A water temperature prediction and alarm system and method

By combining sensor modules and machine learning algorithm models, early prediction and automatic alarm of engine coolant temperature are achieved, solving the problem of time-consuming and labor-intensive manual positioning in existing technologies, and improving detection efficiency and safety.

CN117846766BActive Publication Date: 2026-11-17GUANGXI BOON TECH CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202311614783.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-11-29
Publication Date
2026-11-17
Estimated Expiration
2043-11-29

AI Technical Summary

Technical Problem

Existing technologies require multiple manual fault location operations when detecting excessively high engine coolant temperatures, consuming a large amount of manpower and resources, and subsequent repairs result in economic losses and safety hazards.

Method used

The system uses sensor modules to collect real-time engine data, and uses CNN+LSTM and Bayesian algorithms to optimize the model and predict water temperature. The results are presented in a visual chart and compared with standard data to automatically determine whether to trigger an alarm.

Benefits of technology

It enables early prediction of engine coolant temperature faults, reduces the inspection cost of fault detection, and improves safety and maintenance efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117846766B_ABST
    Figure CN117846766B_ABST
Patent Text Reader

Abstract

This invention discloses a water temperature prediction and alarm system, comprising: a sensor module for collecting real-time vehicle data and acquiring real-time engine parameters; an automatic controller for obtaining water temperature prediction results based on the real-time engine parameters; a visual device for presenting the water temperature prediction results in the form of visual charts; and peripheral devices for determining whether to issue an alarm based on a comparison between the water temperature prediction results and standard data. This invention monitors real-time engine operating parameters, uses an algorithm model to predict whether the engine will experience overheating in the future, and provides maintenance personnel with an intuitive display of the current engine status through visual charts, enabling timely and effective preventative measures and predictive maintenance. This system has significant research and practical value in terms of both engine safety and economy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of automotive technology, and in particular relates to a water temperature prediction and alarm system and method. Background Technology

[0002] In the vehicle system, the coolant temperature indicator light is used to display the temperature of the coolant inside the engine. If the coolant temperature indicator light remains constantly on, it means that the coolant temperature exceeds the specified value. In this case, it is necessary to analyze the fault from various aspects, such as wiring faults, component faults, poor radiator performance, and blockages.

[0003] Currently, the method of using a coolant temperature indicator light to determine if the engine coolant temperature is too high requires checking multiple points during fault location, which is labor-intensive and resource-intensive. Furthermore, by the time this method is used, the coolant temperature problem has already occurred, leading to significant economic losses, lengthy repair times, and substantial safety hazards. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a water temperature prediction and alarm system and method.

[0005] The present invention is achieved through the following technical solutions.

[0006] The present invention provides a water temperature prediction and alarm system, comprising:

[0007] Sensor module: Used to collect real-time data from the vehicle and obtain real-time engine parameters;

[0008] Automatic controller: used to obtain water temperature prediction results based on real-time engine parameters;

[0009] Visual devices: used to present water temperature prediction results in the form of visual charts;

[0010] Peripheral equipment: Used to determine whether to trigger an alarm by comparing the predicted water temperature with standard data.

[0011] Preferably, the sensor module includes a pressure sensor and a water temperature sensor;

[0012] The engine's real-time parameters include real-time vehicle weight data and real-time water temperature data;

[0013] The pressure sensor converts the mechanical pressure input to the vehicle into an electrical output signal and sends the real-time vehicle weight data to the automatic controller.

[0014] The water temperature sensor converts the water temperature of the car engine into an electrical output signal and sends the real-time water temperature data to the automatic controller.

[0015] Preferably, the automatic controller includes a data acquisition circuit, a data processing circuit, a signal control circuit, a data transmission circuit, a motor control circuit, a reset circuit, and a power supply circuit;

[0016] The data acquisition circuit sends the digital signals acquired by the sensor module to the data processing circuit;

[0017] After judging the digital signal, the data processing circuit converts the digital signal into a digital signal suitable for the signal control circuit, inputs an action command to the motor control circuit, and controls the external device.

[0018] The signal control circuit feeds back the current action command to the data processing circuit, causing the data processing circuit to issue a strobe signal suitable for the data transmission circuit, and the engine data is transmitted to the visual device and the peripheral device through the signal transmission circuit.

