A full life cycle electric quantity dynamic scheduling and charging management system for port agv fleet

Abstract

The application discloses a full life cycle electric quantity dynamic scheduling and charging management system for a port AGV fleet, and belongs to the technical field of port scheduling. The application establishes an individualized power consumption correction model for each AGV, extracts a mechanical wear bias coefficient through a historical power consumption ratio sequence, accurately predicts a remaining available operation time length in combination with a current operation state, greatly improves electric quantity estimation reliability, arranges charging time periods of each AGV to be charged staggered among multiple charging positions according to a peak-shaving principle, and real-time checks charging position occupation time sequences, selects a charging path with the lowest total power consumption under the constraint of an arrival time window by fusing a real-time traffic congestion index and a no-load power consumption model; meanwhile, task loads and charging strategies are allocated according to health level differences, global collaborative optimization of AGV fleet electric quantity and operation is realized, port operation efficiency, energy utilization rate and battery full life cycle service length are significantly improved, and operation and maintenance costs are reduced.

Description

Technical Field

[0001] This invention relates to the field of port scheduling technology, and in particular to a dynamic scheduling and charging management system for the entire lifecycle of AGV fleets in ports. Background Technology

[0002] With the continuous advancement of smart port construction, Automated Guided Vehicles (AGVs) have become core equipment for horizontal transportation at container terminals. Port AGV fleets typically need to operate 24 hours a day without interruption, and their power management directly affects overall operational efficiency and operating costs.

[0003] Currently, existing scheduling systems generally use a fixed threshold-triggered charging method, that is, a charging request is issued when the remaining power of the AGV falls below a certain preset value, and then the scheduling center allocates a charging slot. However, existing technology still has the following problems:

[0004] 1. The difference in power consumption between individual AGVs due to mechanical wear and battery aging was not fully considered, resulting in a large deviation in the prediction of the remaining working time, which could easily lead to breakdowns or premature charging.

[0005] 2. When multiple vehicles trigger charging at the same time, the lack of priority coordination and peak-shifting mechanism often leads to congestion at charging spots and queuing, which exacerbates the interruption of operations.

[0006] 3. The charging plan is disconnected from real-time road conditions and task urgency. The route planning does not couple power consumption with dynamic traffic conditions, making it difficult to ensure that vehicles with low power can safely reach the charging location.

[0007] 4. Battery health status is not included in scheduling decisions, and deteriorated batteries continue to perform high-load tasks, accelerating capacity decay and shortening the service life of the entire life cycle. Summary of the Invention

[0008] The purpose of this invention is to provide a dynamic scheduling and charging management system for the entire life cycle of AGV fleets in ports, so as to improve the operational continuity and energy utilization efficiency of port AGV fleets and solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A dynamic power scheduling and charging management system for the entire lifecycle of AGV fleets in ports, including:

[0011] The AGV operation prediction module is used to collect the real-time operating status and power parameters of each AGV. Based on the real-time remaining power and historical power consumption characteristics of each AGV, combined with the operation status identifier, it predicts the available operation time of each AGV.

[0012] The task scheduling module is used to obtain the real-time task queue, and allocate the corresponding transportation task to each AGV based on the available operation time corresponding to each AGV and the estimated operation time of each transportation task in the real-time task queue.

[0013] The charging prediction and scheduling module is used to obtain the task destination coordinates and estimated travel distance of each AGV according to the assigned transportation task, combine the current remaining power of each AGV and the available working time, predict the low power AGVs in the fleet that have charging needs, and generate a staggered charging plan based on priority rules.

[0014] The AGV path planning module is used to allocate corresponding charging positions to each AGV to be charged according to the off-peak charging schedule. It plans the optimal driving route from the current location to the charging position by combining the current traffic congestion of each path segment and the remaining power constraints of the AGV. At the same time, after the AGV completes the charging operation, it plans the optimal return route of the AGV from the charging position back to the work area by combining the real-time task queue of the port and the site road conditions.

[0015] Furthermore, the power parameters include remaining power, voltage, current, temperature, and battery health indicators. The real-time operating status includes the AGV's real-time position coordinates, driving speed, current acceleration, and operation status indicators, wherein the operation status indicators include heavy-load transportation status, empty-load driving status, waiting for instructions status, and charging status.

