Master-slave redundant and model predictive based control architecture for aviation permanent magnet synchronous motor

Abstract

This invention provides a control architecture for aerospace permanent magnet synchronous motors based on master-slave redundancy and model prediction, belonging to the field of aerospace electric propulsion control systems. It aims to solve problems in existing systems such as extreme environment parameter drift, single-fault shutdown, multi-objective control conflicts, and fault diagnosis lag. The control architecture includes a master-slave dual-redundant control unit, a wide-temperature-range adaptive model prediction control unit, an electromagnetic interference adaptive suppression unit, and a multi-dimensional fault-tolerant unit. This invention employs a DSP+FPGA heterogeneous redundancy architecture combined with a 3-out-of-2 voting logic to achieve seamless fault switching; it embeds an extended Kalman filter observer to update motor parameters online, preset takeoff / cruise / landing three modes; it suppresses interference in the 100kHz~10MHz frequency band through adaptive notch filtering and sliding mode variable structure control; and it integrates a long short-term memory network to achieve accurate early fault diagnosis, adapting to aerospace scenarios such as electric general aviation aircraft and drones.

Description

Technical Field

[0001] This invention belongs to the field of aviation electric propulsion control system technology, specifically relating to an aviation permanent magnet synchronous motor control architecture, which is particularly suitable for aviation carriers such as electric general aviation aircraft, medium and large UAVs and electric regional jets. It can operate stably in extreme flight scenarios with high altitude, wide temperature range, strong electromagnetic interference and high safety redundancy requirements, and achieve high-precision and high-reliability control of permanent magnet synchronous motors. Background Technology

[0002] As the aviation industry transforms towards electrification and greening, permanent magnet synchronous motors, with their advantages of high power density and high operating efficiency, have become the core actuators of aviation electric propulsion systems. However, the unique characteristics of the aviation flight environment, such as high altitude, low temperature, low pressure, strong vibration, strong electromagnetic interference, and airworthiness standards, place far more stringent requirements on motor control systems than on the ground. Existing control architectures suffer from numerous technical deficiencies that urgently need to be addressed.

[0003] Parameter drift under extreme environments leads to control inaccuracies: During flight, ambient temperatures can fluctuate drastically between -55℃ and 200℃, accompanied by low-pressure environments below 0.2MPa. Traditional permanent magnet synchronous motor control systems employ fixed-parameter PID control strategies. However, the wide temperature range causes stator resistance drift of 25%~40% and permanent magnet flux attenuation of over 30%. The low-pressure environment further reduces motor heat dissipation efficiency, triggering secondary parameter drift. These problems directly result in speed fluctuations exceeding ±8% and torque response delays exceeding 100ms in traditional control systems. In severe cases, this can cause propulsion system output jitter, affecting flight attitude stability.

[0004] Existing control architectures mostly employ a single controller + single sensor architecture, where a single fault directly triggers a shutdown, failing to meet the "single fault, no failure" airworthiness standard. Even with some redundancy designs, the switching response time is >100ms, easily causing sudden thrust changes. Multi-condition adaptability is weak; a single control strategy cannot balance takeoff, cruise, and landing requirements, resulting in a 40% slower takeoff torque increase or only 85% cruise efficiency, and no electromagnetic compatibility optimization. Fault diagnosis and fault tolerance capabilities are lacking, relying on threshold judgment methods, with an early fault identification rate <40% and a false positive rate >25%. There is no active fault tolerance strategy, and when the core controller crashes, the position sensor fails, or the power module IGBT breaks down, the control signal is directly interrupted, and the motor stops instantly. After a fault, only shutdown or alarm is triggered, resulting in insufficient safety redundancy.

[0005] Multi-objective control conflicts prevent adaptation to all flight conditions: Aviation flight includes three typical flight conditions: takeoff, cruise, and landing. These conditions inherently conflict with the motor control objectives: takeoff requires "maximum torque output" to quickly gain lift, cruise requires "minimum energy consumption" to extend range, and landing requires "smooth deceleration" to ensure landing safety. Traditional control systems employ a single control strategy, which cannot simultaneously address multiple objectives: if cruise energy saving is the optimization objective, the torque increase rate during takeoff is reduced by 40%, failing to meet takeoff distance requirements; if takeoff torque is the objective, motor efficiency during cruise is only 85%, reducing range by 15%. Furthermore, traditional control strategies lack electromagnetic compatibility optimization for avionics systems. Electromagnetic interference in the 100kHz~10MHz frequency band generated during motor operation can easily interfere with navigation and communication systems, failing to meet the DO-160F electromagnetic compatibility standard.

