedback". First, mathematical models of the motor in multiple coordinate systems—three-phase stationary, two-phase stationary, and two-phase synchronous rotating—are established. Variable dimensionality reduction and flux-torque decoupling are achieved through Clark and Park transformations, providing a precise theoretical foundation for subsequent algorithms. Second, an optimized sliding mode observer (SMO) is constructed, employing saturation functions to suppress chattering and introducing a phase-locked loop (PLL) to compensate for phase delay, enabling accurate estimation of rotor electrical angle and angular velocity without position sensors. Subsequently, model predictive current control (MPCC) is designed based on a discretized current prediction model, fully enumerating the eight voltage vectors of the three-phase two-level inverter (IPM). The optimal voltage vector is selected by minimizing the current tracking error objective function. Finally, the discrete optimal vector is converted into a continuous and smooth inverter switching signal through space vector pulse width modulation (SVPWM), and closed-loop feedback is formed by real-time acquisition of the actual motor current, achieving deep collaboration among various algorithm modules. Ultimately, this ensures stable and efficient motor operation under complex logistics conditions, providing reliable drive support for intelligent modular logistics transportation systems. The closed-loop control continues until the permanent magnet synchronous linear motor completes the preset logistics transportation task or receives a stop command from the host computer. When the closed loop ends, the space vector pulse width modulation module outputs the PWM drive signal corresponding to the zero vector, causing the permanent magnet synchronous linear motor to stop running.

[0032] S1: Establish a multi-coordinate mathematical model of a permanent magnet synchronous linear motor (PMSLM).

[0033] 1.1 Model Construction Background and Core Objective: As a core driving component of intelligent modular logistics transportation systems, the permanent magnet synchronous linear motor is essentially a multivariable, strongly coupled nonlinear time-varying system—the stator current, flux linkage, and torque have complex coupling relationships, making it difficult to design efficient control algorithms directly based on the original three-phase stationary coordinate system. Therefore, the core objective of this step is to simplify the system model through coordinate system transformation (Clark transformation, Park transformation), remove coupling terms, and construct a mathematical model that accurately reflects the electromagnetic characteristics and dynamic behavior of the motor. This provides a solid theoretical foundation and accurate system parameter support for subsequent sliding mode observer state estimation and model predictive current control algorithm design.

[0034] 1.2 Model Construction Process for Each Coordinate System

[0035] (1) Basic model of three-phase stationary coordinate system (abc): The three-phase stationary coordinate system is the most primitive coordinate system for describing the electrical characteristics of motors. It directly corresponds to the three-phase stator windings (a phase, b phase, c phase) of the motor. The model construction strictly follows the law of electromagnetic induction and the principle of magnetic flux coupling. It includes two parts: voltage equation and magnetic flux equation, which fully reflect the dynamic relationship between current, voltage and magnetic flux.

[0036] Voltage equation: ; ; ;

[0037] The physical meaning of this equation is: the stator phase voltage consists of two parts, one part being the voltage drop across the stator winding resistance ( The other part is the induced electromotive force generated by the change of stator winding flux over time. ). The three-phase stator phase voltage (unit: V) of the permanent magnet synchronous linear motor is output by the three-phase two-level inverter in the system and is obtained in real time by the voltage sensor installed at the output end of the inverter. It is the input excitation signal for motor operation. The three-phase stator phase current (unit: A) of the permanent magnet synchronous linear motor is collected in real time by a Hall current sensor connected in series in the stator winding. It is the core feedback signal reflecting the motor's operating status and will be directly transmitted to step S2 as the input of the sliding mode observer. The equivalent resistance of the stator winding (unit: Ω) is determined by the conductor material (usually copper), cross-sectional area and length of the motor winding. It can be found by looking up the parameters on the motor nameplate or by experimental measurement using the DC volt-ampere method. It is considered a constant value during motor operation. : The flux linkage of the three-phase stator winding (unit: Wb) is the superposition value of the flux linkage generated by the current and the excitation flux linkage of the permanent magnet, which directly affects the electromagnetic torque output of the motor; t: time (unit: s), used to describe the dynamic change process of the variable.

[0038] Magnetic flux linkage equation: ;

[0039] This equation illustrates the mechanism of flux linkage formation: the first part is the armature reaction flux linkage generated by the stator current through the windings (represented by the product of the inductance matrix and the current vector), and the second part is the excitation flux linkage generated by the rotor permanent magnets (related to the rotor electrical angle). (Related). The self-inductance of the stator winding (unit: H) is the inductance generated by the current in a single winding. Due to the salient pole characteristics of the rotor (the rotor structure of the built-in permanent magnet synchronous linear motor causes uneven air gap), the self-inductance varies with the rotor electrical angle. Periodic changes, specifically expressed as: , , ;

[0040] Mutual inductance between stator windings (unit: H), that is, the inductance generated by the current in one winding in another winding. It also has periodicity and follows a three-phase symmetrical relationship, for example... , The remaining mutual inductances can be derived according to the principle of symmetry; The inductance component generated by the main magnetic flux (unit: H) is determined by the magnetic reluctance of the main magnetic flux path in the motor's magnetic circuit. It is the fundamental part of self-inductance and is obtained through motor magnetic circuit design calculations or experimental measurements. The alternating inductance component (unit: H) caused by rotor rotation reflects the strength of the rotor's salient pole characteristics; the more pronounced the salient pole, the stronger the characteristic. The larger the value, the more it is determined through a combination of motor no-load experiments and finite element simulation. The excitation flux of a permanent magnet (unit: Wb) is determined by the material properties of the permanent magnet (such as the remanence density of NdFeB N35), the size of the magnetic poles, and the magnetic circuit structure. It is the core excitation source for the motor to generate electromagnetic thrust. Rotor electrical angle (unit: rad) is defined as the angle between the axis of the N pole of the rotor permanent magnet and the axis of the stator a-phase winding. It directly reflects the position state of the rotor and is a key state quantity that the sliding mode observer needs to accurately estimate in the subsequent S2 step.