[0016] Furthermore, the specific process for predicting the available operating time of each AGV includes:

[0017] Based on the historical operating data of each AGV, the historical average power consumption rate of each AGV under heavy-load transportation and empty-load driving conditions is extracted.

[0018] Based on the ratio sequence of the actual power consumption of each AGV in each transportation task within a preset historical period to the theoretical power consumption calculated based on the basic power consumption model, the mechanical wear deviation coefficient corresponding to each AGV is determined.

[0019] Based on the mechanical wear deviation coefficient, the historical average power consumption rate under heavy-load transportation and empty-load driving states is corrected to obtain the first power consumption model of each AGV under heavy-load transportation state and the second power consumption model under empty-load driving state.

[0020] Obtain the current remaining power and current operation status identifier of each AGV, match the corresponding first power consumption model or second power consumption model according to the current operation status identifier of the AGV, and calculate the maximum available operation time for the AGV to continue to perform the transportation task under the current remaining power based on the matching result.

[0021] Furthermore, determining the mechanical wear deviation coefficient for each AGV also includes:

[0022] The ratio of the actual power consumption of each AGV in each transportation task within a preset historical period to the theoretical power consumption calculated based on the basic power consumption model is obtained, and an individual power consumption ratio sequence for each AGV is generated.

[0023] Remove abnormal ratios from the individual power consumption ratio sequence that deviate from the midpoint of the sequence by more than a preset deviation threshold to obtain a corrected ratio sequence;

[0024] Calculate the statistical median of the correction ratio sequence and use it as the initial mechanical wear deviation coefficient for the AGV;

[0025] After the AGV completes a preset number of transport tasks, the newly generated ratio is included in the correction ratio sequence. The mechanical wear deviation coefficient of the AGV is updated by performing a sliding weighted average calculation, which assigns the first weight to the ratio within the preset time window before the current time and the second weight to the ratio before the preset time window.

[0026] Furthermore, the charging prediction and scheduling module includes:

[0027] The charging demand prediction unit is used to calculate the first predicted remaining power of each AGV when it finishes the currently assigned transportation task and travels to the task destination, based on the task destination coordinates and estimated travel distance of each AGV and the available working time of each AGV.

[0028] The charging demand determination unit is used to compare the first predicted remaining power with the preset forced charging threshold and the preventive charging threshold, and to classify and mark each AGV in the fleet to generate a charging demand tag queue.

[0029] The off-peak charging unit is used to generate an off-peak charging plan table based on priority rules, according to the charging demand tag queue and the estimated task completion time of each AGV to be charged as the charging request initiation time benchmark.

[0030] The task data interaction unit is used to receive the task endpoint coordinates, estimated travel distance, and estimated task completion time of the transportation task assigned to each AGV. At the same time, it feeds back the generated off-peak charging plan and the occupancy time window of each charging position to the task scheduling module.

[0031] Furthermore, when assigning corresponding transportation tasks, the task scheduling module also includes:

[0032] Before each round of task allocation, the idle time window distribution and number of idle charging positions of each charging position within the current and future preset time windows are obtained based on the task data interaction unit.

[0033] Combining the real-time remaining power, operation status indicators and corresponding power consumption models of each candidate AGV, the estimated total power consumption of each candidate AGV in performing the proposed transportation task is estimated, and the second predicted remaining power after the candidate AGV completes the proposed task is calculated based on the estimated total power consumption.

[0034] The second predicted remaining power is compared with the preset preventive charging threshold. If the second predicted remaining power is lower than the preset preventive charging threshold, the current transportation task to be assigned to the candidate AGV is stopped.

[0035] Furthermore, the generation of the off-peak charging schedule based on priority rules specifically includes:

[0036] The system obtains the tag level, current remaining power, urgency level of the task being performed, and charging position occupancy time window fed back by the task data interaction unit for each AGV to be charged in the charging demand tag queue, and determines the order of charging start time for each AGV to be charged according to priority rules.

[0037] The priority rules include:

[0038] Based on charging demand tags, AGVs marked as vehicles requiring mandatory charging and awaiting dispatch are higher than AGVs marked as vehicles requiring preventative charging and awaiting dispatch.

