Dynamic positioning unmanned energy storage battery switching system

Abstract

The invention relates to the technical field of electric power energy storage battery replacement, in particular to a dynamic positioning unmanned energy storage battery replacement system, which comprises a dynamic positioning module, a power conversion module, a power conversion module, a power conversion module, a power conversion module, a power conversion module, a power conversion module, a power conversion module and a power conversion module, the self-adaptive battery grabbing unit monitors battery deformation based on positioning data and feedback of a six-dimensional force sensor, adjusts grabbing parameters according to the type of the battery, drives a mechanical arm to conduct force-position hybrid control, and adopts a graded force control strategy to adjust the claw distance and the locking force in a self-adaptive mode. The intelligent storage scheduling center dynamically generates a battery replacement instruction through a demand prediction model based on the battery health degree, the load demand and the battery replacement priority; the dynamic fault-tolerant unit monitors battery replacement abnormity and triggers an emergency strategy; and the block chain evidence storage unit writes the key information into an alliance chain, and operation traceability and equipment state right confirmation are carried out. Therefore, the problems of low positioning precision, poor compatibility and the like in the prior art are solved.

Description

Technical Field

[0001] This invention relates to the field of power storage battery swapping technology, specifically a dynamically positioned unmanned power storage battery swapping system. Background Technology

[0002] As a core facility supporting the efficient operation of energy storage power stations and new energy vehicle battery swapping networks, the dynamically positioned unmanned energy storage battery swapping system is prone to problems during long-term operation. These problems include excessive positioning deviation, insufficient grasping adaptability, and battery swapping process bottlenecks, due to factors such as diverse battery models, cabin attitude deviation, and environmental vibration interference. This leads to low battery swapping efficiency, increased equipment collision risk, and even safety accidents such as battery short circuits. As the energy storage industry develops towards large-scale and high-density applications, higher requirements are placed on battery swapping systems in terms of sub-millimeter positioning accuracy, compatibility with multiple battery specifications, unmanned reliability throughout the entire process, and reliable data traceability. There is an urgent need for an intelligent battery swapping system that can integrate multi-source sensor data in real time, dynamically adapt to different battery characteristics, and quickly respond to abnormal operating conditions to ensure the safe and efficient replacement of energy storage batteries in diverse application scenarios.

[0003] However, traditional unmanned battery swapping technology has inherent defects: It relies on a single positioning method, depending solely on vision or lidar perception, without integrating inertial measurement data for dynamic compensation. Positioning errors often exceed 5mm under vibration or temperature drift conditions, leading to a high docking failure rate. The gripping mechanism parameters are fixed, only compatible with 1-2 standard batteries. It is prone to unstable gripping or over-clamping when facing batteries with irregular interfaces or slight deformations, resulting in extremely poor compatibility. The scheduling strategy is crude, allocating batteries only according to "first-come, first-served" or power priority, without considering state of health (SOH) and swapping frequency to optimize inventory, leading to a high-efficiency battery idle rate exceeding 30%. Fault handling is passive; failure of a single sensor or robotic arm triggers system shutdown, lacking a redundancy switching mechanism. The average annual fault impact time far exceeds the stringent requirements for continuous operation of energy storage systems. Data recording is scattered; swapping process parameters are only stored locally, making them susceptible to tampering and difficult to trace, failing to meet the responsibility confirmation requirements in multi-entity collaborative scenarios. Overall, the technology faces multiple challenges: low positioning accuracy, poor compatibility, low reliability, and insufficient data credibility. Summary of the Invention

[0004] This application provides a dynamically positioned unmanned energy storage battery swapping system to solve the problems of low positioning accuracy and poor compatibility in the prior art.

[0005] The first aspect of this application provides a dynamically positioned unmanned energy storage battery swapping system, comprising: a dynamic positioning module, an adaptive battery grasping unit, an intelligent warehouse scheduling center, a dynamic fault-tolerant unit, and a blockchain evidence storage unit; wherein, the dynamic positioning module is used to output the three-dimensional coordinates and attitude angle of the target battery in real time through the fusion perception of LiDAR, visual camera, and IMU inertial measurement unit, combined with a spatiotemporal synchronous calibration algorithm, and dynamically compensate for vibration offset and temperature drift effects; the adaptive battery grasping unit is used to monitor battery deformation in real time using a laser rangefinder based on positioning data and feedback from a six-dimensional force sensor, automatically adjust the grasping parameters according to the battery model, and drive the robotic arm to perform force-position hybrid control. The system employs a hierarchical force control strategy to adaptively adjust the claw distance and locking force. The intelligent warehouse scheduling center dynamically generates battery swapping instructions and allocates batteries based on battery health, load demand, and swapping priority using a demand prediction model. The dynamic fault-tolerant unit monitors abnormalities in the swapping process in real time and triggers emergency strategies, including: real-time correction of the robotic arm's trajectory; automatic switching to a backup mechanism when the trajectory deviation exceeds an absolute value threshold; and switching to local control mode when communication is interrupted. The blockchain evidence storage unit writes the positioning parameters, force control data, and key battery ID information of the swapping process into the consortium blockchain for operation traceability and equipment status confirmation.

[0006] Preferably, the dynamic positioning module includes a sensor fusion unit, a dynamic compensation unit, and a coordinate output unit. The sensor fusion unit is used to fuse data from lidar, vision camera, and IMU inertial measurement unit, and perform data alignment through a spatiotemporal synchronization calibration algorithm. The dynamic compensation unit corrects vibration offset in real time based on a filtering algorithm and dynamically corrects IMU temperature drift error through a temperature drift compensation algorithm. The coordinate output unit is used to output three-dimensional coordinates and attitude angles in real time.

[0007] Preferably, the adaptive battery gripping unit includes a six-dimensional force sensing module, a laser deformation detection unit, and a graded force control actuator. The six-dimensional force sensing module is used to provide real-time feedback of the gripping contact force; the laser deformation detection unit is used to monitor the deformation of the battery casing using a line laser; and the graded force control actuator is used to perform force-position hybrid control through an impedance control algorithm, adopting a three-level force control strategy to dynamically adjust the claw distance and automatically identify and adapt to the battery.

[0008] Preferably, the intelligent warehouse scheduling hub includes a health assessment unit, a demand prediction model, and a dynamic allocation unit. The health assessment unit calculates the state of health (SOH) value based on the number of battery cycles, the rate of change of internal resistance, and the amount of capacity decay, and classifies the health level. The demand prediction model is based on an LSTM neural network and generates a battery swapping plan by combining historical battery swapping data and real-time load demand. The dynamic allocation unit uses a priority algorithm to generate the optimal battery allocation scheme and performs concurrent scheduling of multiple battery warehouses.

[0009] Preferably, the dynamic fault-tolerant unit includes an anomaly monitoring module and a trajectory correction unit, wherein the anomaly monitoring module is used to monitor the joint angle and positioning deviation of the robotic arm in real time; and the trajectory correction unit uses a PID control algorithm to correct the motion path.

[0010] Preferably, the dynamic fault-tolerant unit further includes a redundancy switching unit and a local control module, wherein the redundancy switching unit is used to switch to the backup unit when the main unit fails and to enable the local control mode when communication is interrupted; the local control module is used to perform offline power swapping and maintain basic operation when communication is interrupted.

[0011] Preferably, the blockchain evidence storage unit includes a consortium blockchain writing unit and a smart contract module, wherein the consortium blockchain writing unit is used to write key data into distributed nodes using a consensus algorithm; and the smart contract module is used to automatically perform operation traceability and device status confirmation to generate tamper-proof electronic certificates.

[0012] A second aspect of this application provides a dynamically positioned unmanned energy storage battery swapping method, comprising: acquiring dynamic positioning data, battery parameters, and environmental vibration data; performing spatiotemporal synchronization and filtering processing on the positioning data, battery parameters, and environmental vibration data to generate a battery swapping scenario dataset; evaluating positioning accuracy and battery health level using a sensor fusion algorithm based on the battery swapping scenario dataset, identifying grasping deviations and robotic arm joint faults using an anomaly detection model, and obtaining evaluation results and abnormal positioning data; and dynamically adjusting the grasping claw distance, locking force, and [other parameters] based on the evaluation results, combined with real-time force control feedback and historical battery swapping parameters. The robot arm moves along a trajectory and generates basic battery swapping instructions. Based on these instructions, a weighted approach is used, combining the analytic hierarchy process (AHP) to allocate weights for positioning accuracy, battery health, and load requirements. The basic battery swapping instructions are dynamically optimized based on swapping priorities to generate a target battery swapping scheme. Battery removal and installation operations are performed according to the target scheme. The battery swapping process is monitored in real time through a dynamic fault-tolerance mechanism. When the trajectory deviation exceeds a threshold, the path is corrected. If the main mechanism fails, a backup robot arm is switched on. If communication is interrupted, local control is initiated. Key data from the battery swapping process is written to the consortium blockchain for storage and evidence, generating tamper-proof operation records and equipment status confirmation information.

[0013] Preferably, the sensor fusion algorithm formula is:

[0014] in, Let be the fusion weights for the i-th sensor; Let be the measurement variance of the i-th sensor; The total number of sensors participating in the fusion; For summation index variables; This is the final estimated position after fusion; This represents the original position measurement value of the i-th sensor; The internal resistance of the battery; This refers to the change in voltage. This refers to the change in current. Vibrational energy density; The amplitude of vibration at a specific frequency point f; For the frequency differential component; Battery health status; , , , These are the regression coefficients; It is a temperature decay function; For temperature.

[0015] Preferably, the anomaly detection model formula is:

[0016] in, This is the current measurement value; This is the average value under normal conditions; This represents the standard deviation under normal conditions. The degree of deviation after standardization; The parameter to be detected; It is the lower quartile; It is the upper quartile; It is a robust measure of data dispersion.

[0017] Therefore, this application has the following beneficial effects: This application embodiment integrates data from LiDAR, visual camera, and IMU via a dynamic positioning module, combining spatiotemporal synchronization calibration and dynamic compensation algorithms to effectively offset vibration offset and temperature drift effects, improving positioning accuracy and solving the problem of excessive deviation in traditional single-sensor positioning under complex environments. The adaptive battery gripping unit, relying on six-dimensional force perception and laser deformation detection, achieves adaptive gripping of batteries of different models, even those with slight deformation, through a hierarchical force control strategy and impedance control algorithm. It dynamically adjusts the claw distance and locking force, avoiding battery damage caused by excessive clamping and addressing the pain point of poor compatibility with irregularly shaped batteries. The intelligent warehouse scheduling center, based on battery health assessment and an LSTM demand prediction model, combines priority... The algorithm dynamically allocates batteries, prioritizing those with high health to handle peak demand. This not only improves battery turnover efficiency but also extends overall battery life and reduces inefficient idleness through scientific allocation. The dynamic fault-tolerant unit, through real-time anomaly monitoring and model-predictive control trajectory correction, combined with redundancy switching between primary and backup mechanisms and local offline control mode, responds quickly when the robotic arm deviation exceeds thresholds or communication is interrupted, minimizing the impact of faults and overcoming the limitation of traditional systems where a single module failure leads to shutdown. The blockchain evidence storage unit, through consortium blockchain distributed storage and smart contracts, immutably stores key information such as positioning parameters and force control data during the battery swapping process, achieving end-to-end traceability and equipment status confirmation. This solves the problems of low positioning accuracy and poor compatibility in existing technologies.

