Adaptive robust control method and system for multiphase motor drive system

Abstract

The application discloses a self-adaptive robust control method and system of a multiphase motor driving system, and relates to the technical field of multiphase motor driving control. The method comprises the following steps: collecting the running state data of a double three-phase permanent magnet synchronous motor, calculating the speed error between the current speed and the target speed, and adjusting the gain value of the sliding mode control, the approaching law form and the integral coefficient of the sliding surface; estimating the parameter variation, the unmodeled dynamic quantity and the external disturbance quantity in the multiphase motor driving system through a state observer, and generating an uncertainty compensation quantity; generating a control instruction based on the gain value, the approaching law form, the integral coefficient and the uncertainty compensation quantity, and outputting the control instruction to a six-phase modular multilevel converter to drive the double three-phase permanent magnet synchronous motor to run. The application improves the robustness and dynamic performance of the multiphase motor driving system under the conditions of parameter variation, load disturbance and fault.

Description

Technical Field

[0001] This invention relates to the field of multiphase motor drive control technology, specifically to an adaptive robust control method and system for multiphase motor drive systems. Background Technology

[0002] The dual three-phase permanent magnet synchronous motor consists of two sets of three-phase windings, enabling it to maintain operation even in the event of partial faults in the windings or power devices, thus exhibiting excellent redundancy and fault tolerance. Modular multilevel converters, due to their high output voltage levels, low harmonic content, and strong modular scalability, have been widely used in high-power converter applications. Combining a six-phase modular multilevel converter with a dual three-phase permanent magnet synchronous motor can effectively improve system power density and operational reliability.

[0003] Existing technologies typically employ vector space decomposition or dual dq coordinate transformation, combined with proportional-integral control, model predictive control, or harmonic suppression control, to achieve motor current tracking and torque regulation. In converter control, strategies such as submodule capacitor voltage balancing, circulating current suppression, and carrier phase-shift modulation ensure normal system operation. However, existing control methods still have significant drawbacks: designs based on fixed models and parameters lack effective compensation for model mismatch, leading to decreased current regulation accuracy and increased torque output deviation under parameter changes or fault conditions; their dynamic adjustment capability under large disturbances is insufficient, making it difficult to quickly restore stable operation and prone to current surges or torque fluctuations; while some sliding mode control schemes can enhance robustness, fixed-gain designs are prone to introducing high-frequency chattering in the steady-state phase, affecting current quality; furthermore, there is a lack of a unified control framework between normal and fault operating modes, requiring control structure switching during state transitions, which can easily cause control discontinuities and transient performance degradation. Summary of the Invention

[0004] This invention addresses the technical problems in existing technologies, such as the limited dynamic adaptability of six-phase modular multilevel converters driving dual three-phase permanent magnet synchronous motors, the tendency of sliding mode control to introduce high-frequency chattering, and the lack of a unified control framework for normal and fault operation modes. It provides an adaptive robust control method and system for multiphase motor drive systems.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides an adaptive robust control method for a multiphase motor drive system, comprising: The operating status data of the dual three-phase permanent magnet synchronous motor is collected, the speed error between the current speed and the target speed is calculated, and the gain value of the sliding mode control, the approach law form of the sliding mode control, and the integral coefficient of the sliding surface are adjusted according to the speed error. The parameter changes, unmodeled dynamics, and external disturbances in the multiphase motor drive system are estimated online using a state observer to generate uncertainty compensation quantities. Based on the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount, a control command is generated and output to the six-phase modular multilevel converter to drive the dual three-phase permanent magnet synchronous motor.

[0006] In a second aspect, the present invention provides an adaptive robust control system for a multiphase motor drive system, comprising: The error calculation module is used to collect the operating status data of the dual three-phase permanent magnet synchronous motor, calculate the speed error between the current speed and the target speed, and adjust the gain value of the sliding mode control, the approach law form of the sliding mode control, and the integral coefficient of the sliding surface according to the speed error. The compensation quantity generation module is used to estimate the parameter changes, unmodeled dynamic quantities and external disturbances in the multiphase motor drive system online through the state observer, and generate the uncertainty compensation quantity. The drive control module is used to generate control commands based on the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount, and output the control commands to the six-phase modular multilevel converter to drive the dual three-phase permanent magnet synchronous motor to run.

[0007] The beneficial effects of this invention are: Compared to existing technologies, this invention firstly achieves a strong control mechanism during large errors and a weak switching mechanism during small errors by adaptively adjusting the gain, reaching law form, and integral coefficient of the sliding mode control based on the speed error. This allows for rapid convergence under startup, fault initiation, or large disturbance conditions, and effectively suppresses chattering and current ripple during steady-state operation. Secondly, by estimating parameter changes, unmodeled dynamic quantities, and external disturbances online through a state observer, an uncertainty compensation quantity is generated and fed forward and superimposed into the control law. This enables the system to actively compensate for model mismatch and external disturbances, improving robustness under parameter changes, load disturbances, and fault conditions. Thirdly, by outputting control commands to a six-phase modular multilevel converter to drive dual three-phase permanent magnet synchronous motors, a complete adaptive robust control closed loop is formed. Finally, this invention employs a unified control framework in both normal and fault operating modes, avoiding control structure switching and control discontinuity issues during state transitions, achieving stable, efficient, and smooth control across the entire operating range. Attached Figure Description

[0008] Figure 1 A flowchart illustrating the adaptive robust control method for a multiphase motor drive system provided by this invention; Figure 2 This is a schematic diagram of the adaptive robust control system of the multiphase motor drive system provided by the present invention.

[0009] In the attached diagram, the components represented by each number are as follows: Error calculation module 11, compensation amount generation module 12, drive control module 13. Detailed Implementation

[0010] Example 1, as Figure 1 As shown, this embodiment of the invention provides an adaptive robust control method for a multiphase motor drive system, including: S10: Collect the operating status data of the dual three-phase permanent magnet synchronous motor, calculate the speed error between the current speed and the target speed, and adjust the gain value of the sliding mode control, the approach law form of the sliding mode control, and the integral coefficient of the sliding surface according to the speed error; The six-phase modular multilevel converter adopts a six-phase topology, with each phase including an upper arm and a lower arm. Each arm is composed of multiple half-bridge sub-modules connected in series. Each half-bridge sub-module includes two insulated-gate bipolar transistors, two anti-parallel diodes, and a DC capacitor. The stator of the dual three-phase permanent magnet synchronous motor consists of two sets of three-phase windings. The two sets of three-phase windings are spatially separated by 30 electrical degrees. The neutral points of the two sets of three-phase windings are isolated from each other and are not grounded.

[0011] The multiphase motor drive system includes a six-phase modular multilevel converter and a dual three-phase permanent magnet synchronous motor. The six-phase modular multilevel converter is a converter topology suitable for medium- and high-voltage, high-power power conversion. It achieves high voltage output and low harmonic characteristics by cascading multiple half-bridge submodules, offering advantages such as high modularity, low switching losses, high output voltage levels, and easy expansion. This six-phase modular multilevel converter employs a six-phase topology, with each phase consisting of an upper and lower bridge arm. Each bridge arm is composed of multiple half-bridge submodules connected in series. Each half-bridge submodule includes two insulated-gate bipolar transistors, two anti-parallel diodes, and a DC capacitor. The stator of the dual three-phase permanent magnet synchronous motor consists of two sets of three-phase windings, spatially separated by 30 electrical degrees. The neutral points of the two sets of three-phase windings are isolated and ungrounded.

[0012] First, the operating status data of the dual three-phase permanent magnet synchronous motor is collected. The dual three-phase permanent magnet synchronous motor is a type of multi-phase permanent magnet synchronous motor. Its stator employs two sets of three-phase windings spatially separated by 30 electrical degrees. Compared to traditional three-phase motors, it has higher power density and stronger fault tolerance, making it particularly suitable for scenarios such as offshore wind power where continuous operation is required and maintenance is difficult. This operating status data includes the motor's three-phase current, three-phase voltage, rotor position angle, and speed information, characterizing the motor's current actual operating condition and providing real-time feedback for subsequent control. Based on this operating status data, the speed error between the current speed and the target speed can be calculated.

[0013] Specifically, the current rotational speed refers to the actual mechanical angular velocity of the motor rotor, acquired in real time by a position sensor or speed observer. The target rotational speed refers to the desired rotational speed value set according to the host computer command or system operating requirements. This target rotational speed is set based on power generation demand or operating condition requirements. For example, in offshore wind power applications, the target rotational speed can be dynamically set based on wind speed changes and maximum power point tracking (MPPT) strategies. The rotational speed error between the current and target rotational speeds is calculated. This error is the difference between the target and current rotational speeds and is used to characterize the degree of deviation between the actual and desired rotational speeds of the motor.

