Energy management method and device for amphibious vehicle
By constructing an amphibious working condition identification model and a speed prediction model, combining a multi-objective comprehensive evaluation function, optimizing the torque distribution of engine and motors, the energy management problems of amphibious vehicles under different working conditions are solved, the balance of power responsiveness and fuel economy is achieved, and the energy conversion efficiency and vehicle operation stability are improved.
Patent Information
- Application Number
- CN202510767811.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-10
- Publication Date
- 2025-07-22
AI Technical Summary
Traditional energy management strategies cannot effectively respond to the power and economic needs of amphibious vehicles under different operating conditions, especially when the balance adjustment between low-speed and high-speed power output on land and difficult to achieve, resulting in low energy conversion efficiency and limited operating conditions adaptability.
By constructing an ocean-land working condition identification model and a speed prediction model, combining multi-objective comprehensive evaluation function, the torque distribution of the engine and motor is optimized, and the optimal energy distribution of the power system under different working conditions is achieved. The ELM and LSTM models are used for efficient training and prediction, and the penalty function and optimization factor are introduced for closed-loop feedback adjustment to ensure that the SOC is in the optimal working range.
It improves the speed prediction accuracy and energy distribution efficiency of amphibious vehicles under different operating conditions, meets the needs of power responsiveness and fuel economy, reduces engine response hysteresis, and ensures battery safety and energy management stability.
Smart Images
Figure CN120348274A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of energy management strategies for hybrid vehicles, and particularly to an energy management method and device for an amphibious vehicle. Background Art
[0002] The power system architecture design of wheeled amphibious rescue equipment faces significant technical bottlenecks, specifically manifested as follows: When maneuvering on land, complex terrains impose a rigid demand for low-speed high torque on the traction system, requiring the power system to have high torque reserves and dynamic response capabilities; while when navigating on water, to overcome fluid resistance and achieve stable cruising, the power unit needs to continuously provide power output in the medium and high rotational speed ranges. The power-speed characteristic curves of the power system for the two operating modes have an order-of-magnitude difference. The conventional single power source architecture cannot achieve adaptive adjustment of the power output characteristics under all operating conditions, resulting in a significant reduction in energy conversion efficiency and limited operating condition adaptability.
[0003] In response to the above technical bottlenecks, the hybrid architecture exhibits unique advantages: Through the coupling of the engine and the electric drive system, it can not only meet the torque redundancy requirements during low-speed land maneuvering but also adapt to the power fluctuation characteristics of water operations. However, the operating condition conversion faced by amphibious vehicles requires the power system to seek a balance between power performance and economy, and it is difficult to coordinate the control of the transient response and steady-state output of the power system. The wheeled amphibious vehicle with a hybrid architecture can better meet the power performance and economy indicators under different operating conditions in both land and water domains. However, traditional energy management strategies are mostly developed based on the continuous operating conditions of road vehicles and do not consider the operating condition differences between land and water domains, making it difficult to cope with the unique operating condition heterogeneity of amphibious equipment. Summary of the Invention
[0004] To solve the above problems, the present invention provides an energy management method for an amphibious vehicle, including the steps of: S1: Collect the historical speed sequence of the amphibious vehicle, and calculate and obtain a set of water-land operating condition characteristic parameters based on the historical speed sequence; S2: Construct a water-land operating condition identification model and a speed prediction model, input the set of water-land operating condition characteristic parameters into the water-land operating condition identification model to obtain the predicted water-land operating condition, input the historical speed sequence into the speed prediction model, and obtain the corresponding predicted speed based on the predicted water-land operating condition; S3: Construct a vehicle power model for the amphibious vehicle, input the predicted speed into the vehicle power model, and obtain a set of vehicle parameters, the predicted engine rotational speed, the predicted engine torque, the predicted motor rotational speed, and the predicted motor torque; S4: Construct a multi-objective comprehensive evaluation function based on the predicted water-land operating condition, the set of vehicle parameters, the predicted engine rotational speed, the predicted engine torque, the predicted motor rotational speed, and the predicted motor torque, and obtain the target engine torque and the target motor torque by solving the minimum value of the multi-objective comprehensive evaluation function. S5: Adjust the operation of the amphibious vehicle according to the engine target torque and the motor target torque.
[0005] Optionally: The types of water-land conditions include: low-speed maneuvering condition in water , medium-high speed cruising condition in water , low-speed off-road condition on land , medium-speed maneuvering condition on land and high-speed cruising condition on land .
[0006] Optionally, step S1 specifically includes: S11: Obtain a historical speed sequence, where the historical speed sequence includes a historical vehicle speed sequence and a historical navigation speed sequence; S12: Calculate the land distance feature, land vehicle speed feature, and land acceleration feature based on the historical vehicle speed sequence, and calculate the water distance feature, water navigation speed feature, and water acceleration feature based on the historical navigation speed sequence; S13: Use the land distance feature, land vehicle speed feature, land acceleration feature, water distance feature, water navigation speed feature, and water acceleration feature as the water-land condition feature parameter set.
[0007] Optionally, step S2 specifically includes: S21: Obtain a sample water-land condition feature parameter set and an initial ELM model, and train the initial ELM model with the sample water-land condition feature parameter set to obtain a water-land condition identification model; S22: Obtain a sample water-land condition set and an initial LSTM model, and train the initial LSTM model with the sample water-land condition set to obtain a speed prediction model, where the speed prediction model includes: model, model, model, model, and model; S23: Input the land distance feature, land vehicle speed feature, land acceleration feature, water distance feature, water navigation speed feature, and water acceleration feature in the water-land condition feature parameter set into the water-land condition identification model to obtain a predicted water-land condition, and the predicted water-land condition is or , where i takes an integer from 1 to 2, and j takes an integer from 1 to 3; S24: If the predicted water-land condition is , then input the historical navigation speed sequence into the model to obtain a predicted navigation speed; if the predicted water-land condition is , then input the historical vehicle speed sequence into the The model is used to obtain the predicted vehicle speed; the predicted sailing speed or the predicted vehicle speed is used as the predicted speed.
[0008] Optionally, step S21 specifically includes: S211: The initial ELM model includes an input layer, a hidden layer, and an output layer. The number of neurons in the hidden layer is set according to the number of nodes in the input layer and the regularization constant; S212: Initialize the parameters and weights of the initial ELM model; S213: Input the sample water-land condition characteristic parameter set into the initial ELM model. The sample water-land condition characteristic parameter set is transmitted through the input layer to the hidden layer, and the hidden layer matrix is calculated; the hidden layer matrix is transmitted to the output layer, and the target matrix and the sample predicted water-land condition are calculated; S214: Obtain the sample true water-land condition, and calculate the error function value according to the sample true water-land condition and the sample predicted water-land condition; S215: Adjust the parameters of the initial ELM model according to the error function value, and update the weights of the initial ELM model according to the hidden layer matrix and the target matrix; S216: Repeat steps S213 - S215 until the error function value converges to obtain the water-land condition identification model.
