Method and system for monitoring charging states of different devices

Abstract

The invention discloses a method and system for monitoring the charging states of different devices. The method comprises the steps that a device charging data set related to the charging states is collected; calculating charging stability indexes of charging stability degrees of different devices according to the device charging data set; extracting key charging characteristics with the influence degree on the charging state higher than a preset threshold value in the equipment charging data set; and according to the charging stability index and the key charging characteristics, utilizing a pre-trained charging state identification model based on a deep learning network to predict the charging states of different devices. According to the embodiment of the invention, the method can achieve the efficient prediction of the charging state through the deep learning technology, achieves an intelligent and precise charging monitoring scheme, and improves the safety, effectiveness and user experience of charging.

Description

Technical Field

[0001] The present invention belongs to the technical field of monitoring, and particularly relates to a method and system for monitoring the charging status of different devices. Background Art

[0002] With the wide application of electronic devices, the progress of charging technology has become an important research field. In modern intelligent devices, the monitoring of the charging status directly affects the usage efficiency, safety, and battery life of the devices. Traditional charging monitoring methods mainly rely on the simple collection and analysis of static parameters (such as current, voltage, etc.), and it is difficult to comprehensively reflect the complex status and dynamic changes of the devices during the actual charging process.

[0003] During the charging process, there are various factors affecting the charging status, including device type, charger specifications, environmental temperature and humidity, etc. These factors not only affect the charging speed but also have a significant impact on the charging stability. The evaluation of charging stability is crucial for ensuring the safe use of the battery and extending the service life of the device. Existing research shows that fluctuations during the charging process can lead to problems such as battery overheating and overcharging, which can cause battery damage or device failure. In addition, the instability during the charging process also affects the energy efficiency of the device, increases the waste of electric energy, and causes dissatisfaction and economic losses to users. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and system for monitoring the charging status of different devices to solve the deficiencies in the prior art, and to be able to use deep learning technology for efficient prediction of the charging status, so as to achieve an intelligent and precise charging monitoring solution, thereby improving the safety, effectiveness, and user experience of charging.

[0005] An embodiment of the present application provides a method for monitoring the charging status of different devices, and the method includes: Collecting a device charging data set related to the charging status; Calculating a charging stability index of the charging stability degree of different devices according to the device charging data set; Extracting key charging features in the device charging data set whose influence degree on the charging status is higher than a preset threshold; Predicting the charging status of different devices according to the charging stability index and the key charging features by using a pre-trained charging status recognition model based on a deep learning network.

[0006] Optionally, the calculation formula of the charging stability index is:

[0007] wherein, the is the maximum charging current during the charging process, and the is the minimum charging current during the charging process. is the i-th measured voltage during the charging process, is the average value of all measured voltages, is the maximum ambient temperature during charging. is the minimum ambient temperature during charging. is the average ambient temperature during charging. To avoid the minimum value of the division by zero case, the , , is the influence weight of charging current, charging voltage and ambient temperature on charging stability, and N is the number of measurements during the charging process.

[0008] Optionally, extracting key charging features from the device charging data set whose influence on the charging state is higher than a preset threshold includes: All charging features in the charging data set are regarded as nodes in the charging feature graph to be constructed, and each node represents a charging feature; Creating edges based on the correlation between charging features. If the correlation between two charging features exceeds a preset correlation threshold, an edge is established between the two charging features to form a charging feature graph. The weight of each edge is set to the corresponding correlation value between the two charging features. identifying connected components in the charging feature graph by a graph theory algorithm, wherein the connected components can identify interdependent charging feature groups; For each connected component, calculate the overall weight of the connected component, where the overall weight is the sum or average of the edge weights that constitute the charging characteristics in the connected component. A connected component with a high overall weight indicates that its constituent characteristics are more closely related to the charging state. The charging feature corresponding to the connected component whose overall weight is higher than a preset weight threshold is determined as the key charging feature.

[0009] Optionally, predicting the charging status of different devices according to the charging stability index and the key charging characteristics by using a pre-trained charging status recognition model based on a deep learning network includes: Inputting the charging stability index into a pre-trained first charging state recognition model based on a deep learning network, and outputting the most likely first charging state of the corresponding device and its first probability value; Performing a product operation on the first probability value and a first weight corresponding to the charging stability index to obtain a first weighted probability value; Input the key charging feature into a pre-trained second charging state recognition model based on a deep learning network, and output the most likely second charging state of the corresponding device and its second probability value; Perform a multiplication operation on the second probability value and the second weight corresponding to the key charging feature to obtain a second weighted probability value; Determine the charging state corresponding to the maximum weighted probability value among the first weighted probability value and the second weighted probability value as the final charging state of the corresponding device.

[0010] Another embodiment of the present application provides a system for monitoring the charging states of different devices, and the system includes: An acquisition module, configured to acquire a device charging data set related to the charging state; A calculation module, configured to calculate a charging stability index of the charging stability degree of different devices according to the device charging data set; An extraction module, configured to extract key charging features in the device charging data set whose influence degree on the charging state is higher than a preset threshold; A prediction module, configured to predict the charging states of different devices according to the charging stability index and the key charging features by using a pre-trained charging state recognition model based on a deep learning network.

[0011] Another embodiment of the present application provides a storage medium, in which a computer program is stored, and wherein the computer program is set to execute the method described in any one of the above when running.

[0012] Another embodiment of the present application provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is set to run the computer program to execute the method described in any one of the above.

