Automatic protection method for charging electric vehicles

Abstract

The application discloses an electric vehicle automatic protection type charging method and relates to the technical field of electric vehicle charging, which comprises the following steps: license plate recognition and vehicle model matching are performed on a vehicle entering a charging area, face recognition technology is used to perform double authentication on the identity of a vehicle owner, and permission verification is performed in combination with background information; after the electric vehicle enters the charging area, a charging device is started, charging parking area monitoring data is acquired, and fire risk monitoring is performed; motion object features are distinguished through intelligent recognition algorithms, warning modes are automatically adjusted according to different types of objects, and abnormal events are pushed in real time. The application not only improves the safety during the charging process, but also enhances the intelligent management of charging facilities, and can effectively improve the safety and operation efficiency of the charging station.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric vehicle charging, in particular to an automatic protection type charging method and system for electric vehicles. BACKGROUND

[0002] In recent years, with the rapid development of the electric vehicle industry, the demand for charging of electric vehicles has been increasing. As an important part of electric vehicle infrastructure, charging piles have become a research focus in the field of intelligent transportation and energy management. Traditional charging pile systems mainly focus on providing charging interfaces and managing power supply, but there are still many problems in terms of safety and convenience during the charging process. With the popularization of electric vehicles, especially in public charging places, how to effectively ensure the safety during the charging process, improve the efficiency of equipment use and prevent potential risks has become an important direction of technological development. In recent years, technologies such as license plate recognition, face recognition, and object monitoring have been gradually introduced into intelligent charging management systems, which provides a new solution to improve the safety and automation level of charging piles.

[0003] Although existing charging management systems have integrated some intelligent technologies, such as vehicle recognition based on license plates and simple identity authentication functions, most systems still rely on manual intervention and simple power measurement, and lack the ability to monitor and respond to potential safety hazards during the charging process. The existing technology has the following shortcomings: first, the existing vehicle owner identity verification usually only relies on license plate recognition or account information, lacks a dual authentication mechanism, and may have certain security risks; second, the monitoring of charging equipment and environment mainly relies on manual inspection or simple fire alarm systems, and lacks an intelligent early warning mechanism for potential fire risks such as temperature and smoke; finally, most existing object recognition technologies only stay at the basic motion detection stage and cannot adjust the intelligent warning mode according to the object type, resulting in that the system's response to safety incidents is not timely and not accurate enough. SUMMARY

[0004] Therefore, the problem to be solved by the present application is how to improve the safety and automation level of charging stations by integrating advanced identity authentication, fire monitoring and intelligent warning systems.

[0005] To solve the above technical problems, the present application provides the following technical solutions:

[0006] In a first aspect, the present application provides an automatic protection type charging method for electric vehicles, which includes license plate recognition and vehicle model matching for vehicles entering the charging area, and simultaneously using face recognition technology for dual authentication of the vehicle owner's identity, combined with background information for permission verification.

[0007] After the electric vehicle enters the charging, the charging equipment is started, the charging parking space area monitoring data is acquired, and the fire risk monitoring is carried out;

[0008] The intelligent recognition algorithm is used to distinguish the features of moving objects, the warning mode is automatically adjusted according to different types of objects, and the abnormal events are pushed in real time.

[0009] As a preferred scheme of the electric vehicle automatic protection type charging method, wherein the permission verification comprises the following steps,

[0010] When the vehicle enters, the system generates an authorization code according to the owner reservation information in combination with the owner reservation system, and verifies whether the vehicle has charging permission by comparing the information in the authorization code;

[0011] If the license plate number in the owner reservation information completely matches the current license plate recognition result, and the owner passes the face recognition authentication, then the owner's identity is verified as legal;

[0012] If the license plate number in the owner reservation information partially matches the current license plate recognition result, and the owner passes the face recognition authentication, then the system will additionally perform manual review of the owner's identity to confirm whether the owner is the owner of the vehicle, and only allow charging after the owner confirms;

[0013] If the vehicle type in the owner reservation information matches the current vehicle type, and the owner passes the face recognition authentication, then the owner's reservation time period is verified as overlapping with the current time period;

[0014] If the vehicle type in the owner reservation information does not match the current vehicle type, and the owner passes the face recognition authentication, then the system will remind the owner and require the owner to provide a vehicle type confirmation, and after the owner provides the correct vehicle type and the vehicle meets the electric vehicle standard, the system dynamically allocates a charging parking space according to the current available parking spaces, and allows the vehicle to enter;

[0015] If the license plate number in the owner reservation information matches the current license plate recognition result, and the owner's face recognition passes, and the current time is within the reservation time period, then the verification is passed, and the vehicle is allowed to charge;

[0016] If the license plate number in the owner reservation information matches the current license plate number, and the owner's face recognition fails, then the system starts secondary identity verification, and if the license plate number in the owner reservation information completely matches the current license plate number, but the current vehicle does not meet the electric vehicle standard, then the system refuses to charge and prompts the owner that the vehicle does not meet the charging requirements.

