Mobile energy storage system for passenger car

Abstract

The invention relates to the field of mobile energy storage, in particular to a mobile energy storage system for a passenger car, which comprises a parameter acquisition module, a fault prediction model establishment module, an intelligent diagnosis module, an environment analysis module, an early warning processing module and an adaptive control module, a fault signal is generated according to a preset threshold value; the environment analysis module adjusts the working mode of the energy storage system according to environment parameters through fuzzy rule base and neural network optimization, the early warning processing module carries out graded prompting on a passenger car display screen according to fault signals, and the self-adaptive control module controls the device to operate according to execution numbers and dynamically adjusts the parameters of the energy storage system. Through cooperative work of multiple modules, fault prediction, environment adaptation and intelligent regulation and control of the passenger car energy storage system are achieved, and the safety and reliability of the system are remarkably improved.

Description

Technical Field

[0001] The invention relates to the field of mobile energy storage, and in particular to a mobile energy storage system for a bus. Background Art

[0002] In the context of global energy transformation and increasingly stringent environmental protection requirements, electric buses are an important part of new energy transportation, and the safety, reliability and environmental adaptability of their energy storage systems have become the focus of industry attention; existing systems mostly rely on simple threshold judgments or offline analysis, and cannot capture subtle anomalies in battery operation in real time. For example, potential faults such as battery thermal runaway and capacity attenuation are difficult to warn in advance through a single parameter, resulting in frequent safety accidents;

[0003] Extreme climates and complex working conditions in different regions have significant impacts on the performance of energy storage systems. Traditional systems lack dynamic adjustment strategies, which leads to reduced efficiency and shortened life of batteries in non-ideal environments. Mobile energy storage systems are mostly passive heat dissipation, with low heat dissipation efficiency and delayed response. They lack multi-level early warning mechanisms and adaptive protection measures, and are unable to effectively respond to sudden faults such as battery short circuits and overcharging.

[0004] Therefore, the present invention proposes a mobile energy storage system for buses to solve the above technical problems. Summary of the invention

[0005] In order to solve the technical problems raised by the above background technology, the present invention provides a mobile energy storage system for a bus.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] The present invention is a mobile energy storage system for passenger cars, comprising a parameter acquisition module, a fault prediction model building module, an intelligent diagnosis module, an environmental analysis module, an early warning processing module, an adaptive control module and a database.

[0008] The parameter acquisition module collects corresponding parameter values ​​based on various types of sensors installed in each battery pack, and the sensors installed in the central area of ​​the roof collect corresponding environmental parameter values ​​in real time. The specific process is as follows:

[0009] Various types of sensors are installed in the battery pack of the bus, including temperature sensors, voltage sensors and current sensors. The acquisition frequency of the battery parameter set is set to ten times per second. The temperature sensor is used to collect the temperature parameters of different positions inside the battery in real time to obtain the real-time battery temperature value of the battery; the current sensor is used to collect the current parameters of battery charging and discharging in real time to obtain the real-time current value of the battery; the voltage sensor's differential amplifier circuit and analog-to-digital converter are used to collect the battery's voltage parameters in real time to obtain the battery's real-time voltage value;

[0010] An ambient temperature sensor, humidity sensor and altitude sensor are installed in the central area of ​​the roof, and the frequency of collecting environmental parameters is set to once per minute. The ambient temperature value, humidity value and altitude value of the corresponding bus are collected through the temperature sensor, humidity sensor and altitude sensor respectively, and the results are integrated into an environmental parameter set. The battery parameter set is sent to the fault prediction model establishment module and the intelligent diagnosis module, and the environmental parameter set is sent to the environmental adaptation module.