[0018] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0019] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a schematic diagram of a dynamically positioned unmanned energy storage battery swapping system according to an embodiment of this application. Figure 2 This is a schematic diagram of a dynamic positioning module provided according to an embodiment of this application; Figure 3 This is a schematic diagram of an adaptive battery gripping unit according to an embodiment of this application; Figure 4 This is a schematic diagram of an intelligent warehouse scheduling hub provided according to an embodiment of this application; Figure 5 This is a schematic diagram of a dynamic fault-tolerant unit provided according to an embodiment of this application; Figure 6 This is a schematic diagram of a blockchain evidence storage unit provided according to an embodiment of this application; Figure 7 This is a schematic diagram of a dynamically positioned unmanned energy storage battery swapping system according to an embodiment of this application; Figure 8 This is a structural diagram of the unmanned energy storage battery swapping machine provided according to an embodiment of this application; Figure 9 This is a flowchart of a dynamically positioned unmanned energy storage battery swapping method according to an embodiment of this application; Figure 10 This is a schematic diagram of a dynamically positioned unmanned energy storage battery swapping method according to an embodiment of this application; Figure 11 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The following description, with reference to the accompanying drawings, illustrates an embodiment of a dynamically positioned unmanned energy storage battery swapping system according to this application. Addressing the low positioning accuracy issue mentioned in the background section, this application provides a dynamically positioned unmanned energy storage battery swapping system. In this system, a dynamic positioning module integrates data from LiDAR, a visual camera, and an IMU, combined with spatiotemporal synchronization calibration and dynamic compensation algorithms to effectively counteract vibration offset and temperature drift effects, improving positioning accuracy and solving the problem of excessive deviation in traditional single-sensor positioning under complex environments. The adaptive battery gripping unit, relying on six-dimensional force perception and laser deformation detection, achieves adaptive gripping of batteries of different models, even those with slight deformation, through a hierarchical force control strategy and impedance control algorithm. It dynamically adjusts the claw distance and locking force, avoiding battery damage caused by excessive clamping and solving the pain point of poor compatibility with irregularly shaped batteries. The intelligent warehouse scheduling center is based on electricity... The system employs a battery health assessment and LSTM demand prediction model, combined with a priority algorithm to dynamically allocate batteries. Batteries with high health are prioritized for peak demand, improving battery turnover efficiency and extending overall battery life through scientific allocation, while reducing inefficient idle time. A dynamic fault-tolerant unit, through real-time anomaly monitoring and model-predicted control trajectory correction, along with redundant switching between primary and backup mechanisms and a local offline control mode, responds quickly when the robotic arm deviation exceeds a threshold or communication is interrupted, minimizing the impact of faults and overcoming the limitation of traditional systems where a single module failure results in shutdown. A blockchain-based evidence storage unit, through consortium blockchain distributed storage and smart contracts, immutably stores key information such as positioning parameters and force control data during the battery swapping process, achieving end-to-end traceability and equipment status confirmation. This solves the problems of low positioning accuracy and poor compatibility in existing technologies.

[0022] Figure 1 This is a schematic diagram of a dynamically positioned unmanned energy storage battery swapping system provided in an embodiment of this application.

[0023] This application provides a dynamically positioned unmanned energy storage battery swapping system, the system 10 including: The system includes a dynamic positioning module 100, an adaptive battery grabbing unit 200, an intelligent warehouse scheduling center 300, a dynamic fault-tolerant unit 400, and a blockchain evidence storage unit 500.

[0024] The system includes the following components: a dynamic positioning module 100, which uses multi-sensor fusion sensing combined with a spatiotemporal synchronous calibration algorithm to output the three-dimensional coordinates and attitude angles of the target battery in real time, dynamically compensating for vibration offset and temperature drift effects; an adaptive battery gripping unit 200, which uses positioning data and feedback from a six-dimensional force sensor, monitors battery deformation in real time using a laser rangefinder, automatically adjusts gripping parameters according to the battery model, drives the robotic arm to perform force-position hybrid control, and adopts a hierarchical force control strategy to adaptively adjust the claw distance and locking force; an intelligent warehouse scheduling hub 300, which dynamically generates battery swapping instructions and allocates batteries based on battery health, load demand, and battery swapping priority through a demand prediction model; a dynamic fault-tolerant unit 400, which monitors abnormalities in the battery swapping process in real time and triggers emergency strategies, including: real-time correction of the robotic arm's motion trajectory; automatic switching to a backup mechanism when the trajectory deviation exceeds an absolute value threshold; and switching to local control mode when communication is interrupted; and a blockchain evidence storage unit 500, which writes the positioning parameters, force control data, and key battery ID information of the battery swapping process into the consortium blockchain for operation traceability and equipment status confirmation.

[0025] It is understood that in this embodiment, the dynamic positioning module integrates data from LiDAR, visual camera, and IMU, combined with spatiotemporal synchronization calibration and dynamic compensation algorithms, effectively offsetting vibration offset and temperature drift effects, improving positioning accuracy, and solving the problem of excessive deviation in traditional single-sensor positioning under complex environments. The adaptive battery gripping unit, relying on six-dimensional force perception and laser deformation detection, achieves adaptive gripping of batteries of different models, even those with slight deformation, through a hierarchical force control strategy and impedance control algorithm. It dynamically adjusts the claw distance and locking force, avoiding battery damage caused by excessive clamping and solving the pain point of poor compatibility with irregularly shaped batteries. The intelligent warehouse scheduling center, based on battery health assessment and an LSTM demand prediction model, combines… The system dynamically allocates batteries using a priority algorithm, prioritizing batteries with high health to handle peak demand. This not only improves battery turnover efficiency but also extends overall battery life and reduces inefficient idleness through scientific allocation. The dynamic fault-tolerant unit, through real-time anomaly monitoring and model-predictive control trajectory correction, combined with redundancy switching between primary and backup mechanisms and local offline control mode, responds quickly when the robotic arm deviation exceeds thresholds or communication is interrupted, minimizing the impact of faults and overcoming the limitation of traditional systems where a single module failure leads to shutdown. The blockchain evidence storage unit, through consortium blockchain distributed storage and smart contracts, immutably stores key information such as positioning parameters and force control data during the battery swapping process, achieving full-chain traceability and equipment status confirmation. This solves the problems of low positioning accuracy and poor compatibility in existing technologies.

[0026] In this embodiment of the application, the dynamic positioning module 100 includes: Figure 2 As shown, there are sensor fusion unit, dynamic compensation unit, and coordinate output unit.

[0027] The sensor fusion unit is used to fuse data from lidar, vision camera and IMU inertial measurement unit, and perform data alignment through spatiotemporal synchronization calibration algorithm; the dynamic compensation unit corrects vibration offset in real time based on filtering algorithm, and dynamically corrects IMU temperature drift error through temperature drift compensation algorithm; the coordinate output unit is used to output three-dimensional coordinates and attitude angles in real time.

[0028] It is understood that the sensor fusion unit in this application embodiment achieves accurate alignment of multi-source data by fusing the distance measurement accuracy of LiDAR, the scene texture recognition capability of visual camera, and the dynamic response characteristics of IMU inertial measurement unit, combined with spatiotemporal synchronization calibration algorithm. This avoids the perception limitations of a single sensor in complex environments and provides more comprehensive underlying data for positioning. The dynamic compensation unit relies on filtering algorithm to correct the position offset caused by equipment vibration in real time, and dynamically offsets the measurement error caused by temperature drift of IMU through temperature drift compensation algorithm, resisting the impact of environmental interference on positioning accuracy and ensuring stable measurement accuracy under complex working conditions such as vibration and temperature fluctuation. The coordinate output unit outputs three-dimensional coordinates and attitude angles in real time, providing real-time and accurate position references for operations such as battery gripping and robotic arm movement, ensuring precise coordination of each link in the entire battery swapping process, and improving the reliability and accuracy of dynamic positioning.

[0029] It should be noted that the formula for the spatiotemporal synchronization calibration algorithm is as follows:

[0030] in, This is the sensor's original timestamp; This is the sensor time drift coefficient; This is the initial time offset; To calibrate the start time; The coordinates of three-dimensional points acquired by the IMU; These are the coordinates of a 3D point after transformation to the lidar coordinate system; The rotation matrix from IMU to lidar; This is the translation vector from the IMU to the lidar; This is the calibrated unified timestamp.

[0031] Filtering algorithm formula:

[0032] in, The prior state estimate at time k; The posterior state estimate at time k−1; This is the state transition matrix; Let be the covariance matrix of the prior estimation error; The covariance matrix at time k−1; For process noise covariance; for The transpose of the matrix; The Kalman gain matrix; For measurement matrix; for The transpose of the matrix; To measure the noise covariance matrix; For the posterior state estimate at time k; The sensor measurement value at time k; To measure the residuals; Let k be the posterior covariance matrix at time k; It is the identity matrix; The process noise covariance at the next time step; Forgetting factor; To calculate the covariance of the residuals.

[0033] For example, in energy storage battery swapping scenarios, when a robotic arm performs a battery grasping action, vibrations in the workshop equipment can cause positioning offsets. Simultaneously, the IMU experiences temperature drift due to ambient temperature fluctuations, leading to attitude measurement distortion. The dynamic compensation unit relies on the Extended Kalman Filter (EKF) algorithm. First, based on the IMU's acceleration and angular velocity data, combined with a kinematic model, it predicts the battery's three-dimensional position and attitude (state prediction stage). However, vibration interference causes deviations in this prediction. Subsequently, distance measurements from the LiDAR and texture localization from the vision camera provide real-time position references, while temperature drift data from the temperature sensor is used to correct the IMU's zero-bias error. The EKF dynamically balances the trust weights of the "predicted value" and the "multi-sensor measurements" by calculating the Kalman gain—if the vibration is severe, the gain tilts towards the LiDAR, prioritizing its high-precision position to correct prediction deviations; if the temperature drift is significant, the IMU's zero-bias parameters are adjusted through temperature feedback. After the status update, the output battery coordinates and attitude angles effectively offset the effects of vibration offset and temperature drift, improve positioning accuracy, and ensure that the robotic arm can still accurately grasp the battery under complex working conditions, thus laying a solid core support for the stable operation of unmanned battery swapping.

[0034] In this embodiment of the application, the adaptive battery grasping unit 200 includes: as follows Figure 3 As shown, the six-dimensional force sensing module, laser deformation detection unit, and graded force control actuator are included.

[0035] Among them, the six-dimensional force sensing module is used to provide real-time feedback on the gripping contact force; the laser deformation detection unit is used to monitor the deformation of the battery casing through line laser; and the graded force control actuator is used to perform force-position hybrid control through impedance control algorithm, adopting a three-level force control strategy to dynamically adjust the claw distance and automatically identify and adapt to the battery.

[0036] It is understood that the embodiments of this application use a six-dimensional force sensing module to provide real-time feedback on the contact force between the robotic arm and the battery, accurately capturing force changes during the gripping process and avoiding damage to the battery casing or internal cells due to excessive force. The laser deformation detection unit monitors the minute deformation of the battery casing in real time through line laser scanning, providing data for adjusting gripping parameters and preventing gripping instability caused by abnormal battery shape. The graded force control actuator performs coordinated control of force and position based on an impedance control algorithm, adopting a three-level force control strategy of pre-grip (low force trial), locking (adaptive force value fixed), and holding (dynamic fine-tuning force value). Combined with the dynamic adjustment of the claw distance, it can automatically identify batteries of different sizes and interface types and complete the adaptation. Through real-time perception of force and deformation, the risk of over-grip or unstable gripping is avoided, improving the safety, accuracy, and versatility of gripping operations during unmanned battery swapping.

[0037] It should be noted that the impedance control algorithm formula is as follows:

[0038] in, The target force for the interaction between the robot's end effector and the environment; This is a virtual mass matrix; This is a virtual damping matrix; This is a virtual stiffness matrix; For the desired position; This refers to the actual location; For the desired speed; This refers to the actual speed; For the desired acceleration; This is the actual acceleration.