[0017] As a preferred scheme of the electric vehicle automatic protection charging method, wherein: the monitoring data includes smoke concentration data, temperature change data and flame spectrum data; an infrared thermal imager scans the temperature distribution of the charging device and the vehicle chassis in real time to obtain thermal imaging data of the charging space area;

[0018] The collected smoke concentration data, temperature change data, flame spectrum data and thermal imaging data are normalized to remove noise data and extract effective data features;

[0019] The effective data features are reduced in dimension using principal component analysis (PCA) to extract key feature vectors, and the calculation formula is:

[0020]

[0021] In the formula, n is the number of samples, X is the data matrix, X T is the transpose of the data matrix, C is the covariance matrix, and the feature vectors corresponding to the first k largest eigenvalues are selected as the principal components;

[0022] A convolutional neural network (CNN) combined with a long short-term memory (LSTM) network is used to construct a fire risk prediction model, and the thermal imaging image is processed using the CNN.

[0023] The time series data of the smoke concentration data, temperature change data and flame spectrum data are input into the LSTM network, and the fire risk features in the time series data are output;

[0024] The image features extracted by the CNN and the time series features extracted by the LSTM are combined into a comprehensive feature vector;

[0025] The comprehensive feature vector is processed through a fully connected layer, and the fire risk probability is output, and the risk level is determined according to the fire risk probability to decide the corresponding measures, and the calculation formula is:

[0026] P = sigma (W o F + b o )

[0027] Where P represents the fire risk probability, W o and b o are the weights and biases of the fully connected layer, sigma is the sigmoid function, the output value is mapped to the interval [0, 1], and F is the comprehensive feature vector.

[0028] The historical fire data and normal charging data are used as the training set to train the model, and during the training process, the model is evaluated by the validation set.

[0029] As a preferred scheme of the electric vehicle automatic protection charging method, wherein: the determination of the risk level according to the fire risk probability to determine the corresponding measures comprises calculating a risk preset threshold, and the calculation formula is:

[0030]

[0031] In the formula, P t is the risk preset threshold, P j is the fire risk probability calculated in the jth monitoring, w j is a weight factor, and higher weight is given to more recent data, and R1 is the number of monitoring times.

[0032] When the fire risk probability P is less than or equal to the risk preset threshold P t , the system determines that the current fire risk probability is in a low risk interval, the system enters a low risk warning state, the system collects environmental data of the charging position in real time through temperature sensors, smoke sensors, flame sensors and infrared thermal imagers, and transmits the data to the local controller and the cloud server for storage and secondary analysis, and periodically evaluates the data, if the fire risk probability P is greater than the risk preset threshold P t is continuously monitored in the low risk warning state, the system automatically switches to a medium risk warning state, triggers a medium risk response measure, sends a warning information to a management personnel mobile terminal, prompts that there is a potential fire risk, and suggests that the management personnel perform on-site inspection, simultaneously automatically cuts off the charging power supply, stops the charging process, starts the ventilation system, and continues to monitor the fire risk probability, if the fire risk probability P is greater than the risk preset threshold P t in the medium risk state, the system automatically switches to a high risk warning state.

[0033] When the fire risk probability P is greater than the risk preset threshold P t , when the fire risk probability is in a high risk interval, the system enters a high risk warning state, the system immediately sends an emergency alarm to the management personnel and the fire department, and sends the alarm information to a remote monitoring platform, if the fire risk probability P is greater than the risk preset threshold P t remains unchanged in the high risk state, the system automatically enters an emergency processing mode.

[0034] As a preferred scheme of the electric vehicle automatic protection charging method, wherein: the distinguishing of the motion object features comprises the following steps,

[0035] A high-definition camera is used to collect video frames of the charging area, and the collected images are preprocessed, including grayscale, denoising and standardization.

[0036] A background model is established by a Gaussian mixture model (GMM), and in each pre-processed image, a difference between the current image and the background model is calculated to detect a moving object, and a motion direction and speed of the moving object are calculated by an optical flow method;

[0037] A Kalman filter is used to track a trajectory of the object, and a displacement and speed of the moving object are calculated, a bounding box is generated for each detected object, and motion information of the bounding box is recorded;

[0038] For each moving object, a shape feature and a motion feature are extracted;

[0039] A convolutional neural network (CNN) is used to train an object classification model, features of the moving object are used as input, object classification is performed, a cross-entropy loss function is used, network weight optimization is performed by a standard back propagation algorithm, and the loss function is defined as:

[0040]

[0041] In the formula, L is a loss function value, N is a sample quantity, y i is a true label of the i-th sample, P i is a predicted fire risk probability of the i-th sample;

[0042] Features of the detected moving object are input, object classification is performed by the trained CNN model, an object category is obtained, and a category label and a probability value of the object are output;

[0043] According to the object type recognition result, a warning mode is automatically adjusted to ensure safety of the charging area.

