Power exchange cabinet safety monitoring alarm method, device and readable storage medium

Abstract

The present invention discloses a method, device and readable storage medium for safety monitoring and alarm of a power-swap cabinet, including the following steps: S1, the power-swap cabinet controls the camera to start; S2, the power-swap cabinet captures the video stream through the camera in S1; S3, the power-swap cabinet performs behavior analysis and detection based on a deep learning model. If it is a normal behavior, the power-swap cabinet controls the camera to continue monitoring. If it is an abnormal behavior, the power-swap cabinet triggers an early warning and enters S4; S4, the power-swap cabinet records the event and its detailed information, and then the power-swap cabinet captures it through the camera; S5, the power-swap cabinet uploads the event and detailed information in S4 to cloud storage and notifies the management personnel, who respond to and resolve the event. The present invention has the advantages of real-time monitoring and capturing functions, can reduce the risk of theft and equipment damage, and can prevent users from "cheating to lock".

Description

Technical Field

[0001] The present invention belongs to the technical field of power exchange cabinets, and specifically relates to a power exchange cabinet safety monitoring and alarm method, device and readable storage medium. Background Art

[0002] With the widespread application of two-wheeled electric vehicles and battery technology, battery swap cabinets have become an important infrastructure for two-wheeled electric vehicle battery swaps. However, the existing battery swap cabinet system has the following major safety hazards and management issues during use:

[0003] 1. Risk of theft and equipment damage: Since battery swap cabinets are usually located in public places, they are at risk of vandalism, theft of batteries or equipment, and the existing system lacks effective real-time monitoring and protection measures.

[0004] 2. User "lock-cheating" behavior: During the battery return process, some users close the door lock by improper means, causing the system to mistakenly believe that the door is locked, allowing the user to complete the order normally. But in fact, the door is not completely closed, and the user can still open the door and steal the battery after the order is completed. This "lock-cheating" behavior cannot be effectively identified and prevented by the existing system. Summary of the invention

[0005] In order to solve the deficiencies in the prior art, the present invention provides a method and device for safety monitoring and alarming of a power exchange cabinet, and a readable storage medium.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0007] The battery swap cabinet safety monitoring and alarm method is implemented based on a deep learning model and includes the following steps:

[0008] S1, the battery swap cabinet controls the camera to start;

[0009] S2, the battery swap cabinet captures the video stream through the camera in S1;

[0010] S3: The battery swap cabinet performs behavior analysis and detection based on the deep learning model. If it is normal behavior, the battery swap control camera continues to monitor. If it is abnormal behavior, the battery swap cabinet triggers an early warning and enters S4.

[0011] S4, the power swap cabinet records the event and its detailed information, and then the power swap cabinet captures it through a camera;

[0012] S5. The battery swap cabinet uploads the event and detailed information in S4 to the cloud storage and notifies the management personnel, who respond and resolve the event;

[0013] Among them, S3 uses a deep learning model to input real-time frame images from the camera, extracts the spatial features and temporal features of the image and the interaction features between people and objects in the image after preprocessing, and outputs behavior classification results, abnormal behavior types, target trajectories, and alarm trigger signal data features.

[0014] Preferably, the deep learning model includes a two-stage target detection algorithm R-CNN, a Pose Estimation model, and a behavior classification model, wherein the two-stage target detection algorithm R-CNN is used to detect and locate targets in an image, including people and objects;

[0015] The Pose Estimation model is used to analyze body posture;

[0016] The behavior classification model is used to analyze behavior based on the time series features extracted by the convolutional neural network and combined with the time series data model.

[0017] Preferably, in S3, a deep learning model is used to distinguish image data of images taken by the camera into daytime scenes and nighttime scenes, and then the image data is labeled to generate label data, marked as normal behavior and abnormal behavior, and then the label data is used for data enhancement training.

[0018] Preferably, in S3, the label data is used to train the two-stage target detection algorithm R-CNN to learn the location information of the target, and the behavior classification model is trained by combining the action features and posture features with the labeled time series behavior data.

