Ship main propulsion intelligent control system based on multi-source data fusion

Abstract

The invention discloses a ship main propulsion intelligent control system based on multi-source data fusion, and relates to the technical field of ship main propulsion intelligent control. The problems that in the prior art, dynamic optimization is insufficient, the data fusion capacity is weak, energy efficiency is low, a fault-tolerant mechanism is passive, a control strategy is rigid, fault recovery is slow, the intelligent level is low, and complex environment changes cannot be handled are solved. According to the invention, by introducing a multi-source data fusion intelligent control algorithm, power distribution and battery scheduling of the ship main propulsion system are dynamically optimized, deep fusion is carried out in combination with environment data and equipment data, and the propulsion efficiency and energy management are improved; through multi-dimensional data fusion, system adaptability and fault-tolerant capability are improved, real-time adjustment and intelligent prediction are realized, fuel consumption is reduced, battery aging is reduced, and navigation stability is improved. The system further has an intelligent fault detection and prediction mechanism, manual intervention is reduced, and the fault-tolerant capability and the intelligent level of the system are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control of ship main propulsion, and specifically to an intelligent control system for ship main propulsion based on multi-source data fusion. Background Art

[0002] Traditional ship propulsion control systems mainly adopt threshold control based on a single sensor (such as a speed sensor), and adjust the diesel engine power output through the PID algorithm. The power distribution method is fixed (for example, the diesel engine undertakes 80% of the load, and the rest is supplemented by the battery), but there is a lack of dynamic optimization. At the same time, the phenomenon of data islands is serious, and environmental data (wind speed, sea waves) and equipment status data are processed separately. Its typical defects include low control accuracy (such as the response delay > 2 seconds when the load suddenly changes due to 6-level wind and waves), poor energy consumption efficiency (the diesel engine often operates in a non-economic power area, and the fuel consumption rate increases by 15% - 20%), and short battery life (without an SOC dynamic protection strategy, the battery is frequently deeply discharged, resulting in a cycle life < 1000 times). Early data fusion technologies mainly relied on simple weighted averaging or Kalman filtering to fuse data from similar sensors (such as mean filtering of multiple temperature sensors). Environmental data and equipment data were only statically associated through a rule base (such as IF-THEN statements), resulting in weak heterogeneous data fusion capabilities (such as the modeling error of the dynamic impact of wind speed on propeller thrust > 20%) and poor anti-interference ability (such as high-frequency interference in the current signal causing large fluctuations in the fusion result, with an error of ±5%). Traditional fault tolerance mechanisms adopt hardware redundancy (such as dual generator backups) or simple threshold alarms (such as cutting off the power supply when the temperature exceeds 90°C), but fault recovery relies on manual intervention, and the response time > 10 seconds. Its typical defects include high fault tolerance costs (hardware redundancy increases the system weight and cost, and the standby generator accounts for 8% - 12% of the total ship cost), and insufficient intelligence (unable to perform predictive maintenance, and the sudden failure shutdown rate > 5%). The key defects of the existing technologies include single data fusion dimension (only fusing data from similar sensors, lacking in-depth association of environment + equipment + navigation, such as not dynamically coupling the wave height with the propeller speed, resulting in a decrease in propulsion efficiency of more than 30% when the sea conditions deteriorate), rigid control strategies (fixed power distribution ratios cannot adapt to load fluctuations, resulting in frequent starts and stops of the diesel engine, > 10 times per hour, and a reduction in fuel efficiency of 12% - 18%; charge and discharge scheduling relies on manual experience, SOC protection lags, and accelerates battery aging), poor dynamic environment adaptability (the overshoot of traditional PID control in non-linear and strongly disturbed scenarios > 20%, with insufficient stability, and lacking multi-objective optimization capabilities, making it difficult to balance fuel economy, equipment life, and navigation safety, such as forcing the diesel engine to operate overloaded to ensure power supply), and passive and inefficient fault tolerance mechanisms (fault diagnosis relies on threshold alarms, with a false alarm rate as high as 15%, unable to accurately locate the root cause of the fault, and long standby system switching delays, resulting in a power outage risk for key equipment (such as navigation radar) during the switching period). Summary of the Invention

[0003] In view of the problems mentioned in the above background art, the present invention provides an intelligent control system for ship main propulsion based on multi-source data fusion.

