Electric ship power adjusting device based on self-adaptive control

Abstract

The invention discloses an electric ship power adjusting device based on self-adaptive control. The electric ship power adjusting device comprises a sensor module, a data processing module, a self-adaptive control module, an actuator driving module and a power output module. The sensor module collects state parameters, operation parameters and environment disturbance information, and the state parameters, the operation parameters and the environment disturbance information are input into the self-adaptive control module after being standardized by the data processing module. The self-adaptive control module processes disturbance, non-linear terms and actuator faults through cooperative work of built-in disturbance observation, unknown non-linear approximation, fault-tolerant control and preset performance constraint units, generates self-adaptive control instructions, converts the instructions into driving signals through the actuator driving module, controls the power output module to adjust power, rotating speed and energy distribution, and controls the power output module to output power. Dynamic power adaptation of the electric ship is achieved, and navigation stability is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of marine power control technology, specifically an electric marine power regulation device based on adaptive control. Background Technology

[0002] Electric ships are increasingly being used in inland waterway shipping, marine exploration and other fields due to their advantages of low pollution and high energy efficiency. Their power systems need to adapt to complex working conditions in real time. They need to cope with wave loads and sensor assembly misalignment in the same channel of propulsion and control in the marine environment, as well as handle problems such as partial failures and offset failures of actuators, in order to ensure stable power output.

[0003] Existing electric ship power regulation mostly adopts conventional PID control, basic disturbance observation unit or simple neural network, and some schemes combine preset fault compensation logic; for example, a single type of disturbance is estimated by traditional observation unit, or actuator faults are compensated by fixed parameters to adjust the propulsion motor power and energy storage distribution.

[0004] However, existing technologies have obvious limitations: they are difficult to process matched and unmatched disturbances simultaneously, resulting in limited anti-disturbance effects; the real-time performance of hydrodynamic damping and high-order wave coupling is poor; actuator fault identification lacks dynamic adaptation capabilities and compensation accuracy is insufficient; and the error convergence characteristics of power regulation are not constrained, which can easily cause power output fluctuations and affect the stability of ship navigation. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this invention is to provide an electric ship power regulation device based on adaptive control to solve the problems in the prior art, such as difficulty in synchronously handling matched and unmatched disturbances, poor real-time approximation of electric ship nonlinear terms and high-order coupling terms of disturbances, lack of dynamic adaptation capability and insufficient compensation accuracy of actuator fault identification, and power output fluctuations caused by unconstrained power regulation error convergence characteristics, which affect the stability of ship navigation.

[0006] To achieve the above objectives, embodiments of the present invention disclose an electric ship power regulation device based on adaptive control, comprising: Sensor module, data processing module, adaptive control module, actuator drive module, and power output module; The sensor module is used to collect the electric ship's state parameters, operating parameters, and environmental disturbance information in real time. The state parameters include rotor operating angle, angular velocity, and load input. The environmental disturbance information includes matched disturbances and unmatched disturbances. The data processing module is used to filter and normalize the state parameters, operating parameters and environmental disturbance information collected by the sensor module, and output standardized data. After receiving the standardized data, the adaptive control module generates adaptive control commands through the collaborative processing of the built-in disturbance observation unit, unknown nonlinear approximation unit, fault-tolerant control unit, and preset performance constraint unit. The actuator drive module converts the adaptive control command into a drive signal to control the operating state of the power output module; The power output module includes a propulsion motor and an energy storage management unit, which is used to adjust the output power, speed and energy distribution according to the drive signal to achieve dynamic adaptation of the electric ship's power.

[0007] Furthermore, the disturbance observation unit receives state parameters and environmental disturbance information from the standardized data, and constructs an observation model based on fractional calculus characteristics. The observation gain of the observation model satisfies the following constraints:

[0008]

[0009] in, This is the upper bound of the fractional derivative of the unmatched perturbation; To match the upper bound of the fractional derivative of the perturbation; For unmatched observation gain; To match the observation gain, the amplitude and trend of matched and unmatched disturbances are estimated in real time, and the disturbance estimates are output. Furthermore, the unknown nonlinear approximation unit receives the state parameters, operating parameters and disturbance estimates output by the disturbance observation unit from the standardized data, optimizes the output weights through the least squares method of the extreme learning machine network, approximates the nonlinear terms of the electric ship and the higher-order coupled nonlinear terms in the environmental disturbance in real time, and outputs the nonlinear approximation compensation amount.