[0039] The six-dimensional force sensing module provides real-time feedback on the contact force between the robotic arm and the battery, accurately capturing force changes during the grasping process. It simultaneously detects force values ​​in three translational directions and torques in three rotational directions, forming comprehensive force information monitoring. Data acquisition and transmission are performed with extremely high response speed, enabling real-time feedback on the contact force between the robotic arm and the battery—from the initial contact of the robotic arm's end effector with the battery surface, to the gradual application of clamping force, adjustment of the grasping posture, and even minute force changes that may occur during the movement after grasping, every instantaneous force dynamic can be captured and transmitted to the control system. This not only ensures precise measurement of force magnitude but also clearly reflects changes in force direction. For example, a slight tilt of the battery can cause a shift in the force direction at the robotic arm's contact point, or slight deformation of the battery casing during grasping can cause abnormal force fluctuations. By monitoring force changes, the robotic arm can adjust its motion parameters in real time, preventing damage to the battery casing and internal cells due to excessive force, while also preventing unstable grasping and battery detachment due to insufficient force. This ensures safe, stable, and high-precision battery grasping operations.

[0040] The graded force-controlled actuator is used for force-position hybrid control through an impedance control algorithm. It employs a three-level force control strategy, dynamically adjusting the jaw distance and automatically identifying and adapting to the battery. Based on the impedance control algorithm, force-position hybrid control is achieved by constructing a dynamic balance between force and position (similar to the characteristics of a spring-damped system). This controls the actuator's positional accuracy while simultaneously adjusting the output force in real time, preventing damage to the battery or mechanical structure due to rigid contact. The three-level force control strategy is key to achieving precise operation: The first level is the contact sensing stage, where a very low initial force triggers contact detection when the jaws approach the battery, ensuring gentle contact and avoiding impact. The second level is the clamping and adaptation stage, which automatically switches the force control threshold based on the battery casing material (e.g., hard-shell, soft-pack), applying a slightly higher torque to hard-shell batteries. To ensure a firm grip, the clamping force is reduced for pouch batteries to prevent bulging. The third stage is the stabilization and holding stage, which maintains a constant force value during the gripping and moving process to counteract force fluctuations caused by vibration or posture changes. At the same time, the actuator can dynamically adjust the claw distance through force feedback signals: when the claw contacts the battery, it monitors the difference in force distribution in real time. If the force values ​​on both sides are uneven, it fine-tunes the claw distance until the force is balanced, ensuring that the battery center is aligned. For different sizes (such as 18650 cylindrical batteries and square lithium iron phosphate batteries), no manual preset parameters are required. The battery specifications are automatically identified by the characteristics of force value changes during the three-stage force control process (such as the claw distance position when the preset force value is reached), and then the claw distance is adaptively adjusted. This achieves a fully programmable adaptation from contact to gripping, greatly improving the compatibility with diverse batteries.

[0041] For example, taking the gripping of multiple types of batteries as an example, the graded force control actuator first constructs a millisecond-level dynamic force-position balance system through an impedance control algorithm. When the mechanical gripper approaches an 18650 cylindrical battery with a diameter of 18mm, the first-level force control stage triggers contact sensing with an initial force of less than 5N, and the sensor captures the shell contact signal within 0.1 seconds. After entering the second-level stage, the system automatically calls the hard-shell battery force control model to stabilize the clamping force at 30N±2N, which avoids shell indentation and ensures a firm grip. During the movement, the third-level force control is activated, which monitors and compensates for vibration interference forces of ±5N in real time, with a force value adjustment response delay of no more than 50ms. If switching to a soft-pack lithium battery with a thickness of 5mm, the second-level force control will automatically switch to a safe threshold of 15N±1N, and at the same time, it can adapt to different width specifications of 80mm-120mm through fine adjustment of the gripper distance with a precision of 0.5mm. When encountering a square lithium iron phosphate battery with a side length of 40mm, the actuator can identify the battery type within 0.3 seconds based on the force change curve at the moment of contact (the characteristic force increase point of the hard shell), and automatically lock the claw distance at 42mm±0.1mm. No manual parameter setting is required throughout the process, enabling seamless switching and gripping of multiple battery specifications.

[0042] In this embodiment of the application, the intelligent warehouse scheduling hub 300 includes: Figure 4As shown, there are health assessment unit, demand forecasting model, and dynamic allocation unit.

[0043] The health assessment unit calculates the SOH value based on the number of battery cycles, the rate of change of internal resistance, and the amount of capacity decay, and classifies the health level; the demand forecasting model is based on an LSTM neural network and combines historical battery swapping data and real-time load demand to generate a battery swapping plan; the dynamic allocation unit uses a priority algorithm to generate a target battery allocation scheme and performs concurrent scheduling of multiple battery compartments.

[0044] It is understood that the health assessment unit in this application calculates the SOH value and classifies health levels by quantifying parameters, accurately selects qualified batteries, and ensures the safety and reliability of battery swapping; the demand prediction model generates a battery swapping plan based on LSTM neural network combined with historical and real-time data, matches demand in advance, and reduces waiting time caused by supply and demand imbalance; the dynamic allocation unit uses a priority algorithm to generate a scheme and realizes concurrent scheduling of multiple battery compartments, which can prioritize the response to urgent needs, improve scheduling efficiency and resource utilization, and improve the overall operational efficiency of intelligent warehousing.

[0045] It should be noted that the health assessment unit calculates the SOH value and classifies health levels through quantitative parameters, accurately screening qualified batteries. By accurately collecting data throughout the battery's life cycle, the core quantitative parameters are the number of cycles (e.g., setting 800 cycles as the critical value), the rate of change of internal resistance (a warning is triggered when it exceeds the initial value by 20%), and the amount of capacity decay (marked as abnormal when the remaining capacity is less than 80%). The SOH (health status) value is calculated using the capacity decay method. For example, a battery with 300 cycles, a 10% increase in internal resistance, and 90% remaining capacity has an SOH value of 85 points. Based on the SOH value, the battery is classified into four health levels: excellent (90-100 points), good (70-89 points), requires maintenance (50-69 points), and scrapped (<50 points). During the screening process, high-quality and good-quality batteries are automatically included in the qualified pool and given priority for high-frequency battery swapping scenarios. Batteries awaiting maintenance are marked and then transferred to the testing and repair process. Scrap batteries are locked and a recycling command is triggered. This not only eliminates problems such as a sudden drop in vehicle range and safety hazards after battery swapping by removing batteries with degraded performance, but also accurately matches the battery health status with usage needs, improving the efficiency of battery resource utilization and reducing the cost waste caused by blind replacement.

[0046] Capacity decay method formula:

[0047] in, For battery health; This represents the current maximum discharge capacity. This refers to the battery's factory rated capacity.

[0048] LSTM neural network formula:

[0049] in, The output of the forget gate; It is the sigmoid activation function; Here is the weight matrix for the forget gate; To restore the hidden state of the previous moment and the input at the current moment Concatenate them into a single vector; For the bias term of the forget gate; The output of the input gate; This is the weight matrix of the input gate; This is the bias term for the input gate; Candidate cell state; It is the hyperbolic tangent activation function; The weight matrix for candidate cells; For the bias term of candidate cells; This represents the current state of the cell. This is element-wise multiplication; This represents the cell state at the previous moment; This is the output of the output gate; This is the weight matrix of the output gate; This is the bias term for the output gate; The current hidden state; This is the model's predicted output; This is the weight matrix of the output layer; This is the bias term for the output layer.

[0050] Priority algorithm formula:

[0051] in, Rate the priority of tasks; This represents the total number of impact factors. The index is the ordinal number of the factor; The weight coefficient of the k-th factor; This is the quantized value of the k-th factor.

[0052] The dynamic allocation unit performs intelligent scheduling based on a priority algorithm: It considers the urgency of the battery swapping demand (e.g., automatically increasing priority if vehicle waiting time exceeds 15 minutes), user level (VIP orders have a 20% weight increase), battery health (batteries with SOH ≥ 80% are prioritized), remaining charge (SOC ≥ 70% is more suitable for immediate battery swapping needs), and warehouse status (battery warehouses with a load rate exceeding 80% have their scheduling weight reduced to balance pressure). By processing these indicators and performing a weighted sum, a priority score is generated for each battery-demand pairing, with the highest-scoring combination being prioritized for inclusion in the target allocation scheme. Simultaneously, the system supports concurrent scheduling across multiple battery warehouses, meaning that parallel instruction issuance and path planning can be performed on batteries in different warehouses within the same time period. For example, while warehouse A is processing a VIP order, warehouse B can simultaneously complete the battery dispatch for another regular order, significantly shortening the overall scheduling cycle and increasing the battery swapping service volume per unit time. This ensures rapid response to urgent needs while avoiding efficiency bottlenecks caused by overloading a single warehouse, achieving optimal allocation of warehouse resources.

[0053] For example, in an unmanned battery swapping scenario at a large-scale energy storage power station, the dynamic allocation unit achieves precise battery matching through a three-dimensional scheduling logic of "health-demand-priority": When three battery swapping requests are received simultaneously—an emergency response energy storage container (requiring 10 batteries, high load power, highest priority), a conventional new energy vehicle charging request (requiring 2 batteries, medium priority), and a backup battery warehouse replenishment request (requiring 5 batteries, low priority)—the unit first calls the health assessment data to classify the batteries into Class A (SOH≥90%, low cycle count, stable internal resistance), Class B (70%≤SOH<90%, moderate performance), and Class C (SOH<70%, only suitable for low loads). Combining the "high load will continue for the next hour" conclusion output by the LSTM demand forecasting model, the dynamic allocation unit activates the priority algorithm: 10 Class A batteries are preferentially allocated to the emergency energy storage container to ensure high reliability; 2 Class B batteries are allocated to the new energy vehicle to balance performance and cost; and 5 Class C batteries are used for backup warehouse replenishment to avoid high-health batteries being idle. Meanwhile, the unit references battery swapping frequency records, temporarily storing Class A batteries that have been swapped twice in the last 3 days, and prioritizing the scheduling of Class A batteries that have been idle for more than 48 hours to reduce accelerated aging caused by high-frequency use. The entire process completes instruction generation within 8 seconds, ensuring power supply stability for emergency tasks and improving overall battery turnover efficiency by 35% and extending the average lifespan of a single battery by 12% through differentiated allocation, perfectly solving the resource mismatch and inefficiency problems caused by traditional "sequential allocation".

[0054] In this embodiment of the application, the dynamic fault-tolerant unit 400 includes, as follows: Figure 5 As shown, the anomaly monitoring module and trajectory correction unit are shown.

[0055] The anomaly monitoring module is used to monitor the joint angles and positioning deviations of the robotic arm in real time; the trajectory correction unit uses a PID control algorithm to correct the motion path.

[0056] It is understood that the embodiments of this application use an anomaly monitoring module to capture the joint angle offset and positioning deviation of the robotic arm in real time and identify operational anomalies; the trajectory correction unit dynamically adjusts the motion path according to the PID control algorithm, which can not only accurately avoid trajectory deviation caused by mechanical errors or environmental interference to ensure operational accuracy, but also prevent the risk of anomaly spread in advance, improve the operational stability and task execution reliability of the robotic arm, and reduce the probability of failure and maintenance costs.

[0057] It should be noted that the PID control algorithm formula is as follows:

[0058] in, For output control quantity; This is the error value; This is the proportional gain coefficient; This is the integral gain coefficient; The differential gain coefficient; For time; Let be the derivative of the error e(t); The integral of the error; It is the integral variable.

