Start control system and method for main and auxiliary power hybrid ship

Abstract

The invention discloses a starting control system and method for a main and auxiliary power hybrid ship, and relates to the field of ship power control, and the system comprises a multi-source data collection module which is used for obtaining a multi-source state data set of a ship power system; the fusion processing module is used for performing feature alignment processing on the multi-source state data set based on a multi-modal sensing data fusion algorithm to generate a hybrid power starting parameter set; the dynamic weight distribution module is used for constructing a dynamic weight distribution matrix according to the hybrid power starting parameter set; the cooperative control module is used for generating a main and auxiliary engine combined starting instruction set based on an improved cooperative control algorithm; and the execution and monitoring module is used for executing the main and auxiliary engine combined starting instruction set in a preset time window and monitoring the transient response characteristic of the propulsion system. According to the starting control system and method for the main and auxiliary power hybrid ship, the starting stability and the energy utilization efficiency of the hybrid ship are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of ship power control, and in particular to a start-up control system and method for a main-auxiliary power hybrid ship. Background Art

[0002] The ship power system is the core unit of ship operation, which realizes energy conversion and power output through key equipment such as the main engine, auxiliary engine, shaft system, etc. Existing ship power control technology is mainly based on PID controller and physical model predictive control method, which realizes power output control by adjusting parameters such as fuel injection amount and turbocharging pressure.

[0003] However, traditional methods have significant limitations in the coordinated control of multiple power sources: first, PID control is difficult to adapt to the strong coupling and nonlinear characteristics of the hybrid power system, resulting in response lag and overshoot problems; second, physical model prediction relies on precise mechanism modeling, which is prone to parameter mismatch under dynamic sea conditions and load mutation scenarios, causing torque fluctuations and energy loss. Summary of the invention

[0004] In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a starting control system and method for a main-auxiliary hybrid ship.

[0005] In a first aspect, the present invention provides a start-up control system for a main-auxiliary hybrid ship, the system comprising:

[0006] A multi-source data acquisition module is used to obtain a multi-source state data set of the ship power system, which includes the pressure waveform in the main engine cylinder, the turbine speed timing data, the battery pack temperature gradient distribution and the propeller load spectrum;

[0007] A fusion processing module is used to perform feature alignment processing on a multi-source state data set based on a multi-modal perception data fusion algorithm to generate a hybrid power starting parameter set including a power demand level, an energy allocation weight factor, and an equipment health index;

[0008] Dynamic weight allocation module, used to construct a dynamic weight allocation matrix according to the hybrid power startup parameter set, and calculate the host fuel injection amount correction coefficient, battery pack output power adjustment coefficient and shaft torque smoothing parameter through a double-layer fuzzy inference engine;

[0009] A collaborative control module is used to generate a main and auxiliary engine joint start instruction set based on an improved collaborative control algorithm. The instruction set includes a phased fuel injection pulse width, a turbocharger phase compensation angle, a battery pack multi-stage discharge curve, and a shaft system inertia compensation value;

[0010] The execution and monitoring module is used to execute the main and auxiliary engine joint start-up instruction set within a preset time window and monitor the transient response characteristics of the propulsion system. When the actual torque fluctuation rate exceeds a first preset threshold, a torque rebalancing mechanism based on dynamic inertia compensation is triggered.

[0011] Preferably, the fusion processing module includes:

[0012] Wavelet packet decomposition module, used to perform wavelet packet decomposition on the pressure waveform in the main engine cylinder to extract the low-frequency energy proportion and high-frequency impact factor;

[0013] The harmonic distortion analysis module uses sliding window Fourier transform to analyze the harmonic distortion rate of turbine speed time series data;

[0014] Thermal conduction modeling module, used to establish a two-dimensional thermal conduction model of the temperature gradient distribution of the battery pack and calculate the maximum allowable discharge rate under the maximum temperature difference constraint;

[0015] A convolutional neural network module to identify transient shock signature patterns in the thruster load spectra;

[0016] The dynamic weight update module is used to update the weight coefficient of the dynamic weight allocation model according to the following formula:

[0017] W_t=α·e^{-β·ΔT}+(1-α)·tanh(γ·(P_avg / P_max))

[0018] Among them, α is the temperature sensitivity coefficient, β is the thermal attenuation factor, γ is the power regulation gain, ΔT is the current maximum temperature difference of the battery pack, P_avg is the average load power of the thruster, and P_max is the maximum allowable output power of the battery pack.