[0014] Due to uncertainties such as parameter variations, load disturbances, and winding faults during operation, dual three-phase permanent magnet synchronous motors exhibit complex and variable dynamic characteristics in their speed error. The magnitude and trend of this speed error directly reflect the current control requirements of the system. Therefore, based on this speed error, the gain value of the sliding mode control, the reaching law form of the sliding mode control, and the integral coefficient of the sliding surface can be adjusted.

[0015] Sliding mode control is a nonlinear robust control method that designs a sliding surface and forces the system state to move along the sliding surface, making the system invariant to parameter perturbations and external disturbances. Its gain value determines the switching intensity of the sliding mode control, directly affecting the system's robustness and chattering level; its reaching law shape determines the trajectory characteristics of the system state approaching the sliding surface, affecting the system's dynamic response speed; the sliding surface is a hyperplane defined in the state space, where the system state remains in motion after being driven to this plane; the integral coefficients of the sliding surface are used to introduce integral terms to eliminate steady-state errors, and their values ​​affect the system's steady-state accuracy and resistance to integral saturation.

[0016] Specifically, traditional sliding mode control employs fixed gain and a fixed reaching law, making it difficult to simultaneously meet the requirements of rapid response under large error conditions and stable operation under small error conditions. This invention, by adjusting the gain, reaching law form, and integral coefficient of the sliding mode control in real time based on the speed error, achieves adaptive adjustment: strong control during large error stages to accelerate convergence, and weak control during small error stages to suppress chattering. This ensures robustness while improving the system's dynamic performance and steady-state quality.

[0017] Specifically, firstly, the gain value of the sliding mode control is adjusted according to the aforementioned speed error, including: The gain value is set with a base component and an adaptive component. The base component is a fixed constant, and the adaptive component is set using a linear function relationship. The linear function relationship is that the adaptive component is equal to a first adaptive coefficient multiplied by the absolute value of the speed error. The first adaptive coefficient is a preset positive number used to control the sensitivity of the adaptive component to changes in the speed error. The gain value is obtained by adding the base component and the adaptive component, wherein when the absolute value of the speed error increases, the adaptive component increases accordingly, thereby increasing the gain value; when the absolute value of the speed error decreases, the adaptive component decreases accordingly, thereby decreasing the gain value.

[0018] First, set the base component and adaptive component of the gain value. The gain value is a key parameter in sliding mode control used to determine the switching control strength, and it consists of two components: the base component and the adaptive component.

[0019] The fundamental component is the fixed part of the gain value. This fundamental component is a constant and represents the minimum control strength that sliding mode control must maintain under any operating condition. This fundamental component is set according to the system's rated operating parameters and the desired minimum robustness requirements. For example, it can be determined through simulation or experimental calibration based on parameters such as the rated current and rated torque of the dual three-phase permanent magnet synchronous motor and the DC bus voltage of the six-phase modular multilevel converter, to ensure that the system remains stable under the worst operating conditions.

[0020] The adaptive component is set using a linear function relationship, whereby the adaptive component equals the first adaptive coefficient multiplied by the absolute value of the speed error. The first adaptive coefficient is a preset positive number used to control the sensitivity of the adaptive component to changes in speed error. This first adaptive coefficient is set based on the system's sensitivity requirements to error response and the desired gain adjustment range. For example, it can be determined through theoretical calculations or engineering tuning based on the maximum allowable overshoot during system startup and the maximum allowable chattering amplitude during steady-state operation. This ensures that the gain value, as it changes with speed error, meets the requirements for rapid dynamic response without causing system oscillations due to excessive gain changes.

[0021] Specifically, when the absolute value of the speed error changes, the adaptive component changes in the same direction according to this linear function relationship. That is, the larger the absolute value of the speed error, the larger the value of the adaptive component; the smaller the absolute value of the speed error, the smaller the value of the adaptive component.

[0022] Secondly, the base component and the adaptive component are added together to obtain the gain value. Since the adaptive component changes in the same direction as the absolute value of the speed error, when the absolute value of the speed error increases, the adaptive component increases accordingly, thus increasing the gain value; when the absolute value of the speed error decreases, the adaptive component decreases accordingly, thus decreasing the gain value.

[0023] In this way, the gain value of sliding mode control can be adaptively adjusted according to the magnitude of the speed error. In the stage where the speed error is large, a larger gain value can provide stronger control, enabling the system state to quickly approach the sliding surface, thereby shortening the short-term response time. In the steady-state stage where the speed error is small, a smaller gain value can reduce the switching intensity of sliding mode control, suppress high-frequency chattering from the source, reduce current ripple, and improve the steady-state operation quality of the system.

[0024] Furthermore, adjusting the approach law form of the sliding mode control based on the aforementioned speed error includes: The distance of the system state relative to the sliding surface is divided into three stages: the distance away stage, the proximity stage, and the neighborhood stage. During the distance phase, the exponential approach law is adopted as the dominant approach law, causing the system state to approach the sliding surface at an exponential rate. During the approach phase, a weighted fusion of exponential and power-law approach laws is adopted. As the system state approaches the sliding surface, the weight of the power-law approach law is gradually increased, while the weight of the exponential approach law is gradually decreased. In the neighborhood stage, a power-law approach is used as the dominant approach law, so that the system state enters the sliding surface in a smooth manner. Among them, the weighted fusion method of exponential and power-law approach is adopted, including: A fusion coefficient is set, wherein the fusion coefficient changes continuously between zero and one. When the system state is near the beginning of the approach phase, the fusion coefficient is zero, and when the system state is near the end of the approach phase, the fusion coefficient is one. The weighted component of the exponential reaching law is obtained by multiplying the exponential reaching law by one and subtracting the fusion coefficient. Calculate the power-approach law by multiplying it by the fusion coefficient to obtain the weighted components of the power-approach law; The weighted components of the exponential approach law and the power approach law are added together to obtain the fused approach law.

[0025] First, the distance of the system state relative to the sliding surface is divided into three stages: the distance phase, the proximity phase, and the neighborhood phase. The system state refers to the two-dimensional state variable composed of the speed error and its derivative. In sliding mode control, the speed error is typically chosen as the state variable to describe the deviation of the system's current operating point from the desired operating point. The sliding surface is a hyperplane defined in the state space. This hyperplane is determined by the integral coefficients of the sliding surface. When the system state is driven onto this hyperplane, the system's dynamic behavior is determined by the parameters of the sliding surface and is no longer affected by parameter changes or external disturbances.

[0026] The distance between the system state and the sliding surface is the absolute value of the sliding function value, representing the degree to which the current system operating point deviates from the sliding surface. This distance can be characterized by the magnitude of the sliding function value. The larger the absolute value of the sliding function value, the farther the system state is from the sliding surface, and the smaller the absolute value of the sliding function value, the closer the system state is to the sliding surface.

[0027] Specifically, the distance-away stage, the approach stage, and the neighborhood stage represent three different intervals in which the system state is further away from the sliding surface, from far to near. The distance-away stage refers to the stage where the system state is far from the sliding surface and the absolute value of the sliding function is large; this stage corresponds to the start-up process with large speed errors or the initial recovery stage after a large disturbance. The approach stage refers to the stage where the system state is at a moderate distance from the sliding surface and the absolute value of the sliding function is in the middle range; this stage corresponds to the transition of the system state from far away from the sliding surface to closer to it. The neighborhood stage refers to the stage where the system state is very close to the sliding surface and the absolute value of the sliding function is small; this stage corresponds to the steady-state operation stage where the system state enters a small region near the sliding surface.

[0028] Optionally, the division into three stages is quantified based on a preset threshold for the absolute value of the sliding mode function. Specifically, a first threshold and a second threshold are set, where the first threshold is greater than the second threshold. When the absolute value of the sliding mode function is greater than the first threshold, the system is determined to be in the far-away stage; when the absolute value of the sliding mode function is less than or equal to the first threshold and greater than the second threshold, the system is determined to be in the approach stage; when the absolute value of the sliding mode function is less than or equal to the second threshold, the system is determined to be in the neighborhood stage. The values ​​of the first and second thresholds are set according to the system's requirements for dynamic response speed and steady-state accuracy. For example, the first threshold can be set to 30% to 50% of the rated speed error, and the second threshold can be set to 5% to 10% of the rated speed error, so that the three stages correspond to different control requirements and achieve adaptive adjustment of the approach law form.

