A control system and method for predictive allocation of disturbance rejection of ship motors
By employing a multi-rate cascaded hierarchical collaborative fault-tolerant architecture and a composite disturbance rejection strategy, the problems of thrust distribution inaccuracy and disturbance suppression in ship electric propulsion control under strongly nonlinear environments were solved, achieving high-precision thrust tracking, energy consumption optimization, and rapid fault reconfiguration, thus meeting the requirements of high-order functional safety verification.
Patent Information
- Application Number
- CN202611245331.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-17
- Publication Date
- 2026-11-17
Smart Images

Figure CN123113331A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of ship electric propulsion control technology, specifically to a ship motor disturbance prediction and allocation control system and method, applicable to highly robust dynamic positioning and fault-tolerant control of catamarans and twin-propeller propulsion vessels in complex sea conditions. Background Technology
[0002] The operational reliability and power redundancy of a ship's electric propulsion system and various auxiliary machinery drive systems are directly related to the ship's survivability and operational continuity. In the verification of power systems based on hardware-in-the-loop (HIL), existing simulation methods mostly adopt classic PID or linear sliding mode control models. However, the effectiveness of these controllers is highly dependent on the accurate mathematical model of the controlled object, making it difficult to reproduce the strongly nonlinear physical characteristics of the real marine environment. This leads to frequent false positive results, such as steady-state passage in the laboratory but failure of maneuvering on actual ships.
[0003] Patent CN122339343A discloses a fault-tolerant control method for a ship's electric propulsion system under electrical grid faults. It constructs a model predictive control rolling optimization problem based on a discrete predictive model of a doubly-fed induction motor and introduces a switching frequency penalty term into the cost function to achieve torque tracking under fault conditions. Simultaneously, this scheme employs a strategy combining sliding mode control and a saturation function to suppress system chattering, thereby improving the fault tolerance capability of the propulsion system to a certain extent.
[0004] However, this technical solution still has the following inherent defects: First, its MPC optimization framework adopts a fixed weight configuration, which does not consider the changes in the ship's state from full load to light load and the power output differences of the independent battery packs on the port and starboard sides, and cannot dynamically adapt to the energy efficiency optimization requirements under different operating conditions; Second, its disturbance suppression relies on traditional sliding mode control, which lacks real-time observation and active feedforward compensation mechanisms for high-frequency time-varying disturbances such as wave torque impact and propeller water discharge, resulting in a significant decrease in torque tracking accuracy under severe sea conditions; Third, its electrical model and propeller physical limits are independent of each other, and it does not incorporate constraints such as bus voltage drop, upper limit of battery independent power supply, and motor thermal limit into a unified optimization framework, making it difficult to support high-order functional safety verification at ISO 26262 ASIL C / D level. Summary of the Invention
[0005] This invention addresses the shortcomings of existing ship electric propulsion control technologies, such as inaccurate thrust distribution, insufficient disturbance suppression capabilities, and isolated physical constraint handling under strongly nonlinear environmental disturbances. This invention constructs a multi-rate cascaded hierarchical cooperative fault-tolerant architecture, integrating nonlinear model predictive control with an extended state observer-super-torsional sliding mode composite disturbance rejection strategy. This achieves multi-objective optimization of thrust tracking error, energy consumption, and smoothness, and possesses adaptive reconfiguration capabilities under single-side motor failure.
[0006] To achieve the above objectives, the present invention adopts the following technical solution:
[0007] A control system for predictive allocation of disturbance rejection of marine motors includes: a signal acquisition module, a core processing module, and a hardware-in-the-loop simulation interface module; the signal acquisition module includes a GPS / IMU sensor, a current sensor, a voltage sensor, a temperature sensor, and signal conditioning circuits connected to the output terminals of each sensor; the core processing module includes a digital signal processor, a storage unit connected to the digital signal processor, and a pulse width modulation signal generation circuit connected to the output terminal of the digital signal processor; the hardware-in-the-loop simulation interface module includes an FPGA chip and a digital-to-analog converter, one end of the hardware-in-the-loop simulation interface module is connected to the output terminal of the pulse width modulation signal generation circuit, and the other end is connected to the hardware-in-the-loop simulation platform.