[0019] Preferably, the dynamic weight allocation module includes:

[0020] A three-dimensional weighted space building module defines three optimization dimensions: power response priority, energy efficiency optimization level, and equipment protection level;

[0021] The fuzzy membership calculation module uses trapezoidal membership functions to process boundary conditions in three dimensions;

[0022] The collaborative game optimization module is used to solve the optimal weight combination through the following objective function:

[0023] min(λ1·τ_acc+λ2·E_loss+λ3·W_deg)

[0024] Wherein: τ_acc is the startup time delay, E_loss is the energy loss rate, W_deg is the equipment wear degree, λ1 is the first weight coefficient, λ2 is the second weight coefficient, and λ3 is the third weight coefficient.

[0025] Preferably, the collaborative control module includes:

[0026] The power coupling modeling module is used to establish the main and auxiliary power coupling model based on the shaft system torque transfer function. The shaft system torque transfer function is:

[0027] T_out=K1·(P_engine+P_battery)-K2·dω / dt

[0028] Wherein, T_out is the shaft output torque, P_engine is the host output power, P_battery is the battery pack output power, dω / dt is the shaft angular acceleration change rate, K1 is the mechanical transmission efficiency factor, and K2 is the rotation inertia compensation coefficient;

[0029] The prediction-correction control module is used to adopt a dual-loop control strategy, in which the inner loop controls the proportional relationship between the fuel injection amount and the battery discharge rate, and the outer loop adjusts the coordinated parameters of the turbocharging phase and the shaft inertia compensation.

[0030] Preferably, the system further comprises:

[0031] Adaptive learning module, used to achieve dynamic optimization of control parameters through the following steps:

[0032] Record actual control effect data sets during each startup process, including fuel consumption rate, torque fluctuation rate, battery discharge efficiency and shaft vibration amplitude;

[0033] Construct a parameter optimization model based on deep reinforcement learning, input the actual control effect data set and the current working condition feature vector, and output the adjustment coefficient of the dynamic weight allocation matrix;

[0034] Update the fuzzy rule base of the double-layer fuzzy inference engine, combine the historical fuzzy rule confidence and the gradient of the performance loss function to generate new fuzzy rule confidence;

[0035] The optimized parameters are deployed to the collaborative control module to realize online iteration of the control strategy.

[0036] Preferably, the system further comprises:

[0037] The vibration suppression module is a vibration acceleration sensor array deployed at the key nodes of the shaft system of the hybrid ship, which is used to collect the three-dimensional vibration spectrum; when the amplitude of the characteristic frequency component is detected to exceed the second preset threshold, the shaft system torsional vibration suppression mechanism is activated, and the mechanism includes: adjusting the fuel injection interval phase angle, injecting reverse offset torque pulses, and dynamically adjusting the battery pack output power fluctuation component.

[0038] In a second aspect, a startup control method for a main-auxiliary hybrid ship includes:

[0039] Obtain a multi-source state data set of the ship power system, where the multi-source state data set includes the in-cylinder pressure waveform of the main engine, the turbine speed time-series data, the temperature gradient distribution of the battery pack, and the propeller load spectrum diagram;

[0040] Based on the multi-modal perception data fusion algorithm, perform feature alignment processing on the multi-source state data set to generate a hybrid power start parameter set including the power demand level, the energy distribution weight factor, and the equipment health index;

[0041] Construct a dynamic weight allocation matrix according to the hybrid power start parameter set, and calculate the main engine fuel injection quantity correction coefficient, the battery pack output power adjustment coefficient, and the shafting torque smoothing parameter through a double-layer fuzzy inference engine;

[0042] Generate a combined main and auxiliary engine start instruction set based on the improved cooperative control algorithm, where the instruction set includes the staged fuel injection pulse width, the turbocharging phase compensation angle, the multi-level discharge curve of the battery pack, and the shafting inertia compensation value;

[0043] Execute the combined main and auxiliary engine start instruction set within a preset time window, and monitor the transient response characteristics of the propulsion system. When the actual torque volatility exceeds the first preset threshold, trigger a torque rebalancing mechanism based on dynamic inertia compensation.

[0044] Compared with the prior art, the present invention has the following characteristics and beneficial effects:

[0045] First, the multi-source data acquisition module is used to obtain the pressure waveform in the main engine cylinder, the turbine speed timing data, the battery pack temperature gradient distribution and the propeller load spectrum in real time, which solves the limitation of traditional ship power control relying on single sensor data, ensures the comprehensive perception of mechanical state, thermodynamic characteristics and load dynamics, and provides a high-precision data basis for multi-power source collaborative control. Secondly, based on the multi-modal perception data fusion algorithm of the fusion processing module, feature alignment and deep correlation analysis are performed on heterogeneous data to generate power demand level, energy allocation weight factor and equipment health index, breaking through the static parameter assumption of traditional physical models and realizing dynamic decision optimization under complex working conditions. Furthermore, a three-dimensional weight space is constructed through the dynamic weight allocation module to achieve the global optimal solution between the start time delay, energy loss rate and equipment wear, and improve the system's adaptive ability. In addition, the collaborative control module generates the main and auxiliary engine joint start instruction set based on the improved collaborative control algorithm, and accurately suppresses torque fluctuations and mechanical shocks through phased fuel injection pulse width adjustment, turbocharger phase compensation and shaft inertia dynamic compensation to ensure the smoothness and stability of power output. Finally, the execution and monitoring module executes control instructions within the preset time window and monitors the transient response characteristics in real time. When it is detected that the actual torque fluctuation rate exceeds the limit, the torque rebalancing mechanism of dynamic inertia compensation is immediately triggered to form a closed-loop control link of "perception-decision-execution-feedback" to reduce energy loss, reduce mechanical shock, and improve the starting efficiency and reliability of hybrid ships. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a structural block diagram of a starting control system of a main-auxiliary power hybrid ship mainly embodied in this embodiment.

[0047] Figure 2 It is a flowchart of a method for starting and controlling a main-auxiliary hybrid ship mainly embodied in this embodiment. DETAILED DESCRIPTION

[0048] The present invention is further described in detail below in conjunction with the following examples.

[0049] Reference Figure 1 , the starting control system of the main and auxiliary power hybrid ship, the system includes the following modules:

[0050] A multi-source data acquisition module is used to obtain a multi-source state data set of the ship power system, which includes the pressure waveform in the main engine cylinder, the turbine speed timing data, the battery pack temperature gradient distribution and the propeller load spectrum;

[0051] A fusion processing module is used to perform feature alignment processing on a multi-source state data set based on a multi-modal perception data fusion algorithm to generate a hybrid power starting parameter set including a power demand level, an energy allocation weight factor, and an equipment health index;

[0052] Dynamic weight allocation module, used to construct a dynamic weight allocation matrix according to the hybrid power startup parameter set, and calculate the host fuel injection amount correction coefficient, battery pack output power adjustment coefficient and shaft torque smoothing parameter through a double-layer fuzzy inference engine;

[0053] A collaborative control module is used to generate a main and auxiliary engine joint start instruction set based on an improved collaborative control algorithm. The instruction set includes a phased fuel injection pulse width, a turbocharger phase compensation angle, a battery pack multi-stage discharge curve, and a shaft system inertia compensation value;

[0054] The execution and monitoring module is used to execute the main and auxiliary engine joint start-up instruction set within a preset time window and monitor the transient response characteristics of the propulsion system. When the actual torque fluctuation rate exceeds a first preset threshold, a torque rebalancing mechanism based on dynamic inertia compensation is triggered.

[0055] Specifically, firstly, the multi-source data acquisition module is used to obtain the pressure waveform in the main engine cylinder, the turbine speed timing data, the battery pack temperature gradient distribution and the propeller load spectrum in real time, which solves the limitation of traditional ship power control relying on single sensor data, ensures the comprehensive perception of mechanical state, thermodynamic characteristics and load dynamics, and provides a high-precision data basis for the coordinated control of multiple power sources. Secondly, based on the multi-modal perception data fusion algorithm of the fusion processing module, feature alignment and deep correlation analysis are performed on heterogeneous data to generate power demand level, energy allocation weight factor and equipment health index, breaking through the static parameter assumption of traditional physical models and realizing dynamic decision optimization under complex working conditions. Furthermore, a three-dimensional weight space is constructed through the dynamic weight allocation module to achieve the global optimal solution between the start time delay, energy loss rate and equipment wear, and improve the system's adaptive ability. In addition, the collaborative control module generates the main and auxiliary engine joint start instruction set based on the improved collaborative control algorithm, and accurately suppresses torque fluctuations and mechanical shocks through phased fuel injection pulse width adjustment, turbocharger phase compensation and shaft inertia dynamic compensation to ensure the smoothness and stability of power output. Finally, the execution and monitoring module executes control instructions within the preset time window and monitors the transient response characteristics in real time. When it is detected that the actual torque fluctuation rate exceeds the limit, the torque rebalancing mechanism of dynamic inertia compensation is immediately triggered to form a closed-loop control link of "perception-decision-execution-feedback" to reduce energy loss, reduce mechanical shock, and improve the starting efficiency and reliability of hybrid ships.