[0029] Specifically, during the distance phase, an exponential reaching law is used as the dominant reaching law, causing the system state to approach the sliding surface at an exponential rate. This exponential reaching law is the reaching law form in sliding mode control, and its mathematical expression is that the derivative of the sliding function value is equal to the negative exponential reaching law coefficient multiplied by the sliding function value, where the exponential reaching law coefficient is a positive number. This form determines that the speed at which the system state approaches the sliding surface is directly proportional to the magnitude of the current sliding function value; that is, the farther away from the sliding surface, the faster the approach speed, exhibiting the characteristic of fast approach speed. During the phase where the system state is far from the sliding surface, the exponential reaching law enables the system state to converge rapidly towards the sliding surface, shortening the short-term settling time and meeting the requirement for rapid response of the system during the startup phase or after a large disturbance.

[0030] During the approach phase, a weighted fusion of exponential and power-law approach laws is employed. As the system state approaches the sliding surface, the weight of the power-law approach law is gradually increased, while the weight of the exponential approach law is gradually decreased. The power-law approach law, a form of approach law in sliding mode control, is mathematically expressed as the derivative of the sliding function value equal to the negative power-law coefficient multiplied by a power of the sliding function value, where the power takes values ​​between 0 and 1. This form determines that the speed at which the system state approaches the sliding surface gradually decreases as the distance decreases, exhibiting a smooth approach process. Gradually introducing the power-law approach law during the approach phase allows the system state to approach the sliding surface more gently, preparing for a smooth entry into the sliding surface later. Meanwhile, as the system state approaches the sliding surface, gradually increasing the weight of the power-law approaching law and gradually decreasing the weight of the exponential-law approaching law can gradually transition the dominant role of the approaching process from rapid convergence to smooth approaching. This avoids the system state from crossing the sliding surface and causing oscillations due to the strong switching effect of the exponential-law approaching law when approaching the sliding surface, thus enabling the system state to enter the neighborhood of the sliding surface in a smooth manner.

[0031] Specifically, the weighted fusion method using the exponential and power-law approach includes the following steps: First, a fusion coefficient is set. The fusion coefficient is a proportional factor used to weight the exponential and power-law approach laws. It represents the weight proportion of the power-law approach law in the weighted fusion, while the weight proportion of the exponential approach law is 1 minus the fusion coefficient. This fusion coefficient continuously varies between 0 and 1. When the system state is at the beginning of the approach phase, the fusion coefficient is 0, at which point the weight of the exponential approach law is 1 and the weight of the power-law approach law is 0. When the system state is at the end of the approach phase, the fusion coefficient is 1, at which point the weight of the exponential approach law is 0 and the weight of the power-law approach law is 1.

[0032] Optionally, the change in the fusion coefficient can be continuously determined based on the distance of the system state relative to the sliding surface or the magnitude of the sliding function value. For example, the fusion coefficient can be set as a linear function of the absolute value of the sliding function, that is, the fusion coefficient is equal to the absolute value of the sliding function minus the second threshold, divided by the difference between the first threshold and the second threshold, where the first threshold is the absolute value of the sliding function corresponding to the start of the approach phase, and the second threshold is the absolute value of the sliding function corresponding to the end of the approach phase. This linear function makes the fusion coefficient increase uniformly from 0 to 1 during the approach phase.

[0033] Secondly, the weighted component of the exponential reaching law is obtained by multiplying the exponential reaching law by 1 and subtracting the fusion coefficient. This weighted component represents the proportion of the exponential reaching law in the fused reaching law. As the fusion coefficient increases from 0 to 1, this weighted component gradually decreases from the exponential reaching law itself to 0.

[0034] Simultaneously, the power-law approaching law is multiplied by the fusion coefficient to obtain the weighted component of the power-law approaching law. This weighted component represents the proportion of the power-law approaching law in the fused approaching law. As the fusion coefficient increases from 0 to 1, this weighted component gradually increases from 0 to the power-law approaching law itself.

[0035] Furthermore, the weighted components of the exponential reaching law and the power reaching law are added together to obtain the fused reaching law. This fused reaching law is equivalent to the pure exponential reaching law at the beginning of the approach phase, equivalent to the pure power reaching law at the end of the approach phase, and a weighted combination of the two reaching laws in the middle region of the approach phase, thus achieving a gradual transition from rapid reaching to smooth reaching.

[0036] Furthermore, during the neighborhood phase, a power-law approach is employed as the dominant approach law, ensuring that the system state smoothly enters the sliding surface. Specifically, once the system state has entered the neighborhood of the sliding surface, the power-law approach effectively reduces the intensity of control switching, suppresses chattering, and ensures stable system operation near the sliding surface.

[0037] In summary, through the adaptive adjustment of the aforementioned approach law form, the approach process of sliding mode control can prioritize rapid convergence when far from the sliding surface and smooth entry when approaching the sliding surface, thus balancing the dynamic response speed and steady-state operation quality of the system.

[0038] Further, adjusting the integral coefficient of the sliding surface based on the aforementioned rotational speed error includes: Set a first preset threshold and a second preset threshold, wherein the first preset threshold is greater than the second preset threshold; When the absolute value of the speed error is greater than the first preset threshold, the integral coefficient is set to the first integral value, wherein the first integral value is a low integral value to avoid system overshoot and oscillation caused by integral saturation; When the absolute value of the rotational speed error is less than the second preset threshold, the integral coefficient is set to the second integral value, wherein the second integral value is a high integral value, which is used to enhance the ability to eliminate residual errors and improve the steady-state accuracy of the system. When the absolute value of the speed error is greater than or equal to the second preset threshold and less than or equal to the first preset threshold, the integral coefficient is set to an intermediate value that increases continuously as the absolute value of the speed error decreases, wherein the intermediate value is between the first integral value and the second integral value.

[0039] First, set a first preset threshold and a second preset threshold. The first preset threshold is greater than the second preset threshold. The first preset threshold defines the range with larger speed errors, while the second preset threshold defines the range with smaller speed errors. The specific values ​​of the first and second preset thresholds can be set according to the system's requirements for dynamic response speed and steady-state accuracy. For example, the first preset threshold can be set to 10% to 20% of the rated speed, and the second preset threshold can be set to 1% to 5% of the rated speed.

[0040] When the absolute value of the speed error exceeds the first preset threshold, it indicates that the system is currently in the startup phase or has experienced a significant disturbance, with the speed deviating far from the target value. At this point, the integral coefficient is set to the first integral value. The integral coefficient is the proportionality coefficient of the integral term in the sliding surface, used to determine the strength of the integral action in sliding mode control. The larger the integral coefficient, the more significant the contribution of the integral term to the control quantity, and the stronger the system's ability to eliminate steady-state errors. However, it also makes it easier to cause integral saturation. This first integral value is a low integral value, i.e., a small integral coefficient, used to avoid system overshoot and oscillation caused by integral saturation.

[0041] The first integral value is set based on the maximum allowable overshoot of the system during the startup phase and the output limit range of the actuator. For example, it can be determined through simulation experiments based on the rated current of the dual three-phase permanent magnet synchronous motor and the maximum output voltage of the six-phase modular multilevel converter to ensure that the system does not experience integral saturation during the large error phase. For example, it can be 10% to 30% of the optimal integral coefficient under rated operating conditions.

[0042] During the large error phase, if the integral coefficient is too large, the integral term will accumulate rapidly, causing the control quantity to exceed the actuator's capability range and resulting in integral saturation, which in turn can lead to system overshoot or even oscillation. Using a smaller integral coefficient can limit the accumulation rate of the integral term, allowing the system to remain stable as it rapidly approaches the target value.

[0043] Furthermore, when the absolute value of the speed error is less than the second preset threshold, it indicates that the system has entered the steady-state operation stage and the speed is close to the target value. At this time, the integral coefficient is set to the second integral value. The second integral value is a high integral value, that is, an integral coefficient with a larger value, used to enhance the ability to eliminate residual errors and improve the steady-state accuracy of the system. This second integral value is set according to the system's requirements for steady-state accuracy and the maximum allowable steady-state error. For example, it can be determined by theoretical calculation or experimental calibration based on the allowable speed fluctuation range of the dual three-phase permanent magnet synchronous motor under rated load and the system's tolerance for steady-state error. A maximum integral coefficient that can effectively eliminate residual errors without causing oscillations can be determined, such as 100% to 150% of the optimal integral coefficient under rated operating conditions.

[0044] In the small error stage, a larger integral coefficient can generate sufficient integral action on the small steady-state error, pushing the system state to approach the sliding surface further, thereby eliminating residual deviations and achieving high-precision steady-state tracking.