[0039] When there are multiple vehicles waiting to be scheduled for forced charging at the same time, they are sorted in ascending order based on the first predicted remaining power of each AGV.

[0040] In the same priority AGV queue with the same charging demand label level and similar first predicted remaining power, they are sorted in descending order based on the urgency level of the currently executing task.

[0041] Based on the sorting results of charging start times and the distribution of idle time windows for each charging position, the estimated task completion time of each AGV to be charged is used as the base time for initiating charging requests, and the charging urgency of each AGV is matched with the idle time period of the charging position.

[0042] The charging time slots of each AGV to be charged are allocated to different idle time windows of different charging positions. The charging start time and charging time slot of all AGVs to be charged in the fleet are staggered on the global time axis to generate a staggered charging plan that adapts to the dynamic operation status of the fleet.

[0043] Furthermore, after generating the off-peak charging schedule adapted to the dynamic operating status of the fleet, the process also includes:

[0044] Iterate through the AGVs to be charged corresponding to each charging time window in the off-peak charging plan table, and extract the estimated arrival time of each AGV and the determined target charging position.

[0045] Obtain the real-time charging progress and remaining charging time of the AGV at the target charging position, determine the estimated charging completion time of the AGV, and combine the estimated arrival time of the next AGV waiting to be charged at the target charging position with the difference between the estimated arrival time and the preset clearing time window to obtain the latest idle end time of the charging position.

[0046] The timing relationship between the expected charging completion time and the latest idle deadline of the AGV is compared, and differentiated scheduling is determined based on the comparison results.

[0047] Furthermore, when the AGV path planning module plans the optimal travel path from the current location to the charging station, it performs the following steps:

[0048] Obtain the charging start time and target charging position number corresponding to the AGV to be charged in the off-peak charging plan table. Take the charging start time as the target arrival time and combine it with the preset arrival time window tolerance deviation to generate the planned arrival time window range of the AGV to be charged.

[0049] The current position of the AGV is matched to the current node of the road network topology as the starting point of the path, and the charging berth node corresponding to the target charging position is used as the ending point of the path. Based on the road network connectivity, all passable connecting paths from the starting point to the ending point are traversed to construct each candidate path at the current moment.

[0050] The real-time traffic congestion index and average speed of each path segment of the candidate path are obtained at the current moment. Combined with the second power consumption model of the AGV in the unloaded driving state, the segmented driving power consumption and total driving time of each candidate path in each path segment are calculated respectively.

[0051] From all candidate routes, select candidate routes whose total travel time falls within the planned arrival time window to generate a preliminary route set;

[0052] If the total travel time of all candidate routes is not within the planned arrival time window, the charging start time of the AGV will be adjusted backward by a preset step size, and the adjusted charging start time will be updated synchronously to the off-peak charging schedule.

[0053] Based on the initial path set, calculate the total power consumption of each candidate path in the initial path set, and select the candidate path with the lowest total power consumption as the first preferred path.

[0054] Furthermore, it also includes a battery health management module, used for:

[0055] Real-time collection of battery operation data for each AGV, including charge and discharge capacity of each charge and discharge cycle, temperature rise curve during charging, time-series change trend of battery internal resistance, and cumulative number of charge and discharge cycles.

[0056] The collected battery operation data is used to calculate the health evaluation sub-indicators of each AGV battery and to calculate the comprehensive health score;

[0057] The overall health score of each AGV battery is compared with the preset grading threshold to generate a health level label for each AGV battery. When the overall health score of each AGV battery is lower than the lower limit of the preset grading threshold, a battery replacement warning signal is generated and output.

[0058] At the same time, the health level tags of each AGV are fed back to the AGV operation scheduling module and the charging prediction scheduling module, and the corresponding task allocation strategy and charging strategy are configured according to the health level tags.

[0059] Compared with the prior art, the beneficial effects of the present invention are:

[0060] 1. This invention introduces a mechanical wear deviation coefficient to differentiate the power consumption model of each AGV under heavy load / no load conditions, and uses a sliding weighted average to dynamically update the deviation coefficient, eliminating the influence of temporary interference and individual differences, thus significantly improving the prediction accuracy of remaining available working time, providing a reliable basis for task allocation, and effectively avoiding work interruptions or premature recharging caused by inaccurate power estimation.