[0019] The power supply circuit provides power to the automatic controller; the reset circuit provides automatic reset upon power-on and manual reset.

[0020] Preferably, the signal control circuit uses a 74LS138 chip; the data processing circuit uses an STM32 microcontroller, specifically an STM32F103C8TX chip.

[0021] A water temperature prediction and alarm method includes the following steps:

[0022] Real-time data of the vehicle is collected through sensors to obtain real-time engine parameters;

[0023] The water temperature prediction result is obtained based on the real-time parameters of the engine.

[0024] The water temperature prediction results are presented in the form of visual charts;

[0025] The decision to trigger an alarm is based on a comparison between the predicted water temperature and the standard data.

[0026] Preferably, the specific steps for obtaining the water temperature prediction result based on the engine real-time parameters are as follows: inputting the engine real-time parameters into the algorithm model, performing calculations on the data in the algorithm model, and then obtaining the water temperature prediction result;

[0027] The training process of the algorithm model includes acquiring sample data using sensors, constructing a training dataset based on the sample data, training the CNN+LSTM algorithm model based on the training dataset, and optimizing the algorithm model using the Bayesian algorithm.

[0028] The sample data consists of multiple real-time vehicle data points and data required for feature segmentation indicators.

[0029] Preferably, the specific steps for constructing the training dataset based on the sample data include:

[0030] Data cleaning: Define the sample data as different features, input the different features into MATLAB, use MATLAB to plot the change curves of the same type of data, observe the change curves, and if there are abnormally high or low points on the curves, filter and remove them.

[0031] Feature segmentation: Divide the feature into more specific sub-indicators, verify the sub-indicators and enter them into the database, unify the data of the sub-indicators with the data of the feature, and judge whether there are any anomalies in the sub-indicators based on the feature.

[0032] Feature monitoring: Using the KS-test as a non-parametric test, the feature data is input into the graph of the cumulative probability density function. When the feature data has negative values, the x-axis is transformed by log to obtain the probability density function of the feature data, thus completing the construction of the training dataset. When the probability interval corresponding to the probability density function is abnormal, the training dataset is reconstructed.

[0033] Preferably, the specific steps for training the CNN+LSTM algorithm model based on the training dataset include:

[0034] Load the training dataset and count the number of data points in each class;

[0035] Assign tasks to the training dataset;

[0036] Constructing a CNN+LSTM framework;

[0037] Specify training options, including solver, network size, initial learning rate, and regularization term;

[0038] Build the algorithm model based on the CNN+LSTM framework and training options;

[0039] The algorithm model is tested by classifying the test data using the algorithm model. The classification accuracy of the model is tested by comparing the predictions on the retained test set with the true labels of each test observation.

[0040] Preferably, for the input sequence, the CNN+LSTM framework needs to specify an input layer whose input size matches the input data, and the network options should be set to the length of the shortest sequence in the training data;

[0041] The CNN+LSTM framework uses a two-dimensional CNN architecture, which includes four repeating blocks of convolution, batch normalization, ReLU and max pooling layers.

[0042] The CNN+LSTM framework allocates more filters in the second and fourth convolutional layers;

[0043] In the CNN+LSTM framework, the LSTM layer is assigned a hidden unit. If it is necessary to map the sequence to a single value for prediction, the LSMT layer option is set to output only the last time step.

[0044] In the CNN+LSTM framework, the size of the fully connected layer is equal to the number of classes. If the output needs to be converted into a probability vector, a sky softmax layer is required in the layer.

[0045] The CNN+LSTM framework includes layer classification.

[0046] Preferably, the specific steps of optimizing the algorithm model using the Bayesian algorithm include:

[0047] Construct an objective function; when the output of the objective function is a certain loss, the objective function is negative; when the actual parameters of the algorithm require integer input, the parameter types are specified in the objective function.

[0048] Define the parameter space using a dictionary.

[0049] Define the specific process for optimizing the objective function;

[0050] Define the verification function and perform optimization operations.

[0051] The beneficial effects of this invention are as follows:

[0052] 1. An algorithm model built using sensor data such as pressure and water temperature as input can more effectively and quickly reflect the current engine status.

