Charging pile control method based on multi-mode perception and edge calculation

Abstract

The invention relates to the technical field of battery charging of electric vehicles, and provides a charging pile control method based on multi-modal sensing and edge calculation, and the method comprises the steps: obtaining multi-modal sensing information corresponding to a target charging pile, the multi-modal sensing information comprising environment information, charging pile working information and to-be-charged vehicle information; performing edge calculation on the multi-modal sensing information through an edge calculation unit pre-configured around the target charging pile to obtain an edge calculation result; and the edge calculation result is transmitted to a cloud server, so that the cloud service unit adjusts the power distribution and the charging priority of the target charging pile according to the edge calculation result to obtain a target charging plan, and the target charging plan is fed back to the edge calculation unit to control the target charging pile to charge the to-be-charged vehicle. According to the scheme of the invention, the charging efficiency of the charging pile is improved, and the power load fluctuation is reduced.

Description

Technical Field

[0001] The present application relates to the technical field of battery charging of electric vehicles, and in particular to a charging pile control method based on multimodal sensing and edge computing. Background Art

[0002] As the number of electric vehicles grows, the demand for charging stations increases.

[0003] Existing charging pile management systems usually rely on a single data source for charging scheduling, such as vehicle battery status or user reservation information. However, this system lacks comprehensive consideration of environmental factors and cannot effectively respond to emergencies, such as power supply fluctuations in the area where the charging pile is located, changes in traffic conditions near the charging pile, etc. In addition, centralized data processing methods lead to high network latency and data processing costs. The existing charging pile management method will result in low charging efficiency and large fluctuations in power load.

[0004] Therefore, how to improve the charging efficiency of charging piles and reduce power load fluctuations is a problem that needs to be solved urgently. Summary of the invention

[0005] In view of the above problems existing in the prior art, the purpose of this application is to propose a charging pile control method based on multimodal perception and edge computing, so as to at least solve the technical problem of how to improve the charging efficiency of the charging pile and reduce the fluctuation of power load.

[0006] To achieve the above objectives and other related objectives, the present application provides a charging pile control method based on multimodal perception and edge computing, the method comprising:

[0007] Acquire multimodal sensing information corresponding to the target charging pile, wherein the multimodal sensing information includes environmental information, charging pile working information, and information of the vehicle to be charged;

[0008] By using an edge computing unit pre-configured around the target charging pile, edge computing is performed on the multimodal sensing information to obtain an edge computing result;

[0009] The edge computing result is input to the cloud server, so that the cloud service unit adjusts the power allocation and charging priority of the target charging pile according to the edge computing result, obtains a target charging plan, and feeds back the target charging plan to the edge computing unit to control the target charging pile to charge the vehicle to be charged.

[0010] In some embodiments, the environmental information includes human traffic density, traffic flow, and weather conditions;

[0011] The charging pile working information includes current load, working mode, and fault condition;

[0012] The information of the vehicle to be charged includes the vehicle battery status, the type of the vehicle to be charged and the user reservation information of the vehicle to be charged.

[0013] In some implementations, the acquiring of multimodal perception information corresponding to the target charging pile includes:

[0014] By using cameras installed around the target charging pile, a first video stream is acquired, pedestrians in the first video stream are detected and tracked, and the pedestrian flow density is obtained;

[0015] Acquire a second video stream by using a camera installed on a road around the target charging pile, detect and track vehicles in the second video stream, and obtain the traffic flow;

[0016] The weather conditions are obtained by installing meteorological sensors around the target charging pile and calling a third-party meteorological interface;

[0017] Obtaining a real-time voltage and a real-time current through a voltage sensor and a current sensor installed on a power supply line of the target charging pile, and obtaining the current load according to the real-time voltage and the real-time current;

[0018] Acquiring the working mode through a state detection sensor installed inside the target charging pile;

[0019] By installing fault detection sensors on the power supply unit, communication unit and charging interface of the target charging pile, the working status of the power supply unit, communication unit and charging interface is obtained, and the fault condition is determined based on the working status of the power supply unit, communication unit and charging interface.

[0020] In some implementations, performing edge computing on the multimodal perception information to obtain an edge computing result includes:

[0021] The environmental information is standardized, the charging pile working information is one-hot encoded, and the information of the vehicle to be charged is normalized to obtain the pre-processed multimodal perception information;

[0022] Extracting features corresponding to each modality in the preprocessed multimodal perception information through a pre-built neural network, and fusing the features corresponding to each modality to obtain a fused feature;

[0023] The fusion features are analyzed to obtain edge computing results.

[0024] In some implementations, extracting features corresponding to each modality in the preprocessed multimodal perception information and fusing the features corresponding to each modality to obtain a fused feature includes:

[0025] Extracting environmental features corresponding to the environmental information in the preprocessed multimodal perception information through a convolutional neural network;

[0026] Extracting working features corresponding to the working information of the charging pile in the preprocessed multimodal perception information through a fully connected neural network;

[0027] Extracting, through the embedding layer, features of the vehicle to be charged corresponding to the information of the vehicle to be charged in the preprocessed multimodal sensing information;

[0028] Linearly changing the environmental characteristics, the working characteristics, and the characteristics of the vehicle to be charged respectively to obtain a query and a key;

[0029] Calculate an attention score according to the query and the key, and normalize the attention score to obtain attention weights corresponding to the environmental features, the working features, and the features of the vehicle to be charged;

[0030] According to the attention weights corresponding to the environmental features, the working features, and the features of the vehicle to be charged, weighted processing is performed on the environmental features, the working features, and the features of the vehicle to be charged to obtain the fusion features.

[0031] In some implementations, analyzing the fusion features to obtain edge computing results includes:

[0032] The object to be analyzed in the fusion feature is taken as a node, the object to be analyzed includes the target charging pile and the vehicle to be charged, and the feature vector of the node is the feature vector in the fusion feature;

[0033] Determining edges between a plurality of the nodes according to the relationship structure of the object to be analyzed;

[0034] Initializing the feature vector of each node through the fusion feature to obtain the initialized feature vector of each node;

[0035] For each node, obtain information from adjacent nodes, and update the initialized feature vector according to the information to obtain a target feature vector of each node;

[0036] The target feature vector of each node is transmitted to a pre-trained regression model to obtain the edge computing result.

