Control method and device for hydraulic system of double-clutch gear shifting executing mechanism and vehicle

Abstract

The invention discloses a control method and device for a hydraulic system of a double-clutch gear shifting executing mechanism and a vehicle. The method comprises the steps that real-time sensor signals in the vehicle running process are obtained; identifying each sensor signal to obtain a target characteristic signal related to the hydraulic system; and calling the trained neural network model, and outputting a predicted hydraulic system control strategy according to the sensor signal so as to control the hydraulic system. According to the control method for the hydraulic system of the double-clutch gear shifting executing mechanism of the hybrid electric vehicle, physical modeling is replaced by data driving, the precision is intelligently optimized and improved, the control effect is responded and optimized in real time, detailed parameters such as an oil way and various valve bodies are not needed, and the control efficiency is improved. Therefore, accurate control of the hydraulic system of the DCT gear shifting executing mechanism can be achieved.

Description

Technical Field

[0001] The present application belongs to the field of vehicle control technology, and in particular relates to a method and device for controlling a hydraulic system of a dual-clutch shift actuator, and a vehicle. Background Art

[0002] As a core actuator in the powertrain of both gasoline and hybrid vehicles, the clutch's dynamic performance directly impacts key performance indicators such as starting smoothness, shift quality, and energy transfer efficiency. The P2 architecture commonly used in hybrid vehicles utilizes a dual-clutch transmission (DCT) as a powertrain component, enabling multiple operating modes. In hybrid drive mode, the engine and electric motor work together, with the DCT automatically shifting gears as needed. In direct drive mode, the electric motor is deactivated or assists, with the engine directly driving the DCT to maintain efficient transmission. In recuperation mode, the electric motor generates electricity, recovering energy to the battery, while the DCT downshifts to enhance braking force.

[0003] like Figure 4 As shown in the figure, the vehicle's hybrid mode includes sensor signals such as command, vehicle speed, throttle opening, and clutch input shaft speed. These signals are transmitted to the HCU via a data acquisition system for analysis and processing. The decision signal is then transmitted to the hydraulic control unit, which controls the various actuator solenoid valves, ensuring that each component performs the specified action, thereby ensuring normal vehicle operation. The DCT shift actuation hydraulic system primarily consists of a hydraulic pump, hydraulic cylinder, control valve, clutch, and electronic control unit. The hydraulic pump, typically driven by an electric hydraulic pump, provides the necessary hydraulic oil flow and pressure to ensure the proper operation of the hydraulic cylinder and other components. In the DCT shift actuation hydraulic system, each clutch is equipped with a hydraulic cylinder containing a piston. The piston's movement is controlled by hydraulic oil pressure, thereby disengaging and engaging the clutch. The electronic control unit (hydraulic controller) manages and monitors the entire clutch hydraulic system. It issues control signals based on vehicle speed, engine and motor load, and accelerator pedal position. This control valve adjusts the flow direction, flow rate, and pressure of the hydraulic oil, thereby controlling the engagement and disengagement of the clutch.

[0004] Currently, most DCT shift actuator hydraulic control systems require an analysis of the operating principles of the main oil pressure control system, lubrication system, and clutch control system within the hydraulic system. Components such as the main oil pressure control valve, lubrication control valve, VBS solenoid valve, and VFS solenoid valve are then modeled to construct the hydraulic system. However, the modeling process is difficult due to the difficulty in measuring and acquiring information such as the oil pipeline, valve structure, and parameters, making the modeling process more complex. Consequently, traditional models are poorly adaptable to the complex operating conditions of real vehicles and struggle to cope with the dynamic changes in the hydraulic system under different environments and vehicle conditions, resulting in reduced accuracy in practical applications. Summary of the Invention

[0005] The present application provides a dual-clutch shift actuator hydraulic system control method, device and vehicle, aiming to solve the problem of insufficient hydraulic system control accuracy in the prior art.

[0006] In a first aspect, the present application provides a method for controlling a hydraulic system of a dual-clutch shift actuator, the method comprising: Obtain real-time sensor signals during vehicle operation; Identifying each of the sensor signals to obtain a target characteristic signal related to the hydraulic system; The trained neural network model is called to output a predicted hydraulic system control strategy based on the sensor signal to control the hydraulic system.