[0010] Furthermore, the extreme learning machine network comprises an input layer, a hidden layer, and an output layer. The input layer receives state parameters, operating parameters, and disturbance estimates output by the disturbance observation unit from the standardized data, and maps them to the input signal of the hidden layer. The input weights and biases of the hidden layer are randomly initialized and then fixed. The Sigmoid activation function is used to extract features and perform nonlinear transformation on the input signal. The output layer optimizes the output weights using the least squares method and fuses the output signal of the hidden layer into a nonlinear approximation compensation quantity. This nonlinear approximation compensation quantity is used to offset the nonlinear terms of the electric ship and the higher-order coupled nonlinear terms in the environmental disturbance.

[0011] Furthermore, the fault-tolerant control unit receives the power output deviation of the operating parameters and status parameters in the standardized data, and identifies the fault type and fault parameters of the actuator online through a preset adaptive update law; the fault types include partial faults, bias faults and jamming faults, and generates a fault compensation amount based on the identified fault parameters to offset the power output amplitude attenuation or offset caused by the actuator fault.

[0012] Furthermore, the adaptive update law takes the power output deviation of the operating parameters and state parameters in the standardized data as input, and designs a two-parameter collaborative update rule for the actuator's fault factor and bias fault. The update law adjusts the parameter estimation rate by preset positive gain, and limits the estimated values ​​of the fault factor and bias fault by combining the projection operator. It identifies the fault type by real-time monitoring of the estimation results and changing trends of the two parameters. When the estimated value of the fault factor... The estimated value of bias fault ∈ (0,1) When the value approaches zero, it is determined to be a partial fault; when the estimated value of the fault factor is close to zero... The estimated value of bias fault ∈ (0,1] When the value deviates from zero, it is determined to be a bias fault; when the estimated value of the fault factor is zero... Estimated value of bias fault approaching zero When the value deviates from zero, it is determined to be a jamming fault, and the corresponding estimated fault parameter value is output.

[0013] Furthermore, the preset performance constraint unit receives a state tracking error, which is the difference between the state parameters in the standardized data and the preset reference state parameters for the electric ship's power regulation. This error is then used to construct a performance function.

[0014] Where t is the time variable; The dynamic constraint boundary for the tracking error of the i-th type of state; Let be the initial boundary for the tracking error of the i-th type of state; Let be the steady-state boundary of the tracking error for the i-th state; The convergence rate adjustment parameter is used for the tracking error of the i-th type of state; The preset convergence time is defined as the time for the i-th type of state tracking error; the performance function converges the state tracking error to a value within the preset convergence time. .

[0015] Furthermore, the energy storage management unit adjusts the charging and discharging power according to the adaptive control command. When the ship encounters strong disturbances or actuator failures, it prioritizes the power supply to the propulsion motor and limits the output current of the energy storage unit to a preset range of rated values.

[0016] Compared with the prior art, the beneficial effects of the present invention are: (1) This invention achieves coordinated and accurate processing of multiple disturbances and nonlinear terms. It adopts fractional-order finite-time disturbance observation units to synchronously estimate matched / unmatched disturbances, and combines extreme learning machine networks to approximate the nonlinear terms of electric ship dynamics and high-order coupling terms of disturbances, thus solving the problems of separation of disturbance resistance and nonlinear compensation and poor real-time performance in existing technologies.

[0017] (2) This invention uses a dual-parameter collaborative adaptive update law to identify some faults, bias faults, and jamming faults online and generate targeted compensation quantities, which solves the limitations of existing fixed parameter fault compensation, avoids parameter drift, ensures accurate correction of power output amplitude and offset under fault conditions, and improves system reliability.

[0018] (3) This invention integrates preset performance constraints into ship power regulation. By constructing a performance function to limit the convergence rate, overshoot and steady-state accuracy of state tracking error, it ensures that the error converges to a safe range within a preset time, solves the problem of power output fluctuation caused by the lack of error constraints in the prior art, and ensures the stability and dynamic adaptability of power output during ship navigation. Attached Figure Description

[0019] Figure 1 This is a structural flowchart of the device of the present invention.

[0020] Figure 2 This is a schematic diagram of the disturbance observation unit of the device of the present invention.

[0021] Figure 3 This is a diagram showing the effect of the extreme learning machine network approximation of the device of the present invention.

[0022] Figure 4 This is a diagram illustrating the effect of actuator fault compensation in the device of the present invention.

[0023] Figure 5 This is a diagram illustrating the preset performance constraints of the device of the present invention.