[0008] Furthermore, the core processing module also includes an analog-to-digital converter connected to a digital signal processor. The input of the analog-to-digital converter is connected to the output of the signal conditioning circuit, and the storage unit stores a multi-rate cascaded hierarchical control program.
[0009] Furthermore, the hardware-in-the-loop simulation interface module also includes an analog-to-digital converter (ADC), the output of which is connected to the feedback input of the digital signal processor (DSP), and the input of the digital-to-analog converter (DAC) is connected to the output of the pulse width modulation signal generation circuit.
[0010] The present invention also provides a control method for predictive allocation of disturbance rejection of ship motors, comprising the following steps:
[0011] S1: Establish a multi-rate cascaded hierarchical collaborative fault-tolerant control architecture, which includes a guidance layer, a thrust distribution layer, and a motor execution layer; the guidance layer is based on the ship's nonlinear dynamics model and adaptive guidance algorithm, and generates the desired thrust and desired bow torque according to the ship's real-time attitude information;
[0012] S2: The thrust distribution layer constructs a thrust distributor based on nonlinear model predictive control. It takes the desired thrust and desired throttle torque as tracking targets, incorporates the power constraints of the dual battery packs independently powered, the physical limit constraints of the propulsion motors, and the fault reconfiguration constraints into the quadratic programming cost function, and solves the optimal thrust distribution command in real time through rolling optimization, and outputs the desired torque command of each motor.
[0013] S3: The motor execution layer uses an extended state observer to estimate the lumped disturbance in real time. Based on the estimated value, it generates a feedforward compensation amount by combining the super-twisted sliding mode control law. The desired torque command is then superimposed with the feedforward compensation amount to generate the final torque command, which drives the motor to run.
[0014] S4: Real-time monitoring of motor operating status based on residual analysis. When a fault is detected, an adaptive reconstruction mechanism based on an expert rule base is triggered to adjust the thrust distribution strategy or guidance parameters to achieve fault-tolerant control.
[0015] Furthermore, the adaptive guidance algorithm adopts a forward-looking distance strategy that adapts to ship speed, calculates the desired heading angle based on the lateral trajectory error, and designs a virtual control law based on the backstepping method to calculate the desired bow turning angular velocity command.
[0016] Furthermore, the quadratic programming cost function includes a tracking error penalty term, an energy consumption penalty term, and an output smoothness penalty term, which are adjusted by the corresponding weight matrices.
[0017] Furthermore, the power constraints for independent power supply of the dual battery packs include the maximum allowable output power limit for each battery pack; the physical limit constraints for the propulsion motors include the amplitude limit for the thrust output of each motor and the thrust change rate limit; when a single-side motor fault is detected, the fault reconfiguration constraint limits the thrust on the corresponding side to the product of the rated thrust and the remaining output coefficient.
[0018] Furthermore, the extended state observer treats the unknown dynamics in the motor motion equations and external disturbances as the lumped disturbance, establishes an extended state space model containing the lumped disturbance state for real-time observation, and outputs a real-time estimate of the lumped disturbance.
[0019] Furthermore, the super-twisted sliding mode control law defines the sliding surface based on the motor angular velocity tracking error, adopts an exponential approach law and combines the real-time estimate of the lumped disturbance for feedforward cancellation, and the control law uses a saturation function to replace the sign function to weaken chattering.
[0020] Furthermore, the adaptive reconfiguration mechanism includes: when the fault remaining output coefficient is between 0.3 and 0.8, it is determined to be a derating operation fault, and the output of the healthy side motor is improved by adding a power compensation penalty term of the healthy side motor in the cost function; when the fault remaining output coefficient is zero, it is determined to be a complete failure fault, and the thrust of the fault side is set to zero. The required compensation rudder angle is estimated by the extended state observer and the desired heading is corrected to achieve joint rudder and propeller fault tolerance; when it is determined to be a mechanical lock-up fault, the fault side motor applies short-circuit braking to generate damping torque, and only the healthy side motor propels.