[0004] The object of the present invention can be achieved by the following technical solutions: An intelligent control system for ship main propulsion based on multi-source data fusion, comprising: a data processing and fusion module, a control decision-making module, and an execution control module;

[0005] The data processing and fusion module is used for denoising and data integration of multi-source data, specifically:

[0006] Step 1: Use the Kalman filter to eliminate the noise of the sensor data;

[0007] Step 2: Integrate the data from different sensors; use a multi-layer perceptron for feature acquisition; input the multi-source sensor data into the MLP, calculate the neurons in the hidden layer through the hidden layer to obtain the final output; then perform data fusion according to the output result to obtain the fused data;

[0008] The control decision-making module generates control instructions according to the system state and navigation requirements, specifically: perform optimization analysis of the power distribution algorithm:

[0009] S011: Determine the optimization objectives, including minimizing fuel consumption, optimizing battery charge and discharge, and ensuring that the power demand of the propulsion system is met;

[0010] S012: Input parameters: load demand P load 、maximum power of the diesel generator current generator power P gen (t), state of charge SOC(t) of the battery, and charge and discharge efficiency η bat ;

[0011] S013: Calculation process:

[0012] S131: Perform the optimization objective function;

[0013] S132: Perform the constraint conditions;

[0014] S014: Through the dynamic power distribution strategy and charge and discharge scheduling strategy; if the SOC is higher than the set threshold, generate the first charge and discharge scheduling instruction, if the SOC is lower than the set threshold, generate the second charge and discharge scheduling instruction;

[0015] When the load is less than the minimum value within the preset load demand range, generate a battery charging instruction, and when the load is greater than the maximum value within the preset load demand range, generate a battery discharging instruction;

[0016] The execution control module converts the instructions generated by the control decision-making module into physical actions.

[0017] As a preferred embodiment of the present invention, it further includes a multi-source data acquisition module and a human-computer interaction module;

[0018] The multi-source data acquisition module collects the status data of diesel generators, batteries, and drive motor devices in real time through current sensors, voltage sensors, speed sensors, and temperature sensors; collects environmental data and navigation status data, converts the analog signals output by the sensors into digital signals, and transmits the data to the main controller through the CAN bus and Ethernet;

[0019] The human-computer interaction module provides a real-time status visualization interface, receives operation instructions, triggers alarm prompts, supports historical data query and fault diagnosis interaction, and allows emergency intervention.

[0020] As a preferred embodiment of the present invention, the specific process of obtaining the final output by calculating the hidden layer neurons through the hidden layer is as follows:

[0021] Use a multi-layer perceptron to obtain features; input the multi-source sensor data into the MLP, and set the output as: X = (x1, x2,..., x n ); x1, x2,..., x n represent n sensor data; then perform normalization processing, and calculate the hidden layer neurons through the hidden layer: The formula is: w ij is the weight from the input data x i to the hidden layer neuron h j ; bj is the bias term; f(·) is the activation function; if there are L hidden layers, the calculation formula becomes:

[0022] H (l) = f(W (l) H (l-1) + b (l) ), H (l) is the output of the l-th hidden layer, W (l) and b (l) are the weights and biases of this layer, and the activation function f(x) performs a non-linear transformation; then calculate by the neurons of the output layer:

[0023] y is the final output, and σ(·) is the activation function.

[0024] As a preferred embodiment of the present invention, the specific process of further performing data fusion according to the output result to obtain the fused data is as follows:

[0025] Rule-based fusion: Calculate the battery SOC:

[0026] Calculate the current SOC based on the charge and discharge current and capacity of the battery;

[0027] Perform weighted average fusion to calculate the weighted average value;

[0028] Perform Kalman fusion, including GPS+IMU data fusion and wind speed+ocean current data fusion.

[0029] As a preferred embodiment of the present invention, the specific process of optimizing the objective function is as follows:

[0030] Define the objective function to minimize fuel consumption and battery usage cost: minF = C fuel ×P gen (t)+C battery ×P bat (t), where C fuel is the fuel cost coefficient and C battery is the battery usage cost.

[0031] As a preferred embodiment of the present invention, the specific constraints are as follows:

[0032] Power balance constraint, diesel generator output power constraint, and battery SOC constraint;

[0033] Power balance constraint: P load =P gen (t)+P bat (t); P bat (t) is positive or negative;

[0034] Diesel generator output power constraint: Battery SOC constraint: SOC min ≤SOC(t)≤SOC max .