[0019] Preferably, it also includes a temperature monitoring and early warning step, which includes starting the power swap cabinet control sensor, the power swap cabinet performing real-time temperature detection through the sensor, and the power swap cabinet judging whether the temperature is normal through the real-time temperature detection data. If it is normal, the power swap cabinet controls the sensor to continue monitoring, and if it is abnormal, the power swap cabinet triggers an early warning and issues an early warning notification. The power swap cabinet sends the early warning notification to the management personnel, the management personnel responds and records the temperature data, and then the management personnel handles it.

[0020] The present invention also discloses a power exchange cabinet safety monitoring and alarm device, including a processor and a memory. The processor is used to execute the above-mentioned power exchange cabinet safety monitoring and alarm method by calling a program or instruction stored in the memory.

[0021] The present invention also discloses a readable storage medium, which stores a program or instruction, and the program or instruction executes the above-mentioned power exchange cabinet safety monitoring and alarm method.

[0022] By adopting the above technical solution, the present invention has the following beneficial effects:

[0023] (1) The power exchange cabinet of the present invention is monitored by a camera and then identifies abnormal behavior through a deep learning model. It has real-time monitoring and snapshot functions, which can reduce the risk of theft and equipment damage;

[0024] (2) The power exchange cabinet of the present invention is monitored by a camera and then, when a behavior inconsistent with the normal mode is detected by a deep learning model, an alarm can be automatically triggered to prevent the user from "cheating the lock";

[0025] (3) The present invention also has a temperature monitoring and early warning step, which can determine abnormal temperature behavior and effectively perform temperature detection;

[0026] In summary, the present invention has the advantages of real-time monitoring and snapshot functions, can reduce the risk of theft and equipment damage, and can prevent users from "cheating to lock" and the like. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a schematic diagram of the process of the present invention;

[0028] Figure 2 It is a flow chart of the temperature monitoring and early warning steps of the present invention. DETAILED DESCRIPTION

[0029] The technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.

[0030] The components of the embodiments of the present invention generally described and shown in the drawings herein may be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention.

[0031] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making any creative work shall fall within the scope of protection of the present invention.

[0032] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0033] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0034] Example 1

[0035] In this embodiment, this embodiment proposes a power swap cabinet safety monitoring alarm method, device and readable storage medium, which can improve the safety of the power swap cabinet.

[0036] like Figure 1As shown, in one embodiment of the present invention, the battery swap cabinet safety monitoring and alarm method of the present invention is implemented based on a deep learning model. The deep learning model is a key component based on computer vision and behavior analysis. The specific model includes a convolutional neural network (CNN) and a two-stage target detection algorithm R-CNN, such as Faster R-CNN or Mask R-CNN. These models are used for image feature extraction, target detection, object classification and behavior analysis. In addition, the present invention may also introduce a time series data model (such as LSTM or Transformer) to capture the time series characteristics of user actions, so as to more accurately analyze the behavior. More specifically, the deep learning model includes a two-stage target detection algorithm R-CNN, a Pose Estimation model and a behavior classification model, wherein the two-stage target detection algorithm R-CNN is used to detect and locate targets in the image, including people and objects, and the Pose Estimation model is used to analyze body posture, more specifically the body posture of the user using the battery swap cabinet, and the Pose The Estimation model can be OpenPose or a posture estimation network based on CNN. The behavior classification model is used to analyze behaviors based on the time series features extracted by the convolutional neural network and in combination with the time series data model. The time series data model can be LSTM or Transformer. The hardware of the present invention specifically includes cameras, sensors, and computing units. The cameras include high-resolution cameras and thermal imaging cameras, providing all-weather monitoring capabilities and adapting to various lighting conditions. The sensors can be integrated temperature sensors, motion sensors, and ambient light sensors, which can monitor the internal and surrounding environmental conditions of the power exchange cabinet in real time. The computing unit adopts a high-performance processor with edge computing technology, supports real-time data processing, and reduces delays. The software part of the present invention includes a monitoring management system and a deep learning model. The monitoring management system is responsible for data collection, analysis, and storage, and has a friendly user interface, which is convenient for operators to view the monitoring status in real time. The deep learning model is based on a behavior recognition model of computer vision, which is continuously optimized through training historical data to enhance the detection capability of abnormal behavior.