[0021] Compared with traditional solutions, the present invention has the following advantages:
[0022] (1) This invention abandons the linear assumption of traditional rigid hulls and establishes a deep coupling relationship between the three-degree-of-freedom nonlinear dynamic model of the ship and the propulsion motor model, and introduces a thrust loss mapping mechanism that dynamically changes with the wave direction. This mechanism enables the control system to truly perceive the interaction between the ship's motion and the marine environment, and fundamentally solves the problem of ship stall or severe yaw caused by inaccurate thrust distribution when the sea state changes, which is a problem of conventional methods.
[0023] (2) The NMPC cost function constructed in this invention simultaneously includes three terms: tracking error penalty, energy consumption penalty, and output smoothness penalty. The three terms are dynamically coordinated through an adaptive weight matrix. Compared with a single target tracking strategy, this invention reduces power consumption while ensuring track tracking accuracy, and mitigates mechanical wear by limiting the thrust change rate, thus achieving the best overall balance between maneuverability, economy, and equipment lifespan.
[0024] (3) This invention uses an extended state observer to unify all unknown disturbances (wave torque, propeller water impact, parameter perturbation, and unmodeled dynamics) inside and outside the system into a lumped disturbance for real-time estimation, and combines it with a super-twisted sliding mode control law to achieve feedforward compensation. This mechanism is completely decoupled from the physical parameters of the motor, significantly reduces speed overshoot and greatly shortens settling time under sudden load changes, and is not sensitive to motor parameter perturbations, so there is no need to readjust the controller parameters.
[0025] (4) This invention establishes a digital expert rule base covering three failure modes: derated operation, complete failure, and mechanical lock-up, enabling automatic identification of failure types and accurate matching of reconfiguration strategies. When a single motor fails completely, seamless reconfiguration can be completed within a few hundred milliseconds through rudder angle compensation and desired heading correction, maintaining the ship's course controllability and providing technical support for the propulsion system to meet functional safety certification requirements.
[0026] (5) This invention adopts a cascaded architecture with three levels of different update cycles: guidance layer, thrust distribution layer, and motor execution layer. The fast and slow channels each perform their respective functions: the slow channel ensures macroscopic trajectory planning and energy efficiency optimization, while the fast channel realizes high-frequency disturbance suppression and accurate torque tracking. This architecture takes into account both the efficient use of computing resources and the multi-scale matching of control performance, and has good engineering feasibility. Attached Figure Description
[0027] This manual includes the following figures, which illustrate the following:
[0028] Figure 1 This is a block diagram of the overall hardware structure of the system of the present invention.
[0029] Figure 2 This is the core logic flowchart of the intelligent early warning method of the present invention.
[0030] Figure 3This invention provides a hardware-in-the-loop (HIL) simulation framework. Detailed Implementation
[0031] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, in order to help those skilled in the art to have a more complete, accurate and in-depth understanding of the concept and technical solutions of the present invention, and to facilitate its implementation.
[0032] like Figure 1 The diagram shows the overall architecture of the control system for predictive allocation of disturbances to ship motors according to the present invention. The invention consists of three main parts: a signal acquisition module, a core processing module, and a hardware-in-the-loop simulation interface module. The signal acquisition module includes GPS / IMU sensors, current sensors, voltage sensors, temperature sensors, and signal conditioning circuits connected to the outputs of each sensor. The GPS / IMU sensors monitor the ship's position and heading in real time. The current and voltage sensors monitor the bus voltage of the port and starboard battery packs and the phase current of the port and starboard motors, respectively. The temperature sensors monitor the motor winding temperature and the ambient temperature of the battery compartment. The analog signals output by each sensor are filtered, amplified, and level-converted by the signal conditioning circuits to adapt to the input range of the subsequent analog-to-digital converter. The core processing module includes a digital signal processor, a storage unit connected to the digital signal processor, and a pulse width modulation signal generation circuit connected to the output of the digital signal processor. The storage unit stores multi-rate cascaded hierarchical control programs, including executable code for line-of-sight guidance algorithms, nonlinear model predictive control thrust allocation algorithms, and extended state observer and super-twisted sliding mode composite disturbance rejection algorithms. The core processing module also includes an analog-to-digital converter (ADC) connected to a digital signal processor (DSP). The input of this ADC is connected to the output of a signal conditioning circuit, converting analog voltage signals into digital quantities for real-time calculation by the DSP. A PWM signal generation circuit generates a space vector pulse width modulation waveform based on the final torque control command, used to drive the inverter to control the permanent magnet synchronous motor. The hardware-in-the-loop (HIL) simulation interface module includes an FPGA chip, a digital-to-analog converter (DAC), and an analog-to-digital converter (ADC). One end of the interface module is connected to the output of the PWM signal generation circuit, and the other end is used to connect to the HIL simulation platform. Specifically, the DAC's input is connected to the output of the PWM signal generation circuit, converting digital control signals into analog voltage signals that are input to the virtual model of the ship's electric propulsion system running in the HIL simulation platform. The ADC's output is connected to the feedback input of the DSP, transmitting feedback signals obtained from real-time simulation on the HIL platform, such as motor speed, bus voltage, phase current, and ship position and heading, back to the DSP. The DSP updates the inputs of the hierarchical control calculation unit and the fault diagnosis unit based on the feedback signals, forming a complete closed-loop verification environment.