[0044] As a preferred scheme of the electric vehicle automatic protection type charging method, the object type recognition result includes,

[0045] When the system identifies that there is an animal in the charging area, the system starts an ultrasonic dispersing device, adjusts a frequency and intensity of the ultrasonic wave according to a body type and a category of the animal, and adjusts a direction and intensity of the ultrasonic wave emission according to a motion speed and a relative position of the animal;

[0046] When the system identifies that there is a human in the charging area, the system simultaneously starts a voice alarm and a visual warning, the system reminds personnel through voice broadcasting, safety prompts the personnel entering the charging area, dynamically adjusts a frequency and volume of the voice according to a distance and a stay time of the personnel, and when the human continues to approach the charging area, a warning light flashes, a safety prompt board displays warning information, and the personnel are reminded to keep a safe distance;

[0047] When the system identifies other objects in the charging area, the system initiates a low-frequency warning, either through voice prompts or display screen warnings, to remind the vehicle owner to pay attention to the presence of other objects.

[0048] The present application has the advantages of combining license plate and face recognition technology for dual authentication, integrating fire risk monitoring and intelligent response mechanism, and object type recognition and dynamic warning adjustment, which not only improves the safety during charging, but also enhances the intelligent management of charging facilities. Through accurate identity authentication, real-time fire monitoring and personalized warning mechanism, the safety and operation efficiency of the charging station are effectively improved, which belongs to the technical innovation in the field of electric vehicle charging system and intelligent safety monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor. Among them:

[0050] Figure 1 Flowchart of the electric vehicle automatic protection type charging method in embodiment 1. DETAILED DESCRIPTION

[0051] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0052] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from the description, and those skilled in the art can make similar generalizations without departing from the scope of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0053] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. In this specification, "in one embodiment" does not mean the same embodiment, nor does it mean an embodiment that is separate or selectively excluded from other embodiments.

[0054] Embodiment 1

[0055] Reference Figure 1 For the first embodiment of the present application, the embodiment provides an electric vehicle automatic protection type charging method, as shown in Figure 1 The following contents are included:

[0056] S1: License plate recognition and vehicle model matching for vehicles entering the charging area, while using face recognition technology for double authentication of the owner's identity, combined with background information for permission verification;

[0057] Specifically, through high-definition cameras, multi-angle license plate recognition and vehicle model matching are performed on vehicles entering the charging area, while using face recognition technology for double authentication of the owner's identity, combined with background reservation information for permission verification, to achieve precise access management of the charging area.

[0058] 1. Owner reservation information generates an authorization code

[0059] Before charging, the owner needs to make a reservation through a designated reservation platform (such as a charging pile APP, website, etc.). The reservation content usually includes the owner's information (such as name, owner's ID number or mobile phone number), vehicle information (such as license plate number, vehicle model, etc.), and charging time period and location.

[0060] After the owner completes the reservation, the system will store this information in the background database and generate a unique authorization code based on the specific content of the reservation. The authorization code is not just a normal identification mark, it also contains the owner's reservation information, vehicle information, and reservation time, etc. data, ensuring that each reservation is bound to a specific owner and vehicle.

[0061] When the owner arrives at the charging area, the system will generate an authorization code based on the owner's reservation information and send it to the owner through SMS, APP push, or QR code, etc. for identity verification.

[0062] 2. Face recognition and identity authentication of the owner

[0063] High-definition cameras are installed at the entrance of the charging area. When the owner needs to pass through this area, the system will automatically capture the owner's facial image. The high-definition camera is equipped with infrared sensing and dynamic adjustment functions to ensure clear shooting regardless of the owner's facial orientation.

[0064] When the owner enters the recognition range, the system uses deep learning algorithms (such as convolutional neural networks CNN) to scan the owner's face in real time, extracting facial features (such as eye, nose, mouth, etc. Position and relative proportion).

[0065] The system compares the captured facial features with the facial data uploaded by the owner on the reservation platform. This process uses face recognition technology, through feature point matching, deep feature learning, etc. to determine whether it is the same person. If the match is successful, the system considers the owner to be a legal user and proceeds to the next verification step; if the match fails, the system prompts the owner that the identity verification has failed and cannot continue.

[0066] 3. Matching of owner's identity and reservation information

[0067] After the car owner is identified by facial recognition, the system continues to identify the vehicle's license plate number through high-definition cameras and confirms the vehicle's type through image recognition technology. The system uses character recognition technology (such as OCR technology) to extract the license plate number and compare it with the license plate number provided by the car owner during the reservation.

[0068] The system will also check whether the license plate number and vehicle type information in the car owner's reservation information match the currently identified vehicle. If the match is successful, it indicates that the vehicle has charging privileges during the reservation period; if the license plate or vehicle type does not match the reservation information, the system will reject the car owner's charging request and remind the car owner to check the reservation information or provide modifications.

[0069] 4. Charging privilege confirmation

[0070] After the vehicle and car owner's identity are confirmed, the system will verify whether the car owner is within the reservation time period by comparing the information in the authorization code (such as the car owner's information, reservation period, etc.).

[0071] If all information matches and passes the verification, the system will allow the car owner to enter the charging area, the charging pile will be turned on, and the car owner can start charging operations.

[0072] If any abnormalities are found at any stage (such as mismatched car owner identity, license plate, or reservation time), the system will automatically suspend entry privileges and inform the car owner of the reason for rejection through voice or screen prompts.

[0073] If the license plate number in the car owner's reservation information completely matches the current license plate recognition result, and the car owner passes the facial recognition authentication, the car owner's identity is verified as legitimate.