[0009] Optionally, step S22 specifically includes: S221: The sample water-land condition set includes: the sample water area low-speed maneuvering condition set, the sample water area medium-high speed cruising condition set, the sample land area low-speed off-road condition set, the sample land area medium-speed maneuvering condition set, and the sample land area high-speed cruising condition set; S222: In the offline state, pre-train the initial LSTM model respectively through the sample water area low-speed maneuvering condition set, the sample water area medium-high speed cruising condition set, the sample land area low-speed off-road condition set, the sample land area medium-speed maneuvering condition set, and the sample land area high-speed cruising condition set to obtain model, model, model, model and model; S223: Combine model, model, model, model and model into a speed prediction model.
[0010] Optionally, step S4 specifically includes: S41: Obtain the minimum SOC value, the maximum SOC value, and the current SOC value in the vehicle parameter set, and construct a penalty function according to the minimum SOC value, the maximum SOC value, and the current SOC value ; S42: Calculate the predicted motor power and predicted motor efficiency based on the predicted motor speed and the predicted motor torque ; obtain the charging equivalent factor, predicted power battery charging efficiency, low heating value of fuel, discharge equivalent factor, and predicted power battery discharge efficiency from the vehicle parameter set; S43: Based on the predicted motor torque , the predicted engine speed , the predicted motor power, the predicted motor efficiency, the charging equivalent factor, the predicted power battery charging efficiency, the low heating value of fuel, the discharge equivalent factor, and the predicted power battery discharge efficiency, calculate the initial instantaneous equivalent fuel consumption rate , and correct the initial instantaneous equivalent fuel consumption rate through a penalty function to obtain the optimized instantaneous equivalent fuel consumption rate ; S44: Calculate the predicted engine power based on the predicted engine speed and the predicted engine torque ; obtain the predicted fuel consumption rate from the vehicle parameter set; calculate the engine instantaneous fuel consumption rate based on the predicted engine torque , the predicted engine speed , the predicted fuel consumption rate, and the predicted engine power ; S45: Add the optimized instantaneous equivalent fuel consumption rate and the engine instantaneous fuel consumption rate to construct a fuel economy optimization function ; S46: Obtain the dimension-unifying factor, fuel economy optimization weight factor , dynamic performance weighting factor , total working condition duration, expected working condition speed, and actual vehicle output speed corresponding to the predicted water-land working condition from the vehicle parameter set; construct a dynamic response optimization function based on the total working condition duration, expected working condition speed, and actual vehicle output speed ; S47: Based on the dimension-unifying factor, fuel economy optimization weight factor , dynamic performance weighting factor , the fuel economy optimization function and the dynamic response optimization function , construct a multi-objective comprehensive evaluation function; S48: Solve the multi-objective comprehensive evaluation function by adjusting the values of the charging equivalent factor and the discharging equivalent factor, obtain the optimal values of the charging equivalent factor and the discharging equivalent factor that minimize the multi-objective comprehensive evaluation function, and calculate the engine target torque and the motor target torque according to the optimal values of the charging equivalent factor and the discharging equivalent factor.
[0011] The present invention also provides an energy management device for an amphibious vehicle, which is used to implement the energy management method of the amphibious vehicle. The device includes: An amphibious condition characteristic parameter acquisition module, which is used to collect the historical speed sequence of the amphibious vehicle and calculate the set of amphibious condition characteristic parameters according to the historical speed sequence; A speed prediction module, which is used to input the set of amphibious condition characteristic parameters into the amphibious condition identification model to obtain the predicted amphibious condition, input the historical speed sequence into the speed prediction model, and obtain the corresponding predicted speed based on the predicted amphibious condition; A rotational speed and torque prediction module, which is used to input the predicted speed into the vehicle power model to obtain the set of vehicle parameters, the predicted engine rotational speed, the predicted engine torque, the predicted motor rotational speed, and the predicted motor torque; A target torque acquisition module, which is used to construct a multi-objective comprehensive evaluation function according to the predicted amphibious condition, the set of vehicle parameters, the predicted engine rotational speed, the predicted engine torque, the predicted motor rotational speed, and the predicted motor torque, and obtain the engine target torque and the motor target torque by solving the minimum value of the multi-objective comprehensive evaluation function; An energy management module, which is used to adjust the operation of the amphibious vehicle according to the engine target torque and the motor target torque.
[0012] The present invention also provides an electronic device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the energy management method of the amphibious vehicle.
[0013] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the energy management method of the amphibious vehicle.
[0014] The present invention has the following beneficial effects: 1. Train the initial ELM model through the amphibious condition characteristic parameters to construct an amphibious condition identification model. The ELM model can process high-dimensional data, and during the training process, determine the output layer weights through an analytical method, avoiding the gradient descent and backpropagation processes in traditional neural networks, optimizing the training efficiency, and enabling the obtained amphibious condition identification model after training to accurately identify cluster-type conditions; 2. Build a speed prediction model based on the initial LSTM model. The LSTM model can predict speed and input it to the VCU in advance to reduce the response lag of the engine and ensure the safety of the battery. Train the LSTM model specifically according to different working conditions to obtain LSTM models suitable for different working conditions, and combine various LSTM models into a speed prediction model, enabling the speed prediction model to be applicable to both water and land conditions, improving the accuracy of speed prediction and the general applicability to different working conditions. 3. During the energy management process, introduce a penalty function compensation mechanism to dynamically correct the equivalent fuel consumption rate based on the state of charge level of the battery, forming a closed-loop feedback regulation to ensure that the SOC is maintained within the optimal working range. And introduce the fuel economy optimization weight factor and power performance weighting factor corresponding to different working conditions to participate in the construction of the multi-objective comprehensive evaluation function, enabling the multi-objective comprehensive evaluation function to meet the different requirements for power performance and economy of the amphibious vehicle under different working conditions. 4. Through numerical optimization of the charging equivalent factor and the discharging equivalent factor, obtain the optimal values of the charging equivalent factor and the discharging equivalent factor that minimize the multi-objective comprehensive evaluation function. Then, calculate the engine target torque and the motor target torque based on the optimal values of the charging equivalent factor and the discharging equivalent factor, and make the energy distribution of the amphibious vehicle reach the optimal according to the engine target torque and the motor target torque, enabling the operating state of the amphibious vehicle to meet the requirements of power response and fuel economy. Description of the Drawings
[0015] Figure 1 It is the flowchart of the method of the embodiment of the present invention; Figure 2 It is the schematic diagram of the training process of the water-land condition identification model; Figure 3 It is the schematic diagram of the training process of the speed prediction model; Figure 4 It is the curve graph of the penalty function; Figure 5 For The speed following effect diagram of the working condition; Figure 6 For The curve graph of the SOC and fuel consumption change of the working condition; Figure 7 For WBC 2 The speed following effect diagram of the working condition; Figure 8 For The curve graph of the SOC and fuel consumption change of the working condition; Figure 9 It is the speed comparison diagram of the highest navigation speed condition in the water mode; Figure 10It is a comparison chart of APU power under the maximum speed condition in the water mode; Figure 11 It is a comparison chart of battery current under the maximum speed condition in the water mode; Figure 12 It is a comparison chart of vehicle speed under the catapult start condition in the land mode; Figure 13 It is a comparison chart of APU power under the catapult start condition in the land mode; Figure 14 It is a comparison chart of battery current under the catapult start condition in the land mode; The realization, functional features and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the accompanying drawings. Detailed implementation manners
[0016] Next, in combination with the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.