[0033] like Figure 2 As shown, the method of this invention is implemented based on the above system, and performs hierarchical collaborative fault-tolerant control through the core processing module. The overall control architecture is divided into three levels: guidance layer, thrust allocation layer, and motor execution layer. Each layer operates in cascade with different update cycles. The guidance layer operates at a preset cycle, such as 50ms, and is responsible for calculating the desired thrust and desired bow torque based on the ship's real-time attitude information and target course. The thrust allocation layer operates at a cycle of 20ms and is responsible for optimally allocating the desired commands from the guidance layer to the port and starboard motors. The motor execution layer operates at the fastest cycle, such as 100μs, and is responsible for accurately tracking the torque commands issued by the allocation layer and suppressing various high-frequency disturbances.
[0034] S1: The guidance layer employs a strategy combining an adaptive line-of-sight guidance algorithm with a backstepping virtual control law. For straight-line path tracking, the system first reads GPS data to obtain the ship's current coordinates, reads IMU data to obtain the current heading, and reads target route data to obtain the coordinates of the next waypoint, calculating the lateral trajectory error accordingly. Building upon traditional line-of-sight guidance, this invention adopts a forward-looking distance strategy that adaptively adjusts with ship speed, where the forward-looking distance Δ is calculated according to… Calculate, where Δ min For the minimum forward sight distance, Δ max Where k is the adaptive gain coefficient and U is the ship speed, and k is the maximum forward sight distance, U is the adaptive gain coefficient, and U is the ship speed. At lower ship speeds, the forward sight distance approaches Δ. max To ensure smooth maneuvering at low speeds, the forward sight distance approaches Δ at higher speeds. min To ensure the speed of high-speed tracking. The system then calculates the lateral trajectory error y. e Calculate the desired heading angle , where ψ LOS The line-of-sight guidance angle is given. A virtual control law is designed based on the backstepping method, and the heading tracking error e is defined. ψ =ψ-ψ d The desired bow turning angular velocity command r is calculated by combining the ship's three-degree-of-freedom kinematic model. d =-k1e ψ +ψ d Where k1 is the proportional feedback gain coefficient. The desired longitudinal thrust τ of the final output of the guidance layer. u,d and expected turning torque τ r,d It is transmitted to the thrust distribution layer.