[0035] As a preferred embodiment of the present invention, the specific process by which the execution control module converts the instructions generated by the control decision module into physical actions is as follows:

[0036] When receiving the charge and discharge scheduling instruction 1, increase the power of the diesel generator, adjust the fuel injection volume, increase the speed and output power, cut off the battery discharge circuit, reduce the discharge current, and trigger the trickle charge protection; calculate the power distribution factor α1 to distribute the power between the generator and the battery; the frequency converter adjusts the IGBT switching frequency to match the power supply capacity; monitor the temperature and trigger the cooling system when it exceeds the limit; the HMI displays the power supply status and SOC information and performs fault handling when abnormal;

[0037] When receiving the charge-discharge scheduling instruction two, the generator power is increased to meet the load and charging requirements, while charging the battery; the BMS cuts off the discharge circuit, only allowing small-current discharge, and dynamically adjusts the charging mode; calculates the power distribution factor α2, and optimizes the charging strategies of the generator and the battery; reduces the propulsion power when the load is exceeded to ensure the power supply of key equipment; monitors the charging temperature rise and abnormal charging current, and triggers the cooling fan or cuts off the charging circuit; starts the standby power supply, and switches to the voltage estimation mode when the SOC sensor fails; the HMI updates the power distribution status in real time and provides abnormal alarm prompts;

[0038] When receiving the battery charging instruction, the BMS self-checks the SOC and temperature, and adjusts the cooling or heating when the temperature is abnormal; determines the charging source; disconnects the discharge relay; switches the constant current, constant voltage, and trickle charging modes according to the SOC during the charging process; monitors the charging parameters and triggers overvoltage, overcurrent, and over-temperature protection; performs a soft stop after charging is completed and enters the floating charge or standby state;

[0039] When receiving the battery discharge instruction, the BMS self-checks the SOC and temperature, and pauses the discharge when the SOC is low or the temperature is abnormal; adjusts the discharge power according to the load demand; closes the charging relay, opens the discharge relay, and matches the load type; adopts the constant power or pulse discharge mode to optimize the energy output; monitors the discharge parameters and triggers the abnormal protection mechanism.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] 1. Through the Kalman filter, MLP neural network, and weighted fusion strategy, the present invention performs spatio-temporal alignment and feature extraction on heterogeneous data such as generator operating conditions (current / temperature), environmental parameters (wind speed / wave height), and navigation states (load / speed). For example, the Kalman fusion algorithm using GPS+IMU can reduce the heading estimation error to within 0.5°, and the non-linear feature mapping based on the multi-layer perceptron can effectively identify the SOC estimation deviation caused by battery aging (such as improving the temperature compensation accuracy of the voltage-current curve by 15%). This deep fusion mechanism enables the system to maintain a propeller thrust volatility of less than 2% under sea state 6, significantly better than the 5% - 8% fluctuation performance of traditional PID control.

[0042] 2. By constructing a mixed-integer programming model, the present invention incorporates the fuel cost coefficient (0.3 L / kWh), the battery cycle life cost (0.05 $ / kWh), and the real-time SOC constraint (SOCmin = 20%) into the objective function, and dynamically calculates the power distribution factor α in combination with the rolling horizon optimization algorithm. Measured data shows that under typical working conditions (load demand fluctuates between 80 - 120 kW), the system can adjust the α value in seconds (for example, when α jumps from 0.7 to 0.95, the response time < 500 ms), increasing the proportion of time that the diesel engine operates in the economic power zone (70% - 85% of the rated power) to 92%, and achieving a fuel saving rate of 17.3% compared with the traditional solution. At the same time, through the threshold trigger mechanism (ΔSOC = 5%) of the charge and discharge scheduling command, the lithium battery cycle life can be extended to more than 4000 times, which is 2.5 times higher than that without the protection strategy.

[0043] 3. The present invention adopts a three-level fault tolerance design for the execution layer - decision layer - hardware layer: at the execution layer, when it is detected that the generator is overloaded (greater than 110% of the rated current), the PMS can complete the standby power supply switching within 50 ms (for example, the super capacitor makes up for a 20 kW gap); at the decision layer, if the SOC sensor fails, the system automatically switches to the Kalman prediction mode based on the open circuit voltage (OCV), and the SOC estimation error can be controlled within ±3%; at the hardware layer, a dual CAN bus redundant architecture is adopted, and when the main bus fails, the takeover delay of the slave bus is less than 10 ms. Actual tests show that when the system simulates battery thermal runaway (temperature rise rate > 5 °C / min), the BMS can cut off the charging circuit and fully start the liquid cooling system within 200 ms, stabilizing the battery temperature below the 45 °C safety threshold, and shortening the fault response time by 60% compared with the traditional solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] For the convenience of those skilled in the art to understand, the present invention will be further described below with reference to the accompanying drawings.