[0056] Specifically, the multi-source data acquisition module can obtain the main engine cylinder pressure waveform, turbine speed timing data, battery pack temperature gradient distribution and propeller load spectrum in real time, which solves the limitation of traditional ship power control relying on single sensor data, ensures comprehensive perception of mechanical state, thermodynamic characteristics and load dynamics, and provides a high-precision data basis for multi-power source collaborative control. The workflow of the multi-source data acquisition module covers multi-dimensional data synchronous acquisition and preprocessing. The main engine cylinder pressure waveform can capture the pressure change curve in the combustion chamber in real time through a high-precision cylinder pressure sensor, and its waveform characteristics (such as pressure peak, rising slope) can reflect the fuel combustion efficiency and mechanical work state. For example, the pressure peak lag may indicate ignition timing deviation or poor fuel atomization. The turbine speed timing data can be collected by a magnetoelectric speed sensor to record the change trend of the turbocharger speed over time, which is used to analyze the intake efficiency and turbine dynamic response characteristics. When a sudden drop in speed is detected, the risk of turbo hysteresis or intake blockage can be inferred. The temperature gradient distribution of the battery pack can be generated by a two-dimensional thermal field map of an infrared thermal imager array arranged on the surface of the battery module, combined with the heat conduction equation modeling to evaluate the battery thermal management status in real time. If the temperature in a certain area is significantly higher than the surrounding area (such as a temperature difference of more than 10°C), the local over-temperature protection strategy is triggered. The propeller load spectrum can be generated by the fusion of the torque sensor and the vibration sensor, and the load spectrum characteristics are extracted through time-frequency analysis. For example, a sudden increase in high-frequency narrowband energy may correspond to the propeller cavitation effect, while low-frequency broadband fluctuations are often caused by wave impact. After the four types of data are aligned by time stamps, they are transmitted to the central controller through a high-speed bus to form a multimodal data set that is consistent in time and space, providing a complete input for subsequent fusion processing.

[0057] The specific fusion processing module may include the following submodules:

[0058] Wavelet packet decomposition module, used to perform wavelet packet decomposition on the pressure waveform in the main engine cylinder to extract the low-frequency energy proportion and high-frequency impact factor;

[0059] The harmonic distortion analysis module uses sliding window Fourier transform to analyze the harmonic distortion rate of turbine speed time series data;

[0060] Thermal conduction modeling module, used to establish a two-dimensional thermal conduction model of the battery pack temperature gradient distribution and calculate the maximum allowable discharge rate under the maximum temperature difference constraint;

[0061] A convolutional neural network module to identify transient shock signature patterns in the thruster load spectra;

[0062] The dynamic weight update module is used to update the weight coefficient of the dynamic weight allocation model according to the following formula:

[0063] W_t=α·e^{-β·ΔT}+(1-α)·tanh(γ·(P_avg / P_max))

[0064] Among them, α is the temperature sensitivity coefficient, β is the thermal attenuation factor, γ is the power regulation gain, ΔT is the current maximum temperature difference of the battery pack, P_avg is the average load power of the thruster, and P_max is the maximum allowable output power of the battery pack.

[0065] Specifically, the fusion processing module converts heterogeneous data into quantifiable control parameters through multi-level feature extraction and correlation analysis. The wavelet packet decomposition module decomposes the in-cylinder pressure waveform into multiple frequency bands. The low-frequency energy ratio reflects the combustion stability (such as the risk of knocking when the low-frequency ratio is lower than the threshold), and the high-frequency impact factor characterizes the mechanical impact intensity (such as the sudden increase in high-frequency energy may correspond to piston knocking). The harmonic distortion analysis module calculates the harmonic distortion rate of the turbine speed signal through the sliding window Fourier transform. A high distortion rate indicates turbine blade vibration or airflow turbulence. For example, when the third harmonic amplitude accounts for more than 5%, it is determined that the turbine dynamic balance has failed and the boost pressure needs to be reduced. The heat conduction modeling module simulates the temperature diffusion process of the battery pack based on the heat conduction equation, and dynamically calculates the allowable discharge rate in combination with the maximum temperature difference ΔT. For example, when ΔT reaches 15°C, the maximum discharge rate is limited to 0.8C to avoid thermal runaway. The convolutional neural network module uses a pre-trained CNN model to identify transient impact modes in the propeller load spectrum. For example, wave impact is manifested as a broadband energy surge with a short duration, while the propeller cavitation effect presents a narrowband resonance of a specific frequency. In the formula of the dynamic weight update module, α controls the weight distribution of temperature and power (such as α = 0.6, which means that the temperature difference limitation is emphasized), the exponential term e^{-β·ΔT} causes the weight to decay rapidly at high temperatures, and the tanh function compresses the power ratio (P_avg / P_max) to the [-1,1] interval to avoid weight mutations. For example, when the battery pack temperature difference ΔT rises to 12°C due to poor heat dissipation, the dynamic weight update module automatically increases α to 0.7, so that the control strategy prioritizes limiting the discharge power to a safe range rather than pursuing maximum power output. At the same time, the power distribution weight is smoothly adjusted through the tanh function to avoid shaft system oscillation caused by control command jumps.