[0035] S2: The thrust distribution layer uses the desired thrust and desired bow torque output from the guidance layer as tracking targets to construct a thrust distributor based on nonlinear model predictive control. For a twin-propeller ship, the port thrust T L and starboard thrust T R The mapping relationship between the resultant thrust and the turning moment is τ. u =T L +TR τ r =(T L -T R Let L be the distance from the thruster to the ship's center of gravity. The thrust allocation problem is transformed into a constrained convex optimization problem: minimizing a cost function that includes tracking error penalties, energy consumption penalties, and output smoothness penalties, while satisfying various constraints. The specific form of the cost function is as follows: ,in This is the actual output vector of the thrust distribution layer. The desired value vector output by the guidance layer. Let Δu be the thrust command vector for the port and starboard motors, Δu be the thrust rate of change vector, Q be the tracking accuracy weight matrix, R be the energy consumption weight matrix, S be the output smoothness weight matrix, and T be the output smoothness weight matrix. p For the prediction time domain, the cost function achieves a dynamic balance between tracking accuracy, energy efficiency, and output smoothness through the coordinated configuration of the three weight matrices Q, R, and S. Constraints include three categories: power constraints for independent dual-battery pack power supply, physical limit constraints for the propulsion motor, and fault reconfiguration constraints. Specifically, the power constraints for independent dual-battery pack power supply are... and , where ω L ω R These represent the rotational speeds of the port and starboard propellers, P L,max P R,max These represent the maximum permissible output power of the left and right battery packs, respectively. The physical limit constraints of the propulsion motors include the amplitude limits of the thrust output of each motor. and thrust rate of change limit When a single-side motor fault is detected, the fault reconfiguration constraint limits the thrust on the corresponding side to the product of the rated thrust and the remaining output coefficient. , where 0≤λ≤1. The thrust allocation layer solves this constrained quadratic programming problem in real time, automatically balances tracking accuracy and energy consumption, and automatically reconstructs the thrust allocation weights when a fault occurs, outputting the desired torque command of the port and starboard motors to the motor execution layer.
[0036] S3: After receiving the desired torque command from the thrust distribution layer, the motor execution layer uses i d The vector control strategy for the permanent magnet synchronous motor with a coefficient of 0 transforms the three-phase stationary coordinate system into a two-phase rotating coordinate system using the Clarke-Park transformation, achieving static decoupling of torque and flux linkage, making the electromagnetic torque proportional to the q-axis current. To achieve high-frequency suppression of disturbances, the motor execution layer integrates an extended state observer and a super-torsional sliding mode control law to form a composite disturbance rejection controller. The extended state observer treats the unknown dynamics in the motor's motion equations and external disturbances as a unified lumped disturbance, establishing an extended state-space model that includes the lumped disturbance states. Where x1 is the motor angular velocity, x2 is the derivative of the angular velocity, x3 is the lumped disturbance, u is the control input, b0 is the estimated control gain, and ξ(t) is the rate of change of the disturbance. The designed linear extended state observer is... Where z1, z2, and z3 are the observed values of x1, x2, and d, respectively, and β1, β2, and β3 are the observer gains, tuned using the bandwidth method. ω0 is the observer bandwidth. Z3 is the real-time estimate of the lumped disturbance.
[0037] S4: Based on the obtained total disturbance estimate, the motor execution layer further combines the super-torsion sliding mode control law to generate the feedforward compensation. Define the sliding surface. Where e = ω - ω* is the motor angular velocity tracking error, ω* is the desired angular velocity, and c is the sliding surface coefficient. A control law is designed using an exponential reaching law combined with the lumped disturbance estimate output by the extended state observer. Where k1 and k2 are the reaching law gain coefficients, and sat(s) is the saturation function, defined as sat(s) = s / |s| when |s|>ε, and sat(s) = s / ε when |s|≤ε, where ε is the boundary layer thickness. The selection of the sliding surface coefficient c, reaching law gain k1, and k2 satisfies the Lyapunov stability condition, ensuring that the system tracking error converges to zero within a finite time and that the closed-loop system is uniformly bounded. The final torque command is generated by superimposing the desired torque command and the feedforward compensation, and then converted into an SVPWM waveform by the PWM signal generation circuit to drive the inverter, achieving accurate torque tracking.