[0061] 2. This invention establishes a two-level marking system for forced charging and preventative charging. It integrates multiple priority rules such as remaining power and task urgency, and arranges the charging start time of each AGV in an interleaved manner on the global time axis. It also dynamically matches the charging position idle time window to achieve a time-balanced utilization of charging resources, completely avoids queuing congestion caused by multiple vehicles vying for positions, and ensures the continuity and timeliness of the operation process.

[0062] 3. This invention couples real-time traffic congestion index, remaining battery power safety redundancy, and power consumption model to plan the optimal charging path that meets the arrival time window and has the lowest total power consumption, and supports online replanning to cope with sudden changes in traffic conditions. At the same time, the battery health management module comprehensively evaluates the health of the AGV by integrating multi-dimensional degradation indicators, and configures differentiated task allocation and charging strategies for AGVs with different health levels accordingly, delaying the degradation of deteriorating batteries, and realizing closed-loop full life cycle management from power scheduling and path planning to battery maintenance, maximizing the overall service efficiency of the fleet. Attached Figure Description

[0063] Figure 1 This is a flowchart of the dynamic power scheduling and charging management system for the entire life cycle of the present invention. Detailed Implementation

[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] Please see Figure 1 The present invention provides the following technical solutions:

[0066] A dynamic power scheduling and charging management system for the entire lifecycle of AGV fleets in ports, including:

[0067] The AGV operation prediction module is used to collect the real-time operating status and power parameters of each AGV at a preset sampling frequency. Based on the real-time remaining power and historical power consumption characteristics of each AGV, combined with the operation status identifier, it predicts the available operation time of each AGV and pushes it to the task scheduling module and the charging prediction scheduling module simultaneously.

[0068] The power parameters include remaining power, voltage, current, temperature, and battery health indicators. Real-time operating status includes the AGV's real-time position coordinates, travel speed, current acceleration, and operating status indicators. The operating status indicators include heavy-load transport status, no-load travel status, waiting for instructions status, and charging status.

[0069] The task scheduling module is used to obtain the real-time task queue from the port operation management system, and assign the corresponding transportation task to each AGV based on the available operation time corresponding to each AGV and the estimated operation time of each transportation task in the real-time task queue.

[0070] The charging prediction and scheduling module is used to obtain the task destination coordinates and estimated travel distance of each AGV according to the assigned transportation task, combine the current remaining power of each AGV and the available working time, predict the low power AGVs in the fleet that have charging needs, and generate a staggered charging plan based on priority rules.

[0071] The AGV path planning module is used to allocate corresponding charging positions to each AGV to be charged according to the off-peak charging schedule. Combining the current traffic congestion of each path segment with the remaining power constraints of the AGV, it plans the optimal travel path from the current location to the charging position, ensuring that the total power consumption of the optimal planned path does not exceed the safety redundancy reserved by the AGV's current remaining power. The safety redundancy reserved by the remaining power is calculated and generated by the AGV operation prediction module in combination with battery health, battery temperature, and historical power consumption characteristics. At the same time, after the AGV completes the charging operation, it plans the optimal return path of the AGV from the charging position back to the operation area based on the real-time task queue of the port and the site road conditions, realizing the full life cycle management of AGV power scheduling and operation scheduling.

[0072] In this embodiment, by distinguishing between heavy-load and no-load operating conditions and based on the mechanical wear deviation coefficient, the power consumption model has individual adaptive correction capabilities, accurately tracking the long-term performance degradation of equipment, and significantly improving the reliability of the estimated available operating time and remaining power. Simultaneously, a tiered charging demand labeling and multi-dimensional priority ranking mechanism is established, staggering the charging start times on the global timeline and dynamically adapting to the connection sequence of vehicles charging and waiting to be charged, effectively resolving competition for charging resources and significantly improving the turnover efficiency of charging facilities. Real-time traffic congestion, segmented energy consumption constraints, and off-peak planning time windows are integrated and filtered to ensure that AGVs arrive on time and operate with low energy consumption, enhancing the stability of the scheduling order. The comprehensive health grading results are integrated into task allocation and charging strategies, automatically adapting task intensity and charging procedures for different health states, achieving proactive battery maintenance of degraded batteries, and effectively slowing down the overall battery degradation process of the fleet. This comprehensively improves the operating efficiency, safety, and long-term reliability of the port AGV fleet in continuous and complex operating scenarios.

