Smart battery swap station, control method, equipment and medium compatible with multiple battery packs

Abstract

The present invention relates to the technical field of battery replacement for electric vehicles, and in particular to an intelligent battery replacement station, control method, equipment and medium compatible with multiple battery packs, the method comprising: obtaining vehicle information of a vehicle to be replaced, and retrieving the corresponding battery pack keying and unlocking hole model type and its specific parameters from a database; when the battery replacement vehicle stops at a battery replacement operation platform, obtaining an image of the vehicle chassis and determining the initial posture of the vehicle chassis; processing the initial posture, extracting the center coordinates of the keying and unlocking hole and the normal vector direction of the keying and unlocking hole; comparing the center coordinates of the keying and unlocking hole and the normal vector direction of the keying and unlocking hole with the battery pack keying and unlocking hole model type and its specific parameters retrieved from a database to determine whether the information is consistent; if not, prompting a user to leave the battery replacement operation platform; if consistent, the battery replacement AGV selects the corresponding keying and unlocking operation mode, retrieves the battery in the library, and replaces the vehicle battery; after the battery replacement is completed, a battery replacement report is generated, allowing the user to leave.

Description

Technical Field

[0001] The present invention relates to the field of electric vehicle battery replacement technology, and in particular to an intelligent battery replacement station, control method, equipment and medium compatible with multiple battery packs. Background Art

[0002] In recent years, with the global emphasis on environmental protection and sustainable development, the promotion and use of new energy vehicles (NEVs), especially electric vehicles (EVs), has received strong support from the government. Policy promotion and market demand growth have promoted the development of charging and battery swapping infrastructure, including the demand for efficient and compatible battery swapping technology. However, the current battery swapping station technology is insufficient in terms of compatibility, mainly reflected in the support for different brands, different models and different types of battery packs. Battery swapping stations can often only serve specific vehicles and battery types, limiting their versatility and user range.

[0003] There are various ways to lock the battery pack, including snap-on, bolt-on, and spin-on, which increases the complexity of battery swapping. Different locking methods require specific locking and unlocking methods, and existing battery swapping technologies often cannot cope with this diversity well.

[0004] The information disclosed in this background technology section is only intended to deepen the understanding of the overall background technology of the present invention, and should not be regarded as acknowledging or suggesting in any form that the information constitutes the prior art already known to those skilled in the art. Summary of the invention

[0005] The present invention provides a smart battery swap station, control method, device and medium compatible with multiple battery packs, thereby effectively solving the problems in the background technology.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is: a control method of a smart battery swap station compatible with multiple battery packs, comprising the following steps:

[0007] Obtain vehicle information of the vehicle that needs battery replacement, and retrieve the corresponding battery pack unlocking hole model type and its specific parameters from the database;

[0008] When the battery swap vehicle stops at the battery swap operation platform, an image of the vehicle chassis is acquired to determine the initial position and posture of the vehicle chassis;

[0009] Processing the initial posture to extract the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole;

[0010] Compare the center coordinates of the keypad unlock hole and the normal vector direction of the keypad unlock hole with the keypad unlock hole model type and specific parameters of the battery pack retrieved from the database to determine whether the information is consistent;

[0011] If they are inconsistent, the user is prompted to leave the battery swap operation platform; if they are consistent, the battery swap AGV selects the corresponding locking and unlocking operation mode, retrieves the battery in the warehouse, and replaces the vehicle battery;

[0012] After the battery replacement is completed, a battery replacement report is generated, allowing the user to leave.

[0013] Furthermore, the obtaining of vehicle information of the vehicle requiring battery replacement comprises the following steps:

[0014] Install cameras at the entrance of the battery swap station to capture images of battery swap vehicles entering the station;

[0015] Resize and convert color spaces of images;

[0016] The Canny edge detection algorithm is used to extract the vehicle outline, and the Hough transform method is used to detect the geometric shape in the vehicle outline to obtain the point cloud data of the vehicle;

[0017] The Euclidean clustering algorithm is used to group different objects in the point cloud data, and the key feature points of the vehicle contour are extracted from the clustered point cloud. The image features are fused with the point cloud features to form a complete feature vector.

[0018] Load a pre-trained deep learning model, input the feature vector into the deep learning model, and output the classification result of the vehicle.

[0019] Furthermore, the step of acquiring an image of the vehicle chassis and determining an initial position and posture of the vehicle chassis comprises the following steps:

[0020] Capturing an original image of the vehicle chassis through a visual sensor, and cropping and scaling the captured original image to a set size;

[0021] A deep learning model is used to extract features. The deep learning model performs downsampling on the input image and predicts the position of the center point of the target bounding box and its offset to generate a feature map.

[0022] Predict the position of the target bounding box center point and its offset, as well as the width and height of the target bounding box based on the feature map;

[0023] Obtain the camera's intrinsic and extrinsic parameters, use warpAffine affine transformation to transform the predicted key point positions in the feature map into the spatial coordinates of the original image, and generate point cloud data of the vehicle chassis based on the results of 3D space restoration;

[0024] Perform voxel filtering on point cloud data to remove noise points;

[0025] Perform Euclidean clustering on the filtered point cloud, use principal component analysis (PCA) to analyze the point cloud data, find out the main direction of the point cloud data, and estimate the approximate direction of the vehicle chassis;

[0026] The random sampling consistency RANSAC algorithm is used to fit the point cloud data, exclude outliers, and obtain the plane model of the vehicle chassis;

[0027] Based on the results of PCA analysis and RANSAC fitting, the initial pose of the vehicle chassis is determined.

[0028] Furthermore, the processing of the initial posture to extract the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole includes:

[0029] The initial pose is preprocessed by denoising and contrast enhancement, and the original RGB image is converted into an HSV color space;

[0030] Set the threshold in HSV color space for segmentation and use Canny edge detection algorithm to identify the edges in the image;

[0031] Use opening and closing operations to remove small interfering spots or disconnected edges, use Hough circle transform to detect circular contours in the image, and preliminarily locate the location of the unlocking holes;

[0032] Set the detection parameters, obtain the center coordinates and radius of each detected circle, and use the least squares method to fit a plane model for each initially located unlocking hole;

[0033] In the plane model, an area around the center of the initially located unlocking hole is selected as the region of interest (ROI), and the plane equation of the point set in this area is fitted using the least squares method, and the center position is adjusted according to the plane fitting result;

[0034] Extract the normal vector from the plane equation obtained by least squares plane fitting, and confirm whether the normal vector direction correctly points to the opening direction of the locking and unlocking hole;

[0035] Output the final determined center coordinates of the unlocking hole and the normal vector direction of the unlocking hole.

[0036] Furthermore, the battery-swapping AGV selects a corresponding locking and unlocking operation mode, retrieves a battery from a storage, and replaces the vehicle battery, including:

[0037] Through the spatial inversion design, the battery-swapping AGV can switch between different locking and unlocking modes;

[0038] Use the height switching mechanism to automatically adjust the height of the locking and unlocking mechanism according to the height differences of the battery packs of different models;

[0039] During the locking and unlocking process, adaptive floating is used to make fine adjustments based on the changes in the position of the battery pack.

[0040] Furthermore, the battery-swapping AGV selects a corresponding locking and unlocking operation mode, retrieves a battery from a storage, and replaces the vehicle battery, and further includes:

[0041] The battery replacement operation is performed through the dual AGV collaborative control method, specifically including:

[0042] Connect the independently controlled battery swap AGV in the battery swap station through the communication protocol;

[0043] Time synchronization for battery-swapping AGVs: Use the NTP time synchronization protocol to ensure that the timestamps of all AGVs and station control systems are consistent;

[0044] Perform task planning based on task requirements, AGV current status, vehicle arrival order, and battery pack type, and dynamically assign tasks to each AGV;

[0045] Using a scheduling algorithm, for battery-swap vehicles, one AGV is designated to perform disassembly operations and another AGV to perform assembly operations at the same time.

[0046] Furthermore, the battery replacement operation is performed by the dual AGV collaborative control method, further comprising:

[0047] Autonomous fault monitoring and fault-tolerant control: When one of the AGVs fails, the other AGV switches to stand-alone mode to continue the battery replacement task and sends fault information.

[0048] Furthermore, the battery replacement operation is performed by the dual AGV collaborative control method, further comprising:

[0049] Through the Leader-Follower strategy, the dual AGVs maintain the spacing and angle deviation within the set range, use distance sensors and angle sensors to monitor and adjust the spacing in real time, detect the relative distance between the two AGVs in real time, and use a spacing control algorithm to ensure that the two AGVs maintain a safe distance when traveling on a monorail.

[0050] Furthermore, a method for controlling orderly charging of a battery pack is also provided, comprising the following steps:

[0051] Orderly charging strategy: Based on the improved cuckoo search algorithm CSA, it explores the solution space by random walk and selecting the optimal solution according to the battery SOC, battery health status and scheduled use time of the vehicle, and automatically sorts the battery charging priority;

[0052] Load forecasting and smoothing strategy: Continuously collect real-time data of each battery in the battery swap station, use the combined forecasting model of freshness function and cross entropy based on historical data, receive real-time load data of the power grid, combine its own load forecast, and pre-plan the allocation of charging power;

[0053] Battery pack optimization scheduling strategy: Based on the first-in-first-out (FIFO) principle, ensure that the batteries that enter the warehouse first are used first.

[0054] Furthermore, the orderly charging strategy also includes:

[0055] Monitor the grid status and charging progress in real time. When an abnormal increase in grid load is detected, adjust the charging plan and reduce the charging power until the grid load returns to normal.

[0056] Receive real-time electricity price information from the power grid. When it detects that the electricity price is at a low point, increase the charging power; conversely, reduce or stop charging when the electricity price is at a peak.

[0057] Furthermore, the load forecasting and smoothing strategy also includes:

[0058] A combined prediction model of freshness function and cross entropy based on historical data is used to analyze the impact of past charging patterns, weather conditions, and holiday effects on charging demand.

[0059] Through grey correlation analysis, we find historical days with similar meteorological conditions and day types as the forecast day, and use them as the forecast benchmark. We dynamically adjust the weight of a single forecast model based on the cross entropy algorithm.

[0060] Combine forecasting methods such as time series analysis and machine learning model forecast results to form a final forecast;

[0061] The charging plan for the next day is calculated through the particle swarm algorithm.

[0062] Furthermore, the battery pack optimization scheduling strategy also includes:

[0063] Collect key parameters of battery voltage, current, temperature, and SOC in real time to evaluate the health of the battery;

[0064] Use big data analysis and machine learning technology to predict battery life trends;

[0065] Set a threshold for the battery health status. Once the battery status is detected to be below this threshold, it will be immediately removed from the battery swap service for inspection or repair;

[0066] Maintain more than a set number of fully charged spare batteries to avoid overcharging or discharging a single battery by balancing the use of all batteries.

