Hybrid vehicle cooling fan control method, system, hybrid vehicle and storage medium

Abstract

This invention relates to a control method, system, hybrid vehicle, and storage medium for a cooling fan in a hybrid vehicle, comprising the following steps: S1: acquiring test data and road data under different test conditions; S2: constructing a black-box model based on a recurrent neural network; S3: predicting the overall vehicle state and providing a fan control strategy. This invention can quickly and in real-time predict the optimal fan duty cycle under various operating conditions using a small amount of experimental data and real-vehicle sensor data trained by a neural network model.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of hybrid vehicle cooling fan control, and particularly relates to a hybrid vehicle cooling fan control method and system, a hybrid vehicle and a storage medium. BACKGROUND

[0002] Hybrid vehicles include engines, drive systems, electric motors, and electric control systems. In the case of a constant engine compartment size, the engine compartment layout is more compact, and the hybrid vehicle has multiple driving modes and complex working conditions. Using full working condition tests to calibrate different environmental temperatures, vehicle speeds, and different road conditions under different environmental temperatures, vehicle speeds, and different road conditions poses a huge challenge to engine compartment thermal management.

[0003] For example, patent document CN 110107391A discloses an engine fan post-operation control method, system, and electronic device, which includes collecting operating parameters before the vehicle is parked and turned off, obtaining a post-operation request level corresponding to each operating parameter, obtaining a duty cycle adjustment strategy corresponding to a high post-operation request level, and adjusting the duty cycle of the engine fan in different post-operation time periods according to the duty cycle adjustment strategy. This method can effectively reduce the deterioration of surrounding components caused by engine exhaust temperature, reduce the heat damage of engine exhaust temperature after parking and turning off, and reduce fan energy consumption and fan noise after parking and turning off. For example, patent document CN 112360787A discloses a fan management method for plug-in hybrid vehicles, which actively judges based on IPU circuit temperature signals, vehicle speed, environmental temperature, and other conditions, establishes a fan final request control method based on battery circuit cooling fan control method, motor circuit cooling fan control method, and post-operation cooling fan control method, ensures the normal operation of related new energy components, and appropriately reduces output to reduce energy consumption. However, the above two methods take a long time and cannot control in real time according to the actual vehicle state, making it difficult to meet the needs of vehicle development projects for quick results.

[0004] Therefore, it is necessary to develop a new hybrid vehicle cooling fan control method, system, hybrid vehicle, and storage medium. SUMMARY

[0005] The purpose of the present application is to provide a hybrid vehicle cooling fan control method, system, hybrid vehicle, and storage medium that can quickly and real-time predict the optimal fan duty cycle under each working condition through neural network training models.

[0006] In the first aspect, the hybrid vehicle cooling fan control method includes the following steps:

[0007] S1: Acquire test data and road data under different test conditions, wherein the test data includes vehicle speed, ambient temperature, engine intake air temperature, engine coolant temperature, exhaust temperature, average battery temperature, maximum battery temperature, battery cooling inlet temperature, motor oil temperature, motor coolant temperature, air conditioning operating status, air conditioning temperature, and fan duty cycle; the road data includes gradient.

[0008] S2: Construct a black-box model based on recurrent neural networks, specifically:

[0009] Using the test data and road data obtained in step S1 as inputs and the system state and energy consumption as outputs, a black-box model of a recurrent neural network is established. The weight matrix and threshold matrix of the neural network are obtained through training. The weight matrix and threshold matrix of the neural network represent the characteristic function relationship of the system state with respect to the control input according to the network structure.

[0010] S3: Predict the overall vehicle status and provide a fan control strategy, specifically:

[0011] Using a trained recurrent neural network black-box model, vehicle speed, gradient, ambient temperature, engine intake air temperature, engine coolant temperature, exhaust temperature, average battery temperature, maximum battery temperature, battery cooling inlet temperature, motor oil temperature, motor coolant temperature, air conditioning operating status, air conditioning temperature, and fan duty cycle are used as input conditions. The trained weight matrix and threshold matrix are used to predict the system's state Xpre and energy consumption in the next few time steps according to the network architecture. The linear combination of the system's thermal state performance index and energy consumption is used as the reward value. The optimal fan duty cycle in the next few time steps is found through reinforcement learning algorithm, and the cooling fan is controlled based on the optimal fan duty cycle.

[0012] Optionally, in step S3, while finding the optimal fan duty cycle within a few future time steps using a reinforcement learning algorithm, the input of the neural network is updated with the obtained optimal fan duty cycle for the next time step, the output of the neural network for the next time step is calculated, and the weight matrix and threshold matrix of the neural network are updated in combination with the actual measurement results of the system state at the next time step, so that the neural network can reflect the real state of the system as much as possible in real time.

[0013] Optionally, step S1 specifically includes:

[0014] Test conditions were selected based on the full-condition method.

[0015] Conduct tests according to the determined test conditions, and record test data and road data under different test conditions.

[0016] Optionally, the selection of test conditions based on the full-condition method specifically includes:

[0017] Set a series of ambient temperatures, vehicle speeds, and gradients, compile a number of working conditions corresponding to the full working condition method, and randomly select one-third to one-half of these working conditions as test working conditions.