[0017] Referring to Figure 1 , the present invention provides an energy management method for an amphibious vehicle, including the steps: S1: Collect the historical speed sequence of the amphibious vehicle, and calculate and obtain a set of water-land condition characteristic parameters according to the historical speed sequence; In some embodiments, based on the original condition data after noise filtering, three types of land driving conditions and two types of water navigation conditions that conform to the actual conditions of the amphibious vehicle are determined through cluster analysis.
[0018] The types of water-land conditions include: low-speed maneuvering condition in the water area , medium-high speed cruising condition in the water area , low-speed off-road condition in the land area , medium-speed maneuvering condition in the land area and high-speed cruising condition in the land area .
[0019] In some embodiments: Step S1 specifically includes: S11: Obtain the historical speed sequence, and the historical speed sequence includes the historical vehicle speed sequence and the historical navigation speed sequence; In some embodiments, a sliding window is adopted to collect the historical speed sequence online. The implementation of the sliding window: The present invention designs a data update rule using a fixed-length truncation strategy, and ensures the dynamic adaptability of the feature space by periodically eliminating historical data points and importing new sampled data. The speed data intercepted by the window is input into the land-water condition identification model after feature extraction, and the cluster condition identification is realized through a pre-trained ELM model. Among them, two parameters need to be set for the sliding window: the window size and the update time . is the sample size of the condition data within the window, is the period of the condition data update.
[0020] S12: Calculate the land distance feature, land speed feature, and land acceleration feature based on the historical vehicle speed sequence, and calculate the water area distance feature, water area speed feature, and water area acceleration feature based on the historical navigation speed sequence; In some embodiments, the calculation methods of the land distance feature, land speed feature, land acceleration feature, water area distance feature, water area speed feature, and water area acceleration feature are as follows: The land distance feature is the driving distance , and the dimension km ; The land speed features include: average speed , and the dimension km / h ; maximum speed , and the dimension km / h ; speed standard deviation , and the dimension km / h ; proportion of ≤10km / h , and the dimension ; proportion of 10~30km / h , and the dimension ; proportion of 30~60km / h , and the dimension ; proportion of ≥60km / h , and the dimension ; The land acceleration features include: maximum acceleration , and the dimension m / s² ; acceleration standard deviation , and the dimension m / s² ; average acceleration , and the dimension m / s² ; average deceleration , and the dimension m / s² ; minimum deceleration , and the dimension m / s² ; deceleration standard deviation , and the dimension m / s² ; proportion of acceleration section , Dimension ; Proportion of deceleration section , Dimension ; The water area distance feature is the driving distance , Dimension km ; The water area speed features include: average speed , Dimension km / h ; Maximum speed , Dimension km / h ; Speed standard deviation , Dimension km / h ; Proportion of ≤ 10 km / h , Dimension ; Proportion of 10 - 30 km / h , Dimension ; Proportion of 30 - 50 km / h , Dimension ; Proportion of 30 - 50 km / h , Dimension ; Proportion of ≥ 50 km / h , Dimension ; The water area acceleration features include: maximum acceleration , Dimension m / s² ; Acceleration standard deviation , Dimension m / s² ; Average acceleration , Dimension m / s² ; Proportion of constant speed section , Dimension ; Minimum deceleration , Dimension m / s² ; Deceleration standard deviation , Dimension m / s² ; Average deceleration , Dimension m / s² ; Proportion of acceleration section , Dimension ; Proportion of deceleration section , Dimension ; Among them, N is the number of discrete feature parameters, N a is the number of acceleration section feature parameters, N b is the number of deceleration section feature parameters.
[0021] S13: Take the land area distance feature, land area vehicle speed feature, land area acceleration feature, water area distance feature, water area speed feature and water area acceleration feature as the land - water working condition feature parameter set.
[0022] S2: Construct a land-water condition identification model and a speed prediction model. Input the set of land-water condition characteristic parameters into the land-water condition identification model to obtain the predicted land-water conditions. Input the historical speed sequence into the speed prediction model, and obtain the corresponding predicted speed based on the predicted land-water conditions. In some embodiments: Step S2 specifically includes: S21: Obtain a set of sample land-water condition characteristic parameters and an initial ELM model, and train the initial ELM model with the set of sample land-water condition characteristic parameters to obtain a land-water condition identification model. In some embodiments, the training process of the land-water condition identification model is as Figure 2 shown. The initial ELM model used to achieve cluster condition recognition according to the characteristic parameters belongs to a feedforward neural network, which is particularly suitable for processing high-dimensional data. Its core advantage lies in randomly generating the weights and biases of the hidden layer, and then determining the weights of the output layer through an analytical method, avoiding the gradient descent and backpropagation processes in traditional neural networks, significantly improving the training efficiency. A 7:3 ratio is used for sample allocation, and a benchmark test set is constructed to verify the cross-scenario adaptability of the algorithm.
[0023] Step S21 specifically includes: S211: The initial ELM model includes an input layer, a hidden layer, and an output layer. Set the number of neurons in the hidden layer according to the number of nodes in the input layer and the regularization constant. In some embodiments, the present invention adopts a single hidden layer structure for model selection, and calculates the number of neurons in the hidden layer to be 20 according to the formula.
[0024]
[0025] In the formula: - The number of nodes in the input layer; - The regularization constant.
[0026] S212: Initialize the parameters and weights of the initial ELM model. In some embodiments, the input weights and biases are randomly initialized, and this step is a prerequisite for model training.