[0067] The present invention also includes a smart battery swap station compatible with multiple battery packs, comprising:

[0068] Communication-connected station control system, intelligent battery swapping equipment and intelligent battery management warehouse;

[0069] The intelligent battery exchange equipment includes a plurality of AGVs, and the AGVs include a plurality of locking and unlocking modules corresponding to the locking methods of different vehicle models and battery packs;

[0070] The intelligent battery management compartment stores a plurality of batteries and charges the stored batteries;

[0071] The station control system is used to obtain vehicle information of the vehicle that needs battery replacement, and retrieve the corresponding battery pack unlocking hole model type and its specific parameters from the database; when the battery replacement vehicle stops at the battery replacement operation platform, obtain the image of the vehicle chassis and determine the initial posture of the vehicle chassis; process the initial posture to extract the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole; compare the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole with the battery pack unlocking hole model type and its specific parameters retrieved from the database to determine whether the information is consistent; if inconsistent, prompt the user to leave the battery replacement operation platform;

[0072] If they are consistent, the intelligent battery swapping device controls the battery swapping AGV to select the corresponding locking and unlocking operation mode, retrieves the battery in the intelligent battery management compartment, and replaces the vehicle battery;

[0073] After the battery replacement is completed, the station control system generates a battery replacement report, allowing the user to leave.

[0074] Furthermore, the station control system comprises:

[0075] Technical architecture, which is used for the station control system to communicate in real time with intelligent battery swapping equipment, intelligent battery management warehouse and equipment and systems in the battery swapping station, perceive the whole process information of battery swapping, monitor the environmental information in the battery swapping station, and provide battery swapping monitoring data for the battery swapping station operation system and government management platform;

[0076] Network architecture, which uses TCP / IP network protocol to support access and collection of multiple devices. The station control system uses MySQL data storage and processing and has SQL query algorithm;

[0077] Business architecture, which is used to ensure the smoothness of the automated battery swapping process through battery swapping monitoring, auxiliary equipment monitoring to maintain the stability of the station environment and facilities, safety management to prevent and respond to potential risks, operation analysis to provide decision support, system management to coordinate equipment and user data, and service management to promote seamless connection with external platforms.

[0078] Furthermore, the intelligent battery swapping device further includes:

[0079] The flexible battery swap operation platform is used to automatically adjust the vehicle position, provide a fixed electric vehicle parking operation area, adapt to battery swap vehicles of various types of wheelbases and track widths, and automatically position the battery swap vehicles.

[0080] Furthermore, the AGV also includes a Mecanum wheel, a servo motor, a vision system, a laser radar and a sensor;

[0081] Also includes:

[0082] A spatial transfer mechanism, through which the AGV switches between different locking and unlocking modes;

[0083] Height switching mechanism, used to automatically adjust the height of the locking and unlocking mechanism according to the height differences of battery packs of different vehicle models;

[0084] The adaptive floating mechanism adaptively floats during the locking and unlocking process and performs fine adjustments according to the position changes of the battery pack.

[0085] Furthermore, the intelligent battery management compartment includes:

[0086] Dynamic charging scheduling unit: uses machine learning algorithms to predict battery demand and grid load and dynamically adjust charging plans;

[0087] Battery life optimization unit: used to select the most appropriate charging rate according to the battery type and current status, avoid overcharging or discharging, extend battery life, and formulate regular maintenance plans based on data analysis to ensure the long-term health of the battery pack;

[0088] Flexible storage and dispatching unit: used for linkage control with battery-swapping AGV to realize automatic storage and retrieval of battery packs, and automatically allocate battery packs to charging stations or storage areas according to battery status and charging requirements.

[0089] Furthermore, the intelligent battery management compartment also includes:

[0090] An ordered charging unit for automatically sorting the charging priority of the battery by exploring the solution space through random walks and selecting the optimal solution based on the improved cuckoo search algorithm CSA according to the battery SOC, battery health status and scheduled use time of the vehicle;

[0091] The load forecasting and smoothing unit is used to continuously collect the real-time data of each battery in the battery swap station, and use the combined forecasting model of the freshness function and cross entropy based on historical data to receive the real-time load data of the power grid, and combine its own load forecast to pre-plan the allocation of charging power;

[0092] The battery pack optimization scheduling unit is used to ensure that the batteries that enter the warehouse earliest are used first based on the first-in-first-out (FIFO) principle.

[0093] Furthermore, the orderly charging unit is also used for:

[0094] Monitor the grid status and charging progress in real time. When an abnormal increase in grid load is detected, adjust the charging plan and reduce the charging power until the grid load returns to normal.

[0095] Receive real-time electricity price information from the power grid. When it detects that the electricity price is at a low point, increase the charging power; conversely, reduce or stop charging when the electricity price is at a peak.

[0096] Furthermore, the load forecasting and smoothing unit is also used for:

[0097] A combined prediction model of freshness function and cross entropy based on historical data is used to analyze the impact of past charging patterns, weather conditions, and holiday effects on charging demand.

[0098] Through grey correlation analysis, we find historical days with similar meteorological conditions and day types as the forecast day, and use them as the forecast benchmark. We dynamically adjust the weight of a single forecast model based on the cross entropy algorithm.

[0099] Combine forecasting methods such as time series analysis and machine learning model forecast results to form a final forecast;

[0100] The charging plan for the next day is calculated through the particle swarm algorithm.

[0101] Furthermore, the battery pack optimization scheduling unit is also used for:

[0102] Collect key parameters of battery voltage, current, temperature, and SOC in real time to evaluate the health of the battery;

[0103] Use big data analysis and machine learning technology to predict battery life trends;

[0104] Set a threshold for the battery health status. Once the battery status is detected to be below this threshold, it will be immediately removed from the battery swap service for inspection or repair;

[0105] Maintain more than a set number of fully charged spare batteries to avoid overcharging or discharging a single battery by balancing the use of all batteries.

[0106] Furthermore, the station control system also includes a vehicle information identification unit, which is used to:

[0107] Install cameras at the entrance of the battery swap station to capture images of battery swap vehicles entering the station;

[0108] Resize and convert color spaces of images;

[0109] The Canny edge detection algorithm is used to extract the vehicle outline, and the Hough transform method is used to detect the geometric shape in the vehicle outline to obtain the point cloud data of the vehicle;

[0110] The Euclidean clustering algorithm is used to group different objects in the point cloud data, and the key feature points of the vehicle contour are extracted from the clustered point cloud. The image features are fused with the point cloud features to form a complete feature vector.

[0111] Load a pre-trained deep learning model, input the feature vector into the deep learning model, and output the classification result of the vehicle.

[0112] Furthermore, the station control system further comprises an initial posture recognition unit, which is used to:

[0113] Capturing an original image of the vehicle chassis through a visual sensor, and cropping and scaling the captured original image to a set size;

[0114] A deep learning model is used to extract features. The deep learning model performs downsampling on the input image and predicts the position of the center point of the target bounding box and its offset to generate a feature map.

[0115] Predict the position of the target bounding box center point and its offset, as well as the width and height of the target bounding box based on the feature map;

[0116] Obtain the camera's intrinsic and extrinsic parameters, use warpAffine affine transformation to transform the predicted key point positions in the feature map into the spatial coordinates of the original image, and generate point cloud data of the vehicle chassis based on the results of 3D space restoration;

[0117] Perform voxel filtering on point cloud data to remove noise points;

[0118] Perform Euclidean clustering on the filtered point cloud, use principal component analysis (PCA) to analyze the point cloud data, find out the main direction of the point cloud data, and estimate the approximate direction of the vehicle chassis;

[0119] The random sampling consistency RANSAC algorithm is used to fit the point cloud data, exclude outliers, and obtain the plane model of the vehicle chassis;

[0120] Based on the results of PCA analysis and RANSAC fitting, the initial pose of the vehicle chassis is determined.

[0121] Furthermore, the station control system further comprises an unlocking hole identification unit, which is used to:

[0122] The initial pose is preprocessed by denoising and contrast enhancement, and the original RGB image is converted into an HSV color space;

[0123] Set the threshold in HSV color space for segmentation and use Canny edge detection algorithm to identify the edges in the image;

[0124] Use opening and closing operations to remove small interfering spots or disconnected edges, use Hough circle transform to detect circular contours in the image, and preliminarily locate the location of the unlocking holes;

[0125] Set the detection parameters, obtain the center coordinates and radius of each detected circle, and use the least squares method to fit a plane model for each initially located unlocking hole;

[0126] In the plane model, a smaller area around the center of the initially located unlocking hole is selected as the region of interest (ROI). The plane equation of the point set in this area is fitted using the least squares method, and the center position is adjusted according to the plane fitting result.

[0127] Extract the normal vector from the plane equation obtained by least squares plane fitting, and confirm whether the normal vector direction correctly points to the opening direction of the locking and unlocking hole;

[0128] Output the final determined center coordinates of the unlocking hole and the normal vector direction of the unlocking hole.

[0129] Furthermore, the station control system also includes an AGV control unit, which is used to:

[0130] Connect the independently controlled battery swap AGV in the battery swap station through the communication protocol;

[0131] Time synchronization for battery-swapping AGVs: Use the NTP time synchronization protocol to ensure that the timestamps of all AGVs and station control systems are consistent;

[0132] Perform task planning based on task requirements, AGV current status, vehicle arrival order, and battery pack type, and dynamically assign tasks to each AGV;

[0133] Using a scheduling algorithm, for battery-swap vehicles, one AGV is designated to perform disassembly operations and another AGV to perform assembly operations at the same time.

[0134] Furthermore, the station control system further includes a fault processing unit, which is used to:

[0135] When one of the AGVs fails, the other AGV switches to stand-alone mode to continue the battery replacement task and sends a fault message.

[0136] Furthermore, the station control system further includes a distance control unit, which is used to:

[0137] Through the Leader-Follower strategy, the dual AGVs maintain the spacing and angle deviation within the set range, use distance sensors and angle sensors to monitor and adjust the spacing in real time, detect the relative distance between the two AGVs in real time, and use a spacing control algorithm to ensure that the two AGVs maintain a safe distance when traveling on a monorail.

[0138] Furthermore, the AGV further includes a mode switching unit, which is used to:

[0139] According to the different stages of the battery replacement task, the AGV's disassembly mode, transfer mode, and assembly mode are automatically or manually switched.

[0140] The present invention also includes a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method described above is implemented.

[0141] The present invention also includes a storage medium on which a computer program is stored. When the computer program is executed by a processor, the method described above is implemented.

[0142] The beneficial effects of the present invention are as follows: the present invention uses image processing technology to automatically identify the position of the unlocking holes of the vehicle chassis, combines the parameter information in the database to ensure compatibility, and selects a suitable unlocking operation mode. This battery replacement control method based on intelligent recognition and automated operation can effectively cope with a variety of battery packs and vehicle types, and improve the versatility and efficiency of battery replacement stations. Multiple checks are performed before the unlocking operation to ensure the safety and reliability of the operation and avoid operational errors caused by inconsistent information or misjudgment. BRIEF DESCRIPTION OF THE DRAWINGS

[0143] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0144] Figure 1 is a flow chart of the method in Example 1;

[0145] Figure 2 This is a schematic diagram of the structure of the battery swap station in Example 1;

[0146] Figure 3 This is a topology diagram of a smart battery swap station for passenger cars compatible with multiple locking mode battery packs in Example 2;

[0147] Figure 4This is a schematic diagram of the control topology of a smart battery swap station for passenger vehicles compatible with battery packs with multiple locking modes in Example 2;

[0148] Figure 5 This is a flow chart of battery replacement at a smart battery replacement station for a passenger car compatible with a battery pack with multiple locking modes in Example 2;

[0149] Figure 6 This is the overall system architecture diagram of the station control system in Example 2;

[0150] Figure 7 This is a flow chart of intelligent matching and adaptive locking and unlocking control in Example 2;

[0151] Figure 8 A schematic diagram of the structure of a computer device. DETAILED DESCRIPTION

[0152] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0153] Embodiment 1:

[0154] like Figure 1 As shown: A control method for a smart battery swap station compatible with multiple battery packs includes the following steps:

[0155] Obtain vehicle information of the vehicle that needs battery replacement, and retrieve the corresponding battery pack unlocking hole model type and its specific parameters from the database;

[0156] When the battery swap vehicle stops at the battery swap operation platform, an image of the vehicle chassis is acquired to determine the initial position and posture of the vehicle chassis;

[0157] Process the initial pose to extract the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole;

[0158] Compare the center coordinates of the keypad unlock hole and the normal vector direction of the keypad unlock hole with the keypad unlock hole model type and specific parameters of the battery pack retrieved from the database to determine whether the information is consistent;

[0159] If they are inconsistent, the user is prompted to leave the battery swap operation platform; if they are consistent, the battery swap automatic guided vehicle (AGV) selects the corresponding locking and unlocking operation mode, retrieves the battery from the warehouse, and replaces the vehicle battery;

[0160] After the battery replacement is completed, a battery replacement report is generated, allowing the user to leave.