[0041] (2) Two-phase stationary coordinate system (αβ) model (Clark transformation)

[0042] The model of the three-phase stationary coordinate system exhibits obvious coupling terms (self-inductance, mutual inductance, etc.). The large number of variables (three-phase) makes control algorithm design difficult. Therefore, the Clark transformation is used to transform the three-phase variables (…) The variables are transformed into two-phase stationary coordinate systems (α-axis and β-axis, where the α-axis coincides with the axis of phase a and the β-axis is perpendicular to the α-axis), thereby achieving variable dimensionality reduction and partial decoupling.

[0043] Transformation matrix design (equal amplitude transformation principle):

[0044] To ensure that the amplitudes of voltage and current remain unchanged before and after the transformation (to avoid deviations in energy calculation), an equal-amplitude transformation matrix is ​​used, as follows:

[0045] ;

[0046] Based on this transformation matrix, the conversion formula from three-phase variables to two-phase variables is as follows:

[0047] ;

[0048] Transformed αβ-axis voltage equations: Since the Clark transform is a linear transform, the form of the voltage equations remains consistent, with only the variables replaced by αβ-axis components:

[0049] ; ;

[0050] Stator voltage (unit: V) in a two-phase stationary coordinate system is the voltage signal after Clark transformation, reflecting the voltage excitation of the motor in the αβ axis direction; Stator current in two-phase stationary coordinate system (unit: A) is the current signal after Clark transformation, which is directly transmitted to the sliding mode observer in step S2 as the actual reference value for current estimation. Stator flux linkage (unit: Wb) in a two-phase stationary coordinate system reflects the flux linkage distribution along the αβ axis, providing a basis for estimating the back electromotive force of a sliding mode observer.

[0051] (3) Two-phase synchronous rotating coordinate system (dq) model (Park transformation)

[0052] Although the two-phase stationary coordinate system achieves dimensionality reduction, the magnetic flux and current still have a coupling relationship (due to rotor rotation). (Change). By converting the αβ axis variables into a dq coordinate system that rotates synchronously with the rotor (the d-axis coincides with the N pole axis of the rotor's permanent magnet, and the q-axis is perpendicular to the d-axis and rotates with the rotor), complete decoupling of electromagnetic torque and magnetic flux can be achieved—d-axis current ( The main influence is on the magnetic flux and the q-axis current. It primarily affects torque, enabling independent control of flux and torque.

[0053] Transformation matrix design:

[0054] Park transformation matrix and rotor electrical angle Related, specifically in the form of:

[0055] ;

[0056] Based on this matrix, the formula for converting the αβ axis variable to the dq axis variable is:

[0057] ;

[0058] Transformed dq-axis voltage equation:

[0059] ; ;

[0060] The core value of this equation lies in decoupling: the d-axis voltage equation... In the terms and q-axis voltage equations The term is a rotationally coupled term, which can be compensated for through a control algorithm, thereby achieving... and Independent control. Stator voltage (unit: V) in a two-phase synchronous rotating coordinate system is the target control quantity of the model predictive current control algorithm, which directly determines the magnetic flux and torque output of the motor. Stator current in a two-phase synchronous rotating coordinate system (unit: A). This refers to the magnetic flux component (controlling the air gap magnetic flux of the motor). For the torque component (controlling the electromagnetic thrust of the motor), this scheme adopts... The control strategy (i.e., constant magnetic flux, all stator current used to generate electromagnetic thrust, maximizing thrust output efficiency). The dq-axis stator inductance (unit: H) is derived from the inductance components of the two-phase stationary coordinate system. The specific expression is as follows: (Small air gap on the d-axis, low magnetic reluctance, large inductance) (The q-axis has a large air gap, high magnetic reluctance, and low inductance), which was determined through motor parameter identification experiments; Rotor electrical angular velocity (unit: rad / s) is defined as the rotor electrical angle. The derivative with respect to time ( The speed of the motor is reflected and is a key input parameter for predictive current control in the S3 step model.

[0061] The voltage output of the aforementioned multi-coordinate system model needs to be achieved through an inverter, which is the core hardware connecting the control algorithm and the permanent magnet synchronous linear motor. The control algorithm ultimately needs to generate the on / off sequence of the inverter's power switching elements, and its topology is as follows: Figure 2 As shown in the figure, the meanings of the markings are as follows: N is the DC bus voltage of the inverter, N is the neutral point on the inverter output side, n is the neutral point of the three-phase stator winding of the motor, and A, B, and C correspond to the three-phase stator winding terminals of the permanent magnet synchronous linear motor; the middle three-phase bridge arm (corresponding to...) , , Each contains two switching transistors (e.g.) and For phase a bridge arm upper and lower switch tubes, and For the upper and lower switching transistors of phase b bridge arm, and (This refers to the upper and lower switches of phase c bridge arm). The constraint of 'the upper and lower switches of the same bridge arm must not be turned on simultaneously' must be strictly followed to avoid short-circuit damage to the devices. This rule is the basis for subsequent voltage vector legal generation. Based on the above multi-coordinate system model and inverter topology, the overall control block diagram of the permanent magnet synchronous linear motor is as follows: Figure 4 As shown.

[0062] 1.3 Model Output: This step involves multi-coordinate system modeling, ultimately outputting two types of core data: one type is the inherent parameters of the motor (…). ), serving as the fundamental parameters for all subsequent algorithms; another type is the transformed electrical signal ( ).in, This will be directly passed to step S2, providing input for the current estimation and back EMF observation of the sliding mode observer; This will be passed to step S3 to provide parameter support for the construction of the discretized model of the model predicts current control.

[0063] S2: Construct an optimized sliding mode observer (SMO) to estimate rotor position and speed.