[0027] S213: Input the set of sample land-water condition characteristic parameters into the initial ELM model. The set of sample land-water condition characteristic parameters is transmitted through the input layer to the hidden layer, and the hidden layer matrix is calculated; the hidden layer matrix is transmitted to the output layer, and the target matrix and the sample predicted land-water conditions are calculated. In some embodiments, the hidden layer matrix H output by the hidden layer can be expressed as:
[0028] In the formula: - Input of the sample set; - Activation function; and - Parameters of the hidden layer nodes; S214: Obtain the true water-land working condition of the sample, and calculate the error function value according to the true water-land working condition of the sample and the predicted water-land working condition of the sample; In some embodiments, the error function value can be expressed as:
[0029] In the formula: - True water-land working condition of the sample; - Predicted water-land working condition of the sample.
[0030] S215: Adjust the parameters of the initial ELM model according to the error function value, and update the weights of the initial ELM model according to the hidden layer matrix and the target matrix; In some embodiments, based on the least squares method, solve the output layer weights to minimize the prediction error. The analytical solution of the output weights is:
[0031] In the formula: - Hidden layer matrix; - Target matrix.
[0032] S216: Repeat steps S213 - S215 until the error function value converges to obtain the water-land working condition identification model.
[0033] In some embodiments, use the test set to evaluate the model to ensure that the model has good recognition ability on unseen data.
[0034] S22: Obtain the sample water-land working condition set and the initial LSTM model, and train the initial LSTM model through the sample water-land working condition set to obtain the speed prediction model. The speed prediction model includes: Model, Model, Model, Model and Model; In some embodiments, the training process of the speed prediction model is as Figure 3As shown in the figure, in the power demand model of the amphibious equipment, the speed time series characteristic is the core variable for power demand analysis. To avoid the potential threat of the dynamic system response lag to battery safety, a speed prediction technology with strong real-time performance needs to be introduced. In the field of land vehicles, the speed time series prediction method has been relatively mature. However, the dynamic characteristics of the water operation conditions are complex, and there is a lack of effective prediction means. This invention draws on the land speed time series prediction method, constructs a working condition feature matrix by extracting the speed characteristic parameters, and applies the mature instantaneous working condition identification technology (LSTM-based speed prediction model) in the research of land vehicles to the water operation conditions. This cross-domain method migration provides a feasible path for this invention to break through the bottleneck of water operation condition research.
[0035] Step S22 specifically includes: S221: The sample water-land operation condition set includes: the sample water area low-speed maneuvering operation condition set, the sample water area medium-high speed cruising operation condition set, the sample land area low-speed off-road operation condition set, the sample land area medium-speed maneuvering operation condition set, and the sample land area high-speed cruising operation condition set; S222: In the offline state, the initial LSTM model is pre-trained respectively through the sample water area low-speed maneuvering operation condition set, the sample water area medium-high speed cruising operation condition set, the sample land area low-speed off-road operation condition set, the sample land area medium-speed maneuvering operation condition set, and the sample land area high-speed cruising operation condition set to obtain model, model, model, model and model; In some embodiments, the pre-training steps of the initial LSTM model are as follows: 1) Data preprocessing: Normalize the 5 types of operating modes divided, map the speed / vehicle speed to the [0,1] interval, and eliminate the dimension difference; 2) Parameter configuration: Determine the basic parameter settings such as the number of network layers is 2, the number of neurons in a single layer is 96, the activation function is selected as Sigmoid, the training algorithm is LM, the learning rate is 0.005, the prediction method is selected as single-step multi-step, the gradient threshold is 1, and the Dropout rate is 0.15 according to the model convergence characteristics and sample size.
[0036] 3) Model training: Divide the training and test sets by 7:3, and construct a 3000-second mixed working condition training set to complete model iteration; 4) Real-time prediction: Input real-time operation data online, output the speed prediction value in the future time period, and use the preset water-land operation conditions to verify the prediction effect.
[0037] The single LSTM speed prediction model has good adaptability to typical working conditions, but its prediction accuracy is poor under certain specific working conditions. To further improve the prediction accuracy and generalization ability of the model, the present invention introduces the working condition identification mechanism described above into the speed prediction model, constructs an LSTM speed prediction model considering working condition identification, splices various working condition training sets from the original working condition data set, and trains the LSTM model separately in an offline state to obtain LSTM models applicable to different working conditions, including: the long short-term memory artificial neural network model for low-speed maneuvering working conditions in water areas ( model), the long short-term memory artificial neural network model for medium- and high-speed cruising working conditions in water areas ( model), the long short-term memory artificial neural network model for low-speed off-road working conditions on land ( model), the long short-term memory artificial neural network model for medium-speed maneuvering working conditions on land ( model), and the long short-term memory artificial neural network model for high-speed cruising working conditions on land ( model), and uses the corresponding LSTM model based on the working condition identification result in an online state.
[0038] S223: Combine the model, model, model, model, and model into a speed prediction model.
[0039] S23: Input the land distance feature, land vehicle speed feature, land acceleration feature, water area distance feature, water area speed feature, and water area acceleration feature in the land and water working condition feature parameter set into the land and water working condition identification model to obtain the predicted land and water working condition, and the predicted land and water working condition is or , where i takes an integer from 1 to 2, and j takes an integer from 1 to 3; S24: If the predicted land and water working condition is , input the historical speed sequence into the model to obtain the predicted speed; if the predicted land and water working condition is , input the historical vehicle speed sequence into the model to obtain the predicted vehicle speed; use the predicted speed or the predicted vehicle speed as the predicted speed.
[0040] S3: Construct the vehicle power model of the amphibious vehicle, input the predicted speed into the vehicle power model to obtain the vehicle parameter set, the predicted engine speed, the predicted engine torque, the predicted motor speed, and the predicted motor torque; S4: Construct a multi-objective comprehensive evaluation function based on the predicted water-land working conditions, vehicle parameter set, predicted engine speed, predicted engine torque, predicted motor speed, and predicted motor torque. By solving the minimum value of the multi-objective comprehensive evaluation function, obtain the target torque of the engine and the target torque of the motor. In some embodiments, the traditional energy management (ECMS) strategy generally has inherent defects such as insufficient multi-condition adaptability and poor energy efficiency performance in the water-land dual-mode operation. The core of the ECMS strategy adopted in the present invention lies in the unified quantification system of multi-energy consumption. This algorithm converts the consumption of various energy carriers during the operation of the hybrid power system into fuel consumption, and realizes the dynamic power ratio of the engine and the motor by establishing a multi-objective comprehensive evaluation function, providing a real-time optimization scheme for the energy scheduling of the hybrid amphibious vehicle under complex water-land dual-domain working conditions.
