An intelligent monitoring-based liquid cooling system adaptive adjustment method

Abstract

The application belongs to the technical field of automatic control, and particularly relates to a liquid cooling system adaptive adjustment method based on intelligent monitoring, which comprises the following steps: S1: acquiring actual working temperatures of electric drive components in a vehicle, safety working temperatures of the electric drive components and an actual cooling liquid temperature of a cooling assembly respectively; S2: monitoring working temperature change trends of the electric drive components and calculating total heat dissipation demand information; S3: calculating target cooling liquid temperatures and target cooling liquid flow rates based on the total heat dissipation demand information and the actual cooling liquid temperature respectively, and calculating a PID control amount of the cooling assembly based on the target cooling liquid flow rate; and S4: when it is detected that an engine working temperature of the vehicle reaches a preset temperature threshold, controlling the cooling liquid to perform heat dissipation treatment on the drive components according to the target cooling liquid temperature and the PID control amount. The system can quickly respond, ensures that the electric drive components efficiently operate in a safety working temperature range, reduces performance attenuation caused by temperature fluctuation, and realizes a closed-loop thermal management control mode.

Description

Technical Field

[0001] This invention belongs to the field of automotive automatic control technology, specifically relating to an adaptive adjustment method for a liquid cooling system based on intelligent monitoring. Background Technology

[0002] New energy vehicles, leveraging the technological advantages of their electric drive systems, have demonstrated significant potential in energy conservation, emission reduction, and other aspects, making them a key focus of current automotive industry research and development. Their core electric drive systems (such as drive motors, power batteries, and power electronics) and power transmission systems generate substantial heat during operation. An efficient and stable thermal management system is crucial for ensuring vehicle performance, reliability, and safety. However, under conditions such as high temperatures or continuous high-load operation, existing thermal management systems, especially liquid cooling systems, often face challenges due to insufficient heat dissipation capacity. Once the temperature of critical components (such as the battery pack or drive motor) exceeds a safe threshold, the vehicle's thermal protection mechanism is triggered, forcing it into a power-limiting protection state. This state directly leads to a significant suppression of vehicle power output, severely impairing key dynamic performance indicators such as dynamic acceleration and top speed, while also significantly reducing driving smoothness and comfort, impacting the user experience.

[0003] The control strategies of existing liquid cooling systems in new energy vehicles are typically simple or lack sufficient intelligence. They often rely on preset fixed temperature thresholds or relatively crude on / off control or fixed flow regulation based on local information from a few temperature monitoring points. This passive or semi-passive control method struggles to perceive and respond to the complex and ever-changing thermal state of the vehicle in real time (such as temperature differences between different components, heat accumulation rate, sudden changes in ambient temperature, driving mode switching, etc.), and it is even less capable of predictive intervention before abnormal temperature rises. As a result, the system often intervenes passively only when thermal problems become severe, easily leading to unnecessary and frequent power limiting, or excessive cooling strategies adopted to avoid power limiting, resulting in energy waste.

[0004] Therefore, there is a need for an advanced method that can intelligently monitor the thermal state of the liquid cooling system and key components in real time, and adaptively adjust the cooling intensity and flow distribution, so as to minimize the occurrence of power limitation while ensuring system safety, and thus maintain the vehicle's optimal performance and driving experience. Summary of the Invention

[0005] To address the aforementioned problems in the existing technology, this invention provides an adaptive adjustment method for a liquid cooling system based on intelligent monitoring, which solves the problem that existing automotive cooling control systems, based on passive or semi-passive control methods, are unable to perceive and respond to the complex and ever-changing thermal state of the vehicle in real time and with high accuracy.