[0161] Image processing technology is used to automatically identify the unlocking and unlocking hole positions of the vehicle chassis, combined with parameter information in the database to ensure compatibility and select the appropriate unlocking and unlocking operation method. This battery swap control method based on intelligent recognition and automated operation can effectively cope with a variety of battery packs and vehicle types, improving the versatility and efficiency of battery swap stations. Multiple checks are performed before unlocking and unlocking operations to ensure safe and reliable operations and avoid operational errors caused by inconsistent information or misjudgment.

[0162] In this embodiment, obtaining vehicle information of a vehicle that needs battery replacement includes the following steps:

[0163] Install cameras at the entrance of the battery swap station to capture images of battery swap vehicles entering the station;

[0164] Resize and convert color spaces of images;

[0165] The Canny edge detection algorithm is used to extract the vehicle outline, and the Hough transform method is used to detect the geometric shape in the vehicle outline to obtain the point cloud data of the vehicle;

[0166] The Euclidean clustering algorithm is used to group different objects in the point cloud data, and the key feature points of the vehicle contour are extracted from the clustered point cloud. The image features are fused with the point cloud features to form a complete feature vector.

[0167] Load the pre-trained deep learning model, input the feature vector into the deep learning model, and output the vehicle classification result.

[0168] By combining image processing, point cloud data processing and deep learning technology, efficient acquisition of vehicle information is achieved. This method uses multiple technical means to ensure the accuracy of vehicle classification, especially in environments with diverse vehicle types, which has great advantages. Canny edge detection and Hough transform can effectively identify vehicle contours and geometric features, and the fusion of point cloud data and image data enhances the accuracy of classification through the formation of feature vectors.

[0169] The steps of obtaining an image of the vehicle chassis and determining the initial position and posture of the vehicle chassis include:

[0170] Capturing an original image of the vehicle chassis through a visual sensor, and cropping and scaling the captured original image to a set size;

[0171] The deep learning model is used for feature extraction. The deep learning model downsamples the input image and predicts the position of the target bounding box center point and its offset to generate a feature map.

[0172] Predict the position of the target bounding box center point and its offset, as well as the width and height of the target bounding box based on the feature map;

[0173] Obtain the camera's intrinsic and extrinsic parameters, use warpAffine affine transformation to transform the predicted key point positions in the feature map into the spatial coordinates of the original image, and generate point cloud data of the vehicle chassis based on the results of 3D space restoration;

[0174] Perform voxel filtering on point cloud data to remove noise points;

[0175] Perform Euclidean clustering on the filtered point cloud, use principal component analysis (PCA) to analyze the point cloud data, find out the main direction of the point cloud data, and estimate the approximate direction of the vehicle chassis;

[0176] The random sampling consistency RANSAC algorithm is used to fit the point cloud data, exclude outliers, and obtain the plane model of the vehicle chassis;

[0177] Based on the results of PCA analysis and RANSAC fitting, the initial pose of the vehicle chassis is determined.

[0178] Image cropping, scaling, and feature extraction steps ensure that the model can focus on key chassis areas and improve the accuracy of subsequent steps. The point cloud data generated based on the image is optimized through voxel filtering and Euclidean clustering, which effectively removes noise and extracts useful feature points. PCA analysis combined with the RANSAC fitting method ensures accurate posture estimation of the chassis and adapts to the diversity of chassis of different vehicles, thereby providing high-precision posture estimation for battery swap stations compatible with multiple models. Through multi-step data processing and analysis, the three-dimensional spatial position and posture of the chassis can be accurately reconstructed, providing reliable input for subsequent battery swap operations.

[0179] As a preferred embodiment of the above, the initial posture is processed to extract the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole, including:

[0180] The initial pose is preprocessed by denoising and contrast enhancement, and the original RGB image is converted to HSV color space;

[0181] Set the threshold in HSV color space for segmentation and use Canny edge detection algorithm to identify the edges in the image;

[0182] Use opening and closing operations to remove small interfering spots or disconnected edges, use Hough circle transform to detect circular contours in the image, and preliminarily locate the location of the unlocking holes;

[0183] Set the detection parameters, obtain the center coordinates and radius of each detected circle, and use the least squares method to fit a plane model for each initially located unlocking hole;

[0184] In the plane model, a smaller area around the center of the initially located unlocking hole is selected as the region of interest (ROI). The plane equation of the point set in this area is fitted using the least squares method, and the center position is adjusted according to the plane fitting result.

[0185] Extract the normal vector from the plane equation obtained by least squares plane fitting, and confirm whether the normal vector direction correctly points to the opening direction of the locking and unlocking hole;

[0186] Output the final determined center coordinates of the unlocking hole and the normal vector direction of the unlocking hole.

[0187] Denoising, contrast enhancement and color space conversion improve the visibility of the keyhole and the clarity of the edge. HSV color space segmentation can effectively distinguish the target area from the background and reduce false detection. Canny edge detection identifies the edges in the image, while opening and closing operations help eliminate small interference points and repair discontinuous edges, improving the integrity of edge detection. The position of the keyhole is preliminarily located by detecting the circular contour. This method is very effective for detecting circular structures. The center area of ​​the circle is fitted with a plane through the least squares method, and the center position is adjusted on this basis to ensure positioning accuracy. The normal vector extracted by the fitted plane can accurately determine the direction of the keyhole.

[0188] The accuracy of the extracted locking and unlocking hole center coordinates and normal vector direction is ensured, thereby providing reliable geometric information for subsequent battery replacement operations, especially in scenarios involving different battery pack designs, which helps to improve the versatility and compatibility of the system.

[0189] In this embodiment, the battery-swapping AGV selects the corresponding locking and unlocking operation mode, retrieves the battery in the library, and replaces the vehicle battery, including:

[0190] Through the spatial transposition design, the battery-swapping AGV can switch between different locking and unlocking methods;

[0191] Use the height switching mechanism to automatically adjust the height of the locking and unlocking mechanism according to the height differences of the battery packs of different models;

[0192] During the locking and unlocking process, adaptive floating is used to make fine adjustments based on the changes in the position of the battery pack.

[0193] As a preferred embodiment of the above, the battery-swapping AGV selects the corresponding locking and unlocking operation mode, retrieves the battery in the library, and replaces the vehicle battery, further comprising:

[0194] The battery replacement operation is performed through the dual AGV collaborative control method, specifically including:

[0195] Connect the independently controlled battery swap AGV in the battery swap station through the communication protocol;

[0196] Time synchronization for battery-swapping AGVs: Use the NTP time synchronization protocol to ensure that the timestamps of all AGVs and station control systems are consistent;

[0197] Perform task planning based on task requirements, AGV current status, vehicle arrival order, and battery pack type, and dynamically assign tasks to each AGV;

[0198] Using a scheduling algorithm, for battery-swap vehicles, one AGV is designated to perform disassembly operations and another AGV to perform assembly operations at the same time.

[0199] By introducing a dual AGV system, which is responsible for disassembly and assembly operations respectively, the efficiency of battery replacement is greatly improved. The dual AGV collaboration avoids the time waste of a single AGV in assembling after disassembly, and realizes parallel operation. The use of the NTP time synchronization protocol ensures the time consistency of the AGV and the station control system, which can avoid communication and operation delays between different AGVs. This is a key guarantee for the collaborative work of multiple AGV systems.

[0200] Task planning based on task requirements, AGV status and vehicle arrival order enables more flexible resource scheduling. Dynamic task allocation ensures that each AGV operates optimally according to current conditions, reducing idle time. In battery swap operations, AGVs for disassembly and assembly operations are arranged for specific tasks to ensure the orderliness and maximum efficiency of task allocation. The scheduling algorithm can be dynamically adjusted according to the load condition of the AGV and the urgency of the task to improve the overall work efficiency of the system. Through dual AGV collaboration and an efficient scheduling system, fast and accurate battery replacement operations are achieved, which can significantly shorten the service time of the battery swap station and improve the user experience.

[0201] The battery replacement operation is performed by using a dual AGV collaborative control method, and further includes:

[0202] Autonomous fault monitoring and fault-tolerant control: When one of the AGVs fails, the other AGV switches to stand-alone mode to continue the battery replacement task and sends fault information.

[0203] In this embodiment, the battery replacement operation is performed by the dual AGV collaborative control method, and further includes:

[0204] Through the Leader-Follower strategy, the dual AGVs maintain the spacing and angle deviation within the set range, use distance sensors and angle sensors to monitor and adjust the spacing in real time, detect the relative distance between the two AGVs in real time, and use a spacing control algorithm to ensure that the two AGVs maintain a safe distance when traveling on a monorail.

[0205] By setting one AGV as the leader and the other AGV as the follower, the follower makes corresponding adjustments based on the real-time position and motion status of the leader. This control mode is suitable for the collaborative operation of the dual AGV system, ensuring that the two AGVs can coordinate and perform power exchange tasks on the same track. The distance sensor and angle sensor are used to monitor the relative distance and angle deviation between the AGVs in real time, and the control system is used to make dynamic adjustments to ensure that the set safety distance and reasonable angle deviation are always maintained between the two AGVs. This improves the operational safety and accuracy of the dual AGVs in narrow spaces.

[0206] Through the set spacing control algorithm, the system can dynamically adjust the relative positions of the two AGVs according to the sensor data during the AGV driving process, ensuring that they can operate smoothly in the monorail system and avoid collisions or other interference problems. In the monorail design, space is limited and the two AGVs must maintain a safe distance with precise control. The combination of the leader-follower strategy and the spacing control algorithm ensures the reliability of the system and avoids possible congestion or safety issues.

[0207] As a preferred embodiment of the above, a method for controlling orderly charging of a battery pack is also included, comprising the following steps:

[0208] Orderly charging strategy: Based on the improved cuckoo search algorithm CSA, it explores the solution space by random walk and selecting the optimal solution according to the battery SOC, battery health status and scheduled use time of the vehicle, and automatically sorts the battery charging priority;

[0209] Load forecasting and smoothing strategy: Continuously collect real-time data of each battery in the battery swap station, use the combined forecasting model of freshness function and cross entropy based on historical data, receive real-time load data of the power grid, combine its own load forecast, and pre-plan the allocation of charging power;

[0210] Battery pack optimization scheduling strategy: Based on the first-in-first-out (FIFO) principle, ensure that the batteries that enter the warehouse first are used first.