[0066] The specific dynamic weight allocation module may include the following submodules:

[0067] A three-dimensional weighted space building module defines three optimization dimensions: power response priority, energy efficiency optimization level, and equipment protection level;

[0068] The fuzzy membership calculation module uses trapezoidal membership functions to process boundary conditions in three dimensions;

[0069] The collaborative game optimization module is used to solve the optimal weight combination through the following objective function:

[0070] min(λ1·τ_acc+λ2·E_loss+λ3·W_deg)

[0071] Wherein: τ_acc is the startup time delay, E_loss is the energy loss rate, W_deg is the equipment wear degree, λ1 is the first weight coefficient, λ2 is the second weight coefficient, and λ3 is the third weight coefficient.

[0072] Specifically, the dynamic weight allocation module achieves global optimal decision-making through multi-objective collaborative optimization. The three-dimensional weight space construction module defines three conflicting optimization objectives: the power response priority requires minimizing the startup time delay τ_acc (such as the time from ignition to stable speed), the energy efficiency optimization level requires reducing the energy loss rate E_loss (such as the combined loss of fuel consumption and battery discharge), and the equipment protection level aims to reduce the equipment wear W_deg (such as shaft fatigue accumulation). The fuzzy membership calculation module can use trapezoidal membership functions to process the fuzzy boundaries of each dimension, for example, dividing τ_acc into three fuzzy sets of "short" (0-5 seconds), "medium" (5-10 seconds), and "long" (>10 seconds) to avoid the rigid decision of traditional threshold segmentation. The collaborative game optimization module solves the Pareto optimal solution through the objective function, and the weight coefficient is dynamically adjusted according to the real-time working conditions. For example, in the emergency collision avoidance scenario, λ1 is significantly increased to give priority to shortening the startup time; under battery aging conditions, λ3 is increased to reduce equipment wear. For example, when the shaft vibration sensor detects an abnormal high-frequency component, the collaborative game optimization module automatically increases λ3, and the control strategy prioritizes reducing the fuel injection correction coefficient. Even if this prolongs the startup time, it can reduce shaft wear and avoid mechanical failure.

[0073] The specific collaborative control module may include the following submodules:

[0074] The power coupling modeling module is used to establish the main and auxiliary power coupling model based on the shaft system torque transfer function. The shaft system torque transfer function is:

[0075] T_out=K1·(P_engine+P_battery)-K2·dω / dt

[0076] Wherein, T_out is the shaft output torque, P_engine is the host output power, P_battery is the battery pack output power, dω / dt is the shaft angular acceleration change rate, K1 is the mechanical transmission efficiency factor, and K2 is the rotation inertia compensation coefficient;

[0077] The prediction-correction control module is used to adopt a dual-loop control strategy, in which the inner loop controls the proportional relationship between the fuel injection amount and the battery discharge rate, and the outer loop adjusts the coordinated parameters of the turbocharging phase and the shaft inertia compensation.

[0078] Specifically, the collaborative control module realizes precise power distribution by combining physical models with closed-loop control. In the shaft torque transfer function defined by the power coupling modeling module, K1·(P_engine+P_battery) represents the linear superposition of the host and battery power (such as K1=0.95 corresponds to 5% transmission loss), and K2·dω / dt is the dynamic inertia compensation term, which is used to offset the inertia torque during the acceleration / deceleration of the shaft (such as when dω / dt is negative, the compensation torque is positively superimposed to suppress deceleration fluctuations). The prediction-correction control module adopts a dual-loop structure: the inner loop adjusts the ratio of fuel injection amount to battery discharge rate in real time based on the current power demand (such as 70% of the power is provided by the host and 30% is supplemented by the battery), and the outer loop optimizes the intake efficiency by adjusting the turbocharger phase angle (such as delaying the phase angle to reduce the risk of surge), and dynamically adjusts the K2 value to adapt to the change in shaft inertia.

[0079] Specifically, the execution and detection module is responsible for the physical execution and dynamic feedback of the control instructions. In the execution stage, the phased fuel injection pulse width is converted into the opening and closing timing signal of the injector solenoid valve (such as a pulse width of 5ms corresponding to an injection volume of 10ml), the turbocharger phase compensation angle is converted into the rotation angle instruction of the variable geometry turbine (VGT) blade (such as a phase angle of 60° corresponding to a blade opening of 80%), and the shaft inertia compensation value is converted into the reverse torque output of the motor (such as a compensation value of 50Nm corresponding to a motor output of -50Nm torque). In the monitoring stage, the shaft output torque is collected in real time through a high-precision torque sensor, and the fluctuation rate is calculated. When the fluctuation rate exceeds the threshold, the torque rebalancing mechanism is triggered: the K2 coefficient is dynamically adjusted and a reverse torque pulse is injected. For example, when a positive fluctuation peak is detected, a negative compensation torque is output in the next control cycle.