[0011] The fault prediction model building module builds a fault prediction model for battery energy storage based on the LSTM network and attention mechanism. The specific process is as follows:

[0012] Extract the historical battery parameter set from the database, integrate the data from different sources in the historical battery parameter set in chronological order through the timestamp alignment algorithm to obtain a battery sequence data set with a timestamp, extract the feature vectors of each historical battery temperature value, historical current value and historical voltage value in the historical battery parameters, including the gradient value, maximum temperature value and minimum temperature value of the historical battery temperature value, the current peak value, current mean and current variance value of the historical current value, and the change rate and skewness value of the historical voltage value, integrate all feature vectors into a battery feature vector set, convert the battery feature vector set and the battery sequence data set into multivariate time series data, each time step includes the variables of each historical battery temperature value, historical current value and historical voltage value, use the sliding window technology to divide the time series into samples of fixed length, and divide each sample into a training set, a validation set and a test set according to 7:2:1;

[0013] Select a deep learning framework to build an LSTM network. The LSTM network includes an input layer, multiple LSTM layers, and an output gate. An attention mechanism is added to the output gate. The cross entropy loss function is selected to measure the difference between the model prediction result and the true label. Adam is selected to update the model parameters. The training set data is input into the model, and the loss is calculated by forward propagation. The model parameters are updated by back propagation. After each training cycle, the performance of the model is evaluated using the validation set, and the hyperparameters are adjusted to prevent overfitting. This completes the construction and training of the fault prediction model.

[0014] The intelligent diagnosis module detects the battery parameter set acquired in real time according to the established fault prediction model and outputs the fault prediction value. The specific process is as follows:

[0015] Get the real-time battery parameter set of the bus, set T = 10 time steps, divide the battery parameter set according to T to get the input sequence, send the input sequence to the LSTM network, and pass through the input layer, LSTM layer and output gate in turn, process the input data of each time step, and output the hidden state sequence {h 1 ,h 2....h T}, the attention mechanism calculates the similarity score e at each time step with the learnable query vector t , the formula is e t =υtanh(W a h t +b a ), where W a is the weight matrix, b a is the bias vector, υ is the learnable vector, tanh represents the hyperbolic tangent function, h t Represents the hidden state of the model at any time step; the similarity score is normalized by the softmax function to obtain the attention weight a t , and finally multiply the attention weight and the corresponding hidden state and sum them up to get the output c of the attention mechanism;

[0016] Then the output c of the attention mechanism is calculated through the sigmoid activation function to obtain the fault prediction value;

[0017] The first fault threshold and the second fault threshold are extracted from the database, and the first fault threshold is greater than the second fault threshold. If the fault prediction value is greater than the second fault threshold but less than the first fault threshold, a low fault signal is generated and sent to the early warning module. If the fault prediction value is greater than the first fault threshold, a high fault signal is generated and sent to the early warning module.

[0018] The environmental analysis module performs fuzzy processing based on the environmental parameter set collected by the bus through the established fuzzy rule base, and outputs the corresponding parameter adjustment value according to the neural network. The specific process is as follows:

[0019] The environmental analysis module is provided with a fuzzy rule base building unit, a neural network optimization unit and an output unit;

[0020] The fuzzy rule base building unit first defines fuzzy sets for each parameter, specifically: for temperature parameters, set the temperature range below 10°C as "low temperature", the temperature range between 10°C and 30°C as "normal temperature", and the temperature above 30°C as "high temperature"; for humidity parameters, set the humidity below 30% as "dry", below 30%-70% as "moderate", and below the humidity threshold as "humid"; for altitude parameters, set below 1000m as "low altitude", 1000m-3000m as "medium altitude", and above 3000m as "high altitude";

[0021] like Figure 2As shown, fuzzy rules are set for different temperatures, altitudes and humidity, and the text is expressed in the format of "IF-THEN". Set IF the temperature is high temperature AND the altitude is high altitude AND the humidity is humid THEN reduce the charging current by 60% AND increase the cooling system power by 90% AND start additional cooling measures for the battery; IF the temperature is high temperature AND the altitude is high altitude AND the humidity is moderate THEN reduce the charging current by 44%, increase the cooling system power by 68%, and reduce the charging voltage by 6%; and so on, until setting IF the temperature is low temperature AND the altitude is low altitude AND the humidity is dry THEN start the battery preheating device, the preheating power is 2kW, and reduce the charging current by 20%, and complete the establishment of the fuzzy rule base and store it in the database; It should be noted that the rule IF is a prerequisite, and THEN is an adjustment instruction for the working mode of the energy storage system;

[0022] The neural network optimization unit sets the number of input nodes of the input layer to 9, the number of hidden layers to 2, and the number of output nodes of the output layer to 3. It extracts all the fuzzified historical environmental parameters from the database and divides them into training set, validation set and test set. It trains the neural network by stochastic gradient descent method, adds the triangular membership function of each environmental parameter in the hidden layer, and adds the fuzzy rule base in the output layer to complete the neural network optimization.