[0038] This invention also possesses complete fault diagnosis and adaptive reconfiguration capabilities. The system utilizes a Kalman filter to construct a motor state observer, calculating in real time the residual between the actual output and the model's predicted output. , where y k This is the actual output vector at time k. This is the model prediction output vector at time k. A hard fault is identified when the residual exceeds the adaptive threshold and the duration exceeds a preset time window, triggering an adaptive reconstruction mechanism based on an expert rule base. The reconstruction mechanism covers three typical fault modes: when the residual output coefficient λ is between 0.3 and 0.8, it is identified as a derating fault. The system adds a power compensation penalty term for the healthy motor in the NMPC cost function, forcing the quadratic programming solver to automatically increase the output of the healthy motor to compensate for the thrust loss on the faulty side; when the residual output coefficient λ=0, it is identified as a complete failure fault. The system sets the thrust on the faulty side to zero, estimates the required compensation rudder angle in real time through the extended state observer, and corrects the desired heading of the guidance layer to... The compensation angle T healthyThe system outputs power to the motor on the healthy side, where L is the distance from the thruster to the ship's center of gravity. The hydrodynamic torque generated by the rudder angle deflection works in conjunction with the thrust torque on the healthy side to maintain controllable heading. When a mechanical lock-up fault is detected, the system applies short-circuit braking to the motor on the faulty side to generate damping torque, and propulsion is carried out only by the motor on the healthy side, achieving towing mode operation.
[0039] The multi-rate cascaded architecture of this invention operates in coordination across three different time scales. The guidance layer's update cycle is set to 50ms, matching the GPS update rate. Its operations include reading GPS / IMU data to update the ship's position and heading, reading the target course, and executing an adaptive LOS algorithm to calculate the desired thrust and desired bow torque. The thrust allocation layer's update cycle is set to 20ms. Its operations include reading the desired value output from the guidance layer, reading the current battery SOC and bus voltage, constructing the QP cost function, and calling the solver to solve for the optimal thrust command. The motor execution layer's update cycle is set to 100μs, matching the 10kHz PWM switching frequency. Its operations include reading the torque command, reading the rotor position and speed feedback, running the ESO observer to update the lumped disturbance estimate, running the STSMC control law to calculate the feedforward compensation, and generating an SVPWM duty cycle signal to output to the inverter. These three levels form a complete cascaded relationship, with each fast and slow channel performing its specific function.
[0040] like Figure 3 As shown, in terms of hardware-in-the-loop simulation verification, the HIL simulation interface module of this invention supports a complete closed-loop verification process. First, a virtual model of the ship's electric propulsion system is built in the HIL simulation platform, including a three-degree-of-freedom dynamic model of the ship, a permanent magnet synchronous motor model, and a dual-battery pack model. The core control algorithm is compiled and downloaded to the DSP for execution. The DSP's PWM output is converted into an analog voltage signal by a digital-to-analog converter and input to the motor model interface of the HIL simulation platform. The HIL simulation platform drives the virtual motor according to the input analog voltage signal and calculates various electrical and kinematic quantities in real time. The simulation results are sampled by the analog-to-digital converter and fed back to the DSP's ADC input, forming a complete closed-loop control loop. During the simulation, various predefined fault modes can be injected through the fault injection interface to verify the system's control performance and fault tolerance. Thus, this invention fully realizes the entire implementation process from multi-source signal acquisition, nonlinear dynamic prediction, NMPC optimization solution, ESO-STSMC composite disturbance rejection control to PWM generation and HIL closed-loop verification.
[0041] The role and effect of the embodiments
[0042] This invention achieves high-precision thrust tracking and multi-objective collaborative optimization of ship propulsion systems under complex sea conditions by constructing a three-level, multi-rate cascaded hierarchical collaborative architecture consisting of a guidance layer, a thrust allocation layer, and a motor execution layer. Specifically, the guidance layer employs a line-of-sight guidance algorithm that adaptively adjusts the forward look-ahead distance based on ship speed, combined with a backstepping virtual control law to calculate the desired bow angular velocity command in real time. This allows for dynamic adjustment of guidance parameters based on the ship's real-time attitude and speed, effectively avoiding path tracking overshoot or oscillations caused by a fixed forward look-ahead distance during high-to-low speed switching, thus ensuring the smoothness of trajectory tracking. The thrust allocation layer integrates tracking error penalties, energy consumption penalties, and smoothness penalties into the cost function of nonlinear model predictive control, and uses a rolling optimization solution with an update cycle shorter than that of the guidance layer to achieve optimal thrust allocation, enabling the system to achieve high precision thrust tracking and multi-objective collaborative optimization under complex sea conditions. Under different operating conditions, the weight ratio between tracking accuracy and energy economy can be adaptively adjusted to minimize energy consumption throughout the flight and reduce mechanical wear of the propulsion motor while ensuring maneuverability. The motor execution layer operates with the shortest update cycle. The extended state observer treats wave torque impact, propeller water load drop, motor parameter temperature rise drift and unmodeled dynamics as lumped disturbances for real-time estimation. The estimated values are then fed forward to the super-torsional sliding mode control law to realize an active disturbance rejection mechanism of observation and compensation. This significantly reduces the system's dependence on accurate mathematical models and the complexity of maintenance and debugging.