[0059] In this embodiment, the dynamic fault-tolerant unit 400 further includes a redundancy switching unit and a local control module.

[0060] The redundancy switching unit is used to switch to the backup unit when the main unit fails and to enable the local control mode when communication is interrupted; the local control module is used to perform offline power swapping and maintain basic operations when communication is interrupted.

[0061] It is understood that the redundancy switching unit in this application embodiment, through the switching of the main and backup mechanisms and the switching of the local control mode during communication interruption, cooperates with the offline power swapping support of the local control module and the maintenance of basic operations during communication interruption. This not only enables rapid response to ensure the continuity of system operation when abnormal situations such as main mechanism failure or communication interruption occur, but also improves fault tolerance, reliability and operational stability by maintaining critical operations without interruption and stable maintenance under fault conditions, thereby reducing downtime losses caused by abnormalities.

[0062] In this embodiment of the application, the blockchain evidence storage unit 500 includes, as follows: Figure 6 As shown, the consortium blockchain write unit and smart contract module.

[0063] The consortium blockchain writing unit is used to write key data into distributed nodes using a consensus algorithm; the smart contract module is used to automatically perform operation traceability and device status confirmation, generating tamper-proof electronic certificates.

[0064] It is understood that the consortium blockchain writing unit in this application embodiment uses a consensus algorithm to write key data into distributed nodes, ensuring the distributed and consistent storage of data; the smart contract module automatically completes operation traceability and device status confirmation and generates tamper-proof electronic certificates. Through distributed storage and tamper-proofness, the authenticity and security of data are guaranteed, and the efficiency of operation traceability and rights confirmation is improved by automated processing, thereby enhancing the credibility and reliability of the data.

[0065] It should be noted that the consensus algorithm formula is as follows:

[0066] in, The set of all nodes in the consortium blockchain; This represents the total number of nodes; For a single vertex; For nodes Message value; For a specific message value.

[0067] For example, in the shared charging pile operation scenario, when a user scans the code to start charging, the smart contract module is automatically triggered: it records the charging start time, user identity, charging pile number and other operation information in real time and stores it on the blockchain. At the same time, it connects with the charging pile sensors to obtain real-time power, operating status and other data, and verifies whether the equipment is in normal service status (such as ownership belongs to operator A, monthly maintenance has been completed) according to preset rules to complete the status confirmation. After charging is completed, the contract automatically calculates the fee and generates an electronic certificate containing charging duration, fee details and equipment status confirmation, which is synchronized to the blockchain ledger of the user, operator and regulatory node. The whole process does not require manual intervention, which allows users to trace the entire charging process at any time, and avoids fee disputes or equipment ownership disputes through tamper-proof certificates, thus improving the efficiency of multi-party collaboration.

[0068] This application proposes a dynamic positioning unmanned energy storage battery swapping system. By fusing data from LiDAR, a visual camera, and IMU through a dynamic positioning module, combined with spatiotemporal synchronization calibration and dynamic compensation algorithms, it effectively counteracts vibration offset and temperature drift effects, improving positioning accuracy and solving the problem of excessive deviation in traditional single-sensor positioning under complex environments. The adaptive battery gripping unit, relying on six-dimensional force perception and laser deformation detection, achieves adaptive gripping of batteries of different models, even those with slight deformation, through a hierarchical force control strategy and impedance control algorithm. It dynamically adjusts the claw distance and locking force, avoiding battery damage caused by excessive clamping and addressing the pain point of poor compatibility with irregularly shaped batteries. The intelligent warehouse scheduling center, based on battery health assessment and LSTM demand... A predictive model, combined with a priority algorithm, dynamically allocates batteries, prioritizing those with high health to handle peak demand. This not only improves battery turnover efficiency but also extends overall battery life and reduces inefficient idleness through scientific allocation. A dynamic fault-tolerant unit, through real-time anomaly monitoring and model-predictive control trajectory correction, along with redundant switching between primary and backup mechanisms and local offline control mode, responds quickly when the robotic arm deviation exceeds a threshold or communication is interrupted, minimizing the impact of faults and overcoming the limitation of traditional systems where a single module failure leads to shutdown. A blockchain-based evidence storage unit, through consortium blockchain distributed storage and smart contracts, immutably stores key information such as positioning parameters and force control data during the battery swapping process, achieving end-to-end traceability and equipment status confirmation. This solves the problems of low positioning accuracy and poor compatibility in existing technologies.

[0069] The following will illustrate a dynamic positioning unmanned energy storage battery swapping system through a specific embodiment, such as... Figure 7 As shown, it includes: In the electric truck battery swapping scenario of new energy logistics parks, to meet the daily battery swapping demand, unmanned energy storage battery swapping mechanical structures are deployed, such as... Figure 8As shown, the dynamic positioning module, based on the visual positioning system shown in the figure, integrates a multi-sensor fusion scheme and is deployed at the top of the gantry of the battery swapping station, 5.2 meters above the ground. Coordinating with the movement range of the lateral assembly, it achieves ±120-degree horizontal scanning coverage through a 360-degree rotating bracket, accurately capturing the 3D information of the battery compartment of the electric truck entering the battery swapping area. The sensor fusion unit achieves clock synchronization of all devices via industrial Ethernet (IEEE 802.3 standard), with synchronization errors controlled within 50 microseconds. The visual positioning system uses an industrial-grade global shutter camera (model Baslerac A4096-30uc), with a resolution of 4096×3072, a frame rate of 30fps, and an 8mm fixed-focus lens. The lens optical axis forms a 30-degree angle with the horizontal direction, accurately acquiring the QR code (using DM code, size 10cm×10cm) and feature textures (such as interface outlines and screw hole positions) on the surface of the battery compartment. Image data is transmitted to the image processing unit via the HDMI interface. Feature points are extracted using the SIFT algorithm, achieving a matching accuracy of 99.9% with pre-stored templates. The auxiliary sensing module (such as an IMU inertial measurement unit, model BNO055) is connected to the main controller via the SPI interface. The acceleration measurement range is ±16g, the angular velocity measurement range is ±2000° / s, and the sampling frequency is 100Hz. It captures the vibration and attitude changes of the equipment caused by vehicle driving vibration and wind disturbance (maximum wind speed in the park is 8m / s) in real time. The raw data is converted by 16-bit AD and then input into the fusion algorithm. The data from the three sources are processed by a spatiotemporal synchronization calibration algorithm: timestamp alignment is achieved using a hardware-triggered method, with the visual positioning system triggering synchronous sampling of auxiliary sensors for each frame of data; coordinate system transformation is based on the visual positioning system coordinate system (origin set at the camera center, X-axis along the direction of the transverse assembly movement, Y-axis horizontally, and Z-axis vertically upward), and the pixel coordinates are transformed to three-dimensional spatial coordinates through a preset transformation matrix (based on the camera extrinsic calibration results, the rotation matrix R and translation vector T are obtained through Zhang's calibration method). The IMU data is then corrected for coordinate system rotation using attitude angles (roll angle, pitch angle, yaw angle) predicted by Kalman filtering, ultimately achieving 10 fusion data outputs per second. The output data format is JSON, containing timestamps, three-dimensional coordinates, and confidence parameters. The dynamic compensation unit addresses errors caused by ground vibrations in the park (caused by trucks entering and exiting, with amplitudes of approximately ±5mm and frequencies of 2-10Hz) and environmental temperature variations (-10℃ to 40℃). It employs an extended Kalman filter (EKF) algorithm to correct vibration offsets in real time, with a filtering frequency of 50Hz. The state vector includes position (x, y, z) and velocity (vx, vy, vz). The state noise covariance matrix Q is set to diag[0.01, 0.01, 0.01, 0.001, 0.001, 0.001]. The measurement noise covariance matrix R is dynamically adjusted based on sensor accuracy.The temperature drift compensation algorithm, based on feedback data from the IMU's built-in temperature sensor (measurement accuracy ±0.5℃), establishes a temperature-error model (obtaining error curves within the range of -10℃ to 40℃ through offline calibration, and using cubic polynomial fitting). Every 100ms, it linearly corrects the angular velocity and acceleration output by the IMU, controlling the temperature drift error to within 0.1° / h, ensuring stability during long-term operation. The coordinate output unit sends the processed target battery's three-dimensional coordinates (X, Y, Z) and attitude angles (roll, pitch, yaw) to the central control system via Ethernet TCP protocol, guiding the battery gripping mechanism's actions. The three-dimensional coordinate accuracy reaches ±5mm, the attitude angle accuracy is ±0.5°, and the data transmission delay is ≤100ms. In actual testing, even with a truck parking deviation of ±30cm, it can still stably output accurate coordinates of the battery compartment, with a positioning success rate greater than 95%.