[0023] The output unit collects and sends the environmental parameters acquired in real time to the neural network input layer, obtains the membership degree corresponding to each environmental parameter through the triangular membership function of the hidden layer, and then sends the membership degree to the output layer and matches it with the number of the fuzzy rule library to obtain the corresponding matching execution number, and sends the execution number to the adaptive module;

[0024] The early warning processing module performs prompt processing through the bus display according to the different signals sent by the intelligent diagnosis module. The specific process is as follows:

[0025] If a low fault signal is received, a yellow icon will flash on the bus display screen, and a text message "The current vehicle battery is low fault, it is recommended to stop and check" will be sent to the bus display terminal for a text prompt, which will be displayed for 30 seconds and then turned into a permanent notification;

[0026] When a high fault signal is received, the bus display screen flashes red throughout the entire screen and the buzzer alarm is turned on. The driver needs to manually confirm and release it, and the text "The current car battery has a high fault, turn off the engine and stop the car immediately" is sent to the bus display terminal for a text prompt.

[0027] The adaptive control module controls the operation of the device according to the execution number and provides feedback on the effect after execution. The specific process is as follows:

[0028] After receiving the execution number, parsing the execution operation instruction corresponding to the execution number, and sending the execution operation instruction to each control device in sequence, the control device includes a charging current regulator, a charging voltage regulator, a cooling fan controller and a preheating power regulator. Each control device starts the operating state and adjusts the battery according to the execution operation instruction. The parameter acquisition module monitors the battery data after adjustment. If the adjustment effect is not achieved, the battery continues to be adjusted until the adjustment is completed and the adjustment time is recorded and sent to the database.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] LSTM-attention mechanism model: Through multivariate time series analysis of battery temperature, voltage, current and other parameters, combined with the attention mechanism to highlight key features (such as sudden voltage rise at the end of charging and abnormal temperature rise during high current discharge), potential faults such as battery thermal runaway and capacity decay are predicted in advance; dual threshold warning system: low fault signal prompts the driver to check in time, and high fault signal triggers emergency protection to avoid safety accidents caused by battery failure;

[0031] Environmental adaptive control, improving system robustness, fuzzy neural network optimization: through fuzzy processing of environmental parameters and neural network self-learning, dynamically adjust the charging current, heat dissipation power and preheating strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. The following drawings are not intentionally scaled to the actual size, and the focus is on illustrating the main purpose of the present invention.

[0033] Figure 1 It is a principle block diagram of the present invention.

[0034] Figure 2 It is the fuzzy rule base of the present invention. DETAILED DESCRIPTION

[0035] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present invention.

[0036] Please refer to Figure 1 As shown, the present invention is a mobile energy storage system for buses, including a parameter acquisition module, a fault prediction model building module, an intelligent diagnosis module, an environmental analysis module, an early warning processing module, an adaptive control module and a database.

[0037] The parameter acquisition module collects corresponding parameter values ​​based on various types of sensors installed in each battery pack, and the sensors installed in the central area of ​​the roof collect corresponding environmental parameter values ​​in real time. The specific process is as follows:

[0038] Various types of sensors are installed in the battery pack of the bus, including temperature sensors, voltage sensors and current sensors. The acquisition frequency of the battery parameter set is set to ten times per second. The temperature sensor is used to collect the temperature parameters of different positions inside the battery in real time to obtain the real-time battery temperature value of the battery; the current sensor is used to collect the current parameters of battery charging and discharging in real time to obtain the real-time current value of the battery; the voltage sensor's differential amplifier circuit and analog-to-digital converter are used to collect the battery's voltage parameters in real time to obtain the battery's real-time voltage value;

[0039] An ambient temperature sensor, humidity sensor and altitude sensor are installed in the central area of ​​the roof, and the frequency of collecting environmental parameters is set to once per minute. The ambient temperature value, humidity value and altitude value of the corresponding bus are collected through the temperature sensor, humidity sensor and altitude sensor respectively, and the results are integrated into an environmental parameter set. The battery parameter set is sent to the fault prediction model establishment module and the intelligent diagnosis module, and the environmental parameter set is sent to the environmental adaptation module.