[0045] Furthermore, when the absolute value of the speed error is greater than or equal to the second preset threshold and less than or equal to the first preset threshold, it indicates that the system is in an intermediate stage transitioning from a large error to a small error. At this point, the integral coefficient is set to an intermediate value that continuously increases as the absolute value of the speed error decreases. This intermediate value changes continuously between the first and second integral values. For example, it can be set as a linear function of the absolute value of the speed error, i.e., the integral coefficient equals the first integral value plus the difference between the second and first integral values ​​multiplied by 1 minus the ratio of the absolute value of the speed error to the first preset threshold, allowing the integral coefficient to increase gradually as the error decreases. This continuous transition setting avoids jumps in the integral coefficient at threshold boundaries, thus ensuring the smoothness of the control quantity changes and enabling the system to maintain stable adjustment behavior throughout the transition from transient to steady state.

[0046] By adopting the strategy of adaptively adjusting the integral coefficient according to the magnitude of the rotational speed error, the integral coefficient of the sliding surface can use a low integral value when the rotational speed error is large to avoid overshoot and oscillation caused by integral saturation, use a high integral value when the rotational speed error is small to eliminate residual error and improve steady-state accuracy, and continuously change during the transition phase to achieve smooth adjustment, thereby taking into account both the dynamic response performance and steady-state tracking accuracy of the system.

[0047] S20: Estimate the parameter changes, unmodeled dynamic quantities and external disturbances in the multiphase motor drive system online through the state observer, and generate uncertainty compensation quantities; Secondly, the parameter variations, unmodeled dynamic quantities, and external disturbances in the multiphase motor drive system are estimated online using a state observer. A state observer is a dynamic algorithm that reconstructs the internal state of a system or unmeasurable disturbances based on system input-output information. It can utilize the system's mathematical model and real-time measurement data to estimate physical quantities that cannot be directly measured, thereby providing a basis for control decisions.

[0048] In a multiphase motor drive system, parameter variations refer to the deviations between the actual and nominal values ​​of electrical parameters such as motor stator resistance, stator inductance, and permanent magnet flux linkage due to factors such as temperature changes, magnetic circuit saturation, or aging. Unmodeled dynamic quantities refer to system characteristics that are not included in the mathematical model due to model simplification or neglect of higher-order dynamics, such as core losses, hysteresis effects, inverter nonlinear characteristics, and high-frequency parasitic parameters. External disturbances refer to unknown external inputs acting on the system, including sudden changes in load torque, wind and wave disturbances, and grid voltage fluctuations.

[0049] By estimating the aforementioned parameter changes, unmodeled dynamic quantities, and external disturbances online, these three types of uncertainty can be uniformly treated as a total disturbance for real-time observation, thereby obtaining the degree to which the system deviates from the ideal model during actual operation. This estimation result reflects the comprehensive uncertainty impact on the system at the current moment, providing accurate numerical basis for subsequent proactive compensation. This enables control strategies to specifically offset the negative impact of uncertainty on system performance, enhancing the system's robustness under parameter perturbations, model mismatches, and external disturbances.

[0050] Specifically, the parameter variations, unmodeled dynamic quantities, and external disturbances in the multiphase motor drive system are estimated online using a state observer to generate uncertainty compensation quantities, including: An extended state observer is constructed, wherein the order of the extended state observer is determined according to the system order of the dual three-phase permanent magnet synchronous motor, and the extended state observer expands the parameter changes, unmodeled dynamic quantities and external disturbances into new state variables, which are uniformly observed as the total uncertainty. The collected current, voltage, and speed information of the dual three-phase permanent magnet synchronous motor are sent as input signals to the extended state observer. The extended state observer calculates and outputs an estimate of the total uncertainty in real time based on the input signal and its internal state equation, wherein the internal state equation describes the dynamic relationship between the state variables of the extended state observer and the input signal. A compensation coefficient is set, and the estimated value of the total uncertainty is multiplied by the compensation coefficient to obtain the uncertainty compensation amount. The uncertainty compensation amount is then fed forward and superimposed into the calculation result of the sliding mode control law, so that the uncertainty compensation amount actively offsets the uncertainty influence in the multiphase motor drive system.

[0051] First, an extended state observer is constructed. An extended state observer is an observer structure capable of simultaneously estimating the internal state of a system and external disturbances. Its order is determined based on the system order of the dual three-phase permanent magnet synchronous motor. Optionally, the speed loop of a dual three-phase permanent magnet synchronous motor can typically be simplified to a first-order system; therefore, the extended state observer can be designed as a second-order system, with one order used to estimate the speed state and the other order used to estimate the total uncertainty. The extended state observer expands parameter variations, unmodeled dynamic quantities, and external disturbances into new state variables, uniformly observing them as the total uncertainty. This combines the originally dispersed multiple types of uncertainties into a single lumped variable, simplifying the complexity of the estimation problem.

[0052] Secondly, the acquired current, voltage, and speed information of the dual three-phase permanent magnet synchronous motors are sent as input signals to the extended state observer. Optionally, the current information includes torque current and excitation current components, the voltage information includes stator voltage components, and the speed information includes the current speed value. These input signals reflect the current actual operating state of the system, providing the extended state observer with the basic data for reconstructing the system state and estimating uncertainties.

[0053] The extended state observer calculates and outputs an estimate of the total uncertainty in real time based on the input signal and its internal state equation. The internal state equation is a differential equation describing the time-varying nature of the observer's internal state variables, illustrating the dynamic relationship between the observer's state variables and the input signal. Preferably, this internal state equation is constructed based on a mathematical model of a dual three-phase permanent magnet synchronous motor, and the convergence of the estimation error is ensured by introducing an observer gain matrix. The extended state observer dynamically adjusts the estimated values ​​of the state variables by comparing the deviation between the actual output and the output obtained based on the state estimation, gradually approximating the true state and simultaneously bringing the extended state closer to the true value of the total uncertainty.

[0054] Specifically, the extended state observer compares the actual measured current, voltage, and speed information with the output values ​​calculated by the observer based on the current state estimate. The difference between the two is multiplied by the observer gain matrix and fed back into the differential equation of the state estimate, continuously correcting the state estimate until the estimation error converges to near zero. After the convergence process, the extended state observer can output an accurate estimate of the total uncertainty.

[0055] Next, a compensation coefficient is set. This compensation coefficient is a proportional factor used to adjust the intensity of uncertainty compensation, representing the proportion of the estimated total uncertainty in the compensation process, i.e., the strength of the compensation effect. The compensation coefficient is set according to the system's sensitivity to uncertainty and the stability requirements of the compensation response. For example, it can be set to 1, indicating complete compensation for the estimated total uncertainty; or, based on the system's response characteristics under disturbance conditions, a value less than 1 can be determined through simulation or experimental calibration to avoid overcompensation causing system oscillations, such as a value of 0.8. The estimated total uncertainty is multiplied by the compensation coefficient to obtain the uncertainty compensation amount, and this uncertainty compensation amount is fed forward and superimposed into the calculation result of the sliding mode control law.

[0056] Specifically, the compensation coefficient is used to adjust the strength of the compensation effect. Through feedforward superposition, the uncertainty compensation amount actively intervenes in the control command generation stage. This feedforward superposition method can generate the corresponding compensation control amount based on the total uncertainty estimated in real time by the extended state observer, before the disturbance has a significant impact on the system output. This can offset the uncertainty in the multiphase motor drive system in advance, reduce the interference of parameter changes, unmodeled dynamics and external disturbances on the system control performance, and improve the robustness and anti-disturbance capability of the system.

[0057] In summary, the final uncertainty compensation amount is a correction amount obtained by multiplying the estimated value of the total uncertainty by the compensation coefficient. It represents the comprehensive compensation intensity for the parameter changes, unmodeled dynamics and external disturbances currently existing in the multiphase motor drive system. It is used to actively offset the impact of uncertainty factors on system performance during the control command generation stage, so that the sliding mode control law can obtain additional correction effect on the equivalent model and improve the system's ability to suppress model mismatch and external disturbances.

[0058] S30: Generate control commands based on the gain value, approach law form, integral coefficient, and uncertainty compensation amount, and output the control commands to the six-phase modular multilevel converter to drive the dual three-phase permanent magnet synchronous motor to run.

[0059] Furthermore, control commands are generated based on the aforementioned determined gain value, approach law form, integral coefficient, and obtained uncertainty compensation. These control commands are the final output quantities obtained after calculation using a sliding mode control law and superimposed uncertainty compensation under the current operating state. They include torque current command values ​​and excitation current command values. The torque current command value is used to adjust the motor's electromagnetic torque to control the speed, while the excitation current command value is used to adjust the motor's magnetic field to optimize operating efficiency. This enables precise control of the current and torque of the dual three-phase permanent magnet synchronous motor.

[0060] The control command is output to the six-phase modular multilevel converter. The six-phase modular multilevel converter adjusts the switching state of the half-bridge sub-modules in each phase arm according to the received control command, and generates the corresponding six-phase voltage to drive the dual three-phase permanent magnet synchronous motor to run. This makes the actual speed and current of the motor follow the command value, thereby realizing closed-loop speed control and active compensation for uncertainties.