[0018] Optionally, the neural network is set to dropout value of 0.5. During each computation, only half the parameters of the input features are used for training, which improves training efficiency and avoids overfitting.

[0019] Secondly, the control system for a hybrid vehicle cooling fan according to the present invention includes a controller and a memory, wherein the memory stores a computer-readable program, and the computer-readable program, when invoked by the controller, can execute the steps of the control method for the hybrid vehicle cooling fan as described in the present invention.

[0020] Thirdly, the hybrid vehicle described in this invention employs a hybrid vehicle cooling fan control system as described in this invention.

[0021] Fourthly, the present invention provides a storage medium storing a computer-readable program, which, when invoked, can execute the steps of the control method for a hybrid vehicle cooling fan as described in the present invention.

[0022] The present invention has the following advantages:

[0023] (1) Energy saving and consumption reduction: The vehicle thermal management system can intelligently adjust the start and stop of the cooling fan according to the actual heat load demand of the vehicle, avoiding unnecessary energy consumption. Compared with the continuously running cooling fan, on-demand adjustment can effectively reduce energy consumption and improve fuel economy.

[0024] (2) Noise reduction: The cooling fan will generate noise when it is running, but the vehicle thermal management system can control the start and stop of the cooling fan as needed, reducing unnecessary noise pollution and improving ride comfort.

[0025] (3) Increase component life: Continuously running cooling fans may cause premature wear of components, while adjusting as needed can reduce the usage time of cooling fans, extend the service life of components, and reduce maintenance costs.

[0026] (4) System stability: The vehicle thermal management system can intelligently adjust according to the actual heat load requirements of the vehicle, keep the temperature of the engine and other key components within a suitable range, and improve the stability and reliability of the vehicle system.

[0027] In summary, this invention can quickly and in real-time predict the optimal fan duty cycle under various operating conditions using a neural network training model based on a small amount of experimental data and real vehicle sensor data. While meeting the thermal management requirements of hybrid vehicles, it also reduces energy consumption (reducing energy consumption of hybrid models by 5% to 10% or more), and is more suitable for solving the complex multi-mode fan control problem of hybrid electric vehicles. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a flowchart of this embodiment;

[0030] Figure 2 This is a schematic diagram of the neural network in this embodiment. Detailed Implementation

[0031] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0032] like Figure 1 As shown in this embodiment, a method for controlling a cooling fan in a hybrid vehicle includes the following steps:

[0033] S1: Obtain test data and road data under different test conditions, specifically:

[0034] Test conditions were selected based on the full-condition method.

[0035] Conduct tests according to the determined test conditions, and record test data and road data under different test conditions.

[0036] In this embodiment, the selection of test conditions based on the full-condition method specifically refers to:

[0037] Set a series of ambient temperatures, vehicle speeds, and gradients, compile a number of working conditions corresponding to the full working condition method, and randomly select one-third to one-half of these working conditions as test working conditions.

[0038] The hybrid power operating condition was set with a series of ambient temperatures, including -20℃, -7℃, 0℃, 10℃, 25℃, 35℃, 40℃, and 49℃, totaling eight temperatures. Vehicle speeds were set at 20kph intervals from 0 to 160kph, totaling nine speeds. Gradients were set at 2 degrees intervals from 0 to 10 degrees, totaling six gradients. This resulted in 432 operating conditions corresponding to the full operating condition method. One-third of these conditions, or 144 conditions, were randomly selected for subsequent testing. This random selection ensures that the neural network can be applied across the entire operating condition range.

[0039] The test data includes vehicle speed, ambient temperature, engine intake air temperature, engine coolant temperature, exhaust temperature, average battery temperature, maximum battery temperature, battery cooling inlet temperature, motor oil temperature, motor coolant temperature, air conditioning operating status, air conditioning temperature, and fan duty cycle; the road data includes gradient. See Table 1.

[0040] Table 1:

[0041]

[0042] The above data was normalized, and then randomly divided into training and testing datasets.

[0043] S2: Construct a black-box model based on recurrent neural networks, specifically:

[0044] Using the test data and road data obtained in step S1 as input items, denoted as input sequence X = (x1, x2, ..., x14), 14 features are used as input layers, and system state and energy consumption are used as output items to establish a black box model of recurrent neural network. The weight matrix and threshold matrix of the neural network are obtained through training. The weight matrix and threshold matrix of the neural network represent the characteristic function relationship of system state with respect to control input according to the network structure.

[0045] In this embodiment, the neural network is set to dropout value of 0.5. For example... Figure 2 As shown, during each calculation, half of the parameters of the input features are randomly used for training, and a new half of the parameters are randomly used for the second training. Threshold control is performed using the safety threshold of the control target. This can improve training efficiency and avoid overfitting. The neural network is trained using the experimental data in S1 and converges after about 100,000 iterations.