[0040] The fault prediction model building module builds a fault prediction model for battery energy storage based on the LSTM network and attention mechanism. The specific process is as follows:

[0041] Extract the historical battery parameter set from the database, integrate the data from different sources in the historical battery parameter set in chronological order through the timestamp alignment algorithm to obtain a battery sequence data set with a timestamp, extract the feature vectors of each historical battery temperature value, historical current value and historical voltage value in the historical battery parameters, including the gradient value, maximum temperature value and minimum temperature value of the historical battery temperature value, the current peak value, current mean and current variance value of the historical current value, and the change rate and skewness value of the historical voltage value, integrate all feature vectors into a battery feature vector set, convert the battery feature vector set and the battery sequence data set into multivariate time series data, each time step includes the variables of each historical battery temperature value, historical current value and historical voltage value, use the sliding window technology to divide the time series into samples of fixed length, and divide each sample into a training set, a validation set and a test set according to 7:2:1; it should be noted that the training set is used for model training, the validation set is used to adjust the hyperparameters of the model, and the test set is used to evaluate the final performance of the model;

[0042] A deep learning framework is selected to build an LSTM network. The LSTM network includes an input layer, multiple LSTM layers and an output gate. An attention mechanism is added to the output gate. The cross entropy loss function is selected to measure the difference between the model prediction result and the true label. Adam is selected to update the model parameters. The training set data is input into the model, and the forward propagation is performed to calculate the loss. The model parameters are updated through back propagation. After each training cycle, the performance of the model is evaluated using the validation set, and the hyperparameters are adjusted to prevent overfitting, thereby completing the construction and training of the fault prediction model. It should be noted that the attention mechanism dynamically allocates attention to different parts by calculating the importance weight of each element in the input data, thereby highlighting important information and suppressing irrelevant information. For example, certain specific stages of the battery during the charging and discharging process, such as voltage changes at the end of charging and temperature rise during high current discharge, may be important signs of battery failure. The attention mechanism automatically identifies key features and gives them higher weights to improve the accuracy of prediction.

[0043] The intelligent diagnosis module detects the battery parameter set acquired in real time according to the established fault prediction model and outputs the fault prediction value. The specific process is as follows:

[0044] Get the real-time battery parameter set of the bus, set T = 10 time steps, divide the battery parameter set according to T to get the input sequence, send the input sequence to the LSTM network, and pass through the input layer, LSTM layer and output gate in turn, process the input data of each time step, and output the hidden state sequence {h 1 ,h 2 ....h T}, the attention mechanism calculates the similarity score e at each time step with the learnable query vector t , the formula is e t =υtanh(W a h t +b a ), where W a is the weight matrix, b a is the bias vector, υ is the learnable vector, tanh represents the hyperbolic tangent function, h t Represents the hidden state of the model at any time step; the similarity score is normalized by the softmax function to obtain the attention weight a t , and finally multiply the attention weight and the corresponding hidden state and sum them up to get the output c of the attention mechanism;

[0045] Then the output c of the attention mechanism is calculated through the sigmoid activation function to obtain the fault prediction value;

[0046] The first fault threshold and the second fault threshold are extracted from the database, and the first fault threshold is greater than the second fault threshold. If the fault prediction value is greater than the second fault threshold but less than the first fault threshold, a low fault signal is generated and sent to the early warning module. If the fault prediction value is greater than the first fault threshold, a high fault signal is generated and sent to the early warning module.