[0074] If the license plate number in the car owner's reservation information partially matches the current license plate recognition result (such as some fuzzy characters in license plate recognition), and the car owner passes the facial recognition authentication, the system will additionally conduct manual review of the car owner's identity to confirm whether the car owner is the owner of the vehicle, and only allow charging after the car owner's confirmation.

[0075] If the vehicle type in the car owner's reservation information matches the current vehicle type, and the car owner passes the facial recognition authentication, the system verifies that the car owner's reservation period coincides with the current time period.

[0076] If the vehicle type in the car owner's reservation information does not match the current vehicle type, and the car owner passes the facial recognition authentication, the system will warn the car owner and require confirmation of the vehicle type; if the car owner provides the correct vehicle type, the system will dynamically allocate a parking space based on the current available parking spaces and allow entry.

[0077] If the license plate number in the owner's reservation information matches the current vehicle's license plate number, and the owner's face recognition passes, and the current time is within the reservation period, the verification is passed, and charging is allowed.

[0078] If the license plate number in the owner's reservation information matches the current license plate number, but the owner's face recognition fails, the system will start secondary identity verification: require the owner to present an identity document or verify the mobile phone number.

[0079] If the license plate number in the owner's reservation information matches the current license plate number, and the current vehicle is an electric vehicle type, and the owner's identity is authenticated by face recognition, the system verifies the remaining battery capacity of the vehicle, and arranges the charging period reasonably according to the remaining battery capacity to avoid overcharging.

[0080] If the license plate number in the owner's reservation information matches the current license plate number, but the current vehicle does not meet the electric vehicle standard (such as license plate recognition for hybrid or fuel vehicles), the system refuses to charge and prompts the owner that the vehicle does not meet the charging requirements.

[0081] By performing license plate recognition and vehicle model matching on vehicles entering the charging area, and using face recognition technology for double authentication of the owner's identity, strict identity verification during the charging process is achieved. Compared with traditional license plate recognition systems, double authentication can significantly improve the accuracy and security of identity verification, prevent unauthorized vehicles from using charging equipment, and effectively eliminate safety hazards and resource abuse. This step not only ensures the safety of the charging system, but also improves the utilization efficiency of the charging station resources, ensures that only authorized electric vehicles can charge, and improves user experience and equipment management accuracy.

[0082] In combination with the background information, the owner's reservation system is verified for authority, and after the owner completes identity verification through license plate, vehicle model and face recognition information, the system can automatically match and dynamically allocate charging spaces. This ensures reasonable allocation of charging spaces, avoids human intervention and queuing, and optimizes the overall operation efficiency of the charging station. In addition, the system can handle cases where the owner's identity authentication fails, and through manual review, it ensures that every link in the charging process meets safety standards, which not only effectively prevents charging fraud, but also enhances the automation management capability of the charging system.

[0083] S2: After the electric vehicle enters the charging station, start the charging equipment, obtain the charging space area monitoring data, and perform fire risk monitoring;

[0084] Specifically, a distributed temperature sensing network and an AI intelligent infrared thermal imaging system are used to conduct comprehensive thermal distribution scanning on the charging parking area, and a temperature change curve model of the vehicle charging process is established. When abnormal temperature fluctuations or hot spots are detected, the system will issue a graded warning.

[0085] After the vehicle identity verification is passed, the charging equipment is started, and the system initializes the fire monitoring, small animal driving, and intelligent prediction functions. The fire sensors (smoke sensor, temperature sensor, flame sensor) are activated. The infrared thermal imager is started to scan the temperature distribution of the charging equipment and the vehicle chassis in real time. The small animal activity sensor (infrared induction, microwave radar) is started.

[0086] Each sensor collects data in real time and transmits the data to the local controller through the wireless communication module. The local controller aggregates the data and uploads it to the cloud server through the 5G network for analysis by the intelligent algorithm.

[0087] Fire monitoring and identification: The temperature sensor monitors the temperature changes of the charging parking space in real time. If the temperature exceeds the preset threshold, the system marks it as a potential fire risk point. The smoke sensor detects the smoke concentration in the air. If the smoke concentration exceeds the preset threshold, the system triggers an alarm. The flame sensor detects the spectral characteristics of the flame. If a flame is detected, the system immediately triggers a fire alarm.

[0088] The infrared thermal imager scans the temperature distribution of the charging equipment and the vehicle chassis at a frequency of 10 frames per second, generating real-time thermal images. The system identifies high-temperature areas through image processing algorithms and marks them as potential hot spots. The high-temperature areas are continuously monitored, and if the temperature rises by more than 20°C in a short period of time, the system determines it as a fire risk area.

[0089] The collected temperature, smoke concentration, flame signal, and thermal imaging data are normalized. Noise data is removed, and effective features (such as temperature change rate, smoke concentration change rate, high-temperature area area, etc.) are extracted.

[0090] Temperature features: Extract temperature change rate, temperature peak, temperature gradient, etc. Smoke features: Extract smoke concentration change rate, smoke diffusion speed, etc. Thermal imaging features: Extract high-temperature area area, high-temperature area temperature change rate, texture features of thermal images, etc.