[0006] Fault diagnosis is lagging and has a high false alarm rate, posing significant operational risks: Existing control systems mostly use the "threshold judgment method" for fault diagnosis, which sets fixed thresholds for current and voltage, triggering an alarm when these thresholds are exceeded. This method can only identify serious faults such as "inter-turn short circuits and complete sensor failure," with an identification rate of less than 40% for early faults such as "minor inter-turn short circuits, sensor drift, and slow demagnetization of permanent magnets." Furthermore, affected by extreme environmental noise, the false alarm rate exceeds 25%, frequently leading to "false shutdowns" or "missed fault reports," causing a gradual decrease in thrust during flight and forcing an emergency landing. Summary of the Invention

[0007] To address the core deficiencies of existing aviation permanent magnet synchronous motor control systems in terms of extreme environment adaptability, safety redundancy, multi-objective control, and fault diagnosis, this application provides an aviation permanent magnet synchronous motor control architecture based on master-slave redundancy and model prediction. This architecture achieves "wide-temperature-range high-precision control, single-fault non-perceptible fault tolerance, multi-objective optimization under all operating conditions, and accurate early fault warning," meeting aviation airworthiness standards and improving system reliability.

[0008] This application provides the following technical solution: a control architecture for an aviation permanent magnet synchronous motor based on master-slave redundancy and model prediction, the control architecture including a master-slave dual-redundant control unit, a wide temperature range adaptive model prediction control unit, an electromagnetic interference adaptive suppression unit, and a multi-dimensional fault-tolerant unit;

[0009] The master-slave dual-redundant control unit adopts a heterogeneous architecture consisting of a DSP master controller and an FPGA slave controller. The heterogeneous architecture is equipped with a redundancy switching module with a three-out-of-two voting logic and a triple-redundant sensor group consisting of a position sensor, a current sensor, and a temperature sensor to compare the speed and torque commands of the master and slave controllers and the sensor data in real time, and realize rapid fault detection and seamless switching based on the comparison results.

[0010] The wide-temperature-range adaptive model prediction and control unit incorporates an extended Kalman filter observer and a multi-modal control module. The extended Kalman filter observer is used to collect real-time data on the motor's three-phase current, speed, and temperature, and to update the stator resistance and permanent magnet flux parameters online. The multi-modal control module presets three control modes: takeoff, cruise, and landing. The mode switching response time is <20ms, and each mode achieves differentiated control by optimizing the objective function.

[0011] The electromagnetic interference adaptive suppression unit adopts a combined architecture of an adaptive notch filter module and a sliding mode variable structure control module to suppress electromagnetic interference and reduce the intensity of interference sources; the operating frequency band of the adaptive notch filter module is 100kHz~10MHz, and the switching frequency adjustment range of the sliding mode variable structure control module is 5kHz~20kHz.

[0012] The multi-dimensional fault-tolerant unit includes a data-driven diagnostic module and a model-driven diagnostic module. The data-driven diagnostic module uses a long short-term memory network model to identify faults, while the model-driven diagnostic module calculates the deviation between the observed and actual flux linkage values ​​based on a motor health model to identify faults. The multi-dimensional fault-tolerant unit is also equipped with a fault-tolerant strategy execution module for different fault types.

[0013] According to some embodiments of the present invention, a fault-tolerant control architecture for a permanent magnet synchronous motor based on master-slave redundancy and model prediction is provided in this embodiment. The system architecture includes a master-slave dual-redundant control unit, a wide-temperature-range adaptive model prediction control unit, an electromagnetic interference adaptive suppression unit, and a multi-dimensional fault-tolerant unit.

[0014] To address the issue of insufficient safety redundancy, a master-slave dual-redundant control unit is established. This unit employs a heterogeneous redundancy architecture of a "DSP master controller + FPGA slave controller," coupled with a triple-redundant sensor group and a high-speed redundancy switching module, to achieve rapid fault detection and seamless switching.

[0015] Heterogeneous controller configuration: The main controller uses a TI TMS320C6678 multi-core DSP with a main frequency of 1.2GHz, responsible for real-time speed / torque closed-loop control, multi-modal strategy switching, and data interaction; the slave controller uses a Xilinx Kintex-7 series FPGA with more than 200,000 logic gates, independently acquires sensor data and synchronously runs a simplified control algorithm, comparing it with the main controller output in real time. This heterogeneous architecture avoids "same-source failures" and improves redundancy and reliability.

[0016] Triple Redundant Sensor Group: Equipped with 2 sets of rotary transformers, 3 sets of Hall current sensors, and 2 sets of platinum resistance temperature sensors, all sensor signals are transmitted to the master-slave controller simultaneously to ensure the reliability of status data acquisition.

[0017] High-speed redundancy switching module: Integrates a dedicated logic chip to implement a "two-out-of-three" voting logic, comparing the speed and torque commands of the master and slave controllers with sensor data in real time. Switching is triggered when any of the following conditions are met: the deviation between the master controller output and the slave controller output > 5%; master controller communication interruption > 10μs; abnormal sensor data voting results. The switching process is implemented through hardware circuitry, with a response time ≤ 50μs, and motor speed fluctuation < ±1% after switching.