[0047] The environmental analysis module performs fuzzy processing based on the environmental parameter set collected by the bus through the established fuzzy rule base, and outputs the corresponding parameter adjustment value according to the neural network. The specific process is as follows:

[0048] The environmental analysis module is provided with a fuzzy rule base building unit, a neural network optimization unit and an output unit;

[0049] The fuzzy rule base building unit first defines fuzzy sets for each parameter, specifically: for temperature parameters, set the temperature range below 10°C as "low temperature", the temperature range between 10°C and 30°C as "normal temperature", and the temperature above 30°C as "high temperature"; for humidity parameters, set the humidity below 30% as "dry", below 30%-70% as "moderate", and below the humidity threshold as "humid"; for altitude parameters, set below 1000m as "low altitude", 1000m-3000m as "medium altitude", and above 3000m as "high altitude";

[0050] like Figure 2 As shown, fuzzy rules are set for different temperatures, altitudes and humidity, and the text is expressed in the format of "IF-THEN". Set IF the temperature is high temperature AND the altitude is high altitude AND the humidity is humid THEN reduce the charging current by 60% AND increase the cooling system power by 90% AND start additional cooling measures for the battery; IF the temperature is high temperature AND the altitude is high altitude AND the humidity is moderate THEN reduce the charging current by 44%, increase the cooling system power by 68%, and reduce the charging voltage by 6%; and so on, until setting IF the temperature is low temperature AND the altitude is low altitude AND the humidity is dry THEN start the battery preheating device, the preheating power is 2kW, and reduce the charging current by 20%, and complete the establishment of the fuzzy rule base and store it in the database; It should be noted that the rule IF is a prerequisite, and THEN is an adjustment instruction for the working mode of the energy storage system;

[0051] The neural network optimization unit sets the number of input nodes of the input layer to 9, the number of hidden layers to 2, and the number of output nodes of the output layer to 3. It extracts all the fuzzified historical environmental parameters from the database and divides them into training set, validation set and test set. It trains the neural network by stochastic gradient descent method, adds the triangular membership function of each environmental parameter in the hidden layer, and adds the fuzzy rule base in the output layer to complete the neural network optimization.

[0052] The output unit collects and sends the environmental parameters acquired in real time to the neural network input layer, obtains the membership degree corresponding to each environmental parameter through the triangular membership function of the hidden layer, and then sends the membership degree to the output layer and matches it with the number of the fuzzy rule library to obtain the corresponding matching execution number, and sends the execution number to the adaptive module;

[0053] It should be noted that the value of the triangle membership function is an important basis for fuzzy rule reasoning. In the fuzzy rule base, the "IF-THEN" rule determines the corresponding output according to the membership of the input parameter. By calculating the membership function, the membership degree of the input parameter in each fuzzy set is clarified, and then the corresponding fuzzy output is obtained according to the rule.

[0054] The early warning processing module performs prompt processing through the bus display according to the different signals sent by the intelligent diagnosis module. The specific process is as follows:

[0055] If a low fault signal is received, a yellow icon will flash on the bus display screen, and a text message "The current vehicle battery is low fault, it is recommended to stop and check" will be sent to the bus display terminal for a text prompt, which will be displayed for 30 seconds and then turned into a permanent notification;

[0056] When a high fault signal is received, the bus display screen flashes red throughout the entire screen and the buzzer alarm is turned on. The driver needs to manually confirm and release it, and the text "The current car battery has a high fault, turn off the engine and stop the car immediately" is sent to the bus display terminal for a text prompt.

[0057] The adaptive control module controls the operation of the device according to the execution number and provides feedback on the effect after execution. The specific process is as follows:

[0058] After receiving the execution number, parsing the execution operation instruction corresponding to the execution number, and sending the execution operation instruction to each control device in sequence, the control device includes a charging current regulator, a charging voltage regulator, a cooling fan controller and a preheating power regulator. Each control device starts the operating state and adjusts the battery according to the execution operation instruction. The parameter acquisition module monitors the battery data after adjustment. If the adjustment effect is not achieved, the battery continues to be adjusted until the adjustment is completed and the adjustment time is recorded and sent to the database.

[0059] The above is an explanation of the present invention and should not be considered as a limitation thereof. Although several exemplary embodiments of the present invention have been described, it will be readily appreciated by those skilled in the art that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present invention. Therefore, all such modifications are intended to be included within the scope of the present invention as defined in the claims. It should be understood that the above is an explanation of the present invention and should not be considered as being limited to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The present invention is defined by the claims and their equivalents.

