Automobile emergency starting power supply control system and its method

Abstract

The present application provides an automobile emergency starting power supply control system and its method, which relates to the field of intelligent management of automobile power supplies. First, it collects the voltage time series data of the automobile battery and the internal battery of the automobile emergency power supply, and then transmits this voltage time series data to the MCU control module. The MCU control module deeply analyzes this voltage time series data to obtain corresponding control results, and based on this control result, performs a charging operation on the internal battery of the automobile emergency power supply. In this way, real-time monitoring and intelligent management of the state of the internal battery of the automobile emergency power supply can be achieved, which is beneficial to improving the reliability and practicability of the automobile emergency power supply at critical moments, and further can provide strong guarantee for the normal operation of the automobile.

Description

Technical Field

[0001] This application relates to the field of intelligent management of automotive power supplies, and more specifically, to an automotive emergency starting power supply control system and method thereof. Background Art

[0002] With the continuous improvement of the electrification and intelligence levels of modern automobiles, ensuring the stability and reliability of the vehicle battery system has become particularly important. Especially in cold weather or when parked for a long time, the automotive battery may not be able to start the engine due to insufficient power. Therefore, as a backup power supply solution, the automotive emergency starting power supply can provide necessary power support at critical moments to ensure the normal operation of the vehicle.

[0003] Patent CN108288869B proposed an online emergency power supply control circuit and an online emergency power supply. Specifically, the MCU control module receives the start signal and controls charging and discharging based on the data of the voltage detection module; the start module includes external device signals, remote control, and voltage change-triggered wake-up; the voltage detection module monitors the internal and external battery voltages to ensure automatic charging or auxiliary start when the voltage is below the threshold; the power supply module ensures stable power for the MCU and other components; the charge and discharge control module manages the energy flow between the internal and external batteries to achieve charge and discharge management.

[0004] The judgment of automatically charging the internal battery or providing auxiliary start for the automotive battery in this patent mainly relies on the real-time voltage information provided by the voltage detection module and the preset threshold conditions. However, the fixed voltage thresholds (such as 12.2 volts for the automotive battery and 10 volts for the internal battery) lack flexibility and cannot adapt to different usage scenarios or environmental changes. For example, at extreme temperatures, the optimal operating voltage of the battery may be significantly different from that under normal temperature conditions, but the fixed thresholds cannot be automatically adjusted to cope with these changes, which may lead to improper charging and discharging timing, affecting performance and lifespan. In addition, the system based on fixed thresholds can only react after problems occur and cannot provide early warnings of potential failures, limiting the ability of preventive maintenance and increasing the risk of sudden failures.

[0005] Therefore, an optimized automotive emergency starting power supply control scheme is desired, which can improve the overall reliability of the system by using real-time data for dynamic optimization. Summary of the Invention

[0006] To solve the above technical problems, this application is proposed. Embodiments of this application provide an automotive emergency starting power supply control system and method thereof.

[0007] According to one aspect of the present application, a control system for an automotive emergency starting power supply is provided, which includes: a voltage detection module for collecting the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply; a data transmission module for transmitting the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply to the MCU control module; the MCU control module for analyzing the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply to obtain a control result; a charge and discharge control module for determining whether to charge the internal battery of the automotive emergency power supply based on the control result; wherein, the MCU control module includes:

[0008] a battery voltage time-series analysis unit for respectively performing recursive processing and spatial domain feature extraction on the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply to obtain the spatial domain features of the automotive battery voltage time series and the spatial domain features of the internal battery voltage time series;

[0009] an automotive battery-internal voltage interaction unit for performing fine-grained spatial domain interaction coding on the spatial domain features of the automotive battery voltage time series and the spatial domain features of the internal battery voltage time series to obtain the spatial domain semantic interaction coding features of the automotive battery-internal battery voltage time series;

[0010] a control result generation unit for obtaining a control result based on the spatial domain semantic interaction coding features of the automotive battery-internal battery voltage time series, and controlling the charge and discharge control module based on the control result.

[0011] According to another aspect of the present application, a method for controlling an automotive emergency starting power supply is provided, which is used to execute the control system for an automotive emergency starting power supply as described above, and includes:

[0012] Respectively performing recursive processing and spatial domain feature extraction on the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply to obtain the spatial domain features of the automotive battery voltage time series and the spatial domain features of the internal battery voltage time series;

[0013] Performing fine-grained spatial domain interaction coding on the spatial domain features of the automotive battery voltage time series and the spatial domain features of the internal battery voltage time series to obtain the spatial domain semantic interaction coding features of the automotive battery-internal battery voltage time series;

[0014] Based on the spatial domain semantic interaction coding features of the automotive battery-internal battery voltage time series, obtaining a control result, and controlling the charge and discharge control module based on the control result.

[0015] Compared with the prior art, the automotive emergency starting power supply control system and method provided by the present application first collect the voltage sequence data of the automotive battery and the internal battery of the automotive emergency power supply, and then transmit these voltage sequence data to the MCU control module. The MCU control module deeply analyzes these voltage sequence data to obtain corresponding control results, and based on this control result, performs charging or discharging operations on the automotive battery or the internal battery of the automotive emergency power supply. In this way, real-time monitoring and intelligent management of the states of the automotive battery and the internal battery of the automotive emergency power supply can be achieved, which is beneficial to improving the reliability and practicality of the automotive emergency power supply at critical moments, and further can provide strong guarantee for the normal operation of the vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for description in the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. In the drawings:

[0017] Figure 1 FIG. is a system block diagram of an automotive emergency starting power supply control system according to an embodiment of the present application.

[0018] Figure 2 FIG. is a block diagram of the MCU control module in the automotive emergency starting power supply control system according to an embodiment of the present application.

[0019] Figure 3 FIG. is a block diagram of the battery voltage sequence analysis unit in the automotive emergency starting power supply control system according to an embodiment of the present application.

[0020] Figure 4 FIG. is a block diagram of the automotive battery - internal voltage interaction unit in the automotive emergency starting power supply control system according to an embodiment of the present application.

[0021] Figure 5 FIG. is a block diagram of the control result generation unit in the automotive emergency starting power supply control system according to an embodiment of the present application.

[0022] Figure 6 FIG. is a flowchart of the automotive emergency starting power supply control method according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] Next, exemplary embodiments according to the present application will be described in detail with reference to the drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0024] With the continuous improvement of the electrification and intelligence levels of modern vehicles, the stability and reliability of vehicle battery systems have become increasingly crucial. Especially in cold climates or long-term parking situations, the problem of engine failure due to depleted battery power is particularly prominent. Therefore, as an effective backup power supply solution, an automotive emergency starting power supply can provide necessary power support for vehicles in emergencies, ensuring smooth vehicle operation.

[0025] Patent CN108288869B proposes an online emergency power supply control circuit and an online emergency power supply. Specifically, the MCU control module receives a start signal and precisely controls charging and discharging based on the data of the voltage detection module. The start module integrates external device signals, remote control, and voltage change-triggered wake-up functions; the voltage detection module is responsible for monitoring the internal and external battery voltages to automatically charge or assist in starting when the voltage is below a preset threshold; the power supply module provides stable power for the MCU and other components; the charge and discharge control module is responsible for managing the energy flow between the internal and external batteries.

[0026] However, in this patent, a fixed voltage threshold (such as 12.2 volts for the automotive battery and 10 volts for the internal battery) is used to determine whether to automatically charge the internal battery or assist in starting the automotive battery. This method lacks flexibility and is difficult to adapt to different usage scenarios or environmental changes. Under extreme temperature conditions, the optimal operating voltage of the battery may differ significantly from that at normal temperature, but the fixed threshold cannot be automatically adjusted, which may lead to improper charging and discharging timing, having an adverse impact on the performance and lifespan of the battery. In addition, the system based on a fixed threshold can only respond afterwards and cannot provide early warnings of potential failures, thus limiting the effectiveness of preventive maintenance and increasing the risk of sudden failures.