[0043] This invention also achieves online identification and fault-tolerant recovery of the entire spectrum of faults through a built-in fault diagnosis unit and an adaptive reconfiguration mechanism driven by an expert rule base. The fault diagnosis unit uses a Kalman filter to continuously calculate the residual between the actual motor output and the model prediction. When the residual exceeds the adaptive threshold and continues to exceed the preset time window, the fault type is automatically determined. Differentiated reconfiguration strategies are triggered for three modes: derating operation, complete failure, and mechanical lock-up. During derating, a power compensation penalty term for the healthy side is added to the cost function to automatically increase the output of the healthy side. During complete failure, the rudder angle is estimated and the desired heading is corrected through an extended state observer to maintain course controllability. During mechanical lock-up, a short-circuit brake is applied to the motor on the faulty side and the healthy side propels the system alone. The synergistic effect of the above mechanisms enables the system to complete reconfiguration within a few hundred milliseconds after the complete failure of a single motor, maintaining the ship's course control and providing complete technical support for the propulsion system to meet the functional safety certification requirements of ISO 26262 ASIL C / D level.
[0044] Furthermore, the signal acquisition module, core processing module, and hardware-in-the-loop simulation interface module employed in this invention constitute a complete closed-loop verification hardware platform. The digital signal processor in the core processing module runs the hierarchical control program in real time, and the pulse width modulation signal generation circuit drives the inverter to control the permanent magnet synchronous motor according to the final torque command. At the same time, the hardware-in-the-loop simulation interface module connects the control signal to the virtual controlled object through the FPGA and digital-to-analog converter and sends the simulation feedback signal back to the digital signal processor. This enables the control algorithm to be fully verified and its parameters optimized in the virtual environment, ensuring that its control performance under different sea conditions, different battery charge states, and different fault modes can be evaluated in advance. This effectively reduces the risks and costs of actual ship testing, giving this invention good engineering feasibility and application value.
[0045] The present invention has been described above by way of example with reference to the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the above-described manner. Any non-substantial improvements made using the inventive concept and technical solution; or the direct application of the inventive concept and technical solution to other situations without modification, are all within the protection scope of the present invention.
Claims
1. A control system for predicting and allocating disturbances to ship motors, characterized in that, include: The system comprises a signal acquisition module, a core processing module, and a hardware-in-the-loop simulation interface module. The signal acquisition module includes a GPS / IMU sensor, a current sensor, a voltage sensor, a temperature sensor, and signal conditioning circuits connected to the outputs of each sensor. The core processing module includes a digital signal processor, a storage unit connected to the digital signal processor, and a pulse width modulation (PWM) signal generation circuit connected to the output of the digital signal processor. The hardware-in-the-loop simulation interface module includes an FPGA chip and a digital-to-analog converter. One end of the hardware-in-the-loop simulation interface module is connected to the output of the PWM signal generation circuit, and the other end is connected to the hardware-in-the-loop simulation platform.
2. The control system for predictive allocation of disturbance rejection of ship motors according to claim 1, characterized in that, The core processing module also includes an analog-to-digital converter connected to the digital signal processor. The input of the analog-to-digital converter is connected to the output of the signal conditioning circuit. The storage unit stores a multi-rate cascaded hierarchical control program.
3. The control system for predictive allocation of disturbance rejection of ship motors according to claim 1, characterized in that, The hardware-in-the-loop simulation interface module also includes an analog-to-digital converter (ADC), the output of which is connected to the feedback input of the digital signal processor, and the input of which is connected to the output of the pulse width modulation signal generation circuit.