[0053] 2. Most existing technologies cannot predict water temperature faults. However, an algorithm model for predicting water temperature faults, built using machine learning, can provide early warnings of water temperature faults, thereby avoiding potential engine failures in the future.

[0054] 3. Existing technologies mostly rely on manual troubleshooting to detect engine overheating faults. This invention uses visual charts to compare current data with standard data, which can help maintenance personnel quickly locate the fault and reduce inspection costs during fault detection. Attached Figure Description

[0055] Figure 1 This is a schematic diagram of the system structure according to an embodiment of the present invention;

[0056] Figure 2 This is a schematic diagram of the automatic controller structure according to an embodiment of the present invention;

[0057] Figure 3 This is a flowchart of a method according to an embodiment of the present invention;

[0058] Figure 4 This is a flowchart of the algorithm model training process according to an embodiment of the present invention;

[0059] Figure 5 This is a flowchart illustrating the process of constructing a training dataset based on sample data according to an embodiment of the present invention.

[0060] Figure 6 This is a flowchart of the process of training a CNN+LSTM algorithm model based on a training dataset according to an embodiment of the present invention;

[0061] Figure 7 This is a flowchart of the algorithm model optimization using the Bayesian algorithm according to an embodiment of the present invention;

[0062] Figure 8 This is a schematic diagram illustrating the real-vehicle verification effect of an embodiment of the present invention. Detailed Implementation

[0063] The technical solution of the present invention is further described below, but the scope of protection is not limited to what is described.

[0064] like Figure 1 As shown, a water temperature prediction and alarm system includes:

[0065] Sensor module: Used to collect real-time data from the vehicle and obtain real-time engine parameters;

[0066] The sensor module includes a pressure sensor and a water temperature sensor;

[0067] The engine's real-time parameters include real-time vehicle weight data and real-time water temperature data;

[0068] The pressure sensor converts the mechanical pressure input to the vehicle into an electrical output signal and sends the real-time vehicle weight data to the automatic controller.

[0069] The water temperature sensor converts the water temperature of the car engine into an electrical output signal and sends the real-time water temperature data to the automatic controller.

[0070] In this embodiment, the power supply voltage of the water temperature sensor is a 5V power supply provided by the engine unit, and the signal returned to the engine unit is a linearly changing signal of 1.3V-3.8V.

[0071] Automatic controller: used to obtain water temperature prediction results based on real-time engine parameters;

[0072] like Figure 2 As shown, the automatic controller includes a data acquisition circuit, a data processing circuit, a signal control circuit, a data transmission circuit, a motor control circuit, a reset circuit, and a power supply circuit.

[0073] The data acquisition circuit sends the digital signals acquired by the sensor module to the data processing circuit;

[0074] After judging the digital signal, the data processing circuit converts the digital signal into a digital signal suitable for the signal control circuit, inputs an action command to the motor control circuit, and controls the external device.

[0075] The signal control circuit feeds back the current action command to the data processing circuit, causing the data processing circuit to issue a strobe signal suitable for the data transmission circuit, and the engine data is transmitted to the visual device and the peripheral device through the signal transmission circuit.

[0076] The signal control circuit uses a 74LS138 chip; the data processing circuit uses an STM32 microcontroller, specifically an STM32F103C8TX chip.

[0077] The power supply circuit provides power to the automatic controller; the reset circuit provides automatic reset upon power-on and manual reset.

[0078] Visual devices: used to present water temperature prediction results in the form of visual charts;

[0079] Peripheral equipment: Used to determine whether to trigger an alarm by comparing the predicted water temperature with standard data.

[0080] like Figure 3 As shown, a water temperature prediction and alarm method includes the following steps:

[0081] Real-time data of the vehicle is collected through sensors to obtain real-time engine parameters;

[0082] The water temperature prediction result is obtained based on the real-time parameters of the engine.

[0083] The specific steps for obtaining the water temperature prediction result based on the engine real-time parameters are as follows: input the engine real-time parameters into the algorithm model, and then perform calculations and processing on the data in the algorithm model to obtain the water temperature prediction result;

[0084] like Figure 4 As shown, the training process of the algorithm model includes acquiring sample data using sensors, constructing a training dataset based on the sample data, training the CNN+LSTM algorithm model based on the training dataset, and optimizing the algorithm model using the Bayesian algorithm.