[0013] Compared with the prior art, a method for monitoring the charging states of different devices provided by the present invention acquires a device charging data set related to the charging state; calculates a charging stability index of the charging stability degree of different devices according to the device charging data set; extracts key charging features in the device charging data set whose influence degree on the charging state is higher than a preset threshold; predicts the charging states of different devices according to the charging stability index and the key charging features by using a pre-trained charging state recognition model based on a deep learning network, so that the charging state can be efficiently predicted by using deep learning technology to implement an intelligent and accurate charging monitoring solution, thereby improving the safety, effectiveness and user experience of charging. Description of the Drawings

[0014] Figure 1Hardware block diagram of a computer terminal for a method of monitoring the charging status of different devices provided by an embodiment of the present invention; Figure 2 Flow schematic diagram of a method of monitoring the charging status of different devices provided by an embodiment of the present invention; Figure 3 Structural schematic diagram of a system for monitoring the charging status of different devices provided by an embodiment of the present invention. Detailed implementation manners

[0015] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0016] An embodiment of the present invention first provides a method for monitoring the charging status of different devices. This method can be applied to electronic devices, such as computer terminals, specifically ordinary computers, etc.

[0017] The following takes running on a computer terminal as an example to describe it in detail. Figure 1 Hardware block diagram of a computer terminal for a method of monitoring the charging status of different devices provided by an embodiment of the present invention. As Figure 1 shown, the computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the memory can include a non-volatile storage medium and an internal memory.

[0018] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions. When the program instructions are executed, the processor can execute any method for monitoring the charging status of different devices.

[0019] The processor is used to provide computing and control capabilities to support the operation of the entire computer device.

[0020] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any method for monitoring the charging status of different devices.

[0021] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art can understand that Figure 1 the structure shown in is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0022] It should be understood that the processor may be a Central Processing Unit (CPU), and the processor may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0023] See Figure 2 , embodiments of the present invention provide a method for monitoring the charging status of different devices, which may include the following steps: S201, collect a device charging data set related to the charging status; In this method, the first step is to collect a device charging data set related to the charging status. This process involves collecting a series of charging-related parameters and information from various devices (such as smartphones, tablets, power tools, etc.). These parameters include charging current, charging voltage, ambient temperature, charging time, etc. In addition, factors such as the charger type, usage status, and remaining battery power of the device can also be recorded. To ensure the representativeness and accuracy of the data, the collection should be carried out under different environmental conditions and charging scenarios to form a large-scale and diverse data set. For example, for smartphones, data can be recorded in different situations such as indoors, outdoors, using high-power applications during charging, and standby status to ensure that a wide range of charging characteristics and variations are included. These data will provide a basis for subsequent charging stability analysis and charging status prediction.

[0024] Collecting a device charging data set related to the charging status plays a crucial role in the entire charging status monitoring process. First of all, this data set provides the raw material for the algorithm model, allowing the model to understand various dynamic changes and influencing factors in the charging process by learning and analyzing the characteristics of different charging statuses. Secondly, accurate data collection can help identify potential anomalies in the charging process, such as unstable charging, overheating, etc., thereby improving the safety and user experience of the device. In addition, through the analysis of a large-scale data set, the association between charging characteristics can be automatically discovered, providing a more reliable basis for the calculation of the charging stability index and the extraction of key characteristics. Ultimately, the successful implementation of this link is the basis for realizing intelligent charging status prediction, which can significantly improve the charging efficiency and safety of the device.

[0025] In the specific implementation process, the first step is to select appropriate sensors and data recording devices. This includes current sensors, temperature sensors, voltage sensors, etc. These sensors will be integrated inside the device or charger, or connected through an external module to facilitate real-time monitoring of relevant parameters during device charging. For example, a Hall current sensor with higher precision can be selected to accurately measure the charging current, and a digital temperature sensor can be used to obtain the temperature data of the charging environment. These sensors need to be connected to a central data acquisition module, which is responsible for collecting, storing, and processing the sensor data. In addition, to make the example more intuitive, a small external box can be designed near the charging interface of the smartphone, with all sensors integrated inside. This design can ensure the accuracy of data acquisition without affecting the user experience.

[0026] The second step is to develop a corresponding software system in the data acquisition module to ensure the real-time and reliability of data. The software system needs to be able to automatically identify the charging status of the device and automatically start the data recording function when charging begins. The system will collect current, voltage, and temperature data at regular intervals and store these data in the local database according to timestamps. To improve the reliability of data, the data acquisition frequency can be set, such as once per second, to ensure that subtle changes in charging parameters can be captured during the charging process. And, to avoid data loss, a data backup mechanism can be set to synchronize the real-time data to the cloud or local server. If any abnormal situation occurs during the charging process, users can view the charging history through the application to facilitate subsequent analysis and troubleshooting.

[0027] The third step is to build a user-friendly interface that allows users to intuitively view and manage the charging data. This can be achieved through a mobile application or computer software. When charging, users can view the changing trends of charging current, voltage, and ambient temperature in real time, as well as the real-time calculation results of the charging stability index. The user interface should be designed to be simple and clear, including charts and graphs, so that users can easily understand the various parameters during the charging process. For example, the application can display a real-time monitoring graph showing the curves of charging current and voltage changing over time, and at the same time display the change in ambient temperature. This information can help users judge the charging stability of the device, such as whether there are abnormal fluctuations or overheating conditions. In addition, users can set personalized threshold alerts, and when a certain parameter exceeds the preset range, the system can automatically issue an alert to remind users to take corresponding measures. In this way, users can not only monitor the charging status in real time but also enhance their awareness of proactive management of device charging safety.