[0007] In a second aspect, the present application provides a dual-clutch shift actuator hydraulic system control device, the method comprising: Acquisition module, used to obtain real-time sensor signals during vehicle operation; The computing module is used to call the trained neural network model and output a predicted hydraulic system control strategy according to the sensor signal to control the hydraulic system.

[0008] In a third aspect, the present application provides an electronic control unit, which belongs to a vehicle, and the vehicle also includes a hydraulic system. The electronic control unit is used to execute the dual-clutch shift actuator hydraulic system control method as described above, and generate a control signal for controlling the hydraulic system.

[0009] In a fourth aspect, the present application provides a vehicle comprising a hydraulic system, and the electronic control unit, engine, and motor as described above.

[0010] Compared with the existing technology, this application proposes a control method for the hydraulic system of a dual-clutch shift actuator of a hybrid vehicle based on a machine learning method. It replaces physical modeling with data-driven, improves accuracy through intelligent optimization, and optimizes control effects through real-time response. It does not require detailed parameters such as oil circuits and various valve bodies, and can achieve precise control of the hydraulic system of the DCT shift actuator. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0012] Figure 1 A schematic diagram of the main process of the hydraulic system control method of the dual-clutch shift actuator provided in an embodiment of the present application; Figure 2 for Figure 1 A schematic diagram of a sub-flow diagram of a control method provided in the illustrated embodiment; Figure 3 for Figure 2 A schematic diagram of a sub-flow diagram of a control method provided in the illustrated embodiment; Figure 4 This is a schematic block diagram of the electronic control system of the dual-clutch shift actuator of the hybrid vehicle; Figure 5a This is the root mean square error curve of the training set before optimization; Figure 5b This is the loss function curve of the training set before optimization; Figure 6a This is the root mean square error curve of the training set after optimization; Figure 6b This is the loss function curve of the training set after optimization; Figure 7a for Figure 3 A fitting diagram of the training set results of the neural network model of the illustrated embodiment; Figure 7b for Figure 3 A fitting diagram of the validation set results of the neural network model of the illustrated embodiment; Figure 7c for Figure 3 A test set result fitting diagram of the neural network model of the illustrated embodiment; Figure 8 This is a comparison chart of simulation results of the dual-clutch shift actuator hydraulic system control method of this application compared with the traditional method; Figure 9This is a structural block diagram of a hydraulic system control device for a dual-clutch shift actuator of a hybrid vehicle provided in an embodiment of the present application; Figure 10 A structural block diagram of a vehicle provided in an embodiment of the present application. DETAILED DESCRIPTION

[0013] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0014] It will be understood by those skilled in the art that, unless expressly stated otherwise, the singular forms "a", "an", "said" and "the" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present disclosure refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, "connected" or "coupled" as used herein may include wireless connections or wireless couplings. The term "and / or" used herein includes all or any units and all combinations of one or more associated listed items.

[0015] See Figures 1 to 8 , Figure 1 This is a schematic diagram of the main process of a dual clutch shift actuator hydraulic system control method provided in an embodiment of the present application. The execution subject of the dual clutch shift actuator hydraulic system control method in the embodiment of the present application is a terminal control device, such as a vehicle controller, etc., which can be applied to hybrid vehicles. Figure 1 As shown, the method includes the following steps: S100: Acquire real-time sensor signals during vehicle operation.

[0016] The raw data collected by sensors throughout the vehicle is a "full signal", which contains massive data from various vehicle systems (such as power, chassis, body, entertainment, etc.), but the hydraulic system control model only requires features directly related to the hydraulic system.

[0017] S200, identifying each of the sensor signals to obtain a target characteristic signal related to the hydraulic system; The required target characteristic signals are identified and acquired, wherein the target characteristic signals are selected by a preset method. In one embodiment, the target characteristics include clutch 1 required pressure, clutch 2 required pressure, clutch 2 actual pressure, clutch 1 requested current, clutch 1 requested current change rate, clutch 2 requested current, clutch 2 requested current change rate, main oil circuit target current, main oil circuit pressure request, lubrication target current, lubrication requested flow, engine speed, and hydraulic oil temperature.