[0024] Figure 6 This is a diagram illustrating the current control effect of the energy storage unit in the device of the present invention. Detailed Implementation

[0025] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Please see Figure 1 , Figure 2 , Figure 3 , Figure 4 , Figure 5 and Figure 6 This application provides a detailed description of the technical solutions provided in each embodiment.

[0027] This application provides an adaptive control-based electric ship power regulation device, specifically including: like Figure 1 As shown, the device of this invention consists of five parts: a sensor module, a data processing module, an adaptive control module, an actuator drive module, and a power output module. It adopts a hardware architecture of distributed sensors and centralized control, and is adapted to the humid, vibration, and electromagnetic interference conditions of 200-500 ton inland waterway cargo ships or marine exploration vessels. The core hardware is integrated into the power control cabinet in the cockpit, and the various functional modules realize data interaction through an anti-interference CAN bus. The overall protection level of the device reaches IP65, which meets the environmental adaptability requirements of long-term ship navigation.

[0028] The overall physical structure and installation layout of the adaptive control-based electric ship power regulation device are as follows: The power control cabinet for the bridge is installed in the power control area at the rear of the ship's bridge, measuring 800mm × 600mm × 400mm. It features a built-in cooling fan and dust filter, and the cabinet is made of 304 stainless steel with pre-installed shock-absorbing pads at the bottom to prevent the impact of ship movement on the core hardware. The cabinet integrates a dual-core computing unit for the data processing module and adaptive control module, as well as a power amplifier circuit for the actuator drive module. All modules are connected via copper busbars and shielded cables, and the electromagnetic compatibility design complies with the IEC 60945 marine electrical standard.

[0029] The distributed sensors are functionally dispersed and installed in key parts of the ship, and all sensors are waterproof and sealed. Propulsion motor-related sensors: the photoelectric encoder E6B2-CWZ6C is mounted on the rotor shaft of the propulsion motor and fixed coaxially to the shaft via a coupling; the gyroscope sensor ADXRS650 is attached to the stator housing of the propulsion motor; the torque sensor is connected in series at the motor-propulsion shaft connection flange. Actuator-related sensors: the current sensor ACS712-20A and the voltage sensor LV25-P are integrated into the junction box at the actuator drive input end and fixed by snap-fit ​​installation. Environmental disturbance sensors: the pressure sensor MS5837-30BA and the tilt sensor SCA100T are jointly encapsulated in a titanium alloy waterproof housing, installed on the outer side of the hull 0.5m below the waterline, and fixed by welding with a stainless steel bracket. Sealed joints are used to protect the cables passing through the hull.

[0030] The power output module uses a YVP200L-4 permanent magnet synchronous motor for propulsion, which is horizontally mounted on a motor base in the middle of the ship's engine room and connected to the propeller shaft via a flexible coupling. The energy storage management unit consists of a lithium battery pack, a charge and discharge controller, and a current sensor, and is installed in the energy storage compartment at the rear of the engine room. The battery pack adopts a modular design with 20 cells connected in series and 5 modules connected in parallel, and a heat dissipation channel and an interface for fire extinguishing devices are reserved.

[0031] The core function of the sensor module is to provide accurate and real-time physical quantity data for subsequent control. The hardware selection and installation are strictly adapted to the dynamic navigation scenarios of electric ships. The sensor module converts the physical quantities of ship status, operation, and environmental disturbances into electrical signals through mechanical coupling, electromagnetic induction, piezoelectric effect, and MEMS technology. All sensor signal cables use shielded twisted-pair cables, with a length controlled within 5m. The cables are laid along the inside of the ship's keel to avoid strong interference sources such as engines. The sensor power supply uses isolated power supply modules to avoid power supply noise affecting the acquisition accuracy. The status parameter acquisition frequency is 1kHz, the operation parameter acquisition frequency is 2kHz, and the environmental disturbance information acquisition frequency is 500Hz to ensure coverage of the changing frequencies of dynamic disturbances such as wave load and hull tilt.