[0018] The wide-temperature-range adaptive model predictive control (MPC) unit replaces traditional PID control with model predictive control algorithms, embeds an extended Kalman filter (EKF) observer to achieve online parameter identification, and combines with a multi-modal control module to adapt to all flight conditions.

[0019] The EKF parameter online identification module, based on the motor mathematical model, collects three-phase current, speed, and temperature data in real time. It iteratively calculates the actual values ​​of stator resistance and permanent magnet flux linkage every 10ms using the EKF algorithm, dynamically correcting the control model parameters. For a wide temperature range of -55℃ to 200℃, it corrects the positive temperature coefficient characteristics of the stator resistance using a temperature compensation algorithm and corrects the change in the permanent magnet energy product using a flux linkage decay model.

[0020] Multimodal MPC control module: Presets three control modes: "takeoff," "cruise," and "landing." Each mode achieves differentiated control through optimized objective functions. Takeoff mode: Objective function is "maximizing torque output," with weighting coefficients tilted towards torque tracking error, ensuring the motor reaches 95% of rated torque within one second, shortening takeoff distance. Cruise mode: Objective function is "minimizing energy consumption," incorporating an efficiency map model and dynamically adjusting dq-axis current distribution through an MPC algorithm to keep the motor operating in its highest efficiency range. Landing mode: Objective function is "minimizing the rate of change of rotational speed," limiting the rate of torque change to achieve smooth deceleration.

[0021] Modal adaptive switching logic: By collecting signals of aircraft throttle opening, flight altitude, and speed, the current operating condition is automatically determined and the mode is switched. The switching process adopts a "weighted smooth transition" algorithm to avoid torque shock during mode switching.

[0022] The electromagnetic interference adaptive suppression unit adopts a "interference tracking-suppression" closed-loop architecture, combining adaptive notch filtering and sliding mode variable structure control to accurately suppress electromagnetic interference and reduce the intensity of interference sources.

[0023] Adaptive notch filter module: It tracks the frequency of electromagnetic interference signals in the 100kHz~10MHz band generated during motor operation in real time through a phase-locked loop, dynamically adjusts the center frequency and quality factor of the notch filter, and attenuates the interference signal by more than 40dB without affecting the normal transmission of control signals.

[0024] Sliding mode variable structure control module: As a means of suppressing interference sources, it dynamically adjusts the IGBT switching frequency to keep the switching noise frequency away from the sensitive frequency band of the avionics system (such as 1MHz~5MHz of the navigation system). The switching frequency adjustment is coordinated and controlled by the MPC algorithm to ensure that it does not affect the motor control accuracy.

[0025] Electromagnetic compatibility closed-loop optimization: The integrated electromagnetic interference sensor collects the radiated interference intensity of the motor housing in real time. When the interference value exceeds the DO-160F Class 3 standard (≤40dBμV / m), the notch filter depth is automatically increased and the switching frequency is adjusted to form an "interference detection-suppression-feedback" closed loop.

[0026] The multi-dimensional fault-tolerant unit constructs a data-driven + model-driven integrated diagnostic system, and adaptively executes fault-tolerant strategies based on fault types to achieve full-process coverage of "early warning - precise location - fault-tolerant control":

[0027] Data-driven diagnostic module: Based on a Long Short-Term Memory (LSTM) network, it trains current, voltage, and temperature time-series data samples under normal motor operating conditions to establish a normal operation mode model. During real-time operation, it calculates the deviation between the current time-series data and the model output. When the deviation is greater than 3σ (σ is the standard deviation of normal samples), it triggers an early warning, which can identify faults such as "minor inter-turn short circuits and sensor drift".

[0028] Model-driven diagnostic module: Based on the parameters such as flux linkage and resistance output by the EKF observer, it compares them with the motor health model (calibrated at the factory) to calculate flux linkage deviation and resistance deviation. When the flux linkage deviation is >2%, it is determined to be a permanent magnet demagnetization fault; when the resistance deviation is >5%, it is determined to be a winding insulation aging fault, thus achieving precise fault type location.

[0029] Fault-Tolerant Strategy Execution Module: Executes differentiated strategies for different fault types. Position sensor fault: Switches to EKF-based sensorless control mode, estimating rotor position using current and speed data; IGBT single-transistor fault: Reconstructs the PWM modulation strategy, shuts down the faulty bridge arm, and adjusts the switching timing of the remaining bridge arms to maintain motor output at over 75% of rated power; Controller fault: Triggers master-slave redundancy switching, seamlessly taking over control from the controller; Early fault: Adjusts current distribution using the MPC algorithm to compensate for performance degradation, while simultaneously sending a warning signal to the flight control system to prompt ground maintenance.