[0061] Specifically, control commands are generated based on the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount, including: A working condition identification unit is set up, wherein the working condition identification unit identifies the current operating stage of the system as one of the following based on the collected speed error, current error, error change rate and fault diagnosis signal of the dual three-phase permanent magnet synchronous motor: startup stage, steady-state operation stage, fault recovery stage or parameter change stage. A weight allocation unit is set up, wherein the weight allocation unit is used to allocate corresponding weight values ​​to the gain value, the approach law form, the integral coefficient and the uncertainty compensation amount according to the operating stage identified by the operating condition identification unit. The weight values ​​are used to represent the degree of influence of each parameter in the final control command. The working condition identification unit and the weight allocation unit are integrated to form a coordination controller, wherein the weight allocation unit is connected to the working condition identification unit and is used to receive the operating stage identified by the working condition identification unit; The coordination controller evaluates the current operating stage of the system in real time, and determines the weights of the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount in the final control command based on the operating stage. Based on the weights and preset priority order, the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount are fused and calculated to output the final control command, wherein the control command includes the torque current command value.

[0062] First, a working condition identification unit is set up. This unit identifies the current operating stage of the system based on the collected speed error, current error, error change rate, and fault diagnosis signals of the dual three-phase permanent magnet synchronous motor. Specifically, the speed error reflects the deviation between the current speed and the target speed; the current error reflects the difference between the actual current and the commanded current; the error change rate reflects the trend of error change over time; and the fault diagnosis signals reflect fault states such as open circuits in windings and abnormal power devices. By integrating the above information, the working condition identification unit can accurately determine the current operating state of the system, providing a basis for the subsequent weight allocation of control parameters.

[0063] Optionally, the operating condition identification unit quantifies and identifies the operating phase using preset judgment thresholds and rules. When the fault diagnosis signal indicates an open circuit in the winding or a power device fault, the operating condition identification unit identifies the current operating phase as a fault recovery phase. When the fault diagnosis signal is normal and the absolute value of the speed error is greater than the start-up threshold, the operating condition identification unit identifies the current operating phase as a start-up phase, which can be set to 10% to 20% of the rated speed. When the fault diagnosis signal is normal, the absolute value of the speed error is less than the steady-state threshold, the absolute value of the current error is less than the steady-state current threshold, and the error change rate remains within a preset stable range for more than a preset stabilization time, the operating condition identification unit identifies the current operating phase as a steady-state operating phase, where the steady-state threshold can be set to 1% to 5% of the rated speed, the steady-state current threshold can be set to 2% to 5% of the rated current, and the stabilization time can be set to 0.5 to 2 seconds. When the fault diagnosis signal is normal and the absolute value of the speed error is between the start-up threshold and the steady-state threshold, or when the rate of change of the current error exceeds the preset rate of change threshold, the operating condition identification unit identifies the current operating stage as a parameter change stage. This rate of change threshold can be set according to the maximum normal rate of change of the system under rated operating conditions, for example, it can be set to 120% to 150% of the rated current rate of change. Through the above quantitative judgment rules, the operating condition identification unit can identify the current operating stage of the system in real time and accurately, providing a reliable decision basis for the weight allocation unit.

[0064] Secondly, a weight allocation unit is set up. This unit assigns corresponding weight values ​​to the gain value, reaching law form, integral coefficient, and uncertainty compensation amount based on the operating stage identified by the operating condition identification unit. The weight value represents the degree of influence of each parameter in the final control command; a larger weight value indicates a greater contribution of that parameter in the control command generation process. Different operating stages have different requirements for control emphasis. For example, the startup stage emphasizes rapid response, the steady-state stage emphasizes accuracy maintenance, the fault recovery stage emphasizes robustness, and the parameter change stage emphasizes adaptive compensation. Therefore, weight allocation is needed to achieve targeted adjustments to the control strategy.

[0065] Specifically, the weights of the gain value, the approach law form, the integral coefficient, and the uncertainty compensation in the final control command are determined according to the operational phase, including: When the operation phase is the startup phase, the weights of the gain value and the approach law form are set to be higher than the weights of other parameters, the weight of the integral coefficient is set to be lower than the weights of other parameters, and the weight of the uncertainty compensation is set to be between the weight of the gain value and the weight of the integral coefficient, so that the system mainly uses strong control and supplements integral action in the startup phase. When the operating phase is a steady-state operating phase, the weight of the integral coefficient is set to be higher than the weight of other parameters, the weight of the approach law form is set to be lower than the weight of other parameters, and the weight of the gain value and the weight of the uncertainty compensation amount are set to be between the weight of the integral coefficient and the weight of the approach law form, so that the system mainly uses integral regulation and weakens the switching effect in the steady-state operating phase. When the operation phase is the fault recovery phase, the weights of the gain value, the approach law form, and the uncertainty compensation amount are all set to be higher than the weights of the integral coefficient, and the weight of the uncertainty compensation amount is not lower than the weights of the gain value and the approach law form. The weight of the integral coefficient is set to be lower than the weights of other parameters, so that the system can quickly restore stability by focusing on strong control and uncertainty compensation during the fault recovery phase. When the operation phase is the parameter change phase, the weight of the uncertainty compensation amount is set to be higher than the weight of other parameters, and the weights of the gain value, the approach law form, and the integral coefficient are set to be equal and lower than the weight of the uncertainty compensation amount, so that the system mainly uses uncertainty compensation and the various control parameters play a balanced role in the parameter change phase.

[0066] First, during the startup phase, the weights of the gain value and the reaching law morphology are set higher than those of other parameters, while the weight of the integral coefficient is set lower than that of other parameters. The weight of the uncertainty compensation is set between the weights of the gain value and the integral coefficient. Optionally, for example, during startup, the weights of the gain value and the reaching law morphology are both set to 40%, both higher than the weights of other parameters, allowing the system to drive rapid convergence with strong control. The weight of the integral coefficient is set to 0%, lower than the weights of other parameters, to avoid overshoot and oscillations caused by integral saturation. The weight of the uncertainty compensation is set to 20%, between the weights of the gain value and the integral coefficient, providing some compensation capability without affecting the dominant role of rapid response. Through this weight allocation, the system achieves rapid and stable startup performance during startup, primarily relying on strong control and secondarily on integral action.

[0067] Secondly, during the steady-state operation phase, the weight of the integral coefficient is set higher than that of other parameters, the weight of the reaching law morphology is set lower than that of other parameters, and the weights of the gain value and the uncertainty compensation amount are set between the weights of the integral coefficient and the reaching law morphology. Optionally, for example, during the steady-state operation phase, the weight of the integral coefficient is set to 60%, higher than the weights of other parameters, to enhance the ability to eliminate residual errors and improve steady-state accuracy; the weight of the reaching law morphology is set to 0%, lower than the weights of other parameters, to weaken the switching effect and reduce chattering; the weight of the gain value is set to 20%, and the weight of the uncertainty compensation amount is set to 20%, both between the weights of the integral coefficient and the reaching law morphology, maintaining a certain robustness to cope with small disturbances without disrupting steady-state stability due to excessive control. Through the above weight allocation, the system primarily uses integral regulation and weakens the switching effect during the steady-state operation phase, achieving a balance between high accuracy and stable operation.

[0068] Simultaneously, during the fault recovery phase, the weights of the gain value, the approach law form, and the uncertainty compensation are all set higher than the integral coefficient, and the weight of the uncertainty compensation is no lower than the weights of the gain value and the approach law form. The weight of the integral coefficient is set lower than the weights of other parameters. Optionally, for example, during the fault recovery phase, the weights of the gain value, approach law form, and uncertainty compensation are set to 30%, all higher than the integral coefficient, and the uncertainty compensation is no lower than the weights of the gain value and the approach law form, enabling the system to quickly restore stability through a combination of strong control and active compensation. The weight of the integral coefficient is set to 0%, lower than the weights of other parameters, to avoid overshoot caused by excessively rapid accumulation of the integral term. Through the above weight allocation, the system primarily relies on strong control and uncertainty compensation during the fault recovery phase to achieve rapid stability restoration.

[0069] Furthermore, when the operation phase involves parameter changes, the weight of the uncertainty compensation is set higher than that of other parameters, while the weights of the gain value, the approach law form, and the integral coefficient are set equal and lower than the weight of the uncertainty compensation. Optionally, for example, during the parameter change phase, the weight of the uncertainty compensation is set to 70%, higher than the weights of other parameters, to fully utilize the state observer's ability to estimate and compensate for parameter changes and proactively adapt to model changes; the weights of the gain value, approach law form, and integral coefficient are set to 10%, equal and lower than the weight of the uncertainty compensation, ensuring that sliding mode control maintains basic robustness and adjustment capability while avoiding excessive weighting of a single parameter that could lead to extreme control behavior. Through this weight allocation, the system primarily relies on uncertainty compensation during the parameter change phase, with all control parameters acting in a balanced manner, achieving smooth adaptation of the control system to parameter changes.