[0018] S300 , calling a trained neural network model and outputting a predicted hydraulic system control strategy according to the sensor signal to control the hydraulic system.

[0019] In an exemplary embodiment, Figure 2 As shown, the training method of the neural network model includes steps S310 to S330: S310, obtaining communication data of the hydraulic controller during the actual vehicle test operation; The raw data collected from actual vehicle testing is a "full signal," containing massive amounts of data from various vehicle systems (such as power, chassis, body, entertainment, etc.), but data-driven models (such as hydraulic system control models) only require features directly related to the hydraulic system.

[0020] S320, preprocessing the hydraulic controller communication data to obtain a training set for model training; Filter out data irrelevant to the hydraulic system (such as vehicle temperature and audio volume) to reduce data redundancy (for example, raw data may contain hundreds of signals, but after filtering, a certain number of core hydraulic signals are retained). Irrelevant signals increase the computational cost of model training and may even introduce noise, interfering with the model's learning of key patterns.

[0021] In an exemplary embodiment, Figure 3 As shown, step S320 includes steps S321 to S323: S321, extracting signals related to the hydraulic system from the hydraulic controller communication data, and removing irrelevant signals and constant value signals to form a first data set; The communication data of the hydraulic controller during the actual vehicle test was collected, the hydraulic-related signals were extracted, and the irrelevant and constant-value signals were eliminated to form a data set of 48 signals in total, as shown in Table 1.

[0022]

[0023] During the actual road driving experiment, the test vehicle was equipped with multiple types of sensors to collect different feature data in real time. The sampling frequencies of the sensors were different, as shown in Table 2.

[0024]

[0025] Therefore, it is necessary to align the timestamps of each data point. For different sensor data points within the same data file, we select the intersection of their timestamps and determine the start and end time points of this intersection as the synchronization period for all data. Then, we set a fixed sampling frequency within this synchronization period for sampling frequency synchronization, as shown in Table 3. After processing, the sampling interval is unified to 0.01s, laying the foundation for data-driven model training.

[0026]

[0027] S322, using a 3σ detection method to clean the first data set to remove outliers, to obtain a second data set; Since the extracted data may contain erroneous values ​​or missing values, resulting in low quality of the extracted data, in order to improve the data quality, the data needs to be cleaned to reduce interference factors in subsequent analysis.

[0028] The 3-Sigma Rule is an outlier detection method based on statistical principles. It uses the mean and standard deviation of the data to determine whether there are outliers. The principle of the 3-Sigma Rule is as follows: 1. Assume that the data set (or sample) is X, which contains n data points; 2. Calculate the mean (μ) and standard deviation (σ) of the data set; 3. According to the properties of the normal distribution, approximately 68% of the data points fall within the range of ±1σ of the mean, approximately 95% of the data points fall within the range of ±2σ of the mean, and approximately 99.7% of the data points fall within the range of ±3σ of the mean; 4. Using the 3σ principle, data points in the data set that differ from the mean by more than 3 standard deviations are defined as outliers; 5. Check whether there are outliers in the data set. If so, mark them as outliers.

[0029] The 48 signal data extracted in the first step are imported for detection. During the data processing, these outliers should be deleted to obtain the second data set.

[0030] S323. Use a random forest algorithm to perform feature selection on the second data set to determine important features that affect the control of the hydraulic system, so as to obtain a training set for model training.

[0031] Data feature selection is a key step in data preprocessing, which aims to identify the features from the raw data that are most helpful for model prediction.

[0032] Feature selection is a process of ranking features, using a series of rules to obtain the relative relationship of feature importance. The main process of tree model feature selection is as follows: 1. Dataset Preparation: First, you need to prepare a dataset containing features and a target variable. Features can be numerical or discrete, while the target variable is typically a continuous value or discrete category. 2. Tree Model Construction: This paper uses the random forest tree model algorithm to partition nodes based on the selected features. 3. Feature Evaluation and Ranking: Evaluate and rank the features according to the requirements of the feature selection method. By calculating the purity or information gain of each feature, you can determine the importance of the feature.