[0030] The master-slave dual-redundant control unit serves as the core coordination node of the entire architecture, undertaking the central functions of data distribution, command execution, and fault response. On the one hand, its triple-redundant sensor group (including 2 sets of rotary transformers, 3 sets of Hall current sensors, and 2 sets of platinum resistance temperature sensors) collects key status data such as motor position, current, and temperature in real time, and distributes them to the wide-temperature-range adaptive model predictive control unit and the multi-dimensional fault-tolerant unit through a synchronous transmission mechanism, providing accurate data support for subsequent analysis, optimization, and fault diagnosis. On the other hand, this unit receives the optimal control command output by the wide-temperature-range adaptive model predictive control unit, drives the power module to achieve motor control after SVPWM modulation, and triggers redundancy switching based on the "two out of three" voting logic according to the diagnostic results generated by the multi-dimensional fault-tolerant unit, ensuring that control is not interrupted in single fault scenarios.

[0031] The wide-temperature-range adaptive model predictive control unit takes real-time operating data (three-phase current, speed, and temperature) and flight control commands (throttle opening, altitude, and speed) transmitted from the master-slave dual-redundant control unit as input. It dynamically corrects the stator resistance and permanent magnet flux parameters every 10ms through an extended Kalman filter observer to offset the parameter drift effects in a wide temperature range of -55℃ to 200℃. At the same time, it automatically completes multi-mode switching for takeoff, cruise, and landing, and generates the optimal dq-axis current command to feed back to the master-slave dual-redundant control unit. In addition, this unit also receives interference detection data from the electromagnetic interference adaptive suppression unit and coordinates the adjustment of the IGBT switching frequency to avoid interference affecting control accuracy.

[0032] The electromagnetic interference adaptive suppression unit collects electromagnetic interference data of the motor in the 100kHz~10MHz frequency band in real time through a dedicated sensor. It adopts a combination of "adaptive notch filtering + sliding mode variable structure control": the notch filtering module tracks the interference frequency through a phase-locked loop and attenuates the interference signal by >40dB; the sliding mode control module dynamically adjusts the switching frequency to avoid the sensitive frequency band of avionics; at the same time, the interference detection data is synchronized to the wide temperature range adaptive model prediction control unit, forming a closed-loop optimization of "interference detection-suppression-feedback" to ensure that the electromagnetic radiation complies with the DO-160F Class 3 standard.

[0033] The multi-dimensional fault-tolerant unit constructs a "data-driven + model-driven" integrated diagnostic system based on the status data transmitted by the master-slave dual-redundant control unit: the data-driven module identifies early faults with data deviation > 3σ through a long short-term memory network (LSTM), and the model-driven module locates the fault type by comparing the observed flux linkage value with the actual value, and then generates a differentiated fault-tolerant strategy to be fed back to the master-slave dual-redundant control unit for execution.

[0034] The four main units interact with each other via the CAN-FD bus, forming a complete closed loop of "data acquisition - analysis and optimization - control execution - fault tolerance". This ensures the stable operation of the aviation permanent magnet synchronous motor in extreme scenarios such as high altitude, wide temperature range, and strong electromagnetic interference, and meets the aviation airworthiness requirement of "no failure under single fault".

[0035] This invention achieves the following advantages compared to existing technologies through a collaborative architecture design of four major units: master-slave redundancy, adaptive MPC, electromagnetic interference suppression, and fault tolerance:

[0036] (1) The present invention proposes a fault-tolerant control architecture for permanent magnet synchronous motor based on master-slave redundancy and model prediction. It adopts online parameter identification and dynamic correction mechanism, which can stably adapt to a wide range of environments in aviation scenarios, from high altitude, low temperature and low pressure to high temperature in the engine compartment. It effectively avoids control performance fluctuations caused by environmental changes and ensures that the motor can maintain stable output under various extreme conditions.

[0037] (2) The present invention proposes a fault-tolerant control architecture for permanent magnet synchronous motor based on master-slave redundancy and model prediction. It adopts a heterogeneous master-slave redundancy architecture and high-speed switching design. When a single fault occurs in the controller, sensor or power module, it can achieve seamless and rapid switching. At the same time, it maintains core operating performance through differentiated fault-tolerant strategies, which fully meets the safety requirements of "single fault does not fail" in the aviation field and provides reliable protection for flight safety.

[0038] (3) The present invention proposes a fault-tolerant control architecture for permanent magnet synchronous motor based on master-slave redundancy and model prediction. Relying on multi-modal control strategy, it can accurately match the core requirements of different flight conditions such as takeoff, cruise and landing. During takeoff, it can quickly output sufficient power; during cruise, it can optimize energy consumption to extend the range; and during landing, it can achieve smooth deceleration. At the same time, through electromagnetic interference suppression design, it avoids interference to the avionics system and meets the requirements of aviation electromagnetic compatibility.