[0046] S3: Predict the overall vehicle status and provide a fan control strategy, specifically:

[0047] Using a trained recurrent neural network (RNN) black-box model, vehicle speed, gradient, ambient temperature, engine intake air temperature, engine coolant temperature, exhaust temperature, average battery temperature, maximum battery temperature, battery cooling inlet temperature, motor oil temperature, motor coolant temperature, air conditioning operating status, air conditioning temperature, and fan duty cycle are used as input conditions. The trained weight and threshold matrices, based on the network architecture, predict the system's state Xpre and energy consumption over several future time steps. A linear combination of system thermal performance indicators (such as thermal comfort) and energy consumption is used as the reward value. A reinforcement learning algorithm is employed to find the optimal fan duty cycle over these future time steps, and the cooling fan is controlled based on this optimal duty cycle. Simultaneously, the obtained optimal fan duty cycle for the next time step is used to update the neural network input, and the output of the neural network at the next time step is calculated. The weight and threshold matrices of the neural network are then updated based on the actual measurement results of the system state at the next time step, ensuring that the neural network can reflect the true system state as realistically as possible in real time.

[0048] One-third of the operating conditions are randomly selected as test conditions. The predicted system state Xpre is compared with the energy consumption of the traditional fan control method to confirm the training effect. The energy consumption of the traditional fan control method is defined as Q0, and the energy consumption of the neural network training is defined as Q1. The energy consumption change is: (Q0-Q1) / Q0.

[0049] Experiments have shown that the prediction error of the system state Xpre using this method is within 2%, which meets the requirements. The fan control logic provided by this method can achieve real-time coordination of the system, and the energy consumption under test conditions is reduced by more than 5% to 10%.

[0050] In this embodiment, a control system for a hybrid vehicle cooling fan includes a controller and a memory. The memory stores a computer-readable program, which, when invoked by the controller, can execute the steps of the hybrid vehicle cooling fan control method described in this embodiment.

[0051] In this embodiment, a hybrid vehicle employs a hybrid vehicle cooling fan control system as described in this embodiment.

[0052] In this embodiment, a storage medium stores a computer-readable program that, when invoked, can execute the steps of the control method for the cooling fan of a hybrid vehicle as described in this embodiment.

[0053] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A control method for a cooling fan in a hybrid vehicle, characterized in that: Includes the following steps: S1: Acquire test data and road data under different test conditions, wherein the test data includes vehicle speed, ambient temperature, engine intake air temperature, engine coolant temperature, exhaust temperature, average battery temperature, maximum battery temperature, battery cooling inlet temperature, motor oil temperature, motor coolant temperature, air conditioning operating status, air conditioning temperature, and fan duty cycle; the road data includes gradient. S2: Construct a black-box model based on a recurrent neural network. Specifically, using the test data and road data obtained in step S1 as inputs and the system state and energy consumption as outputs, establish a black-box model of a recurrent neural network. Obtain the weight matrix and threshold matrix of the neural network through training. The weight matrix and threshold matrix of the neural network represent the characteristic function relationship of the system state with respect to the control input according to the network structure. S3: Predict the overall vehicle status and provide a fan control strategy. Specifically, using a trained recurrent neural network black-box model, the system takes vehicle speed, gradient, ambient temperature, engine intake air temperature, engine coolant temperature, exhaust temperature, average battery temperature, maximum battery temperature, battery cooling inlet temperature, motor oil temperature, motor coolant temperature, air conditioning operating status, air conditioning temperature, and fan duty cycle as input conditions. Based on the trained weight matrix and threshold matrix, the system predicts the state Xpre and energy consumption of the system in the next few time steps according to the network architecture. The linear combination of the system thermal state performance index and energy consumption is used as the reward value. The optimal fan duty cycle in the next few time steps is found through a reinforcement learning algorithm, and the cooling fan is controlled based on the optimal fan duty cycle. In step S3, while finding the optimal fan duty cycle within a few future time steps using the reinforcement learning algorithm, the input of the neural network is updated with the obtained optimal fan duty cycle at the next moment, the output of the neural network at the next moment is calculated, and the weight matrix and threshold matrix of the neural network are updated in combination with the actual measurement results of the system state at the next moment. The neural network is set to dropout value of 0.

5.

2. The control method for the cooling fan of a hybrid vehicle according to claim 1, characterized in that: Step S1 specifically involves: Test conditions were selected based on the full-condition method. Conduct tests according to the determined test conditions, and record test data and road data under different test conditions.

3. The control method for the cooling fan of a hybrid vehicle according to claim 2, characterized in that: The selection of test conditions based on the full-condition method is specifically as follows: Set a series of ambient temperatures, vehicle speeds, and gradients, compile a number of working conditions corresponding to the full working condition method, and randomly select one-third to one-half of these working conditions as test working conditions.

4. A control system for a cooling fan in a hybrid vehicle, characterized in that: It includes a controller and a memory, wherein the memory stores a computer-readable program that, when invoked by the controller, can execute the steps of the control method for the cooling fan of a hybrid vehicle as described in any one of claims 1 to 3.

5. A hybrid vehicle, characterized in that: The control system for the hybrid vehicle cooling fan as described in claim 4 is adopted.

6. A storage medium, characterized in that: It contains a computer-readable program that, when invoked, can perform the steps of the control method for the cooling fan of a hybrid vehicle as described in any one of claims 1 to 3.

Images