[0073] In this embodiment, the specific process of predicting the available operating time of each AGV includes:

[0074] Based on the historical operating data of each AGV, including historical power parameters and historical real-time operating status, the historical average power consumption rate of each AGV under heavy-load transportation and empty-load driving conditions is extracted.

[0075] Based on the ratio sequence of the actual power consumption of each AGV in each transportation task within a preset historical period to the theoretical power consumption calculated based on the basic power consumption model, the mechanical wear deviation coefficient corresponding to each AGV is determined.

[0076] Based on the mechanical wear deviation coefficient, the historical average power consumption rate under heavy-load transportation and empty-load driving states is individually differentiated to obtain the first power consumption model of each AGV under heavy-load transportation state and the second power consumption model under empty-load driving state.

[0077] Obtain the current remaining power and current operation status identifier of each AGV, match the corresponding first power consumption model or second power consumption model according to the current operation status identifier of the AGV, and calculate the maximum available operation time that the AGV can continue to perform transportation tasks under the current remaining power based on the matching result and the corresponding power consumption model.

[0078] In this embodiment, the formula for calculating the maximum available time is as follows:

[0079] First power consumption model:

[0080]

[0081] Second power consumption model:

[0082]

[0083] In the formula, This represents the maximum available operating time under heavy load conditions. This represents the maximum available operating time under no-load conditions. Real-time remaining battery power of the AGV; The safety redundancy is the amount of power reserved for the preset safe return trip. This is the corrected real-time power consumption under heavy load, i.e., the output value of the first power consumption model. , This refers to the basic power consumption rate per unit time obtained from the historical heavy-load operation data of this AGV. This is the corrected real-time no-load power consumption rate, i.e., the output value of the second power consumption model. , This refers to the basic power consumption rate per unit time obtained from the historical no-load operation data of the AGV. This is the mechanical wear deviation coefficient.

[0084] In this embodiment, determining the mechanical wear deviation coefficient corresponding to each AGV further includes:

[0085] The ratio of the actual power consumption of each AGV in each transportation task within a preset historical period to the theoretical power consumption calculated based on the basic power consumption model is obtained, and an individual power consumption ratio sequence for each AGV is generated.

[0086] Remove abnormal ratios from the individual power consumption ratio sequence that deviate from the midpoint of the sequence by more than a preset deviation threshold to obtain a corrected ratio sequence;

[0087] Calculate the statistical median of the correction ratio sequence and use it as the initial mechanical wear deviation coefficient for the AGV;

[0088] After the AGV completes a preset number of transport tasks, the newly generated ratio is included in the correction ratio sequence. The mechanical wear deviation coefficient of the AGV is updated by performing a sliding weighted average calculation, which assigns the first weight to the ratio within the preset time window before the current time and the second weight to the ratio before the preset time window. The first weight is higher than the second weight.

[0089] In this embodiment, abnormal ratios that deviate from the median value by more than a preset deviation threshold are removed to eliminate temporary interference, such as power consumption data spikes caused by sudden headwinds or temporary uneven loads, thereby improving the robustness of the deviation coefficient. The statistical median value is used as the initial deviation coefficient to avoid the mean value being affected by extreme values, thus more robustly representing the long-term individual characteristics of the AGV. Based on the time-varying dynamic update mechanism, the sliding weighted average update reflects the gradual change of AGV mechanical wear over time, enabling the model to match the current physical state in real time.

[0090] In this embodiment, the mechanical wear deviation coefficient enables individualized adaptive correction of the power consumption model for each AGV, accurately distinguishing between heavy-load and no-load conditions and reflecting the gradual aging characteristics of the equipment accumulated during operation. This ensures that the predicted available operating time always matches the current physical state. Simultaneously, a preset safe return trip energy is introduced as a hard deduction item in the calculation of the maximum available operating time, ensuring that the prediction result naturally includes return trip energy redundancy. This guarantees the safe return capability of the AGV after task execution from the source, significantly improving the scientific nature and operational safety of the port AGV fleet's energy management throughout its entire life cycle.