[0037] In some implementations, adjusting the power allocation and charging priority of the target charging pile according to the edge computing result to obtain a target charging plan includes:

[0038] Determine the state of the environment and executable actions according to the edge computing result, wherein the state of the environment includes the power state of the target charging pile and the state of the vehicle to be charged, and the executable actions include adjusting the power allocation of the charging pile and setting the charging priority;

[0039] Evaluate the effect of the executable action according to a preset reward function, wherein the preset reward function includes a charging efficiency reward function, a user satisfaction reward function, and a cost-effectiveness reward function;

[0040] Determine an expected reward corresponding to taking the executable action under the state of the environment, and determine a target action according to the effect of the executable action and the expected reward;

[0041] Based on the target action, the target charging plan is obtained.

[0042] In one embodiment of the present application, a charging pile control device based on multimodal perception and edge computing is also provided, and the device includes:

[0043] A multimodal sensing information acquisition module, used to acquire multimodal sensing information corresponding to a target charging pile, wherein the multimodal sensing information includes environmental information, charging pile working information, and information of a vehicle to be charged;

[0044] An edge computing module, configured to perform edge computing on the multimodal sensing information through edge computing units pre-configured around the target charging pile to obtain edge computing results;

[0045] The charging plan module is used to input the edge computing result to the cloud server so that the cloud service unit adjusts the power allocation and charging priority of the target charging pile according to the edge computing result, obtains the target charging plan, and feeds back the target charging plan to the edge computing unit to control the target charging pile to charge the vehicle to be charged.

[0046] In one embodiment of the present application, a computer-readable storage medium is also provided, wherein the computer-readable storage medium includes a stored computer program, wherein the computer program executes the above-mentioned charging pile control method based on multimodal perception and edge computing when running.

[0047] In one embodiment of the present application, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the above-mentioned charging pile control method based on multimodal perception and edge computing through the computer program.

[0048] Beneficial effects of the present invention:

[0049] First, multimodal perception information corresponding to a target charging pile is obtained, wherein the multimodal perception information includes environmental information, charging pile working information, and information of a vehicle to be charged; then, edge computing is performed on the multimodal perception information through an edge computing unit pre-configured around the target charging pile to obtain an edge computing result; finally, the edge computing result is input to a cloud server, so that the cloud service unit adjusts the power allocation and charging priority of the target charging pile according to the edge computing result, obtains a target charging plan, and feeds back the target charging plan to the edge computing unit to control the target charging pile to charge the vehicle to be charged. In this application, the edge computing unit can process multimodal perception information in real time close to the data source, reducing the delay in data transmission. The charging pile can make rapid adjustments based on the latest environmental and working status information, improving the response speed and accuracy of decision-making; through real-time data processing, the charging pile can dynamically adjust the charging strategy to avoid unnecessary waiting time and waste of resources, thereby improving charging efficiency; by analyzing environmental information, future electricity demand and environmental changes can be predicted, and the working mode of the charging pile can be adjusted in advance; by monitoring the working information of the charging pile, potential problems can be discovered and solved in a timely manner to ensure that the charging pile is always in the best working state; by analyzing the information of the vehicle to be charged, personalized charging services can be provided for different types of vehicles, and the charging plan can be optimized; the cloud server performs global optimization based on the edge computing results and generates a target charging plan. From a global perspective, the cloud server can balance the load of each charging pile, avoid local overload or idle conditions, and improve the overall charging efficiency. The cloud server can dynamically adjust the power distribution and charging priority of the charging pile based on real-time data to ensure efficient use of resources. In addition, the cloud server can perform load balancing based on global data to avoid situations where some charging piles are overloaded while other charging piles are idle. By reasonably allocating charging tasks, the power load can be smoothed and fluctuations can be reduced. By dynamically adjusting the power distribution and charging priority of the charging piles, it can flexibly respond to sudden power demands and avoid load surges in a short period of time. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0051] Figure 1 is a schematic diagram of an application environment of a charging pile control method based on multimodal perception and edge computing shown in an exemplary embodiment of the present application;

[0052] Figure 2 It is a flow chart of a charging pile control method based on multimodal perception and edge computing shown in an exemplary embodiment of the present application;

[0053] Figure 3 is a schematic diagram of a charging pile control device based on multimodal perception and edge computing shown in an exemplary embodiment of the present application;

[0054] Figure 4 It is a schematic diagram of the structure of an electronic device shown in an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0056] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0057] In one embodiment of the present application, a charging pile control method based on multimodal sensing and edge computing is provided. Optionally, as an optional implementation, the above-mentioned charging pile control method based on multimodal sensing and edge computing can be applied to, but not limited to, Figure 1 in the environment shown. Figure 1 is a schematic diagram of an application environment of a charging pile control method based on multimodal perception and edge computing shown in an exemplary embodiment of the present application, see Figure 1, the implementation environment includes a charging pile 101, a charging pile control terminal 102, and a cloud server 103. The charging pile 101, the charging pile control terminal 102, and the cloud server 103 can, but are not limited to, communicate through a network, and the cloud server 103 can, but are not limited to, perform operations on the database, for example, write data operations or read data operations. The above-mentioned charging pile control terminal 102 can, but is not limited to, include a human-computer interaction screen, a processor, and a memory. The above-mentioned human-computer interaction screen can, but is not limited to, be used to display the target charging plan. The above-mentioned processor can, but is not limited to, be used to respond to the above-mentioned human-computer interaction operation, perform corresponding operations, or generate corresponding instructions, and send the generated instructions to the cloud server 103. The above-mentioned memory is used to store relevant storage data, such as environmental information, charging pile working information, and information of vehicles to be charged. The above-mentioned charging pile control terminal 102 can also, but is not limited to, include an edge computing unit and a data processing terminal.

[0058] As an optional method, data may be collected by the charging pile control terminal 102 , for example, multimodal sensing information corresponding to the target charging pile may be collected and preprocessed.