[0070] The adaptive battery gripping unit is integrated into the battery gripping mechanism, working in conjunction with the lateral movement assembly (for adjusting horizontal position) and the rotation assembly (for adjusting gripping angle). Relying on a six-axis motion system (repeatability ±0.1mm, maximum load 500kg, working radius 3.2m), it meets the battery gripping needs of trucks of different tonnages. This unit achieves adaptive gripping for three mainstream battery models in the park (50kWh, 80kWh, and 100kWh, with dimensions of 1200mm×600mm×300mm, 1500mm×700mm×350mm, and 1800mm×800mm×400mm respectively). A six-dimensional force sensor (model ATIMini45) is integrated between the mechanical gripper and the end flange of the mechanism, measuring a range of ±5000N (force) and ±500N・m (torque), with a sampling frequency of 1kHz. It provides real-time feedback of the contact force between the mechanical gripper and the battery via an EtherCAT bus. The force sensor has a resolution of 0.1N and a linearity of 0.5%FS. When a lateral force (perpendicular to the gripping direction) exceeding 500N is detected, the compliant control mode is immediately triggered. By adjusting the damping coefficients of each joint in the lateral and slewing assemblies (increasing from 50N·s / m to 100N·s / m), the force is flexibly buffered to prevent the battery casing from being deformed due to excessive force. A laser rangefinder (model KeyenceIL-1000) is installed on both sides of the mechanical gripper (50mm from the gripper tip). The laser wavelength is 650nm, the measurement range is 0-5m, the accuracy is ±0.1mm, and the sampling frequency is 500Hz. The deformation of the battery casing is monitored in real time using the laser triangulation method (the measurement point is located at the midpoint of the long side of the battery), with a maximum allowable deformation of 3mm. If the detected deformation exceeds the threshold, the system automatically adjusts the gripping point (moving it 50mm towards both ends of the battery) and reduces the locking force by 10%. At the same time, it issues an audible and visual alarm through the visual monitoring system, prompting the background to check the battery status. The graded force control actuator consists of a mechanical claw driven by a servo motor (model Yaskawa SGM7G). The claw body is made of high-strength aluminum alloy (7075-T6) and the surface is covered with a 3mm thick polyurethane anti-slip pad (Shore hardness 60A) to prevent slippage during gripping. The actuator employs a three-level force control strategy: for a 50kWh battery (weighing 200kg), the jaw spacing is adjusted to 400mm, the locking force is controlled at 800-1000N (via closed-loop control using a force sensor), and the gripping speed is 50mm / s; for an 80kWh battery (weighing 320kg), the jaw spacing is 550mm, the locking force is 1200-1500N, and the gripping speed is 40mm / s; for a 100kWh battery (weighing 400kg), the jaw spacing is 650mm, the locking force is 1800-2200N, and the gripping speed is 30mm / s. This is achieved through an impedance control algorithm. Achieving hybrid force-position control: During the gripping process, the unit first rapidly approaches the battery in position control mode (speed decreases to 20mm / s when 100mm away). When the force sensor detects a contact force of 50N, it instantly switches to force control mode, smoothly applying the locking force according to a preset force curve (linearly increasing to 80% of the target force within 0-0.5s, then increasing to the target force in 0.5s), ensuring a shock-free gripping process (maximum impact force ≤200N). In practical applications, this unit identifies the model markings (QR code or characters) on the battery surface through a vision positioning system, achieving 100% accuracy in identifying different battery models and a 99.8% gripping success rate. No battery damage due to improper gripping was observed during a six-month operation period. The alignment accuracy of the battery mounting holes reaches ±1mm, meeting the requirements for automatic insertion and removal. The intelligent warehousing and scheduling hub is deployed in the central control system of the battery swapping station (using an industrial-grade server, configured with an Intel Xeon Gold 6348 processor (28 cores and 56 threads), 128GB DDR4 memory, 2TB NVMe solid-state drive, and redundant power supply). It is connected to the pallet assembly (200 battery storage compartments, divided into three zones: A, B, and C, storing high-quality, sub-high-quality, and batteries awaiting maintenance, respectively) and the push assembly (roller design, running speed of 2m / s, positioning accuracy of ±5mm) and 80 spare batteries via an industrial Ethernet (ring network structure, redundant design). Battery (stored in a constant temperature chamber, temperature controlled at 25±2℃): The health assessment unit collects data through the battery management system (BMS) at a frequency of 10 seconds / time, including parameters such as cycle number, internal resistance (measurement accuracy ±5%), and capacity (measured using constant current discharge method, error ±2%), and calculates the SOH value (health status): taking the minimum value of three indicators, which are the percentage of current capacity to rated capacity, 100% minus the internal resistance change rate (the difference between current internal resistance and initial internal resistance divided by the initial internal resistance), and 100% minus the ratio of cycle number to the designed cycle number. When a battery cycle count exceeds 1200, the internal resistance change rate exceeds 20%, or the capacity decay exceeds 20%, and the SOH value is below 80%, it is identified as a battery requiring maintenance. It is automatically dispatched to area C of the pallet assembly, and a maintenance work order is generated. Batteries with an SOH value of 80%-90% are considered suboptimal and stored in area B, prioritized for allocation to light trucks with lower range requirements (daily average mileage ≤100km). Batteries with an SOH value above 90% are considered high-quality and stored in area A, prioritized for allocation to heavy trucks (daily average mileage ≥200km). The demand prediction model is built on an LSTM neural network, trained using the TensorFlow framework. It has 128 neurons in the input layer, 3 hidden layers with 64 neurons each (using ReLU activation function), and 1 neuron in the output layer (linear activation function). Model input features include battery swapping time-series data from the past 7 days (number of swaps per hour), weather conditions (sunny / cloudy / rainy, coded as 0 / 1 / 2), logistics order volume (ratio of daily orders to historical average), and holiday identifiers (0 / 1), which are standardized before being input into the model. The training dataset contains 180 days of historical data, divided into training and validation sets in an 8:2 ratio. The Adam optimizer (learning rate 0.001) was used for 500 iterations, reducing the mean squared error of the validation set to below 0.8. The accuracy of predicting battery swapping demand for the next 4 hours reached over 85%, with prediction results updated hourly. The dynamic allocation unit generates scheduling instructions based on the prediction results and real-time battery swapping requests (submitted via the park's app or vehicle terminal) using a priority algorithm: emergency orders (e.g., truck battery ≤ 20%) have a priority coefficient of 1.5, regular orders 1.0, and reserved orders (reserved more than 2 hours in advance) 0.8. When calculating the scheduling priority of each request, the priority coefficient is multiplied by (1 plus the remaining battery percentage), and the results are sorted from highest to lowest.Dispatch instructions are sent via industrial Ethernet (1Gbps transmission rate) to the battery gripping mechanism, lateral movement assembly, rotary assembly, and push assembly. RFID tags identify battery locations, enabling automated battery transfer from the pallet assembly storage area to the battery swapping station. During system operation, by dynamically adjusting battery storage strategies (e.g., pre-transferring 30% of spare batteries to a buffer area near the battery swapping station during peak hours), battery inventory turnover rate increases by 30% (from 5 times / month to 6.5 times / month). Battery waiting time during peak hours (8:00-10:00, 14:00-16:00) is reduced to less than 5 minutes, and storage space utilization reaches 90%, achieving optimal allocation of storage resources.

[0071] The dynamic fault-tolerant unit adopts a distributed monitoring architecture, deploying 28 sensors in key areas such as the battery gripping mechanism joints (6 joints), the push assembly motors (12 units), the communication assembly (4 modules), and sensor interfaces (6 locations) to monitor the equipment status in real time. Sensor data is aggregated to the fault-tolerant controller (using a PLC, model Siemens S7-1214C) via a CAN bus, with a sampling period of 10ms. The anomaly monitoring module performs real-time analysis of the collected data: the joint angle of the battery gripping mechanism is monitored by an absolute encoder (accuracy ±0.01°). When the deviation exceeds ±3mm (absolute threshold) for three consecutive cycles, it is determined to be a trajectory deviation; the push assembly motor current (measurement range 0-50A, accuracy ±0.1A) exceeds 120% of the rated value and lasts for 100ms, it is determined to be an overload; data from the communication assembly... A packet loss rate exceeding 5% (100 consecutive data packets) is considered a communication anomaly. Upon detecting a deviation, the trajectory correction unit immediately activates a PID control algorithm with a proportional gain of 5.0 (for adjusting speed), an integral gain of 0.1 (for eliminating steady-state error), and a derivative gain of 0.5 (for suppressing overshoot). The control output range is ±10V. Real-time correction is achieved by adjusting the output torque (maximum torque 300N·m) of the servo motors at each joint of the battery gripping mechanism. The correction response time is ≤50ms, and the deviation can be controlled within ±1mm within two cycles when the deviation is ≤10mm. The redundancy switching unit is equipped with dual battery gripping mechanism drive systems (main and backup PLCs, servo amplifiers) and communication assemblies (main 5G, backup WiFi 6), using a hot backup mode. The main and backup devices operate synchronously with a state difference of ≤1ms. When the main mechanism detects a fault signal three times consecutively (such as motor overcurrent, encoder malfunction, or communication timeout), the switching logic is triggered, and the activation of the backup mechanism and the transfer of control authority are completed within 0.5 seconds. During the switching process, the position commands of each joint of the battery gripping mechanism are kept unchanged to ensure that the end position fluctuation is ≤ ±0.5mm and there is no mechanical impact. The local control module is automatically activated when communication is interrupted (judgment criterion: communication interruption with the central control system exceeds 2 seconds). This module uses an embedded controller (STM32H743), pre-stores the battery swapping process (XML format) and battery parameter database (offline update cycle of 24 hours), and can independently perform basic operations: battery gripping (positioning accuracy ±10mm), pallet assembly compartment positioning (based on preset coordinates) and simple battery swapping actions (insertion and extraction force controlled at 300-500N) are completed through local sensors. After communication is restored (three consecutive normal communications are detected), the operation data is automatically synchronized to the central control system to ensure data integrity. In actual operation, the unit successfully handled 12 battery grabbing mechanism trajectory deviations (maximum deviation 8mm), 3 main mechanism fault switchings (the system operated normally after the switching), and 5 communication interruption events (the longest interruption time was 15 minutes), without causing any interruption to the battery swapping process, and the system availability reached 99.9%. The blockchain evidence storage unit is built on the Hyperledger Fabric 2.4 consortium blockchain, consisting of 7 nodes: battery swapping station operators, battery manufacturers (2), logistics companies (3), and regulatory agencies (1). Each node uses a server (configured with Intel Xeon E5-2680v4, 64GB RAM, and 4TB hard drive), interconnected via an encrypted dedicated line (100Mbps bandwidth). It employs the PBFT consensus algorithm (7 nodes, fault tolerance threshold of 2, consensus latency ≤500ms). The consortium blockchain writing unit packages key data from the battery swapping process and uploads it to the blockchain every 100ms. This data includes: positioning parameters (3D coordinates, attitude angle, timestamp), force control data (grabbing force, torque, claw distance), battery ID (unique code), operation time, device status code, etc. Each data entry is approximately 512 bytes in size. After encryption using the SHA-256 hash algorithm, it is organized into blocks (size controlled within 1MB). Each block includes a block header (pre-order hash, maker's hash, etc.). The system consists of a battery swapping module (including battery ID, timestamp, and block body, transaction data), with block generation taking approximately 2 seconds and on-chain data retention for 5 years. The smart contract module, written in Go, predefines three types of contracts: an operation traceability contract for chained storage of the entire battery swapping process data, supporting historical data queries via battery ID (query response time ≤ 1 second) or vehicle VIN code (associated with the data), including the time of each swap, operator (automatically identified by the system), and device status; a status confirmation contract for real-time synchronization of battery SOH value and ownership information (via smart contract address mapping), automatically sending maintenance reminders to battery manufacturer nodes when the battery SOH value falls below 80% (off-chain notification + on-chain notification); and a certificate generation contract for generating electronic certificates within 10 seconds of battery swapping completion, containing fields such as swapping time, battery parameters (capacity, SOH value), operator (system ID), and device number, in PDF / A format (compliant with long-term archiving standards), embedding on-chain hash values, and supporting offline verification (using dedicated tools to verify hash consistency). Within six months of system operation, a total of 1.2 million data entries were uploaded to the blockchain, reaching a block height of 12,000. The on-chain data storage space occupied was approximately 1.2TB. The electronic credentials were recognized during regulatory inspections, achieving reliable data storage and efficient traceability.

[0072] In summary, this application's embodiment, through a dynamic positioning module relying on multi-sensor fusion and precise algorithms, can still stably output ±5mm three-dimensional coordinates and ±0.5° attitude angle even with a truck parking deviation of ±30cm, achieving a positioning success rate of over 95%, providing a precise spatial benchmark for subsequent grasping; the adaptive battery grasping unit, targeting three mainstream battery models (50kWh, 80kWh, and 100kWh), achieves 100% model recognition and a 99.8% grasping success rate through six-dimensional force sensing, laser ranging, and a tiered force control strategy. It has operated for six months without battery damage, and the ±1mm hole alignment accuracy meets the requirements for automatic insertion and removal, balancing grasping safety and compatibility; the intelligent warehouse scheduling center achieves precise battery allocation through health-level hierarchical management, combined with LST... The M-model achieves over 85% accuracy in 4-hour demand forecasting, dynamically optimizing scheduling strategies to increase battery inventory turnover by 30% (from 5 times / month to 6.5 times / month), reducing peak-hour waiting time to within 5 minutes, and achieving 90% warehouse space utilization, significantly improving resource allocation efficiency. The dynamic fault-tolerant unit, with its distributed monitoring, rapid trajectory correction, and redundant switching design, successfully handled 12 trajectory deviations, 3 main mechanism failures, and 5 communication interruptions, ensuring 99.9% system availability and significantly reducing the impact of failures on the battery swapping process. The blockchain evidence storage unit, based on a 7-party node consortium blockchain, has accumulated 1.2 million data entries, enabling trusted traceability throughout the battery swapping process through smart contracts. Electronic certificates have gained regulatory approval, solidifying the foundation of data credibility. Overall, the system demonstrates outstanding effectiveness in improving battery swapping efficiency, ensuring equipment reliability, optimizing resource management, and enhancing data credibility, comprehensively meeting the diverse and high-frequency battery swapping needs of new energy logistics vehicles.

[0073] Next, referring to the accompanying drawings, a dynamic positioning unmanned energy storage battery swapping method is described according to an embodiment of this application.