[0085] The sample data consists of multiple real-time vehicle data points and data required for feature segmentation indicators.

[0086] like Figure 5 As shown, the specific steps for constructing the training dataset based on the sample data include:

[0087] Data cleaning: Define the sample data as different features, input the different features into MATLAB, use MATLAB to plot the change curves of the same type of data, observe the change curves, and if there are abnormally high or low points on the curves, filter and remove them.

[0088] Feature segmentation: Divide the feature into more specific sub-indicators, verify the sub-indicators and enter them into the database, unify the data of the sub-indicators with the data of the feature, and judge whether there are any anomalies in the sub-indicators based on the feature.

[0089] Specifically, in this embodiment, the subdivided indicators of feature quantity A include: cargo box dimensions a, loaded cargo b, loading tonnage c, and total vehicle weight d. The length, width, and height of the cargo box (a) are measured, and the data is recorded in a table or provided to the user for verification. The loaded cargo (b) is reviewed, and cargo descriptions or photos are obtained from the cargo provider for later comparison and inspection. The vehicle is calibrated on a weighbridge both empty and fully loaded, and the results are verified against the standard loading tonnage (c) and total vehicle weight (d), with the data recorded in a table for feedback. Finally, the data for a, b, c, and d are unified into the data for feature quantity A. The end user only needs to use the normality or abnormality of feature quantity A to determine if there are any problems in this process.

[0090] The detailed indicators of characteristic quantity B include: engine thermostat opening temperature e, engine coolant temperature f collected by the engine coolant temperature sensor, radiator operation g, and fan operation h. Among these, e, f, g, and h are real-time data, and reasonable reference values ​​need to be determined based on characteristic quantity B and the model of the sensor collecting this data. The engine thermostat opening temperature e is obtained by photographing its exterior and providing the user with a sample to verify the parameter values ​​of the model under normal operating conditions. The results are compared with those measured by the vehicle and existing thermostats. Small deviations are generally due to environmental factors and characteristic quantity B and can be ignored; large deviations indicate a problem with the result of e. The coolant temperature f collected by the coolant temperature sensor is recorded, and the results collected by different coolant temperature sensors are compared to select the coolant temperature sensor that performs best for the engine under the condition that the vehicle meets characteristic quantity A. The water tank's operational status (g) is evaluated based on whether it enters the main circulation loop. This standard requires the temperature difference between the upper and lower parts of the water tank measured by a water temperature sensor. If the collected temperature difference gradually decreases and stabilizes, this stable value is influenced by characteristic quantity B, and is generally considered stable at 5-15 degrees Celsius, indicating that the water tank is operating normally. The fan's operational status (h) requires information on the type of fan directly connected to the phone, including direct-connect fans, silicone oil fans, and electrically controlled silicone oil fans. Further measurements of the fan's blade count, diameter, and length are also needed. Data recorded in a table or a photo of the fan should be provided to the user for verification. Finally, the data from e, f, g, and h are unified as a reference for characteristic quantity B. The user can then use the normality or abnormality of characteristic quantity B to determine if there is a problem in this step. If a problem exists, it can be quickly located based on e, f, g, and h.

[0091] Feature monitoring: Using the KS-test as a non-parametric test, the feature data is input into the graph of the cumulative probability density function. When the feature data has negative values, the x-axis is transformed by log to obtain the probability density function of the feature data, thus completing the construction of the training dataset. When the probability interval corresponding to the probability density function is abnormal, the training dataset is reconstructed.

[0092] like Figure 6 As shown, the specific steps for training the CNN+LSTM algorithm model based on the training dataset include:

[0093] Load the training dataset and count the number of data points in each class;

[0094] The training dataset is assigned tasks, such as 60% of the data for training, 20% for validation, and 20% for testing.

[0095] Constructing a CNN+LSTM framework;

[0096] For the input sequence, the CNN+LSTM framework requires specifying an input layer whose input size matches the input data. To ensure that the network supports the training data, the network option should be set to the length of the shortest sequence in the training data.

[0097] The CNN+LSTM framework is used to construct spatial relationships in a one-dimensional image sequence. It requires a two-dimensional CNN architecture, which includes four repeating blocks of convolution, batch normalization, ReLU and max pooling layers.