[0059] As an optional method, the following steps in the charging pile control method based on multimodal perception and edge computing may be performed on the charging pile control terminal 102:

[0060] Acquire multimodal sensing information corresponding to the target charging pile, wherein the multimodal sensing information includes environmental information, charging pile working information, and information of the vehicle to be charged;

[0061] By using an edge computing unit pre-configured around the target charging pile, edge computing is performed on the multimodal sensing information to obtain an edge computing result;

[0062] The edge computing result is input to the cloud server, so that the cloud service unit adjusts the power allocation and charging priority of the target charging pile according to the edge computing result, obtains a target charging plan, and feeds back the target charging plan to the edge computing unit to control the target charging pile to charge the vehicle to be charged.

[0063] As an optional method, the cloud server 103 can adjust the power distribution and charging priority of the target charging pile according to the edge computing results, obtain the target charging plan, and feed back the target charging plan to the edge computing unit to control the target charging pile to charge the vehicle to be charged.

[0064] In the above manner, the edge computing unit can process multimodal perception information in real time near the data source, reducing the delay in data transmission. The charging pile can make rapid adjustments based on the latest environmental and working status information, improving the response speed and accuracy of decision-making. Through real-time data processing, the charging pile can dynamically adjust the charging strategy to avoid unnecessary waiting time and waste of resources, thereby improving charging efficiency. By analyzing environmental information, future electricity demand and environmental changes can be predicted, and the working mode of the charging pile can be adjusted in advance. By monitoring the working information of the charging pile, potential problems can be discovered and solved in a timely manner to ensure that the charging pile is always in the best working state. By analyzing the information of the vehicle to be charged, personalized charging services can be provided for different types of vehicles, and the charging plan can be optimized. The cloud server performs global optimization based on the edge computing results and generates a target charging plan. From a global perspective, the cloud server can balance the load of each charging pile, avoid local overload or idleness, and improve the overall charging efficiency. The cloud server can dynamically adjust the power distribution and charging priority of the charging pile based on real-time data to ensure efficient use of resources. In addition, the cloud server can perform load balancing based on global data to avoid situations where some charging piles are overloaded while other charging piles are idle. By reasonably allocating charging tasks, the power load can be smoothed and fluctuations can be reduced. By dynamically adjusting the power distribution and charging priority of the charging piles, it can flexibly respond to sudden power demands and avoid load surges in a short period of time.

[0065] Optionally, in this embodiment, the network may include but is not limited to: a wireless network, wherein the wireless network includes: Bluetooth, WIFI and other networks that implement wireless communication. The cloud server may be a single server or a server cluster consisting of multiple servers. The above is only an example and is not limited in this embodiment.

[0066] As an optional example, this embodiment does not limit the execution subject of the above-mentioned charging pile control method based on multimodal sensing and edge computing. Some or all steps of the above-mentioned charging pile control method based on multimodal sensing and edge computing can be executed on the charging pile control terminal 102.

[0067] In one embodiment of the present application, a charging pile control method based on multimodal perception and edge computing is provided. Figure 2 is a flow chart of a charging pile control method based on multimodal perception and edge computing shown in an exemplary embodiment of the present application, see Figure 2 The charging pile control method based on multimodal perception and edge computing includes the following steps S210 to S230:

[0068] In step S210, multimodal sensing information corresponding to the target charging pile is obtained, where the multimodal sensing information includes environmental information, charging pile working information, and information of the vehicle to be charged.

[0069] Among them, multimodal perception information refers to information obtained from multiple different types of data sources, which together describe the various environments and states around the target charging pile. The purpose of multimodal perception information is to provide comprehensive and real-time data support for more accurate analysis and decision-making.

[0070] Among them, environmental information refers to data that describes the environmental conditions around the charging pile, including but not limited to the following: crowd density, using cameras or other sensors to monitor the crowd flow near the charging pile in real time to understand the crowd activities; traffic flow, using cameras or other sensors to monitor the traffic flow near the charging pile in real time to understand the vehicle traffic conditions; weather conditions, using meteorological sensors to monitor the temperature, humidity, wind speed and other meteorological parameters near the charging pile in real time.

[0071] Among them, the charging pile working information refers to the data that describes the working status of the charging pile itself, including but not limited to the following: current load, real-time monitoring of the current load of the charging pile through current and voltage sensors to understand the power output of the charging pile; working mode, real-time monitoring of the working mode of the charging pile (such as standby, charging, fault, etc.) through status monitoring sensors; fault condition, real-time monitoring of the status of key components of the charging pile through fault detection sensors to understand whether there is a fault; historical data, recording the historical working data of the charging pile, including past loads, working modes and fault records.

[0072] Among them, the information of the vehicle to be charged refers to the relevant data describing the vehicle to be charged and its user, including but not limited to the following: vehicle battery status, real-time monitoring of the battery's SOC (State of Charge), SOH (State of Health) and temperature through the vehicle's battery management system (BMS); vehicle type, identifying the type of vehicle to be charged through a camera or other sensor; user appointment information, collecting the user's appointment information through the user application, including appointment time, vehicle type, charging requirements, etc.

[0073] In step S220, edge computing is performed on the multimodal perception information by an edge computing unit pre-configured around the target charging pile to obtain an edge computing result.

[0074] Among them, the edge computing unit refers to a computing device deployed near the data source (such as a charging pile) to process and analyze data collected from sensors and other data sources in real time. The main features and functions of the edge computing unit include: close deployment, the edge computing unit is usually deployed near the location where the data is generated, such as around the charging pile, to reduce the delay in data transmission; high-performance computing capability, the edge computing unit is equipped with sufficient computing resources to perform complex real-time data processing and analysis; low-power design, in order to adapt to outdoor or embedded environments, the edge computing unit usually has a low-power design to extend its service life; network connection, the edge computing unit has a network connection function and can transmit the processing results to a cloud server or other central node; security, the edge computing unit usually has security mechanisms, such as encrypted transmission, access control, etc., to protect the security and privacy of data.

[0075] Among them, edge computing refers to data processing and analysis on the edge devices where the data is generated, rather than transmitting all data to a central server or cloud server for processing.

[0076] In step S230, the edge computing result is input to the cloud server, so that the cloud service unit adjusts the power allocation and charging priority of the target charging pile according to the edge computing result, obtains the target charging plan, and feeds back the target charging plan to the edge computing unit to control the target charging pile to charge the vehicle to be charged.