[0028] Exemplarily, a smart phone is selected as the test device. On this device, multiple sensors are integrated to capture data related to the charging process. First, a high-precision Hall effect current sensor is installed on the back of the phone to monitor the change of the charging current in real time. Near the charging interface, a digital temperature sensor is installed to detect the ambient temperature during charging. In addition, a voltage sensor needs to be set on the output port of the charger to record the change of the charging voltage. These sensors are connected to the smart phone through a small data acquisition module to ensure seamless collection of relevant data during charging.

[0029] Next, a data acquisition system is built. This system needs to have an automatic start function. When the smart phone is connected to the charger, the system will automatically identify and activate the data acquisition program. The data acquisition software is designed to record the current, voltage and temperature parameters at regular intervals, and the sampling frequency is set to once per second. In this way, the system can capture the dynamic changes during the charging process. In terms of data storage, a local database is used to store the collected data, and at the same time, regular backups are implemented to the cloud to ensure data security. In addition, the software also has an anomaly detection function, which can monitor data fluctuations. If an anomaly (such as a sudden increase in current or too high temperature) is detected, the system will immediately issue an alarm.

[0030] To improve the user experience, a user-friendly interface is developed, enabling users to intuitively view the real-time data during charging. The interface includes multiple charts showing the change trends of the charging current, voltage and ambient temperature, as well as the charging stability index. When charging, users can monitor these parameters in real time through the application and set personalized thresholds. When a certain parameter exceeds the set range, the system will immediately notify the user. In addition, users can also view historical data and analyze the charging performance under different charging scenarios, such as the charging efficiency in high or low temperature environments, so as to provide data support for optimizing subsequent charging habits.

[0031] S202, calculate the charging stability index of the charging stability degree of different devices according to the device charging data set; The calculation of the charging stability index (CSI) is based on multiple key parameters during the charging process of the device, including the maximum and minimum values of the charging current, the fluctuation of the charging voltage, and the change of the ambient temperature. This calculation method reflects their impact on charging stability by assigning different weights to these parameters. Specifically, the change range of the charging current and the standard deviation of the charging voltage provide information about the fluctuations during the charging process, while the change of the ambient temperature affects the thermal management and safety of the device. By incorporating these factors into a unified index, CSI not only provides an intuitive stability indicator but also enables comparison and analysis of the charging performance between different devices.

[0032] The main significance of calculating the charging stability index lies in enhancing the safety and efficiency of the battery charging process. By real-time monitoring and evaluating the stability of the charging process, users and device manufacturers can promptly identify potential charging problems, such as overheating, overloading, or current fluctuations, which may lead to a decline in battery performance or damage. The effective application of CSI can provide necessary feedback to users, enabling them to take appropriate measures during charging and avoid unnecessary risks. At the same time, for device manufacturers, it also provides valuable data support for optimizing charging algorithms and designing more efficient chargers, helping to improve the user experience and the market competitiveness of the device.

[0033] Taking a smartphone as an example, during charging, the sensors built into the device will record the charging current, voltage, and ambient temperature in real time. When the user connects the charger, the system will automatically collect this data. Suppose during a charging process, the recorded charging current ranges from 1A (C_max) to 0.2A (C_min), the charging voltage fluctuates around 5V (V_i), and the ambient temperature changes between 10°C and 40°C. By inputting these data into the calculation formula of the charging stability index, the CSI value can be obtained.

[0034] In practical applications, if the CSI value shows high charging stability in a specific charging scenario, it means that the device has good charging performance under this environmental condition, and users can use it with confidence. However, if the CSI value is low, it may indicate that there are significant current fluctuations or overheating risks during the charging process of the device. Users need to pay attention to the charging method or environmental conditions and may even need to replace the charger. This method is not only used for users' charging management but also provides a basis for product optimization by device manufacturers, improving the overall performance and safety of the device.

[0035] Specifically, a calculation formula for the charging stability index can be:

[0036] This formula combines multiple indicators related to the stability of the charging process and quantifies them into a comprehensive index, the charging stability index (CSI). In this way, the charging stability of the device can be evaluated from multiple dimensions (current, voltage, and temperature), providing important reference information for users and manufacturers. This information not only helps to monitor the charging status of the device in real time but also provides data support for device optimization and risk management.

[0037] Among them, the is the maximum charging current during the charging process, which reflects the highest energy consumption state of the device during charging. A higher maximum current may indicate that the device encounters a larger load during charging, which may affect the charging stability. The is the minimum charging current during the charging process. Together with C_{max}, it shows the fluctuation range of the charging current. An overly low minimum current value may indicate an unstable charging process and insufficient current supply, which may lead to a decrease in charging efficiency or abnormal charging. The is the measured voltage at the i-th time during the charging process. The greater the voltage fluctuation, the higher the instability of the charging process may be. The is the average value of all measured voltages, which serves as a reference. The difference from the individual measured value V_i is used to reflect the degree of voltage dispersion, thereby evaluating the voltage stability of the entire charging process. The is the maximum ambient temperature during the charging process. Excessive temperature will affect battery performance and may pose safety hazards. The is the minimum ambient temperature during the charging process. Lower temperatures may lead to a decrease in battery charging efficiency. The is the average ambient temperature during the charging process, providing a comprehensive perspective on the temperature fluctuation and helping to monitor the overall impact of temperature on charging. The is a very small value to avoid division by zero, which is a very small constant, usually used to avoid division by zero during calculations and ensure the stability and effectiveness of the formula. The and the and the are the influence weights of the charging current, charging voltage, and ambient temperature on charging stability. N is the number of measurements during the charging process.