[0006] The objective of this invention can be achieved through the following technical solution: an adaptive adjustment method for a liquid cooling system based on intelligent monitoring, comprising the following steps: S1: Obtain the actual operating temperature, safe operating temperature, and actual coolant temperature of the electric drive components in the vehicle, respectively. The electric drive components include a power battery pack and a drive motor. S2: Monitor the operating temperature change trend of the electric drive component, and calculate the total heat dissipation requirement of the electric drive component based on the actual operating temperature, safe operating temperature and temperature change trend; S3: Calculate the target coolant temperature and target coolant flow rate based on the total heat dissipation demand information and the actual coolant temperature, and calculate the PID control quantity of the cooling component based on the target coolant flow rate; S4: When the engine operating temperature of the vehicle is detected to reach a preset temperature threshold, the coolant is controlled to dissipate heat from the drive components according to the target coolant temperature and the PID control quantity.

[0007] Preferably, in S4, when the engine operating temperature of the vehicle is detected to reach a preset temperature threshold, the method further includes controlling the electronic thermostat in the cooling assembly to be in an open state, and controlling the coolant to dissipate heat from the electric drive component based on the open state of the electronic thermostat.

[0008] Preferably, the open state includes: calculating the opening deviation value of the electronically controlled thermostat based on the actual coolant temperature and the target coolant temperature using a pre-trained learning model; The actual opening value of the electronically controlled thermostat is adjusted according to the opening deviation value to keep the coolant temperature at the target coolant temperature.

[0009] Preferably, the learning model is a neural network model, and the training data includes historical coolant temperature, historical target coolant temperature, and the opening deviation value of the electronically controlled thermostat required based on the historical coolant temperature and the historical target coolant temperature.

[0010] Preferably, S4 also includes: The operating temperature of the drive motor is monitored in real time. When the operating temperature of the drive motor is lower than the set first lower limit, the valve of the electronically controlled thermostat is kept closed. When the operating temperature of the drive motor reaches the set first upper temperature limit, the main valve of the electronically controlled thermostat is opened to cool the engine.

[0011] Preferably, in step S3, a pre-trained LSTM-GNN model is used to calculate the total heat dissipation requirements of the electric drive components based on the actual operating temperature, safe operating temperature, and temperature change trend. This includes constructing two independent LSTM networks to extract the temporal features of the two electric drive components respectively. The GNN network uses three electric drive components as nodes, and the node features are the output vectors of the corresponding LSTM branches. The edge weights are determined by the physical distance and thermal conductivity between two electric drive components.

[0012] Preferably, in the GNN network, the weights of neighboring nodes are dynamically allocated through attention coefficients, wherein the formula for calculating the attention coefficients is: ; It is the attention coefficient of node i to its neighboring node j. For normalization function, Let W be the model activation function and W be the feature transformation matrix. For the feature concatenation operation, the feature vectors of nodes i and j after W transformation are concatenated column by column to form a joint feature vector with dimension 2.

[0013] Preferably, in S3, the loss function of the LSTM-GNN model is modified, and the modification formula is as follows: ; in, This is the raw loss without adjusting for ambient temperature. The deviation between the predicted and actual total heat dissipation demand is calculated using the mean square error (MSE). For environmental correction items, The current ambient temperature is collected in real time by a temperature sensor located near the drive motor. The current ambient temperature is [temperature value], while the baseline temperature is [temperature value] during model training. It is the absolute deviation between the ambient temperature and the reference temperature. It is the temperature influence coefficient.

[0014] Preferably, in step S1, the actual operating temperature and coolant temperature are acquired using a temperature sensor array, and the acquired temperature data is preprocessed, including outlier filtering and time synchronization. The 3σ criterion is used to remove temperature sensor array jump values ​​caused by faults, and an interpolation algorithm is used to make up the removed jump values. Temperature data collected by the temperature sensor array is synchronized via the vehicle's CAN bus timestamp.