[0077] Among them, power allocation refers to the reasonable allocation of the output power of the charging pile according to the current working status of the charging pile, environmental conditions and the needs of the vehicle to be charged. Reasonable power allocation can improve charging efficiency, optimize resource utilization and reduce energy waste. When allocating power, the output power of each charging pile can be dynamically adjusted according to real-time data to ensure efficient use of resources; the load of each charging pile can be balanced through a global optimization algorithm to avoid local overload or idleness; energy consumption can be minimized and energy efficiency can be improved while ensuring charging needs; when a charging pile failure is detected, the power of other charging piles can be automatically adjusted to ensure the continuity of the charging task.

[0078] Among them, charging priority refers to determining which vehicles should be charged first based on factors such as vehicle charging needs, user reservation information, and environmental conditions. Reasonable charging priority can improve user satisfaction and optimize the charging experience. Priority can be determined based on user reservation information and charging needs. For example, emergency charging requests can be processed first; priority can be determined based on the battery status and charging needs of the vehicle. For example, vehicles with low battery power can be charged first; charging priority can be adjusted based on environmental information (such as pedestrian density, traffic flow, and weather conditions). For example, more charging requests can be processed first during peak hours; future charging needs can be predicted based on the user's charging history data, and charging priority can be adjusted in advance.

[0079] By adopting the above-mentioned embodiments provided by the present application, the edge computing unit can process multimodal perception information in real time near the data source, reducing the delay of data transmission, and the charging pile can make adjustments quickly according to the latest environment and working status information, thereby improving the response speed and the accuracy of decision-making; through real-time data processing, the charging pile can dynamically adjust the charging strategy to avoid unnecessary waiting time and waste of resources, thereby improving the charging efficiency; by analyzing environmental information, it is possible to predict future electricity demand and environmental changes, and adjust the working mode of the charging pile in advance; by monitoring the working information of the charging pile, it is possible to timely discover and solve potential problems to ensure that the charging pile is always in the best working state; by analyzing the information of the vehicle to be charged, it is possible to provide personalized charging services for different types of vehicles and optimize the charging plan; the cloud server performs global optimization based on the edge computing results and generates a target charging plan. From a global perspective, the cloud server can balance the loads of each charging pile, avoid local overload or idleness, and improve the overall charging efficiency. The cloud server can dynamically adjust the power distribution and charging priority of the charging pile according to real-time data to ensure efficient use of resources. In addition, the cloud server can perform load balancing based on global data to avoid situations where some charging piles are overloaded while other charging piles are idle. By reasonably allocating charging tasks, the power load can be smoothed and fluctuations can be reduced. By dynamically adjusting the power distribution and charging priority of the charging piles, it can flexibly respond to sudden power demands and avoid load surges in a short period of time.

[0080] In one embodiment of the present application, the environmental information includes human traffic density, traffic flow, and weather conditions;

[0081] The charging pile working information includes current load, working mode, and fault condition;

[0082] The information of the vehicle to be charged includes the vehicle battery status, the type of the vehicle to be charged and the user reservation information of the vehicle to be charged.

[0083] In one embodiment of the present application, the step of obtaining multimodal perception information corresponding to the target charging pile includes:

[0084] By using cameras installed around the target charging pile, a first video stream is acquired, pedestrians in the first video stream are detected and tracked, and the pedestrian flow density is obtained;

[0085] Acquire a second video stream by using a camera installed on a road around the target charging pile, detect and track vehicles in the second video stream, and obtain the traffic flow;

[0086] The weather conditions are obtained by installing meteorological sensors around the target charging pile and calling a third-party meteorological interface;

[0087] Obtaining a real-time voltage and a real-time current through a voltage sensor and a current sensor installed on a power supply line of the target charging pile, and obtaining the current load according to the real-time voltage and the real-time current;

[0088] Acquiring the working mode through a state detection sensor installed inside the target charging pile;

[0089] By installing fault detection sensors on the power supply unit, communication unit and charging interface of the target charging pile, the working status of the power supply unit, communication unit and charging interface is obtained, and the fault condition is determined based on the working status of the power supply unit, communication unit and charging interface.

[0090] The first video stream of the area around the charging pile is obtained through the camera, and the pedestrian detection and tracking algorithm is used to identify and track pedestrians in the video. The second video stream is obtained through the camera installed on the road around the charging pile, and the vehicle detection and tracking algorithm is used to identify and track the vehicle in the video. The weather conditions near the charging pile, including temperature, humidity, wind speed, etc., are obtained by calling the third-party meteorological API (interface). It helps predict the use demand of the charging pile, optimize the charging plan, and ensure that the equipment works in a suitable environment.

[0091] Among them, the voltage and current of the charging pile are monitored in real time through the voltage sensor and current sensor installed on the power supply line of the charging pile, and the current load is calculated. The working mode of the charging pile (such as standby, charging, fault, etc.) is monitored in real time through the state detection sensor installed inside the charging pile; the state detection sensor (such as switch sensor, state register) is used to collect data, and the working mode is determined through logical judgment. The fault detection sensor installed on the charging pile power supply unit, communication unit and charging interface monitors the working status of these components in real time to determine whether there is a fault. Fault detection sensors (such as temperature sensors, current sensors, state registers) are used to collect data, and the fault condition is determined through logical judgment and threshold comparison.

[0092] In this embodiment, through real-time data processing and analysis, the charging pile can quickly make adjustments based on the latest environment and working status information, avoiding unnecessary waiting time and waste of resources, thereby improving charging efficiency. By dynamically adjusting the power distribution and charging priority of the charging pile, the load of each charging pile can be balanced, local overload or idleness can be avoided, and resource utilization can be optimized. Through global optimization and dynamic adjustment, the power load can be smoothed, the load surge in a short period of time can be reduced, the power load fluctuation can be reduced, and the stability of the power grid can be improved. Through personalized charging services and optimized charging plans, the charging needs of users can be met, and user satisfaction and charging experience can be improved. By real-time monitoring of charging pile faults and timely processing, the normal operation of the charging pile can be ensured and reliability can be enhanced. Through edge computing and secure communication protocols, the security and privacy of data can be protected and the risk of data leakage can be reduced.