[0045] Figure 1 It is a principle block diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention. It should be understood that the terms "including" and "comprising" used in the specification and claims of this disclosure indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0047] It should also be understood that the terms used in this disclosure are only for the purpose of describing specific embodiments and are not intended to limit the disclosure. As used in this disclosure and claims, the singular forms of "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should also be further understood that the term "and / or" used in this disclosure and claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations.

[0048] See also Figure 1 As shown, a ship main propulsion intelligent control system based on multi-source data fusion includes: a multi-source data acquisition module, a data processing module, a control decision module, an execution control module and a human-computer interaction module;

[0049] The multi-source data acquisition module collects the status data (such as voltage, current, speed, temperature, etc.) of diesel generators, batteries, drive motors and other equipment in real time through current sensors, voltage sensors, speed sensors, temperature sensors, etc. It is also used to collect environmental data (such as wind speed, wave height) and navigation status data (such as speed, load); converts the analog signal output by the sensor into a digital signal, and transmits the data to the main controller through the CAN bus and Ethernet;

[0050] The data processing module retrieves multi-source data in the main controller for denoising:

[0051] Step 1: Filter and denoise the collected multi-source data: eliminate the noise of sensor data by using Mann filter;

[0052] Step 2: Integrate the data from different sensors; use a multi-layer perceptron (MLP) to obtain features; input the multi-source sensor data into the MLP, and set the output to: X = (x1, x2, ..., x n ), where x1, x2, ..., x n Represents n sensor data, such as battery current, voltage, temperature, motor torque, etc.; then performs standardization processing, such as normalization (0-1) or Z-score standardization processing, to ensure that the distribution of all input data is consistent and improve training stability; then, the hidden layer neurons are calculated through the hidden layer: the formula is: Where: w ij is the input data x i To the hidden layer neuron h j The weight of ; bj is the bias term; f(·) is the activation function, usually ReLU: f(x) = max(0,x); if there are L hidden layers, the calculation formula becomes: H (l) =f(W(l) H (l-1) + b (l) ), where H (l) is the output of the l-th hidden layer, W (l) and b (l) are the weights and biases of this layer, and the activation function f(x) performs a non-linear transformation; then the neurons of the output layer calculate: where: y is the final output, σ(·) is the activation function; then data fusion is performed according to the output result, and data fusion includes:

[0053] Rule-based fusion, which is applicable to cases where there is a clear mathematical relationship between data, such as: battery SOC calculation:

[0054] Calculate the current SOC through the charge and discharge current and capacity of the storage battery;

[0055] Weighted average fusion, which is applicable to calculating the weighted average of the same type of data provided by multiple sensors (such as the readings of multiple current sensors) to improve accuracy;

[0056] Kalman fusion, which is applicable to data fusion of dynamic changes, such as: GPS + IMU data fusion: improving the accuracy of position, speed, and heading estimation; wind speed + ocean current data fusion: predicting the true navigation state of the ship;

[0057] The control decision module generates control instructions according to the system state and navigation requirements, specifically: performing optimization analysis of the power distribution algorithm:

[0058] S011: Determine the optimization objectives, including minimizing fuel consumption, optimizing the charge and discharge of the storage battery, and ensuring that the power demand of the propulsion system is met;

[0059] S012: Input parameters, including the load demand P load (total power demand of the current ship propulsion system), maximum power of the diesel generator (rated maximum power output capacity of the diesel generator), power P gen (t) (actual power output of the diesel generator at time t), state of charge (SOC) of the storage battery SOC(t) (current charge state of the storage battery, usually between SOC min and SOC max ), and charge and discharge efficiency η bat ;

[0060] S013: Calculation process:

[0061] S131: Optimize the objective function, define the objective function, and minimize fuel consumption and storage battery usage cost: minF = C fuel × Pgen (t) + C battery ×P bat (t), where C fuel is the fuel cost coefficient, representing the fuel consumption rate of the diesel generator, C battery is the battery usage cost, considering battery life attenuation and the cycle cost of the battery;