[0015] The beneficial effects of this invention are as follows: By correlating the actual temperature, desired operating temperature, and temperature change trend of the electric drive components, the system calculates the heat dissipation requirements of the electric drive components. Based on these requirements, it calculates the amount and rate of coolant. When the temperature of the electric drive components reaches a certain threshold, the system can react quickly to ensure that the electric drive components operate efficiently within a safe operating temperature range, reducing performance degradation caused by temperature fluctuations and achieving a closed-loop thermal management control method. Furthermore, by controlling the electronic thermostat, the system prevents coolant from flowing into the radiator before the engine warms up, effectively protecting the engine's thermal energy and solving the problem of low reliability in vehicle heat dissipation regulation, thereby improving the safety and reliability of vehicle operation. Attached Figure Description

[0016] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0017] Figure 1 This is a flowchart of the method steps of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0019] Please see Figure 1 This embodiment provides an adaptive adjustment method for a liquid cooling system based on intelligent monitoring, including the following steps: S1: Use a temperature sensor array to obtain the actual operating temperature, safe operating temperature, and actual coolant temperature of the electric drive components in the vehicle. The electric drive components include the power battery pack and the drive motor, including: Temperature sensor arrays are deployed on the power battery pack and drive motor to acquire the required temperatures. Digital temperature sensors are installed at the inlet and outlet of the vehicle's cooling circuit to collect the inlet and outlet temperatures of the coolant in real time, and the outlet temperature of the coolant within the cooling assembly is set as the actual coolant temperature. The safe operating temperature of the electric drive components is the temperature at which the electric drive components can operate safely, specifically set by the vehicle manufacturer based on the component model.

[0020] This also includes preprocessing the acquired temperature data, including outlier filtering and time synchronization. The 3σ criterion is used to remove temperature sensor array jump values ​​caused by faults, and an interpolation algorithm is used to make up the removed jump values. Temperature data collected by the temperature sensor array is synchronized via the vehicle's CAN bus timestamp.

[0021] S2: Monitor the operating temperature change trend of electric drive components, and calculate the total heat dissipation requirements of electric drive components based on the actual operating temperature, safe operating temperature and temperature change trend; The monitoring of the operating temperature change trend of the electric drive components includes: obtaining the temperature of the electric drive components using a sliding window, calculating the temperature rate based on the temperature change, using the temperature change rate as the temperature change trend, and calculating the total heat dissipation requirements of the electric drive components based on the actual operating temperature, safe operating temperature, and temperature change trend using a pre-trained machine learning model.

[0022] S3: Calculate the target coolant flow rate based on the total heat dissipation demand information and the actual coolant temperature, and calculate the target speed of the water pump in the cooling component based on the target coolant flow rate. In this step, the actual coolant temperature flowing through the engine from the vehicle's cooling components is first detected; this temperature reflects the current state of the engine. Simultaneously, a desired coolant temperature is also obtained; this is an ideal value set by the vehicle manufacturer based on the vehicle's operating efficiency and safety.

[0023] The calculated target coolant flow rate is filtered to prevent excessively drastic changes in the target value. Of course, when the target value shows an increasing or decreasing trend, different time constants can be selected to control whether the trend is increasing or decreasing. The aforementioned water pump can be configured as an electronic water pump, a mechanical water pump, etc. The coolant flow rate is determined by the pump's rotational speed. Therefore, after determining the target coolant flow rate, the system calculates the required pump speed (i.e., the target speed) based on the specific pump's flow characteristics, and generates a PID control input to control the pump's target speed. The electronically controlled thermostat is a key component controlling the coolant circulation path. Its main function is to prevent coolant from flowing into the radiator before the engine warms up, effectively protecting the engine's thermal energy. Once the engine reaches a certain operating temperature threshold, the thermostat will open. By allowing the engine to warm up quickly, engine wear can also be reduced. This temperature threshold is usually set based on the engine's safe operating temperature. In other words, during a cold start, the thermostat is usually closed, allowing the coolant to circulate within the engine for rapid heating; once the engine reaches its normal operating temperature, the thermostat opens, allowing coolant to flow through the radiator for heat dissipation. However, a simple electronically controlled thermostat cannot control the actual opening and closing based on the actual coolant temperature and the desired target coolant temperature.