[0064] 2.1 Background and Core Objective of the Observer Construction: In intelligent modular logistics transportation systems, the installation space of permanent magnet synchronous linear motors is usually limited, and the operating environment is complex (frequent start-stop, vibration). Installing physical position sensors (such as encoders) on the motor shaft end presents problems such as maintenance difficulties, high costs, and weak anti-interference capabilities. Therefore, the core objective of this step is to construct a sensorless sliding mode observer. Based on the αβ axis voltage and current signals output from step S1, the rotor electrical angle can be accurately estimated without physical sensors. and rotor electric angular velocity Meanwhile, to address the shortcomings of traditional sliding mode observers, such as severe low-speed chattering, insufficient high-speed stability, and significant phase delay, improvements are made by optimizing the switching function and introducing a phase-locked loop (PLL). This enhances the robustness and estimation accuracy of the observer, providing reliable rotor state information for model predictive current control in step S3.

[0065] 2.2 Optimize the sliding mode observer model construction and training process

[0066] (1) Current estimation model design (core observation logic)

[0067] The sliding mode observer estimates the back electromotive force and rotor state by driving the switching function through the current estimation error. Its core components are as follows: Figure 5 As shown. The core idea of ​​the sliding mode observer is to construct an observation model that matches the actual dynamic characteristics of the motor current, and then force the observed current through the "switching characteristic" of sliding mode control. Tracking actual current When the system enters sliding mode, the switching term in the observation model can be equivalent to the back electromotive force of the actual motor (including rotor position information). Based on the αβ-axis voltage equation of step S1, the current estimation equation of the sliding mode observer is designed as follows:

[0068] ;

[0069] ;

[0070] The structure of the observation model is consistent with the αβ axis voltage equations of the actual motor, except that sliding mode control terms are introduced. and —The function of this control item is to eliminate the error between the observed current and the actual current by rapidly switching, forcing , . The αβ-axis current (in A) estimated by the sliding mode observer is the output of the observation model, used to compare with the actual current. Compare and generate error signals; Sliding mode gain (unit: V·s / A) is a key parameter that determines the response speed and stability of the sliding mode observer. The larger the gain, the faster the observed current tracks the actual current, but the more severe the chattering at low speeds. The smaller the gain, the less chattering, but it may not meet the stability requirements at high speeds. The initial value is 50~100, and it needs to be optimized through experimental debugging. The αβ axis current estimation error (unit: A) is the input signal of the sliding mode control term, which directly reflects the degree of deviation between the observed current and the actual current. : Saturation function (replacing the sign function of the traditional sliding mode observer) The core function is to suppress the chattering problem inherent in sliding mode control—the output of the sign function is step-like (-1 or 1), which causes drastic fluctuations in the observed current; while the saturation function within the boundary layer ( The output is linear, while the output outside the boundary layer is constant, allowing for smooth switching and reducing chattering. Its specific expression is:

[0071] Where a is the boundary layer thickness of the saturation function (unit: A), and the value ranges from 0.05 to 0.1 A. The larger a is, the better the chattering suppression effect, but the current estimation accuracy will decrease slightly; the smaller a is, the higher the estimation accuracy, but the chattering suppression effect is weakened. A balance needs to be found between chattering and accuracy.

[0072] (2) Back EMF observation (sliding mode equivalent control principle)

[0073] When the sliding mode observer enters a stable sliding mode state, the current estimation error... The current converges to near 0, at which point it is observed. With actual current Almost identical, the observed value of the back electromotive force can be derived by subtracting the current estimation equation from the actual αβ-axis voltage equation in step S1. .

[0074] Detailed derivation process:

[0075] 1. Equations for the αβ axis voltages of a real motor (Step S1): ( (the actual back electromotive force).

[0076] 2. Current estimation equation for sliding mode observer: ;

[0077] 3. In sliding mode Subtracting the two equations, we get: ;

[0078] 4. Final back electromotive force observation:

[0079] ; ;

[0080] The physical meaning and core function of back electromotive force: Back electromotive force is the induced electromotive force generated when the permanent magnet cuts the stator windings during the rotor motion of the motor. Its phase is related to the rotor electrical angle. Directly related, the specific relationship is as follows Therefore, the observed back electromotive force... It is the core carrier for extracting rotor position and speed information. The ultimate goal of this step is to indirectly obtain information through back electromotive force. .

[0081] (3) Observer stability verification (Lyapunov function method)

[0082] To ensure the stable operation of the sliding mode observer under different speeds and load conditions (with convergence of current estimation errors), the stability of the observer needs to be verified using Lyapunov stability theory, and the sliding mode gain needs to be determined. The range of values ​​for .

[0083] Stability verification process:

[0084] 1. Define the Lyapunov function. This function is a positive definite function. Only when and (when V=0).

[0085] 2. Differentiate V with respect to time: ,in , ;

[0086] 3. Substitute the current estimation equation and the actual voltage equation into... Simplifying, we get: ;

[0087] 4. To ensure stability, the following must be met. (The Lyapunov function is non-increasing). Combining this with the characteristics of saturated functions, the stability condition is derived: .

[0088] Observer training process: A permanent magnet synchronous linear motor model and a sliding mode observer model are built in the Simulink simulation environment to simulate typical operating conditions of a logistics transportation system (speed range: 50~1500rad / s, load fluctuation range: ±20% of rated load), and the sliding mode gain is gradually adjusted. And boundary layer thickness a:

[0089] 1. Initial Setup =50, a=0.08A, observe the current estimation error. The convergence status;

[0090] 2. If chattering is severe at low speeds (error fluctuation > ±0.05A), increase a to 0.1A, or decrease it. Up to 40;

[0091] 3. If the error cannot converge at high speed (error > ±0.03A), then increase the... Up to 60-80;

[0092] 4. Repeatedly adjust until the current estimation error converges to within ±0.01A and the rotor electrical angle estimation error is ≤0.5rad under the full speed range and load fluctuations, to ensure the stability and accuracy of the observer.

[0093] 2.3 Rotor position and speed extraction (phase-locked loop optimization)

[0094] Through back electromotive force observations While the rotor electrical angle can be initially calculated, directly using the arctangent function introduces a phase delay (due to the continuous nature of the saturation function), and the harmonic components in the back electromotive force affect the smoothness of the speed estimation. Therefore, a second-order phase-locked loop (PLL) is introduced to correct the initially estimated rotor position and simultaneously extract the smoothed rotor electrical angular velocity.