[0062] S132: Set the constraint conditions including: power balance constraint, diesel generator output power constraint, and battery SOC constraint; Power balance constraint: P load = P gen (t) + P bat (t); where P bat (t) may be positive (discharging) or negative (charging);

[0063] Diesel generator output power constraint: Battery SOC constraint: SOC min ≤ SOC(t) ≤ SOC max ;

[0064] S014: Dynamic power distribution strategy: To optimize power distribution, set the power distribution factor α and substitute it into the formula to get: P gen (t) = αP load , P bat (t) = (1 - α)P load ; where α is calculated through an optimization algorithm to ensure that power distribution achieves the lowest fuel consumption while satisfying the constraints;

[0065] Charge and discharge scheduling strategy: If SOC is higher than the set threshold, i.e., SOC > SOCmax - ΔSOC, generate charge and discharge scheduling instruction one; When reducing battery charging, give priority to using the generator for power supply, limit battery discharge, and avoid deep discharge damage to battery life; If SOC is lower than the set threshold, i.e., SOC < SOCmin + ΔSOC, then generate charge and discharge scheduling instruction two; Limit battery discharge, prevent SOC from being too low to supply power, increase the generator power, and give priority to charging the battery;

[0066] When the load is less than the minimum value within the preset load demand range, generate a battery charging instruction; The diesel generator maintains economic power output, and the excess power is used to charge the battery; When the load is greater than the maximum value within the preset load demand range, generate a battery discharge instruction; The battery discharges to assist the generator in power supply and reduce the high-load operation of the diesel engine;

[0067] The execution control module converts the instructions generated by the control decision module into physical actions, specifically:

[0068] When receiving the charge-discharge scheduling instruction 1, control the power of the diesel generator to increase. The main controller sends a power increase instruction to the diesel generator ECU, and the ECU adjusts the fuel injection amount (for example, increases the opening of the solenoid valve of the fuel injection pump) to increase the diesel engine speed and output power; the output power of the generator gradually increases from the current value (such as 50 kW) to the target value (such as 90 kW);

[0069] Cut off the discharge circuit: The main controller sends a **"discharge limit" instruction** to the battery BMS; the BMS controls the discharge contactor to disconnect (or reduces the duty cycle of the DC-DC converter) to make the discharge current approach 0, and the output power of the battery drops from the current value (such as 30 kW) to the preset safety value (such as below 5 kW), and the BMS continuously monitors the change of SOC. If the SOC continues to rise due to self-discharge or a small load, trigger trickle charge protection (such as limiting the charging current ≤ 1 A);

[0070] The main controller calculates the power distribution factor α1 according to the optimization algorithm (for example, α1 = 0.95), and distributes the power according to the formula: P gen (t) = α1 × P load ,P bat (t) = (1 - α1) × P load ; If the load demand is 100 kW, the generator undertakes 95 kW, and the battery only provides 5 kW (or reverse charging); and fine-tunes the value of α1 through the PID controller to ensure a smooth transition of the generator power (such as adjusting α1 ± 0.01 per second);

[0071] Propulsion system cooling linkage: If the load demand changes, the frequency converter adjusts the IGBT switching frequency to match the power supply capacity of the generator (such as the motor speed increases from 1000 rpm to 1200 rpm); to keep the propeller thrust stable and avoid ship vibration caused by sudden power increase; when the temperature sensor detects that the generator temperature > 80 °C, trigger the cooling water pump and fan to start (through the relay to close the circuit) (such as the coolant flow rate increases from 50 L / min to 80 L / min to ensure the generator heat dissipation);

[0072] Finally, update the HMI display on the operation interface to show the current status, such as text like "Generator power supply: 95%", "Battery discharge is limited, SOC = 91%", etc.; when the alarm indicator light (yellow) indicates "High SOC status"; if the generator is detected to be overloaded (such as current > 110% of the rated value), the fault diagnosis module immediately executes: cut off the fuel supply (the ECU triggers an emergency shutdown), and switch to the standby power supply (such as another generator).

[0073] When receiving the charge-discharge scheduling instruction two, the main controller sends a "boost power + charge" instruction to the diesel generator ECU. The ECU increases the fuel injection volume, boosts the output power to meet the load demand, and charges the battery at the same time. If the generator was originally operating at the economic power (such as 60% of the rated power), it is now boosted to 85% - 95% of the rated power (for example, from 80kW to 120kW). The engine charges the battery through a rectifier or a bidirectional DC-DC converter, and the charging current is set by the BMS (such as a current limit of 0.2C, that is, 20A for a 100Ah battery).