[0093] In one embodiment of the present application, performing edge computing on the multimodal perception information to obtain an edge computing result includes:

[0094] The environmental information is standardized, the charging pile working information is one-hot encoded, and the information of the vehicle to be charged is normalized to obtain the pre-processed multimodal perception information;

[0095] Extracting features corresponding to each modality in the preprocessed multimodal perception information through a pre-built neural network, and fusing the features corresponding to each modality to obtain a fused feature;

[0096] The fusion features are analyzed to obtain edge computing results.

[0097] Among them, the numerical features in the environmental information (such as pedestrian density, traffic flow, temperature, humidity, wind speed, etc.) are converted into a standard normal distribution with a mean of 0 and a standard deviation of 1. Standardization can make features of different scales comparable and improve the effects of subsequent feature extraction and model training.

[0098] Among them, the category features (such as working mode, fault code, etc.) in the working information of the charging pile are converted into one-hot encoding. One-hot encoding can convert category features into numerical features, making them suitable for machine learning models.

[0099] Among them, a min-max scaler is used to convert the numerical features in the information of the vehicle to be charged into a range of 0 to 1. Normalization processing can make features of different scales comparable and improve the effect of subsequent feature extraction and model training.

[0100] Among them, a feature extraction network of multimodal perception information is constructed, and the features of each modal perception information are extracted using a fully connected layer or an attention mechanism, and these features are fused to generate a comprehensive feature representation. Feature extraction and fusion can capture the complex relationship between multimodal information, generate a comprehensive feature representation, and improve the prediction and decision-making capabilities of the model.

[0101] Among them, the fusion features are analyzed using regression models or classification models to generate edge computing results, such as the power allocation plan of the charging pile, the charging priority list, etc. The edge computing results can be directly used to control the behavior of the charging pile to improve real-time performance and response speed.

[0102] In this embodiment, the dimensional differences between different features are eliminated through standardization, unique hot encoding and normalization processing, which improves the efficiency and accuracy of data processing. The use of deep learning models to extract the features of multimodal perception information and perform feature fusion can capture the complex relationship between data, generate a comprehensive feature representation, and improve the prediction and decision-making capabilities of the model. The edge computing unit performs real-time data processing and analysis close to the data source, reducing the delay in data transmission and improving real-time performance and response speed.

[0103] In one embodiment of the present application, extracting features corresponding to each modality in the preprocessed multimodal perception information and fusing the features corresponding to each modality to obtain fused features includes:

[0104] Extracting environmental features corresponding to the environmental information in the preprocessed multimodal perception information through a convolutional neural network;

[0105] Extracting working features corresponding to the working information of the charging pile in the preprocessed multimodal perception information through a fully connected neural network;

[0106] Extracting, through the embedding layer, features of the vehicle to be charged corresponding to the information of the vehicle to be charged in the preprocessed multimodal sensing information;

[0107] Linearly changing the environmental characteristics, the working characteristics, and the characteristics of the vehicle to be charged respectively to obtain a query and a key;

[0108] Calculate an attention score according to the query and the key, and normalize the attention score to obtain attention weights corresponding to the environmental features, the working features, and the features of the vehicle to be charged;

[0109] According to the attention weights corresponding to the environmental features, the working features, and the features of the vehicle to be charged, weighted processing is performed on the environmental features, the working features, and the features of the vehicle to be charged to obtain the fusion features.

[0110] Among them, multiple layers of convolutional layers and pooling layers are used to gradually extract local and global features in environmental information, capturing complex patterns in environmental information, such as the spatial distribution of pedestrian flow density and the changing trend of traffic flow.

[0111] Among them, a fully connected neural network (FCN) is used to extract high-level features in the working information of charging piles, and to capture the complex relationships in the working information of charging piles, such as the changing trend of the current load and the conversion rules of the working mode.

[0112] Among them, the embedding layer is used to map discrete category features (such as vehicle type and user reservation information) into a low-dimensional space to generate a continuous feature vector. The discrete category features are converted into numerical features to make them suitable for subsequent feature extraction and fusion.

[0113] Among them, the feature vector is converted into query and key through linear transformation (such as fully connected layer) to provide input for the subsequent attention mechanism. Query and key are two important parts of the attention mechanism, which are used to calculate the attention score. The query vector and key vector represent different aspects of the feature vector respectively. The attention score is calculated by the inner product of the query and the key, reflecting the similarity between the features.

[0114] Among them, when calculating the attention score by the inner product of the query vector and the key vector, the dot product of the query vector and the key vector is calculated to generate an attention score matrix to reflect the similarity and importance between different features.

[0115] Among them, through normalization processing, the attention score is converted into attention weight, which indicates the importance of different features and is used for subsequent weighted processing.

[0116] Among them, the attention weight is multiplied by the corresponding feature vector, and then the sum is generated to generate a fusion feature, generating a comprehensive feature representation for subsequent analysis and decision-making.

[0117] In this embodiment, the environmental features, working features and features of the vehicle to be charged are extracted respectively through the convolutional neural network, the fully connected neural network and the embedding layer, which can capture the complex patterns and relationships in the multimodal perception information and generate high-level feature representations. The use of the attention mechanism to weight the features of different modalities can highlight important features, suppress unimportant features, and generate more comprehensive and effective fusion features. The fusion features can better reflect the overall characteristics of multimodal perception information and improve the accuracy and robustness of subsequent analysis and decision-making. The edge computing unit performs real-time data processing and analysis close to the data source, reducing the delay in data transmission and improving real-time performance and response speed.

[0118] In one embodiment of the present application, analyzing the fusion features to obtain edge computing results includes:

[0119] The object to be analyzed in the fusion feature is taken as a node, the object to be analyzed includes the target charging pile and the vehicle to be charged, and the feature vector of the node is the feature vector in the fusion feature;

[0120] Determining edges between a plurality of the nodes according to the relationship structure of the object to be analyzed;

[0121] Initializing the feature vector of each node through the fusion feature to obtain the initialized feature vector of each node;

[0122] For each node, obtain information from adjacent nodes, and update the initialized feature vector according to the information to obtain a target feature vector of each node;

[0123] The target feature vector of each node is transmitted to a pre-trained regression model to obtain the edge computing result.