[0211] The improved Cuckoo Search Algorithm (CSA) optimizes the battery charging order by random walk and selecting the optimal solution based on the battery status (SOC, health status) and the scheduled use time of the vehicle. This strategy uses the global search capability of CSA to automatically sort the charging priority of the battery under complex charging requirements. The improved version of CSA helps to speed up convergence and avoid local optimal problems, thereby improving battery management efficiency and ensuring that those batteries with low SOC, poor health status or priority use can be charged first.

[0212] Combined prediction model of freshness function and cross entropy: Load forecasting is performed by collecting real-time data of all batteries in the battery swap station (such as battery SOC, voltage, temperature, etc.) and combining it with historical load data. This model can improve the prediction accuracy, especially for effective estimation of short-term load fluctuations. Combining the real-time load situation of the power grid and its own predicted data, the charging power distribution of the battery swap station is dynamically adjusted to avoid excessive charging load during peak hours and effectively utilize power resources during off-peak hours. Through this strategy, the impact on the power grid can be reduced and a more stable and efficient charging process can be achieved.

[0213] First-in-first-out (FIFO) principle: ensures that the batteries in the warehouse are used in the order they enter. This simple and effective scheduling strategy avoids the performance degradation that may be caused by long-term idleness of batteries, while also ensuring the orderly use of batteries. Combined with the orderly charging strategy, the FIFO principle can further optimize the charging and use cycle to ensure that each battery operates within a healthy working cycle.

[0214] Among them, the orderly charging strategy also includes:

[0215] Monitor the grid status and charging progress in real time. When an abnormal increase in grid load is detected, adjust the charging plan and reduce the charging power until the grid load returns to normal.

[0216] Receive real-time electricity price information from the power grid. When it detects that the electricity price is at a low point, increase the charging power; conversely, reduce or stop charging when the electricity price is at a peak.

[0217] The system continuously monitors the load status of the power grid. When the load on the power grid increases abnormally (for example, during peak hours or when there is a sudden surge in power demand), the system will automatically adjust the charging plan of the battery swap station. When the load on the power grid is too high, the system will reduce the charging power or even suspend the charging of some batteries to help the power grid return to normal load levels. After the load on the power grid returns to normal, the system will gradually restore the charging power to ensure the continuity of the battery charging task.

[0218] The system receives and analyzes real-time electricity price information released by the power grid. When electricity prices are at a low point (such as late at night or during periods of low electricity demand), the system will increase the charging power appropriately to use the low-priced electricity from the power grid for charging. This not only reduces charging costs, but also improves battery charging efficiency. During peak electricity price periods (such as during the day or during periods of high electricity demand), the system will reduce charging power based on real-time electricity price information, and even suspend charging when necessary, to avoid large-scale charging during peak electricity price periods, thereby reducing operating costs.

[0219] Load forecasting and smoothing strategies also include:

[0220] A combined prediction model of freshness function and cross entropy based on historical data is used to analyze the impact of past charging patterns, weather conditions, and holiday effects on charging demand.

[0221] Through grey correlation analysis, we find historical days with similar meteorological conditions and day types as the forecast day, and use them as the forecast benchmark. We dynamically adjust the weight of a single forecast model based on the cross entropy algorithm.

[0222] Combine forecasting methods such as time series analysis and machine learning model forecast results to form a final forecast;

[0223] The charging plan for the next day is calculated through the particle swarm algorithm.

[0224] According to the time decay law of historical data, the freshness function is applied to weight the data of historical charging modes to ensure that the latest charging data has a greater impact on the prediction results, and the impact of older data in the past gradually decreases. The cross-entropy method is used to evaluate the reference value of historical data for current predictions, quantify the differences between the prediction results of multiple prediction models and the actual results, and dynamically adjust the weights of different prediction models according to the cross-entropy results to make the overall prediction more accurate.

[0225] Using the grey correlation analysis method, we find historical days with similar meteorological conditions and day types (such as working days, weekends or holidays) to the forecast day as the forecast benchmark. This analysis evaluates the correlation between the characteristics of historical days and forecast days, selects the closest historical data, and thus improves the accuracy of the forecast. Combined with the cross entropy algorithm, based on the historical data of similar days, we continuously adjust the weight of a single forecast model and dynamically optimize the forecast results. This process ensures that the model can adapt to different changes in conditions and provide more reliable forecasts.

[0226] Use time series analysis forecasting models (such as ARIMA or LSTM) to analyze trends and periodicity of past charging demand and capture long-term and short-term patterns. Apply machine learning models (such as random forests or neural networks) for nonlinear forecasting, considering the potential impact of multiple factors such as weather, holidays, and social events on future charging demand. Combine the forecast results of time series analysis and machine learning models to generate the final load forecast. The combination of multiple models can balance the advantages and disadvantages of different forecasting methods and improve the stability and accuracy of the overall forecast.

[0227] Based on the results of load forecasting, the particle swarm optimization algorithm (PSO) is used to calculate and adjust the charging plan for the next day. PSO seeks the global optimal solution by simulating the foraging process of bird flocks in nature, and is suitable for multi-objective optimization problems. The particle swarm algorithm calculates the optimal charging power allocation plan based on the results of load forecasting to ensure that the charging tasks are reasonably arranged within the allowable range of the power grid load and avoid excessive load during peak hours.

[0228] By combining multiple models and methods, the prediction results have higher accuracy, especially when facing complex environmental changes, the prediction strategy can be better adjusted. Through grey correlation analysis and cross entropy weight adjustment, the prediction model can flexibly respond to the influence of factors such as weather changes and holiday effects. The particle swarm algorithm ensures the global optimal solution of the charging plan, achieves efficient charging power distribution, and avoids grid overload.

[0229] The battery pack optimization scheduling strategy also includes:

[0230] Collect key parameters of battery voltage, current, temperature, and SOC in real time to evaluate the health of the battery;

[0231] Use big data analysis and machine learning technology to predict battery life trends;

[0232] Set a threshold for the battery health status. Once the battery status is detected to be below this threshold, it will be immediately removed from the battery swap service for inspection or repair;

[0233] Maintain more than a set number of fully charged spare batteries to avoid overcharging or discharging a single battery by balancing the use of all batteries.

[0234] Monitor the battery voltage, current, temperature and state of charge (SOC) in real time. These key parameters can help evaluate the current health status of the battery and provide real-time feedback during the charging process. By analyzing these key parameters, the health status of the battery can be evaluated to ensure that the battery is operating in good working conditions. Use big data technology to collect historical usage data and real-time monitoring data of the battery. By analyzing factors such as the battery's charge and discharge mode and working environment, a battery life prediction model is constructed, and machine learning algorithms (such as regression analysis, support vector machines, or deep learning models) are applied to predict the battery life trend. The model can identify early signs of battery aging and predict future performance changes.

[0235] Set a threshold for the battery health status, such as the battery's minimum voltage, maximum temperature, or minimum SOC. Once the battery status is detected to be below the set threshold, the system will automatically remove the battery from the battery swap service. Batteries below the threshold will be inspected and repaired to ensure that only batteries in good health are used for the battery swap service. This reduces the impact of faulty batteries on the battery swap service and improves the reliability of the system.

[0236] Maintaining more than a set number of fully charged spare batteries ensures that faulty batteries can be replaced or recharged immediately when needed. By using all batteries evenly, overcharging or discharging of a single battery is avoided. Balanced use can extend the battery life and reduce uneven battery performance degradation.

[0237] By real-time monitoring and prediction of battery health, battery problems can be promptly handled, the effective use time of the battery can be extended, and the replacement frequency can be reduced. Only healthy batteries are ensured to be used for battery replacement services, improving the reliability of battery replacement and user satisfaction. Through big data analysis and machine learning prediction, battery maintenance and management are optimized, reducing operational interruptions caused by battery failures, thereby saving operating costs. Setting health status thresholds and real-time monitoring can promptly detect and handle potential battery safety issues, ensure that the battery operates within a safe range, and reduce safety hazards.

[0238] like Figure 2 As shown, this embodiment also includes a smart battery swap station compatible with multiple battery packs, including:

[0239] Communication-connected station control system, intelligent battery swapping equipment and intelligent battery management warehouse;

[0240] The intelligent battery exchange equipment includes several AGVs, which include several locking and unlocking modules corresponding to the locking methods of different vehicle models and battery packs;

[0241] The intelligent battery management compartment stores a number of batteries and charges the stored batteries;

[0242] The station control system is used to obtain the vehicle information of the vehicle that needs to be replaced, and retrieve the corresponding battery pack unlocking hole model type and its specific parameters from the database; when the battery replacement vehicle stops at the battery replacement operation platform, obtain the image of the vehicle chassis and determine the initial posture of the vehicle chassis; process the initial posture, extract the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole; compare the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole with the battery pack unlocking hole model type and its specific parameters retrieved from the database to determine whether the information is consistent; if inconsistent, prompt the user to leave the battery replacement operation platform;

[0243] If they match, the smart battery swapping device controls the battery swapping AGV to select the corresponding unlocking and unlocking operation mode, retrieves the battery in the smart battery management compartment, and replaces the vehicle battery;

[0244] After the battery replacement is completed, the station control system generates a battery replacement report, allowing the user to leave.

[0245] The station control system includes:

[0246] Technical architecture: The technical architecture is used for real-time communication between the station control system and the intelligent battery swapping equipment, the intelligent battery management warehouse, and the equipment and systems in the battery swapping station, to perceive the whole process information of the battery swapping, monitor the environmental information in the battery swapping station, and provide battery swapping monitoring data for the battery swapping station operation system and the government management platform;

[0247] Network architecture: The network architecture adopts TCP / IP network protocol, supports the access and collection of multiple devices, and the station control system adopts MySQL data storage and processing, and has SQL query algorithm;

[0248] Business architecture: Business architecture is used to ensure the smoothness of the automated battery swapping process through battery swapping monitoring, auxiliary equipment monitoring to maintain the stability of the station environment and facilities, safety management to prevent and respond to potential risks, operation analysis to provide decision support, system management to coordinate equipment and user data, and service management to promote seamless integration with external platforms.

[0249] Among them, the intelligent battery replacement equipment also includes:

[0250] The flexible battery swap operation platform is used to automatically adjust the vehicle position, provide a fixed electric vehicle parking operation area, adapt to battery swap vehicles of various types of wheelbases and track widths, and automatically position the battery swap vehicles.

[0251] As a preferred embodiment of the above, the AGV further includes a Mecanum wheel, a servo motor, a vision system, a laser radar and a sensor;

[0252] Also includes:

[0253] Spatial transfer mechanism: AGV uses the spatial transfer mechanism to switch between different locking and unlocking methods;

[0254] Height switching mechanism, used to automatically adjust the height of the locking and unlocking mechanism according to the height differences of battery packs of different vehicle models;

[0255] Adaptive floating mechanism: During the locking and unlocking process, the adaptive floating mechanism floats adaptively and makes fine adjustments according to the position changes of the battery pack.

[0256] The smart battery management compartment includes:

[0257] Dynamic charging scheduling unit: uses machine learning algorithms to predict battery demand and grid load and dynamically adjust charging plans;

[0258] Battery life optimization unit: used to select the most appropriate charging rate according to the battery type and current status, avoid overcharging or discharging, extend battery life, and formulate regular maintenance plans based on data analysis to ensure the long-term health of the battery pack;

[0259] Flexible storage and dispatching unit: used for linkage control with battery-swapping AGV to realize automatic storage and retrieval of battery packs, and automatically allocate battery packs to charging stations or storage areas according to battery status and charging requirements.