[0038] S203, extract the key charging features in the device charging dataset whose influence on the charging state is higher than the preset threshold; In this method, the step of extracting the key charging features in the device charging dataset whose influence on the charging state is higher than the preset threshold is to identify the features closely related to the charging state by establishing a charging feature map. First, all charging features in the charging dataset are regarded as nodes in the charging feature map, and each node represents a charging feature. Then, based on the preset correlation threshold, the correlation between charging features is analyzed to determine whether to establish edge connections between features. The connected components formed by these connections are identified through graph theory algorithms, thereby finding groups of interdependent features. This process can effectively screen out those features that play a key role in charging state prediction, making subsequent analysis and model construction more accurate and efficient.

[0039] Extracting key charging features with a high impact on the charging state helps to screen out the most representative and predictive features from complex charging data, thereby improving the performance and reliability of the charging state recognition model. By focusing on these key features, data redundancy can be significantly reduced, the training efficiency of the model can be improved, and faster response times and higher prediction accuracy can be achieved. This not only optimizes the charging management of the device but also provides a safer and more efficient charging experience for users.

[0040] Specifically, all charging features in the charging dataset can be regarded as nodes in the charging feature graph to be constructed, and each node represents a charging feature. At this stage, all charging features collected from the device charging dataset (such as current, voltage, temperature, etc.) are regarded as nodes in the graph. This graph structure can lay the foundation for modeling the relationships between subsequent features. By defining the nodes, the mutual influence and correlation of each charging feature can be clearly understood. By representing all charging features as nodes, the relationships between charging features can be systematically analyzed, which helps to better understand the interactions between features in subsequent steps and provides a clear framework for subsequent feature selection and model training.

[0041] First, the charging dataset needs to be preprocessed to extract all relevant charging features. These features may include current (I), voltage (V), temperature (T), and charging time, etc. Once the list of features to be processed is confirmed, an empty graph structure can be established, usually using an adjacency list or an adjacency matrix to represent the relationships between nodes and edges in the graph. In the data structure, a node is created for each charging feature, and a unique identifier (such as a number, letter, or string) is assigned to these nodes for easy subsequent reference.

[0042] Next, all these charging feature nodes need to be added to the node set of the charging feature graph. If the adjacency list storage method is used, a hash table or dictionary can be created, with the feature identifier as the key and the storage structure of the corresponding feature node as the value. In this process, the basic information of the feature, such as name, type, unit, etc., can be considered to be saved to enhance the readability of the graph and the convenience of subsequent analysis.

[0043] In this way, the constructed charging feature graph becomes a comprehensive graph structure, which can provide a data basis for subsequent feature relationship analysis and graph theory algorithms. Throughout the process, the unity of feature names should be ensured to avoid search confusion caused by inconsistent names and lay a solid foundation for subsequent steps.

[0044] Edges are created based on the correlation between charging characteristics. If the correlation between two charging characteristics exceeds a preset correlation threshold, an edge is established between these two charging characteristics to form a charging characteristic graph, and the weight of each edge is set to the correlation value between the corresponding two charging characteristics; This process calculates the correlation between each charging characteristic (such as using the Pearson correlation coefficient or other statistical methods) to determine the relationship between them. If the correlation between two characteristics exceeds the preset threshold, an edge is established between them. The weight of each edge is determined by its correlation value, thus forming a charging characteristic graph where the edge weight represents the degree of mutual influence between characteristics. By constructing edges based on correlation, it is possible to effectively identify those mutually related charging characteristics and form a network structure between the characteristics. This helps to find out which characteristics may have a significant impact on the change of the charging state, thus guiding the further optimization of the model.

[0045] In the process of creating the edges of the characteristic graph, it is first necessary to select an appropriate correlation calculation method, such as statistical methods like the Pearson correlation coefficient and Spearman rank correlation. Traverse each pair of charging characteristics in the dataset and calculate the correlation value between them. This requires the use of an appropriate statistical toolkit (such as NumPy or Pandas in Python) to achieve. For example, the corr() function of Pandas can be used to calculate the correlation matrix between each characteristic in the data frame.

[0046] Then, for the calculated correlation matrix, traverse all the characteristic pairs and check whether the correlation value of each pair of characteristics exceeds the preset correlation threshold. This step requires setting a reasonable threshold logically, which can be determined based on industry standards, previous research, or empirical data. If the correlation between two characteristics breaks through the threshold, an edge is established between the corresponding nodes in the characteristic graph, and its weight is set to their correlation value to represent the dependence relationship between these two characteristics.

[0047] When storing these edges, an adjacency list can be used to add the edges to the corresponding structure of the characteristic nodes. For example, if a Python dictionary is used to store the adjacency list, the weight of the edge can be stored in a nested list or tuple in the dictionary. After completing this process, the charging characteristic graph will be represented as a weighted directed graph (or undirected graph), and the weight of the edge directly reflects the correlation between the characteristics, which is convenient for subsequent analysis.

[0048] Identify the connected components in the charging characteristic graph through graph theory algorithms, and the connected components can identify groups of mutually dependent charging characteristics; Graph theory algorithms are used to identify the connected components in the charging feature graph, which are sets of nodes that are connected together. These connected components represent a high degree of correlation between features and are related to the possible impact on the charging state. This step helps to reveal the internal relationship patterns of charging features and find the set of relevant features. By finding the connected components of charging features, the interdependence between features can be understood more clearly, which will provide a basis for subsequent weight calculation and feature selection. At the same time, it can effectively reduce the number of features and lower the complexity of the model.

[0049] To identify the connected components in the charging feature graph, graph traversal algorithms such as depth-first search (DFS) or breadth-first search (BFS) can be used. First, an unvisited node is selected as the starting node according to the selected algorithm and marked as visited. Then, for each adjacent node of this node, the depth-first search or level traversal is recursively called until all reachable nodes are visited.