[0074] After receiving the instruction, the BMS immediately cuts off the discharge circuit contactor (physically disconnects) to prohibit the battery from discharging. If the system needs emergency discharge (such as a sudden increase in load), the BMS only allows a very small current discharge (such as ≤5% of the rated current). The charging circuit contactor closes, and the charger switches the charging mode according to the SOC status: when the SOC is extremely low (such as SOC < 20%), it quickly charges at a constant current (such as 0.3C) to a safe threshold; when the SOC rises above 25%, it switches to the constant voltage mode to prevent overcharging.

[0075] Similarly, according to the optimization algorithm, the dynamic power distribution factor α2 is calculated and adjusted. The objective function is adjusted to minimize battery discharge + meet the charging demand, and the power distribution formula is adjusted to: P gen (t) = α2 × P load +P charge ,P bat (t) = 0 (discharge prohibited), where P charge is the charging power; ensure that the generator power covers the load and charging demand; the PMS sends instructions to the generator and BMS through the CAN bus: such as the generator target power: P_gen = 120kW (including 90kW load + 30kW charge), the upper limit of the battery charging power: P_charge_max = 30kW (dynamically adjusted by the BMS according to SOC and temperature).

[0076] If the load demand exceeds the maximum capacity of the generator (such as a sudden increase in load), the execution module temporarily reduces the power of the propulsion motor (such as restricting the speed through a frequency converter) to give priority to ensuring the power supply of key equipment (such as the navigation system).

[0077] If the battery temperature sensor detects a charging temperature rise (such as > 40°C), the cooling fan is triggered to start forcibly (controlled by a relay circuit) to maintain the battery temperature ≤ 35°C.

[0078] If an abnormal charging current is detected (e.g., fluctuation > ±10%), the BMS immediately cuts off the charging circuit and alarms via the CAN bus. If the generator is overloaded, the fault-tolerant module activates the backup power supply (e.g., another generator or supercapacitor) to temporarily make up for the power gap; if the BMS detects a fault in the SOC sensor, it switches to the voltage estimation mode (based on the open-circuit voltage curve) and triggers an HMI alarm to prompt "SOC estimation degraded".

[0079] The interface display: The HMI is updated to **"Low SOC Protection Mode"**, showing: Generator power: 120kW (load 90kW + charging 30kW), Battery status: SOC = 23%, Charging current = 30A, corresponding text; if the alarm indicator light (red) flashes to prompt "Low battery charge, charging"; allows a manual override command (e.g., forced battery discharge in an emergency), but a security password needs to be entered and the risk confirmed.

[0080] When receiving a battery charging instruction, the Battery Management System (BMS) first conducts a self-check and reads the State of Charge (SOC). If SOC ≥ 100%, the charging process is terminated; if SOC is below this threshold, the system will further read the battery temperature data. If the temperature exceeds the safe range (too high or too low), the BMS will suspend charging and activate the temperature regulation mechanism (such as turning on the cooling fan or heating device), while continuously monitoring the battery voltage to avoid overvoltage or short-circuit risks. Subsequently, the system determines the available charging sources (diesel generator, shore power, photovoltaic system, etc.): If a diesel generator is selected, its load status needs to be checked first to avoid overload; if shore power or photovoltaic power is available, the external power supply is preferentially connected. After the power source is confirmed, the BMS will double-check the relay status - disconnect the discharge relay to prevent short-circuit risks caused by parallel charging and discharging, and at the same time ensure that the charging relay is in the off state to avoid the impact of direct connection of high-voltage power. The charging execution stage includes the following processes: First, close the charging relay to connect the battery to the charging power source, and then adjust the electrical energy parameters according to the power source type. If it is a DC power source (such as photovoltaic or DC shore power), the DC / DC converter will perform voltage stabilization (for example, step down 400VDC to 48VDC); if it is an AC power source (such as a diesel generator or AC shore power), it needs to be converted to DC power through an AC-DC rectifier (such as rectify 380VAC to 400VDC), and then adjusted to the appropriate voltage through the DC / DC converter (such as 48VDC). The charging mode is dynamically switched according to SOC: The constant current mode (SOC ≤ 80%) realizes rapid charge replenishment by setting a constant current; the constant voltage mode (SOC 80% - 100%) gradually reduces the current to prevent overcharging; the trickle mode (SOC approaching 100%) maintains floating charge with a microcurrent to compensate for self-discharge losses. During the entire charging process, the BMS monitors the temperature / voltage / current parameters in real time, and ensures safety through three mechanisms: overvoltage protection (disconnect charging and alarm when exceeding the limit), overcurrent protection (automatically reduce the current or disconnect in case of abnormal current), and overtemperature protection (trigger cooling or emergency shutdown). When charging is completed, the system will execute a soft stop process after SOC reaches 100% or the current drops to the threshold: gradually reduce the current to make the battery adapt to the zero charge state → turn off the charging relay to cut off the circuit → switch to the floating charge mode or standby state according to the grid connection status.