[0024] Therefore, in one embodiment, the open state includes: performing learning model control on the electronic thermostat based on the actual coolant temperature and the target coolant temperature to obtain the opening deviation value of the electronic thermostat, thereby obtaining the opening deviation value of the electronic thermostat based on the learning characteristics of the learning model.

[0025] S4: When the engine operating temperature of the vehicle is detected to reach the preset temperature threshold, the electronic thermostat in the cooling system is turned on, and the coolant is controlled to dissipate heat from the electric drive components based on the water pump at the target speed and the turned-on electronic thermostat.

[0026] Here, the opening state of the electronic thermostat includes: calculating the opening deviation value of the electronic thermostat based on the actual coolant temperature and the target coolant temperature using a pre-trained learning model. The learning model can be a neural network model or other learning models. The training data for the learning model are historical coolant temperature, historical target coolant temperature, and the opening deviation value of the electronic thermostat required based on the historical coolant temperature and the historical target coolant temperature. The actual opening value of the electronically controlled thermostat is adjusted according to the opening deviation value to maintain the coolant temperature at the target coolant temperature. After the thermostat opens, the coolant circulates between the engine and radiator at the target speed driven by the water pump; after absorbing heat inside the engine, the coolant transfers heat to the atmosphere through heat exchange as it flows through the radiator, thus achieving heat dissipation.

[0027] This embodiment calculates the heat dissipation requirements of the electric drive component by associating its actual temperature, desired operating temperature, and temperature change trend. Based on the heat dissipation requirements, it calculates the amount and rate of coolant. When the temperature of the electric drive component reaches the temperature threshold, the system can react quickly to ensure that the electric drive component operates efficiently within the safe operating temperature range, reduce performance degradation caused by temperature fluctuations, and realize a closed-loop thermal management control method, thereby improving the safety and reliability of vehicle operation.

[0028] To prevent the cooling assembly from overcooling the drive component, in this embodiment, S4 further includes: The operating temperature of the drive motor is monitored in real time. When the operating temperature of the drive motor is lower than the set first lower limit, the valve of the electronically controlled thermostat is kept closed. When the operating temperature of the drive motor reaches the set first upper temperature limit, the main valve of the electronically controlled thermostat is opened to cool the engine.