Claims

1. A mobile energy storage system for a passenger car, comprising a parameter acquisition module, a fault prediction model building module, an intelligent diagnosis module, an environmental analysis module, an early warning processing module, an adaptive control module and a database, characterized in that: The fault prediction model building module builds a fault prediction model for battery energy storage based on the LSTM network and attention mechanism; the environmental analysis module is equipped with a fuzzy rule base building unit, a neural network optimization unit and an output unit, which fuzzifies the collected environmental parameter set through the established fuzzy rule base and adjusts the corresponding parameter values ​​according to the neural network output; The intelligent diagnosis module detects the real-time battery parameter set according to the established fault prediction model and outputs the fault prediction value. Specifically, the real-time battery parameter set of the bus is obtained, T = 10 time steps are set, the battery parameter set is divided according to T to obtain the input sequence, the input sequence is sent to the LSTM network, and the input data of each time step is processed through the input layer, LSTM layer and output gate in turn, and the hidden state sequence {h1,h2....h T }, the attention mechanism calculates the similarity score e at each time step with the learnable query vector t , the formula is e t =υtanh(W a h t +b a ), where W a is the weight matrix, b a is the bias vector, υ is the learnable vector, tanh represents the hyperbolic tangent function, h t Represents the hidden state of the model at any time step; the similarity score is normalized by the softmax function to obtain the attention weight a t , and finally multiply the attention weight and the corresponding hidden state and sum them up to get the output c of the attention mechanism; Then the output c of the attention mechanism is calculated through the sigmoid activation function to obtain the fault prediction value; The first fault threshold and the second fault threshold are extracted from the database, and the first fault threshold is greater than the second fault threshold. If the fault prediction value is greater than the second fault threshold but less than the first fault threshold, a low fault signal is generated and sent to the early warning module. If the fault prediction value is greater than the first fault threshold, a high fault signal is generated and sent to the early warning module.

2. A mobile energy storage system for a bus according to claim 1, characterized in that: The fault prediction model building module builds a fault prediction model for battery energy storage based on the LSTM network and attention mechanism. The specific process is as follows: Extract the historical battery parameter set from the database, integrate the data from different sources in the historical battery parameter set in chronological order through the timestamp alignment algorithm to obtain a battery sequence data set with a timestamp, extract the feature vectors of each historical battery temperature value, historical current value and historical voltage value in the historical battery parameters, including the gradient value, maximum temperature value and minimum temperature value of the historical battery temperature value, the current peak value, current mean and current variance value of the historical current value, and the change rate and skewness value of the historical voltage value, integrate all feature vectors into a battery feature vector set, convert the battery feature vector set and the battery sequence data set into multivariate time series data, each time step includes the variables of each historical battery temperature value, historical current value and historical voltage value, use the sliding window technology to divide the time series into samples of fixed length, and divide each sample into a training set, a validation set and a test set according to 7:2:1; Select a deep learning framework to build an LSTM network. The LSTM network includes an input layer, multiple LSTM layers, and an output gate. Add an attention mechanism in the output gate, select the cross entropy loss function to measure the difference between the model prediction result and the true label, and select Adam to update the model parameters. Input the training set data into the model, perform forward propagation to calculate the loss, and update the model parameters through back propagation. After each training cycle, use the validation set to evaluate the performance of the model and adjust the hyperparameters to prevent overfitting, thereby completing the construction and training of the fault prediction model.

3. A mobile energy storage system for a bus according to claim 1, characterized in that: The environmental analysis module performs fuzzy processing based on the environmental parameter set collected by the bus through the established fuzzy rule base. The specific process is as follows: The fuzzy rule base building unit first defines fuzzy sets for each parameter, specifically: for temperature parameters, set the temperature range below 10°C as "low temperature", the temperature range between 10°C and 30°C as "normal temperature", and the temperature above 30°C as "high temperature"; for humidity parameters, set the humidity below 30% as "dry", below 30%-70% as "moderate", and below the humidity threshold as "humid"; for altitude parameters, set below 1000m as "low altitude", 1000m-3000m as "medium altitude", and above 3000m as "high altitude"; Fuzzy rules are set for different temperatures, altitudes and humidity. The text is expressed in the "IF-THEN" format. Set IF the temperature is high temperature AND the altitude is high altitude AND the humidity is humid THEN reduce the charging current by 60% AND increase the cooling system power by 90% AND start additional cooling measures for the battery; IF the temperature is high temperature AND the altitude is high altitude AND the humidity is moderate THEN reduce the charging current by 44%, increase the cooling system power by 68%, and reduce the charging voltage by 6%; and so on, until setting IF the temperature is low temperature AND the altitude is low altitude AND the humidity is dry THEN start the battery preheating device with a preheating power of 2kW and reduce the charging current by 20%, and complete the establishment of the fuzzy rule base and store it in the database.