[0050] In this process, a list or set needs to be maintained to store the nodes in each connected component. Whenever a complete DFS or BFS traversal is completed, the set of visited nodes can be stored as a new connected component in the result set. This result set includes all the connected components identified from the graph, and each connected component consists of a series of interconnected charging feature nodes.

[0051] To ensure a complete traversal of the graph, the unvisited nodes need to be looped until all nodes are included in at least one connected component. Finally, a set including all connected components will be obtained, which will be directly used for subsequent analysis and screening of charging features to ensure the effectiveness and accuracy of the method.

[0052] For each connected component, calculate the overall weight of the connected component, where the overall weight is the sum or average of the edge weights of the charging features that make up the connected component. A connected component with a high overall weight indicates that the features it consists of have a closer relationship with the impact on the charging state; In this process, the edge weights in each connected component need to be calculated to obtain the overall weight of the connected component. This can be achieved by summing or averaging the edge weights that make it up. The higher the overall weight, the greater the potential impact of the feature group on the charging state. Calculating the overall weight helps to quantify the influence of the feature group on the charging state. This step can significantly improve the accuracy of feature selection, making the model pay more attention to those feature combinations with high weights during training, thereby improving the prediction accuracy.

[0053] When calculating the overall weight of connected components, first traverse each identified connected component. For each connected component, initialize the overall weight to zero, and then accumulate the weights of each edge in that connected component. If the sum of edge weights is selected for calculation, the weight of each edge is accumulated into the overall weight; if the average value is calculated, the number of edges also needs to be recorded during the accumulation for subsequent average value calculation.

[0054] This step also requires effective verification of edge weights to ensure that the edge weights are not lost during data storage. During the accumulation, the detailed information of the edges, such as the connected characteristic nodes, can be considered for recording to facilitate subsequent debugging and data analysis.

[0055] Once the weights of all connected components are calculated, the results can be stored in a data structure (such as a dictionary), where the keys are the identifiers of the connected components and the values are their overall weights. Next, these weights of the connected components can be compared with a preset weight threshold to determine which connected components have higher influence.

[0056] Determine the charging characteristics corresponding to the connected components whose overall weight is higher than the preset weight threshold as the key charging characteristics.

[0057] After determining the overall weight of each connected component, compare these weights with the previously set weight threshold. The connected components with weights higher than this threshold will be selected, and their corresponding charging characteristics will be regarded as key charging characteristics. These characteristics will be of great significance for predicting the charging state. By selecting key charging characteristics, it is possible to effectively focus on the characteristics that have a significant impact on the charging state, thereby reducing the complexity of model training and improving the accuracy and efficiency of the model. At the same time, the selection of these characteristics provides an important basis for subsequent charging state prediction and device optimization.

[0058] In this step, first, according to the aforementioned calculation results, a suitable weight threshold needs to be defined. This can be set based on historical data analysis or the performance feedback of a machine learning model. Next, traverse the overall weight data of all connected components and check whether each weight value is higher than this threshold. If so, add its corresponding charging characteristics to the set of key charging characteristics.

[0059] The characteristics corresponding to these high-weight connected components can be stored by establishing a simple set or list. For subsequent use, these characteristics and their related connection information (such as connected component identifiers, weights, etc.) can also be stored together to form a dictionary for subsequent model construction and charging state evaluation.

[0060] Once these key charging characteristics are screened out, they will be used as input data and fed into the subsequent deep learning model. This step is crucial because features with high weights usually have a significant impact on the prediction of the charging state. By optimizing the selection of features, the performance and accuracy of the model can be improved, thereby more effectively monitoring and predicting the charging state of the device.

[0061] S204, according to the charging stability index and the key charging characteristics, use a pre-trained charging state recognition model based on a deep learning network to predict the charging states of different devices.

[0062] According to the charging stability index and the key charging characteristics, using a pre-trained charging state recognition model based on a deep learning network can effectively identify the actual charging states of different devices. This process combines charging stability and feature influence degree, ensuring that the model can more accurately reflect the state of the device during the charging process. By regarding the charging stability index as a direct reflection of device safety and charging efficiency, and the profound insights provided by the key charging characteristics, the model can not only analyze the current charging situation, but also make accurate predictions for the charging states in future similar scenarios through learning historical data.

[0063] The importance of this process lies in that it enhances the monitoring and management capabilities of the charging process, helps users understand the charging state of the device in real time, thereby improving charging efficiency and extending battery life. At the same time, the deep learning-based model can self-update, and with the addition of more data, continuously improve the recognition accuracy, providing users with a dynamic and intelligent charging management solution, reducing the limitations of traditional charging monitoring methods, and enhancing the intelligent level of the device.

[0064] Specifically, the charging stability index can be input into a pre-trained first charging state recognition model based on a deep learning network to output the most likely first charging state of the corresponding device and its first probability value; The core of this step is to use the calculated charging stability index as input and pass it to the deep learning model. This model has been trained with a large amount of data and has learned how to map different degrees of charging stability to specific charging states. After the input, the model will evaluate the charging state of the device according to the learned features and patterns, and generate a corresponding first probability value, indicating the likelihood of this state occurring. By directly inputting the charging stability index into the model, the preliminary charging state of the device can be obtained quickly and accurately. This step emphasizes the importance of the charging stability index in charging state evaluation, ensuring that the model can use real-time data for dynamic prediction, and thus providing more timely and practical charging state information for users.

[0065] First, the charging stability index needs to be standardized to ensure it is within the input range expected by the deep learning model. Standardization can be achieved through the calculation of the mean and standard deviation, or by using Min - Max normalization to scale the index to the range of [0, 1]. Standardization not only improves the computational efficiency of the model but also enhances its adaptability to data of different scales, avoiding problems such as vanishing or exploding gradients caused by overly large or small numerical values.