[0039] (4) The present invention proposes a fault-tolerant control architecture for permanent magnet synchronous motor based on master-slave redundancy and model prediction. It adopts a "data + model" fusion diagnostic system, which can accurately identify early potential faults such as permanent magnet demagnetization and winding aging and provide early warning. At the same time, it can actively compensate for performance degradation through control strategy adjustment, reduce the risk of sudden faults, and reduce the difficulty and cost of operation and maintenance. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic diagram of the control architecture of an aviation permanent magnet synchronous motor according to an embodiment of the present invention;

[0042] Figure 2 This is a flowchart of the master-slave dual-redundant control unit in an embodiment of the present invention;

[0043] Figure 3 This is a flowchart of the algorithm for the wide-temperature-range adaptive MPC unit in an embodiment of the present invention;

[0044] Figure 4 This is a schematic diagram illustrating the output variation and EKF identification accuracy under high-altitude and low-temperature conditions in an embodiment of the present invention.

[0045] Figure 5 This is a schematic diagram of the power output curve after fault injection during the cruise phase in an embodiment of the present invention, used to compare the output stability before and after the fault and after fault tolerance. Detailed Implementation

[0046] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0047] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. This application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0048] like Figure 1 As shown, this embodiment of the invention provides a control architecture for an aircraft permanent magnet synchronous motor based on master-slave redundancy and model prediction. The control architecture includes a master-slave dual-redundant control unit, a wide-temperature-range adaptive model prediction control unit, an electromagnetic interference adaptive suppression unit, and a multi-dimensional fault-tolerant unit.

[0049] The master-slave dual-redundant control unit adopts a heterogeneous architecture consisting of a DSP master controller and an FPGA slave controller. The heterogeneous architecture is equipped with a redundancy switching module with a "two out of three" voting logic, and is equipped with a triple-redundant sensor group consisting of a position sensor, a current sensor, and a temperature sensor to compare the speed and torque commands of the master and slave controllers and the sensor data in real time, and realize rapid fault detection and seamless switching based on the comparison results.

[0050] The wide-temperature-range adaptive model prediction and control unit incorporates an extended Kalman filter observer and a multi-modal control module. The extended Kalman filter observer is used to collect real-time data on the motor's three-phase current, speed, and temperature, and to update the stator resistance and permanent magnet flux parameters online. The multi-modal control module presets three control modes: takeoff, cruise, and landing. The mode switching response time is <20ms, and each mode achieves differentiated control by optimizing the objective function.

[0051] The electromagnetic interference adaptive suppression unit adopts a combined architecture of an adaptive notch filter module and a sliding mode variable structure control module to suppress electromagnetic interference and reduce the intensity of interference sources; the operating frequency band of the adaptive notch filter module is 100kHz~10MHz, and the switching frequency adjustment range of the sliding mode variable structure control module is 5kHz~20kHz.

[0052] The multi-dimensional fault-tolerant unit includes a data-driven diagnostic module and a model-driven diagnostic module. The data-driven diagnostic module uses a long short-term memory network model to identify faults, while the model-driven diagnostic module calculates the deviation between the observed and actual flux linkage values ​​based on a motor health model to identify faults. The multi-dimensional fault-tolerant unit is also equipped with a fault-tolerant strategy execution module for different fault types.

[0053] The master-slave dual-redundant control unit is the core coordination node. On one hand, it collects motor position, current, temperature, and other status data through a triple-redundant sensor group and synchronously transmits this data to the wide-temperature-range adaptive model prediction control unit and the multi-dimensional fault-tolerant unit. Simultaneously, it triggers redundancy switching based on the diagnostic results of the multi-dimensional fault-tolerant unit. The wide-temperature-range adaptive model prediction control unit, based on real-time operating data and flight control commands provided by the master-slave dual-redundant control unit, dynamically corrects parameters and completes multi-mode switching through an extended Kalman filter observer. It also receives interference detection data from the electromagnetic interference adaptive suppression unit to collaboratively adjust the IGBT switching frequency. The electromagnetic interference adaptive suppression unit collects real-time motor electromagnetic interference data, uses adaptive notch filtering and sliding mode variable structure control to suppress interference, and synchronizes the interference detection data to the wide-temperature-range adaptive model prediction control unit to form a closed-loop optimization. The multi-dimensional fault-tolerant unit, based on the status data transmitted by the master-slave dual-redundant control unit, uses "data + model" fusion to diagnose and identify faults and generate fault-tolerant strategies, which are then fed back to the master-slave dual-redundant control unit for execution. These four units are connected via CAN. The FD bus enables data interaction, forming a complete closed loop of "data acquisition - analysis and optimization - control execution - fault tolerance", which together ensures the stable operation of aviation permanent magnet synchronous motors.

[0054] The master-slave dual-redundant control unit proposed in this invention, such as Figure 2As shown. The process is as follows: After the system is powered on, the DSP master controller and FPGA slave controller synchronously perform hardware self-test, algorithm self-test, and parameter self-test. After the self-test is passed, it enters the standby state. Subsequently, a triple redundant sensor group consisting of 2 sets of rotary transformers, 3 sets of Hall current sensors, and 2 sets of platinum resistance temperature sensors synchronously transmits the collected motor position, current, and temperature data to the master and slave controllers. The master and slave controllers compare the consistency of speed commands, torque output, and sensor data every 2ms in real time, and monitor the communication link. If the "two out of three" voting logic is triggered, the slave controller quickly and seamlessly takes over the control authority through hardware circuitry. When there is no fault, the master controller receives the dq axis current command output by the wide temperature range adaptive MPC unit and drives the power module through SVPWM modulation to realize motor control.