[0032] The data processing module employs a dedicated chip and hardware circuit architecture. The core chip is the STM32H743VIT6, paired with an FPGA for parallel filtering. It includes a built-in 16-channel ADC (for receiving analog signals) and an 8-channel UART interface for receiving digital signals. Iterative calculations are implemented through FPGA hardware logic to suppress noise in angular velocity, matched, and unmatched disturbance signals. A resistor-capacitor filter circuit performs low-pass filtering on the actuator voltage signal, with a cutoff frequency of 159Hz, effectively reducing steady-state errors caused by voltage ripple. Hardware multipliers and adders linearly map the filtered signal to the [-1,1] interval; the mapping formula is embedded in the hardware logic, providing standardized input for the adaptive control module. Standardized data is transmitted to the adaptive control module via a CAN FD bus. The bus interface uses a TJA1057 chip, and the bus terminals are matched with 120Ω resistors to ensure reliable data transmission in the complex electromagnetic environment of a ship.

[0033] The adaptive control module adopts a dual-core hardware architecture of DSP and FPGA to achieve hardware acceleration and parallel processing of the algorithm, ensuring the real-time performance and accuracy of control commands. The main control chip uses a TITMS320F28335 DSP, which is responsible for the operation of core algorithms such as disturbance observation, nonlinear approximation, fault-tolerant control, and preset performance constraints. The auxiliary processing chip uses a Xilinx XC7A35T FPGA, which is responsible for instruction fusion, PWM signal generation, and hardware protection logic implementation. The storage unit has an external 256MB DDR3 memory and 16MB Flash. The interface circuit uses two CAN FD interfaces, one Ethernet interface for debugging and data upload, and four analog output interfaces for backup.

[0034] The built-in disturbance observation unit is based on a fractional calculus-based observation model implemented by the floating-point unit of a DSP. The constraints of the observation gain are fixed by hardware logic, and it can output the estimated values ​​of matched and unmatched disturbances in real time.

[0035] The built-in disturbance observation unit receives the state parameters and environmental disturbance information from the standardized data, constructs an observation model based on fractional calculus characteristics, and the observation gain satisfies the following constraints:

[0036]

[0037] in, The upper bound of the fractional derivative of the unmatched disturbance is based on actual engineering test data of hull tilting disturbance, which quantifies the maximum rate of change of this type of disturbance. To match the upper bound of the fractional derivative of the disturbance, the limit range of dynamic changes of this type of disturbance was clarified based on the measured statistical results derived from wave-loaded disturbances; g is the unmatched observation gain. To match the observation gain, the core function of the constraint formula is to ensure that the observation gain is large enough to track the dynamics of the disturbance in real time by quantifying the maximum rate of change of the disturbance, while avoiding observation lag or error accumulation caused by insufficient gain. This is achieved by substituting... and The engineering measured parameters were used to calculate the results. and This result provides a minimum threshold for selecting the observation gain. Based on this observation model and gain constraints, the disturbance observation unit can synchronously and with high accuracy estimate the amplitude and variation trend of matched and unmatched disturbances in real time, where the estimated value of the matched disturbance is... The maximum estimation error is ≤0.008 N·m, and the mismatched disturbance estimate is... The maximum estimation error is ≤0.01 N·m, which solves the technical defect of existing technologies that are difficult to process two types of disturbances simultaneously; and The data is transmitted to an unknown nonlinear approximation unit, which helps the unit accurately capture the nonlinear terms of the electric ship and the higher-order coupling terms of the disturbance, thus improving the accuracy of the nonlinear approximation compensation; at the same time, and The data is transmitted to the fault-tolerant control unit, which can effectively eliminate the interference of disturbances on actuator fault identification, ensure the accuracy of fault identification, and lay the core foundation for the anti-disturbance performance and control stability of the entire power regulation system.

[0038] like Figure 2 The image shows the ELM nonlinear approximation effect. Figure 2 The graph contains two subgraphs, which verify the accuracy of the perturbation estimation; Figure 1 To match the observed disturbances, the solid blue line represents the actual matching disturbance, expressed as noise of 0.3sin(2πt / 3) + 0.02, and the dashed red line represents the observed matching disturbance. The observed values ​​have a 50ms delay, and are superimposed with 0.008 N·m of random noise. The maximum estimation error is less than or equal to 0.008 N·m, achieving accurate tracking of wave-like matching disturbances. Figure 2 For the observation of mismatched disturbances, the solid blue line represents the actual mismatched disturbance, and the dashed red line represents the observed mismatched disturbance. A 5% overshoot occurs between t=2.05-2.25s, marked by yellow shading. The maximum estimation error is less than or equal to 0.01 N·m, which satisfies the requirement of real-time estimation of disturbance amplitude and trend.