[0124] The relationship structure between the objects to be analyzed is defined, that is, which objects are associated or interacted with each other. The relationship between the objects to be analyzed is represented by a graph structure, where nodes represent the objects to be analyzed (such as target charging piles and vehicles to be charged), and edges represent the relationship between nodes (such as the connection between charging piles and vehicles), helping the model understand the interactions between different objects and improving the accuracy of the analysis.

[0125] When obtaining feature information from adjacent nodes directly connected to the current node, the message passing mechanism in the graph neural network (GNN) is used to obtain feature information from adjacent nodes, capture the dependency between nodes, and enrich the feature representation of the current node.

[0126] Among them, using the aggregation and update mechanism in the graph neural network (GNN), the initialized feature vector of the current node is updated according to the information obtained from the adjacent nodes to generate the target feature vector, which can generate a richer and more accurate node feature representation and improve the prediction ability of the model.

[0127] Among them, the regression model is a machine learning model used to predict continuous values, such as linear regression, decision tree regression, neural network regression, etc. The deep learning framework is used to build and train the regression model, and the edge computing results such as the power allocation scheme and charging priority of the charging pile are predicted based on the target feature vector of the node.

[0128] Among them, the regression model is used to predict the node feature vector to generate the power allocation scheme of the charging pile, charging priority and other results. The specific edge computing results are generated to control the behavior of the charging pile and improve the real-time performance and response speed.

[0129] In this embodiment, by using the message passing mechanism of the graph neural network, information is obtained from adjacent nodes and the node feature vector is updated, which can generate a richer and more accurate feature representation and improve the prediction ability of the model. Using a graph structure to represent the relationship between the objects to be analyzed can better capture the interaction between different objects, generate a comprehensive feature representation, and improve the robustness of the model. By predicting the updated node feature vector through a regression model and generating specific edge computing results, the prediction accuracy and generalization ability of the model can be improved.

[0130] In one embodiment of the present application, adjusting the power allocation and charging priority of the target charging pile according to the edge computing result to obtain a target charging plan includes:

[0131] Determine the state of the environment and executable actions according to the edge computing result, wherein the state of the environment includes the power state of the target charging pile and the state of the vehicle to be charged, and the executable actions include adjusting the power allocation of the charging pile and setting the charging priority;

[0132] Evaluate the effect of the executable action according to a preset reward function, wherein the preset reward function includes a charging efficiency reward function, a user satisfaction reward function, and a cost-effectiveness reward function;

[0133] Determine an expected reward corresponding to taking the executable action under the state of the environment, and determine a target action according to the effect of the executable action and the expected reward;

[0134] Based on the target action, the target charging plan is obtained.

[0135] The state of the environment describes various parameters of the current system operation status, including the power state of the target charging pile and the state of the vehicle to be charged. The executable action is the operation that can be taken to adjust the working state of the charging pile.

[0136] Among them, the reward function is a function used to evaluate the effect of executable actions, which guides decision-making by quantifying different aspects of performance. Charging efficiency reward function: evaluates the amount of charging completed by the charging pile in unit time, the higher the better. User satisfaction reward function: evaluates the user's charging experience, such as waiting time, charging speed, etc., the higher the better. Cost-effectiveness reward function: evaluates the cost-effectiveness of the charging process, such as electricity costs, equipment maintenance costs, etc., the lower the better. According to the preset reward function, when evaluating the effect of the executable action, the value of each reward function can be calculated according to the current environmental state and the action taken to help select the optimal action to achieve the best overall effect.

[0137] Among them, expected return refers to the total reward expected to be obtained after taking a certain action under the current environmental state. Estimating the long-term benefits after taking a certain action through dynamic programming or reinforcement learning methods can help evaluate the long-term effects of different actions and select the best action. Consider the effect of the current action and the expected return to select the best action. Use the optimization algorithm to select the action with the highest expected return to ensure that the action taken can achieve the best overall effect in the current environment. Generate a specific charging plan based on the selected target action, convert the target action into a specific power allocation plan and charging priority list, guide the actual operation of the charging pile, and achieve optimized charging management.

[0138] For example, assuming that in a smart charging station, the edge computing unit adjusts the power allocation and charging priority of the charging pile according to the edge computing result through the following steps to generate a target charging plan:

[0139] a. Determine the state of the environment and the actions that can be performed:

[0140] The current power status of charging pile A: output power is 50kW, working mode is charging, and load is 80%;

[0141] Status of vehicle V1: battery SOC (State of Charge) is 30%, SOH (State of Health) is 90%, and the appointment time is 10:00;

[0142] Status of vehicle V2: battery SOC is 40%, SOH is 85%, and the appointment time is 10:30;

[0143] Status of vehicle V3: battery SOC is 50%, SOH is 80%, and the reservation time is 11:00;

[0144] Adjust the power distribution of charging pile A: distribute the power to V1, V2, and V3;

[0145] Set charging priority: determine the charging order of V1, V2, and V3;

[0146] b. Evaluate the effect of executable actions based on the reward function:

[0147] User satisfaction reward function:

[0148] Action 1: User satisfaction is 75%;

[0149] Action 2: User satisfaction is 85%;

[0150] Cost-Benefit Reward Function:

[0151] Action 1: cost-effectiveness is 70%;

[0152] Action 2: Cost-effectiveness is 80%;

[0153] Effect of action 1: Power is distributed to V1, V2, and V3, and the charging efficiency is 85%;

[0154] Effect of action 2: V1 is charged first, and the charging efficiency is 90%;

[0155] C. Determine expected returns:

[0156] Expected return of action 1: R1 = 0.85 × 0.75 × 0.70;

[0157] Expected return of action 2: R2 = 0.90 × 0.85 × 0.80;

[0158] d. Determine the target action:

[0159] Choose target action: Choose the action with the highest expected reward2.

[0160] e. Generate target charging plan:

[0161] The power distribution of charging pile A: V150kW, V220kW, V310kW.

[0162] Charging priority: V1>V2>V3.