[0027] Based on this, this application proposes an automotive emergency starting power supply control system. Figure 1 The system block diagram of the automotive emergency starting power supply control system according to an embodiment of this application is as follows. As Figure 1 shown, in the automotive emergency starting power supply control system 100, it includes: a voltage detection module 110 for collecting the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply; a data transmission module 120 for transmitting the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply to the MCU control module; the MCU control module 130 for analyzing the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply to obtain a control result; a charge and discharge control module 140 for determining whether to charge the internal battery of the automotive emergency power supply based on the control result.

[0028] In the embodiment of the present application, the voltage detection module 110 is used to collect the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply. It should be understood that the time-series data of the automotive battery voltage is a data sequence reflecting the voltage change of the automotive battery over a period of time. It can reflect the voltage fluctuations of the automotive battery under different working conditions (such as starting, driving, parking, etc.), and can provide basic information for subsequent analysis of the performance state of the automotive battery. The time-series data of the internal battery voltage of the automotive emergency power supply is similar to the time-series data of the automotive battery voltage, which is the voltage change sequence of the internal battery within the corresponding time period. By collecting these data, the charging history of the internal battery, the change trend of the remaining power, and whether there are potential performance problems can be understood, which is an indispensable part of comprehensively evaluating the state of the automotive emergency power supply system. Generally speaking, the automotive battery is the main power source for the normal operation of the vehicle, while the internal battery of the emergency power supply is the backup guarantee when the automotive battery has problems. Obtaining the voltage time-series data of both can monitor the operation status of the entire battery system in real time, ensuring a clear understanding of the vehicle's power supply at any time. For example, during a long-distance vehicle journey, by continuously monitoring the time-series data of the automotive battery voltage, it can be timely detected whether the battery power is sufficient to support the remaining journey. If the power is insufficient, charging can be planned in advance or the emergency power supply can be used to supplement the power; at the same time, the voltage time-series data of the internal battery of the emergency power supply can also reflect its own preparation status, ensuring reliable operation when needed.

[0029] Collecting the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply involves elaborate design and operation in many aspects to ensure the accuracy, integrity, and effectiveness of the data, thereby providing a solid foundation for the stable operation of the entire system. The following is a detailed description of its implementation process:

[0030] First of all, the selection and connection of hardware devices are crucial. For the automotive battery and the internal battery of the emergency power supply, appropriate voltage sensors need to be carefully selected. These sensors not only need to have high precision, high sensitivity, and excellent stability, but also need to be perfectly compatible with the battery characteristics and the automotive electrical system. For example, for the automotive battery, a sensor with a measurement range of 0-16V and an accuracy of ±0.05V may be selected. Its high input impedance can avoid excessive consumption of battery power, ensuring the accuracy of measurement and the battery performance is not affected. In the connection process, the sensors must be firmly connected to the positive and negative electrodes of the battery according to the correct polarity, ensuring a firm connection and an extremely small contact resistance. At the same time, protection measures such as series fuses are taken to prevent short-circuit abnormalities. For the internal battery of the emergency power supply, it is also necessary to ensure that the sensor is correctly connected and the wiring is reasonable to avoid electromagnetic interference. The collected voltage signal is transmitted to the subsequent data acquisition circuit or device through high-quality wires.

[0031] At the same time, the reasonable configuration of data acquisition equipment is indispensable. A microcontroller or dedicated data acquisition card with sufficient sampling accuracy (at least 12 bits) and speed (more than 1000 times per second) must be selected to receive and process the analog signal from the voltage sensor. During the initial configuration process, the sampling parameters must be accurately set, such as setting the sampling frequency to 500 times per second initially, and flexibly adjusting it according to actual needs. At the same time, a high-speed serial communication interface (such as SPI or I2C interface) must be configured to ensure that data can be quickly transmitted to the MCU control module.

[0032] In the process of data collection, strict data verification and preprocessing algorithms must also be incorporated. By setting a reasonable voltage range threshold, the validity of the collected data can be checked and abnormal data (such as data that exceeds the normal working voltage range) can be identified in a timely manner. These abnormalities may be caused by factors such as sensor failure or interference. For abnormal data, interpolation correction or direct abandonment can be performed based on the trend of the previous and subsequent data, and the abnormal situation can be recorded in detail to provide a basis for subsequent in-depth analysis. At the same time, the collected data is preprocessed by filtering and other means to effectively remove noise interference and significantly improve the accuracy and smoothness of the data.

[0033] Finally, it is crucial to reasonably determine the time and frequency of data collection to fully and accurately grasp the battery status. At each stage of the vehicle's operation, carefully plan the data collection time according to different working conditions. At the moment of vehicle startup, the battery load suddenly changes, causing the voltage to change sharply. At this time, focus on collecting the data before and after, which will help to deeply analyze the battery's startup performance and health status; during driving, regularly collect data (such as every 1 minute) to monitor the discharge characteristics and remaining power change trends during normal operation of the battery; continue to collect data for a period of time after parking to observe the battery's self-discharge. In addition, dynamically adjust the acquisition frequency according to the battery's working status and system requirements. In the stage of drastic voltage changes (such as startup or fast charging), increase the acquisition frequency to 1000 times per second to accurately capture the details of dynamic changes; in the relatively stable voltage state (such as long-term parking or stable driving), appropriately reduce it to 100 times per second, while ensuring monitoring needs while optimizing system resource utilization and data storage capacity.

[0034] Through the above comprehensive and detailed implementation process, reliable voltage timing data of the car battery and the internal battery of the emergency power supply can be collected.

[0035] In the embodiment of the present application, the data transmission module 120 is configured to transmit the automotive battery voltage time-series data and the internal battery voltage time-series data of the automotive emergency power supply to the MCU control module. It should be understood that the states of these two batteries, namely the automotive battery and the internal battery of the emergency power supply, are interrelated and jointly affect the power supply of the vehicle. The automotive emergency starting power supply control system needs to precisely manage the states of these two batteries to ensure reliable power support for the vehicle in various situations. Therefore, in the present application, it is necessary to transmit the automotive battery voltage time-series data and the internal battery voltage time-series data to the MCU control module so that all key data can be processed in a core control module, thereby enabling a more accurate grasp of the states of the automotive battery and the internal battery of the emergency power supply, which is conducive to realizing the intelligent management of different batteries to better serve the actual power consumption needs of the vehicle.

[0036] In the embodiment of the present application, the MCU control module 130 is configured to analyze the automotive battery voltage time-series data and the internal battery voltage time-series data of the automotive emergency power supply to obtain a control result.

[0037] Correspondingly, in the MCU control module, the technical concept of the present application is to use data analysis and processing algorithms based on deep learning to recursively process and extract spatial domain features from the automotive battery voltage time-series data and the internal battery voltage time-series data of the automotive emergency power supply, and based on the fine-grained spatial semantic interaction representation between the spatial domain features of the automotive battery voltage time-series and the spatial domain features of the internal battery voltage time-series, intelligently determine whether to charge the internal battery of the automotive emergency power supply and control the charge and discharge control module. In this way, the subtle patterns and trends of voltage changes can be captured, and the charging strategy can be dynamically adjusted to ensure optimal performance in various usage scenarios. This not only enhances the flexibility of the system but also solves the problem that the fixed threshold method cannot cope with environmental changes, thereby effectively protecting the battery performance and extending its service life.