[0039] The unknown nonlinear approximation unit is hardware-accelerated using the input, hidden, and output layers of an Extreme Learning Machine (ELM) network via an FPGA's DSP48E1 unit. The sigmoid activation function in the hidden layer is fixed using a lookup table, and the least-squares optimization of the output weights is implemented using the DSP. This unknown nonlinear approximation unit receives state parameters, operating parameters, and disturbance estimates from the disturbance observation unit in standardized data. It then performs nonlinear approximation through the ELM network, but not through iterative iteration of the disturbance estimates. The disturbance estimates are only one of the network's real-time input features, synchronously input to the network along with other operating parameters. Their role is to reflect the instantaneous state of the current environmental disturbance, continuously updated as the ship's navigation conditions change, but not repeatedly iterated or corrected. The core of the ELM network's approximation lies in fixing the hidden layer parameters and performing a one-time weight optimization. The weights and biases from the input layer to the hidden layer are fixed after random initialization. The hidden layer performs instantaneous feature extraction and nonlinear transformation of the input signal using the sigmoid activation function, while the output layer directly optimizes the output weights using the least-squares method in one step, without iterative processes. Finally, the results are fused to generate the nonlinear approximation compensation quantity Δf.

[0040] The network structure of the extreme learning machine network includes an input layer, a hidden layer, and an output layer; the input layer receives rotor operating angle, angular velocity, actuator current, actuator voltage, and estimated matching disturbance value. Unmatched perturbation estimates The input signal is mapped to the hidden layer. The hidden layer input weights and biases are randomly initialized and then fixed, and feature extraction is performed using the Sigmoid activation function. The output layer optimizes the output weights using the least squares method, fusing them into a nonlinear approximation compensation quantity Δf. This quantity approximates the nonlinear terms of the electric ship and the higher-order coupled nonlinear terms in environmental disturbances in real time. The nonlinear approximation compensation quantity Δf is transmitted to the control command fusion stage to counteract the adverse effects of the nonlinear terms and higher-order coupled nonlinear terms on power regulation.

[0041] like Figure 3 The image shows the approximation effect of the Extreme Learning Machine network. Figure 3 It contains two subplots to verify the approximation error effect; Figure 1 To compare the nonlinear term with the approximation value, the solid blue line represents the actual nonlinear term, expressed as 0.2sin(2πt / 2) + 0.15tanh(3t) + 0.05t0.8, while the dashed orange line represents the approximation value from the ELM network. A slight, discernible discrepancy exists between the two, reflecting the approximation characteristics in practical engineering, rather than an ideal fit. Figure 2 The green solid line represents the relative approximation error, while the red solid line represents the upper limit of the error, which is 3%. The maximum error is less than or equal to 2.95%, and a nonlinear approximation compensation quantity Δf is generated through an extreme learning machine network.

[0042] The fault-tolerant control unit employs a dual-parameter estimation method based on an adaptive update law, implemented via a DSP interrupt service routine. The projection operator limits the parameter range through hardware logic, and fault identification results are fed back to the status indicator light in real time via a GPIO interface. The fault-tolerant control unit receives the power output deviation of the operating parameters and status parameters from standardized data and uses an adaptive update law to achieve fault identification and compensation. The adaptive update law takes the power output deviation of the operating parameters and state parameters in the standardized data as input, and designs a two-parameter collaborative update rule for the actuator's fault factors and bias faults; the update law estimates the rate by adjusting the preset positive gain parameter, and the preset positive gain is based on the fault update factor. and bias fault update gain First, a lower bound needs to be determined based on Lyapunov stability theory to ensure asymptotic stability of the system. Then, a preliminary range is obtained through MATLAB / Simulink simulation optimization. Finally, the optimal value is determined by hardware platform experiments, satisfying the conditions of estimation error ≤0.05 and no oscillatory overshoot. At the same time, the update law combined with the projection operator limits the estimated value of the fault factor. Estimates of ∈(0,1) and bias faults ∈[-2V,2V], where The interval is defined based on the physical meaning of actuator power output attenuation. For completely normal Completely ineffective. The range is set based on the rated bias error and withstand voltage limit of hardware such as the IGBT driver module. Fault types are identified by real-time monitoring of the estimated results and trends of these two parameters. When the estimated value of the fault factor... The estimated value of bias fault ∈ (0,1) When the value approaches zero, it is determined to be a partial fault; when the estimated value of the fault factor is close to zero... The estimated value of bias fault ∈ (0,1] When the value deviates from zero, it is determined to be a bias fault; when the estimated value of the fault factor is zero... Estimated value of bias fault approaching zero When the value deviates from zero, it is determined to be a jamming fault, and the corresponding estimated fault parameter value is output.