[0070] Furthermore, the operating condition identification unit and the weight allocation unit are integrated to form a coordination controller. The weight allocation unit is connected to the operating condition identification unit and is used to receive the operating stage identified by the operating condition identification unit. As the upper-level decision-making body, the coordination controller is responsible for coordinating the intensity of the effects of various control parameters according to the current operating state, so that the control system can adaptively adjust the control strategy to adapt to different operating conditions.

[0071] By coordinating the controller to evaluate the current operating stage of the system in real time, the weights of the gain value, reaching law form, integral coefficient, and uncertainty compensation in the final control command are determined according to the operating stage. This process realizes a two-layer adjustment mechanism for control parameters, from bottom-level adaptive adjustment to top-level weight coordination, enabling the control system to have both the ability to adjust individual parameters according to error changes and the overall optimization capability of multiple parameters working together.

[0072] Finally, based on the weights and a preset priority order, the gain value, reaching law form, integral coefficient, and uncertainty compensation are fused and calculated to output the final control command. This control command includes the torque and current command values. During the fusion calculation, in addition to considering the weight allocation of each parameter, a priority order is introduced to handle potential control direction conflicts, ensuring that the final output control command maintains consistency and stability of control behavior while meeting multi-objective optimization requirements.

[0073] Specifically, based on the weights and a preset priority order, the gain value, the reaching law form, the integral coefficient, and the uncertainty compensation amount are fused and calculated to output the final control command, including: Based on the weights, the calculation results of the gain value, the calculation results of the reaching law form, the calculation results of the integral coefficient, and the calculation results of the uncertainty compensation amount are weighted and summed to obtain a preliminary control command, wherein the control command includes a torque current command value. Determine whether there is a directional conflict among the calculation results of the gain value, the calculation results of the reaching law form, the calculation results of the integral coefficient, and the calculation results of the uncertainty compensation amount, wherein the directional conflict refers to some calculation results being in the positive direction while other calculation results are in the negative direction; When there is a directional conflict, the dominant parameter is determined according to a preset priority order. The calculation result of the dominant parameter is used as a benchmark, and the calculation results of the others that are opposite to the dominant parameter are set to zero and then weighted and summed again to obtain the corrected control command. The priority order is as follows: the uncertainty compensation amount has the highest priority, the gain value has the second highest priority, the approach law form has the third highest priority, and the integral coefficient has the lowest priority.

[0074] First, based on the weights, the calculated results of the gain value, the reaching law form, the integral coefficient, and the uncertainty compensation are weighted and summed to obtain the preliminary control command. This control command includes the torque and current command value. Specifically, the weighted summation is performed by multiplying the calculated results of each parameter by their corresponding weight value and then summing the products to obtain the preliminary torque and current command value. This step achieves the coordinated integration of multiple control parameters under different operating conditions, enabling the final command to comprehensively reflect the regulating effects of each parameter.

[0075] Secondly, it's crucial to determine if there are directional conflicts among the calculated results of the gain value, the approach law form, the integral coefficient, and the uncertainty compensation. A directional conflict occurs when some calculated results are in a positive direction while others are in a negative direction. In motor control, a positive direction typically indicates an increase in torque current to increase speed, while a negative direction indicates a decrease in torque current to reduce speed. When the signs of the calculated results for different control parameters are inconsistent, it indicates a contradiction between different control objectives. For example, the gain value might require an increase in torque current for a rapid response, while the integral coefficient might require a decrease in torque current to suppress overshoot. Simply performing a weighted sum in this case could lead to control commands canceling each other out or producing inappropriate control effects.

[0076] Specifically, when directional conflicts exist, the dominant parameter is determined according to a preset priority order. The calculation result of the dominant parameter is used as the benchmark, and the calculation results of the other parameters with opposite directions are set to 0. Then, a weighted sum is performed again to obtain the corrected control command. The priority order is as follows: uncertainty compensation quantity has the highest priority, followed by gain value, then the approach law form, and the integral coefficient has the lowest priority.

[0077] Specifically, the priority order is set based on the following: the uncertainty compensation amount comes from the real-time estimation of the state observer, directly reflecting the comprehensive uncertainty impact currently experienced by the system, and has the highest reliability and timeliness, therefore it has the highest priority; the gain value determines the switching intensity of sliding mode control and plays a dominant role in rapid response, so it has the next highest priority; the approach law morphology affects the path of the system state towards the sliding surface, so it has the next lowest priority; the integral coefficient is mainly used to eliminate steady-state residual error and has the lowest priority in dynamic adjustment. Through the above priority ordering and conflict handling mechanism, the control system can prioritize the effective execution of the most important control objective when the directions of multi-parameter calculation results are inconsistent, avoid mutual contradictions in control commands, and ensure that the final output control commands have a clear and consistent control intent.

[0078] In summary, the embodiments of this application have at least the following technical effects: This invention firstly achieves an adaptive adjustment mechanism by adaptively adjusting the gain, reaching law form, and integral coefficient of the sliding mode control based on the rotational speed error. This mechanism provides strong control during large errors and weak switching during small errors, enabling rapid convergence under startup, fault initiation, or large disturbance conditions. It effectively suppresses chattering and current ripple in the steady-state phase, overcoming the difficulty of balancing response speed and chattering in traditional sliding mode control. Secondly, by estimating parameter changes, unmodeled dynamic quantities, and external disturbances online using a state observer, an uncertainty compensation quantity is generated and fed forward and superimposed into the sliding mode control law. This gives the system an active compensation capability for model mismatch and external disturbances, improving its robustness under parameter changes, load disturbances, and fault conditions. Thirdly, by fusing the gain, reaching law form, integral coefficient, and uncertainty compensation quantity to generate control commands, a unified control framework is constructed. This eliminates the need to switch control structures between normal and fault operating modes, avoiding control discontinuity and transient performance degradation.

[0079] Finally, the combined effect of the above mechanisms enables the dual three-phase permanent magnet synchronous motors, driven by a six-phase modular multilevel converter, to maintain rapid response, steady-state accuracy, and smooth operation under normal operation and various abnormal conditions, thereby improving the overall performance and engineering reliability of the multiphase motor drive system.

[0080] Example 2, as Figure 2As shown, based on the same inventive concept as the adaptive robust control method for a multiphase motor drive system provided in Embodiment 1, this embodiment of the invention also provides an adaptive robust control system for a multiphase motor drive system, including: Error calculation module 11 is used to collect the operating status data of dual three-phase permanent magnet synchronous motor, calculate the speed error between the current speed and the target speed, and adjust the gain value of sliding mode control, the approach law form of sliding mode control and the integral coefficient of sliding mode surface according to the speed error. The compensation quantity generation module 12 is used to estimate the parameter changes, unmodeled dynamic quantities and external disturbances in the multiphase motor drive system online through the state observer, and generate the uncertainty compensation quantity; The drive control module 13 is used to generate control commands based on the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount, and output the control commands to the six-phase modular multilevel converter to drive the dual three-phase permanent magnet synchronous motor to run.

[0081] Specifically, the error calculation module 11 is used for: The six-phase modular multilevel converter adopts a six-phase topology, with each phase including an upper arm and a lower arm. Each arm is composed of multiple half-bridge sub-modules connected in series. Each half-bridge sub-module includes two insulated-gate bipolar transistors, two anti-parallel diodes, and a DC capacitor. The stator of the dual three-phase permanent magnet synchronous motor consists of two sets of three-phase windings. The two sets of three-phase windings are spatially separated by 30 electrical degrees. The neutral points of the two sets of three-phase windings are isolated from each other and are not grounded.

[0082] Specifically, firstly, the gain value of the sliding mode control is adjusted according to the aforementioned speed error, including: The gain value is set with a base component and an adaptive component. The base component is a fixed constant, and the adaptive component is set using a linear function relationship. The linear function relationship is that the adaptive component is equal to a first adaptive coefficient multiplied by the absolute value of the speed error. The first adaptive coefficient is a preset positive number used to control the sensitivity of the adaptive component to changes in the speed error. The gain value is obtained by adding the base component and the adaptive component, wherein when the absolute value of the speed error increases, the adaptive component increases accordingly, thereby increasing the gain value; when the absolute value of the speed error decreases, the adaptive component decreases accordingly, thereby decreasing the gain value.