[0038] Specifically, Figure 2 is a block diagram of the MCU control module in the automotive emergency starting power supply control system according to the embodiment of the present application. As Figure 2As shown, the MCU control module 130 includes: a battery voltage timing analysis unit 131, configured to perform recursive processing and spatial domain feature extraction on the automotive battery voltage timing data and the internal battery voltage timing data of the automotive emergency power supply respectively to obtain the spatial domain features of the automotive battery voltage timing and the spatial domain features of the internal battery voltage timing; an automotive battery-internal voltage interaction unit 132, configured to perform fine-grained spatial domain interaction encoding on the spatial domain features of the automotive battery voltage timing and the spatial domain features of the internal battery voltage timing to obtain the spatial domain semantic interaction encoding features of the automotive battery-internal battery voltage timing; a control result generation unit 133, configured to obtain a control result based on the spatial domain semantic interaction encoding features of the automotive battery-internal battery voltage timing, and control the charge and discharge control module based on the control result.

[0039] In the embodiment of the present application, the battery voltage timing analysis unit 131 is configured to perform recursive processing and spatial domain feature extraction on the automotive battery voltage timing data and the internal battery voltage timing data of the automotive emergency power supply respectively to obtain the spatial domain features of the automotive battery voltage timing and the spatial domain features of the internal battery voltage timing. Specifically, Figure 3 is a block diagram of the battery voltage timing analysis unit in the automotive emergency starting power supply control system according to the embodiment of the present application. As Figure 3 shown, the battery voltage timing analysis unit 131 includes: a battery voltage timing data recurrence graph calculation sub-unit 1311, configured to calculate the recurrence graphs of the automotive battery voltage timing data and the internal battery voltage timing data of the automotive emergency power supply respectively to obtain the automotive battery voltage timing recurrence graph and the internal battery voltage timing recurrence graph; a battery voltage timing spatial domain feature extraction sub-unit 1312, configured to pass the automotive battery voltage timing recurrence graph and the internal battery voltage timing recurrence graph through a voltage timing spatial domain feature extractor based on MCNN to obtain the automotive battery voltage timing spatial domain feature map as the spatial domain features of the automotive battery voltage timing and the internal battery voltage timing spatial domain feature map as the spatial domain features of the internal battery voltage timing.

[0040] In the embodiment of the present application, the recursive graph calculation subunit 1311 of the battery voltage time series data is configured to calculate the recursive graphs of the automotive battery voltage time series data and the internal battery voltage time series data of the automotive emergency power supply respectively to obtain the automotive battery voltage time series recursive graph and the internal battery voltage time series recursive graph. It should be understood that the internal batteries of the automotive battery and the emergency power supply may exhibit complex non-linear dynamic characteristics, such as voltage fluctuations or abnormal patterns. Therefore, in order to gain a deeper understanding of the time series coding trends of the battery over time, to reveal the subtle changes and dynamic characteristics of the battery behavior, and thus better understand the time series dynamic behavior of the battery, in the technical solution of the present application, the recursive graphs of the automotive battery voltage time series data and the internal battery voltage time series data of the automotive emergency power supply are calculated respectively to obtain the automotive battery voltage time series recursive graph and the internal battery voltage time series recursive graph. It can be understood that the recursive graph is a technique for analyzing time series data. It visualizes the repeatability of the system state by comparing the points in the time series with their own states at different time points. Specifically, the recursive graph shows the similarity between the state at a certain time point t and the state at another time point s. If the two states are very close, a point is marked at the corresponding position in the graph; otherwise, no mark is made. This graphical representation method can reveal the hidden patterns and structures inside the time series data, including periodicity and chaotic behavior. In this way, the dynamic characteristics and internal complex pattern changes of the battery can be more accurately identified.

[0041] In the automotive emergency starting power supply control system, calculating the recursive graphs of the automotive battery and the internal battery voltage time series data of the emergency power supply is a complex and crucial task. Its specific implementation process involves various theoretical knowledge and technical operations, aiming to obtain the recursive graph information that can accurately reflect the battery state by deeply analyzing the dynamic characteristics of the voltage time series data, so as to provide strong support for subsequent system control and decision-making. The following is a detailed elaboration of its implementation process:

[0042] The core of recursive graph calculation lies in state space reconstruction and similarity judgment. First, for the voltage time series data of the automotive battery and the internal battery of the emergency power supply, accurately selecting appropriate time delay parameter p and embedding dimension m is a crucial starting step. There are various methods for determining the time delay p, and the autocorrelation function method is one of the commonly used means. Taking the automotive battery voltage time series data as an example, by calculating the autocorrelation function and observing the change of the function value with the time delay. When the autocorrelation function value first drops to close to 0, it indicates that the correlation between data points is relatively low at this time, and the corresponding time delay is the more appropriate p value. In actual analysis, it may be found that when p = 5 (unit: sampling interval), the autocorrelation function of the automotive battery voltage time series data presents this characteristic, thus determining the value of p.

[0043] The selection of the embedding dimension m usually relies on technical means such as the false nearest neighbor method. For the time series data of the battery voltage inside the emergency power supply, the false nearest neighbor method is used for calculation. At different values of the embedding dimension, observe the change in the proportion of false nearest neighbors. When the proportion of false nearest neighbors decreases significantly, it indicates that the selected embedding dimension at this time can better capture the internal structure of the data. Through careful calculation and analysis, it may be determined that when m = 7, the internal structure of the time series data of the battery voltage inside the emergency power supply can be presented more accurately.

[0044] After determining p and m, perform state space reconstruction on the voltage time series data. Taking the time series data of the automotive battery voltage as V(t) as an example, the reconstructed vector sequence can be expressed as This process transforms the one-dimensional voltage time series data into a multi-dimensional vector sequence, thus laying a foundation for subsequent recurrence plot calculation.

[0045] When constructing the recurrence plot, set a suitable threshold ε to judge the similarity between two state vectors. The selection basis of the threshold is usually related to the statistical characteristics of the data. For example, the threshold ε = 0.5×standard deviation (time series data of automotive battery voltage) can be taken. For the vector sequence reconstructed from the time series data of the automotive battery voltage Calculate any two vectors and The distance between them is generally calculated using the Euclidean distance formula. By comparing the distance d ij with the threshold ε, construct the recurrence plot of the automotive battery voltage time series. If d ij < ε, mark it as 1 at the corresponding position in the recurrence plot, indicating that the two states are similar; otherwise, mark it as 0, indicating dissimilarity. The same calculation process is applied to the time series data of the battery voltage inside the emergency power supply, thus obtaining the recurrence plot of the internal battery voltage time series.

[0046] Implementing the above calculation process requires the aid of efficient algorithms and tools. In terms of algorithm selection, fast nearest neighbor search algorithms (such as the k-d tree algorithm or the ball tree algorithm) can significantly improve the calculation efficiency, especially suitable for processing large-scale voltage time series data. Taking the k-d tree algorithm as an example, its core principle is to reasonably divide the data space and organize the data into a tree structure. During the calculation process, when searching for the nearest neighbor point, the target point area can be quickly located according to the tree structure, thus greatly reducing the calculation complexity. When applying the k-d tree algorithm, according to the characteristics of the time series data of the automotive battery and the battery voltage inside the emergency power supply, further optimize the algorithm parameters. For example, if the time series data of the automotive battery voltage fluctuates greatly, when selecting the splitting dimension in the k-d tree algorithm, give priority to the dimension with larger data changes for splitting, which can more effectively narrow the search range and improve the search efficiency.