[0033] 4. Feature selection and model training: Rank features by importance and select the most important features as input features to train the tree model. In a decision tree, select important features as nodes for partitioning; in a random forest, select important features as the basis for subtree partitioning. 5. Model evaluation and optimization: Evaluate the model using a test dataset and optimize and adjust the model based on the evaluation results. If the model performance is not ideal, repeat the iterative process of feature selection and model training.

[0034] This paper uses a random forest to construct a supervised tree model, performing feature evaluation and ranking to identify important features with high influencing factors. Taking the clutch K1 tree model as an example, parameters affecting clutch K1 hydraulic pressure were first extracted from the collected data. The tree model was then used to analyze the resulting importance evaluation factors, ranked from highest to lowest, as shown in Table 4.

[0035]

[0036] 13 important characteristic signals are selected as shown in Table 5. The 13 important characteristic signals are used as model inputs, and the actual oil pressure of the clutch is used as the output of the model.

[0037]

[0038] As shown in Table 5, the important features include clutch 1 demand pressure, clutch 2 demand pressure, clutch 2 actual pressure, clutch 1 requested current, clutch 1 requested current change rate, clutch 2 requested current, clutch 2 requested current change rate, main oil circuit target current, main oil circuit pressure request, lubrication target current, lubrication request flow, engine speed and hydraulic oil temperature.

[0039] S330: Input the training set into a preset neural network model for training to obtain a trained neural network model.

[0040] Training data is fed into the network, and predictions are calculated through each layer. Loss calculation: Based on the predicted values ​​and the true labels, the loss value for the current batch is calculated using a loss function. Backpropagation: Based on the loss value, the gradient of each layer's parameters (i.e., the degree to which the parameters affect the loss) is calculated using the chain rule. Parameter update: The optimizer updates the network weights and biases based on the gradient and the learning rate (e.g., w = w - learning rate × gradient). Iterative training: Repeat the "forward propagation → loss calculation → backpropagation → parameter update" process until the loss converges or the preset number of training cycles is reached.

[0041] In an exemplary embodiment, an LSTM neural network is used to construct an initial network model, and a preset LSTM neural network model is trained by calling a MATLAB tool and using a dung beetle optimization algorithm to obtain a trained LSTM neural network model.

[0042] Specifically, the dataset division principle adopted is: the first 80% of the data is the training set, 80%-90% is the validation set, and the last 10% is the test set. The data needs to be standardized before training.

[0043] Taking the K1 data-driven model as an example, the data standardization process is as follows: 1. Separate the training set, validation set, and test set and mark the labels, respectively 、 、 、 、 、 ; 2. Get the mean and variance of the training set features and labels, respectively 、 as well as 、 、 ; 3. Standardize the data and record the standardized training set, validation set, and test set, which are 、 、 、 、 、 、 ; LSTM is an improvement on the traditional RNN. Its structure can alleviate problems such as excessive weight influence and vanishing gradients during RNN training, and it has better processing capabilities for long time series. Compared to RNN, LSTM has three additional gates: input gate, forget gate, and output gate.

[0044] Forget Gate Through the first forget gate, it is decided to discard the information from the current state input, and the activation function is used to selectively forget the information.

[0045] ; in is the weight parameter of the forget gate link input and hidden neurons, and the input feature is dimensional, the number of neurons in the hidden layer is ,but is a The matrix, The bias of the hidden layer neurons for the forget gate is the same as the number of hidden layer neurons.

[0046] Input Gate The input gate mainly decides which information to adopt and update.

[0047] ; ; It is also an activation function, and the output is limited to [-1,1]. are the parameters connecting the input and hidden layer neurons, are bias respectively.

[0048] Update cell status The new state is obtained by adding the information passing through the forget gate and the new information passing through the input gate.

[0049] ; Output Gate Sigmod determines the output part and then multiplies it with the updated cell state to get the final output.