[0041] Step S4 specifically includes: S41: Obtain the minimum SOC value, the maximum SOC value, and the current SOC value from the vehicle parameter set, and construct a penalty function based on the minimum SOC value, the maximum SOC value, and the current SOC value ; In some embodiments, the ECMS strategy usually only considers a single index of equivalent fuel consumption and does not consider the constraint requirements for maintaining the state of charge of the power battery. Therefore, the present invention additionally constructs a penalty function compensation mechanism, dynamically corrects the equivalent fuel consumption rate according to the state of charge level of the battery, forms a closed-loop feedback regulation, and ensures that the SOC is maintained in the optimal working range.
[0042] Evaluate the influence of different combinations of charge-discharge equivalent factors on the maintenance of SOC, and propose the amplitude of the deviation of the SOC of the power battery from the efficient working range:
[0043] where is the minimum SOC value, is the maximum SOC value, is the current SOC value; When ΔSOC is greater than or equal to zero:
[0044] When ΔSOC is less than zero:
[0045] The curve of the penalty function is as Figure 4As shown, the mathematical model of the SOC penalty function constructed based on the above formula can calibrate the curve shape to improve the control accuracy by adjusting fitting coefficients such as a, b, c, d, and e. This model quantitatively characterizes the degree of deviation of the battery state of charge from the central value of the calibrated interval. Among them, the SOC offset is used as the core input parameter, and the gradient characteristics of the penalty function are configured through multi-dimensional coefficient optimization. The fitting coefficient adopted in the present invention is .
[0046] S42: Calculate the predicted motor power and predicted motor efficiency based on the predicted motor speed and the predicted motor torque . Obtain the charging equivalent factor, predicted power battery charging efficiency, low heating value of fuel, discharge equivalent factor, and predicted power battery discharge efficiency from the vehicle parameter set; S43: Calculate the initial instantaneous equivalent fuel consumption rate according to the predicted motor torque , the predicted engine speed , the predicted motor power, the predicted motor efficiency, the charging equivalent factor, the predicted power battery charging efficiency, the low heating value of fuel, the discharge equivalent factor, and the predicted power battery discharge efficiency, and correct the initial instantaneous equivalent fuel consumption rate through the penalty function to obtain the optimized instantaneous equivalent fuel consumption rate ; In some embodiments, when the power battery is in the charging mode and the discharging mode, the operating states of the corresponding ISG motors are completely different, and the energy flow directions have a large difference. It is impossible to accurately calculate the instantaneous fuel consumption when the power battery is charging and discharging simultaneously. It is necessary to discuss the charging and discharging conditions separately: Initial instantaneous equivalent fuel consumption rate in the case of power battery charging:
[0047] In the formula: - Charging equivalent factor; - Predicted motor power; - Predicted motor efficiency; - Predicted power battery charging efficiency; - Low heating value of fuel; Initial instantaneous equivalent fuel consumption rate in the case of power battery discharging:
[0048] In the formula: - Discharge equivalent factor; - Predicted power battery discharge efficiency; The operation logic of the dynamic adjustment mechanism of the power battery SOC penalty function is as follows: 1) When the battery SOC is in the middle of the reasonable range, the penalty coefficient approaches the reference value of 1. At this time, the equivalent fuel consumption rate remains unchanged, and the system maintains the conventional energy distribution mode.
[0049] 2) When the SOC approaches the lower threshold, the penalty factor increases significantly: in the discharge condition, the equivalent fuel consumption rate increases, triggering a positive adjustment of the SOC usage cost and suppressing the release of battery energy; in the charging condition, the equivalent fuel consumption rate decreases, forming a negative compensation for the SOC replenishment cost and accelerating the battery energy recharge.
[0050] 3) When the SOC approaches the upper boundary, the penalty factor decays in the opposite direction: in the discharge condition, the equivalent fuel consumption rate decreases, promoting the consumption of SOC through cost reduction; in the charging condition, the equivalent fuel consumption rate increases, restricting the storage of SOC by means of increased replenishment cost.
[0051] S44: Calculate the predicted engine power according to the predicted engine speed and the predicted engine torque Calculate the predicted engine power. Obtain the predicted fuel consumption rate from the vehicle parameter set. According to the predicted engine torque the predicted engine speed the predicted fuel consumption rate and the predicted engine power, calculate the instantaneous fuel consumption rate of the engine ; In some embodiments, according to the universal characteristic diagram of the engine, select the current output power and fuel consumption rate to calculate the instantaneous fuel consumption rate of the engine:
[0052] Where: - Predicted engine power; - Fuel consumption rate.
[0053] S45: Add the optimized instantaneous equivalent fuel consumption rate and the instantaneous fuel consumption rate of the engine to construct the fuel economy optimization function ; S46: Obtain the dimension-unifying factor, fuel economy optimization weight factor the dynamic performance weighting factor the total duration of the working condition, the expected speed of the working condition, and the actual output vehicle speed from the vehicle parameter set; according to the total duration of the working condition, the expected speed of the working condition and the actual output vehicle speed, construct the dynamic response optimization function ; S47: According to the dimension-unifying factor, fuel economy optimization weight factor the dynamic performance weighting factor the fuel economy optimization function Fuel economy optimization function and dynamic response optimization function , construct a multi-objective comprehensive evaluation function; In some embodiments, the construction of the multi-objective comprehensive evaluation function is as follows: Fuel economy optimization function:
[0054] Dynamic response optimization function:
[0055] In the formula: - The total duration of the working condition at time t; - The expected speed of the working condition at time t; - The actual output vehicle speed of the vehicle Linearly weight and fuse the above optimization functions, and then construct a multi-objective comprehensive evaluation function :
[0056] In the formula: - Fuel economy optimization weight factor; - Power performance weighting factor; - Dimension unification factor (used for dimensionless processing of parameters, and its value K = 0.001 is set according to the system characteristics).
[0057] In the present invention, different values of the optimization weight factor are adopted according to the different requirements of the amphibious vehicle for power performance and economy under different working conditions. For example, for the low-speed maneuvering working condition in the water mode characterized by the working condition, before the ship speed crosses the still water resistance peak, it is required to be able to frequently accelerate and decelerate sharply. The driver needs quick dynamic response and good vehicle speed following effect. Therefore, for the working condition, the proportion of should be increased. The values of the optimization weight factor under each working condition are shown in Table 1 below.
[0058] Table 1 Values of the corresponding optimization weight factor under different working conditions
[0059] S48: Solve the multi-objective comprehensive evaluation function by adjusting the values of the charging equivalent factor and the discharging equivalent factor, obtain the optimal values of the charging equivalent factor and the discharging equivalent factor that minimize the multi-objective comprehensive evaluation function, and calculate the engine target torque and the motor target torque according to the optimal values of the charging equivalent factor and the discharging equivalent factor.