[0043] The fault identification is based on a dual-parameter collaborative update rule, according to the partial fault, bias fault, and jamming fault. The two parameters include a failure factor. and bias fault The fault factor The update law is:

[0044] in, Update the gain for the fault factor; The time update rate for the fault factor estimates; This refers to the angular velocity tracking error. This is an estimate of the failure factor; Here, c is the slider function, and c is the switching gain. This is the projection operator.

[0045] The bias fault The update law is:

[0046] in, Bias fault update gain; Time update rate of bias fault estimates; This is an estimate of the bias fault. This refers to the actuator current deviation.

[0047] Limited by the projection operator and The logic determines the type of fault. The fault compensation is based on the identified fault parameters. and Generate a fault compensation amount Δτ to offset the attenuation or offset of power output amplitude caused by actuator failure; output error constraint correction signal to the control command fusion stage.

[0048] like Figure 4 The image shows the effect of actuator fault compensation, with a scenario of 1.2V bias fault in the actuator at t=2s. Figure 4 It contains two sub-graphs to verify the effect of actuator fault compensation; Figure 1 For torque output comparison, the solid black line represents the commanded torque, the dashed red line represents the output amplitude deviation without compensation faults, and the solid green line represents the output after fault-tolerant compensation. If all three values ​​coincide before t=2s, there is no fault. After t=2s, the uncompensated output deviates from the command by 8 N·m. After compensation, the output returns to near the command, achieving dynamic compensation. Figure 2 For torque deviation comparison, the red dashed line represents the uncompensated deviation, and the green solid line represents the deviation after compensation. The red dashed line represents the upper limit of compensation accuracy of 0.6 Nd·m; the deviation after compensation is less than or equal to 0.55 Nd·m, which meets the technical requirements for offsetting power output offset and online identification of fault types and parameters by the adaptive update law.

[0049] The performance constraint unit calculates the performance function in parallel using the FPGA's arithmetic logic unit. Real-time comparison of the state tracking error is performed using a hardware comparator. When the error exceeds the constraint boundary, an emergency adjustment interrupt is immediately triggered on the DSP. The preset performance constraint unit receives the state tracking error and the difference between the state parameters in the standardized data and the reference state parameters. It then constructs a performance function to implement error constraints.

[0050] Where t is a time variable, corresponding to the real-time time progress of the ship's navigation, ensuring that the constraint logic is synchronized with the actual working conditions; It serves as the dynamic constraint boundary for the i-th type of state tracking error. Its core feature is that it changes dynamically over time rather than being a fixed value, which can adapt to the error characteristics of the entire process of a ship from startup, sudden changes in operating conditions to steady-state navigation. The initial boundary for the i-th type of state tracking error corresponds to scenarios where errors are prone to be large, such as the initial stage of ship startup and the instant of working condition switching. It needs to be set to a relatively loose initial threshold to avoid the initial stage error exceeding the constraint range and causing the system to adjust incorrectly. The steady-state boundary of the tracking error of the i-th type of state is the maximum allowable error limit when the ship is sailing stably. This value is determined based on engineering indicators such as the ship's sailing stability requirements and the control accuracy of the propulsion motor, and directly determines the steady-state accuracy of the power output. Let be the convergence rate adjustment parameter for the tracking error of the i-th type of state. Physically, it controls how quickly the error converges to the steady-state boundary. The larger the value, the faster the convergence rate, but it is necessary to avoid excessive values ​​that could cause system oscillations. The value should be optimized in conjunction with the response speed of the dynamic system. The preset convergence time for the tracking error of the i-th type of state is a key indicator set according to the requirements of ship navigation safety and real-time power adaptation. It clearly requires that the error converge to the steady-state range within this time to ensure the timeliness of power adjustment. The performance function, in physical terms, constructs a dynamic constraint boundary that is initially loose and gradually tightens through exponential decay characteristics. At the initial time t=0, This design adapts to the actual situation of large errors during the startup phase. As time progresses, the error smoothly converges towards the steady-state boundary, avoiding the rigid limitation of dynamic errors by fixed boundaries and solving the adjustment fluctuation problem caused by the lack of error constraints in existing technologies. Its technical effects are significant: firstly, through flexible parameter configuration, it can be specifically adapted to the adjustment needs of different state parameters, possessing strong versatility; secondly, combined with measured data, this function can ensure that the state tracking error converges within the preset time. Inner precise convergence to Within the specified range, the dynamic constraint boundary is never exceeded at any time, effectively suppressing fluctuations in power output and ensuring the stability of electric ships under complex operating conditions. At the same time, the clear error convergence characteristics provide a clear basis for the generation of instructions by the adaptive control module, avoiding control parameter drift caused by unconstrained errors, and further improving the dynamic adaptability and control accuracy of the entire power regulation system.