[0037] The present invention specifically comprises the following steps:

[0038] S1, the battery swap cabinet controls the camera to start;

[0039] S2, the battery swap cabinet captures the video stream through the camera in S1;

[0040] S3, the battery swap cabinet performs behavior analysis and detection based on the deep learning model. If it is normal behavior, the battery swap control camera continues to monitor. If it is abnormal behavior, the battery swap cabinet triggers an early warning and enters S4. In S3, the image data taken by the camera is distinguished by the deep learning model, and the image data is distinguished into daytime scenes and nighttime scenes. Then the image data is labeled to generate label data, which is marked as normal behavior and abnormal behavior. Then, the label data is used for data enhancement training. In S3, the label data is used to train the two-stage target detection algorithm R-CNN, learn the location information of the target, and train the behavior classification model through the labeled time series behavior data combined with the action features and posture features;

[0041] When performing specific behavior analysis and detection in the present invention, data input and preprocessing are required, wherein the data input is image data, that is, data captured by a camera, specifically images obtained from a high-resolution camera and a thermal imaging camera, which are divided into daytime scenes and nighttime scenes, and then these data are labeled, and normal behaviors (such as correct locking, normal battery replacement) and abnormal behaviors (such as non-compliant unlocking, long-term uncompleted operations, etc.) need to be marked to generate labeled data, and then random cropping, rotation, lighting adjustment and other technologies are used to enhance the diversity of training data, thereby performing data enhancement and improving the robustness of the deep learning model;

[0042] Then, the deep learning model is trained. First, the two-stage target detection algorithm R-CNN is trained using the annotated image data (label data) to learn the location information of the target (including the location information of people and batteries). Then, the behavior classification model is trained by combining the annotated time series behavior data with the action features and posture features. The cross entropy loss function is used for classification optimization. At the same time, IoU (Intersection over Union) is used as the target detection performance evaluation indicator to optimize the deep learning model.

[0043] S4, the power swap cabinet records the event and its detailed information, and then the power swap cabinet captures it through a camera;

[0044] S5. The battery swap cabinet uploads the event and detailed information in S4 to the cloud storage and notifies the management personnel, who respond and resolve the event;

[0045] Among them, S3 uses a deep learning model to input real-time frame images from the camera, and extracts the spatial features (such as position, shape or action) and temporal features (such as action change trajectory) of the image after preprocessing, as well as the interactive features of people and objects in the image (such as the position relationship between the hand and the door lock, the movement trajectory of the battery, etc.), and outputs the behavior classification results, abnormal behavior types, target trajectories, and alarm trigger signal data features. Among them, the behavior classification results include whether the user's current behavior is normal or abnormal, and the abnormal behavior types include abnormal locking action, operation timeout, improper battery placement, etc. The target trajectory is to track the movement trajectory of the user and the object and provide continuous spatial information. The alarm trigger signal is to trigger an alarm and save key information when abnormal behavior is detected;

[0046] Continue to refer to Figure 1 After the manager in S5 of the present invention responds to and resolves the incident, the power swap cabinet will continue to record temperature data to ensure the continued safe use of the power swap cabinet.

[0047] like Figure 2 As shown, the present invention also includes a temperature monitoring and early warning step, which includes starting the power swap cabinet control sensor, the power swap cabinet performing real-time temperature detection through the sensor, and the power swap cabinet judging whether the temperature is normal through the real-time temperature detection data. If normal, the power swap cabinet controls the sensor to continue monitoring, and if abnormal, the power swap cabinet triggers an early warning and issues an early warning notification. The power swap cabinet sends the early warning notification to the management personnel, the management personnel responds and records the temperature data, and then the management personnel handles it.