[0098] The CNN+LSTM framework allocates more filters in the second and fourth convolutional layers;

[0099] In the CNN+LSTM framework, to clearly show the long-term dependencies in a one-dimensional image sequence, the LSTM layer needs to be allocated a hidden unit. If it is necessary to map the sequence to a single value for prediction, the LSMT layer option is set to output only the last time step.

[0100] In the CNN+LSTM framework, the size of the fully connected layer is equal to the number of classes. If the output needs to be converted into a probability vector, a sky softmax layer is required in the layer.

[0101] The CNN+LSTM framework includes layer classification.

[0102] Specify training options, including solver, network size, initial learning rate, and regularization term;

[0103] In this embodiment, the Adam solver is used to train a network with a mini-batch size of 32 for three epochs. The network is trained with an initial learning rate of 0.01 and the learning rate is reduced in stages after two epochs. To avoid overfitting the training data, a regularization term of 2.0 is specified for L0005.

[0104] Build the algorithm model based on the CNN+LSTM framework and training options;

[0105] The algorithm model is tested by classifying the test data using the algorithm model. The classification accuracy of the model is tested by comparing the predictions on the retained test set with the true labels of each test observation.

[0106] like Figure 7 As shown, the specific steps for optimizing the algorithm model using the Bayesian algorithm include:

[0107] Construct an objective function; when the output of the objective function is a certain loss, the objective function is negative; when the actual parameters of the algorithm require integer input, the parameter types are specified in the objective function.

[0108] Define the parameter space using a dictionary.

[0109] Define the specific process for optimizing the objective function;

[0110] Define the verification function and perform optimization operations.

[0111] The water temperature prediction results are presented in the form of visual charts;

[0112] The decision to trigger an alarm is based on a comparison between the predicted water temperature and the standard data.

[0113] Let TP be the number of vehicles predicted to malfunction and actually malfunctioning, FN be the number of vehicles predicted to be normal but actually malfunctioning, FP be the number of vehicles predicted to malfunction but actually normal, and TN be the number of vehicles predicted to be normal and actually normal. After real-vehicle verification, the number of vehicles participating in the experiment was 195, of which 12 were TP, 2 were FN, 4 were FP, and 177 were TN. The calculated recall rate is TP / (TP+FN) = 0.75, and the precision rate is TP / (TP+FP) = 0.86, achieving a good expected result.

[0114] In addition, such as Figure 8 As shown in the figure, after verification on a real vehicle, it can be seen that when the method described in this invention is used for measurement, the vehicle failure rate tends to stabilize as the cumulative driving time increases, which proves the effectiveness and efficiency of this invention.

[0115] This invention monitors real-time engine operating parameters, uses an algorithm model to predict whether the engine will experience overheating in the future, and provides maintenance personnel with a visual chart to intuitively display the current engine status, enabling timely and effective preventive measures and predictive maintenance. It has significant research significance and practical value in terms of both engine safety and economy.