[0095] (1) Rotor electrical angle Preliminary estimates and PLL correction

[0096] Preliminary estimation: Based on the phase relationship of the back electromotive force, the equivalent rotor position is calculated using the arctangent function. : The derivation of this formula is based on the expression for back electromotive force ( ). Dividing the two equations, we get Thus, the solution is obtained. .

[0097] Phase-locked loop (PLL) calibration:

[0098] A second-order phase-locked loop consists of a phase detector (PD), a loop filter (LF), and a voltage-controlled oscillator (VCO). Its core function is to track... The phase change compensates for the phase delay and suppresses harmonic interference.

[0099] 1. Phase detector: Calculates a preliminary estimated position. PLL output position angular error ;

[0100] 2. Loop filter: A PI controller is used to filter angle errors. Perform filtering and amplification to output control signals. The PI controller parameters are designed as follows: =2h, (Where h is the distance from the pole to the origin in the complex plane, ranging from 5 to 10. The smaller h is, the lower the cutoff frequency of the PLL and the stronger the anti-harmonic interference capability, but the slower the dynamic response; the larger h is, the faster the dynamic response, but the weaker the anti-interference capability. In this scheme, h=8 is chosen.)

[0101] 3. Voltage-controlled oscillator: This transducers the control signal... The corrected rotor electrical angle is obtained by integration. Simultaneously output rotor electric angular velocity .

[0102] (2) Rotor electric angular velocity Extraction

[0103] Rotor electric angular velocity There are two ways to obtain it: one is to directly use the output of a PLL voltage-controlled oscillator. (Good smoothness and strong anti-interference ability); secondly, the corrected rotor electrical angle Perform numerical differentiation ( The preferred approach in this scheme avoids amplifying high-frequency noise through differential operations, thus ensuring the smoothness of the speed signal.

[0104] 2.4 Data Output and Step Connection

[0105] This step optimizes the sliding mode observer to ultimately output two core state variables: rotor electrical angle. and rotor electric angular velocity These two state variables will be directly passed to the model predictive current control algorithm in step S3, where... Coordinate transformations used for Park transformation and inverse Park transformation. Used for constructing discretized current prediction models and calculating the outer velocity loop reference current, achieving precise connection between steps S2 and S3.

[0106] S3: Design Model Predictive Current Control (MPCC) to determine the optimal voltage vector

[0107] 3.1 Background and Core Objective of Control Algorithm Design: The operating conditions of intelligent modular logistics transportation systems are complex and variable. Permanent magnet synchronous linear motors need to cope with frequent load fluctuations (such as changes in cargo weight) and motor parameter drift (such as increased resistance due to temperature rise), as well as other external and internal disturbances. Traditional PI control algorithms struggle to balance dynamic response speed and steady-state control accuracy in nonlinear, highly disturbed environments. Therefore, the core objective of this step is to design a model predictive current control algorithm based on the dq-axis mathematical model in step S1 and the rotor state variables estimated in step S2. By using the logic of "predicting future current state - enumerating all possible control inputs - minimizing the objective function", the optimal voltage vector is selected in real time, enabling the motor current to quickly track the reference current, improving the system's robustness to load fluctuations and parameter drift, while taking into account dynamic response and steady-state accuracy.

[0108] 3.2 Details of Model Predictive Current Control Model Construction

[0109] (1) Current prediction model (forward Euler discretization)

[0110] Model predictive control is a control method based on discretized models. It requires discretizing the continuous-time dq-axis voltage equations in step S1 to predict the current state at future moments. Considering the one-step delay in the digital control system (the control quantity calculated at the current moment can only be applied to the motor at the next moment), to ensure control accuracy, the current at moment k+2 needs to be predicted (to compensate for the one-step delay).

[0111] Discretization method selection: Forward Euler discretization is adopted (low computational cost, good real-time performance, suitable for the rapid control requirements of logistics systems), with the discretization step size being the sampling period. (Unit: s), value range 100~200μs (matching the inverter switching frequency, switching frequency 5~10kHz).

[0112] Derivation of Discretized Current Prediction Model:

[0113] 1. Continuous-time dq-axis voltage equation (S1 step):

[0114] ; ;

[0115] 2. Forward Euler discretization formula: ;

[0116] 3. Substituting into the continuity equation, the predicted current value at time k+1 is derived:

[0117] ;

[0118] ;

[0119] 4. Compensation for 1-step delay: Based on the current prediction value at time k+1, further predict the current at time k+2 (the control quantity at time k+1). (Applies at time k+2)

[0120] ;

[0121] ;

[0122] : Discrete time step, k is the current time, k+1 is the next sampling time, k+2 is the next sampling time; : The actual dq-axis current at the current time (k) (obtained by the Park transformation in step S1); Predicted dq-axis current at time k+1; The predicted dq-axis current at time k+2 is the core evaluation object of the objective function; The dq axis voltage control quantity at time k+1 is determined by the switching state of the inverter; The rotor electric angular velocity at time k+1 (predicted by the sliding mode observer in step S2, assuming...) The variation is small within the sampling period and can be approximated as ).

[0123] (2) Definition of objective function (minimizing current tracking error)

[0124] The core of model predictive current control is "selecting the optimal voltage vector to optimize the control objective". The core control objective of this scheme is "the current quickly tracks the reference current" (the current tracking accuracy directly determines the control accuracy of the motor torque and thrust). Therefore, the objective function is defined as the sum of squares of the current tracking error (the square term can amplify the error, enhance the penalty for large deviations, and avoid the cancellation of positive and negative errors).