[0083] Furthermore, adjusting the approach law form of the sliding mode control based on the aforementioned speed error includes: The distance of the system state relative to the sliding surface is divided into three stages: the distance away stage, the proximity stage, and the neighborhood stage. During the distance phase, the exponential approach law is adopted as the dominant approach law, causing the system state to approach the sliding surface at an exponential rate. During the approach phase, a weighted fusion of exponential and power-law approach laws is adopted. As the system state approaches the sliding surface, the weight of the power-law approach law is gradually increased, while the weight of the exponential approach law is gradually decreased. In the neighborhood stage, a power-law approach is used as the dominant approach law, so that the system state enters the sliding surface in a smooth manner. Among them, the weighted fusion method of exponential and power-law approach is adopted, including: A fusion coefficient is set, wherein the fusion coefficient changes continuously between zero and one. When the system state is near the beginning of the approach phase, the fusion coefficient is zero, and when the system state is near the end of the approach phase, the fusion coefficient is one. The weighted component of the exponential reaching law is obtained by multiplying the exponential reaching law by one and subtracting the fusion coefficient. Calculate the power-approach law by multiplying it by the fusion coefficient to obtain the weighted components of the power-approach law; The weighted components of the exponential approach law and the power approach law are added together to obtain the fused approach law.

[0084] Further, adjusting the integral coefficient of the sliding surface based on the aforementioned rotational speed error includes: Set a first preset threshold and a second preset threshold, wherein the first preset threshold is greater than the second preset threshold; When the absolute value of the speed error is greater than the first preset threshold, the integral coefficient is set to the first integral value, wherein the first integral value is a low integral value to avoid system overshoot and oscillation caused by integral saturation; When the absolute value of the rotational speed error is less than the second preset threshold, the integral coefficient is set to the second integral value, wherein the second integral value is a high integral value, which is used to enhance the ability to eliminate residual errors and improve the steady-state accuracy of the system. When the absolute value of the speed error is greater than or equal to the second preset threshold and less than or equal to the first preset threshold, the integral coefficient is set to an intermediate value that increases continuously as the absolute value of the speed error decreases, wherein the intermediate value is between the first integral value and the second integral value.

[0085] The compensation amount generation module 12 is specifically used for: Specifically, the parameter variations, unmodeled dynamic quantities, and external disturbances in the multiphase motor drive system are estimated online using a state observer to generate uncertainty compensation quantities, including: An extended state observer is constructed, wherein the order of the extended state observer is determined according to the system order of the dual three-phase permanent magnet synchronous motor, and the extended state observer expands the parameter changes, unmodeled dynamic quantities and external disturbances into new state variables, which are uniformly observed as the total uncertainty. The collected current, voltage, and speed information of the dual three-phase permanent magnet synchronous motor are sent as input signals to the extended state observer. The extended state observer calculates and outputs an estimate of the total uncertainty in real time based on the input signal and its internal state equation, wherein the internal state equation describes the dynamic relationship between the state variables of the extended state observer and the input signal. A compensation coefficient is set, and the estimated value of the total uncertainty is multiplied by the compensation coefficient to obtain the uncertainty compensation amount. The uncertainty compensation amount is then fed forward and superimposed into the calculation result of the sliding mode control law, so that the uncertainty compensation amount actively offsets the uncertainty influence in the multiphase motor drive system.

[0086] Specifically, the drive control module 13 is used for: Specifically, control commands are generated based on the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount, including: A working condition identification unit is set up, wherein the working condition identification unit identifies the current operating stage of the system as one of the following based on the collected speed error, current error, error change rate and fault diagnosis signal of the dual three-phase permanent magnet synchronous motor: startup stage, steady-state operation stage, fault recovery stage or parameter change stage. A weight allocation unit is set up, wherein the weight allocation unit is used to allocate corresponding weight values ​​to the gain value, the approach law form, the integral coefficient and the uncertainty compensation amount according to the operating stage identified by the operating condition identification unit. The weight values ​​are used to represent the degree of influence of each parameter in the final control command. The working condition identification unit and the weight allocation unit are integrated to form a coordination controller, wherein the weight allocation unit is connected to the working condition identification unit and is used to receive the operating stage identified by the working condition identification unit; The coordination controller evaluates the current operating stage of the system in real time, and determines the weights of the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount in the final control command based on the operating stage. Based on the weights and preset priority order, the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount are fused and calculated to output the final control command, wherein the control command includes the torque current command value.

[0087] Specifically, the weights of the gain value, the approach law form, the integral coefficient, and the uncertainty compensation in the final control command are determined according to the operational phase, including: When the operation phase is the startup phase, the weights of the gain value and the approach law form are set to be higher than the weights of other parameters, the weight of the integral coefficient is set to be lower than the weights of other parameters, and the weight of the uncertainty compensation is set to be between the weight of the gain value and the weight of the integral coefficient, so that the system mainly uses strong control and supplements integral action in the startup phase. When the operating phase is a steady-state operating phase, the weight of the integral coefficient is set to be higher than the weight of other parameters, the weight of the approach law form is set to be lower than the weight of other parameters, and the weight of the gain value and the weight of the uncertainty compensation amount are set to be between the weight of the integral coefficient and the weight of the approach law form, so that the system mainly uses integral regulation and weakens the switching effect in the steady-state operating phase. When the operation phase is the fault recovery phase, the weights of the gain value, the approach law form, and the uncertainty compensation amount are all set to be higher than the weights of the integral coefficient, and the weight of the uncertainty compensation amount is not lower than the weights of the gain value and the approach law form. The weight of the integral coefficient is set to be lower than the weights of other parameters, so that the system can quickly restore stability by focusing on strong control and uncertainty compensation during the fault recovery phase. When the operation phase is the parameter change phase, the weight of the uncertainty compensation amount is set to be higher than the weight of other parameters, and the weights of the gain value, the approach law form, and the integral coefficient are set to be equal and lower than the weight of the uncertainty compensation amount, so that the system mainly uses uncertainty compensation and the various control parameters play a balanced role in the parameter change phase.

[0088] Specifically, based on the weights and a preset priority order, the gain value, the reaching law form, the integral coefficient, and the uncertainty compensation amount are fused and calculated to output the final control command, including: Based on the weights, the calculation results of the gain value, the calculation results of the reaching law form, the calculation results of the integral coefficient, and the calculation results of the uncertainty compensation amount are weighted and summed to obtain a preliminary control command, wherein the control command includes a torque current command value. Determine whether there is a directional conflict among the calculation results of the gain value, the calculation results of the reaching law form, the calculation results of the integral coefficient, and the calculation results of the uncertainty compensation amount, wherein the directional conflict refers to some calculation results being in the positive direction while other calculation results are in the negative direction; When there is a directional conflict, the dominant parameter is determined according to a preset priority order. The calculation result of the dominant parameter is used as a benchmark, and the calculation results of the others that are opposite to the dominant parameter are set to zero and then weighted and summed again to obtain the corrected control command. The priority order is as follows: the uncertainty compensation amount has the highest priority, the gain value has the second highest priority, the approach law form has the third highest priority, and the integral coefficient has the lowest priority.

Claims

1. An adaptive robust control method for a multiphase motor drive system, characterized in that, The multiphase motor drive system includes a six-phase modular multilevel converter and dual three-phase permanent magnet synchronous motors, and the method includes: The operating status data of the dual three-phase permanent magnet synchronous motor is collected, the speed error between the current speed and the target speed is calculated, and the gain value of the sliding mode control, the approach law form of the sliding mode control, and the integral coefficient of the sliding surface are adjusted according to the speed error. The parameter changes, unmodeled dynamics, and external disturbances in the multiphase motor drive system are estimated online using a state observer to generate uncertainty compensation quantities. Based on the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount, a control command is generated and output to the six-phase modular multilevel converter to drive the dual three-phase permanent magnet synchronous motor.

2. The adaptive robust control method for a multiphase motor drive system according to claim 1, characterized in that, The six-phase modular multilevel converter adopts a six-phase topology, with each phase including an upper arm and a lower arm. Each arm is composed of multiple half-bridge sub-modules connected in series. Each half-bridge sub-module includes two insulated-gate bipolar transistors, two anti-parallel diodes, and a DC capacitor. The stator of the dual three-phase permanent magnet synchronous motor consists of two sets of three-phase windings. The two sets of three-phase windings are spatially separated by 30 electrical degrees. The neutral points of the two sets of three-phase windings are isolated from each other and are not grounded.