[0080] In some embodiments, the system may further include:

[0081] Adaptive learning module, used to achieve dynamic optimization of control parameters through the following steps:

[0082] Record actual control effect data sets during each startup process, including fuel consumption rate, torque fluctuation rate, battery discharge efficiency and shaft vibration amplitude;

[0083] Construct a parameter optimization model based on deep reinforcement learning, input the actual control effect data set and the current working condition feature vector, and output the adjustment coefficient of the dynamic weight allocation matrix;

[0084] Update the fuzzy rule base of the double-layer fuzzy inference engine, combine the historical fuzzy rule confidence and the gradient of the performance loss function to generate new fuzzy rule confidence;

[0085] The optimized parameters are deployed to the collaborative control module to realize online iteration of the control strategy.

[0086] Specifically, the adaptive learning module continuously optimizes the control strategy through data-driven. The deep reinforcement learning model uses historical control effect data as state input and dynamic weight allocation matrix adjustment coefficient as action output, which motivates the system to strike a balance between shortening startup time, reducing energy consumption and reducing wear. The fuzzy rule base is updated based on the gradient direction of the performance loss function (such as the mean square error of torque fluctuation). If a rule frequently causes E_loss to increase, its confidence η is reduced (such as from 0.9 to 0.7). Example: When a ship operates in a low sea condition area for a long time, the adaptive learning module can automatically reduce λ1 (power response weight) to 0.4 and increase λ2 (energy efficiency weight) to 0.5 through historical data analysis, forming a control strategy with energy saving as the core and reducing average energy consumption.

[0087] In some embodiments, the system may further include:

[0088] The vibration suppression module is a vibration acceleration sensor array deployed at the key nodes of the shaft system of the hybrid ship, which is used to collect the three-dimensional vibration spectrum; when the amplitude of the characteristic frequency component is detected to exceed the second preset threshold, the shaft system torsional vibration suppression mechanism is activated, and the mechanism includes: adjusting the fuel injection interval phase angle, injecting reverse offset torque pulses, and dynamically adjusting the battery pack output power fluctuation component.

[0089] Specifically, the vibration suppression module suppresses the resonance of the shaft system through active control. The vibration acceleration sensor array collects XYZ three-axis vibration signals, and extracts characteristic frequency components (such as the natural frequency of the shaft system and its harmonics) through fast Fourier transform. When a certain frequency amplitude exceeds the limit, the suppression mechanism is activated: the phase angle of the fuel injection interval is adjusted to change the excitation frequency, such as changing from uniform injection to 120° injection interval, and injecting reverse torque pulses to offset vibration energy, such as phase-synchronous output of reverse pulses, and dynamically adjusting the battery power fluctuation component to destroy the resonance condition. For example, in the high-speed cruising stage, the vibration suppression module detects that the amplitude of the 2nd order harmonic of the shaft system reaches 0.6g (threshold 0.5g), and immediately adjusts the fuel injection interval from 180° to 90°, and injects reverse pulses through the motor to reduce the vibration amplitude to below 0.3g within 2 seconds to avoid fatigue damage to the coupling.

[0090] A start-up control method for a main-auxiliary hybrid ship, referring to Figure 2, including the following steps: S1: obtaining a multi-source state data set of a ship power system, the multi-source state data set includes a main engine cylinder pressure waveform, a turbine speed timing data, a battery pack temperature gradient distribution and a propeller load spectrum; S2: performing feature alignment processing on the multi-source state data set based on a multimodal perception data fusion algorithm, and generating a hybrid power starting parameter set including a power demand level, an energy allocation weight factor and an equipment health index; S3: constructing a dynamic weight allocation matrix according to the hybrid power starting parameter set, and calculating the main engine fuel injection amount correction coefficient, the battery pack output power adjustment coefficient and the shaft system torque smoothing parameter through a double-layer fuzzy inference engine; S4: generating a main and auxiliary engine joint start instruction set based on an improved collaborative control algorithm, the instruction set including a staged fuel injection pulse width, a turbocharger phase compensation angle, a battery pack multi-stage discharge curve and a shaft system inertia compensation value; S5: executing the main and auxiliary engine joint start instruction set within a preset time window, and monitoring the transient response characteristics of the propulsion system, and triggering a torque rebalancing mechanism based on dynamic inertia compensation when the actual torque fluctuation rate exceeds a first preset threshold.