[0125] Objective function expression: ;

[0126] Detailed explanation of the objective function:

[0127] 1. Weighting coefficient settings: Because this scheme adopts... The control strategy (constant magnetic flux). and The control priorities are equally important, therefore weighting coefficients are set. , (If torque control is required, it can be set) );

[0128] 2. Reference current source: The magnetic flux component (unit: A) in the dq axis current reference value is fixed at 0A to ensure constant air gap magnetic flux of the motor and maximize thrust output efficiency. The torque component (unit: A) in the dq-axis current reference value is calculated by the speed outer-loop PI controller and used to track the speed requirements of the logistics system. The expression for the speed outer-loop PI controller is:

[0129] ;

[0130] in, The rotor reference angular velocity (unit: rad / s, issued by the system's host computer based on the transportation task) is required by the logistics transportation system. , (This was determined through experimental debugging to ensure that the speed response has no overshoot and the steady-state error is ≤2%).

[0131] 3.3 Voltage Vector Enumeration and Optimal Selection

[0132] The switching state of a three-phase two-level inverter determines the type of output voltage vector. Each bridge arm (phase a, phase b, phase c) of the inverter has two switching states (1 indicates the upper bridge arm is on and the lower bridge arm is off; 0 indicates the upper bridge arm is off and the lower bridge arm is on), therefore there are a total of The various switch combinations correspond to eight basic voltage vectors (six active vectors and two zero vectors), and their spatial distribution is as follows: Figure 3 As shown. Model predictive current control requires enumerating all 8 voltage vectors, calculating the objective function value for each vector, and selecting the vector with the smallest objective function as the optimal voltage vector. .

[0133] (1) Complete enumeration of 8 basic voltage vectors (including switching states and dq axis voltages)

[0134]

[0135] The switching states (1 / 0) of the inverter's phase a, phase b, and phase c bridge arms are the final control signals that drive the inverter. The inverter's DC bus voltage (unit: V) is provided by the power supply module of the logistics system (e.g., 380V DC bus), and is a constant value measured and calibrated by a voltage sensor; Effective vector: The output voltage is not zero, which can drive the motor to generate electromagnetic thrust. The six effective vectors are evenly distributed in the αβ coordinate system (adjacent vectors are 60° apart), covering all possible voltage directions; Zero vector: The output voltage is zero, and the motor has no thrust output. It is used to regulate the average voltage during the switching cycle and reduce current ripple and switching losses.

[0136] (2) The complete process of selecting the optimal voltage vector

[0137] 1. Input parameter acquisition: Obtain the actual dq-axis current at the current time (k). (S1 step Park transformation output), rotor electric angular velocity (S2 step sliding mode observer output), rotational speed reference value (Sent from the system's host computer), DC bus voltage (Actual measurement);

[0138] 2. Calculate the reference current Calculated via speed outer loop PI controller , Fixed at 0A;

[0139] 3. Enumerate 8 voltage vectors: Extract the dq-axis voltage of each voltage vector in sequence ( );

[0140] 4. Current prediction calculation: Calculate the current for each vector. Substitute into the discretized model of S3.2(1), and calculate the corresponding... ;

[0141] 5. Objective function evaluation: For each voltage vector, calculate the value of the objective function J (e.g., ...). correspond , correspond , ..., correspond );

[0142] 6. Optimal Vector Selection: Compare the eight J values ​​and select the voltage vector with the smallest J value as the optimal voltage vector. (For example, if) ,but );

[0143] 7. Output the optimal vector: Record Corresponding switch state ( ), and pass it to step S4.

[0144] 3.4 Data Output: This step uses a model-predictive current control algorithm to ultimately output the optimal voltage vector. and their corresponding switching states ( The switch status information will be directly transmitted to the space vector pulse width modulation module in step S4, serving as the core basis for generating the inverter drive signal and realizing the logical closed loop of "rotor state estimation - current prediction control - voltage vector selection".

[0145] S4: Generate inverter switching signals based on SVPWM and implement closed-loop feedback.

[0146] 4.1 SVPWM Module Design Background and Core Purpose

[0147] The optimal voltage vector output by step S3 It is a discrete combination of switch states (such as...) (Corresponding to switch state 110), directly outputting the discrete switch state to the inverter would cause drastic fluctuations in motor current (abrupt switching state), resulting in significant torque ripple and noise, affecting the operational stability of the logistics transportation system. Therefore, the core objective of this step is to construct a space vector pulse width modulation (SVPWM) module to convert the discrete optimal voltage vector... The signal is converted into a continuous, smooth PWM (Pulse Width Modulation) drive signal. By adjusting the duration of the voltage vector, the average value of the inverter output voltage is made equal to... This achieves smooth voltage output; at the same time, the actual operating current of the motor is fed back to step S1 in real time, forming a complete closed-loop control and improving the system's ability to resist external disturbances.

[0148] 4.2 Details of Space Vector Pulse Width Modulation (SVPWM) Model Construction and Application

[0149] The core principle of SVPWM is "volt-second balance"—within a sampling period Within this system, through the combined action of two adjacent effective voltage vectors and a zero vector, the volt-second value of the synthesized voltage is made equal to that of the optimal voltage vector. The volt-second value, thereby achieving the control of The equivalent output. The specific process includes three stages: sector determination, duty cycle calculation, and switch signal generation.

[0150] (1) Sector determination (determining the region where the voltage vector is located)

[0151] First, the optimal voltage vector The dq-axis voltage is converted to the αβ-axis voltage (through the inverse Park transform, based on the rotor electrical angle output in step S2). Then, based on the sign and magnitude relationship of the αβ axis voltages, determine... The sector in the αβ coordinate system (the αβ coordinate system is divided into 6 sectors by 6 effective voltage vectors, each sector is 60°).

[0152] Inverse Park transform formula (converts dq-axis voltage to αβ-axis voltage):

[0153] ; ;

[0154] in For the optimal voltage vector Voltage components along the αβ axis.

[0155] Complete judgment conditions for 6 sectors:

[0156]

[0157] The core function of sector identification: to clarify The adjacent valid vectors are then used to combine these two adjacent valid vectors with the zero vector to achieve [the following]. The equivalent output.

[0158] (2) Duty cycle calculation (volt-second balance principle)

[0159] Duty cycle refers to the time interval between sampling cycles for each voltage vector. The duration of action within the time frame is calculated based on the volt-second balance principle, using adjacent effective vectors ( ) and zero vector ( The duration of action is determined to ensure that the volt-second value of the synthesized voltage is equal to the applied voltage. The volt-second value.