[0091] Use principal component analysis (PCA) to reduce the dimensionality of the effective data features, extract the key feature vectors, and the calculation formula is:

[0092]

[0093] In the formula, n is the number of samples, X is the data matrix, and X TFor the transpose of the data matrix, select the eigenvectors corresponding to the top k largest eigenvalues as the principal components.

[0094] A convolutional neural network (CNN) combined with a long short-term memory network (LSTM) is used to build a fire risk prediction model.

[0095] A convolutional neural network (CNN) is used to process the thermal imaging images. Local features such as the shape and size of high-temperature regions are extracted. Pooling (usually max pooling) is used to reduce the dimensionality of the features while retaining the most important information. After multiple layers of convolution and pooling, image features related to fire risk are extracted.

[0096] Time series data of temperature, smoke concentration, and flame signals are input into a long short-term memory network (LSTM). Capture dynamic changes: LSTM can remember past inputs and determine if current changes are abnormal. For example, a rapid rise in temperature or a sudden increase in smoke concentration.

[0097] LSTM outputs a feature vector representing fire risk features in time series data.

[0098] Combine image features and time series features to comprehensively assess fire risk. Merge image features extracted by CNN and time series features extracted by LSTM into a comprehensive feature vector.

[0099] Process the comprehensive feature vector through a fully connected layer, and finally output a value between 0 and 1 representing the probability of fire occurrence. If this probability value is close to 1, it means the fire risk is high; if it is close to 0, it means the fire risk is low. The system decides whether to take measures (such as issuing an alarm or starting a fire extinguishing device), and the formula is:

[0100] P = σ(W o F + b o )

[0101] Where P represents the fire risk probability, W o and b o are the weights and biases of the fully connected layer, σ is the sigmoid function that maps the output value to the [0, 1] interval, and F is the comprehensive feature vector.

[0102] Use historical fire data and normal charging data as the training set to train the model. The loss function uses the cross-entropy loss function, and the optimizer selects the Adam optimizer. During training, the model is evaluated by the validation set to prevent overfitting.

[0103] The real-time collected feature data is input into the trained model, and the model outputs the fire risk probability. If the fire risk probability P exceeds the risk preset threshold, the system judges that there is a high fire risk.

[0104]

[0105] In the formula, P t is the risk preset threshold, P j is the fire risk probability calculated in the jth monitoring, w j is the weight factor, which gives higher weight to more recent data, and R1 is the number of monitoring times.

[0106] When the fire risk probability P≤risk preset threshold P t , the fire risk is in the low risk interval, and the system enters the low risk warning state. In this state, the system mainly takes real-time monitoring and data recording measures. Specifically, the system collects environmental data of the charging parking space in real time through temperature sensors, smoke sensors, flame sensors, and infrared thermal imagers, and transmits these data to the local controller and the cloud server for storage and analysis. At the same time, the system will regularly evaluate the data to ensure that the risk probability does not suddenly rise. If the risk probability has a rising trend during continuous monitoring in the low risk state, the system will automatically switch to the medium risk warning state and trigger the corresponding response measures. This low risk warning and response mechanism can timely discover potential problems in the early stage of fire risk, provide basic data support for subsequent warning and response, and ensure that the system can take action quickly when the risk probability rises.

[0107] When the fire risk probability P> risk preset threshold P t , the medium risk probability is in the medium risk interval, and the system enters the medium risk warning state. In this state, the system not only continues to monitor and record data in real time, but also takes a series of proactive warning and local control measures. Specifically, the system sends warning information to the mobile terminal of the manager, prompting that there is a potential fire risk, and suggests that the manager conducts on-site inspection. At the same time, the system automatically cuts off the charging power supply to stop the charging process to prevent the fire risk from further expanding. In addition, the system also starts the ventilation system to reduce the temperature and smoke concentration of the charging parking space, reducing the possibility of fire. If the risk probability continues to rise in the medium risk state, the system will automatically switch to the high risk warning state and trigger higher level response measures. This risk warning and response mechanism can take local control measures in time in the middle stage of fire risk, effectively reduce the possibility of fire, and ensure the safety of personnel and equipment.

[0108] When the fire risk probability is in the high-risk interval, the system enters a high-risk warning state. In this state, the system will take the highest level of warning and emergency response measures. Specifically, the system will immediately send an emergency alert to the management personnel and the fire department, notifying them to take action as soon as possible. At the same time, the system will automatically start the fire extinguishing device (such as a carbon dioxide fire extinguisher or a sprinkler system) to extinguish the fire on the charging spot. In addition, the system will send the alarm information to the remote monitoring platform through the 5G network or the Internet of Things technology, ensuring that the management personnel and the fire department can receive the alarm in time and conduct emergency command. If the risk probability does not decrease in the high-risk state, the system will automatically enter the emergency handling mode and link with the emergency command system of the fire department to provide real-time data and video monitoring information of the fire scene, assisting the fire department in fire extinguishing and rescue. This high-risk warning and response mechanism can quickly take emergency measures when the fire risk is high, minimizing the damage to personnel and property caused by fire, and ensuring that the fire is controlled and extinguished in time.