[0260] Among them, the intelligent battery management warehouse also includes:

[0261] An ordered charging unit for automatically sorting the charging priority of the battery by exploring the solution space through random walks and selecting the optimal solution based on the improved cuckoo search algorithm CSA according to the battery SOC, battery health status and scheduled use time of the vehicle;

[0262] The load forecasting and smoothing unit is used to continuously collect the real-time data of each battery in the battery swap station, and use the combined forecasting model of the freshness function and cross entropy based on historical data to receive the real-time load data of the power grid, and combine its own load forecast to pre-plan the allocation of charging power;

[0263] The battery pack optimization scheduling unit is used to ensure that the batteries that enter the warehouse earliest are used first based on the first-in-first-out (FIFO) principle.

[0264] In this embodiment, the orderly charging unit is further used for:

[0265] Monitor the grid status and charging progress in real time. When an abnormal increase in grid load is detected, adjust the charging plan and reduce the charging power until the grid load returns to normal.

[0266] Receive real-time electricity price information from the power grid. When it detects that the electricity price is at a low point, increase the charging power; conversely, reduce or stop charging when the electricity price is at a peak.

[0267] As a preferred embodiment of the above, the load forecasting and smoothing unit is further used for:

[0268] A combined prediction model of freshness function and cross entropy based on historical data is used to analyze the impact of past charging patterns, weather conditions, and holiday effects on charging demand.

[0269] Through grey correlation analysis, we find historical days with similar meteorological conditions and day types as the forecast day, and use them as the forecast benchmark. We dynamically adjust the weight of a single forecast model based on the cross entropy algorithm.

[0270] Combine forecasting methods such as time series analysis and machine learning model forecast results to form a final forecast;

[0271] The charging plan for the next day is calculated through the particle swarm algorithm.

[0272] The battery pack optimization scheduling unit is also used to:

[0273] Collect key parameters of battery voltage, current, temperature, and SOC in real time to evaluate the health of the battery;

[0274] Use big data analysis and machine learning technology to predict battery life trends;

[0275] Set a threshold for the battery health status. Once the battery status is detected to be below this threshold, it will be immediately removed from the battery swap service for inspection or repair;

[0276] Maintain more than a set number of fully charged spare batteries to avoid overcharging or discharging a single battery by balancing the use of all batteries.

[0277] In this embodiment, the station control system further includes a vehicle information identification unit, which is used to:

[0278] Install cameras at the entrance of the battery swap station to capture images of battery swap vehicles entering the station;

[0279] Resize and convert color spaces of images;

[0280] The Canny edge detection algorithm is used to extract the vehicle outline, and the Hough transform method is used to detect the geometric shape in the vehicle outline to obtain the point cloud data of the vehicle;

[0281] The Euclidean clustering algorithm is used to group different objects in the point cloud data, and the key feature points of the vehicle contour are extracted from the clustered point cloud. The image features are fused with the point cloud features to form a complete feature vector.

[0282] Load the pre-trained deep learning model, input the feature vector into the deep learning model, and output the vehicle classification result.

[0283] The station control system also includes an initial posture recognition unit for:

[0284] Capturing an original image of the vehicle chassis through a visual sensor, and cropping and scaling the captured original image to a set size;

[0285] The deep learning model is used for feature extraction. The deep learning model downsamples the input image and predicts the position of the target bounding box center point and its offset to generate a feature map.

[0286] Predict the position of the target bounding box center point and its offset, as well as the width and height of the target bounding box based on the feature map;

[0287] Obtain the camera's intrinsic and extrinsic parameters, use warpAffine affine transformation to transform the predicted key point positions in the feature map into the spatial coordinates of the original image, and generate point cloud data of the vehicle chassis based on the results of 3D space restoration;

[0288] Perform voxel filtering on point cloud data to remove noise points;

[0289] Perform Euclidean clustering on the filtered point cloud, use principal component analysis (PCA) to analyze the point cloud data, find out the main direction of the point cloud data, and estimate the approximate direction of the vehicle chassis;

[0290] The random sampling consistency RANSAC algorithm is used to fit the point cloud data, exclude outliers, and obtain the plane model of the vehicle chassis;

[0291] Based on the results of PCA analysis and RANSAC fitting, the initial pose of the vehicle chassis is determined.

[0292] As a preferred embodiment of the above, the station control system further includes a locking and unlocking hole identification unit, which is used to:

[0293] The initial pose is preprocessed by denoising and contrast enhancement, and the original RGB image is converted to HSV color space;

[0294] Set the threshold in HSV color space for segmentation and use Canny edge detection algorithm to identify the edges in the image;

[0295] Use opening and closing operations to remove small interfering spots or disconnected edges, use Hough circle transform to detect circular contours in the image, and preliminarily locate the location of the unlocking holes;

[0296] Set the detection parameters, obtain the center coordinates and radius of each detected circle, and use the least squares method to fit a plane model for each initially located unlocking hole;

[0297] In the plane model, a smaller area around the center of the initially located unlocking hole is selected as the region of interest (ROI). The plane equation of the point set in this area is fitted using the least squares method, and the center position is adjusted according to the plane fitting result.

[0298] Extract the normal vector from the plane equation obtained by least squares plane fitting, and confirm whether the normal vector direction correctly points to the opening direction of the locking and unlocking hole;

[0299] Output the final determined center coordinates of the unlocking hole and the normal vector direction of the unlocking hole.

[0300] The station control system also includes an AGV control unit, which is used to:

[0301] Connect the independently controlled battery swap AGV in the battery swap station through the communication protocol;

[0302] Time synchronization for battery-swapping AGVs: Use the NTP time synchronization protocol to ensure that the timestamps of all AGVs and station control systems are consistent;

[0303] Perform task planning based on task requirements, AGV current status, vehicle arrival order, and battery pack type, and dynamically assign tasks to each AGV;

[0304] Using a scheduling algorithm, for battery-swap vehicles, one AGV is designated to perform disassembly operations and another AGV to perform assembly operations at the same time.

[0305] The station control system also includes a fault handling unit for:

[0306] When one of the AGVs fails, the other AGV switches to stand-alone mode to continue the battery replacement task and sends a fault message.

[0307] In this embodiment, the station control system further includes a distance control unit, which is used to:

[0308] Through the Leader-Follower strategy, the dual AGVs maintain the spacing and angle deviation within the set range, use distance sensors and angle sensors to monitor and adjust the spacing in real time, detect the relative distance between the two AGVs in real time, and use a spacing control algorithm to ensure that the two AGVs maintain a safe distance when traveling on a monorail.

[0309] As a preferred embodiment of the above, the AGV further includes a mode switching unit for:

[0310] According to the different stages of the battery replacement task, the AGV's disassembly mode, transfer mode, and assembly mode are automatically or manually switched.

[0311] Embodiment 2:

[0312] like Figure 3 As shown, this embodiment includes a passenger car intelligent battery swap station compatible with multi-locking mode battery packs, including: a station control system, intelligent battery swap equipment, an intelligent battery management warehouse, and an auxiliary system;

[0313] The station control system is responsible for overall coordination and control, and communicates in real time with the intelligent battery swapping equipment, intelligent battery management warehouse, auxiliary systems and other equipment in the battery swapping station. It coordinates the control of the dual adaptive locking and unlocking battery swapping AGV to replace battery packs with multiple locking methods. It has the functions of equipment monitoring, task scheduling, safety management and data analysis, collects information on the battery swapping process, monitors environmental parameters, and ensures the safety and efficiency of the battery swapping process.

[0314] Intelligent battery replacement equipment, used to select the corresponding locking and unlocking module according to the battery pack with different locking methods to autonomously replace the battery of the vehicle;

[0315] Intelligent battery management warehouse, which is used to centrally and orderly charge the batteries in the warehouse and manage the storage battery packs. It also optimizes the charging strategy to extend the battery life and improve the charging efficiency, and intelligently interacts with the power grid to relieve the pressure on the grid.

[0316] Auxiliary systems, including but not limited to power distribution systems, environmental controllers, logistics and supply chain systems, fire protection and safety systems, access control and monitoring systems, ventilation and lighting systems, waste treatment and recycling systems, etc.

[0317] like Figure 4 As shown in the figure, the station control system, as the brain of the smart battery swap station, integrates an advanced central processor and is responsible for collecting, processing and analyzing information from various devices in the battery swap station, including data from smart battery swap equipment, smart battery management warehouse and various sensors, to ensure the smoothness and efficiency of the entire battery swap process. It is connected to all smart devices in the station through a high-speed communication bus to achieve real-time data exchange and command transmission, ensuring the collaborative work between devices, such as AGV location information update, battery pack status check, etc.

[0318] The AGV is equipped with interchangeable locking and unlocking devices, which can automatically identify and switch to the corresponding module according to the different locking methods of the battery pack, so as to achieve compatibility with battery packs with multiple locking methods. It is equipped with high-precision cameras and sensors to identify the characteristics of the vehicle chassis and the position of the battery pack to ensure the accuracy and safety of the locking and unlocking operation. It adopts advanced scheduling algorithms to intelligently adjust the charging sequence and rate of the battery pack according to the grid load, battery status and user needs, realize the first-in-first-out principle of the battery, avoid overcharging and discharging of the battery, and extend the battery life. It monitors the status of each battery pack in real time, including parameters such as voltage, current, and temperature, and timely warns of potential faults to ensure the safe operation of the battery. It uses machine learning algorithms to predict battery demand and grid load, plan charging strategies in advance, realize peak shaving and valley filling, reduce grid pressure, and improve the overall efficiency of the power system. It monitors environmental parameters such as temperature, humidity, and air quality in the battery swap station to maintain suitable battery storage and charging conditions and reduce the impact of environmental factors on battery performance. It is equipped with fire alarm, smoke detection, emergency shutdown and other safety systems to ensure the safety of personnel and equipment during the operation of the battery swap station.

[0319] Through the intelligent dispatching of the station control system, the battery swap process is highly automated, and it is compatible with a variety of battery locking methods, which greatly improves the service scope and efficiency of the battery swap station. The orderly charging strategy not only balances the load of the power grid and realizes the optimal allocation of power resources, but also reduces the charging cost, bringing economic benefits to users and power companies. By reducing the fluctuation of power demand by shaving peaks and filling valleys, it helps the stable operation of the power system, improves the utilization efficiency of power resources, and promotes the sustainable development of the power grid.

[0320] The second aspect is the control method of the intelligent battery swap station for passenger cars that is compatible with battery packs with multiple locking modes, including: intelligent matching and adaptive unlocking control method, dual AGV collaborative control method, and orderly charging control method for battery packs;

[0321] Intelligent matching and adaptive unlocking control method is used to solve the compatibility of battery packs and the accuracy of unlocking and unlocking operations in battery swap stations. This method uses deep learning-based algorithms and image processing technology to accurately identify the vehicle chassis and battery pack unlocking holes, ensuring that battery packs of different models and different locking methods can be accurately matched and operated;

[0322] The dual AGV collaborative control method is used for the station control system to control two independent AGVs to work collaboratively on a single track. Through the communication bus, the two AGVs share position information and task status in real time. One is responsible for disassembly and the other is responsible for assembly, realizing parallel processing to improve battery replacement efficiency and safety.