[0047] In terms of programming implementation, programming languages such as Python and C++ can be selected to implement the recursive graph calculation algorithm. Taking Python as an example, its rich scientific computing libraries such as NumPy and SciPy provide strong support for the calculation process. The array operation function of NumPy can conveniently process voltage time series data and implement operations such as vector calculation; the spatial search module of SciPy can be used to implement the fast nearest neighbor search algorithm, simplifying the complexity of programming. At the same time, with the help of professional data processing and analysis tools (MATLAB), the calculation results can be visualized and verified. MATLAB provides rich plotting functions that can intuitively plot the recursive graphs of the automotive battery and the internal battery voltage time series. By observing the structure and characteristics of the recursive graphs, the rationality of the calculation results can be initially judged. In addition, the built-in statistical analysis tools in MATLAB can also be used to deeply verify and analyze the calculation results to ensure the accuracy of the calculation.

[0048] In the actual calculation process, parameter adjustment and optimization are the key links to improve the quality of recursive graph calculation. Sensitivity analysis of parameters such as time delay, embedding dimension, and threshold is an essential step. By changing the parameter values and carefully observing the structural changes of the recursive graphs, the optimal value range of the parameters can be determined. Taking the voltage time series data of automotive batteries as an example, when p changes between 3 and 7, the periodic structure of the recursive graph will change significantly. Through in-depth analysis, it is found that when p = 5, the main periodic characteristics of the battery voltage can be clearly shown. For the embedding dimension m, when m changes between 5 and 9, the detailed degree of the description of the system dynamic characteristics by the recursive graph is different. After comparative research, when m = 7 is determined, the internal structure of the data can be better captured without excessive increase in the calculation complexity.

[0049] In addition to sensitivity analysis, an adaptive parameter adjustment strategy also needs to be formulated according to the working state and environmental changes of the battery. The electrochemical characteristics of automotive batteries will change at different working temperatures, which will in turn affect the voltage change law. When the temperature rises, the voltage fluctuation of the battery may intensify. At this time, the threshold can be appropriately reduced to more sensitively capture the similarity structure in the voltage change. At the same time, the time delay is dynamically adjusted according to the frequency of voltage change to ensure that the reconstructed state space can accurately reflect the dynamic characteristics of the battery. To achieve more accurate parameter optimization, a parameter optimization model can also be established by combining historical data and real-time monitoring data. The parameters are automatically optimized through machine learning algorithms such as neural networks and support vector machines. The model is trained with historical data to enable the model to learn the optimal value law of parameters under different working conditions, and then the real-time monitoring data is input into the model to obtain the optimal parameter values in the current situation, so as to achieve the adaptive adjustment of parameters, further improve the accuracy and effectiveness of recursive graph calculation, and provide more reliable guarantee for the stable operation of the automotive emergency starting power control system.

[0050] In the embodiment of the present application, the battery voltage time - series spatial - domain feature extraction sub - unit 1312 is configured to obtain a vehicle battery voltage time - series spatial - domain feature map as the vehicle battery voltage time - series spatial - domain feature and an internal battery voltage time - series spatial - domain feature map as the internal battery voltage time - series spatial - domain feature by passing the vehicle battery voltage time - series recurrence graph and the internal battery voltage time - series recurrence graph through a voltage time - series spatial - domain feature extractor based on MCNN. Correspondingly, considering that the vehicle battery voltage time - series recurrence graph and the internal battery voltage time - series recurrence graph contain spatial feature information. For example, the diagonal structure in the recurrence graph usually represents the periodic behavior of the system, which may correspond to the charging cycle or other regularly occurring events; the disordered or randomly distributed points may indicate that the system is in a chaotic state or experiencing abnormal conditions, and this non - periodicity may be caused by voltage changes due to factors such as temperature fluctuations and load changes. Based on this, in the technical solution of the present application, the vehicle battery voltage time - series recurrence graph and the internal battery voltage time - series recurrence graph are passed through a voltage time - series spatial - domain feature extractor based on MCNN to utilize the powerful pattern recognition ability of MCNN to more precisely capture subtle changes and long - term spatial - time dependencies from complex voltage time - series data, and obtain the vehicle battery voltage time - series spatial - domain feature map and the internal battery voltage time - series spatial - domain feature map. In particular, MCNN (Multi - Channel Convolutional Neural Network) is a deep - learning model specifically designed to process multi - dimensional data. It extends the traditional Convolutional Neural Network (CNN) by introducing multiple input channels to simultaneously process different types or sources of data. Each channel can be regarded as an independent feature map, allowing the network to capture information in the input data from different perspectives. In this way, different channels can be used to optimize for specific types of features. For example, one channel focuses on capturing and mining the periodic time - series patterns of the battery, and another channel focuses on the non - periodic changes in the battery state, such as short - term continuous fluctuations, etc., thereby improving the understanding and analysis of the local spatial structure and temporal correlation in the recurrence graph.

[0051] In the embodiment of the present application, the vehicle battery - internal voltage interaction unit 132 is configured to perform fine - grained spatial - domain interaction coding on the vehicle battery voltage time - series spatial - domain feature and the internal battery voltage time - series spatial - domain feature to obtain a vehicle battery - internal battery voltage time - series spatial - domain semantic interaction coding feature. Specifically, Figure 4 is a block diagram of the vehicle battery - internal voltage interaction unit in the vehicle emergency starting power supply control system according to the embodiment of the present application. As Figure 4As shown, the automotive battery - internal voltage interaction unit 132 includes: a battery voltage sharing feature interaction subunit 1321, configured to perform feature decomposition on the automotive battery voltage temporal - spatial domain feature map and the internal battery voltage temporal - spatial domain feature map, and calculate the shared semantic features between the set of automotive battery voltage temporal local - spatial domain semantic feature matrices and the set of internal battery voltage temporal local - spatial domain semantic feature matrices after decomposition to obtain a set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices; a battery voltage semantic compensation text subunit 1322, configured to calculate the semantic compensation text features of each automotive battery - internal battery voltage fine - grained local shared semantic feature matrix in the set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices to obtain a set of automotive battery - internal battery voltage semantic compensation text feature matrices; a battery voltage feature aggregation subunit 1323, configured to perform weighted aggregation on the set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices based on the set of automotive battery - internal battery voltage semantic compensation text feature matrices to obtain an automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding feature map as the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding feature.

[0052] It should be understood that the automotive battery voltage temporal - spatial domain features and the internal battery voltage temporal - spatial domain features exhibit complex non - linear dynamic characteristics in different time periods, and there is also shared semantic information between them. Therefore, in order to be able to dig deeper into the interaction between the automotive battery and the internal battery of the emergency power supply and capture the complex dynamic relationship between them, in the technical solution of this application, fine - grained spatial domain interaction coding is performed on the automotive battery voltage temporal - spatial domain features and the internal battery voltage temporal - spatial domain features to obtain an automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding feature. In particular, the purpose of the fine - grained spatial domain interaction coding is to finely adjust the feature interaction process through prompt learning, thereby using the prior knowledge of large - scale models to perform detailed semantic enhancement on the feature interaction, so as to generate richer and more accurate feature expressions.