[0050] ; ; ; The Dung Beetle Optimizer (DBO) is a new swarm intelligence optimization algorithm inspired by the ball-rolling, dancing, foraging, stealing, and reproduction behaviors of dung beetles. In nature, dung beetles roll their feces into a ball and use celestial bodies for navigation, keeping the ball rolling in a straight line. However, if there is no light at all, the dung beetle's path is no longer a straight line. Many natural factors, such as rough terrain, can also cause dung beetles to deviate from their original direction. Furthermore, dung beetles can reorient themselves through dance. Dung balls can also serve as a breeding ground for eggs. Furthermore, stealing occurs, with some dung beetles competing for food and taking it for themselves. The main principles of the DBO algorithm are as follows: 1. Simulating the behavior of dung beetles rolling dung balls: The DBO algorithm optimizes by simulating the behavior of dung beetles rolling dung balls. Dung beetles choose different directions and paths when rolling their dung balls to find the optimal path. The DBO algorithm treats the solution space of the optimization problem as the path of a dung ball. Individuals search for the optimal solution within the solution space, similar to the behavior of dung beetles rolling their dung balls.

[0051] 2. Dung beetle individuals and populations: In the DBO algorithm, each individual represents a solution, and they search for the optimal solution through their own behavior and interactions with other individuals. A population is composed of multiple individuals, who work together to improve search results.

[0052] 3. Rolling and Position Updates: In the DBO algorithm, individuals search the solution space through rolling and position updates. Each individual rolls according to its position and velocity in each iteration. The direction and distance of the roll are influenced by both the individual and the global optimal solution. The individual's position is updated based on the rolling results, gradually approaching the optimal solution.

[0053] 4. Fitness evaluation: In the DBO algorithm, the fitness of each individual is evaluated based on the value of the objective function. The better the fitness value, the closer the individual's position is to the optimal solution.

[0054] 5. Knowledge transfer and update: In the DBO algorithm, individuals transfer and update knowledge to improve search performance. Individuals can update their own behavior strategies by observing the positions and fitness values ​​of other individuals.

[0055] By simulating the rolling dung ball behavior of dung beetles, the interaction between individuals and populations, rolling and position updates, fitness evaluation, and knowledge transfer and updating, the DBO algorithm can search for optimal or near-optimal solutions in the solution space. It possesses a certain global search capability and convergence properties, making it suitable for solving a variety of optimization problems.

[0056] The neural network training is implemented by calling the MATLAB trainNetwork function, and the dung beetle optimization algorithm is used to adjust the number of hidden units, maximum training cycle, initial learning rate, and L2 parameter of the LSTM network. is the mean square error (MSE) of LSTM for the training set, test set, and validation set. The lower the MSE, the better. ; It has been verified that the LSTM training and optimization is carried out using the clutch K1 data as an example. The results of the training set root mean square error curve and loss function curve before and after optimization are as follows: Figure 5a 、 Figure 5b and Figure 6a 、 Figure 6b As shown, the fitting results of each set after LSTM algorithm optimization are as follows Figure 7a 、 Figure 7b and Figure 7c The accuracy indicators are shown in Table 6.

[0057]

[0058] By comparing the root mean square error curves and loss function curves of the training set before and after using the dung beetle optimization algorithm, we can see that when the number of iterations reaches 50, the optimized root mean square error of the training set is reduced by 31.8%, and the optimized loss function of the training set is reduced by 51.2%. Overall, the root mean square error of the training set and the loss function of the training set converge faster.

[0059] The trained neural network model is imported into Simulink, and the input and output interfaces are set and connected. The model serves as the clutch pressure controller and hydraulic model, inputting the control signal issued by the hydraulic controller during operation, thereby simulating the output of the clutch shift execution hydraulic control system pressure in real time.

[0060] The same input signal is imported into the neural network model and the AMEsim model for simulation, and their results are compared together. Figure 8 As shown in the figure, after calculation, the average absolute percentage error of the simulation results of the neural network model is 6.5991%, and the average absolute percentage error of the simulation results of the AMEsim model is 11.8718%. It can be seen that the average error percentage of the simulation results of the neural network model is reduced by 44.4% compared with the AMEsim model, with higher accuracy, and the response is faster during the oil filling stage, which is closer to the actual vehicle data.