[0091] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A start-up control system for a main-auxiliary hybrid ship, characterized in that: Includes the following modules: A multi-source data acquisition module is used to obtain a multi-source state data set of a ship power system, wherein the multi-source state data set includes a pressure waveform in a main engine cylinder, a turbine speed timing data, a battery pack temperature gradient distribution, and a propeller load spectrum; A fusion processing module, used for performing feature alignment processing on the multi-source state data set based on a multi-modal perception data fusion algorithm to generate a hybrid power starting parameter set including a power demand level, an energy allocation weight factor and an equipment health index; A dynamic weight allocation module, used to construct a dynamic weight allocation matrix according to the hybrid power startup parameter set, and calculate the host fuel injection amount correction coefficient, battery pack output power adjustment coefficient and shaft system torque smoothing parameter through a double-layer fuzzy inference engine; A collaborative control module, used to generate a main and auxiliary engine joint start instruction set based on an improved collaborative control algorithm, wherein the instruction set includes a phased fuel injection pulse width, a turbocharger phase compensation angle, a battery pack multi-stage discharge curve, and a shaft system inertia compensation value; The execution and monitoring module is used to execute the main and auxiliary engine joint start-up instruction set within a preset time window and monitor the transient response characteristics of the propulsion system. When the actual torque fluctuation rate exceeds a first preset threshold, a torque rebalancing mechanism based on dynamic inertia compensation is triggered.

2. The starting control system of the main-auxiliary hybrid ship according to claim 1, characterized in that: The fusion processing module includes: A wavelet packet decomposition module, used to perform wavelet packet decomposition on the main engine cylinder pressure waveform to extract the low-frequency energy proportion and high-frequency impact factor; A harmonic distortion analysis module, which uses a sliding window Fourier transform to analyze the harmonic distortion rate of the turbine speed time series data; A heat conduction modeling module, used to establish a two-dimensional heat conduction model of the temperature gradient distribution of the battery pack and calculate the maximum allowable discharge rate under the maximum temperature difference constraint; A convolutional neural network module for identifying transient impact characteristic patterns in the thruster load spectrum; The dynamic weight update module is used to update the weight coefficient of the dynamic weight allocation model according to the following formula: W_t=α·e^{-β·ΔT}+(1-α)·tanh(γ·(P_avg / P_max)) Among them, α is the temperature sensitivity coefficient, β is the thermal attenuation factor, γ is the power regulation gain, ΔT is the current maximum temperature difference of the battery pack, P_avg is the average load power of the thruster, and P_max is the maximum allowable output power of the battery pack.

3. The starting control system of the main-auxiliary hybrid ship according to claim 2, characterized in that: The dynamic weight allocation module comprises: A three-dimensional weighted space building module defines three optimization dimensions: power response priority, energy efficiency optimization level, and equipment protection level; A fuzzy membership calculation module, using a trapezoidal membership function to process the boundary conditions of the three dimensions; The collaborative game optimization module is used to solve the optimal weight combination through the following objective function: min(λ1·τ_acc+λ2·E_loss+λ3·W_deg) Wherein: τ_acc is the startup time delay, E_loss is the energy loss rate, W_deg is the equipment wear degree, λ1 is the first weight coefficient, λ2 is the second weight coefficient, and λ3 is the third weight coefficient.

4. The starting control system of the main-auxiliary hybrid ship according to claim 3, characterized in that: The collaborative control module includes: The power coupling modeling module is used to establish the main and auxiliary machine power coupling model based on the shaft system torque transfer function, where the shaft system torque transfer function is: T_out=K1·(P_engine+P_battery)-K2·dω / dt Wherein, T_out is the shaft output torque, P_engine is the host output power, P_battery is the battery pack output power, dω / dt is the shaft angular acceleration change rate, K1 is the mechanical transmission efficiency factor, and K2 is the rotation inertia compensation coefficient; The prediction-correction control module is used to adopt a dual-loop control strategy, in which the inner loop controls the proportional relationship between the fuel injection amount and the battery discharge rate, and the outer loop adjusts the coordinated parameters of the turbocharging phase and the shaft inertia compensation.

5. The starting control system of the main-auxiliary hybrid ship according to claim 4, characterized in that: The system further comprises: Adaptive learning module, used to achieve dynamic optimization of control parameters through the following steps: Record actual control effect data sets during each startup process, including fuel consumption rate, torque fluctuation rate, battery discharge efficiency and shaft vibration amplitude; Constructing a parameter optimization model based on deep reinforcement learning, inputting the actual control effect data set and the current operating condition feature vector, and outputting the adjustment coefficient of the dynamic weight allocation matrix; Updating the fuzzy rule base of the two-layer fuzzy inference engine, combining the historical fuzzy rule confidence and the gradient of the performance loss function, and generating a new fuzzy rule confidence; The optimized parameters are deployed to the collaborative control module to achieve online iteration of the control strategy.