[0081] When receiving a battery discharge instruction, the BMS also starts a self-check program and reads the SOC value: if SOC ≤ 20%, discharging is prohibited to prevent over-discharge damage. Meanwhile, the battery temperature is monitored, and when it is abnormal, discharging is paused and the temperature control device is started. The discharge strategy is dynamically adjusted according to the load demand: when the power demand is low, the discharge power is reduced to optimize the lifespan; when the power demand is high, the battery is jointly powered with a diesel generator. During the discharge execution stage, the charging relay needs to be closed first, and then the discharge relay is turned on. The power conversion device is used to match the load type: for DC loads (such as propulsion motors), the DC / DC converter boosts the battery voltage for output (such as 48 VDC → 400 VDC); for AC loads (such as ship equipment), the DC-AC inverter converts direct current into alternating current (such as 48 VDC → 380 VAC). The discharge process adopts multi-mode dynamic regulation: under normal conditions, a constant power output is maintained; short-term high power demand (such as ship acceleration) triggers a pulse discharge mode to avoid overloading of the diesel engine; when the SOC is lower than 30%, it automatically switches to the low power mode to extend the endurance. During the entire discharge cycle, the BMS continuously monitors key parameters and implements over-discharge / over-temperature protection to ensure that the system operates within the safety threshold, and finally returns to the standby state according to the instruction or condition change to wait for the next operation cycle.

[0082] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to only the specific embodiments. Obviously, many modifications and changes can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principle and practical application of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. An intelligent control system for ship main propulsion based on multi-source data fusion, comprising: A data processing and fusion module, a control decision-making module, and an execution control module; characterized in that: The data processing and fusion module is used to perform denoising processing and data integration on multi-source data. Specifically: Step 1: Use a Kalman filter to eliminate the noise of sensor data; Step 2: Integrate the data from different sensors; use a multi-layer perceptron for feature acquisition; input the multi-source sensor data into the MLP, calculate the neurons in the hidden layer through the hidden layer to obtain the final output; then perform data fusion according to the output result to obtain the fused data; The control decision-making module generates control instructions according to the system state and navigation requirements. Specifically: perform optimization analysis of the power distribution algorithm: S011: Determine the optimization objectives, including minimizing fuel consumption, optimizing battery charge and discharge, and ensuring that the power demand of the propulsion system is met; S012: Input parameter: load demand P load , maximum power of diesel generator Current generator power P gen (t), state of charge SOC(t) of the battery and charge and discharge efficiency η bat ; S013: Calculation process: S131: Perform the optimization objective function; S132: Perform the constraint conditions; S014: Through the dynamic power distribution strategy and charge and discharge scheduling strategy; if the SOC is higher than the set threshold, generate the first charge and discharge scheduling instruction, if the SOC is lower than the set threshold, generate the second charge and discharge scheduling instruction; When the load is less than the minimum value within the preset load demand range, generate a battery charging instruction, and when the load is greater than the maximum value within the preset load demand range, generate a battery discharging instruction; The execution control module converts the instructions generated by the control decision-making module into physical actions.

2. The intelligent control system for the main propulsion of a ship based on multi-source data fusion according to claim 1, wherein It also includes a multi-source data acquisition module and a human-computer interaction module; The multi-source data acquisition module collects the status data of diesel generators, batteries, and drive motor equipment in real time through current sensors, voltage sensors, speed sensors, and temperature sensors; collects environmental data and navigation status data, converts the analog signals output by the sensors into digital signals, and transmits the data to the main controller through the CAN bus and Ethernet; The human-computer interaction module provides a real-time status visualization interface, receives operation instructions, triggers alarm prompts, supports historical data query and fault diagnosis interaction, and allows emergency intervention.