[0323] Orderly charging control method for battery packs, where the station control system controls the charging optimization, maintenance and storage of battery packs, prolongs battery life, improves charging efficiency, uses algorithms to predict battery demand and grid load, intelligently adjusts charging plans, and avoids charging during peak hours;

[0324] like Figure 5 As shown in the figure, when an electric vehicle enters the entrance area of ​​a battery swap station, the vehicle identification sensor at the entrance automatically captures and analyzes the vehicle model and license plate information, and then sends key data to the station control system. The station control system verifies whether the license plate, model, and battery pack ID information of the battery swap vehicle are consistent, including the consistency of the license plate and model, the consistency of the battery pack ID and model, the legitimacy of the vehicle user, and the legitimacy of the battery pack.

[0325] When the vehicle information is found to be inconsistent, the station control system starts to respond and guides the driver to leave the battery swap station in time to avoid unnecessary detention. When the vehicle information is verified to be correct, the station control system enters the preparatory battery swap stage to confirm the battery pack information, vehicle configuration, inventory location of the battery to be replaced, and temporary storage space for recycled batteries, etc., to ensure that the intelligent battery swap device can accurately perform the next task. With the confirmation of the preparatory battery swap instruction, the station control system sends a gate opening signal to guide the user to enter the battery swap area smoothly. The driver follows the instructions and parks the vehicle at the designated location of the battery swap operation platform.

[0326] The adaptive unlocking and battery-swapping AGV equipped with a high-pixel camera recognizes the vehicle chassis shape, locates the unlocking hole of the vehicle battery pack, determines the locking type, transmits the information to the station control system, and matches it with the vehicle information. When it is found that the vehicle information does not match the locking type, the driver is guided to leave the operating platform and wait for further confirmation by the personnel in the battery swap station; when the vehicle information matches the locking type, the station control system controls the dual adaptive unlocking and battery-swapping AGVs to select the corresponding appropriate unlocking and battery-swapping module. The battery swap process begins, and the 1# adaptive unlocking and battery-swapping AGV moves to the bottom of the vehicle to remove the exhausted battery and place it on the exchange platform. After the 1# adaptive unlocking and battery-swapping AGV leaves the battery swapping operation platform, the 2# adaptive unlocking and battery-swapping AGV quickly arrives and installs the fully charged battery to the bottom of the vehicle to complete the battery swap process.

[0327] The station control system obtains the operating parameters and status of the battery swapping equipment in real time, and performs fault monitoring and alarm processing according to the safety strategy. The system generates process data such as battery swapping monitoring records and alarm records. After the battery swapping is completed, the station control system will generate a battery swapping record and establish a corresponding relationship record between the battery pack and the vehicle. The system prompts that the battery swapping is completed and opens the gate to allow the vehicle to leave.

[0328] like Figure 6 As shown, the overall system architecture of the station control system includes dual AGV collaborative control, orderly battery pack charging, network architecture, technical architecture, business architecture, and dual AGV collaborative control, including: dual AGV system, time synchronization, spacing control, mode switching, task allocation, collaborative operation, autonomous fault monitoring and fault-tolerant control;

[0329] The dual AGV system is based on a single-track "elastic" dual AGV system and consists of two independently controlled carts (AGVs). The communication between the dual AGVs and between the AGV and the station control system is connected via RS485 communication lines to form a triangular network structure. The station control system is the master node, responsible for task allocation and monitoring status, while the AGV is the slave node, receiving instructions and uploading status. Each cart can complete the task of replacing the vehicle battery pack autonomously or through upper centralized control, and is connected to the station control system through a communication protocol, and can autonomously perceive the environment and make decisions;

[0330] Time synchronization: through time synchronization technology, ensure that the two AGVs can accurately synchronize when performing tasks, avoid task conflicts caused by time deviation, and use NTP (Network Time Protocol) time synchronization protocol to ensure that the timestamps of all AGVs and station control systems are consistent, so that the time error between the two AGVs is kept at the millisecond level;

[0331] Distance control: Through the Leader-Follower strategy, the two AGVs maintain a certain distance and angle deviation. The distance sensor and angle sensor are used to monitor and adjust the distance in real time, detect the relative distance between the two AGVs in real time, and use the distance control algorithm to ensure that the two AGVs maintain a certain safe distance when driving on the monorail, ensure the stability of the formation, and prevent collisions.

[0332] Mode switching: A mode switching system based on vision and inertial navigation is designed to allow the AGV to automatically or manually switch between disassembly mode, transfer mode, assembly mode and other working modes according to different stages of the battery replacement task. The mode switching is performed through the control panel of the station control system to ensure that the AGV can switch working modes efficiently according to task requirements.

[0333] Task allocation: The station control system plans tasks based on task requirements, the current status of the AGV, the order of vehicle arrival, the type of battery pack, and other factors. It dynamically assigns specific tasks to each AGV and uses a scheduling algorithm to designate one AGV for disassembly operations and another for assembly operations, ensuring reasonable task allocation, avoiding resource waste, and improving battery replacement efficiency.

[0334] Collaborative operation, through the communication bus, the two AGVs share location information, task status and other data in real time to ensure the efficiency of collaborative work. A collaborative strategy is adopted to ensure that during the point change process, one AGV is responsible for disassembling the depleted battery pack and the other AGV is responsible for assembling the fully charged battery pack to achieve parallel processing.

[0335] Autonomous fault monitoring and fault-tolerant control. The AGV system is equipped with a multi-level safety protection mechanism, including laser obstacle avoidance, emergency shutdown, safety edge contact and limit switch, to ensure that each AGV has autonomous fault monitoring capabilities and can monitor its own operating status and health status in real time. Once a fault is found, the system can automatically diagnose the type and cause of the fault and take corresponding measures immediately. When one AGV fails, the other AGV can switch to stand-alone mode to continue to complete the power replacement task. The system can send fault information to the station control system in time to prompt maintenance personnel to perform maintenance. When the fault is more complicated or the system cannot handle it automatically, the system can issue a request for manual intervention to troubleshoot the fault. The AGV collaborative control system is designed with redundancy and recovery mechanisms, which can adjust the formation or task allocation through the re-dispatch of the station control system to ensure the continuity of the task when a single AGV fails.

[0336] Orderly charging control method for battery packs, including: orderly charging strategy, load forecasting and smoothing strategy, and battery pack optimization scheduling strategy;

[0337] The orderly charging strategy is based on the improved Cuckoo Search Algorithm (CSA). It explores the solution space by random walk and selecting the optimal solution according to the battery SOC, battery health status and scheduled use time of the vehicle, automatically sorts the battery charging priority, monitors the grid status and charging progress in real time, and immediately adjusts the charging plan and reduces the charging power until the grid load returns to normal once an abnormal increase in grid load is detected. The station control system receives real-time electricity price information released by the grid. When it detects that the electricity price is at a low point, it automatically increases the charging power. Conversely, it reduces charging or even stops charging when the electricity price is at a peak to avoid high electricity bills.

[0338] Load forecasting and smoothing strategy: The station control system continuously collects real-time data of each battery in the battery swap station, including battery SOC (state of charge), health status, charging rate, temperature and other parameters, and uses a combined prediction model based on historical data freshness function and cross entropy to analyze the impact of past charging patterns, weather conditions, holiday effects, etc. on charging demand. The freshness function takes into account the time decay effect of data, while the cross entropy algorithm optimizes the weight of the prediction model and improves the accuracy and stability of the prediction; through gray correlation analysis, find historical days with similar characteristics such as meteorological conditions and day types as the prediction day, as the basis for prediction, dynamically adjust the weight of a single prediction model based on the cross entropy algorithm, and combine the prediction results of multiple prediction methods (such as time series analysis and machine learning models) to form the final prediction; the station control system receives real-time load data from the power grid, combines its own load forecast, and pre-plans the distribution of charging power to avoid large-scale charging during the peak load period of the power grid. Through the particle swarm algorithm, it calculates the charging plan for the next day to ensure that the charging load is distributed during the period when the power grid load is lower, such as at night;

[0339] The battery pack optimization scheduling strategy is based on the first-in-first-out (FIFO) principle to ensure that the batteries that enter the warehouse first are used first, avoiding performance degradation caused by long-term idleness of batteries. Batteries in better health are given priority to ensure the quality of battery replacement service. At the same time, batteries in poor health are maintained or retired to prevent them from dragging down overall performance. Efficient battery replacement paths are designed to reduce the time for robotic arms or manual battery replacement and improve battery replacement efficiency. The station control system collects key parameters such as battery voltage, current, temperature, SOC, etc. in real time to evaluate the health of the battery. It uses big data analysis and machine learning technology to predict the life trend of the battery, detect potential faults in a timely manner, arrange preventive maintenance or replacement, and set a threshold for the battery health status. Once the battery status is detected to be lower than this threshold, it will be immediately removed from the battery replacement service for inspection or repair. The station control system maintains a certain number of fully charged spare batteries to cope with sudden charging needs or grid abnormalities. By using all batteries in a balanced manner, it avoids overcharging or discharging of a single battery and extends the overall life of the battery.

[0340] Technical architecture: The station control system communicates in real time with the adaptive unlocking and locking battery swapping AGV, environmental controller, and vehicle identification equipment, fully perceives the entire process information of battery swapping, monitors the environmental information in the battery swapping station, and provides battery swapping monitoring data for the battery swapping station operation system and government management platform, making the entire battery swapping process measurable, observable, and controllable, fully ensuring the efficiency and reliability of the charging and swapping process, as well as the safety of equipment and personnel.

[0341] Network architecture, the station control system uses TCP / IP network protocol. High-level protocol standardization can provide a variety of reliable network services, and then support the access and collection of multiple devices. The station control system uses MySQL data storage and processing, and has an optimized SQL query algorithm to effectively improve the query speed. Other systems access the station control system through Ethernet and obtain battery replacement related business data from the station control system. The industrial computer of the station control system is equipped with four Ethernet ports, which are used to independently connect to intelligent battery replacement equipment, industrial switches, other system access, and expansion ports; the four Ethernet ports are managed and configured through a soft firewall to ensure the security of the station control system.

[0342] like Figure 7 As shown in the figure, when a vehicle enters the battery swap station, a high-definition camera installed at the entrance of the battery swap station is used to capture the vehicle image, and the acquired image is resized and converted in color space to meet the needs of subsequent processing. The Canny edge detection algorithm is used to extract the vehicle contour, and the Hough transform and other methods are used to detect the geometric shape in the vehicle contour. The point cloud data of the vehicle and its chassis are obtained, and the Euclidean clustering algorithm is used to group different objects in the point cloud data. The key feature points of the vehicle contour are extracted from the clustered point cloud, and the image features are fused with the point cloud features to form a complete feature vector. The pre-trained deep learning model is loaded, and the fused feature vector is input into the model. The model outputs the classification result of the vehicle, that is, the vehicle model. The identified vehicle model information is used to query the relevant information in the database, and the corresponding battery pack unlocking hole model type and its specific parameters are retrieved from the database. The identified vehicle model information and the specific parameters of the unlocking hole are uploaded to the station control system.