[0066] Next, the standardized charging stability index is passed as an input feature to the pre - trained deep learning network. During this process, it is necessary to ensure that the input data format meets the requirements of the model. Use an appropriate input layer design to ensure that the model can correctly receive the charging stability index and process it effectively. The network calculates the activation values of the hidden layer through forward propagation and uses corresponding activation functions (such as ReLU or sigmoid) to non - linearly transform the output to capture the complex relationship between the charging state and the stability index.

[0067] Finally, the model generates the predicted charging state and its corresponding probability values through the output layer. This output usually passes through a softmax layer to ensure that the sum of the probabilities of all states is 1. The output results include the probability values corresponding to each possible charging state and the most likely first charging state. At this time, the first charging state and its probability value will be saved for use in subsequent steps, thus constituting a preliminary assessment of the device's charging state.

[0068] Perform a multiplication operation on the first probability value and the first weight corresponding to the charging stability index to obtain a first weighted probability value; This step uses the obtained first probability value and its corresponding weight to perform a multiplication operation to generate a first weighted probability value. The purpose of this weighted calculation is to adjust the contribution degree of the probability value through the weight, ensuring that the charging state more relevant to charging stability has a higher influence, thereby enhancing the credibility of the prediction result.

[0069] Through the weighted processing, the complexity and diversity of the charging state are actually reflected, emphasizing the influence of different factors in the prediction of the charging state. This weighted method can improve the accuracy of the model, making the final decision better reflect the real situation, reducing the risk of misjudgment caused by a single prediction, and providing a more reliable basis for decision - making.

[0070] At this stage, first, it is necessary to determine the weight values related to the charging stability index. These weights are generally set during the pre - training of the model to map the influence of charging stability on state prediction. The weight values can be obtained based on historical data analysis or set through expert experience to ensure that they reflect the importance of the stability index in different charging environments.

[0071] Next, multiply the first probability value and the corresponding weights item by item to calculate the weighted probability value. This process emphasizes the weight distribution of charging stability in state prediction, aiming to highlight the contribution of more important features to the final state. The product result will generate a weighted probability, which helps with subsequent analysis and provides a basis for decision-making.

[0072] Finally, store the obtained first weighted probability value in the system, usually in the form of a variable or data structure. This weighted value will not only serve as a benchmark for subsequent comparisons but also provide data support for the final decision. An effective storage method can ensure the efficient execution of subsequent steps and the accurate output of results, laying a solid foundation for the entire prediction process of the system.

[0073] Input the key charging features into a pre-trained second charging state recognition model based on a deep learning network to output the most likely second charging state of the corresponding device and its second probability value; In this step, the extracted key charging features are used in another deep learning model to further analyze the charging state of the device. This model has learned how to integrate multiple charging features during training to form a comprehensive assessment of the charging state. What is output is the second charging state related to the charging features and its probability value. This level of analysis can supplement and enrich the understanding of the charging state. Especially in complex charging environments, a single feature may not fully reflect the actual state of the device. By combining multiple key features, the model improves the accuracy of state judgment and also provides more comprehensive prediction information.

[0074] First, perform preprocessing on the key charging features to ensure that the data format meets the input requirements of the second deep learning model. This step may include standardization or feature scaling to adapt the data to the training and inference processes of the network. Reasonable data preprocessing not only improves the training efficiency of the model but also enhances the model's robustness to outliers or noisy data.

[0075] Next, input the processed key charging features into the second deep learning model to perform forward propagation calculations. The multiple layers of neurons inside the model will process the input features, calculate the hidden states through weighted sums and biases of each layer, and apply a non-linear activation function to generate the output. Different from the first model, the second model will focus on using the relationships between multiple key features to judge the charging state and form a more comprehensive assessment result.

[0076] Finally, the model outputs the second charging state corresponding to the input features and its probability value. The output usually undergoes softmax processing to standardize the probabilities of possible charging states into a comparable form. The obtained output result will be recorded and passed to the next step for comparison and comprehensive analysis with the results of the previous step to ensure the continuity and logic of the prediction process.

[0077] Multiplying the second probability value by a second weight corresponding to the key charging feature to obtain a second weighted probability value; In this step, the second probability value is multiplied by its corresponding weight to obtain a new weighted probability value. This process is intended to adjust the probability value based on the criticality and relevance of the charging characteristics, thereby ensuring that the actual contribution of different characteristics to the charging status can be reflected in the final decision. By weighting the probability values, the accuracy of the charging status recognition model is further improved, ensuring that important charging characteristics occupy an appropriate proportion in the final judgment. This method allows the model to flexibly respond to state changes under different conditions when facing a complex charging environment, thereby improving the intelligence of charging management.

[0078] In this step, the weight values associated with the second probability value need to be extracted from the model or data source. The weight values are usually set during the model training phase and are derived by analyzing the influence of historical charging data on different features. This process helps ensure that the prediction of the charging state is well supported statistically and empirically.

[0079] Then, the second probability value and the corresponding weight are multiplied item by item to calculate the second weighted probability value. This process will reflect the role of different charging characteristics in state prediction, ensuring that important characteristics can obtain higher weights in the prediction, thereby affecting the accuracy of the final decision.

[0080] Finally, the calculated second weighted probability value is stored in an appropriate data structure for easy comparison with the first weighted probability value. This weighted value paves the way for the next steps, ensuring that the complexity and influencing factors of the state of charge can be fully reflected in the results, ensuring the comprehensiveness of the prediction process.