[0163] In this embodiment, by evaluating the charging efficiency reward function, the optimal power allocation scheme is selected to improve the charging efficiency of the charging pile. By evaluating the user satisfaction reward function, the charging priority is optimized, the user's waiting time is reduced, and user satisfaction is improved. By evaluating the cost-effectiveness reward function, the most cost-effective charging scheme is selected to reduce the cost of the charging process. Taking into account the charging efficiency, user satisfaction and cost-effectiveness, the optimal charging plan is generated to optimize resource utilization and avoid local overload or idle situations. The edge computing unit performs real-time data processing and analysis close to the data source, reducing the delay in data transmission and improving real-time performance and response speed.

[0164] It can be seen from the above embodiments that the edge computing unit can process multimodal perception information in real time near the data source, reducing the delay of data transmission. The charging pile can make adjustments quickly according to the latest environment and working status information, improving the response speed and decision accuracy; through real-time data processing, the charging pile can dynamically adjust the charging strategy to avoid unnecessary waiting time and waste of resources, thereby improving the charging efficiency; by analyzing environmental information, it is possible to predict future electricity demand and environmental changes, and adjust the working mode of the charging pile in advance; by monitoring the working information of the charging pile, potential problems can be discovered and solved in time to ensure that the charging pile is always in the best working state; by analyzing the information of the vehicle to be charged, personalized charging services can be provided for different types of vehicles, and the charging plan can be optimized; the cloud server performs global optimization based on the edge computing results and generates a target charging plan. From a global perspective, the cloud server can balance the load of each charging pile, avoid local overload or idleness, and improve the overall charging efficiency. The cloud server can dynamically adjust the power distribution and charging priority of the charging pile according to real-time data to ensure efficient use of resources. In addition, the cloud server can perform load balancing based on global data to avoid situations where some charging piles are overloaded while other charging piles are idle. By reasonably allocating charging tasks, the power load can be smoothed and fluctuations can be reduced. By dynamically adjusting the power distribution and charging priority of the charging piles, it can flexibly respond to sudden power demands and avoid load surges in a short period of time.

[0165] In one embodiment of the present application, a charging pile control device based on multimodal perception and edge computing is also provided. Figure 3 is a schematic diagram of a charging pile control device based on multimodal perception and edge computing shown in an exemplary embodiment of the present application, see Figure 3 , the device comprises:

[0166] The multimodal sensing information acquisition module 301 is used to acquire the multimodal sensing information corresponding to the target charging pile, wherein the multimodal sensing information includes environmental information, charging pile working information, and information of the vehicle to be charged;

[0167] The edge computing module 302 is used to perform edge computing on the multimodal sensing information through edge computing units pre-configured around the target charging pile to obtain edge computing results;

[0168] The charging plan module 303 is used to input the edge computing results to the cloud server so that the cloud service unit adjusts the power allocation and charging priority of the target charging pile according to the edge computing results, obtains the target charging plan, and feeds back the target charging plan to the edge computing unit to control the target charging pile to charge the vehicle to be charged.

[0169] In the charging pile control device based on multimodal perception and edge computing in the embodiment of the present application, the edge computing unit can process multimodal perception information in real time near the data source, reducing the delay of data transmission. The charging pile can make adjustments quickly according to the latest environment and working status information, thereby improving the response speed and accuracy of decision-making; through real-time data processing, the charging pile can dynamically adjust the charging strategy to avoid unnecessary waiting time and waste of resources, thereby improving the charging efficiency; by analyzing environmental information, it is possible to predict future electricity demand and environmental changes, and adjust the working mode of the charging pile in advance; by monitoring the working information of the charging pile, it is possible to timely discover and solve potential problems to ensure that the charging pile is always in the best working state; by analyzing the information of the vehicle to be charged, it is possible to provide personalized charging services for different types of vehicles and optimize the charging plan; the cloud server performs global optimization based on the edge computing results to generate a target charging plan. From a global perspective, the cloud server can balance the loads of each charging pile, avoid local overload or idleness, and improve the overall charging efficiency. The cloud server can dynamically adjust the power distribution and charging priority of the charging pile according to real-time data to ensure efficient use of resources. In addition, the cloud server can perform load balancing based on global data to avoid situations where some charging piles are overloaded while other charging piles are idle. By reasonably allocating charging tasks, the power load can be smoothed and fluctuations can be reduced. By dynamically adjusting the power distribution and charging priority of the charging piles, it can flexibly respond to sudden power demands and avoid load surges in a short period of time.

[0170] For the specific embodiments of the charging pile control device based on multimodal sensing and edge computing in the present application, reference can be made to the examples shown in the above-mentioned charging pile control method based on multimodal sensing and edge computing, which will not be repeated here in this example.

[0171] In one embodiment of the present application, an electronic device for implementing the above-mentioned charging pile control method based on multimodal sensing and edge computing is also provided. The electronic device includes a memory and a processor, the memory stores a computer program, and the processor is configured to execute the above-mentioned charging pile control method based on multimodal sensing and edge computing through the computer program.

[0172] See also Figure 4 , Figure 4 404. It is a schematic diagram of the structure of an electronic device shown in an exemplary embodiment of the present application. Computer system 400 includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to the program stored in read-only memory (ROM) 402 or the program loaded from storage part 408 to random access memory (RAM) 403, such as executing the method described in the above embodiment. In RAM 403, various programs and data required for system operation are also stored. CPU 401, ROM 402 and RAM 403 are connected to each other via bus 404. Input / output (I / O) interface 405 is also connected to bus 404.

[0173] The following components are connected to the I / O interface 405: an input section 406 including a keyboard, a mouse, etc.; an output section 407 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the I / O interface 405 as needed. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as needed so that a computer program read therefrom is installed into the storage section 408 as needed.

[0174] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication section 409, and / or installed from a removable medium 411. When the computer program is executed by a central processing unit (CPU) 401, various functions defined in the system of the present application are executed.

[0175] It should be noted that the computer-readable medium shown in the embodiment of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, wherein a computer-readable computer program is carried. This propagated data signal can take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. A computer program contained on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0176] The flowchart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operations of the system, method and computer program product according to various embodiments of the present application. Wherein, each box in the flowchart or block diagram can represent a module, a program segment, or a part of the code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0177] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. The names of these units do not, in some cases, constitute limitations on the units themselves.

[0178] Another aspect of the present application further provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein the computer program executes the above-mentioned charging pile control method based on multimodal perception and edge computing when it is run. The computer-readable storage medium may be included in the electronic device described in the above embodiment, or may exist independently without being assembled into the electronic device.