[0061] This application utilizes a neural network to propose a control method for the hydraulic system of a dual-clutch shift actuator in a hybrid electric vehicle. This method eliminates the need for detailed parameters such as the oil circuit and various valve bodies. Instead, it extracts and processes multi-dimensional data from the hydraulic controller's communication signals, including clutch demand pressure, demand current, and its rate of change, main oil circuit target pressure and demand current, main oil circuit target lubrication flow and demand current, engine speed, and oil temperature. This method constructs an LSTM neural network model and integrates it with the Dung Mantis optimization algorithm to train the model, thereby achieving precise control of the DCT shift actuator hydraulic system. The model's accuracy is verified using bench-measured data. Because the extracted and trained data are all collected from actual vehicle testing, the established model maintains sufficient accuracy even under complex real-world operating conditions.

[0062] like Figure 9 As shown, the present application also proposes a hybrid vehicle dual-clutch shift actuator hydraulic system control device 100, which includes an acquisition module 101 and a calculation module 102. The acquisition module 101 is used to obtain real-time sensor signals during vehicle operation; the calculation module 102 is used to call a trained neural network model and output a predicted hydraulic system control strategy based on the sensor signal to control the hydraulic system.

[0063] like Figure 10 As shown, the present application also proposes an electronic control unit 402, which belongs to a vehicle 400. The vehicle also includes a hydraulic system 401. The electronic control unit 402 is used to execute the hybrid vehicle dual-clutch shift actuator hydraulic system control method as described above, and generate a control signal for controlling the hydraulic system.

[0064] In addition, an embodiment of the present application further provides a vehicle 400 , which includes a hydraulic system 401 , the above-mentioned electronic control unit 402 , an engine 403 , and a motor 404 .

[0065] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the flowcharts of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0066] The present application is described by way of some embodiments. It is understood by those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present application. In addition, under the guidance of the present application, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present application. Therefore, the present application is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of the present application are within the scope protected by the present application.

Claims

1. A dual-clutch shift actuator hydraulic system control method, characterized in that: The method comprises: Obtain real-time signals from various sensors during vehicle operation; Identifying each of the sensor signals to obtain a target characteristic signal related to the hydraulic system; The trained neural network model is called to output a predicted hydraulic system control strategy based on the sensor signal to control the hydraulic system.

2. The method according to claim 1, characterized in that The training method of the neural network model includes: Obtaining communication data of the hydraulic controller during actual vehicle test operation; Preprocessing the hydraulic controller communication data to obtain a training set for model training; The training set is input into a preset neural network model for training to obtain a trained neural network model.

3. The method according to claim 2, characterized in that The preprocessing of the hydraulic controller communication data to obtain a training set for model training includes: Extracting signals related to the hydraulic system from the hydraulic controller communication data, and removing irrelevant signals and constant value signals to form a first data set; Using a 3σ detection method to clean the first data set to remove outliers, so as to obtain a second data set; A random forest algorithm is used to perform feature selection on the second data set to determine important features that affect the control of the hydraulic system, so as to obtain a training set for model training.

4. The method according to claim 3, characterized in that The important features include at least one of clutch 1 required pressure, clutch 2 required pressure, clutch 2 actual pressure, clutch 1 requested current, clutch 1 requested current change rate, clutch 2 requested current, clutch 2 requested current change rate, main oil circuit target current, main oil circuit pressure request, lubrication target current, lubrication requested flow, engine speed and hydraulic oil temperature.

5. The method according to any one of claim 2, characterized in that Inputting the training set into a preset neural network model for training to obtain a trained neural network model includes: The preset LSTM neural network model is trained by calling the MATLAB tool and using the dung beetle optimization algorithm to obtain a trained LSTM neural network model.

6. A dual-clutch shift actuator hydraulic system control device, characterized in that: include: Acquisition module, used to obtain real-time sensor signals during vehicle operation; The computing module is used to call the trained neural network model and output a predicted hydraulic system control strategy according to the sensor signal to control the hydraulic system.

7. An electronic control unit, characterized in that: The electronic control unit belongs to a vehicle, which also includes a hydraulic system. The electronic control unit is used to execute the method according to any one of claims 1 to 5 and generate a control signal for controlling the hydraulic system.

8. A vehicle, characterized in that: The vehicle includes a hydraulic system, and the electronic control unit, engine, and motor according to claim 7.

Images