[0053] Specifically, in the embodiment of this application, the battery voltage sharing feature interaction subunit is configured to: perform feature decomposition on the automotive battery voltage temporal - spatial domain feature map and the internal battery voltage temporal - spatial domain feature map along the channel dimension of the automotive battery voltage temporal - spatial domain feature map and the internal battery voltage temporal - spatial domain feature map to obtain the set of automotive battery voltage temporal local - spatial domain semantic feature matrices and the set of internal battery voltage temporal local - spatial domain semantic feature matrices, and this process can be expressed as:

[0054] Decouple(F1)={F11 , F 12 ,..., F 1i ,..., F 1n}

[0055] Decouple(F2) = {F 21 , F 22 ,..., F 2i ,..., F 2n}

[0056] Wherein, F1 and F2 are the automotive battery voltage time - series spatial domain feature map and the internal battery voltage time - series spatial domain feature map respectively, Decouple(·) is an operation for feature decomposition of the feature map, F 11 , F 12 , F 1i and F 1n are respectively the 1st, 2nd, ith and nth automotive battery voltage time - series local spatial domain semantic feature matrices in the set of automotive battery voltage time - series local spatial domain semantic feature matrices, F 21 , F 22 , F 2i and F 2n are respectively the 1st, 2nd, ith and nth internal battery voltage time - series local spatial domain semantic feature matrices in the set of internal battery voltage time - series local spatial domain semantic feature matrices;

[0057] Performing shared semantic feature extraction with a siamese structure on each pair of corresponding channel - dimension automotive battery voltage time - series local spatial domain semantic feature matrices and internal battery voltage time - series local spatial domain semantic feature matrices in the set of automotive battery voltage time - series local spatial domain semantic feature matrices and the set of internal battery voltage time - series local spatial domain semantic feature matrices to obtain the set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices, this process can be expressed as:

[0058]

[0059] Wherein, F 1i is the ith automotive battery voltage time - series local spatial domain semantic feature matrix in the set of automotive battery voltage time - series local spatial domain semantic feature matrices, F 2i is the ith internal battery voltage time - series local spatial domain semantic feature matrix in the set of internal battery voltage time - series local spatial domain semantic feature matrices, ⊙ and are element - wise addition, element - wise multiplication and element - wise subtraction respectively according to positions, Concat(·;·;·) is concatenation processing, Cov 1×1 (·) is point - convolutional encoding, sigmoid(·) is the sigmoid function, Si Yes F 1i and F 2i The car battery-internal battery voltage fine-grained local shared semantic feature matrix.

[0060] It should be understood that the automotive battery voltage time series spatial domain feature map and the internal battery voltage time series spatial domain feature map contain rich and complex information, which reflects the status of the battery voltage from different angles. By performing feature decomposition on these two feature maps along the channel dimension, the various types of information originally integrated in the feature map can be separated according to the different abstract levels corresponding to the channels, thereby mining out the more detailed and targeted feature information about the battery voltage hidden therein, so as to understand the battery voltage characteristics more comprehensively and deeply, and this feature decomposition method can also provide a basis for subsequent fine-grained analysis. Next, the shared semantic features of the automotive battery voltage time series local spatial domain semantic feature matrix and the internal battery voltage time series local spatial domain semantic feature matrix corresponding to the channel dimension in each set of the automotive battery voltage time series local spatial domain semantic feature matrix and the internal battery voltage time series local spatial domain semantic feature matrix are extracted with twin structures to find out the shared semantic information between the automotive battery voltage and the internal battery voltage local feature matrix under each set of the corresponding channel dimension in the two sets, that is, to mine the common features between them. By clarifying these common characteristics, we can have a deeper understanding of the common characteristics of the vehicle battery and the internal battery in the same dimension in terms of voltage performance, which can provide key basis for subsequent understanding of the internal battery status and determining whether the internal battery needs to be charged.

[0061] Specifically, in the embodiment of the present application, the battery voltage semantic compensation text subunit is used to: perform semantic compensation decoding based on a large language on each of the automobile battery-internal battery voltage fine-grained local shared semantic feature matrices in the set of the automobile battery-internal battery voltage fine-grained local shared semantic feature matrices to obtain a set of automobile battery-internal battery voltage semantic compensation text descriptions. This process can be expressed as:

[0062] T i =LLM{S i}

[0063] Among them, S i Yes F 1i and F 2i The car battery-internal battery voltage fine-grained local shared semantic feature matrix between them, LLM is the semantic compensation decoding operation, T i For S i Corresponding car battery-internal battery voltage semantic compensation text description;

[0064] Perform text convolutional semantic encoding on each automotive battery - internal battery voltage semantic compensation text description in the set of automotive battery - internal battery voltage semantic compensation text descriptions to obtain the set of automotive battery - internal battery voltage semantic compensation text feature matrices. This process can be expressed as:

[0065] M i =TextCNN{T i}

[0066] where T i is the automotive battery - internal battery voltage semantic compensation text description corresponding to S i , TextCNN is text convolutional encoding, and M i is the automotive battery - internal battery voltage semantic compensation text feature matrix corresponding to S i .

[0067] It should be understood that by performing semantic compensation decoding based on large language for each automotive battery - internal battery voltage fine - grained local shared semantic feature matrix in the set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices, the powerful language understanding and generation capabilities of the large language model can be utilized to generate corresponding semantic representations according to the feature information in the automotive battery - internal battery voltage fine - grained local shared semantic feature matrix. This is beneficial to increasing the interpretability and understandability of the input features. At the same time, the large language model can also supplement the input information with the knowledge stored in itself, enhancing the semantic richness of the features, which helps to have a more comprehensive understanding of the relationship between the automotive battery and the internal battery voltage, as well as their respective state performances. Subsequently, perform text convolutional semantic encoding on each automotive battery - internal battery voltage semantic compensation text description in the set of automotive battery - internal battery voltage semantic compensation text descriptions to achieve semantic feature alignment of each automotive battery - internal battery voltage semantic compensation text description, so as to more conveniently apply the semantic compensation information contained in the automotive battery - internal battery voltage semantic compensation text description to the obtained automotive battery - internal battery voltage fine - grained local shared semantic feature matrix, thereby further expanding and deepening the understanding of the battery voltage state.

[0068] Specifically, in the embodiment of the present application, the battery voltage feature aggregation sub - unit is used to: perform fine - grained semantic interaction compensation on each pair of corresponding automotive battery - internal battery voltage fine - grained local shared semantic feature matrix and automotive battery - internal battery voltage semantic compensation text feature matrix in the set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices and the set of automotive battery - internal battery voltage semantic compensation text feature matrices to obtain the set of automotive battery - internal battery voltage local interaction fine - grained semantic enhancement feature matrices. This process can be expressed as:

[0069]

[0070] Among them, S i is F 1i and F 2i is the fine-grained local shared semantic feature matrix of the automotive battery - internal battery voltage between F and F, M i T is M i 's transposed matrix, is matrix multiplication, D is the scale of the said M i , that is, the width of the matrix multiplied by the height of the matrix, softmax(·) is the softmax function, α and β are weighted coefficients respectively, S bi is the i-th automotive battery - internal battery voltage local interaction fine-grained semantic enhancement feature matrix in the set of automotive battery - internal battery voltage local interaction fine-grained semantic enhancement feature matrices;

[0071] Aggregating the set of the automotive battery - internal battery voltage local interaction fine-grained semantic enhancement feature matrices along the channel dimension to obtain the automotive battery - internal battery voltage temporal-spatial domain semantic interaction encoding feature map, this process can be expressed as:

[0072] F f = couple{S b1 , S b2 ,..., S bi ,..., S bn}

[0073] Among them, S b1 , S b2 , S bi and S bn are respectively the 1st, 2nd, i-th and n-th automotive battery - internal battery voltage local interaction fine-grained semantic enhancement feature matrices in the set of automotive battery - internal battery voltage local interaction fine-grained semantic enhancement feature matrices, n is the number of feature matrices in the set of the automotive battery - internal battery voltage local interaction fine-grained semantic enhancement feature matrices, couple{·,·,...,·} is the aggregation of the set of feature matrices along the channel dimension, F f is the automotive battery - internal battery voltage temporal-spatial domain semantic interaction encoding feature map.