[0179] Another aspect of the present application further provides a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the charging pile control method based on multimodal perception and edge computing provided in each of the above embodiments.

[0180] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A charging pile control method based on multimodal perception and edge computing, characterized in that: The method comprises: Acquire multimodal sensing information corresponding to the target charging pile, wherein the multimodal sensing information includes environmental information, charging pile working information, and information of the vehicle to be charged; By using an edge computing unit pre-configured around the target charging pile, edge computing is performed on the multimodal sensing information to obtain an edge computing result; The edge computing result is input to the cloud server, so that the cloud service unit adjusts the power allocation and charging priority of the target charging pile according to the edge computing result, obtains a target charging plan, and feeds back the target charging plan to the edge computing unit to control the target charging pile to charge the vehicle to be charged.

2. The charging pile control method based on multimodal perception and edge computing according to claim 1, characterized in that: The environmental information includes human traffic density, traffic flow, and weather conditions; The charging pile working information includes current load, working mode, and fault condition; The information of the vehicle to be charged includes the vehicle battery status, the type of the vehicle to be charged and the user reservation information of the vehicle to be charged.

3. The charging pile control method based on multimodal perception and edge computing as claimed in claim 2, characterized in that: The acquiring of multimodal sensing information corresponding to the target charging pile includes: By using cameras installed around the target charging pile, a first video stream is acquired, pedestrians in the first video stream are detected and tracked, and the pedestrian flow density is obtained; A second video stream is acquired by installing a camera on a road around the target charging pile, and vehicles in the second video stream are detected and tracked to obtain the traffic flow; The weather conditions are obtained by installing meteorological sensors around the target charging pile and calling a third-party meteorological interface; Obtaining a real-time voltage and a real-time current through a voltage sensor and a current sensor installed on a power supply line of the target charging pile, and obtaining the current load according to the real-time voltage and the real-time current; Acquiring the working mode through a state detection sensor installed inside the target charging pile; By installing fault detection sensors on the power supply unit, communication unit and charging interface of the target charging pile, the working status of the power supply unit, communication unit and charging interface is obtained, and the fault condition is determined based on the working status of the power supply unit, communication unit and charging interface.

4. The charging pile control method based on multimodal perception and edge computing according to claim 1, characterized in that: The performing edge computing on the multimodal perception information to obtain an edge computing result includes: The environmental information is standardized, the charging pile working information is one-hot encoded, and the information of the vehicle to be charged is normalized to obtain the pre-processed multimodal perception information; Extracting features corresponding to each modality in the preprocessed multimodal perception information through a pre-built neural network, and fusing the features corresponding to each modality to obtain a fused feature; The fusion features are analyzed to obtain edge computing results.

5. The charging pile control method based on multimodal perception and edge computing as claimed in claim 4, characterized in that: The extracting the features corresponding to each modality in the preprocessed multimodal perception information and fusing the features corresponding to each modality to obtain a fused feature includes: Extracting environmental features corresponding to the environmental information in the preprocessed multimodal perception information through a convolutional neural network; Extracting working features corresponding to the working information of the charging pile in the preprocessed multimodal perception information through a fully connected neural network; Extracting, through the embedding layer, features of the vehicle to be charged corresponding to the information of the vehicle to be charged in the preprocessed multimodal sensing information; Linearly changing the environmental characteristics, the working characteristics, and the characteristics of the vehicle to be charged respectively to obtain a query and a key; Calculate an attention score according to the query and the key, and normalize the attention score to obtain attention weights corresponding to the environmental features, the working features, and the features of the vehicle to be charged; According to the attention weights corresponding to the environmental features, the working features, and the features of the vehicle to be charged, weighted processing is performed on the environmental features, the working features, and the features of the vehicle to be charged to obtain the fusion features.

6. The charging pile control method based on multimodal perception and edge computing as claimed in claim 4, characterized in that: The analyzing the fusion features to obtain edge computing results includes: The object to be analyzed in the fusion feature is taken as a node, the object to be analyzed includes the target charging pile and the vehicle to be charged, and the feature vector of the node is the feature vector in the fusion feature; Determining edges between a plurality of the nodes according to the relationship structure of the object to be analyzed; Initializing the feature vector of each node through the fusion feature to obtain the initialized feature vector of each node; For each node, obtain information from adjacent nodes, and update the initialized feature vector according to the information to obtain a target feature vector of each node; The target feature vector of each node is transmitted to a pre-trained regression model to obtain the edge computing result.

7. The charging pile control method based on multimodal perception and edge computing according to claim 1, characterized in that: The step of adjusting the power allocation and charging priority of the target charging pile according to the edge computing result to obtain a target charging plan includes: Determine the state of the environment and executable actions according to the edge computing result, wherein the state of the environment includes the power state of the target charging pile and the state of the vehicle to be charged, and the executable actions include adjusting the power allocation of the charging pile and setting the charging priority; Evaluate the effect of the executable action according to a preset reward function, wherein the preset reward function includes a charging efficiency reward function, a user satisfaction reward function, and a cost-effectiveness reward function; Determine an expected reward corresponding to taking the executable action under the state of the environment, and determine a target action according to the effect of the executable action and the expected reward; Based on the target action, the target charging plan is obtained.

8. A charging pile control device based on multimodal perception and edge computing, characterized in that: The device comprises: A multimodal sensing information acquisition module, used to acquire multimodal sensing information corresponding to a target charging pile, wherein the multimodal sensing information includes environmental information, charging pile working information, and information of a vehicle to be charged; An edge computing module, configured to perform edge computing on the multimodal sensing information through edge computing units pre-configured around the target charging pile to obtain edge computing results; The charging plan module is used to input the edge computing result to the cloud server so that the cloud service unit adjusts the power allocation and charging priority of the target charging pile according to the edge computing result, obtains the target charging plan, and feeds back the target charging plan to the edge computing unit to control the target charging pile to charge the vehicle to be charged.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein the computer program executes the charging pile control method based on multimodal perception and edge computing described in any one of claims 1 to 7 when running.

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 execute the charging pile control method based on multimodal perception and edge computing according to any one of claims 1 to 7 through the computer program.

Images