6. The starting control system of the main-auxiliary hybrid ship according to claim 1, characterized in that: The system further comprises: A vibration suppression module is deployed in a vibration acceleration sensor array at a key node of the shaft system of the hybrid ship, and is used to collect a three-dimensional vibration spectrum; when it is detected that the amplitude of the characteristic frequency component exceeds a second preset threshold, a shaft system torsional vibration suppression mechanism is activated, and the mechanism includes: adjusting the fuel injection interval phase angle, injecting a reverse offset torque pulse, and dynamically adjusting the battery pack output power fluctuation component.

7. A startup control method for a main-auxiliary hybrid ship, characterized in that: The following steps are involved: Acquire a multi-source state data set of a ship power system, wherein the multi-source state data set includes a pressure waveform in a main engine cylinder, a turbine speed timing data, a battery pack temperature gradient distribution, and a propeller load spectrum; Performing feature alignment processing on the multi-source state data set based on a multimodal sensing data fusion algorithm to generate a hybrid power starting parameter set including a power demand level, an energy allocation weight factor, and an equipment health index; A dynamic weight distribution matrix is ​​constructed according to the hybrid power starting parameter set, and a main engine fuel injection amount correction coefficient, a battery pack output power adjustment coefficient and an axle system torque smoothing parameter are calculated through a double-layer fuzzy inference engine; Generate a main and auxiliary engine joint start instruction set based on an improved cooperative control algorithm, the instruction set including phased fuel injection pulse width, turbocharger phase compensation angle, battery pack multi-stage discharge curve and shaft system inertia compensation value; The main and auxiliary engine joint start instruction set is executed within a preset time window, and the transient response characteristics of the propulsion system are monitored. When the actual torque fluctuation rate exceeds a first preset threshold, a torque rebalancing mechanism based on dynamic inertia compensation is triggered.

8. The startup control method of a main-auxiliary hybrid ship according to claim 7, characterized in that: The steps of performing feature alignment processing on the multi-source state data set based on a multimodal sensing data fusion algorithm to generate a hybrid power starting parameter set including a power demand level, an energy allocation weight factor and an equipment health index are specifically as follows: Performing wavelet packet decomposition on the main engine cylinder pressure waveform to extract low-frequency energy proportion and high-frequency impact factor; Using sliding window Fourier transform to analyze the harmonic distortion rate of the turbine speed time series data; Establishing a two-dimensional heat conduction model of the temperature gradient distribution of the battery pack, and calculating the maximum allowable discharge rate under the maximum temperature difference constraint; identifying a transient impact characteristic mode in the thruster load spectrum; Update the weight coefficient of the dynamic weight allocation model according to the following formula: W_t=α·e^{-β·ΔT}+(1-α)·tanh(γ·(P_avg / P_max)) Among them, α is the temperature sensitivity coefficient, β is the thermal attenuation factor, γ is the power regulation gain, ΔT is the current maximum temperature difference of the battery pack, P_avg is the average load power of the thruster, and P_max is the maximum allowable output power of the battery pack.

9. The startup control method of a main-auxiliary hybrid ship according to claim 8, characterized in that: The steps of constructing a dynamic weight distribution matrix according to the hybrid power starting parameter set and calculating the host fuel injection amount correction coefficient, battery pack output power adjustment coefficient and shaft system torque smoothing parameter through a double-layer fuzzy inference engine are specifically as follows: Define three optimization dimensions: power response priority, energy efficiency optimization level, and equipment protection level; Using trapezoidal membership functions to process the boundary conditions in the three dimensions; The optimal weight combination is solved by the following objective function: min(λ1·τ_acc+λ2·E_loss+λ3·W_deg) Wherein: τ_acc is the startup time delay, E_loss is the energy loss rate, W_deg is the equipment wear degree, λ1 is the first weight coefficient, λ2 is the second weight coefficient, and λ3 is the third weight coefficient.

10. The startup control method of a main-auxiliary hybrid ship according to claim 9, characterized in that: The main and auxiliary engine joint start instruction set is generated based on the improved cooperative control algorithm, and the instruction set includes the steps of phased fuel injection pulse width, turbocharger phase compensation angle, battery pack multi-stage discharge curve and shaft system inertia compensation value, specifically: The main and auxiliary machine power coupling model is established based on the shaft system torque transfer function, and the shaft system torque transfer function is: T_out=K1·(P_engine+P_battery)-K2·dω / dt Wherein, T_out is the shaft output torque, P_engine is the host output power, P_battery is the battery pack output power, dω / dt is the shaft angular acceleration change rate, K1 is the mechanical transmission efficiency factor, and K2 is the rotation inertia compensation coefficient; A dual-loop control strategy is adopted. The inner loop controls the proportional relationship between the fuel injection amount and the battery discharge rate, and the outer loop adjusts the coordinated parameters of the turbocharging phase and shaft inertia compensation.

Images