[0343] When the vehicle parks at the battery swap operation platform, the original image of the vehicle chassis to be photographed is captured by the visual sensor, the captured original image is cropped and scaled to a uniform size, and a deep learning model (such as DLA-34) is used for feature extraction. The model downsamples the input image, predicts the position of the center point of the target bounding box and its offset, generates a feature map, and predicts the position of the center point of the target bounding box and its offset, as well as the width and height of the target bounding box based on the feature map. The camera's intrinsic and extrinsic parameters are set so that the information in the two-dimensional image can be converted into information in three-dimensional space. The warpAffine affine transformation is used to convert the predicted key point positions in the feature map into the spatial coordinates of the original image. Based on the results of the three-dimensional space restoration, the point cloud data of the vehicle chassis is generated. The point cloud data is voxel filtered to remove noise points and make the point cloud smoother. The filtered point cloud is clustered Euclideanly to group the point clouds belonging to the same object together. The point cloud data is analyzed using principal component analysis (PCA) to find the main direction of the point cloud data and estimate the approximate direction of the vehicle chassis. The random sampling consensus (RANSAC) algorithm is used to fit the point cloud data to eliminate the influence of outliers and obtain the planar model of the vehicle chassis. Based on the results of PCA analysis and RANSAC fitting, the initial position and posture of the vehicle chassis are determined.

[0344] The initial posture of the vehicle chassis is preprocessed by denoising and contrast enhancement, and the original RGB image is converted into HSV (hue, saturation, value) color space. According to the color characteristics of the unlocking holes, the threshold is set in the HSV space for segmentation, and the unlocking hole area is highlighted. The Canny edge detection algorithm is applied to identify the edges in the image. The opening and closing operations are used to remove small interference spots or disconnected edges. The Hough circle transform is used to detect the circular contour in the image, so as to preliminarily locate the position of the unlocking holes. Appropriate detection parameters are set, such as the radius range of the circle, the accumulator threshold, etc., to obtain the center coordinates and radius of each detected circle. For each initially located unlocking hole, the least squares method is used to fit a plane model to further optimize the center position. A smaller area around the initially located center is selected as the region of interest (ROI), and the least squares method is used to fit the plane equation of the point set in this area. The center position is adjusted according to the plane fitting result to ensure that it is located on the best fitting plane. Extract the normal vector from the plane equation obtained by least squares plane fitting, confirm whether the normal vector direction correctly points to the opening direction of the locking and unlocking hole, and output the final determined center coordinates of the locking and unlocking hole and the normal vector direction of the locking and unlocking hole.

[0345] Based on the vehicle model information obtained from the multi-model / multi-model battery pack information identification process, the model type of the unlocking hole to be retrieved is clarified based on the vehicle model information query results, and the battery pack unlocking hole model type and its specific parameters that match the vehicle model are retrieved from the battery swap station database. The identified unlocking hole information (such as the center position, normal direction, etc.) is compared with the information in the database to ensure that the unlocking hole model type is completely consistent with the information recorded in the database. Based on the matching unlocking hole model type, determine the battery swap AGV to select the appropriate unlocking operation method.

[0346] Through the spatial transposition design, the battery-swapping AGV can quickly switch between different locking and unlocking methods in a limited space. The height switching mechanism is adopted, so that the mechanism can automatically adjust the height of the locking and unlocking mechanism according to the height difference of the battery packs of different models, achieve precise matching, break through the precision design challenge of the locking and unlocking mechanism with large tonnage push and pull forces in a very small space, and ensure the stability and reliability of the mechanism. The three locking and unlocking mechanisms can be quickly switched to meet the locking and unlocking requirements of battery packs of different models. Through the highly integrated locking and unlocking mechanism adaptive floating function, it is ensured that during the locking and unlocking process, the mechanism can be fine-tuned according to the position changes of the battery pack to ensure the accuracy and safety of the locking and unlocking operation.

[0347] See also Figure 8 A computer device 400 provided in an embodiment of the present application includes: a processor 410 and a memory 420, wherein the memory 420 stores a computer program executable by the processor 410, and when the computer program is executed by the processor 410, the above method is executed.

[0348] The embodiment of the present application further provides a storage medium 430 on which a computer program is stored. When the computer program is run by the processor 410, the above method is executed.

[0349] Among them, the storage medium 430 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable red-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, disk or optical disk.

[0350] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. "Multiple" means two or more, unless otherwise clearly and specifically defined.

[0351] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like 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 an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0352] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0353] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0354] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0355] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0356] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0357] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present invention. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A control method for a smart battery swap station compatible with multiple battery packs, characterized in that: The steps include: Obtain vehicle information of the vehicle that needs battery replacement, and retrieve the corresponding battery pack unlocking hole model type and its specific parameters from the database; When the battery swap vehicle stops at the battery swap operation platform, an image of the vehicle chassis is acquired to determine the initial position and posture of the vehicle chassis; Processing the initial posture to extract the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole; Compare the center coordinates of the keypad unlock hole and the normal vector direction of the keypad unlock hole with the keypad unlock hole model type and specific parameters of the battery pack retrieved from the database to determine whether the information is consistent; If they are inconsistent, the user is prompted to leave the battery swap operation platform; if they are consistent, the battery swap automatic guided vehicle AGV selects the corresponding locking and unlocking operation mode, retrieves the battery in the warehouse, and replaces the vehicle battery; After the battery swap is completed, a battery swap report is generated, allowing the user to leave; The battery-swapping automatic guided vehicle AGV selects a corresponding locking and unlocking operation mode, retrieves the battery from the warehouse, and replaces the vehicle battery, including: Through the spatial transposition design, the battery-swapping automatic guided vehicle AGV can switch between different locking and unlocking modes; Use the height switching mechanism to automatically adjust the height of the locking and unlocking mechanism according to the height differences of the battery packs of different models; During the locking and unlocking process, the adaptive floating function is used to make fine adjustments according to the position changes of the battery pack. The battery-swapping automatic guided vehicle AGV selects a corresponding locking and unlocking operation mode, retrieves a battery from a storage, and replaces the vehicle battery, further comprising: The battery replacement operation is performed through the dual AGV collaborative control method, specifically including: Connect the independently controlled battery swap AGV in the battery swap station through the communication protocol; Time synchronization for battery-swapping AGVs: Use the NTP time synchronization protocol to ensure that the timestamps of all AGVs and station control systems are consistent; Perform task planning based on task requirements, AGV current status, vehicle arrival order, and battery pack type, and dynamically assign tasks to each AGV; Using a scheduling algorithm, for battery-swap vehicles, one AGV is designated to perform disassembly operations and another AGV to perform assembly operations; The battery replacement operation is performed by the dual AGV collaborative control method, and further includes: Autonomous fault monitoring and fault-tolerant control: When one AGV fails, the other AGV switches to stand-alone mode to continue the battery replacement task and sends fault information; The battery replacement operation is performed by the dual AGV collaborative control method, and further includes: Through the Leader-Follower strategy, the dual AGVs maintain the spacing and angle deviation within the set range, use distance sensors and angle sensors to monitor and adjust the spacing in real time, detect the relative distance between the two AGVs in real time, and use a spacing control algorithm to ensure that the two AGVs maintain a safe distance when traveling on a monorail.

2. The control method of a smart battery swap station compatible with multiple battery packs according to claim 1, characterized in that: The step of obtaining vehicle information of a vehicle that needs battery replacement includes the following steps: Install cameras at the entrance of the battery swap station to capture images of battery swap vehicles entering the station; Resize and convert color spaces of images; The Canny edge detection algorithm is used to extract the vehicle outline, and the Hough transform method is used to detect the geometric shape in the vehicle outline to obtain the point cloud data of the vehicle; The Euclidean clustering algorithm is used to group different objects in the point cloud data, and the key feature points of the vehicle contour are extracted from the clustered point cloud. The image features are fused with the point cloud features to form a complete feature vector. Load a pre-trained deep learning model, input the feature vector into the deep learning model, and output the classification result of the vehicle.

3. The control method of a smart battery swap station compatible with multiple battery packs according to claim 1, characterized in that: The step of acquiring an image of the vehicle chassis and determining an initial position and posture of the vehicle chassis comprises the following steps: Capturing an original image of the vehicle chassis through a visual sensor, and cropping and scaling the captured original image to a set size; A deep learning model is used to extract features. The deep learning model performs downsampling on the input image and predicts the position of the center point of the target bounding box and its offset to generate a feature map. Predict the position of the target bounding box center point and its offset, as well as the width and height of the target bounding box based on the feature map; Obtain the camera's intrinsic and extrinsic parameters, use warpAffine affine transformation to transform the predicted key point positions in the feature map into the spatial coordinates of the original image, and generate point cloud data of the vehicle chassis based on the results of 3D space restoration; Perform voxel filtering on point cloud data to remove noise points; Perform Euclidean clustering on the filtered point cloud, use principal component analysis (PCA) to analyze the point cloud data, find out the main direction of the point cloud data, and estimate the approximate direction of the vehicle chassis; The random sampling consistency RANSAC algorithm is used to fit the point cloud data, exclude outliers, and obtain the plane model of the vehicle chassis; Based on the results of PCA analysis and RANSAC fitting, the initial pose of the vehicle chassis is determined.

4. The control method of a smart battery swap station compatible with multiple battery packs according to claim 1, characterized in that: The processing of the initial posture to extract the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole includes: The initial pose is preprocessed by denoising and contrast enhancement, and the original RGB image is converted into an HSV color space; Set the threshold in HSV color space for segmentation and use Canny edge detection algorithm to identify the edges in the image; Use opening and closing operations to remove small interfering spots or disconnected edges, use Hough circle transform to detect circular contours in the image, and preliminarily locate the location of the unlocking holes; Set the detection parameters, obtain the center coordinates and radius of each detected circle, and use the least squares method to fit a plane model for each initially located unlocking hole; In the plane model, an area around the center of the initially located unlocking hole is selected as the region of interest (ROI), the plane equation of the point set in this area is fitted using the least squares method, and the center position is adjusted according to the plane fitting result; Extract the normal vector from the plane equation obtained by least squares plane fitting, and confirm whether the normal vector direction correctly points to the opening direction of the locking and unlocking hole; Output the final determined center coordinates of the unlocking hole and the normal vector direction of the unlocking hole.

5. The control method of a smart battery swap station compatible with multiple battery packs according to claim 1, characterized in that: Also included is a battery pack orderly charging control method, comprising the following steps: Orderly charging strategy: Based on the improved cuckoo search algorithm CSA, it explores the solution space by random walk and selecting the optimal solution according to the battery SOC, battery health status and scheduled use time of the vehicle, and automatically sorts the battery charging priority; Load forecasting and smoothing strategy: Continuously collect real-time data of each battery in the battery swap station, use the combined forecasting model of freshness function and cross entropy based on historical data, receive real-time load data of the power grid, combine its own load forecast, and pre-plan the allocation of charging power; Battery pack optimization scheduling strategy: Based on the first-in-first-out (FIFO) principle, ensure that the batteries that enter the warehouse first are used first.

6. The control method of a smart battery swap station compatible with multiple battery packs according to claim 5, characterized in that: The orderly charging strategy also includes: Monitor the grid status and charging progress in real time. When an abnormal increase in grid load is detected, adjust the charging plan and reduce the charging power until the grid load returns to normal. Receive real-time electricity price information released by the power grid. When it is detected that the electricity price is at a low point, increase the charging power; conversely, reduce or stop charging when the electricity price is at a peak.