[0109] By collecting smoke concentration, temperature change and flame spectrum data in the charging area, and combining with infrared thermal imaging images, the fire risk during charging is monitored and analyzed in real time. Through principal component analysis (PCA) for dimension reduction of multi-dimensional data, and through the deep learning model combining convolutional neural network (CNN) and long short-term memory network (LSTM), the fire risk is predicted and risk level is judged, so as to realize intelligent warning of fire risk. The fire prevention ability during charging is significantly improved, the limitations of human monitoring are reduced, and through the automatic alarm system and emergency response mechanism, intervention can be carried out before the occurrence of fire risk, ensuring the safe operation of charging facilities.

[0110] S3: Differentiate the characteristics of moving objects through intelligent recognition algorithm, automatically adjust the warning mode according to different types of objects, and push the abnormal events in real time.

[0111] Specifically, high-precision infrared motion sensor arrays and ultrasonic emitting devices are arranged around the charging area. Through intelligent recognition algorithm, the characteristics of small animal activities are distinguished, and the ultrasonic frequency and sound pressure are automatically adjusted according to different types of small animals to achieve precise dispersion while avoiding interference to charging vehicles and personnel.

[0112] Infrared induction sensors and microwave radars monitor small animal activities around the charging spot in real time. If small animals are detected approaching charging equipment or vehicle chassis, the system marks it as an interference event. Start the ultrasonic repellent device to emit ultrasonic signals with a frequency of 40 kHz to drive away small animals. If the ultrasonic repellent is ineffective, start the light flashing device to interfere with the vision of small animals through strong light flashing, forcing them to leave.

[0113] The system records the time, location, and measures taken to deter the small animal interference events. If multiple instances of small animal interference are detected consecutively, the system sends an alert to the management personnel, suggesting further measures (such as installing protective nets).

[0114] The results of each fire monitoring and small animal deterrence event are fed back into the intelligent algorithm model. Using an incremental learning method, the model is dynamically updated to optimize fire risk prediction and small animal deterrence strategies.

[0115] The model is periodically retrained, incorporating new fire data and small animal interference data to improve its accuracy and robustness.

[0116] Through simulation testing and actual operation data comparison, the model parameters are optimized to ensure the reliability of the system in different environments.

[0117] High-definition cameras are used to capture video frames (e.g., 30 frames per second) of the charging area. The captured images are preprocessed, including grayscale conversion, noise removal (using Gaussian filtering), and standardization (normalizing pixel values to the [0, 1] interval).

[0118] Motion object detection is performed based on the combination of optical flow and background subtraction methods. In each frame, the difference between the current image and the background model is calculated to detect moving objects. The background model can be modeled using a Gaussian Mixture Model (GMM) to adapt to changes in lighting and dynamic scenes. Combined with the background subtraction result, the optical flow method is used to further calculate the motion direction and speed of the moving object. The optical flow method can use the Hough transform to analyze the motion trajectory of the object and track the motion pattern of the moving object.

[0119] Kalman filtering is used to track the trajectory of the object's motion, calculating the displacement and speed of the moving object. A bounding box is generated for each detected object, and the motion information of the bounding box is recorded as input for subsequent classification and warning.

[0120] For each moving object, its shape features (such as contour, size, shape) and motion features (such as speed, direction) are extracted. Methods such as HOG (Histogram of Oriented Gradients) and SIFT (Scale-Invariant Feature Transform) can be used for feature extraction.

[0121] A convolutional neural network (CNN) is used to train the object classification model. The CNN model uses the extracted features of the moving object as input for object classification. Assuming that the object types to be classified include "animal," "human," and "other objects," a training set can be prepared for each category. Using the cross-entropy loss function, the network weights are optimized through the standard backpropagation algorithm. The loss function is defined as:

[0122]

[0123] where L is the loss function value, N is the number of samples, y i is the true label of the i-th sample, P i is the predicted fire risk probability of the i-th sample.

[0124] Input the detected features of the moving object, pass through the trained CNN model for object type classification, and obtain the object category.

[0125] Output the category label of the object (such as animal, person, other) and the related probability value (indicating the confidence that the object belongs to this category).

[0126] According to the identification result of the object type, automatically adjust the warning mode to ensure the safety of the charging area.

[0127] Animal objects: When the system identifies animals in the charging area, the system will start the ultrasonic dispersing device, and adjust the frequency and intensity of the ultrasonic waves according to the size and type of the animal. For example, small animals may require higher frequency ultrasonic waves, while large animals use lower frequency. The system can also adjust the emission direction and intensity of the ultrasonic waves according to the speed and relative position of the animal, so as to maximize the dispersing effect and avoid interference to the surrounding people or equipment.

[0128] Human objects: If a human enters the charging area, the system will simultaneously start voice alarm and visual warning, and the system will remind the personnel through voice broadcast, such as "Please keep away from the charging pile", to ensure that the personnel entering the charging area receive timely safety prompts. The system will dynamically adjust the frequency and volume of the voice according to the distance and time of the personnel, to ensure that the warning sound is enough to arouse attention.

[0129] When a human approaches the charging area, the warning light will flash and the safety prompt board will display obvious warning information, reminding the personnel to maintain a certain safety distance. The system will adjust the flashing frequency and brightness of the warning light according to the degree of the personnel approaching the charging device and the time of staying, to arouse enough attention.