[0060] In some embodiments, the system equivalent fuel consumption model and the power unit optimal torque solution model are as follows:
[0061]
[0062] wherein: - System instantaneous equivalent fuel consumption rate (g / s); - Optimized instantaneous equivalent fuel consumption rate (g / s); - Engine instantaneous fuel consumption rate (g / s); - Engine target torque (Nm) under the optimal equivalent fuel consumption condition; - Motor target torque (Nm) corresponding to the optimal point of equivalent fuel consumption; - Engine predicted speed; - Engine predicted torque; - Motor predicted torque; Boundary conditions:
[0063] After completing the construction of the optimization equation and the setting of constraint conditions, it is necessary to solve the engine and ISG torque distribution scheme corresponding to the minimized objective function, and convert the calculation results into the control parameters of the execution unit.
[0064] To further improve the dynamic regulation accuracy of the charge and discharge equivalent factor in the ECMS strategy, the present invention introduces the aforementioned operating condition identification model and the PSO algorithm to optimize the ECMS algorithm, and further develops an equivalent factor dynamic regulation system with environmental perception ability: by real-time identifying the characteristics of water and land operation scenarios, and according to the weight ratio of power demand and economic indicators, online optimize the numerical combination of charge and discharge factors.
[0065] Design of the charge and discharge factor optimization process: First, determine the initial value and optimization range of the charge and discharge factor array , and according to preliminary tests and the power system parameters of the amphibious vehicle, determine the optimization parameter table as shown in Table 2 below: Table 2 Optimization Parameter Table
[0066] The core control parameter configuration scheme of the PSO algorithm is shown in Table 3 for details.
[0067] Table 3 Key Parameter Settings of the Particle Swarm Optimization Algorithm
[0068] Through the MIL offline optimization method, the present invention can obtain the optimal parameter set for water and land conditions as shown in Table 4, and form an ECMS charge and discharge factor configurable parameter database to support online operating condition identification and dynamic matching.
[0069] Table 4 Optimal parameter set of charge and discharge factors under water and land conditions
[0070] S5: Adjust the operation of the amphibious vehicle according to the engine target torque and the motor target torque.
[0071] In some embodiments, to compare the dynamic performance and economic performance of the energy management method (predictive ECMS) designed by the present invention with those of the traditional rule-based energy management strategy, a simulation method is used to compare and verify the dynamic performance, SOC maintenance effect, and fuel economy performance of the two strategies, and further verify the advantages of the strategy designed by the present invention through real vehicle tests.
[0072] Simulation verification: (1) System modeling To conform to the longitudinal motion state and energy consumption of the actual amphibious vehicle during water operation, a simulation model of its dynamics and components related to the energy management strategy is built in the Matlab-Simulink software, and the model accuracy is optimized according to the data obtained from actual tests; for the accurate assessment requirements of the land operation conditions of the amphibious vehicle, an energy flow model of the hybrid system is built using AVL Cruise, and joint simulation is carried out through Cruise and Simulink to realize the real-time verification of the control strategy and multi-dimensional performance analysis under dynamic conditions.
[0073] (2) Simulation design The on-land mode simulation selects the on-land medium-speed maneuvering condition that has requirements for both dynamic performance and economy. The water mode selects , and the dynamic performance, SOC maintenance effect, and economic performance of the two strategies are compared and verified through simulation.
[0074] (3) Simulation results ① The speed following effect of the condition is as shown in Figure 5 . The change curves of SOC and fuel consumption of the condition are as shown in Figure 6 .
[0075] In terms of dynamic performance, when the required power exceeds the rated output power, the predictive ECMS strategy has a large speed error, but in other intervals, the speed error is basically controlled within 2 km / h Within this range, the cumulative error has also decreased by 40.48% compared to the regular strategy, and the dynamic optimization effect is obvious. In terms of SOC maintenance with the predictive ECMS strategy, the APU can dynamically adjust the charging and discharging current of the ISG according to the current working conditions to achieve a better SOC maintenance effect. In terms of economy, in terms of fuel economy, the fuel consumption rate of the predictive ECMS is relatively stable, and the final fuel consumption is low.
[0076] ② WBC 2 The speed following effect of the working conditions is as Figure 7 shown, The change curves of SOC and fuel consumption of the working conditions are as Figure 8 shown.
[0077] The speed following effect of the predictive ECMS is better than that of the regular type during high-speed cruising and medium-speed slightly fluctuating driving, especially in the speed range with a resistance peak. The SOC maintenance effect of the predictive ECMS strategy is significantly better than that of the regular type and can basically be maintained at about 50%. The fuel consumption of the predictive ECMS strategy is also significantly lower than that of the traditional regular strategy.
[0078] Real vehicle test: In view of the idealized constraints of modeling parameters and the simplified treatment of working conditions in the numerical simulation environment, it is difficult to fully reflect the true operating efficiency of the energy management strategy. To further verify the advantages of the predictive energy management strategy designed in the present invention over the traditional regular energy management strategy in terms of dynamic response, a real vehicle comparative test is carried out based on a certain hybrid amphibious prototype vehicle.
[0079] (1)Test conditions: The on-vehicle data acquisition system is implemented based on the PCAN protocol stack. The land test relies on the closed test site of Chibi Lushui Lake, and the water area verification selects the open water dynamic test platform constructed in the water test area of Chishui Lake.
[0080] (2)Test design: Build a real vehicle verification technology closed loop: Burn the regular strategy and the predictive ECMS strategy into the VCU main control unit through the data interface respectively. After completing the basic verification, use the dynamic charge and discharge adjustment method to accurately calibrate the battery SOC to the 50% reference value, and then perform heat engine pre-treatment to make the power assembly reach the steady-state working condition temperature to establish a standardized test environment for subsequent strategy performance evaluation.
[0081] ①Test design for the maximum speed condition in the water mode: The maximum speed of the amphibious vehicle in the water mode is designed to be 60 km / h. The test specialist accelerates from a speed of 0 to 60 km / h at full throttle, and compares the power output characteristics of the regular strategy and the predictive ECMS strategy according to the test data results to verify the optimization effect.
[0082] ②Onshore mode catapult start test design: Based on the performance target system established by the power system parameter matching design, set the acceleration performance benchmark for the onshore catapult mode: the time from full throttle 0 - 80 km / h ≤ 5 seconds. Conduct a comparison test under the emergency acceleration condition to evaluate the dynamic response optimization ability of the rule control and predictive ECMS strategies.
[0083] (3)Test results: ①Test results of the maximum speed condition in the water mode: Regarding the extreme speed performance of the rule-based and predictive ECMS energy strategies in the water mode, conduct a comparison test on the test vehicle, control the initial SOC of the battery to be 50% and the water environment conditions to be similar. The comparison of the speeds under the maximum speed condition in the water mode is as Figure 9 shown, the comparison of the APU powers under the maximum speed condition in the water mode is as Figure 10 shown, and the comparison of the battery currents under the maximum speed condition in the water mode is as Figure 11 shown.