[0055] The algorithm for the wide-temperature-range adaptive MPC unit proposed in this invention is as follows: Figure 3 As shown. The process is as follows: First, two types of core data are obtained from the master-slave dual-redundant control unit, namely real-time operating data: three-phase current, motor speed and winding temperature, and flight control command data: throttle opening, flight altitude and speed. The current signal is preprocessed by wavelet packet decomposition and the temperature signal by low-pass filtering. Then, the extended Kalman filter observer iteratively updates the stator resistance and permanent magnet flux every 10ms to dynamically correct the motor control model parameters to adapt to a wide temperature range of -55℃ to 200℃. Subsequently, the current operating condition is determined based on the flight control command data, and the three modes of takeoff, cruise and landing are quickly switched. The takeoff mode aims to maximize torque output, the cruise mode aims to minimize energy consumption and the landing mode aims to minimize the rate of change of speed. The "weighted smooth transition" algorithm is used to avoid torque shock during the switching. Then, the optimal dq axis current command is generated by rolling time domain optimization. At the same time, the interference detection data of the electromagnetic interference adaptive suppression unit is received. If the interference exceeds DO-160F Class 3, the IGBT switching frequency is adjusted synchronously. Finally, the optimized dq axis current command is sent to the master-slave dual redundant control unit through the CAN FD bus to complete one closed loop of the algorithm process.

[0056] After completing the hardware deployment and software integration of the four major units—master-slave dual-redundant control, wide-temperature-range adaptive model predictive control (MPC), adaptive electromagnetic interference suppression, and multi-dimensional fault tolerance—the system must first build a stable operating environment through initialization and self-testing processes. Considering the high reliability requirements of the control architecture in aviation scenarios, the startup phase must prioritize verifying the functions of core components, calibrating key parameters, and confirming the smooth operation of communication links. This ensures that the master-slave controller, triple-redundant sensor group, MPC / EKF / LSTM, and other core algorithm models are all in a normal and ready state, providing a prerequisite for subsequent full-condition control performance verification and typical fault tolerance response testing. This also avoids potential anomalies during the startup phase that could lead to inaccurate control or fault tolerance failure in subsequent operating conditions.

[0057] System initialization and self-test process Step 1: The system is powered on, and the master and slave controllers start synchronously and perform three self-tests: (1) Hardware self-test: Detect the controller I / O port and sensor signal path. If there is an abnormality, the red indicator light will be lit; (2) Algorithm self-test: Load the MPC, EKF and LSTM models and verify the integrity of the models; (3) Parameter calibration: Read the motor health baseline parameters and store them in Flash. Step 2: After the self-test is passed, the master controller sends a "ready" signal to the flight control system through the CAN FD bus and enters the standby state to wait for the operating condition command.

[0058] Full-condition control performance verification: Three typical aviation operating conditions were set, executed sequentially, and data was collected to verify control accuracy and adaptability.

[0059] Operating Condition 1: High-altitude, low-temperature takeoff (simulated altitude 1000m, temperature -10℃). The flight control system sends a "takeoff" command (throttle opening 100%), and the system executes: ① Mode switching: The MPC unit quickly switches to the takeoff mode, and the objective function weights are tilted towards torque; ② Parameter identification: The EKF collects three-phase current, speed, and temperature data every 10ms and corrects the stator resistance and flux linkage; ③ Output control: The output voltage is modulated by SVPWM, and the motor speed reaches 2000rpm within 1s.

[0060] Condition 2: Cruise at an altitude of 10,000 meters (simulated altitude 10km, temperature -55℃, pressure 0.2MPa, vibration 1500Hz), the flight control system sends the "cruise" command, and the system executes: (1) Mode switching: automatically switches to cruise mode based on altitude and speed signals, and the weight coefficient is adjusted to "efficiency priority"; (2) Efficiency optimization: the MPC algorithm calls the efficiency map and dynamically adjusts the dq axis current ratio to 3:1; (3) Interference suppression: the EMI sensor detects the initial radiation value, the notch filter automatically tracks the interference frequency, the sliding mode switch frequency is adjusted, and the interference is quickly reduced.

[0061] Condition 3: Near-ground landing (simulated altitude 500m, temperature 45℃, throttle opening 20%), the flight control system sends a "landing" command, and the system executes: (1) Mode switching: quickly switch to landing mode and limit the torque change rate; (2) Smooth deceleration: predict the speed change through MPC and gradually reduce the output torque, with no overshoot during the deceleration process; (3) Landing adaptation: the torque increases rapidly at the moment of contact with the ground and then drops rapidly to 0, with the landing impact acceleration ≤2g.