[0074] like Figure 9 As shown, this dynamic positioning unmanned energy storage battery swapping method includes the following steps: In step S101, dynamic positioning data, battery parameters, and environmental vibration data are acquired.

[0075] It is understood that the embodiments of this application can accurately grasp the real-time positional relationship between the battery swapping equipment and the object to be swapped by acquiring dynamic positioning data, ensuring accurate docking of unmanned operation; acquiring battery parameters can clarify the requirements and compatibility of the battery to be swapped, ensuring stable operation of the equipment after battery swapping and optimizing battery management; acquiring environmental vibration data can monitor the stability of the operating environment in real time, avoid operational deviations or safety risks caused by vibration interference in advance, and improve the reliability and intelligence level of battery swapping.

[0076] In step S102, the positioning data, battery parameters and environmental vibration data are spatiotemporally synchronized and filtered to generate a battery swapping scenario dataset. Based on the battery swapping scenario dataset, the positioning accuracy and battery health level are evaluated by a sensor fusion algorithm, and the anomaly detection model is used to identify grasping deviations and robotic arm joint faults to obtain evaluation results and abnormal positioning data.

[0077] Among them, the battery swapping scenario dataset refers to a collection of multi-source dynamic data collected during the battery swapping process of electric vehicles or energy storage devices, including battery status, user behavior, environmental parameters, and battery swapping equipment operation information.

[0078] It is understood that the embodiments of this application utilize battery swapping scenario datasets and sensor fusion algorithms to accurately assess positioning accuracy and battery health levels, providing data for positioning calibration and efficient battery scheduling in unmanned battery swapping; at the same time, it helps the anomaly detection model effectively identify grasping deviations and robotic arm joint failures, providing early warnings of potential risks to ensure battery swapping safety, and optimizing the entire battery swapping process through data-driven optimization, thereby improving intelligent decision-making capabilities and operational reliability.

[0079] It should be noted that, based on the multi-source data (including dynamic positioning data, battery parameters, and environmental vibration data) collected in the battery swapping scenario dataset after spatiotemporal synchronization and filtering, the sensor fusion algorithm integrates information collected by different sensing devices (such as the position coordinates of the positioning module, the distance data of the lidar, and the state parameters of the battery sensor) to achieve data complementarity and error cancellation. In terms of positioning accuracy assessment, by combining dynamic positioning data and environmental vibration data, the impact of vibration interference and equipment drift on positioning is quantitatively analyzed, the deviation between the actual position and the target position is calculated, and it is determined whether the accuracy threshold for robotic arm grasping and battery docking is met. In terms of battery health level assessment, the battery's voltage, current, cycle count, and internal resistance parameters are integrated, and historical charging and discharging data are combined with current state characteristics to construct a health assessment model. This model quantifies the battery's degradation degree, endurance, and safety redundancy, and outputs accurate positioning accuracy level and battery health quantification results.

[0080] Health assessment model formula:

[0081] in, Battery health status; , , , , For parameter weights;

[0082] This represents the current actual usable capacity of the battery. This refers to the battery's rated capacity. This is the battery's rated internal resistance; This represents the current internal resistance of the battery. This is the cyclic decay coefficient; This refers to the number of battery cycles. This is the current battery voltage. This refers to the battery's nominal voltage. This refers to the charging and discharging current. The duration of the current; It is a current-time characteristic function; This is for residual correction.

[0083] For example, a smart battery swapping operator, relying on a battery swapping scenario dataset, integrated multi-dimensional data such as battery voltage / internal resistance / cycle count, robotic arm positioning coordinates, and environmental vibration waveforms. Through a sensor fusion algorithm, it: on the one hand, correlated dynamic positioning data with battery parameters, quantified the millimeter-level deviation of the robotic arm's grasping alignment to assess positioning accuracy, and simultaneously fused historical charge-discharge curves to construct a health model, outputting battery degradation levels and range redundancy; on the other hand, used an anomaly detection model to analyze the correlation characteristics between vibration data and positioning deviations, identifying potential faults such as robotic arm joint wear and grasping misalignment. This practice reduced battery swapping alignment error to ±0.5mm, increased battery health assessment accuracy to 92%, and provided early warnings of 37 equipment anomalies, driving the battery swapping process from "experience-based decision-making" to "data-driven" upgrading.

[0084] In this embodiment of the application, the sensor fusion algorithm formula is as follows:

[0085] in, Let be the fusion weights for the i-th sensor; Let be the measurement variance of the i-th sensor; The total number of sensors participating in the fusion; For summation index variables; This is the final estimated position after fusion; This represents the original position measurement value of the i-th sensor; The internal resistance of the battery; This refers to the change in voltage. This refers to the change in current. Vibrational energy density; The amplitude of vibration at a specific frequency point f; For the frequency differential component; Battery health status; , , , These are the regression coefficients; It is a temperature decay function; For temperature.

[0086] It is understood that the embodiments of this application integrate positioning data (multi-source positioning information from LiDAR, vision, and IMU), battery parameters (voltage, cycle count), and environmental vibration data to eliminate errors such as ranging deviation under strong light by LiDAR, feature loss under low light by vision, and temperature drift error by IMU. The consistency and reliability of the data are achieved through spatiotemporal synchronization and filtering. The accuracy of positioning and battery health level assessment are improved through cross-validation of multi-source data, providing more comprehensive input for the anomaly detection model, improving the sensitivity and accuracy of grasping deviation and robotic arm joint failure identification, providing data support for accurate decision-making and stable operation of the entire battery swapping process, and reducing operational risks caused by data distortion.

[0087] For example, in the actual operation of an energy storage battery swapping station, the sensor fusion algorithm achieves accurate assessment through multi-dimensional data collaboration: First, it fuses the 3D point cloud data of the battery from the lidar (distance accuracy ±2mm), the texture features of the battery interface from the vision camera (recognition error <0.5mm), and the motion posture data of the robotic arm from the IMU (angular velocity error <0.1° / s). After spatiotemporal synchronization (timestamp alignment error ≤1ms, coordinate transformation deviation <0.3mm) and Kalman filtering (filtering out high-frequency noise caused by vibration), a positioning dataset containing the battery position, posture, and motion trajectory is generated. At the same time, it incorporates the voltage, cycle count, and internal resistance data uploaded by the battery management system, combined with the workshop vibration amplitude (<0.2g) collected by the environmental vibration sensor. The algorithm calculates the battery health level (SOH assessment error ≤ 3%) through a weighted fusion model. Finally, it compares the fused positioning data with the preset path (deviation threshold ± 0.5 mm), identifies a 0.7 mm grasping offset caused by visual reflection, and combines the joint angle sensor data of the robotic arm (sampling frequency 1 kHz) to find an abnormal jamming of 0.3° in a certain joint through residual analysis. The whole process not only makes up for the limitations of a single sensor (such as the ranging deviation of LiDAR in strong light and the loss of visual features in low light), but also improves the positioning accuracy (fusion error ≤ 0.3 mm) and the anomaly identification efficiency (fault response time < 50 ms) through cross-validation of multi-source data, providing a highly reliable evaluation basis for battery swapping decisions.

[0088] In this embodiment of the application, the anomaly detection model formula is:

[0089] in, This is the current measurement value;

[0090] This is the average value under normal conditions; This represents the standard deviation under normal conditions. The degree of deviation after standardization; The parameter to be detected; It is the lower quartile; It is the upper quartile; It is a robust measure of data dispersion.

[0091] It is understood that the embodiments of this application accurately identify positional deviations exceeding thresholds (such as 0.7mm deviation) and abnormal states of robotic arm joints (such as 0.3° jamming) during the grasping process by analyzing abnormal patterns in data such as positioning deviations and robotic arm joint movements. This allows for real-time capture of potential fault risks during the battery swapping process, preventing battery collisions and interface damage caused by grasping deviations, or robotic arm jamming and shutdown caused by joint failures. This not only improves operational safety but also reduces fault diagnosis time, ensuring the continuous and stable operation of the battery swapping system. At the same time, it provides accurate abnormal location data for subsequent dynamic adjustments (such as trajectory correction and redundancy switching), enhancing anti-interference capabilities and fault tolerance.

[0092] In step S103, based on the evaluation results, combined with real-time force control feedback and historical battery swapping parameters, the gripper distance, locking force and robotic arm movement trajectory are dynamically adjusted, and basic battery swapping instructions are generated. Based on the basic battery swapping instructions, the weights of positioning accuracy, battery health and load requirements are allocated using the analytic hierarchy process. The basic battery swapping instructions are dynamically optimized according to the battery swapping priority to generate the target battery swapping scheme.

[0093] Among them, the Analytic Hierarchy Process (AHP) is a systematic analysis method that decomposes complex decision problems into levels such as objectives, criteria, and alternatives, determines the weight of each factor by pairwise comparisons, and then makes a comprehensive judgment to arrive at the optimal decision.

[0094] It is understood that the embodiments of this application, by using the analytic hierarchy process, can decompose key factors such as positioning accuracy, battery health, and load requirements into a clear hierarchical structure. By comparing each factor pairwise, the weight of each factor under different battery swapping priorities can be scientifically quantified, avoiding subjective decision-making bias, effectively balancing operational accuracy (positioning accuracy), battery reliability (health), and the immediate needs of the equipment (load), optimizing basic battery swapping instructions, generating target battery swapping solutions that are more in line with actual scenario needs, and improving the rationality of unmanned battery swapping decisions and execution efficiency.

[0095] It should be noted that the evaluation results (including a comprehensive judgment of key indicators such as positioning accuracy, battery health, size compatibility, and equipment operating status), combined with real-time force control feedback (using a six-dimensional force sensor to monitor the force and pressure distribution data of the robotic arm in contact with the battery in real time, ensuring that the magnitude and distribution of force during the grasping process meet safety standards, avoiding excessive force that could damage the battery casing or cause poor contact) and historical battery swapping parameters (covering the optimal claw distance and locking force range for different battery models, the efficient movement path of the robotic arm in various scenarios, and deviation corrections that occurred in previous battery swapping processes) and historical battery swapping parameters (covering the optimal claw distance and locking force range for different battery models, the efficient movement path of the robotic arm in various scenarios, and deviation corrections that occurred in previous battery swapping processes). Based on accumulated case data, dynamic multi-dimensional adjustments are made: For the gripper distance, real-time calibration is performed according to the actual battery size (including individual differences) to ensure precise matching between the gripper and the battery contact position; for the locking force, adaptive settings are made within an adjustable range of 50-200Nm based on the battery material hardness, shell load-bearing capacity, and real-time force feedback, ensuring a firm grip without detachment while preventing excessive tightening that could deform the shell; the robotic arm's motion trajectory is adjusted in real-time using positioning deviation compensation, obstacle avoidance, and historical optimal paths to ensure a smooth and precise process from gripping to docking. Finally, these adjustment parameters are integrated and transformed into specific execution commands, i.e., basic battery swapping commands, providing standardized and precise action guidelines for subsequent battery disassembly, transportation, and installation.

[0096] Analytic Hierarchy Process (AHP) formula:

[0097] in, For the judgment matrix; This is the weight vector; To determine the largest eigenvalue of a matrix; As a consistency indicator; To determine the order of a matrix; Consistency ratio; It is a random consistency indicator.

[0098] In step S104, battery removal and installation operations are performed according to the target battery swapping scheme. The battery swapping process is monitored in real time through a dynamic fault tolerance mechanism. When the trajectory deviation exceeds the threshold, the path is corrected. When the main mechanism fails, the backup robotic arm is switched. When communication is interrupted, local control is started. Key data of the battery swapping process is written into the consortium blockchain for storage and evidence, generating tamper-proof operation records and equipment status confirmation information.