[0048] The present invention also discloses a power exchange cabinet safety monitoring and alarm device, including a processor and a memory. The processor is used to execute the above-mentioned power exchange cabinet safety monitoring and alarm method by calling a program or instruction stored in the memory.

[0049] The present invention also discloses a readable storage medium, which stores a program or instruction, and the program or instruction executes the above-mentioned power exchange cabinet safety monitoring and alarm method.

[0050] The present invention includes a real-time monitoring and capture module, an AI "lock fraud" behavior recognition module, and a temperature monitoring and early warning module. The real-time monitoring and capture module uses a camera to capture dynamic images around the power exchange cabinet in real time, and uses computer vision technology to analyze the video stream. When the present invention recognizes abnormal behavior (such as abnormal interaction between people and objects, rapid movement, etc.), it automatically captures and records the key images and timestamps of abnormal events, which is convenient for subsequent tracking and investigation;

[0051] The AI ​​module for identifying "fraudulent lock" behavior analyzes historical video data to establish a baseline model of normal behavior and a characteristic model of "fraudulent lock" behavior. When the battery is returned in the battery swap cabinet, the module monitors the user's behavior in real time. If a behavior that deviates significantly from the normal mode is detected (such as hand movements after the door lock is closed), the present invention will automatically trigger an alarm and record relevant data for subsequent analysis.

[0052] The temperature monitoring and early warning module uses thermal imaging technology to monitor temperature changes inside and outside the battery swap cabinet. Especially in low-light environments, thermal imaging technology can effectively identify people or equipment that are abnormally close. The temperature sensor monitors the temperature of the battery in the cabinet in real time to ensure that the battery is within a safe range. If overheating or overcooling is detected, the present invention will automatically send an early warning notification to the management personnel so that potential safety risks can be dealt with in a timely manner.

[0053] In the present invention, all monitoring videos, sensor data and alarm records are stored in the cloud in real time to ensure the security and accessibility of the data and support subsequent analysis and backtracking. The present invention adopts data mining technology to analyze historical data, discover potential safety hazards and behavior patterns, and provide a basis for subsequent optimization. The early warning triggered by the power exchange cabinet of the present invention has a variety of alarm methods, including SMS, APP push, e-mail, etc., to ensure that managers can quickly receive alarm information under any circumstances. The present invention can also display monitoring screens, alarm records and equipment status in real time, support multi-level alarm information feedback, and facilitate managers to respond quickly and take measures.

[0054] A plurality of temperature sensors are installed inside the power exchange cabinet of the present invention to monitor the temperature inside the cabinet in real time. These sensors can quickly and accurately obtain the temperature data of the battery storage area or charging area, and transmit these data to the processing unit of the present invention. At the same time, the power exchange cabinet is also equipped with a thermal imaging camera, which generates images by capturing the infrared radiation released by the object. Each object emits a certain amount of infrared rays, and its intensity varies with temperature. The thermal imaging camera uses these infrared rays to identify the temperature distribution of the object and generates a thermal map of different colors, and the color corresponds to the high and low temperature. At night or in low light environments, traditional optical cameras are difficult to capture images clearly, while thermal imaging cameras can ignore light conditions and provide clear thermal maps by detecting the temperature changes of objects, thereby monitoring the temperature changes of the battery in the power exchange cabinet. When the temperature sensor detects that the temperature in the cabinet is abnormal (such as overheating or overcooling), the present invention will combine the picture of the thermal imaging camera to further analyze the source of the temperature abnormality. If the temperature abnormality occurs in the battery storage area, the present invention will focus on monitoring the changes in the thermal map near the battery to determine whether there is a potential danger. If the thermal imaging picture shows that the temperature of a certain area inside the power exchange cabinet suddenly rises, this may mean that the battery is overheated, on fire or in other dangerous situations. When the present invention confirms that the temperature is abnormal and meets the early warning conditions, the alarm information will be sent to relevant managers through various means (such as text messages, APP push, email, etc.). The notification content includes key information such as the specific location of the abnormal temperature, temperature change trends, real-time monitoring images, etc., so that managers can take quick action. In this entire monitoring present invention, the model will continuously optimize itself. The present invention will store the trigger records of each temperature anomaly, combine thermal imaging data and temperature sensor historical data, and conduct data mining and analysis. By analyzing these historical data, the present invention can further optimize the temperature anomaly recognition model, reduce false alarms and improve the accuracy of early warnings. The present invention has the advantages of high security, fast response speed, real-time and convenient temperature monitoring, all-round monitoring, intelligent control, high recognition accuracy, and improved user experience.