[0062] The diagram illustrates the output variation and EKF identification accuracy under high-altitude and low-temperature conditions. Figure 4As shown. Simulation results show that, under ideal assumptions, the motor speed starts from 0 and accurately reaches the rated value of 2000 rpm within 1 second with a speed tracking error of less than 1%; the torque responds quickly to the rated value of 477 N·m within 0.1 seconds and is output stably with a fluctuation amplitude of less than ±0.5%; the identification error of the stator resistance by the extended Kalman filter algorithm is less than 0.3%, and the steady-state fluctuation of the permanent magnet flux linkage is less than 0.0005 Wb.

[0063] It should be noted that the linear changes in speed and torque in the simulation are the result of idealization and simplification, which differs from the nonlinear characteristics of actual takeoff conditions. However, this simplified design can focus on verifying the core performance of the control architecture. Overall, it verifies that the architecture has the technical characteristics of meeting dynamic response standards, accurate parameter identification, and strong system robustness in the simulation scenario, meeting the core requirement of "no failure under single fault" in aviation airworthiness standards. It provides theoretical support for subsequent refined simulation, engineering implementation, and experimental verification under actual operating conditions by introducing nonlinear factors.

[0064] Typical fault tolerance performance verification: Under cruise conditions, three types of high-frequency faults from aviation scenarios are artificially injected to verify the fault tolerance response.

[0065] Fault 1: Position sensor failure. Operation: Disconnect one AD2S1210 rotary transformer signal line. The fault tolerance unit detects the current timing deviation of 3.2σ (exceeding the threshold of 3σ) through LSTM and quickly locates the "position sensor failure". It automatically switches to EKF no-position sensor mode and estimates the rotor position by current and speed.

[0066] Fault 2: Main controller crash operation: An abnormal command is sent through the JTAG interface to trigger a DSP crash. The redundancy switching module detects the main controller communication interruption and triggers a "two out of three" vote; the main controller output is quickly cut off and the controller is seamlessly taken over.

[0067] Fault 3: IGBT single tube breakdown operation: The A-phase IGBT is triggered to break down by a pulse signal; the model driving module detects the harmonic distortion rate of the A-phase current and locates the "IGBT single tube fault" within 10ms; the PWM modulation strategy is reconstructed, the A-phase bridge arm is shut down, and the output is maintained by bipolar modulation of the BC phase.

[0068] The power output curve diagram after fault injection during the cruise phase is shown below. Figure 5As shown, the simulation results indicate that during the 0-3s pre-fault stable cruise phase, the power output curve stabilizes near the rated value of 80kW, with only minor fluctuations within ±0.5%, verifying the power stability of the architecture when there is no fault. After an open-circuit fault is actively injected into the main channel power module at 3s, the power output curve drops instantly to trigger the fault-tolerant control mechanism. In the 3-10s post-fault phase, the master-slave redundant control architecture completes channel switching in less than 45μs, and the power output quickly converges to 62.4kW, which is 78% of the pre-fault rated value, and stabilizes with a fluctuation of less than ±0.3%, meeting the core requirement of "no failure under single fault". The simulation also verifies that the architecture has the technical characteristics of rapid fault detection, seamless channel switching, and stable power output.

[0069] This invention discloses a fault-tolerant control architecture for aviation permanent magnet synchronous motors based on master-slave redundancy and model prediction. Its protection scope covers the structural design, functional implementation and coordination mechanism of the four core units. Specifically: A master-slave dual-redundant control unit employs a heterogeneous architecture consisting of a DSP master controller (such as the TI TMS320C6678) and an FPGA slave controller (such as the Xilinx Kintex-7). It features a redundant switching module with "two out of three" voting logic and a triple-redundant sensor group composed of two sets of rotary transformers, three sets of Hall current sensors, and two sets of platinum resistance temperature sensors, enabling seamless fault switching and reliable data acquisition. A wide-temperature-range adaptive model predictive control unit incorporates an extended Kalman filter observer that iteratively updates stator resistance and permanent magnet flux linkage parameters every 10ms, as well as a multi-mode control module with preset takeoff, cruise, and landing modes and a switching response of <20ms, adapting to a wide temperature range of -55℃ to 200℃. An electromagnetic interference adaptive suppression unit combines an adaptive notch filter module with an operating frequency band of 100kHz to 10MHz and attenuation >40dB with a sliding mode variable structure control module with a switching frequency adjustment range of 5kHz to 20kHz, ensuring that output interference complies with the DO-160F standard Class 1. Level 3 requirements; multi-dimensional fault-tolerant units, including a data-driven diagnostic module based on long short-term memory networks, a model-driven diagnostic module based on motor health models, and a differentiated fault-tolerant strategy execution module for position sensor, IGBT, and controller faults. These four units exchange data via the CAN FD bus, forming a closed loop of "data acquisition - analysis and optimization - control execution - fault tolerance," protecting the technical solutions for addressing extreme environmental parameter drift, single-fault shutdown, multi-objective control conflicts, and fault diagnosis lag in aviation permanent magnet synchronous motor control. This encompasses architecture composition, hardware selection, algorithm design, parameter thresholds, and collaborative logic.