[0081] The charging state corresponding to the maximum weighted probability value between the first weighted probability value and the second weighted probability value is determined as the final charging state of the corresponding device.

[0082] In the last step, the first weighted probability value is compared with the second weighted probability value, and the charging state corresponding to the maximum value is selected as the final prediction result of the device. This process ensures that the state analysis from multiple angles can be effectively integrated to obtain the most likely charging state. This step emphasizes the model's ability to comprehensively evaluate the charging state, which can effectively integrate the outputs from different models to ensure the reliability of the final prediction. By selecting the maximum weighted probability value, the system can propose the charging state with the most reference value, thereby achieving more intelligent decision support in battery management and monitoring.

[0083] At this stage, first obtain the calculated first weighted probability value and second weighted probability value from the previous steps. These two values will form the basis for the final assessment of the device's charging status. To ensure the accuracy of the assessment, it is necessary to retain the corresponding status information of both for subsequent comparison.

[0084] Then, by comparing the first and second weighted probability values, select the maximum value among them and its corresponding charging status. This process ensures the selection of the most representative status among the multi-model prediction results, reflecting the true situation of the device under the current charging conditions. This comparison can be achieved through simple conditional judgments to ensure the efficiency and accuracy of the decision-making process.

[0085] Finally, output the determined final charging status to the user interface or the relevant control system to ensure that the user can obtain the charging status information of the device in a timely manner. The output of this step can not only provide real-time information feedback to the user, but also provide decision-making support for the subsequent operation and management of the device, enhancing the intelligence level and response ability of the overall charging management system.

[0086] It can be seen that collecting a device charging data set related to the charging status; calculating a charging stability index for the charging stability degree of different devices according to the device charging data set; extracting key charging features in the device charging data set whose influence degree on the charging status is higher than a preset threshold; and predicting the charging status of different devices according to the charging stability index and the key charging features by using a pre-trained charging status recognition model based on a deep learning network, so as to be able to efficiently predict the charging status by using deep learning technology, implement an intelligent and precise charging monitoring scheme, and thus improve the safety, effectiveness and user experience of charging.

[0087] Another embodiment of the present invention provides a system for monitoring the charging status of different devices. Refer to Figure 3 , the system may include: A collection module 301 for collecting a device charging data set related to the charging status; A calculation module 302 for calculating a charging stability index for the charging stability degree of different devices according to the device charging data set; An extraction module 303 for extracting key charging features in the device charging data set whose influence degree on the charging status is higher than a preset threshold; A prediction module 304 for predicting the charging status of different devices according to the charging stability index and the key charging features by using a pre-trained charging status recognition model based on a deep learning network.

[0088] It can be seen that a device charging dataset related to the charging state is collected; according to the device charging dataset, a charging stability index for calculating the charging stability degree of different devices is calculated; key charging features in the device charging dataset whose influence degree on the charging state is higher than a preset threshold are extracted; according to the charging stability index and the key charging features, a charging state recognition model based on a deep learning network that is pre-trained is used to predict the charging states of different devices, so that the deep learning technology can be used for efficient prediction of the charging state, so as to implement an intelligent and precise charging monitoring solution, thereby improving the safety, effectiveness and user experience of charging.

[0089] An embodiment of the present invention further provides a storage medium, in which a computer program is stored, and wherein the computer program is configured to execute the steps in any one of the above method embodiments when running.

[0090] Specifically, in this embodiment, the above storage medium may be configured to store a computer program for executing the following steps: S201, collect a device charging dataset related to the charging state; S202, calculate a charging stability index for calculating the charging stability degree of different devices according to the device charging dataset; S203, extract key charging features in the device charging dataset whose influence degree on the charging state is higher than a preset threshold; S204, according to the charging stability index and the key charging features, use a pre-trained charging state recognition model based on a deep learning network to predict the charging states of different devices.

[0091] It can be seen that a device charging dataset related to the charging state is collected; according to the device charging dataset, a charging stability index for calculating the charging stability degree of different devices is calculated; key charging features in the device charging dataset whose influence degree on the charging state is higher than a preset threshold are extracted; according to the charging stability index and the key charging features, a charging state recognition model based on a deep learning network that is pre-trained is used to predict the charging states of different devices, so that the deep learning technology can be used for efficient prediction of the charging state, so as to implement an intelligent and precise charging monitoring solution, thereby improving the safety, effectiveness and user experience of charging.

[0092] An embodiment of the present invention further provides an electronic device, including a memory and a processor, a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0093] Specifically, the above-mentioned electronic device may further include a transmission device and an input / output device. Among them, the transmission device is connected to the above-mentioned processor, and the input / output device is connected to the above-mentioned processor.

[0094] Specifically, in this embodiment, the above-mentioned processor may be set to execute the following steps through a computer program: S201, collect a device charging data set related to the charging state; S202, calculate a charging stability index for the charging stability of different devices according to the device charging data set; S203, extract key charging features in the device charging data set whose influence on the charging state is higher than a preset threshold; S204, predict the charging states of different devices according to the charging stability index and the key charging features, using a pre-trained charging state recognition model based on a deep learning network.

[0095] It can be seen that collecting a device charging data set related to the charging state; calculating a charging stability index for the charging stability of different devices according to the device charging data set; extracting key charging features in the device charging data set whose influence on the charging state is higher than a preset threshold; predicting the charging states of different devices according to the charging stability index and the key charging features, using a pre-trained charging state recognition model based on a deep learning network, so as to be able to efficiently predict the charging state using deep learning technology, to implement an intelligent and precise charging monitoring solution, thereby improving the safety, effectiveness and user experience of charging.