[0084] The test vehicle can accelerate to the maximum speed at full throttle under both strategies. The acceleration time for the rule-based strategy to reach the maximum speed is 36.71 s, and the predictive ECMS only takes 33.12 s, an improvement of 9.93%. Moreover, during the acceleration process, there is a faster torque response and a safe and stable battery continuous output ability.
[0085] ②Test results of the catapult start condition in the onshore mode: Adjust the test vehicle to the onshore mode and conduct a catapult start comparison experiment on a closed road to analyze the dynamic performance of the rule-based and predictive ECMS energy management strategies. The comparison of the vehicle speeds under the catapult start condition in the onshore mode is as Figure 12 shown, the comparison of the APU powers under the catapult start condition in the onshore mode is as Figure 13 shown, and the comparison of the battery currents under the catapult start condition in the onshore mode is as Figure 14 shown.
[0086] In the catapult start condition in the onshore mode, the predictive ECMS can effectively reduce the engine response delay. Moreover, after the condition is identified as low-speed maneuvering, there is a more aggressive power output, and the acceleration is completed while ensuring the safety of the battery, with the dynamic performance improved by 17.81%.
[0087] The present invention also provides an energy management device for an amphibious vehicle, which is used to implement the energy management method of the amphibious vehicle. The device includes: An amphibious condition characteristic parameter acquisition module, which is used to collect the historical speed sequence of the amphibious vehicle and calculate the amphibious condition characteristic parameter set according to the historical speed sequence; A speed prediction module, which is used to input the set of water-land condition characteristic parameters into the water-land condition identification model to obtain the predicted water-land condition, input the historical speed sequence into the speed prediction model, and obtain the corresponding predicted speed based on the predicted water-land condition; A rotational speed and torque prediction module, which is used to input the predicted speed into the vehicle power model to obtain a set of vehicle parameters, the predicted engine rotational speed, the predicted engine torque, the predicted motor rotational speed, and the predicted motor torque; An objective torque acquisition module, which is used to construct a multi-objective comprehensive evaluation function according to the predicted water-land condition, the set of vehicle parameters, the predicted engine rotational speed, the predicted engine torque, the predicted motor rotational speed, and the predicted motor torque, and obtain the engine objective torque and the motor objective torque by solving the minimum value of the multi-objective comprehensive evaluation function; An energy management module, which is used to adjust the operation of the amphibious vehicle according to the engine objective torque and the motor objective torque.
[0088] An embodiment of the present application provides an electronic device, including a processor and a memory; the memory stores a computer program, wherein the computer program, when executed by the processor, implements the energy management method of the amphibious vehicle in any of the above solutions.
[0089] Specifically, the processor may include, for example, a general microprocessor, an instruction set processor, and / or a related chipset, and / or a dedicated microprocessor (such as an application specific integrated circuit (ASIC)), etc. The processor may also include on-board memory for caching purposes. The processor may be a single processing unit or multiple processing units for performing different actions of the method flow according to the embodiment of the present application.
[0090] The memory may be, for example, any medium capable of containing, storing, transmitting, propagating, or transferring instructions. For example, the memory may include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, components, or propagation media. Specific examples of the memory include: magnetic storage devices, such as magnetic tapes or hard disk drives (HDDs); optical storage devices, such as compact discs (CD-ROMs); may also be, such as random access memory (RAM) or flash memory; and / or wired / wireless communication links.
[0091] The present application also provides a computer-readable medium, on which a computer program is stored, and when the program is executed by the processor, the energy management method of the amphibious vehicle in any of the above solutions is implemented. The computer-readable medium may be included in the device / device / system described in the above embodiment; or may exist separately and not be assembled into the device / device / system. The above computer-readable medium carries one or more programs, and when the one or more programs are executed, the method according to the embodiment of the present application is implemented.
[0092] According to an embodiment of the present application, a computer-readable medium may be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, and this computer-readable medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted by any appropriate medium, including but not limited to: wireless, wired, optical cable, radio frequency signal, etc., or any suitable combination of the above.
[0093] Those skilled in the art can understand that the features recited in the various embodiments and / or claims of the present application can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly recited in the present application. In particular, without departing from the spirit and teachings of the present application, the features recited in the various embodiments and / or claims of the present application can be combined and / or combined in various ways. All such combinations and / or combinations fall within the scope of the present application. Therefore, the scope of the present application should not be limited to the above embodiments, but should be determined not only by the appended claims but also by the equivalents of the appended claims. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included within the protection scope of the present application.
Claims
1. An energy management method for an amphibious vehicle, characterized in that, Including the steps: S1: Collect the historical speed sequence of the amphibious vehicle, and calculate the set of water-land condition characteristic parameters based on the historical speed sequence; S2: Construct a water-land condition identification model and a speed prediction model. Input the set of water-land condition characteristic parameters into the water-land condition identification model to obtain the predicted water-land condition. Input the historical speed sequence into the speed prediction model, and obtain the corresponding predicted speed based on the predicted water-land condition; S3: Construct a vehicle power model for the amphibious vehicle, input the predicted speed into the vehicle power model, and obtain the set of vehicle parameters, the predicted engine speed, the predicted engine torque, the predicted motor speed, and the predicted motor torque; S4: Construct a multi-objective comprehensive evaluation function based on the predicted water-land condition, the set of vehicle parameters, the predicted engine speed, the predicted engine torque, the predicted motor speed, and the predicted motor torque. By solving the minimum value of the multi-objective comprehensive evaluation function, obtain the target engine torque and the target motor torque; S5: Adjust the operation of the amphibious vehicle according to the target engine torque and the target motor torque.
2. The energy management method for the amphibious vehicle according to claim 1, wherein: The types of water-land working conditions include: low-speed maneuvering condition in water area , medium-high speed cruising condition in water area , low-speed off-road condition on land , medium-speed maneuvering condition on land and high-speed cruising condition on land .
3. The energy management method of the amphibious vehicle according to claim 1, characterized in that Step S1 specifically includes: S11: Obtain the historical speed sequence, which includes the historical vehicle speed sequence and the historical navigation speed sequence; S12: Calculate the land distance characteristic, the land vehicle speed characteristic, and the land acceleration characteristic based on the historical vehicle speed sequence, and calculate the water distance characteristic, the water navigation speed characteristic, and the water acceleration characteristic based on the historical navigation speed sequence; S13: Use the land distance characteristic, the land vehicle speed characteristic, the land acceleration characteristic, the water distance characteristic, the water navigation speed characteristic, and the water acceleration characteristic as the set of water-land condition characteristic parameters.