[0029] To improve the dynamic calculation capability of the heat dissipation requirements of electric drive components and to better predict the system's heat dissipation requirements under complex operating conditions, in this embodiment, in S2, a pre-trained LSTM-GNN prediction model is used to dynamically predict the total heat dissipation requirements of the electric drive components by fusing temporal features and the coupling relationship between components. Specifically, this includes: S21: Obtain the extended dataset and preprocess the extended dataset and the dataset obtained in S1; The extended dataset includes: ambient temperature, vehicle speed, accelerator pedal opening, battery SOC (state of charge), and motor output power; Preprocessing includes normalizing the temperature sequence dataset obtained in S1 to eliminate dimensional differences and improve the fusion efficiency between data; data augmentation: expanding the dataset by time series shifting and adding noise to avoid overfitting of the model. Data set partitioning: Divide the dataset into training, validation, and test sets in a 7:2:1 ratio; S22: Design LSTM sub-networks, constructing three independent LSTM feature branches for the three electric drive components: power battery pack, drive motor, and motor controller, respectively: power battery pack LSTM, drive motor LSTM, and motor controller LSTM. LSTM temporal feature sequence is a deep learning structure specifically designed for processing time series data. Its core function is to capture long-term dependencies through memory units and gating mechanisms, and extract the dynamic evolution features of the input data from the time dimension. In this step, the LSTM network is used to extract temporal features from the acquired dataset, and the corresponding temporal feature vector is given to obtain the subsequent temporal feature sequence data. The three LSTM temporal feature branches are used to extract the temporal features of the three electric drive components respectively. Specifically, the input of the power battery pack LSTM is the temperature data of 10 consecutive time steps (each step is 1 second apart), and the output is a temporal feature vector with a dimension of 64. In the design of the LSTM network for the power battery pack LSTM, each branch contains two LSTM layers (with 128 and 64 hidden units respectively), and a Dropout layer (with dropout rate set to 0.2) is added to suppress overfitting. Finally, the feature dimension is compressed through a fully connected layer. The drive motor LSTM and the motor controller LSTM are also constructed in the same way. S23: Construct and initialize a GNN network, using the three electric drive components as nodes, with node features being the output vectors of the corresponding LSTM branches; initialize the network with edge weights based on the physical distance and thermal conductivity coefficient between the three electric drive components. Furthermore, using GAT (Graph Attention Network), the weights of neighboring nodes are dynamically allocated through attention coefficients. The attention coefficients of the GNN spatial feature aggregation layer are: ; in, It is the attention coefficient of node i to its neighboring node j, and its value ranges from 1 to 2. This is used to quantify the feature contribution of node j to node i. The larger the coefficient, the higher the weight of node j's features in the aggregation process of node i. This is a normalization function used to convert the attention scores of all neighboring nodes into a probability distribution, ensuring that the sum of the attention coefficients of all neighbors of node i is 1. The activation function for the model. The learnable parameter vector of the attention mechanism (with a dimension of 1×F, where F is the feature dimension) is used to perform a linear transformation on the concatenated feature vector and calculate the similarity score between node i and node j. By training adaptive learning, key correlation features between nodes (such as the consistency of temperature change trends) are captured; W is the feature transformation matrix (dimension 1). (where D is the original feature dimension of the node and F is the transformed feature dimension), used to map the original features of the node to a new feature space, enhancing the expressive power of the features. and These are the original feature vectors (of dimension D) of nodes i and j, including the temporal features of that node. For the feature concatenation operation, the feature vectors of nodes i and j after transformation by W (both with dimension F) are concatenated column by column to form a joint feature vector with dimension 2F, which is used to calculate the interaction features between the two nodes. S24: Combine temporal and spatial features to obtain fused features, and calculate the predicted total heat dissipation requirements of the electric drive components based on the fused features. When using a pre-trained LSTM-GNN model to calculate the total heat dissipation requirement of the electric drive components, to improve the prediction accuracy under different ambient temperatures, it is necessary to introduce an adjustment to the loss function based on ambient temperature. Here, the ambient temperature is set to the temperature surrounding the electric drive components. This is because in summer, when the air conditioning is on in the car, it affects the ambient temperature around the drive motor, thus having a cooling effect. Therefore, when calculating the total heat dissipation requirement, the model's loss function needs to be calculated based on the real-time ambient temperature around the drive motor. The loss function is: ; in, This is the raw loss without adjusting for ambient temperature. The deviation between the predicted and actual total heat dissipation demand is calculated using the mean square error (MSE). For environmental correction items, The current ambient temperature (unit: °C) is collected in real time by a temperature sensor located near the drive motor. The ambient temperature is 25℃ (default, normal operating conditions), which is the reference temperature for model training. It is the absolute deviation between the ambient temperature and the reference temperature, quantifying the difference between the current environment and the baseline operating conditions. It is the temperature influence coefficient (empirical value 0.02), which controls the amplification of the loss weight by the environmental temperature deviation, and is obtained by fitting experimental data (determined by analyzing the rate of change of heat dissipation efficiency at different temperatures). By modifying the loss function of the LSTM-GNN model, the model can automatically assign higher weights based on the ambient temperature samples around the drive motor during training, thus solving the problem of low prediction accuracy of traditional models under normal operating conditions.