[0160] Taking sector 2 as an example (adjacent effective vectors) The duty cycle calculation process is as follows:

[0161] 1. Volt-second equilibrium equation: ( (time of action of zero vector);

[0162] 2. Substituting the α and β axis voltages into the equation, we can decompose it into volt-second balance equations for the α and β axes.

[0163] 3. Solve the simultaneous equations to obtain the duration of action of adjacent effective vectors. and the action time of the zero vector :

[0164] ; ; ;

[0165] General rules for duty cycle calculation:

[0166] 1. The duty cycle calculation for all sectors is based on the volt-second balance principle. The voltage component expressions in the equations differ slightly only because adjacent effective vectors are different, but the calculation logic is the same.

[0167] 2. Zero vector action time It is obtained by subtracting the duration of two effective vectors from the sampling period. To avoid sudden current changes caused by frequent switching states, [the following is omitted]. Distribute equally to the two zero vectors ( effect , effect );

[0168] 3. If the calculated result is... or Negative numbers ( If the vector is close to the sector boundary, set it to 0 and recalculate the duration of another valid vector to ensure that the duration is non-negative.

[0169] Adjacent valid vectors The duration of action (in seconds) determines the contribution of the effective vector to the synthesized voltage; The total duration of the zero vector (in seconds) is used to adjust the average voltage within the sampling period and reduce current ripple.

[0170] (3) Switch signal generation (timing optimization)

[0171] Based on the sector judgment results and duty cycle calculation results, the timing sequence of the PWM wave is designed according to the principle of "minimizing the number of switching state transitions" to reduce the switching losses of the inverter switching transistors and ensure smooth output voltage.

[0172] With sector 2 (adjacent effective vectors) Zero vector Taking as an example, the timing design of the switching signal is as follows (one sampling period) Inside):

[0173] 1. Initial stage: Output zero vector (Switch status 000), duration ;

[0174] 2. First effective vector stage: Output (Switch status 100), duration ;

[0175] 3. Second effective vector stage: Output (Switch status 110), duration ;

[0176] 4. Final Stage: Output zero vector (Switch status 111), duration ;

[0177] 5. Repeat the above timing sequence to form a continuous PWM drive signal.

[0178] The core principles of timing design:

[0179] 1. Adjacent switch states only change the switch state of one bridge arm (e.g., (000) → (100) Only change phase a bridge arm), reduce switching losses;

[0180] 2. The zero vectors are placed at the beginning and end of the sampling period, so that the effective vector's duration is concentrated in the middle, thus improving the voltage synthesis accuracy;

[0181] 3. The timing design of all sectors follows the above principles, and the switching order of the switching states is adjusted only according to the differences between adjacent valid vectors.

[0182] 4.3 Closed-loop feedback and algorithm synergy integration

[0183] The final output of this step is the inverter's PWM drive signal, which drives the inverter to output a smooth three-phase voltage to the permanent magnet synchronous linear motor, controlling the motor to generate electromagnetic thrust that meets the needs of logistics transportation. Simultaneously, to achieve robust control of the system, a complete closed-loop feedback mechanism must be constructed to feed back the actual operating status of the motor to the preceding steps in real time, enabling the coordinated operation of each algorithm.

[0184] (1) Closed-loop feedback process

[0185] 1. Current Acquisition: The actual three-phase stator current of the motor is acquired in real time using Hall effect current sensors. ;

[0186] 2. Signal conversion: [This refers to the process of converting signals into signals.] Passed to step S1, transformed by Clark transformation shaft current Then, it is converted into dq-axis current through Park transformation. ;

[0187] 3. Status Feedback: Feedback is sent to step S2 as the actual current reference value for the sliding mode observer, used to correct the current estimation error and the back EMF observation value; Feedback is sent to step S3 as the current at the current moment for model-predicted current control, which is used to update the current prediction model;

[0188] 4. Dynamic adjustment: If the logistics system experiences load fluctuations or motor parameter drift, the deviation between the actual current and the reference current will be transmitted to step S3 through the feedback mechanism. The model predicts the current control, which will re-enumerate the voltage vector, select a new optimal vector, and adjust the inverter output voltage to enable the current to quickly track the reference value and achieve disturbance rejection control.

[0189] (2) Algorithm collaborative fusion mechanism

[0190] The algorithms for each step do not run independently, but rather achieve deep collaboration through data flow and feedback mechanisms. The core collaborative relationships are as follows:

[0191] 1. Collaboration between S1 and S2: S1 provides... Shaft voltage and current are the input basis for the S2 sliding mode observer, and the rotor electrical angle estimated by S2 is... It also feeds back into the Park transformation of S1 (Park transformation matrix dependency) This ensures the accuracy of coordinate system transformations;

[0192] 2. Synergy between S2 and S3: S2 estimation It is the key state variable for predictive current control in the S3 model — Used for coordinate transformation of voltage vectors, Used for constructing current prediction models; if the speed estimation deviation of S2 is >5 rad / s, S3 will automatically adjust the parameters of the speed outer loop PI controller (increase the speed). This improves dynamic response speed and compensates for speed estimation errors.

[0193] 3. Coordination between S3 and S4: Optimal voltage vector output by S3 It is the input of S4SVPWM. S4 converts the discrete vector into a continuous PWM signal through duty cycle calculation and timing design; if zero vector ( or S4 will automatically extend the zero vector action time. This reduces motor energy consumption;

[0194] 4. Closed-loop collaboration throughout the entire process: Through closed-loop feedback of “S1→S2→S3→S4→S1”, the entire process of “motor modeling → state estimation → control decision → drive output → state feedback” is coordinated to ensure the robustness and stability of the system under complex working conditions.

[0195] Overview of core contributions and data flow at each step

[0196]

[0197] In some embodiments, an adaptive compensation algorithm is designed to correct motor parameters in real time based on the current tracking error of model-predicted current control, so that the corrected parameters... By consistently matching the actual characteristics of the motor, the parameter adaptive compensation algorithm is deeply integrated with MPCC and sliding mode observer (SMO) to further improve the robustness of the system.