7. The control method of a smart battery swap station compatible with multiple battery packs according to claim 5, characterized in that: The load forecasting and smoothing strategy also includes: A combined prediction model of freshness function and cross entropy based on historical data is used to analyze the impact of past charging patterns, weather conditions, and holiday effects on charging demand. Through grey correlation analysis, we find out historical days with similar meteorological conditions and day type characteristics as the forecast day, and use them as the forecast benchmark. Based on the cross entropy algorithm, we dynamically adjust the weight of the single forecast model. Combine forecasting methods such as time series analysis and machine learning model forecast results to form a final forecast; The charging plan for the next day is calculated through the particle swarm algorithm.

8. The control method of a smart battery swap station compatible with multiple battery packs according to claim 5, characterized in that: The battery pack optimization scheduling strategy also includes: Collect key parameters of battery voltage, current, temperature, and SOC in real time to evaluate the health of the battery; Use big data analysis and machine learning technology to predict battery life trends; Set a threshold for the battery health status. Once the battery status is detected to be below this threshold, it will be immediately removed from the battery swap service for inspection or repair; Maintain more than a set number of fully charged spare batteries to avoid overcharging or discharging a single battery by balancing the use of all batteries.

9. A smart battery swap station compatible with multiple battery packs, characterized in that: include: Communication-connected station control system, intelligent battery swapping equipment and intelligent battery management warehouse; The intelligent battery exchange equipment includes several automatic guided vehicles (AGVs), and the AGVs include several locking and unlocking modules corresponding to the locking methods of different vehicle models and battery packs; The intelligent battery management compartment stores a plurality of batteries and charges the stored batteries; The station control system is used to obtain vehicle information of the vehicle that needs battery replacement, and retrieve the corresponding battery pack unlocking hole model type and its specific parameters from the database; when the battery replacement vehicle stops at the battery replacement operation platform, obtain the image of the vehicle chassis and determine the initial posture of the vehicle chassis; process the initial posture to extract the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole; compare the center coordinates of the unlocking hole and the normal vector direction of the unlocking hole with the battery pack unlocking hole model type and its specific parameters retrieved from the database to determine whether the information is consistent; if inconsistent, prompt the user to leave the battery replacement operation platform; If they are consistent, the intelligent battery swapping device controls the battery swapping AGV to select the corresponding locking and unlocking operation mode, retrieves the battery in the intelligent battery management compartment, and replaces the vehicle battery; After the battery replacement is completed, the station control system generates a battery replacement report, allowing the user to leave; The AGV also includes a Mecanum wheel, a servo motor, a vision system, a laser radar and a sensor; Also includes: A spatial transfer mechanism, through which the AGV switches between different locking and unlocking modes; Height switching mechanism, used to automatically adjust the height of the locking and unlocking mechanism according to the height differences of battery packs in different vehicle models; An adaptive floating mechanism, which is adaptively floating during the locking and unlocking process and is fine-tuned according to the position change of the battery pack; The station control system also includes an AGV control unit, which is used to: Connect the independently controlled battery swap AGV in the battery swap station through the communication protocol; Time synchronization for battery-swapping AGVs: Use the NTP time synchronization protocol to ensure that the timestamps of all AGVs and station control systems are consistent; Perform task planning based on task requirements, AGV current status, vehicle arrival order, and battery pack type, and dynamically assign tasks to each AGV; Using a scheduling algorithm, for battery-swap vehicles, one AGV is designated to perform disassembly operations and another AGV to perform assembly operations; The station control system further includes a fault processing unit, which is used to: When one of the AGVs fails, the other AGV switches to stand-alone mode to continue the battery replacement task and sends a fault message; The station control system further includes a distance control unit, which is used for: Through the Leader-Follower strategy, the dual AGVs maintain the spacing and angle deviation within the set range, use distance sensors and angle sensors to monitor and adjust the spacing in real time, detect the relative distance between the two AGVs in real time, and use a spacing control algorithm to ensure that the two AGVs maintain a safe distance when traveling on a monorail.

10. The smart battery swap station compatible with multiple battery packs according to claim 9, characterized in that: The station control system comprises: Technical architecture, which is used for the station control system to communicate in real time with intelligent battery swapping equipment, intelligent battery management warehouse and equipment and systems in the battery swapping station, perceive the whole process information of battery swapping, monitor the environmental information in the battery swapping station, and provide battery swapping monitoring data for the battery swapping station operation system and government management platform; Network architecture, which uses TCP / IP network protocol to support access and collection of multiple devices. The station control system uses MySQL data storage and processing and has SQL query algorithm; Business architecture, which is used to ensure the smoothness of the automated battery swapping process through battery swapping monitoring, auxiliary equipment monitoring to maintain the stability of the station environment and facilities, safety management to prevent and respond to potential risks, operation analysis to provide decision support, system management to coordinate equipment and user data, and service management to promote seamless connection with external platforms.

11. The smart battery swap station compatible with multiple battery packs according to claim 9, characterized in that: The intelligent battery replacement device also includes: The flexible battery swap operation platform is used to automatically adjust the vehicle position, provide a fixed electric vehicle parking operation area, adapt to battery swap vehicles of various types of wheelbases and track widths, and automatically position the battery swap vehicles.

12. The smart battery swap station compatible with multiple battery packs according to claim 9, characterized in that: The intelligent battery management compartment comprises: Dynamic charging scheduling unit: uses machine learning algorithms to predict battery demand and grid load and dynamically adjust charging plans; Battery life optimization unit: used to select the most appropriate charging rate according to the battery type and current status, avoid overcharging or discharging, extend battery life, and formulate regular maintenance plans based on data analysis to ensure the long-term health of the battery pack; Flexible storage and dispatching unit: used for linkage control with battery-swapping AGV to realize automatic storage and retrieval of battery packs, and automatically allocate battery packs to charging stations or storage areas according to battery status and charging requirements.

13. The smart battery swap station compatible with multiple battery packs according to claim 12, characterized in that: The intelligent battery management compartment also includes: An ordered charging unit for automatically sorting the charging priority of the battery by exploring the solution space through random walks and selecting the optimal solution based on the improved cuckoo search algorithm CSA according to the battery SOC, battery health status and scheduled use time of the vehicle; The load forecasting and smoothing unit is used to continuously collect the real-time data of each battery in the battery swap station, and use the combined forecasting model of the freshness function and cross entropy based on historical data to receive the real-time load data of the power grid, and combine its own load forecast to pre-plan the allocation of charging power; The battery pack optimization scheduling unit is used to ensure that the batteries that enter the warehouse earliest are used first based on the first-in-first-out (FIFO) principle.

14. The smart battery swap station compatible with multiple battery packs according to claim 13, characterized in that: The orderly charging unit is also used for: Monitor the grid status and charging progress in real time. When an abnormal increase in grid load is detected, adjust the charging plan and reduce the charging power until the grid load returns to normal. Receive real-time electricity price information from the power grid, and increase charging power when it detects that the electricity price is at a low point; Conversely, reduce or stop charging when electricity prices are high.

15. The smart battery swap station compatible with multiple battery packs according to claim 13, characterized in that: The load forecasting and smoothing unit is also used for: A combined prediction model of freshness function and cross entropy based on historical data is used to analyze the impact of past charging patterns, weather conditions, and holiday effects on charging demand. Through grey correlation analysis, we find out historical days with similar meteorological conditions and day type characteristics as the forecast day, and use them as the forecast benchmark. Based on the cross entropy algorithm, we dynamically adjust the weight of the single forecast model. Combine forecasting methods such as time series analysis and machine learning model forecast results to form a final forecast; The charging plan for the next day is calculated through the particle swarm algorithm.

16. The smart battery swap station compatible with multiple battery packs according to claim 13, characterized in that: The battery pack optimization scheduling unit is also used for: Collect key parameters of battery voltage, current, temperature, and SOC in real time to evaluate the health of the battery; Use big data analysis and machine learning technology to predict battery life trends; Set a threshold for the battery health status. Once the battery status is detected to be below this threshold, it will be immediately removed from the battery swap service for inspection or repair; Maintain more than a set number of fully charged spare batteries to avoid overcharging or discharging a single battery by balancing the use of all batteries.

17. The smart battery swap station compatible with multiple battery packs according to claim 9, characterized in that: The station control system also includes a vehicle information identification unit, which is used to: Install cameras at the entrance of the battery swap station to capture images of battery swap vehicles entering the station; Resize and convert color spaces of images; The Canny edge detection algorithm is used to extract the vehicle outline, and the Hough transform method is used to detect the geometric shape in the vehicle outline to obtain the point cloud data of the vehicle; The Euclidean clustering algorithm is used to group different objects in the point cloud data, and the key feature points of the vehicle contour are extracted from the clustered point cloud. The image features are fused with the point cloud features to form a complete feature vector. Load a pre-trained deep learning model, input the feature vector into the deep learning model, and output the classification result of the vehicle.

18. The smart battery swap station compatible with multiple battery packs according to claim 9, characterized in that: The station control system further includes an initial posture recognition unit, which is used to: Capturing an original image of the vehicle chassis through a visual sensor, and cropping and scaling the captured original image to a set size; A deep learning model is used to extract features. The deep learning model performs downsampling on the input image and predicts the position of the center point of the target bounding box and its offset to generate a feature map. Predict the position of the target bounding box center point and its offset, as well as the width and height of the target bounding box based on the feature map; Obtain the camera's intrinsic and extrinsic parameters, use warpAffine affine transformation to transform the predicted key point positions in the feature map into the spatial coordinates of the original image, and generate point cloud data of the vehicle chassis based on the results of 3D space restoration; Perform voxel filtering on point cloud data to remove noise points; Perform Euclidean clustering on the filtered point cloud, use principal component analysis (PCA) to analyze the point cloud data, find out the main direction of the point cloud data, and estimate the approximate direction of the vehicle chassis; The random sampling consistency RANSAC algorithm is used to fit the point cloud data, exclude outliers, and obtain the plane model of the vehicle chassis; Based on the results of PCA analysis and RANSAC fitting, the initial pose of the vehicle chassis is determined.

19. The smart battery swap station compatible with multiple battery packs according to claim 9, characterized in that: The station control system also includes a locking and unlocking hole identification unit, which is used to: The initial pose is preprocessed by denoising and contrast enhancement, and the original RGB image is converted into an HSV color space; Set the threshold in HSV color space for segmentation and use Canny edge detection algorithm to identify the edges in the image; Use opening and closing operations to remove small interfering spots or disconnected edges, use Hough circle transform to detect circular contours in the image, and preliminarily locate the location of the unlocking holes; Set the detection parameters, obtain the center coordinates and radius of each detected circle, and use the least squares method to fit a plane model for each initially located unlocking hole; In the plane model, an area around the center of the initially located unlocking hole is selected as the region of interest (ROI), the plane equation of the point set in this area is fitted using the least squares method, and the center position is adjusted according to the plane fitting result; Extract the normal vector from the plane equation obtained by least squares plane fitting, and confirm whether the normal vector direction correctly points to the opening direction of the locking and unlocking hole; Output the final determined center coordinates of the unlocking hole and the normal vector direction of the unlocking hole.

20. The smart battery swap station compatible with multiple battery packs according to claim 9, characterized in that: The AGV further includes a mode switching unit for: According to the different stages of the battery replacement task, the AGV's disassembly mode, transfer mode, and assembly mode are automatically or manually switched.

21. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 9 is implemented.

22. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 9 is implemented.

Images