Claims

1. A water temperature prediction alarm method applied to a water temperature prediction alarm system, characterized by, include: Sensor module: Used to collect real-time data from the vehicle and obtain real-time engine parameters; The sensor module includes a pressure sensor and a water temperature sensor; The engine's real-time parameters include real-time vehicle weight data and real-time water temperature data; The pressure sensor converts the mechanical pressure input to the vehicle into an electrical output signal and sends the real-time vehicle weight data to the automatic controller. The water temperature sensor converts the water temperature of the car engine into an electrical output signal and sends the real-time water temperature data to the automatic controller. Automatic controller: used to obtain water temperature prediction results based on real-time engine parameters; The automatic controller includes a data acquisition circuit, a data processing circuit, a signal control circuit, a data transmission circuit, a motor control circuit, a reset circuit, and a power supply circuit. The data acquisition circuit sends the digital signals acquired by the sensor module to the data processing circuit; After judging the digital signal, the data processing circuit converts the digital signal into a digital signal suitable for the data transmission circuit, inputs an action command to the motor control circuit, and controls the external device. The signal control circuit feeds back the current action command to the data processing circuit, causing the data processing circuit to issue a strobe signal suitable for the data transmission circuit, and the engine data is transmitted to the visual device and peripheral devices through the data transmission circuit. The power supply circuit provides power to the automatic controller; the reset circuit provides automatic reset upon power-on and manual reset. Visual devices: used to present water temperature prediction results in the form of visual charts; Peripheral equipment: used to determine whether to trigger an alarm by comparing the predicted water temperature with standard data; Real-time data of the vehicle is collected through sensors to obtain real-time engine parameters; The water temperature prediction result is obtained based on the real-time parameters of the engine. The specific steps for obtaining the water temperature prediction result based on the engine real-time parameters are as follows: input the engine real-time parameters into the algorithm model, and then perform calculations and processing on the data in the algorithm model to obtain the water temperature prediction result; The training process of the algorithm model includes acquiring sample data using sensors, constructing a training dataset based on the sample data, training the CNN+LSTM algorithm model based on the training dataset, and optimizing the algorithm model using the Bayesian algorithm. The specific steps for constructing the training dataset based on the sample data include: Data cleaning: Define the sample data as different features, input the different features into MATLAB, use MATLAB to plot the change curves of the same type of data, observe the change curves, and if there are abnormally high or low points on the curves, filter and remove them. Feature segmentation: Divide the feature into more specific sub-indicators, verify the sub-indicators and enter them into the database, unify the data of the sub-indicators with the data of the feature, and judge whether there are any anomalies in the sub-indicators based on the feature. Feature monitoring: Using the KS-test as a non-parametric test, the feature data is input into the graph of the cumulative probability density function. When the feature data has negative values, the x-axis is transformed by log to obtain the probability density function of the feature data, thus completing the construction of the training dataset. When the probability interval corresponding to the probability density function is abnormal, the training dataset is reconstructed. The sample data consists of multiple real-time vehicle data and data required for feature segmentation indicators; The water temperature prediction results are presented in the form of visual charts; The decision to trigger an alarm is based on a comparison between the predicted water temperature and the standard data.

2. The water temperature prediction and alarm method as described in claim 1, characterized in that, The signal control circuit uses a 74LS138 chip; the data processing circuit uses an STM32 microcontroller, specifically an STM32F103C8TX chip.

3. The water temperature prediction and alarm method as described in claim 1, characterized in that, The specific steps for training the CNN+LSTM algorithm model based on the training dataset include: Load the training dataset and count the number of data points in each class; Assign tasks to the training dataset; Constructing a CNN+LSTM framework; Specify training options, including solver, network size, initial learning rate, and regularization term; Build the algorithm model based on the CNN+LSTM framework and training options; The algorithm model is tested by classifying the test data using the algorithm model. The classification accuracy of the model is tested by comparing the predictions on the retained test set with the true labels of each test observation.

4. The water temperature prediction and alarm method as described in claim 3, characterized in that, For the input sequence, the CNN+LSTM framework requires specifying an input layer whose input size matches the input data, and the network options should be set to the length of the shortest sequence in the training data. The CNN+LSTM framework uses a two-dimensional CNN architecture, which includes four repeating blocks of convolution, batch normalization, ReLU and max pooling layers. The CNN+LSTM framework allocates more filters in the second and fourth convolutional layers; In the CNN+LSTM framework, the LSTM layer is assigned a hidden unit. If it is necessary to map the sequence to a single value for prediction, the LSMT layer option is set to output only the last time step. In the CNN+LSTM framework, the size of the fully connected layer is equal to the number of classes. If the output needs to be converted into a probability vector, a sky softmax layer is required in the layer. The CNN+LSTM framework includes layer classification.

5. The water temperature prediction and alarm method as described in claim 1, characterized in that, The specific steps for optimizing the algorithm model using the Bayesian algorithm include: Construct an objective function; when the output of the objective function is a certain loss, the objective function is negative; when the actual parameters of the algorithm require integer input, the parameter types are specified in the objective function. Define the parameter space using a dictionary. Define the specific process for optimizing the objective function; Define the verification function and perform optimization operations.

Citation Information

Patent Citations

  • Automobile engine coolant temperature display, early-warning and alarming system and method

    CN104417367A

  • Thermal management system and thermal management method and device based on neural network

    CN112459890A

  • Vehicle cooling loop detection method and device, computer equipment and storage medium

    CN113465935A