[0051] like Figure 5 The image shows the effect of preset performance constraints. Taking the rotor operating angle state tracking error as an example, the error constraint capability is verified. The red dashed lines represent the upper and lower boundaries of the performance function, and the expression is: The solid blue line represents the actual state tracking error; the orange vertical line represents the preset convergence time. The green dashed line represents ; The actual error converges to [-0.5°, 0.5°] at t=2s, the overshoot is less than or equal to 4.8%, and there is no moment when it exceeds the boundary of the performance function, which meets the technical requirement of converging to the steady-state boundary within the preset convergence time.

[0052] Substituting the nonlinear approximation compensation amount Δf, fault compensation amount Δτ, and error constraint correction signal output from the above four units into the sliding mode control law, adaptive control commands are generated. :

[0053] in, This refers to the angular velocity tracking error. This is the derivative of the angular velocity tracking error; This is the proportionality coefficient; These are the differential coefficients; It is a saturation function, limited to 0~50 N·m to avoid torque overload; finally, it is converted into a PWM instruction and transmitted to the actuator drive module.

[0054] The actuator drive module serves as a bridge connecting control commands and power output. Its core function is to convert adaptive control commands into drive signals for the propulsion motor, while also possessing a comprehensive hardware protection mechanism. The hardware utilizes IGBT power modules paired with an IR2110 driver chip, and includes built-in overcurrent, overvoltage, and overheat detection circuits. Mounted on a heat dissipation plate within the power control cabinet, it is equipped with an independent cooling fan. Thermal grease is applied between the IGBT module and the heat sink to ensure stable operation in the high-temperature environment of the ship.

[0055] The power output module includes a propulsion motor and an energy storage management unit, which is used to adjust the output power, speed and energy distribution according to the drive signal to achieve dynamic adaptation of the electric ship's power.

[0056] The propulsion motor is a YVP200L-4 permanent magnet synchronous motor with a built-in 16-bit rotary transformer, forming a closed-loop speed control. Based on the drive signal output from the actuator drive module, it adjusts the output power and speed, accurately converting torque commands into actual power. Figure 3 This provides an implementation basis for the implementation of fault compensation in China.

[0057] The energy storage management unit includes charge / discharge regulation, priority protection, and current limiting. The charge / discharge regulation receives adaptive control commands and adjusts the charge / discharge power. The priority protection ensures that when the ship encounters strong disturbances or actuator failures, unnecessary loads are cut off, prioritizing the power supply to the propulsion motor. The current limiting restricts the output current of the energy storage unit within a preset range of its rated value, specifically 1.2 times the rated current of 80A, to prevent overcurrent damage. The energy storage management unit feeds back the battery status to the sensor module, forming an energy closed loop and providing a basis for the adaptive control module to adjust energy distribution.

[0058] like Figure 6The diagram shows the current control effect of the energy storage unit. A strong disturbance combined with an actuator failure was used as the test conditions to verify the function of the energy storage management unit. The solid green line represents the rated current of the energy storage unit (80A), and the dashed red line represents the safe current limit of 96A. The solid blue line represents the actual output current. 0-3s represents normal operating conditions, with the current fluctuating between 60-80A, meeting conventional power requirements. 3-5s represents a strong disturbance, with the current rising to 92A, prioritizing power supply to the propulsion motor. 7-10s represents an actuator failure, with the current stabilizing at 88A, never exceeding the safe limit. This meets the technical requirements of prioritizing the propulsion motor and limiting the output current under strong disturbances or failures.

[0059] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An electric ship power regulation device based on adaptive control, characterized in that, include: Sensor module, data processing module, adaptive control module, actuator drive module, and power output module; The sensor module is used to collect the electric ship's state parameters, operating parameters, and environmental disturbance information in real time. The state parameters include rotor operating angle, angular velocity, and load input. The environmental disturbance information includes matched disturbances and unmatched disturbances. The data processing module is used to filter and normalize the state parameters, operating parameters and environmental disturbance information collected by the sensor module, and output standardized data. After receiving the standardized data, the adaptive control module generates adaptive control commands through the collaborative processing of the built-in disturbance observation unit, unknown nonlinear approximation unit, fault-tolerant control unit, and preset performance constraint unit. The actuator drive module converts the adaptive control command into a drive signal to control the operating state of the power output module; The power output module includes a propulsion motor and an energy storage management unit, which is used to adjust the output power, speed and energy distribution according to the drive signal to achieve dynamic adaptation of the electric ship's power.