[0070] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A control architecture for an aircraft permanent magnet synchronous motor based on master-slave redundancy and model prediction, characterized in that, The control architecture includes a master-slave dual-redundant control unit, a wide-temperature-range adaptive model prediction control unit, an electromagnetic interference adaptive suppression unit, and a multi-dimensional fault-tolerant unit. The master-slave dual-redundant control unit adopts a heterogeneous architecture consisting of a DSP master controller and an FPGA slave controller. The heterogeneous architecture is equipped with a redundancy switching module with a three-out-of-two voting logic and a triple-redundant sensor group consisting of a position sensor, a current sensor, and a temperature sensor to compare the speed and torque commands of the master and slave controllers and the sensor data in real time, and realize rapid fault detection and seamless switching based on the comparison results. The wide-temperature-range adaptive model prediction and control unit incorporates an extended Kalman filter observer and a multi-modal control module. The extended Kalman filter observer is used to collect real-time data on the motor's three-phase current, speed, and temperature, and to update the stator resistance and permanent magnet flux parameters online. The multi-modal control module presets three control modes: takeoff, cruise, and landing. The mode switching response time is <20ms, and each mode achieves differentiated control by optimizing the objective function. The electromagnetic interference adaptive suppression unit adopts a combined architecture of an adaptive notch filter module and a sliding mode variable structure control module to suppress electromagnetic interference and reduce the intensity of interference sources; the operating frequency band of the adaptive notch filter module is 100kHz~10MHz, and the switching frequency adjustment range of the sliding mode variable structure control module is 5kHz~20kHz. The multi-dimensional fault-tolerant unit includes a data-driven diagnostic module and a model-driven diagnostic module. The data-driven diagnostic module uses a long short-term memory network model to identify faults, while the model-driven diagnostic module calculates the deviation between the observed and actual flux linkage values ​​based on a motor health model to identify faults. The multi-dimensional fault-tolerant unit is also equipped with a fault-tolerant strategy execution module for different fault types.

2. The control architecture for an aircraft permanent magnet synchronous motor according to claim 1, characterized in that, The DSP master controller uses a TI TMS320C6678 multi-core DSP chip, and the FPGA slave controller uses a Xilinx Kintex-7 series FPGA chip.

3. The control architecture for an aircraft permanent magnet synchronous motor according to claim 1, characterized in that, The triple redundant sensor group includes two sets of rotary transformers for position detection, three sets of Hall current sensors for current detection, and two sets of platinum resistance temperature sensors for temperature detection.

4. The control architecture for an aircraft permanent magnet synchronous motor according to claim 1, characterized in that, The parameter update period of the extended Kalman filter observer is 10ms, and the temperature drift error of the stator resistance and the attenuation error of the permanent magnet flux linkage are corrected by iterative calculation.

5. The control architecture for an aircraft permanent magnet synchronous motor according to claim 1, characterized in that, The optimization objectives for each mode of the multi-modal control module are different: the takeoff mode aims to maximize torque output, the cruise mode aims to minimize motor energy consumption, and the landing mode aims to achieve smooth deceleration.

6. The control architecture for an aircraft permanent magnet synchronous motor according to claim 1, characterized in that, The adaptive notch filter module tracks the frequency of electromagnetic interference signals in real time through a phase-locked loop and dynamically adjusts the filter parameters to achieve precise suppression of interference signals.

7. The control architecture for an aircraft permanent magnet synchronous motor according to claim 1, characterized in that, The fault warning threshold of the multi-dimensional fault-tolerant unit includes: data deviation of the long short-term memory network model > σ is the standard deviation of the normal sample, and the deviation between the observed and actual values ​​of the flux linkage in the motor health model is >2%.

8. The control architecture for an aircraft permanent magnet synchronous motor according to claim 1, characterized in that, The specific strategies of the fault-tolerant strategy execution module are as follows: when the position sensor fails, switch to the sensorless control mode; when a single IGBT tube fails, maintain motor operation by reconstructing the PWM modulation strategy; when the main controller fails, switch to slave controller control through the redundancy switching module.

9. The control architecture for an aircraft permanent magnet synchronous motor according to claim 1, characterized in that, The output electromagnetic interference level of the adaptive electromagnetic interference suppression unit meets the Class 3 requirements of the DO-160F standard.

10. The control architecture for an aircraft permanent magnet synchronous motor according to claim 1, characterized in that, In the master-slave dual-redundant control unit, the master controller and the slave controller synchronously collect and compare motor status data. When the data deviation is greater than 5%, the redundancy switching module is triggered.

Images