[0198] Model building of parameter adaptive compensation algorithm

[0199] (1) Definition of core parameter error

[0200] The estimation error of motor parameters is defined as the difference between the actual parameters and the nominal parameters:

[0201] ;

[0202] Stator resistance, d-axis inductance, and q-axis inductance during actual motor operation (dynamically changing with operating conditions); : The nominal parameters of the motor at the time of manufacture (initial values, provided by the motor nameplate); : Parameter error (which needs to be estimated using an adaptive algorithm).

[0203] (2) Adaptive law design (based on current prediction error)

[0204] The core of the parameter adaptive compensation algorithm is to derive the adaptive update law of parameter error based on the current prediction error of the MPCC. Let the predicted current at time k+2 in the MPCC be... The actual feedback current is The current prediction error is defined as:

[0205] ;

[0206] Based on model reference adaptive theory, an adaptive update law (proportional-integral type) for design parameter errors is proposed:

[0207] ;

[0208] ;

[0209] ;

[0210] : Estimated value of parameter error (output of adaptive algorithm); Adaptive gain (all positive numbers, determined through stability analysis, initial value taken as follows) ); : Sign function (enhances the algorithm's response speed to large errors); The differential of the predicted d / q-axis current at time k (calculated via forward difference): ); : Sampling period (consistent with the discretization step size of MPCC).

[0211] Corrected parameter calculation:

[0212] The corrected parameters output by the adaptive algorithm are the sum of the nominal parameters and the parameter error estimates:

[0213] ;

[0214] The process of integration and interaction with algorithms:

[0215] (1) Interaction with Model Predictive Current Control (MPCC)

[0216] 1. Input Interaction: The parameter adaptive algorithm will correct the parameters. Input the current prediction model of MPCC, replace the original nominal parameters, and the corrected prediction model is as follows:

[0217] ;

[0218] ;

[0219] 2. Output Interaction: Current Prediction Error Calculated by MPCC As input to the parameter adaptation algorithm, it drives the parameter error estimate. The update forms a positive cycle of "parameter correction → improved prediction accuracy → reduced error → further parameter optimization".

[0220] (2) Interaction with Sliding Mode Observer (SMO)

[0221] The current estimation model of the sliding mode observer also depends on the motor inductance parameters, and the parameter adaptive algorithm will correct the inductance. (because It can be deduced The corrected current estimation equation is obtained by inputting SMO:

[0222] ;

[0223] ;

[0224] This improves the accuracy of current estimation in the SMO, making the observed back electromotive force more accurate. More accurate, thus optimizing rotor electrical angle and rotational speed The estimation results, and more accurate ones. It will also feed back into the voltage vector conversion and current prediction of MPCC, achieving synergistic optimization of the three algorithms.

[0225] Model training and application process:

[0226] (1) Training process (offline parameter tuning)

[0227] Build a motor model with parameter drift in Simulink (simulating temperature-induced drift). Increase by 20%, magnetic saturation leads to Reduce by 15%), and tune the adaptive gain using the following steps:

[0228] 1. Fixed Only the proportional item is retained. Gradually increase the gain until the parameter error is reached. It converges to within 5%, but without oscillation;

[0229] 2. With a fixed proportional gain, gradually increase the integral gain. ), until the steady-state parameter error converges to within 2%;

[0230] 3. Finalize the gain combination: ; ; .

[0231] (2) Application process (online real-time correction)

[0232] 1. Initialization: When the system starts, Use nominal parameters ;

[0233] 2. Real-time calculation: for each sampling period Internal MPCC output current prediction error The parameter adaptive algorithm updates the parameters according to the adaptive law. ;

[0234] 3. Parameter limiting: To avoid over-correction of parameters, constraints are set. ;

[0235] 4. Collaborative Output: The corrected parameters are synchronously input into MPCC and SMO to update their internal models, completing one collaborative optimization.

[0236] Core role and contribution:

[0237] Parameter drift compensation: through real-time correction This solves the problem of decreased prediction accuracy of traditional MPCC when parameters change, and enables the current tracking error to be controlled within 5% even in the extreme case of 20% parameter drift;

[0238] Multi-algorithm synergistic gain: The interaction with SMO improves the accuracy of rotor state estimation, and the interaction with MPCC improves the accuracy of current control, forming a closed-loop synergy of "parameter correction → state estimation optimization → control decision optimization", which improves the overall adaptability of the system to complex logistics conditions.

[0239] Enhanced robustness: It can adapt to changes in motor characteristics without human intervention, reducing thrust fluctuations and efficiency declines caused by parameter mismatch, and extending the fault-free operation time of the logistics transportation system.