2. The electric ship power regulation device based on adaptive control according to claim 1, characterized in that, The disturbance observation unit receives state parameters and environmental disturbance information from the standardized data, and constructs an observation model based on fractional calculus characteristics. The observation gain of the observation model satisfies the following constraints: in, This is the upper bound of the fractional derivative of the unmatched perturbation; To match the upper bound of the perturbation fractional derivative; For unmatched observation gain; To match the observation gain, the amplitude and trend of matched and unmatched disturbances are estimated in real time, and the disturbance estimates are output.

3. The electric ship power regulation device based on adaptive control according to claim 1, characterized in that, The unknown nonlinear approximation unit receives the state parameters, operating parameters and disturbance estimates output by the disturbance observation unit from the standardized data, optimizes the output weights through the least squares method of the extreme learning machine network, approximates the nonlinear terms of the electric ship and the higher-order coupled nonlinear terms in the environmental disturbance in real time, and outputs the nonlinear approximation compensation amount.

4. The electric ship power regulation device based on adaptive control according to claim 3, characterized in that, The extreme learning machine network comprises an input layer, a hidden layer, and an output layer. The input layer receives state parameters, operating parameters, and disturbance estimates output by the disturbance observation unit from the standardized data, and maps them to the input signal of the hidden layer. The input weights and biases of the hidden layer are randomly initialized and then fixed. The Sigmoid activation function is used to extract features and perform nonlinear transformation on the input signal. The output layer optimizes the output weights using the least squares method and fuses the output signal of the hidden layer into a nonlinear approximation compensation quantity. This nonlinear approximation compensation quantity is used to offset the nonlinear terms of the electric ship and the higher-order coupled nonlinear terms in the environmental disturbance.

5. The electric ship power regulation device based on adaptive control according to claim 1, characterized in that, The fault-tolerant control unit receives the power output deviation of the operating parameters and status parameters in the standardized data, and identifies the fault type and fault parameters of the actuator online through a preset adaptive update law. The fault types include partial faults, bias faults and jamming faults, and generates a fault compensation amount based on the identified fault parameters to offset the power output amplitude attenuation or offset caused by the actuator fault.

6. The electric ship power regulation device based on adaptive control according to claim 5, characterized in that, The adaptive update law takes the power output deviation of the operating parameters and state parameters in the standardized data as input, and designs a two-parameter collaborative update rule for the actuator's fault factor and bias fault. The update law adjusts the parameter estimation rate by preset positive gain, and limits the estimated values ​​of the fault factor and bias fault by combining the projection operator. It identifies the fault type by real-time monitoring of the estimation results and changing trends of the two parameters. When the estimated value of the fault factor... The estimated value of bias fault ∈ (0,1) When the value approaches zero, it is determined to be a partial fault; when the estimated value of the fault factor is close to zero... The estimated value of bias fault ∈ (0,1] When the value deviates from zero, it is determined to be a bias fault; when the estimated value of the fault factor is zero... Estimated value of bias fault approaching zero When the value deviates from zero, it is determined to be a jamming fault, and the corresponding estimated fault parameter value is output.

7. The electric ship power regulation device based on adaptive control according to claim 1, characterized in that, The preset performance constraint unit receives the state tracking error, which is the difference between the state parameters in the standardized data and the preset reference state parameters of the electric ship's power regulation. A performance function is constructed based on this error. Where t is the time variable; The dynamic constraint boundary for the tracking error of the i-th type of state; Let be the initial boundary for the tracking error of the i-th type of state; Let be the steady-state boundary of the tracking error for the i-th state; The convergence rate adjustment parameter is used for the tracking error of the i-th type of state; The preset convergence time is defined as the time for the i-th type of state tracking error; the performance function converges the state tracking error to a value within the preset convergence time. .

8. The electric ship power regulation device based on adaptive control according to claim 1, characterized in that, The energy storage management unit adjusts the charging and discharging power according to the adaptive control command. When the ship encounters strong disturbances or actuator failures, it prioritizes the power supply to the propulsion motor and limits the output current of the energy storage unit to a preset range of rated values.

Images