4. A mobile energy storage system for a bus according to claim 3, characterized in that: The neural network optimization unit trains and optimizes the neural network, specifically: The number of input nodes in the input layer is set to 9, the number of hidden layers is set to 2, and the number of output nodes in the output layer is set to 3. All the fuzzified historical environmental parameters in the database are extracted and divided into training set, verification set and test set. The neural network is trained by stochastic gradient descent method, and the triangular membership function of each environmental parameter is added to the hidden layer, and the fuzzy rule base is added to the output layer to complete the neural network optimization.

5. A mobile energy storage system for a bus according to claim 3, characterized in that: The output unit outputs the corresponding parameter adjustment value, specifically: The environmental parameters acquired in real time are collected and sent to the neural network input layer. The membership degree corresponding to each environmental parameter is obtained through the triangular membership function of the hidden layer. The membership degree is then sent to the output layer and matched with the number of the fuzzy rule library to obtain the corresponding matching execution number, which is then sent to the adaptive module.

6. A mobile energy storage system for a bus according to claim 1, characterized in that: The warning processing module performs prompt processing through the display of the bus according to the different signals sent by the intelligent diagnosis module. The specific process is as follows: If a low fault signal is received, a yellow icon will flash on the bus display screen, and a text message "The current vehicle battery is low fault, it is recommended to stop and check" will be sent to the bus display terminal for a text prompt, which will be displayed for 30 seconds and then turned into a permanent notification; When a high fault signal is received, the bus display screen flashes red and the buzzer alarm is turned on. The driver needs to manually confirm and release it, and the text "The current car battery has a high fault, shut down the engine and stop the car immediately" is sent to the bus display terminal for a text prompt.

7. A mobile energy storage system for a bus according to claim 1, characterized in that: The adaptive control module operates according to the execution number control device, and feeds back and stores the adjusted effect. The specific process is as follows: After receiving the execution number, parsing the execution operation instruction corresponding to the execution number, and sending the execution operation instruction to each control device in sequence, the control device includes a charging current regulator, a charging voltage regulator, a cooling fan controller and a preheating power regulator. Each control device starts the operating state and adjusts the battery according to the execution operation instruction. The parameter acquisition module monitors the battery data after adjustment. If the adjustment effect is not achieved, the battery continues to be adjusted until the adjustment is completed and the adjustment time is recorded and sent to the database.

8. A mobile energy storage system for a bus according to claim 1, characterized in that: The parameter acquisition module collects corresponding parameter values ​​according to various types of sensors installed in each battery pack, and the sensors installed in the central area of ​​the roof collect corresponding environmental parameter values ​​in real time. The specific process is as follows: Various types of sensors are installed in the battery pack of the bus, including temperature sensors, voltage sensors and current sensors. The acquisition frequency of the battery parameter set is set to ten times per second. The temperature parameters of different positions inside the battery are collected in real time through the temperature sensor to obtain the real-time battery temperature value of the battery. The current sensor collects the current parameters of battery charging and discharging in real time to obtain the real-time current value of the battery; the voltage sensor collects the voltage parameters of the battery in real time through the differential amplifier circuit and analog-to-digital converter to obtain the real-time voltage value of the battery; An ambient temperature sensor, humidity sensor and altitude sensor are installed in the central area of ​​the roof, and the frequency of collecting environmental parameters is set to once per minute. The ambient temperature value, humidity value and altitude value of the corresponding bus are collected through the temperature sensor, humidity sensor and altitude sensor respectively, and the results are integrated into an environmental parameter set. The battery parameter set is sent to the fault prediction model establishment module and the intelligent diagnosis module, and the environmental parameter set is sent to the environmental adaptation module.

Images