4. The energy management method of the amphibious vehicle according to claim 3, characterized in that, Step S2 specifically includes: S21: Obtain the sample set of water-land condition characteristic parameters and the initial ELM model, and train the initial ELM model with the sample set of water-land condition characteristic parameters to obtain the water-land condition identification model; S22: Obtain the sample water-land working condition set and the initial LSTM model, and train the initial LSTM model with the sample water-land working condition set to obtain a speed prediction model, where the speed prediction model includes: model, model, model, model, and model; S23: Input the land distance feature, land vehicle speed feature, land acceleration feature, water area distance feature, water area speed feature, and water area acceleration feature in the land-water working condition characteristic parameter set into the land-water working condition identification model to obtain the predicted land-water working condition. The predicted land-water working condition is or , where i takes an integer from 1 to 2, and j takes an integer from 1 to 3; S24: If the predicted water-land working condition is , input the historical speed sequence into model to obtain the predicted speed; if the predicted water-land working condition is , input the historical vehicle speed sequence into model to obtain the predicted vehicle speed; use the predicted speed or the predicted vehicle speed as the predicted velocity.
5. The energy management method of the amphibious vehicle according to claim 4, characterized in that, Step S21 specifically includes: S211: The initial ELM model includes an input layer, a hidden layer, and an output layer. Set the number of neurons in the hidden layer according to the number of nodes in the input layer and the adjustment constant; S212: Initialize the parameters and weights of the initial ELM model; S213: Input the sample set of water-land condition characteristic parameters into the initial ELM model. The sample set of water-land condition characteristic parameters is transmitted through the input layer to the hidden layer, and the hidden layer matrix is calculated; the hidden layer matrix is transmitted to the output layer, and the target matrix and the sample predicted water-land condition are calculated; S214: Obtain the sample true water-land condition, and calculate the error function value according to the sample true water-land condition and the sample predicted water-land condition; S215: Adjust the parameters of the initial ELM model according to the error function value, and update the weights of the initial ELM model according to the hidden layer matrix and the target matrix; S216: Repeat steps S213 - S215 until the error function value converges to obtain the water-land condition identification model.
6. The energy management method of the amphibious vehicle according to claim 4, characterized in that, Step S22 specifically includes: S221: The sample water-land operation condition set includes: the sample water area low-speed maneuvering operation condition set, the sample water area medium-high speed cruising operation condition set, the sample land area low-speed off-road operation condition set, the sample land area medium-speed maneuvering operation condition set, and the sample land area high-speed cruising operation condition set; S222: In the offline state, pre-train the initial LSTM model respectively through the low-speed maneuvering condition set of the sample water area, the medium-high speed cruising condition set of the sample water area, the low-speed off-road condition set of the sample land area, the medium-speed maneuvering condition set of the sample land area, and the high-speed cruising condition set of the sample land area to obtain model, model, model, model, and model; S223: Combine model, model, model, model and model into a speed prediction model.
7. The energy management method of the amphibious vehicle according to claim 1, characterized in that Step S4 specifically includes: S41: Obtain the minimum SOC value, the maximum SOC value, and the current SOC value from the set of vehicle parameters, and construct a penalty function based on the minimum SOC value, the maximum SOC value, and the current SOC value ; S42: Predict the rotational speed of the motor and the predicted torque of the motor Calculate the predicted motor power and the predicted motor efficiency, and obtain the charging equivalent factor, the predicted charging efficiency of the power battery, the low calorific value of fuel, the discharge equivalent factor, and the predicted discharge efficiency of the power battery from the vehicle parameter set; S43: Predict the torque of the motor , predict the rotational speed of the engine , predict the motor power, predict the motor efficiency, charging equivalent factor, predict the charging efficiency of the power battery, low calorific value of fuel, discharge equivalent factor and predict the discharge efficiency of the power battery, and calculate the initial instantaneous equivalent fuel consumption rate , through the penalty function correct the initial instantaneous equivalent fuel consumption rate to obtain the optimized instantaneous equivalent fuel consumption rate ; S44: Predict engine speed and predicted engine torque to calculate the predicted engine power, obtain the predicted fuel consumption rate from the vehicle parameter set, and calculate the engine instantaneous fuel consumption rate based on the predicted engine torque , predicted engine speed , predicted fuel consumption rate, and predicted engine power ; S45: Optimize the instantaneous equivalent fuel consumption rate and the engine's instantaneous fuel consumption rate and add them together to construct a fuel economy optimization function ; S46: Obtain the dimension-unified factor, fuel economy optimization weight factor , power performance weighting factor , total working condition duration, expected working condition speed, and actual output vehicle speed of the vehicle; construct a dynamic response optimization function based on the total working condition duration, expected working condition speed, and actual output vehicle speed of the vehicle ; S47: Construct a multi-objective comprehensive evaluation function based on the dimension-unifying factor, the fuel economy optimization weight factor , the power performance weighting factor , the fuel economy optimization function and the power response optimization function ; S48: Solve the multi-objective comprehensive evaluation function by adjusting the values of the charging equivalent factor and the discharging equivalent factor, obtain the optimal charging equivalent factor value and the optimal discharging equivalent factor value that make the multi-objective comprehensive evaluation function the minimum value, and calculate the engine target torque and the motor target torque according to the optimal charging equivalent factor value and the optimal discharging equivalent factor value.
8. An energy management device for an amphibious vehicle, which is used to implement the energy management method of the amphibious vehicle according to any one of claims 1 to 7, characterized in that, The device includes: A water-land operation condition characteristic parameter acquisition module, which is used to collect the historical speed sequence of the amphibious vehicle and calculate the water-land operation condition characteristic parameter set according to the historical speed sequence; A speed prediction module, which is used to input the water-land operation condition characteristic parameter set into the water-land operation condition identification model to obtain the predicted water-land operation condition, input the historical speed sequence into the speed prediction model, and obtain the corresponding predicted speed based on the predicted water-land operation condition; A rotational speed and torque prediction module, which is used to input the predicted speed into the vehicle power model to obtain the vehicle parameter set, the predicted engine rotational speed, the predicted engine torque, the predicted motor rotational speed, and the predicted motor torque; A target torque acquisition module, which is used to construct a multi-objective comprehensive evaluation function according to the predicted water-land operation condition, the vehicle parameter set, the predicted engine rotational speed, the predicted engine torque, the predicted motor rotational speed, and the predicted motor torque, and obtain the engine target torque and the motor target torque by solving the minimum value of the multi-objective comprehensive evaluation function; An energy management module, which is used to adjust the operation of the amphibious vehicle according to the engine target torque and the motor target torque.
9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the energy management method of the amphibious vehicle according to any one of claims 1 to 7.
10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the energy management method of the amphibious vehicle according to any one of claims 1 to 7.