[0074] It should be understood that by performing fine-grained semantic interaction compensation on each corresponding automotive battery-internal battery voltage fine-grained local shared semantic feature matrix and automotive battery-internal battery voltage semantic compensation text feature matrix in the set of automotive battery-internal battery voltage fine-grained local shared semantic feature matrices and the set of automotive battery-internal battery voltage semantic compensation text feature matrices, the fine-grained shared semantic features between the automotive battery voltage time-series spatial domain feature map and the internal battery voltage time-series spatial domain feature map, as well as the semantic compensation features between the automotive battery voltage time-series spatial domain feature map and the internal battery voltage time-series spatial domain feature map, can interact with each other at the fine-grained level. This can improve the integrity of the fine-grained interaction between the two feature maps. Finally, the set of automotive battery-internal battery voltage local interaction fine-grained semantic enhancement feature matrices is aggregated along the channel dimension to obtain the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature map. In this way, it helps to elevate the feature information mined and enhanced at the fine-grained level from the local level to a more globally-visioned feature level, so as to better grasp the overall semantic interaction information of the automotive battery and the internal battery voltage in the time-series and spatial domains.

[0075] In the embodiment of the present application, the control result generation unit 133 is configured to obtain a control result based on the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature, and control the charge and discharge control module based on the control result. Specifically, Figure 5 It is a block diagram of the control result generation unit in the automotive emergency start power supply control system according to the embodiment of the present application. As Figure 5 shown, the control result generation unit 133 includes: an internal battery charging judgment subunit 1331, configured to obtain the control result by passing the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature map through a charging mode controller based on a classifier, where the control result is used to indicate whether to charge the internal battery of the automotive emergency power supply; a charging control result response subunit 1332, configured to, in response to the control result being to charge the internal battery of the automotive emergency power supply, the MCU control module controls the charge and discharge control module to charge the internal battery of the automotive emergency power supply.

[0076] In the embodiment of the present application, the internal battery charging judgment subunit 1331 is configured to obtain the control result by passing the automotive battery-internal battery voltage time-sequence spatial domain semantic interaction encoding feature map through a classifier-based charging mode controller, and the control result is used to indicate whether to charge the internal battery of the automotive emergency power supply. Specifically, in the embodiment of the present application, the internal battery charging judgment subunit is configured to: expand each automotive battery-internal battery voltage time-sequence spatial domain semantic interaction encoding feature matrix in the automotive battery-internal battery voltage time-sequence spatial domain semantic interaction encoding feature map into a one-dimensional feature vector according to a row vector or a column vector and then cascade them to obtain a charging control classification feature vector; perform fully connected encoding on the charging control classification feature vector using the fully connected layer of the classifier to obtain a charging control fully connected encoding feature vector; input the charging control fully connected encoding feature vector into the Softmax classification function of the classifier to obtain the probability values of the automotive battery-internal battery voltage time-sequence spatial domain semantic interaction encoding feature map belonging to each classification label, where the classification labels include those for indicating that the internal battery of the automotive emergency power supply needs to be charged and those for indicating that the internal battery of the automotive emergency power supply does not need to be charged; and determine the classification label corresponding to the maximum value among the probability values as the control result.

[0077] It should be understood that the automotive battery-internal battery voltage time-sequence spatial domain semantic interaction encoding feature map contains rich battery state information. To map this abstract and complex feature representation form to a specific charging decision (i.e., the control result), it is necessary to input the automotive battery-internal battery voltage time-sequence spatial domain semantic interaction encoding feature map into a classifier-based charging mode controller. Specifically, a classifier is a machine learning model that can analyze and judge according to the features of the input data and map it to different categories. It is trained with a large amount of labeled data, which includes the voltage time-sequence data of different automotive batteries and internal batteries and the corresponding charging decisions. During the training process, the classifier continuously adjusts its own parameters (weights, biases, etc.) to minimize the error between the prediction result and the actual labeled result. After training, when a new automotive battery-internal battery voltage time-sequence spatial domain semantic interaction encoding feature map is input, the classifier analyzes and judges the feature map according to the learned patterns and parameters, and outputs the corresponding charging decision result, that is, whether to charge the internal battery of the automotive emergency power supply, so as to accurately control the charging of the internal battery of the emergency power supply.

[0078] Here, considering that the automotive battery voltage time-series spatial domain feature map and the internal battery voltage time-series spatial domain feature map respectively represent the image semantic features of the time-series recurrence maps of the automotive battery voltage and the internal battery voltage of the automotive emergency power supply, when performing cross-domain fine-grained feature global interaction encoding, due to the cross-domain fine-grained semantic interaction differences caused by the differences in the distribution of image semantic features resulting from the differences in the time-series distribution of data sources, the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature map will have a probability convergence and divergence based on different feature encoding interactivity, thus affecting the accuracy of the control result obtained through the charging mode controller based on the classifier.

[0079] Preferably, obtaining a control result by passing the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature map through a charging mode controller based on a classifier includes:

[0080] Determining the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding distribution probability value p obtained by passing the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature map through the charging mode controller based on the classifier, where the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding distribution probability value p represents the probability value of charging the internal battery of the automotive emergency power supply;

[0081] Performing maximum normalization on the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature map to obtain an automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding probability feature map F;

[0082] Calculating the power function of each eigenvalue of the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding probability feature map with one minus the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding distribution probability value as the exponent to obtain a first automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding convergence feature map, and this process can be expressed as:

[0083] F1 = F ⊙(1-p)

[0084] where F represents the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding probability feature map, p represents the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding distribution probability value, ⊙ represents element-wise multiplication, and F ⊙(1-p) represents calculating the power function of each eigenvalue in F with one minus p as the exponent, and F1 represents the first automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding convergence feature map;

[0085] Taking the exponential calculation unit eigenmap of the difference between each eigenvalue of the eigenmap of the point difference between the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding distribution probability value and the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding probability eigenmap to obtain the second automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding convergence eigenmap, this process can be expressed as:

[0086]

[0087] where F represents the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding probability eigenmap, p represents the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding distribution probability value, F I represents the unit eigenmap, and ⊙ respectively represent point - by - point subtraction and point - by - point multiplication by position, represents the power function of each eigenvalue of the eigenmap obtained by point - by - point subtraction of F I and F with p as the exponent, and F2 represents the second automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding convergence eigenmap;

[0088] Multiplying the difference between the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding probability eigenmap and the value of one minus the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding distribution probability value by position to obtain the first automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding restricted eigenmap, this process can be expressed as:

[0089] F3 = F⊙(1 - p)

[0090] where F represents the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding probability eigenmap, p represents the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding distribution probability value, ⊙ represents point - by - point multiplication by position, and F3 represents the first automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding restricted eigenmap;

[0091] Multiplying the point - difference eigenmap by the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding distribution probability value by position to obtain the second automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding restricted eigenmap, this process can be expressed as:

[0092]

[0093] where F represents the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding probability eigenmap, p represents the automotive battery - internal battery voltage temporal - spatial domain semantic interaction coding distribution probability value, F I represents the unit eigenmap, ⊖ and ⊙ respectively represent pointwise subtraction and pointwise multiplication by position. F4 represents the semantic interaction coding limit feature map of the second automotive battery - internal battery voltage time - series spatial domain;