[0099] Among them, the dynamic fault tolerance mechanism refers to the mechanism that monitors abnormal situations in real time during system operation and dynamically handles faults through emergency strategies such as trajectory correction, backup mechanism switching or local control to ensure the continuous and stable operation of the system.

[0100] It is understood that the embodiments of this application utilize a dynamic fault-tolerant mechanism to monitor the battery swapping process in real time during battery removal and installation. By promptly correcting trajectory deviations, switching to backup robotic arms, and activating local control, various anomalies are dynamically handled, effectively avoiding battery swapping interruptions caused by faults, improving operational reliability and stability, reducing downtime and operational risks. At the same time, the consortium blockchain ensures the traceability of battery swapping data and the confirmation of equipment status, guaranteeing the security and credibility of the entire process.

[0101] It should be noted that when the robotic arm performs battery grabbing, transfer, and installation actions, it compares the motion trajectory data collected in real time by LiDAR, vision camera, and IMU inertial measurement unit with the planned path. If the position or posture deviation exceeds the preset threshold (such as ±0.5mm required for battery swapping docking accuracy), the PID control algorithm is immediately triggered. Combined with feedback from the six-dimensional force sensor and dynamic positioning compensation information, the motion parameters (angular velocity, displacement compensation, etc.) of the robotic arm joints are dynamically adjusted to iteratively correct the path in real time. This ensures that the robotic arm accurately reaches the target position, avoids battery collisions and docking failures caused by trajectory deviations, and ensures the continuity and reliability of the battery swapping operation.

[0102] According to the embodiments of this application, a dynamic positioning unmanned energy storage battery swapping method is proposed. This method integrates data from LiDAR, visual cameras, and IMU through a dynamic positioning module, combined with spatiotemporal synchronization calibration and dynamic compensation algorithms. This effectively counteracts vibration offset and temperature drift effects, improving positioning accuracy and solving the problem of excessive deviation in traditional single-sensor positioning under complex environments. The adaptive battery gripping unit relies on six-dimensional force perception and laser deformation detection. Through a hierarchical force control strategy and impedance control algorithm, it achieves adaptive gripping of batteries of different models, even those with slight deformation. It dynamically adjusts the claw distance and locking force, avoiding battery damage caused by excessive clamping and addressing the pain point of poor compatibility with irregularly shaped batteries. The intelligent warehouse scheduling center is based on battery health assessment and LSTM. The demand forecasting model, combined with a priority algorithm, dynamically allocates batteries, prioritizing those with high health to handle peak demand. This not only improves battery turnover efficiency but also extends overall battery life and reduces inefficient idleness through scientific allocation. The dynamic fault-tolerant unit, through real-time anomaly monitoring and model-predictive control trajectory correction, along with redundant switching between primary and backup mechanisms and local offline control modes, responds quickly when the robotic arm deviation exceeds thresholds or communication is interrupted, minimizing the impact of faults and overcoming the limitation of traditional systems where a single module failure leads to shutdown. The blockchain evidence storage unit, through consortium blockchain distributed storage and smart contracts, immutably stores key information such as positioning parameters and force control data during the battery swapping process, achieving full-chain traceability and equipment status confirmation. This solves the problems of low positioning accuracy and poor compatibility in existing technologies.

[0103] The following will illustrate an adaptive adjustment method for railway backup power supply through a specific embodiment, such as... Figure 10 As shown, it includes: For a 200MW / 400MWh containerized energy storage power station, targeting the unmanned battery swapping needs of 10kV lithium iron phosphate battery packs (single cell specifications: L2000mm×W800mm×H400mm, weight 500kg), a dual-mode GPS and BeiDou satellite positioning module was selected for dynamic positioning data acquisition. The main reason is that this dual-mode module can fuse signals from two satellite systems, effectively improving the reliability and accuracy of positioning, especially in complex environments such as densely populated urban areas or canyons, reducing positioning errors caused by signal blockage from a single satellite system. Simultaneously, a high-precision inertial measurement unit (IMU) is used. The IMU can calculate position and attitude by measuring acceleration and angular velocity when satellite signals are briefly lost or interfered with, thus achieving continuous positioning output. The sampling frequency is set to 10Hz because a 10Hz sampling frequency meets the real-time requirements, promptly capturing changes in the position and attitude of the battery swapping vehicle and robotic arm, without causing excessive data volume due to an excessively high sampling frequency, thus avoiding increasing the burden on subsequent processing. This system collects real-time position and attitude information of the battery swapping vehicle and robotic arm, including longitude, latitude, altitude, heading angle, pitch angle, and roll angle. Battery parameters are acquired through CAN bus communication with the Battery Management System (BMS). The CAN bus, with its high reliability, real-time performance, and flexibility, is well-suited for data transmission in automotive and industrial control applications. This communication method is used to acquire battery parameters such as voltage, current, SOC (State of Charge), SOH (State of Health), and temperature. A 1-second sampling interval is set because these battery parameters change relatively slowly; a 1-second interval effectively reflects changes in battery status while avoiding data redundancy. Environmental vibration data is collected using vibration sensors installed on key parts of the battery swapping platform and robotic arm. Piezoelectric vibration sensors are selected due to their high sensitivity, fast response, and wide measurement range, enabling accurate capture of vibration signals during the battery swapping process. The sampling frequency was set to 100Hz. Since the movement of the robotic arm, the gripping and installation of the battery during the battery swapping process will generate relatively frequent vibrations, the 100Hz sampling frequency can capture subtle vibration changes and provide sufficient data support for subsequent anomaly detection.

[0104] Spatiotemporal synchronization and filtering of collected positioning data, battery parameters, and environmental vibration data are performed to generate an accurate and reliable dataset for battery swapping scenarios. Spatiotemporal synchronization is based on the system timestamp, aligning data from different sensors to the same timeline. Specifically, an independent clock is set for each sensor and calibrated using the system master clock to ensure that the time error of each sensor is controlled within 1ms. When data from each sensor is received, its timestamp is extracted and compared with the system master clock's timestamp. Interpolation or extrapolation is used to adjust the data from different sensors to the same time point, achieving spatiotemporal synchronization. The synchronized data is then processed using appropriate filtering algorithms. For positioning data, a Kalman filter is used to remove noise. The state equation and observation equation of the Kalman filter are set according to the characteristics of the positioning system. State variables include position, velocity, and acceleration, while observation variables are values ​​measured by satellite positioning and IMU. Process noise covariance and observation noise covariance are initially set based on the sensor performance parameters and adjusted using adaptive algorithms in practical applications to improve the filtering effect. After Kalman filtering, the noise level of the positioning data can be reduced by more than 30%, significantly improving positioning accuracy. A moving average filter is used to process the environmental vibration data, with a sliding window size of 10 sampling points. When a new data point enters the window, the oldest data point is removed, and the average of all data points within the window is calculated as the current filtering result. This filtering method effectively smooths high-frequency noise in the vibration data while preserving the data's trend, making the environmental vibration data smoother and more accurate. Through the above processing, a battery swapping scenario dataset containing positioning information, battery parameters, and environmental vibration information is generated, providing high-quality data support for subsequent evaluation and decision-making.

[0105] Based on the battery swapping scenario dataset, a sensor fusion algorithm is used to evaluate positioning accuracy and battery health level, and an anomaly detection model is used to identify grasping deviations and robotic arm joint malfunctions. For positioning accuracy evaluation, an Extended Kalman Filter (EKF) sensor fusion algorithm is employed. EKF can handle nonlinear systems and is suitable for evaluating positioning accuracy by combining data from multiple sensors such as satellite positioning and IMU. Positioning data from the battery swapping scenario dataset is used as input, and the EKF algorithm fuses information from different sensors to calculate the covariance matrix of the positioning error, thereby evaluating positioning accuracy. When the trace of the covariance matrix is ​​less than a set threshold (e.g., 0.5 m²), the positioning accuracy is considered to meet the battery swapping requirements. For battery health level evaluation, a particle filter sensor fusion algorithm is used. Particle filtering can handle non-Gaussian and nonlinear system models, making it suitable for assessing battery health status. Data such as voltage, current, temperature, and SOC from battery parameters are input into the particle filter algorithm. By establishing a battery aging model, the SOH of the battery is estimated, and the battery health level is then classified into four levels: excellent, good, average, and poor. Among the parameters, a State of Interest (SOH) greater than 80% is considered excellent, 60%-80% is good, 40%-60% is average, and less than 40% is poor. For identifying grasping deviations and robotic arm joint malfunctions, a neural network-based anomaly detection model is employed. First, a large amount of grasping data and robotic arm joint operation data from normal battery swapping processes are collected, including grasping position, force, joint angle, and rotational speed, as training samples. Then, a deep neural network model is constructed, containing an input layer, hidden layers, and an output layer. The input layer contains the collected data, and the output layer provides the judgment result of normal or abnormal conditions. The model is trained using the training samples, and the network parameters are adjusted to enable the model to accurately identify normal states. When real-time data from the battery swapping scenario dataset is input into the model, if the model outputs an anomaly, further analysis is conducted to determine whether it is a grasping deviation or a robotic arm joint malfunction. For grasping deviation, the deviation between the actual grasping position and the target position is calculated. When the deviation is greater than 5mm, it is determined that there is grasping deviation. For robotic arm joint failure, the variation patterns of joint angle, rotation speed and other data are analyzed. When fluctuations or sudden changes occur that exceed the normal range, it is determined that there is a joint failure.

[0106] Based on the evaluation results, and combining real-time force control feedback with historical battery swapping parameters, the gripper distance, locking force, and robotic arm trajectory are dynamically adjusted to generate basic battery swapping commands. These commands are then optimized using the analytic hierarchy process (AHP) to generate the target battery swapping scheme. Real-time force control feedback is acquired through force sensors mounted on the robotic arm's gripper, enabling real-time monitoring of the magnitude and direction of force during the gripping process. Historical battery swapping parameters include data such as gripper distance, locking force, robotic arm trajectory, and battery swapping time from past swapping processes. Based on the positioning accuracy evaluation results, when the positioning accuracy is high, the restrictions on the robotic arm's trajectory can be appropriately relaxed; when the positioning accuracy is low, the error range of the robotic arm's trajectory needs to be narrowed. Based on the battery health level, for batteries with low health levels, the locking force needs to be reduced to avoid damage to the battery. Combined with real-time force control feedback, when the gripping force is too large, the gripper distance is increased and the locking force is decreased; when the gripping force is too small, the gripper distance is decreased and the locking force is increased. These adjustments generate basic battery swapping commands. The AHP is then used to optimize these basic battery swapping commands. First, the hierarchical structure is determined. The target layer is for generating the optimal battery swapping scheme; the criteria layer includes positioning accuracy, battery health, and load requirements; and the scheme layer consists of different combinations of battery swapping command parameters. Next, a judgment matrix is ​​constructed. Expert scoring is used to determine the weights of each criterion layer factor relative to the target layer, and the weights of each scheme layer parameter combination relative to each criterion layer factor. For example, the weights for positioning accuracy, battery health, and load requirements are set to 0.4, 0.3, and 0.3, respectively. Then, the weight vector is calculated and a consistency check is performed to ensure the rationality of the judgment matrix. Finally, based on the weights, the comprehensive score of each scheme layer parameter combination is calculated, and the parameter combination with the highest score is selected as the optimized battery swapping command, generating the target battery swapping scheme.