3. The adaptive robust control method for a multiphase motor drive system according to claim 1, characterized in that, Adjusting the gain value of the sliding mode control based on the aforementioned speed error includes: The gain value is set with a base component and an adaptive component. The base component is a fixed constant, and the adaptive component is set using a linear function relationship. The linear function relationship is that the adaptive component is equal to a first adaptive coefficient multiplied by the absolute value of the speed error. The first adaptive coefficient is a preset positive number used to control the sensitivity of the adaptive component to changes in the speed error. The gain value is obtained by adding the base component and the adaptive component, wherein when the absolute value of the speed error increases, the adaptive component increases accordingly, thereby increasing the gain value; when the absolute value of the speed error decreases, the adaptive component decreases accordingly, thereby decreasing the gain value.

4. The adaptive robust control method for a multiphase motor drive system according to claim 1, characterized in that, Adjusting the approach law form of sliding mode control based on the aforementioned speed error includes: The distance of the system state relative to the sliding surface is divided into three stages: the distance away stage, the proximity stage, and the neighborhood stage. During the distance phase, the exponential approach law is adopted as the dominant approach law, causing the system state to approach the sliding surface at an exponential rate. During the approach phase, a weighted fusion of exponential and power-law approach laws is adopted. As the system state approaches the sliding surface, the weight of the power-law approach law is gradually increased, while the weight of the exponential approach law is gradually decreased. In the neighborhood stage, a power-law approach is used as the dominant approach law, so that the system state enters the sliding surface in a smooth manner. Among them, the weighted fusion method of exponential and power-law approach is adopted, including: A fusion coefficient is set, wherein the fusion coefficient changes continuously between zero and one. When the system state is near the beginning of the approach phase, the fusion coefficient is zero, and when the system state is near the end of the approach phase, the fusion coefficient is one. The weighted component of the exponential reaching law is obtained by multiplying the exponential reaching law by one and subtracting the fusion coefficient. Calculate the power-approach law by multiplying it by the fusion coefficient to obtain the weighted components of the power-approach law; The weighted components of the exponential approach law and the power approach law are added together to obtain the fused approach law.

5. The adaptive robust control method for a multiphase motor drive system according to claim 1, characterized in that, Adjusting the integral coefficient of the sliding surface based on the aforementioned rotational speed error includes: Set a first preset threshold and a second preset threshold, wherein the first preset threshold is greater than the second preset threshold; When the absolute value of the speed error is greater than the first preset threshold, the integral coefficient is set to the first integral value, wherein the first integral value is a low integral value to avoid system overshoot and oscillation caused by integral saturation; When the absolute value of the rotational speed error is less than the second preset threshold, the integral coefficient is set to the second integral value, wherein the second integral value is a high integral value, which is used to enhance the ability to eliminate residual errors and improve the steady-state accuracy of the system. When the absolute value of the speed error is greater than or equal to the second preset threshold and less than or equal to the first preset threshold, the integral coefficient is set to an intermediate value that increases continuously as the absolute value of the speed error decreases, wherein the intermediate value is between the first integral value and the second integral value.

6. The adaptive robust control method for a multiphase motor drive system according to claim 1, characterized in that, The system generates uncertainty compensation quantities by estimating parameter variations, unmodeled dynamic quantities, and external disturbances in a multiphase motor drive system online using a state observer, including: An extended state observer is constructed, wherein the order of the extended state observer is determined according to the system order of the dual three-phase permanent magnet synchronous motor, and the extended state observer expands the parameter changes, unmodeled dynamic quantities and external disturbances into new state variables, which are uniformly observed as the total uncertainty. The collected current, voltage, and speed information of the dual three-phase permanent magnet synchronous motor are sent as input signals to the extended state observer. The extended state observer calculates and outputs an estimate of the total uncertainty in real time based on the input signal and its internal state equation, wherein the internal state equation describes the dynamic relationship between the state variables of the extended state observer and the input signal. A compensation coefficient is set, and the estimated value of the total uncertainty is multiplied by the compensation coefficient to obtain the uncertainty compensation amount. The uncertainty compensation amount is then fed forward and superimposed into the calculation result of the sliding mode control law, so that the uncertainty compensation amount actively offsets the uncertainty influence in the multiphase motor drive system.

7. The adaptive robust control method for a multiphase motor drive system according to claim 1, characterized in that, Based on the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount, control commands are generated, including: A working condition identification unit is set up, wherein the working condition identification unit identifies the current operating stage of the system as one of the following based on the collected speed error, current error, error change rate and fault diagnosis signal of the dual three-phase permanent magnet synchronous motor: startup stage, steady-state operation stage, fault recovery stage or parameter change stage. A weight allocation unit is set up, wherein the weight allocation unit is used to allocate corresponding weight values ​​to the gain value, the approach law form, the integral coefficient and the uncertainty compensation amount according to the operating stage identified by the operating condition identification unit. The weight values ​​are used to represent the degree of influence of each parameter in the final control command. The working condition identification unit and the weight allocation unit are integrated to form a coordination controller, wherein the weight allocation unit is connected to the working condition identification unit and is used to receive the operating stage identified by the working condition identification unit; The coordination controller evaluates the current operating stage of the system in real time, and determines the weights of the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount in the final control command based on the operating stage. Based on the weights and preset priority order, the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount are fused and calculated to output the final control command, wherein the control command includes the torque current command value.

8. The adaptive robust control method for a multiphase motor drive system according to claim 7, characterized in that, The weights of the gain value, the approach law form, the integral coefficient, and the uncertainty compensation in the final control command are determined according to the aforementioned operating phases, including: When the operation phase is the startup phase, the weights of the gain value and the approach law form are set to be higher than the weights of other parameters, the weight of the integral coefficient is set to be lower than the weights of other parameters, and the weight of the uncertainty compensation is set to be between the weight of the gain value and the weight of the integral coefficient, so that the system mainly uses strong control and supplements integral action in the startup phase. When the operating phase is a steady-state operating phase, the weight of the integral coefficient is set to be higher than the weight of other parameters, the weight of the approach law form is set to be lower than the weight of other parameters, and the weight of the gain value and the weight of the uncertainty compensation amount are set to be between the weight of the integral coefficient and the weight of the approach law form, so that the system mainly uses integral regulation and weakens the switching effect in the steady-state operating phase. When the operation phase is the fault recovery phase, the weights of the gain value, the approach law form, and the uncertainty compensation amount are all set to be higher than the weights of the integral coefficient, and the weight of the uncertainty compensation amount is not lower than the weights of the gain value and the approach law form. The weight of the integral coefficient is set to be lower than the weights of other parameters, so that the system can quickly restore stability by focusing on strong control and uncertainty compensation during the fault recovery phase. When the operation phase is the parameter change phase, the weight of the uncertainty compensation amount is set to be higher than the weight of other parameters, and the weights of the gain value, the approach law form, and the integral coefficient are set to be equal and lower than the weight of the uncertainty compensation amount, so that the system mainly uses uncertainty compensation and the various control parameters play a balanced role in the parameter change phase.

9. The adaptive robust control method for a multiphase motor drive system according to claim 7, characterized in that, Based on the weights and a preset priority order, the gain value, reaching law form, integral coefficient, and uncertainty compensation amount are fused and calculated to output the final control command, including: Based on the weights, the calculation results of the gain value, the calculation results of the reaching law form, the calculation results of the integral coefficient, and the calculation results of the uncertainty compensation amount are weighted and summed to obtain a preliminary control command, wherein the control command includes a torque current command value. Determine whether there is a directional conflict among the calculation results of the gain value, the calculation results of the reaching law form, the calculation results of the integral coefficient, and the calculation results of the uncertainty compensation amount, wherein the directional conflict refers to some calculation results being in the positive direction while other calculation results are in the negative direction; When there is a directional conflict, the dominant parameter is determined according to a preset priority order. The calculation result of the dominant parameter is used as a benchmark, and the calculation results of the others that are opposite to the dominant parameter are set to zero and then weighted and summed again to obtain the corrected control command. The priority order is as follows: the uncertainty compensation amount has the highest priority, the gain value has the second highest priority, the approach law form has the third highest priority, and the integral coefficient has the lowest priority.

10. An adaptive robust control system for a multiphase motor drive system, characterized in that, An adaptive robust control method for implementing the multiphase motor drive system according to any one of claims 1-9 includes: The error calculation module is used to collect the operating status data of the dual three-phase permanent magnet synchronous motor, calculate the speed error between the current speed and the target speed, and adjust the gain value of the sliding mode control, the approach law form of the sliding mode control, and the integral coefficient of the sliding surface according to the speed error. The compensation quantity generation module is used to estimate the parameter changes, unmodeled dynamic quantities and external disturbances in the multiphase motor drive system online through the state observer, and generate the uncertainty compensation quantity. The drive control module is used to generate control commands based on the gain value, the approach law form, the integral coefficient, and the uncertainty compensation amount, and output the control commands to the six-phase modular multilevel converter to drive the dual three-phase permanent magnet synchronous motor to run.

Images