4. The control method for a ship motor disturbance rejection prediction and allocation control system according to any one of claims 1-3, characterized in that, Includes the following steps: S1: Establish a multi-rate cascaded hierarchical collaborative fault-tolerant control architecture, which includes a guidance layer, a thrust distribution layer, and a motor execution layer; the guidance layer is based on the ship's nonlinear dynamics model and adaptive guidance algorithm, and generates the desired thrust and desired bow torque according to the ship's real-time attitude information; S2: The thrust distribution layer constructs a thrust distributor based on nonlinear model predictive control. The desired thrust and desired throttle torque are used as tracking targets. The power constraints of the dual battery packs independently powered, the physical limit constraints of the propulsion motors, and the fault reconfiguration constraints are all incorporated into the quadratic programming cost function. The optimal thrust distribution command is solved in real time through rolling optimization, and the desired torque command of each motor is output. S3: The motor execution layer uses an extended state observer to estimate the lumped disturbance in real time. Based on the estimated value, it generates a feedforward compensation amount by combining the super-twisted sliding mode control law. The desired torque command is then superimposed with the feedforward compensation amount to generate the final torque command, which drives the motor to run. S4: Real-time monitoring of motor operating status based on residual analysis. When a fault is detected, an adaptive reconstruction mechanism based on an expert rule base is triggered to adjust the thrust distribution strategy or guidance parameters to achieve fault-tolerant control.
5. The control method for predictive allocation of disturbance rejection of ship motors according to claim 4, characterized in that, The adaptive guidance algorithm employs a forward-looking distance strategy that adapts to ship speed, calculates the desired heading angle based on the lateral trajectory error, and designs a virtual control law based on the backstepping method to calculate the desired bow turning angular velocity command.
6. The control method for predictive allocation of disturbance rejection of ship motors according to claim 4, characterized in that, The quadratic programming cost function includes a tracking error penalty term, an energy consumption penalty term, and an output smoothness penalty term, which are adjusted by the corresponding weight matrix.
7. The control method for predictive allocation of disturbance rejection of ship motors according to claim 6, characterized in that, The power constraints for the independent power supply of the dual battery packs include the maximum allowable output power limit of each battery pack; the physical limit constraints for the propulsion motors include the amplitude limit of the thrust output of each motor and the thrust change rate limit; when a single-side motor fault is detected, the fault reconstruction constraint limits the thrust on the corresponding side to the product of the rated thrust and the remaining output coefficient.
8. The control method for predictive allocation of disturbance rejection of ship motors according to claim 4, characterized in that, The extended state observer treats the unknown dynamics in the motor motion equations and external disturbances as the lumped disturbance, establishes an extended state space model containing the lumped disturbance state for real-time observation, and outputs a real-time estimate of the lumped disturbance.
9. The control method for predictive allocation of disturbance rejection of ship motors according to claim 8, characterized in that, The super-twisted sliding mode control law defines the sliding surface based on the motor angular velocity tracking error, adopts an exponential approach law and combines the real-time estimate of the lumped disturbance for feedforward cancellation, and uses a saturation function to replace the sign function to reduce chattering.
10. The control method for predictive allocation of disturbance rejection of ship motors according to claim 4, characterized in that, The adaptive reconfiguration mechanism includes: when the fault remaining output coefficient is between 0.3 and 0.8, it is determined to be a derating operation fault, and the output of the healthy side motor is improved by adding a power compensation penalty term of the healthy side motor in the cost function; when the fault remaining output coefficient is zero, it is determined to be a complete failure fault, and the thrust of the fault side is set to zero. The required compensation rudder angle is estimated by the extended state observer and the desired heading is corrected to achieve joint rudder and propeller fault tolerance; when it is determined to be a mechanical lock-up fault, the fault side motor applies short-circuit braking to generate damping torque, and only the healthy side motor propels.
Citation Information
Patent Citations
Fault-tolerant control method for motor propulsion system under ship power grid fault
CN122339343A