[0130] Other objects: For objects classified as "other objects" (such as paper, wind-blown leaves, etc.), the system will start a low-frequency warning, and the system will remind the personnel to pay attention to the existence of the object through a short voice prompt or display screen display warning information. If the identified object stays for a very short time or its speed is too low, the system will automatically ignore or reduce the warning level, reducing unnecessary interference and resource waste.

[0131] If there are multiple objects in the charging area at the same time, the system will set priorities according to the types and threat levels of the objects. Generally, humans and animals have a higher threat level, so the corresponding warnings will be triggered first.

[0132] The system automatically adjusts the warning scheme according to the dynamic changes of the objects. For example, when the moving track of the animal becomes irregular, the system will adjust the ultrasonic frequency to enhance the dispersing effect.

[0133] Abnormal events are pushed in real time to the monitoring center, property management department, user APP and fire department, and the push information includes event type, location coordinates, on-site image and disposal suggestion, and an emergency disposal process is automatically generated to ensure rapid coordinated response of all parties.

[0134] By collecting video frames of the charging area through high-definition cameras, using Gaussian Mixture Model (GMM) combined with optical flow method for object detection, and tracking the motion track of the object through Kalman filtering, the system can accurately identify and classify animals, humans and other objects in the charging area. According to the different types of objects, the system automatically adjusts the warning mode. Through dynamic monitoring and intelligent analysis of the charging area, the system's response to different objects is more accurate and effective, avoiding unnecessary warnings and improving the safety of users and devices.

[0135] The embodiment also provides a computer device suitable for the automatic protection type charging method, including a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions to realize the automatic protection type charging method proposed in the above embodiment.

[0136] The computer device can be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be realized through Wi-Fi, operator network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device. In addition, the input device can also be an external keyboard, touchpad or mouse, etc.

[0137] The embodiment also provides a storage medium on which a computer program is stored, the program being executed by a processor to implement the automatic protection charging method proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk, or an optical disk.

[0138] To sum up, the application improves the safety of the charging process and enhances the intelligent management of the charging facility by combining the license plate and face recognition technologies for double authentication, integrating the fire risk monitoring and intelligent response mechanism, and identifying the object type and adjusting the dynamic warning. The application effectively improves the safety and operation efficiency of the charging station through accurate identity authentication, real-time fire monitoring, and personalized warning mechanism, and belongs to the technical innovation in the field of electric vehicle charging systems and intelligent safety monitoring.

[0139] It should be noted that the above embodiments are only used to illustrate the technical solutions of the application but not limit the application. Although the application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the application can be modified or replaced equivalently without departing from the spirit and scope of the application, and all of them should be covered in the scope of the claims of the application.