[0055] This embodiment does not impose any formal limitation on the shape, material, structure, etc. of the present invention. Any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention shall fall within the protection scope of the technical solution of the present invention.

Claims

1. A safety monitoring and alarm method for a power exchange cabinet, which is implemented based on a deep learning model and is characterized in that: The following steps are involved: S1, the battery swap cabinet controls the camera to start; S2, the battery swap cabinet captures the video stream through the camera in S1; S3: The battery swap cabinet performs behavior analysis and detection based on the deep learning model. If it is normal behavior, the battery swap control camera continues to monitor. If it is abnormal behavior, the battery swap cabinet triggers an early warning and enters S4. S4, the power swap cabinet records the event and its detailed information, and then the power swap cabinet captures it through a camera; S5. The battery swap cabinet uploads the event and detailed information in S4 to the cloud storage and notifies the management personnel, who respond and resolve the event; Among them, S3 uses a deep learning model to input real-time frame images from the camera, extracts the spatial features and temporal features of the image and the interaction features between people and objects in the image after preprocessing, and outputs behavior classification results, abnormal behavior types, target trajectories, and alarm trigger signal data features.

2. The method for safety monitoring and alarming of a power exchange cabinet according to claim 1 is characterized in that: The deep learning model includes a two-stage target detection algorithm R-CNN, a Pose Estimation model, and a behavior classification model, wherein the two-stage target detection algorithm R-CNN is used to detect and locate targets in images, including people and objects; The Pose Estimation model is used to analyze body posture; The behavior classification model is used to analyze behavior based on the time series features extracted by the convolutional neural network and combined with the time series data model.

3. The method for safety monitoring and alarming of a power exchange cabinet according to claim 2 is characterized in that: In the S3, a deep learning model is used to distinguish the image data of the images taken by the camera into daytime scenes and nighttime scenes, and then the image data is labeled to generate label data, which is marked as normal behavior and abnormal behavior, and then the label data is used for data enhancement training.

4. The method for safety monitoring and alarming of a power exchange cabinet according to claim 3 is characterized in that: In S3, the label data is used to train the two-stage target detection algorithm R-CNN to learn the location information of the target, and the behavior classification model is trained by combining the action features and posture features with the labeled time series behavior data.

5. The method for safety monitoring and alarming of a power exchange cabinet according to claim 1 is characterized in that: It also includes a temperature monitoring and early warning step, which includes starting the power swap cabinet control sensor, the power swap cabinet performing real-time temperature detection through the sensor, and the power swap cabinet judging whether the temperature is normal through the real-time temperature detection data. If it is normal, the power swap cabinet controls the sensor to continue monitoring. If it is abnormal, the power swap cabinet triggers an early warning and issues an early warning notification. The power swap cabinet sends the early warning notification to the management personnel, the management personnel responds and records the temperature data, and then the management personnel handles it.

6. The battery exchange cabinet safety monitoring and alarm device is characterized by: It includes a processor and a memory, and the processor is used to execute the power exchange cabinet safety monitoring and alarm method as described in any one of claims 1 to 5 by calling the program or instructions stored in the memory.

7. A readable storage medium, characterized in that: The storage medium stores a program or instruction, and the program or instruction executes the power exchange cabinet safety monitoring and alarm method as described in any one of claims 1 to 5.

Images