3. The intelligent control system for the main propulsion of a ship based on multi-source data fusion according to claim 1, characterized in that, The specific process of calculating the neurons in the hidden layer through the hidden layer to obtain the final output is: Feature acquisition is performed using a multi-layer perceptron; multi-source sensor data is input into the MLP, and the output is set as: X = (x1, x2,..., x n ); x1, x2,..., x n represent n sensor data; then normalization processing is carried out, and the calculation of the hidden layer neurons is performed through the hidden layer: the formula is: w ij is the weight from the input data x i to the hidden layer neuron h j ; bj is the bias term; f(·) is the activation function; If there are L hidden layers, the calculation formula becomes: H (l) = f(W (l) H (l-1) + b (l) ), where H (l) is the output of the l-th hidden layer, W (l) and b (l) are the weights and biases of this layer, and the activation function f(x) performs a non-linear transformation; then the neurons in the output layer calculate: where y is the final output and σ(·) is the activation function.

4. The intelligent control system for ship main propulsion based on multi-source data fusion according to claim 3, characterized in that, The specific process of performing data fusion according to the output result to obtain the fused data is: Rule-based fusion: Calculate the battery SOC; Calculate the current State of Charge (SOC) based on the charge and discharge current and capacity of the battery; Weighted average fusion, calculate the weighted average value; Kalman fusion, GPS+IMU data fusion; Wind speed+ocean current data fusion.

5. The intelligent control system for the main propulsion of a ship based on multi-source data fusion according to claim 1, characterized in that, The specific process of performing the optimization objective function is: Define the objective function to minimize fuel consumption and battery usage cost: minF = C fuel ×P gen (t) + C battery ×P bat (t), where C fuel is the fuel cost coefficient and C battery is the battery usage cost.

6. The intelligent control system for the main propulsion of a ship based on multi-source data fusion according to claim 1, characterized in that The specific content of performing the constraint conditions is: Power balance constraint, diesel generator output power constraint, and battery SOC constraint; Power balance constraint: P load = P gen (t) + P bat (t); P bat (t) is positive or negative; Output power constraint of diesel generator: Battery SOC Constraint: SOC min ≤SOC(t)≤SOC max .

7. The intelligent control system for the main propulsion of a ship based on multi-source data fusion according to claim 1, wherein The specific process of the execution control module converting the instructions generated by the control decision-making module into physical actions is: When receiving charge-discharge scheduling instruction 1, increase the power of the diesel generator, adjust the fuel injection volume, increase the speed and output power, cut off the battery discharge circuit, reduce the discharge current, and trigger trickle charge protection; calculate the power distribution factor α1 and distribute the power between the generator and the battery; the frequency converter adjusts the IGBT switching frequency to match the power supply capacity; monitor the temperature and trigger the cooling system when it exceeds the limit; the HMI displays the power supply status and SOC information and performs fault handling when abnormal; When receiving charge-discharge scheduling instruction 2, increase the generator power to meet the load and charging requirements and charge the battery at the same time; the BMS cuts off the discharge circuit and only allows small-current discharge, and dynamically adjusts the charging mode; calculate the power distribution factor α2 and optimize the charging strategy between the generator and the battery; reduce the propulsion power when the load exceeds the limit to ensure power supply for critical equipment; monitor abnormal charging temperature rise and charging current and trigger the cooling fan or cut off the charging circuit; start the standby power supply and switch to the voltage estimation mode when the SOC sensor fails; the HMI updates the power distribution status in real time and provides abnormal alarm prompts; When receiving the battery charging instruction, the BMS self-checks the SOC and temperature, and adjusts cooling or heating when the temperature is abnormal; determine the charging source; disconnect the discharge relay; During the charging process, switch between constant current, constant voltage, and trickle modes according to the SOC; monitor the charging parameters and trigger overvoltage, overcurrent, and overtemperature protection; After charging is completed, perform a soft stop and enter the floating charge or standby state; When receiving the battery discharge instruction, the BMS self-checks the SOC and temperature and suspends discharge when the SOC is low or the temperature is abnormal; adjust the discharge power according to the load demand; turn off the charging relay and turn on the discharge relay to match the load type; adopt a constant power or pulse discharge mode to optimize the energy output; monitor the discharge parameters and trigger the abnormal protection mechanism.

Images