[0107] The system performs battery removal and installation operations according to the target battery swapping plan. A dynamic fault-tolerant mechanism monitors the battery swapping process in real time, and key data from the process is stored on the consortium blockchain for notarization. During battery removal and installation, the robotic arm operates according to parameters such as the motion trajectory, gripper distance, and locking force specified in the target battery swapping plan. The dynamic fault-tolerant mechanism monitors various status information during the battery swapping process in real time. When a trajectory deviation exceeds a threshold (e.g., 3mm), a path correction algorithm is immediately activated to adjust the robotic arm's motion trajectory based on the deviation between the current and target positions. When the main mechanism malfunctions, it automatically switches to a backup robotic arm to ensure the continuous operation of the battery swapping process. When communication is interrupted, a local control mode is activated, and the robotic arm continues to perform operations according to the pre-stored battery swapping plan. Once communication is restored, the operation data is uploaded to the control system. Key data in the battery swapping process includes swapping time, battery number, location information, battery parameters, robotic arm operation parameters, and fault handling records. This data is stored on the consortium blockchain using Hyperledger Fabric technology, which features high performance, high security, and scalability. First, critical data is encrypted to ensure confidentiality. Then, data is written into blocks on the consortium blockchain via smart contracts. Each block contains the hash value of the previous block, forming an immutable chain structure. Simultaneously, multiple nodes in the consortium blockchain verify and reach consensus on the data, ensuring consistency and reliability. The generated immutable operation records and device status confirmation information can be used for subsequent traceability, auditing, and fault analysis.

[0108] In summary, this embodiment of the application, through a GPS and BeiDou dual-mode positioning module combined with a high-precision IMU and a 10Hz sampling frequency, not only solves the positioning blind spot problem in complex environments such as urban high-rise buildings and canyons, but also achieves continuous and accurate capture of the position and attitude of the battery swapping vehicle and the robotic arm, providing a fundamental guarantee for the grasping and positioning of large-size, heavy-load batteries. By using the CAN bus to collect core battery parameters at 1-second intervals, and combining this with 100Hz high-frequency sampling by a piezoelectric vibration sensor, the battery status and environmental vibration details are comprehensively grasped, laying a data foundation for subsequent analysis. In the data processing stage, the application of 1ms-level spatiotemporal synchronization, Kalman filtering (reducing positioning noise by more than 30%), and moving average filtering techniques significantly improves the accuracy and reliability of the dataset, ensuring the scientific nature of subsequent evaluation decisions. In the evaluation and identification stage, extended Kalman filtering provides quantitative evaluation of positioning accuracy, particle filtering provides accurate classification of battery health levels (SOH four-level grading standard), and a neural network-based anomaly detection model provides real-time identification of 5mm-level grasping deviations and robotic arm joint faults, enabling early warning and accurate judgment of potential risks during the battery swapping process. In the instruction generation and optimization phase, the gripper distance, locking force, and motion trajectory are dynamically adjusted by combining real-time force control feedback and historical parameters. Furthermore, the weight allocation of multiple criteria (positioning accuracy, battery health, and load requirements) is optimized using the analytic hierarchy process (AHP), enabling the battery swapping solution to adapt to different battery health states (e.g., reducing locking force for batteries with low health levels) while also considering positioning accuracy and load requirements, significantly improving the solution's adaptability and efficiency. In the execution and evidence storage phase, a dynamic fault-tolerance mechanism (trajectory deviation correction within 3mm, main / backup robotic arm switching, and local control during communication interruptions) effectively ensures the continuity and security of the battery swapping process. Meanwhile, the immutable evidence storage of the Hyperledger Fabric consortium blockchain enables traceability of data throughout the entire battery swapping process and confirmation of equipment status, meeting the stringent requirements of the energy storage industry for operational standardization and accountability. Overall, this method not only achieves fully unmanned battery swapping for large-scale energy storage, significantly reducing the cost and operational risks of manual intervention, but also significantly improves swapping efficiency, stability, and safety through a closed-loop design of high-precision sensing, intelligent decision-making, and reliable execution, providing key technical support for the large-scale operation of large-scale energy storage power stations.

[0109] Figure 11 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: The memory 1101, the processor 1102, and the computer program stored on the memory 1101 and executable on the processor 1102.

[0110] When the processor 1102 executes the program, it implements a dynamic positioning unmanned energy storage battery swapping method provided in the above embodiments.

[0111] Furthermore, electronic devices also include: Communication interface 1103 is used for communication between memory 1101 and processor 1102.

[0112] The memory 1101 is used to store computer programs that can run on the processor 1102.

[0113] The memory 1101 may include high-speed RAM (Random Access Memory) and may also include non-volatile memory, such as at least one disk storage.

[0114] If the memory 1101, processor 1102, and communication interface 1103 are implemented independently, then the communication interface 1103, memory 1101, and processor 1102 can be interconnected via a bus to complete communication between them. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 11 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0115] Optionally, in a specific implementation, if the memory 1101, processor 1102, and communication interface 1103 are integrated on a single chip, then the memory 1101, processor 1102, and communication interface 1103 can communicate with each other through an internal interface.

[0116] The processor 1102 may be a CPU (Central Processing Unit), an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of this application.

[0117] In the description of this specification, the references to "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0118] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0119] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0120] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any of the following techniques known in the art, or a combination thereof: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0121] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0122] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A dynamically positioned unmanned energy storage battery swapping system, characterized in that, include: The system includes a dynamic positioning module, an adaptive battery grasping unit, an intelligent warehouse scheduling center, a dynamic fault-tolerant unit, and a blockchain evidence storage unit; among which, The dynamic positioning module is used to output the three-dimensional coordinates and attitude angle of the target battery in real time through the fusion perception of multiple sensors and the spatiotemporal synchronous calibration algorithm, and to dynamically compensate for vibration offset and temperature drift effect. The adaptive battery gripping unit is used to monitor battery deformation in real time using a laser rangefinder based on positioning data and feedback from a six-dimensional force sensor. It automatically adjusts gripping parameters according to the battery model, drives the robotic arm to perform force-position hybrid control, and adopts a hierarchical force control strategy to adaptively adjust the claw distance and locking force. The intelligent warehouse scheduling center is used to dynamically generate battery swapping instructions and allocate batteries based on battery health, load demand and battery swapping priority through a demand prediction model. The dynamic fault-tolerant unit is used to monitor abnormalities in the battery swapping process in real time and trigger emergency strategies. The emergency strategies include: real-time correction of the robotic arm's motion trajectory; automatic switching to the backup mechanism when the trajectory deviation exceeds the absolute value threshold; and switching to local control mode when communication is interrupted. The blockchain evidence storage unit is used to write the positioning parameters, force control data, and key information of the battery ID during the battery swapping process into the consortium blockchain for operation traceability and equipment status confirmation.

2. The dynamically positioned unmanned energy storage battery swapping system according to claim 1, characterized in that, The dynamic positioning module includes a sensor fusion unit, a dynamic compensation unit, and a coordinate output unit. The sensor fusion unit is used to fuse data from lidar, vision camera, and IMU inertial measurement unit, and align the data through a spatiotemporal synchronization calibration algorithm. The dynamic compensation unit corrects vibration offset in real time based on a filtering algorithm and dynamically corrects IMU temperature drift error through a temperature drift compensation algorithm. The coordinate output unit is used to output three-dimensional coordinates and attitude angles in real time.

3. The unmanned energy storage battery swapping system with dynamic positioning according to claim 1, characterized in that, The adaptive battery gripping unit includes a six-dimensional force sensing module, a laser deformation detection unit, and a graded force control actuator. The six-dimensional force sensing module is used to provide real-time feedback of the gripping contact force; the laser deformation detection unit is used to monitor the deformation of the battery casing using a line laser; and the graded force control actuator is used to perform force-position hybrid control through an impedance control algorithm, adopting a three-level force control strategy to dynamically adjust the claw distance and automatically identify and adapt to the battery.

4. The unmanned energy storage battery swapping system with dynamic positioning according to claim 1, characterized in that, The intelligent warehouse scheduling hub includes a health assessment unit, a demand forecasting model, and a dynamic allocation unit. The health assessment unit calculates the state of health (SOH) value based on the number of battery cycles, the rate of change of internal resistance, and the amount of capacity decay, and classifies the health level. The demand forecasting model is based on an LSTM neural network and combines historical battery swapping data and real-time load demand to generate a battery swapping plan. The dynamic allocation unit uses a priority algorithm to generate a target battery allocation scheme and performs concurrent scheduling of multiple battery warehouses.

5. The unmanned energy storage battery swapping system with dynamic positioning according to claim 1, characterized in that, The dynamic fault-tolerant unit includes an anomaly monitoring module and a trajectory correction unit. The anomaly monitoring module is used to monitor the joint angle and positioning deviation of the robotic arm in real time. The trajectory correction unit uses a PID control algorithm to correct the motion path.

6. The unmanned energy storage battery swapping system with dynamic positioning according to claim 1, characterized in that, The dynamic fault-tolerant unit also includes a redundancy switching unit and a local control module. The redundancy switching unit is used to switch to the backup unit when the main unit fails and to enable the local control mode when communication is interrupted. The local control module is used to perform offline battery swapping and maintain basic operations when communication is interrupted.

7. The unmanned energy storage battery swapping system with dynamic positioning according to claim 1, characterized in that, The blockchain evidence storage unit includes a consortium blockchain writing unit and a smart contract module. The consortium blockchain writing unit is used to write key data into distributed nodes using a consensus algorithm. The smart contract module is used to automatically perform operation traceability and device status confirmation, generating tamper-proof electronic certificates.

8. A method for applying a dynamically positioned unmanned energy storage battery swapping system according to any one of claims 1-7, characterized in that, The method includes: Acquire dynamic positioning data, battery parameters, and environmental vibration data; The positioning data, battery parameters and environmental vibration data are spatiotemporally synchronized and filtered to generate a battery swapping scenario dataset. Based on the battery swapping scenario dataset, the positioning accuracy and battery health level are evaluated by a sensor fusion algorithm, and the anomaly detection model is used to identify grasping deviations and robotic arm joint faults to obtain evaluation results and abnormal positioning data. Based on the evaluation results, combined with real-time force control feedback and historical battery swapping parameters, the gripper distance, locking force and robotic arm movement trajectory are dynamically adjusted, and basic battery swapping instructions are generated. Based on the basic battery swapping instructions, the weights of positioning accuracy, battery health and load requirements are allocated using the analytic hierarchy process. The basic battery swapping instructions are dynamically optimized according to the battery swapping priority to generate a target battery swapping scheme. According to the target battery swapping scheme, battery removal and installation operations are performed. The battery swapping process is monitored in real time through a dynamic fault tolerance mechanism. When the trajectory deviation exceeds the threshold, the path is corrected. When the main mechanism fails, the backup robotic arm is switched. When communication is interrupted, local control is started. Key data of the battery swapping process is written into the consortium blockchain for storage and evidence, generating tamper-proof operation records and equipment status confirmation information.

9. A dynamically positioned unmanned energy storage battery swapping method according to claim 8, characterized in that, The formula for the sensor fusion algorithm is: ; in, Let be the fusion weights for the i-th sensor; Let be the measurement variance of the i-th sensor; The total number of sensors participating in the fusion; For summation index variables; This is the final estimated position after fusion; This represents the original position measurement value of the i-th sensor; The internal resistance of the battery; This refers to the change in voltage. This refers to the change in current. Vibrational energy density; The amplitude of vibration at a specific frequency point f; For the frequency differential component; Battery health status; , , , These are the regression coefficients; It is a temperature decay function; For temperature.

10. A dynamically positioned unmanned energy storage battery swapping method according to claim 8, characterized in that, The formula for the anomaly detection model is: ; in, This is the current measurement value; This is the average value under normal conditions; This represents the standard deviation under normal conditions. The degree of deviation after standardization; The parameter to be detected; It is the lower quartile; It is the upper quartile; It is a robust measure of data dispersion.

Images