Claims

1. An automatic protective charging method for electric vehicles, characterized in that: The application relates to a fire risk prediction method and device for an electric vehicle charging area. The method comprises the following steps: When the electric vehicle enters the charging area, license plate recognition and vehicle model matching are performed on the vehicle, and face recognition technology is used to double-authenticate the identity of the vehicle owner, and permission verification is performed in combination with background information; After the electric vehicle enters the charging area, the charging equipment is started, charging parking area monitoring data is acquired, and fire risk monitoring is performed; The monitoring data comprises smoke concentration data, temperature change data and flame spectrum data; an infrared thermal imager scans the temperature distribution of the charging equipment and the vehicle chassis in real time, and charging parking area thermal imaging data are obtained; The acquired smoke concentration data, temperature change data and flame spectrum data and the thermal imaging data are normalized, effective data features are extracted after noise data are removed; In the formula, n is the sample number, X is a data matrix, X T is the transpose of the data matrix, C is a covariance matrix, and the eigenvectors corresponding to the first k largest eigenvalues are selected as principal components. Principal component analysis (PCA) is used to reduce the dimensionality of the effective data features, and a key feature vector is extracted, and the calculation formula is as follows: A convolutional neural network (CNN) combined with a long short-term memory (LSTM) network is used to construct a fire risk prediction model, and the CNN is used to process thermal imaging images; The time series data of the smoke concentration data, the temperature change data and the flame spectrum data are input into the LSTM, and fire risk features in the time series data are output; The image features extracted by the CNN and the time series features extracted by the LSTM are combined into a comprehensive feature vector; P = σ(W o F + b o ) where P represents the fire risk probability, W o and b o are the weights and bias of the fully connected layer, respectively, σ is a sigmoid function that maps the output value to the interval [0, 1], and F is the comprehensive feature vector. The comprehensive feature vector is processed through a full connection layer, and a fire risk probability is output, the risk level is determined according to the fire risk probability, and corresponding measures are taken, and the calculation formula is as follows: Historical fire data and normal charging data are used as a training set to train the model, and the model is evaluated through a validation set during the training process; where P t is a risk preset threshold, P j is a fire risk probability calculated in the jth monitoring, w j is a weight factor, giving higher weight to more recent data, R1 is the number of monitoring times; When the fire risk probability P is less than the risk preset threshold P t , the system determines that the current fire risk probability is in a low risk interval, the system enters a low risk warning state, the system collects environmental data of the charging parking space in real time through the temperature sensor, the smoke sensor, the flame sensor and the infrared thermal imager, and transmits the data to the local controller and the cloud server for storage and secondary analysis, and regularly evaluates the data. If the fire risk probability P is greater than the risk preset threshold P t in the low risk warning state, the system automatically switches to a medium risk warning state, triggers a medium risk response measure, sends warning information to the management personnel mobile terminal, prompts that there is a potential fire risk, and suggests that the management personnel perform on-site inspection, automatically cuts off the charging power supply, stops the charging process, starts the ventilation system, and continues to monitor the fire risk probability. When the fire risk probability P is greater than the risk preset threshold P t in the medium risk state, the system will automatically switch to a high risk warning state. When the fire risk probability P > risk preset threshold P t When the fire risk probability is in the high risk interval, the system enters the high risk warning state, the system immediately sends an emergency alarm to the management personnel and the fire department, and sends the alarm information to the remote monitoring platform. If the fire risk probability P > risk preset threshold P t does not decrease, the system automatically enters the emergency processing mode; The risk level is determined according to the fire risk probability, and corresponding measures are taken, which comprises the following steps: An intelligent recognition algorithm is used to distinguish the features of moving objects, and the warning mode is automatically adjusted according to different types of objects, and abnormal events are pushed in real time; The method for distinguishing the features of moving objects comprises the following steps: A high-definition camera is used to collect video frames of the charging area, and the collected images are preprocessed, including grayscale, denoising and standardization; A Gaussian mixture model (GMM) is used to establish a background model, the difference between the current image and the background model is calculated in each preprocessed image, moving objects are detected, and the motion direction and speed of the moving objects are calculated through an optical flow method; Kalman filtering is used to track the trajectory of the objects, the displacement and speed of the moving objects are calculated, a bounding box is generated for each detected object, and the motion information of the bounding box is recorded; The shape features and motion features of each moving object are extracted; In the formula, L is a loss function value, N is the number of samples, yi i is the true label of the i th sample, P i is the probability of the i th sample; A convolutional neural network (CNN) is used to train an object classification model, the features of the moving objects are used as input, object classification is performed, a cross-entropy loss function is used, network weight optimization is performed through a standard back propagation algorithm, and the loss function is defined as follows: The features of the detected moving objects are input, the object type is classified through the trained CNN model, the object category is obtained, and the category label and probability value of the object are output. According to the recognition result of the object type, the warning mode is automatically adjusted to ensure the safety of the charging area.

2. The automatic protection type charging method for electric cars according to claim 1, characterized by: The permission verification includes the following steps, When the vehicle enters, in combination with the owner reservation system, the system generates an authorization code according to the owner reservation information, and verifies whether the vehicle has charging permission by comparing the information in the authorization code; If the license plate number in the owner reservation information completely matches the current license plate recognition result, and the owner passes the face recognition authentication, the owner's identity is verified to be legal; If the license plate number in the owner reservation information partially matches the current license plate recognition result, and the owner passes the face recognition authentication, the system will additionally conduct manual review of the owner's identity to confirm whether the owner is the owner of the vehicle, and only allow charging after the owner confirms; If the vehicle model in the owner reservation information matches the current vehicle model, and the owner passes the face recognition authentication, continue to verify whether the owner's reservation time period coincides with the current time period; If the vehicle model in the owner reservation information does not match the current vehicle model, and the owner passes the face recognition authentication, the system will remind the owner and require the owner to provide a vehicle model confirmation. After the owner provides the correct vehicle model and the vehicle meets the electric vehicle standard, the system dynamically allocates a charging parking space according to the current available parking space and allows the vehicle to enter; If the license plate number in the owner reservation information matches the current license plate recognition result, and the owner's face recognition passes, and the current time is within the reservation time period, the verification is passed and the vehicle is allowed to charge; If the license plate number in the owner reservation information matches the current license plate number, and the owner's face recognition fails, the system initiates a secondary identity verification. If the license plate number in the owner reservation information completely matches the current license plate number, but the current vehicle does not meet the electric vehicle standard, the system denies charging and prompts the owner that the vehicle does not meet the charging requirements.

3. The automatic protection type charging method for electric cars according to claim 2, characterized by: The recognition result of the object type includes, When the system identifies that there is an animal in the charging area, the system starts the ultrasonic dispersing device, adjusts the frequency and intensity of the ultrasonic waves according to the size and type of the animal, and adjusts the emission direction and intensity of the ultrasonic waves according to the movement speed and relative position of the animal; When the system identifies that there is a human in the charging area, the system simultaneously starts voice alarm and visual warning. The system reminds the personnel through voice broadcast and provides safety prompts to the personnel entering the charging area. The system dynamically adjusts the frequency and volume of the voice according to the distance and residence time of the personnel. When the human continues to approach the charging area, the warning light flashes and the safety prompt board displays warning information to remind the personnel to maintain a safe distance; When the system identifies that there is another object in the charging area, the system starts a low-frequency warning, which reminds the owner to pay attention to the existence of other objects through voice prompts or display screen display of warning information.

Images