Claims

1. A robust current control method for permanent magnet synchronous linear motor, characterized in that, The method comprises the following steps: S1: Obtain the three-phase stator phase voltage and the three-phase stator phase current of the permanent magnet synchronous linear motor, establish the motor mathematical model in the three-phase stationary coordinate system, the two-phase stationary coordinate system and the two-phase synchronous rotating coordinate system, convert the three-phase voltage and current into the voltage and current in the two-phase stationary coordinate system through the Clark transformation, and then convert the voltage and current into the voltage and current in the two-phase synchronous rotating coordinate system through the Park transformation to obtain the motor inherent parameters; S2: Input the voltage and current in the two-phase stationary coordinate system into the optimized sliding mode observer, suppress the chattering through the saturation function, combine the phase delay compensation of the phase-locked loop, and estimate the rotor electric angle and the rotor electric angular velocity; S3: Input the motor inherent parameters, the current in the two-phase synchronous rotating coordinate system, the rotor electric angle and the rotor electric angular velocity into the model predictive current control algorithm, construct the current prediction model based on the forward Euler discretization, compensate the one-step delay of the digital control, define the target function of the current tracking error minimization, determine all the voltage vectors of the three-phase two-level inverter, calculate the target function value corresponding to each voltage vector, select the voltage vector with the minimum target function value as the optimal voltage vector, and obtain the switching state corresponding to the optimal voltage vector; S4: Input the rotor electric angle and the optimal voltage vector into the space vector pulse width modulation module, convert the optimal voltage vector into the voltage component in the two-phase stationary coordinate system through the inverse Park transformation, judge the sector where the voltage component is located, calculate the action time of the adjacent effective voltage vector and the zero vector based on the volt-second balance principle, generate the PWM driving signal of the inverter according to the principle of the minimum switching state switching frequency, and drive the permanent magnet synchronous linear motor to operate; Real-time acquisition of the three-phase stator actual current of the permanent magnet synchronous linear motor during operation is performed, the three-phase stator actual current is taken as the input of step S1, a closed-loop control is formed, steps S1 to S4 are continuously executed, and the robust current control of the permanent magnet synchronous linear motor is realized; In the model predictive current control algorithm of step S3, the target function is designed by multi-objective optimization, in addition to the current tracking error, a switching loss penalty term is additionally introduced, the penalty coefficient is dynamically adjusted to balance the current control accuracy and the inverter switching loss, the penalty coefficient is adaptively adjusted according to the rotor electric angular velocity, the penalty coefficient is increased to reduce the switching frequency under high-speed working conditions, and the penalty coefficient is reduced to improve the control accuracy under low-speed working conditions; The voltage vector enumeration adopts the pruning strategy, the adjacent three voltage vectors with high probability of being the optimal solution are selected based on the current deviation direction and the rotor electric angular velocity, the target function of the selected voltage vectors is calculated, the enumeration calculation amount is reduced, the real-time performance of the control algorithm is improved, and the selection basis is the matching degree of the direction of the voltage vector and the current deviation compensation demand; The current prediction model introduces a parameter adaptive correction term, the inductance and resistance values in the motor inherent parameters are adjusted in real time based on the deviation between the feedback actual current and the predicted current, the prediction error caused by the parameter drift is compensated, and the weight of the correction term is increased with the increase of the deviation.

2. The robust current control method of a permanent magnet synchronous linear motor according to claim 1, characterized in that, In step S2, the saturation function adopts linear output in the boundary layer and constant output outside the boundary layer, and the thickness of the boundary layer is determined through simulation debugging to make the current estimation error converge to a preset range, thereby suppressing the inherent chattering problem of the sliding mode control.

3. The robust current control method of a permanent magnet synchronous linear motor according to claim 1, characterized in that, In step S2, the phase-locked loop includes a phase discriminator, a loop filter and a voltage-controlled oscillator, the phase discriminator is used to calculate the angle error between the preliminary estimated rotor electrical angle and the output rotor electrical angle of the phase-locked loop, the loop filter is used to filter and amplify the angle error, and the voltage-controlled oscillator is used to output the corrected rotor electrical angle and the rotor electrical angular velocity, the proportional coefficient and the integral coefficient of the loop filter are designed by the pole placement method to balance the anti-interference ability and the dynamic response speed.

4. The robust current control method of a permanent magnet synchronous linear motor according to claim 1, characterized in that, In step S3, the current prediction model predicts the current at k+1 time step through forward Euler discretization of the voltage equation in the two-phase synchronous rotating coordinate system, and then predicts the current at k+2 time step based on the current at k+1 time step, wherein k is the current discrete time step, the two-step prediction compensates for the one-step delay of the digital control system, and ensures that the action time of the control quantity is consistent with the prediction time of the current.

5. The robust current control method of a permanent magnet synchronous linear motor according to claim 1, wherein, In step S3, the weight coefficients of the target function are all set to 1, wherein the magnetic flux component reference current is fixed to zero, and the torque component reference current is calculated by a speed outer loop PI controller, the input of the speed outer loop PI controller is the difference between the rotor reference angular velocity and the rotor electrical angular velocity output by step S2, and the rotor reference angular velocity is issued by the system host computer according to the logistics transportation task.

6. The robust current control method of a permanent magnet synchronous linear motor according to claim 1, wherein, In step S3, all voltage vectors of the three-phase two-level inverter include 6 effective voltage vectors and 2 zero vectors, a total of 8 voltage vectors, the switching state corresponding to each voltage vector is determined, and the current at k+2 time step and the target function value under each switching state are calculated one by one, k is the current discrete time step.

7. The robust current control method of a permanent magnet linear motor according to claim 1, wherein, In step S4, the sector judgment is based on the positive and negative and amplitude relationship of the voltage components in the two-phase static coordinate system, the two-phase static coordinate system is divided into 6 sectors, each sector corresponds to two adjacent effective voltage vectors, and the optimal voltage vector and the effective voltage vector cooperatively synthesized are determined according to the numerical characteristics of the voltage components.

8. The robust current control method of a permanent magnet synchronous linear motor according to claim 1, wherein, In step S4, the action time of the adjacent effective voltage vector is solved by the volt-second balance principle, the total action time of the zero vector is the sampling period minus the action time of the two adjacent effective voltage vectors, and the action time of the zero vector is evenly distributed to the two zero vectors and placed at the beginning and end of the sampling period to reduce the switching state switching frequency.

9. The robust current control method of a permanent magnet synchronous linear motor according to claim 1, wherein, In the space vector pulse width modulation module of step S4, a fast sector judgment algorithm is adopted, the optimal voltage vector sector is directly determined through the sign combination and amplitude comparison of the voltage components in the two-phase static coordinate system, so as to shorten the sector judgment time, the sign combination reflects the positive and negative directions of the voltage components, and the amplitude comparison reflects the relative size of the voltage components. The duty ratio of the PWM driving signal adopts a dynamic adjustment mechanism, the DC bus voltage of the inverter is collected in real time, if the bus voltage fluctuates, the action time of the adjacent effective voltage vector and the zero vector is corrected based on the fluctuation amplitude, so that the average value of the synthesized voltage always tracks the optimal voltage vector, and the influence of the bus voltage fluctuation on the control accuracy is offset.

Images