[0094] After multiplying the first automotive battery - internal battery voltage time - series spatial domain semantic interaction coding convergence feature map and the second automotive battery - internal battery voltage time - series spatial domain semantic interaction coding convergence feature map point - by - point, perform point - addition with the first automotive battery - internal battery voltage time - series spatial domain semantic interaction coding limit feature map and the second automotive battery - internal battery voltage time - series spatial domain semantic interaction coding limit feature map to obtain the optimized automotive battery - internal battery voltage time - series spatial domain semantic interaction coding feature map. This process can be expressed as:

[0095]

[0096] Among them, F1 represents the first automotive battery - internal battery voltage time - series spatial domain semantic interaction coding convergence feature map, F2 represents the second automotive battery - internal battery voltage time - series spatial domain semantic interaction coding convergence feature map, and F3 represents the first automotive battery - internal battery voltage time - series spatial domain semantic interaction coding limit feature map; ⊕ and ⊙ respectively represent pointwise addition and pointwise multiplication by position. F4 represents the semantic interaction coding limit feature map of the second automotive battery - internal battery voltage time - series spatial domain, and F' represents the optimized automotive battery - internal battery voltage time - series spatial domain semantic interaction coding feature map;

[0097] Input the optimized automotive battery - internal battery voltage time - series spatial domain semantic interaction coding feature map into the classifier - based charging mode controller to obtain the control result.

[0098] That is, by using the cross-entropy formal power series of the regression analysis probability of the semantic interaction coding feature map in the time-series space domain of the automotive battery - internal battery voltage as the probability distribution convergence limit, the class probability convergence approximation of the semantic interaction coding feature map in the time-series space domain of the automotive battery - internal battery voltage is carried out on the basis of the combination of the feature set distribution and the probability density distribution of the semantic interaction coding feature map in the time-series space domain of the automotive battery - internal battery voltage, so as to further use the probability distribution cross-entropy of the semantic interaction coding feature map in the time-series space domain of the automotive battery - internal battery voltage as the objective function to guide the limit recovery strategy, realizing the common agility of the convergence of the feature set with probability convergence divergence of the semantic interaction coding feature map in the time-series space domain of the automotive battery - internal battery voltage to the probability density distribution space, and improving the accuracy of the control result obtained by the charging mode controller based on the classifier for the semantic interaction coding feature map in the time-series space domain of the automotive battery - internal battery voltage.

[0099] In the embodiment of the present application, the charging control result response sub-unit 1332 is configured to, in response to the control result being to charge the internal battery of the automotive emergency power supply, the MCU control module controls the charge and discharge control module to charge the internal battery of the automotive emergency power supply. In this way, it can ensure that the automotive emergency power supply is always in the best state, improve the system reliability and safety, and this reasonable charging operation also helps to extend the service life of the internal battery of the automotive emergency power supply and maintain its good performance.

[0100] In summary, the MCU control module is clearly described. It uses data analysis and processing algorithms based on deep learning to recursively process and extract spatial domain features from the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply, and thus intelligently determines whether to charge the internal battery of the automotive emergency power supply according to the fine-grained spatial semantic interaction representation between the time-series space domain features of the automotive battery voltage and the time-series space domain features of the internal battery voltage, and controls the charge and discharge control module. In this way, it can capture the subtle patterns and trends of voltage changes and dynamically adjust the charging strategy to ensure the best performance in various usage scenarios. This not only enhances the flexibility of the system but also solves the problem that the fixed threshold method cannot cope with environmental changes, thus effectively protecting the battery performance and extending its service life.

[0101] In the embodiment of the present application, the charge and discharge control module 140 is configured to determine whether to charge the internal battery of the automotive emergency power supply based on the control result. It should be understood that as a backup power supply for vehicles, the core function of the automotive emergency starting power supply is to quickly and reliably provide starting power when the vehicle battery has insufficient power or fails, enabling the vehicle to operate normally. The charge and discharge control module can accurately charge the internal battery of the automotive emergency power supply according to the control result generated by the MCU control module. For example, after the vehicle has been parked for a long time, the vehicle battery's power decreases due to natural discharge and other reasons. The charge and discharge control module starts the charging program for the internal battery according to the control result to ensure that the internal battery has sufficient power, which greatly improves the success rate and reliability of vehicle emergency starting.

[0102] In summary, the automotive emergency starting power supply control system 100 based on the embodiment of the present application is elucidated. It first collects the voltage time series data of the vehicle battery and the internal battery of the automotive emergency power supply, and then transmits this voltage time series data to the MCU control module. The MCU control module deeply analyzes this voltage time series data to obtain the corresponding control result, and based on this control result, performs charging or discharging operations on the vehicle battery or the internal battery of the automotive emergency power supply. In this way, real-time monitoring and intelligent management of the states of the vehicle battery and the internal battery of the automotive emergency power supply can be achieved, which is beneficial to improving the reliability and practicality of the automotive emergency power supply at critical moments, and further can provide strong guarantee for the normal operation of the vehicle.

[0103] In the embodiment of the present application, there is also provided an automotive emergency starting power supply control method for implementing the above-mentioned automotive emergency starting power supply control system. Figure 6 The flowchart of the automotive emergency starting power supply control method according to the embodiment of the present application is as follows. As Figure 6 shown, in the automotive emergency starting power supply control method, it includes: S110, respectively performing recursive processing and spatial domain feature extraction on the voltage time series data of the vehicle battery and the voltage time series data of the internal battery of the automotive emergency power supply to obtain the spatial domain features of the vehicle battery voltage time series and the spatial domain features of the internal battery voltage time series; S120, performing fine-grained spatial domain interaction encoding on the spatial domain features of the vehicle battery voltage time series and the spatial domain features of the internal battery voltage time series to obtain the semantic interaction encoding features of the vehicle battery-internal battery voltage time series in the spatial domain; S130, based on the semantic interaction encoding features of the vehicle battery-internal battery voltage time series in the spatial domain, obtaining a control result, and based on the control result, controlling the charge and discharge control module.

[0104] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned automotive emergency starting power supply control method have been described above with reference to Figures 1 to 5has been introduced in detail in the description of the automotive emergency starting power supply control system, and therefore, its repeated description will be omitted.

[0105] In summary, the automotive emergency starting power supply control method based on the embodiments of the present application is clarified. First, the voltage time series data of the automotive battery and the internal battery of the automotive emergency power supply are collected, and then these voltage time series data are transmitted to the MCU control module. The MCU control module deeply analyzes these voltage time series data to obtain corresponding control results, and based on this control result, charging or discharging operations are performed on the automotive battery or the internal battery of the automotive emergency power supply. In this way, real-time monitoring and intelligent management of the states of the automotive battery and the internal battery of the automotive emergency power supply can be realized, which is beneficial to improving the reliability and practicability of the automotive emergency power supply at critical moments, and further can provide strong guarantee for the normal operation of the vehicle.

Claims

1. An emergency starting power supply control system for an automobile, characterized in that, Including: A voltage detection module, configured to collect the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply; A data transmission module, configured to transmit the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply to the MCU control module; The MCU control module, configured to analyze the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply to obtain a control result; A charge and discharge control module, configured to determine whether to charge the internal battery of the automotive emergency power supply based on the control result; wherein, the MCU control module includes: A battery voltage time-series analysis unit, configured to perform recursive processing and spatial domain feature extraction on the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply respectively to obtain the spatial domain feature of the automotive battery voltage time series and the spatial domain feature of the internal battery voltage time series; An automotive battery-internal voltage interaction unit, configured to perform fine-grained spatial domain interaction coding on the spatial domain feature of the automotive battery voltage time series and the spatial domain feature of the internal battery voltage time series to obtain the spatial domain semantic interaction coding feature of the automotive battery-internal battery voltage time series; A control result generation unit, configured to obtain a control result based on the spatial domain semantic interaction coding feature of the automotive battery-internal battery voltage time series, and control the charge and discharge control module based on the control result.