[0030] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An adaptive adjustment method for a liquid cooling system based on intelligent monitoring, characterized in that: Includes the following steps: S1: Obtain the actual operating temperature, safe operating temperature, and actual coolant temperature of the electric drive components in the vehicle, respectively. The electric drive components include a power battery pack and a drive motor. S2: Monitor the operating temperature change trend of the electric drive component, and calculate the total heat dissipation requirement of the electric drive component based on the actual operating temperature, safe operating temperature and temperature change trend; S3: Calculate the target coolant temperature and target coolant flow rate based on the total heat dissipation demand information and the actual coolant temperature, and calculate the PID control quantity of the cooling component based on the target coolant flow rate; S4: When the engine operating temperature of the vehicle is detected to reach a preset temperature threshold, the coolant is controlled to dissipate heat from the electric drive components according to the target coolant temperature and the PID control quantity. In step S3, a pre-trained LSTM-GNN model is used to calculate the total heat dissipation requirements of the electric drive components based on the actual operating temperature, safe operating temperature, and temperature change trend. This includes constructing two independent LSTM networks to extract the temporal features of the two electric drive components. The GNN network uses two electric drive components as nodes, and the node features are the output vectors of the corresponding LSTM branches. The edge weights are determined by the physical distance and thermal conductivity between the two electric drive components. In GNN networks, neighboring node weights are dynamically assigned using attention coefficients. The formula for calculating the attention coefficients is as follows: ； It is the attention coefficient of node i to its neighboring node j. For normalization function, Let W be the model activation function and W be the feature transformation matrix. For the feature concatenation operation, the feature vectors of nodes i and j after transformation by W are concatenated column by column to form a joint feature vector of dimension 2. and These are the original feature vectors of nodes i and j; S3 includes modifying the loss function of the LSTM-GNN model, with the following formula: ； in, This is the raw loss without adjusting for ambient temperature. The deviation between the predicted and actual total heat dissipation demand is calculated using the mean square error (MSE). For environmental correction items, The current ambient temperature is collected in real time by a temperature sensor located near the drive motor. The current ambient temperature is [temperature value], while the baseline temperature is [temperature value] during model training. It is the absolute deviation between the ambient temperature and the reference temperature. It is the temperature influence coefficient.

2. The adaptive adjustment method for a liquid cooling system based on intelligent monitoring according to claim 1, characterized in that: In S4, when the engine operating temperature of the vehicle is detected to reach a preset temperature threshold, the system further includes controlling the electronic thermostat in the cooling assembly to be in an open state, and controlling the coolant to dissipate heat from the electric drive component based on the open state of the electronic thermostat.

3. The adaptive adjustment method for a liquid cooling system based on intelligent monitoring according to claim 2, characterized in that: The opening state includes: calculating the opening deviation value of the electronically controlled thermostat based on the actual coolant temperature and the target coolant temperature using a pre-trained learning model; The actual opening value of the electronically controlled thermostat is adjusted according to the opening deviation value to keep the coolant temperature at the target coolant temperature.

4. The adaptive adjustment method for a liquid cooling system based on intelligent monitoring according to claim 3, characterized in that: The learning model is a neural network model, and the training data includes historical coolant temperature, historical target coolant temperature, and the opening deviation value of the electronically controlled thermostat required based on the historical coolant temperature and the historical target coolant temperature.

5. The adaptive adjustment method for a liquid cooling system based on intelligent monitoring according to claim 2, characterized in that: S4 also includes: The operating temperature of the drive motor is monitored in real time. When the operating temperature of the drive motor is lower than the set first lower limit, the valve of the electronically controlled thermostat is kept closed. When the operating temperature of the drive motor reaches the set first upper temperature limit, the main valve of the electronically controlled thermostat is opened to cool the engine.

6. The adaptive adjustment method for a liquid cooling system based on intelligent monitoring according to claim 1, characterized in that: In S1, a temperature sensor array is used to acquire the actual operating temperature and the coolant temperature, respectively. The acquired temperature data is then preprocessed, including outlier filtering and time synchronization. The 3σ criterion is used to remove temperature sensor array jump values ​​caused by faults, and an interpolation algorithm is used to make up the removed jump values. Temperature data collected by the temperature sensor array is synchronized via the vehicle's CAN bus timestamp.

Images