[0096] The above has detailed the structure, features and effects of the present invention according to the illustrated embodiments. The above is only the preferred embodiment of the present invention, but the present invention is not limited to the scope defined by the drawings. Any changes made according to the concept of the present invention, or equivalent embodiments modified into equivalent changes, still within the spirit covered by the description and drawings, should be within the protection scope of the present invention.

Claims

1. A method for monitoring the charging status of different devices, characterized in that, The method includes: Collecting a device charging data set related to the charging state; Calculating a charging stability index for the charging stability degree of different devices according to the device charging data set; Extracting key charging features in the device charging data set whose influence degree on the charging state is higher than a preset threshold; Predicting the charging states of different devices according to the charging stability index and the key charging features by using a pre-trained charging state recognition model based on a deep learning network.

2. The method according to claim 1, characterized in that, The calculation formula of the charging stability index is: ； Among them, the is the maximum charging current during the charging process, the is the minimum charging current during the charging process, the is the i-th measured voltage during the charging process, the is the average value of all measured voltages, the is the maximum ambient temperature during the charging process, the is the minimum ambient temperature during the charging process, the is the average ambient temperature during the charging process, the is a minimum value to avoid division by zero, the , the , the are the influence weights of the charging current, charging voltage and ambient temperature on the charging stability, and N is the number of measurements during the charging process.

3. The method according to claim 2, wherein The extracting of the key charging features in the device charging data set whose influence degree on the charging state is higher than a preset threshold includes: Regarding all charging features in the charging data set as nodes in a charging feature graph to be constructed, and each node represents a charging feature; Creating edges according to the correlation between charging features. If the correlation between two charging features exceeds a preset correlation threshold, an edge is established between these two charging features to form a charging feature graph, and the weight of each edge is set to the correlation value between the corresponding two charging features; Identifying connected components in the charging feature graph through a graph theory algorithm, and the connected components can identify groups of mutually dependent charging features; For each connected component, calculating the overall weight of the connected component, where the overall weight is the sum or average of the edge weights of the charging features that make up the connected component. A connected component with a high overall weight indicates that the relationship between its constituent features and the charging state is closer; Determining the charging features corresponding to the connected components whose overall weight is higher than a preset weight threshold as the key charging features.

4. The method according to claim 3, characterized in that, The predicting of the charging states of different devices according to the charging stability index and the key charging features by using a pre-trained charging state recognition model based on a deep learning network includes: Inputting the charging stability index into a pre-trained first charging state recognition model based on a deep learning network to output the most likely first charging state of the corresponding device and its first probability value; Performing a multiplication operation on the first probability value and the first weight corresponding to the charging stability index to obtain a first weighted probability value; Inputting the key charging features into a pre-trained second charging state recognition model based on a deep learning network to output the most likely second charging state of the corresponding device and its second probability value; Performing a multiplication operation on the second probability value and the second weight corresponding to the key charging features to obtain a second weighted probability value; Determining the charging state corresponding to the maximum weighted probability value among the first weighted probability value and the second weighted probability value as the final charging state of the corresponding device.

5. A system for monitoring the charging status of different devices, characterized in that, The system includes: A collecting module for collecting a device charging data set related to the charging state; A calculating module for calculating a charging stability index for the charging stability degree of different devices according to the device charging data set; An extracting module for extracting key charging features in the device charging data set whose influence degree on the charging state is higher than a preset threshold; A predicting module for predicting the charging states of different devices according to the charging stability index and the key charging features by using a pre-trained charging state recognition model based on a deep learning network.

6. The system according to claim 5, characterized in that, The calculation formula of the charging stability index is: ； Among them, the is the maximum charging current during the charging process, the is the minimum charging current during the charging process, the is the i-th measured voltage during the charging process, the is the average value of all measured voltages, the is the maximum ambient temperature during the charging process, the is the minimum ambient temperature during the charging process, the is the average ambient temperature during the charging process, the is a minimum value to avoid division by zero, the , the , the are the influence weights of the charging current, charging voltage, and ambient temperature on the charging stability, and N is the number of measurements during the charging process.

7. The system according to claim 6, characterized in that, The extraction module is specifically used for: All charging features in the charging data set are regarded as nodes in the charging feature graph to be constructed, and each node represents a charging feature; Creating edges based on the correlation between charging features. If the correlation between two charging features exceeds a preset correlation threshold, an edge is established between the two charging features to form a charging feature graph. The weight of each edge is set to the corresponding correlation value between the two charging features. identifying connected components in the charging feature graph by a graph theory algorithm, wherein the connected components can identify interdependent charging feature groups; For each connected component, the overall weight of the connected component is calculated, where the overall weight is the sum or average of the edge weights of the charging features in the connected component. A connected component with a high overall weight indicates that its constituent features are consistent with the connected component. The impact of the charging state is more relevant; The charging feature corresponding to the connected component whose overall weight is higher than a preset weight threshold is determined as the key charging feature.

8. The system according to claim 7, wherein The prediction module is specifically used for: Inputting the charging stability index into a pre-trained first charging state recognition model based on a deep learning network, and outputting the most likely first charging state of the corresponding device and its first probability value; Performing a product operation on the first probability value and a first weight corresponding to the charging stability index to obtain a first weighted probability value; Input the key charging feature into a pre-trained second charging state recognition model based on a deep learning network, and output the most likely second charging state of the corresponding device and its second probability value; Multiplying the second probability value by a second weight corresponding to the key charging feature to obtain a second weighted probability value; The charging state corresponding to the maximum weighted probability value between the first weighted probability value and the second weighted probability value is determined as the final charging state of the corresponding device.

9. A storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 4 when executed.

10. An electronic device, comprising a memory and a processor, characterized in that, A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 4.

Images