2. The automotive emergency starting power supply control system according to claim 1, wherein The battery voltage time-series analysis unit includes: A battery voltage time-series data recurrence graph calculation sub-unit, configured to calculate the recurrence graphs of the time-series data of the automotive battery voltage and the time-series data of the internal battery voltage of the automotive emergency power supply respectively to obtain the recurrence graph of the automotive battery voltage time series and the recurrence graph of the internal battery voltage time series; A battery voltage time-series spatial domain feature extraction sub-unit, configured to obtain the spatial domain feature map of the automotive battery voltage time series as the spatial domain feature of the automotive battery voltage time series and the spatial domain feature map of the internal battery voltage time series as the spatial domain feature of the internal battery voltage time series by passing the recurrence graph of the automotive battery voltage time series and the recurrence graph of the internal battery voltage time series through a voltage time-series spatial domain feature extractor based on MCNN.

3. The automotive emergency starting power supply control system according to claim 2, wherein, The automotive battery-internal voltage interaction unit includes: A battery voltage shared feature interaction sub-unit, configured to perform feature decomposition on the spatial domain feature map of the automotive battery voltage time series and the spatial domain feature map of the internal battery voltage time series, and calculate the shared semantic features between the set of local spatial domain semantic feature matrices of the automotive battery voltage time series after decomposition and the set of local spatial domain semantic feature matrices of the internal battery voltage time series to obtain the set of fine-grained local shared semantic feature matrices of the automotive battery-internal battery voltage; A battery voltage semantic compensation text sub-unit, configured to calculate the semantic compensation text features of each fine-grained local shared semantic feature matrix of the automotive battery-internal battery voltage in the set of fine-grained local shared semantic feature matrices of the automotive battery-internal battery voltage to obtain the set of semantic compensation text feature matrices of the automotive battery-internal battery voltage. The battery voltage feature aggregation subunit is used to perform weighted aggregation on the set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices based on the set of automotive battery - internal battery voltage semantic compensation text feature matrices to obtain the automotive battery - internal battery voltage temporal - spatial domain semantic interaction encoding feature map as the automotive battery - internal battery voltage temporal - spatial domain semantic interaction encoding feature.

4. The automotive emergency starting power supply control system according to claim 3, wherein, The battery voltage shared feature interaction subunit is used for: Performing feature decomposition on the automotive battery voltage temporal - spatial domain feature map and the internal battery voltage temporal - spatial domain feature map along the channel dimension of the automotive battery voltage temporal - spatial domain feature map and the internal battery voltage temporal - spatial domain feature map to obtain the set of automotive battery voltage temporal local spatial domain semantic feature matrices and the set of internal battery voltage temporal local spatial domain semantic feature matrices; Performing shared semantic feature extraction with a twin structure on each pair of corresponding channel - dimension automotive battery voltage temporal local spatial domain semantic feature matrices and internal battery voltage temporal local spatial domain semantic feature matrices in the set of automotive battery voltage temporal local spatial domain semantic feature matrices and the set of internal battery voltage temporal local spatial domain semantic feature matrices to obtain the set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices.

5. The automotive emergency starting power supply control system according to claim 4, characterized in that The battery voltage semantic compensation text subunit is used for: Performing semantic compensation decoding based on large language on each automotive battery - internal battery voltage fine - grained local shared semantic feature matrix in the set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices to obtain the set of automotive battery - internal battery voltage semantic compensation text descriptions; Performing text convolutional semantic encoding on each automotive battery - internal battery voltage semantic compensation text description in the set of automotive battery - internal battery voltage semantic compensation text descriptions to obtain the set of automotive battery - internal battery voltage semantic compensation text feature matrices.

6. The automotive emergency starting power supply control system according to claim 5, characterized in that, The battery voltage feature aggregation subunit is used for: Performing fine - grained semantic interaction compensation on each pair of corresponding automotive battery - internal battery voltage fine - grained local shared semantic feature matrices and automotive battery - internal battery voltage semantic compensation text feature matrices in the set of automotive battery - internal battery voltage fine - grained local shared semantic feature matrices and the set of automotive battery - internal battery voltage semantic compensation text feature matrices to obtain the set of automotive battery - internal battery voltage local interaction fine - grained semantic enhanced feature matrices; Aggregating the set of automotive battery - internal battery voltage local interaction fine - grained semantic enhanced feature matrices along the channel dimension to obtain the automotive battery - internal battery voltage temporal - spatial domain semantic interaction encoding feature map.

7. The automotive emergency starting power supply control system according to claim 6, characterized in that, The control result generation unit includes: The internal battery charging judgment subunit is used to obtain the control result by passing the automotive battery - internal battery voltage temporal - spatial domain semantic interaction encoding feature map through a charging mode controller based on a classifier, and the control result is used to indicate whether to charge the internal battery of the automotive emergency power supply. A charging control result response subunit, configured to, in response to the control result being to charge the internal battery of the vehicle emergency power supply, control, by the MCU control module, the charge and discharge control module to charge the internal battery of the vehicle emergency power supply.

8. The automotive emergency starting power supply control system according to claim 7, wherein The internal battery charging judgment subunit is configured to: Expand each of the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature matrices in the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature map into one-dimensional feature vectors by row vectors or column vectors and then cascade them to obtain a charging control classification feature vector; Perform fully connected encoding on the charging control classification feature vector using the fully connected layer of the classifier to obtain a charging control fully connected encoding feature vector; Input the charging control fully connected encoding feature vector into the Softmax classification function of the classifier to obtain the probability values of the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature map belonging to each classification label, where the classification labels include those for indicating that the internal battery of the vehicle emergency power supply needs to be charged and those for indicating that the internal battery of the vehicle emergency power supply does not need to be charged; Determine the classification label corresponding to the maximum value among the probability values as the control result.

9. A control method for an automotive emergency starting power supply, which is used to execute the automotive emergency starting power supply control system as described in claim 1, characterized in that, The vehicle emergency starting power supply control method includes: Respectively perform recursive processing and spatial domain feature extraction on the automotive battery voltage time-series data and the internal battery voltage time-series data of the vehicle emergency power supply to obtain an automotive battery voltage time-series spatial domain feature and an internal battery voltage time-series spatial domain feature; Perform fine-grained spatial domain interaction encoding on the automotive battery voltage time-series spatial domain feature and the internal battery voltage time-series spatial domain feature to obtain an automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature; Based on the automotive battery-internal battery voltage time-series spatial domain semantic interaction encoding feature, obtain a control result, and based on the control result, control the charge and discharge control module.

10. The automotive emergency starting power supply control method according to claim 9, characterized in that, Respectively performing recursive processing and spatial domain feature extraction on the automotive battery voltage time-series data and the internal battery voltage time-series data of the vehicle emergency power supply to obtain an automotive battery voltage time-series spatial domain feature and an internal battery voltage time-series spatial domain feature includes: Respectively calculate the recurrence graphs of the automotive battery voltage time-series data and the internal battery voltage time-series data of the vehicle emergency power supply to obtain an automotive battery voltage time-series recurrence graph and an internal battery voltage time-series recurrence graph; Pass the automotive battery voltage time-series recurrence graph and the internal battery voltage time-series recurrence graph through a voltage time-series spatial domain feature extractor based on MCNN to obtain an automotive battery voltage time-series spatial domain feature map as the automotive battery voltage time-series spatial domain feature and an internal battery voltage time-series spatial domain feature map as the internal battery voltage time-series spatial domain feature.

Images