[0091] In this embodiment, the charging prediction and scheduling module includes:

[0092] The charging demand prediction unit is used to calculate the first predicted remaining power of each AGV when it finishes the currently assigned transportation task and travels to the task destination, based on the task destination coordinates and estimated travel distance of each AGV and the available working time of each AGV.

[0093] The charging demand determination unit is used to compare the first predicted remaining power with preset forced charging thresholds and preventative charging thresholds, and to classify and label each AGV in the fleet to generate a charging demand tag queue, wherein:

[0094] If the first predicted remaining battery power is lower than the preset forced charging threshold, then the AGV is marked as a vehicle to be dispatched for forced charging.

[0095] If the first predicted remaining power is between the preset forced charging threshold and the preventive charging threshold, then the AGV is marked as a preventive charging vehicle to be dispatched.

[0096] If the first predicted remaining battery power is higher than the preventative charging threshold, then the AGV is marked as a vehicle that does not need charging.

[0097] The off-peak charging unit is used to generate an off-peak charging plan table containing the charging start time and expected charging duration of each AGV based on priority rules, according to the charging demand tag queue and the estimated task completion time of each AGV to be charged as the charging request initiation time benchmark.

[0098] The task data interaction unit is used to receive the task endpoint coordinates, estimated travel distance, and estimated task completion time of the transportation task assigned to each AGV. At the same time, it feeds back the generated off-peak charging plan and the occupancy time window of each charging position to the task scheduling module.

[0099] In this embodiment, the first predicted remaining power is obtained by subtracting the estimated power consumption required to perform the current task from the current remaining power. The estimated power consumption required to perform the current task is calculated by matching the corresponding first power consumption model or second power consumption model with the current working status identifier of the AGV and combining it with the estimated travel distance.

[0100] In this embodiment, the estimated task completion time of each AGV to be charged is calculated based on the task start time and the estimated operation duration.

[0101] In this embodiment, when the task scheduling module assigns the corresponding transportation task, it further includes:

[0102] Before each round of task allocation, the idle time window distribution and number of idle charging positions of each charging position within the current and future preset time windows are obtained based on the task data interaction unit.

[0103] Combining the real-time remaining power, operation status indicators and corresponding power consumption models of each candidate AGV, the estimated total power consumption of each candidate AGV in performing the proposed transportation task is estimated, and the second predicted remaining power after the candidate AGV completes the proposed task is calculated based on the estimated total power consumption.

[0104] The second predicted remaining power is compared with the preset preventive charging threshold. If the second predicted remaining power is lower than the preset preventive charging threshold, the current transportation task to be assigned to the candidate AGV will be stopped, thus eliminating the risk of insufficient power and no redundant power to support the return trip after the AGV task is executed.

[0105] In this embodiment, the heavy load / no load condition is determined by the current working status identifier of the candidate AGV, the corresponding first power consumption model / second power consumption model is called to obtain the real-time corrected power consumption rate, and the estimated total power consumption of the candidate task is calculated by combining the estimated working time of the candidate task to be assigned, and the estimated total power consumption of the candidate task is subtracted by the current remaining power to obtain the second predicted remaining power.

[0106] In this embodiment, charging demand prediction, graded handling, staggered scheduling and pre-task allocation safety verification are tightly coupled, realizing the transformation of the scheduling mode from passively discovering low power at the end of task execution to actively avoiding power shortage risks at the source of task allocation, thereby improving the energy safety margin and charging resource utilization efficiency of the port AGV fleet in continuous operation scenarios.

[0107] In this embodiment, the step of generating a staggered charging schedule based on priority rules specifically includes:

[0108] The system obtains the tag level, current remaining power, urgency level of the task being performed, and charging position occupancy time window fed back by the task data interaction unit for each AGV to be charged in the charging demand tag queue, and determines the order of charging start time for each AGV to be charged according to priority rules.

[0109] The priority rules include:

[0110] Based on charging demand tags, AGVs marked as vehicles requiring mandatory charging and awaiting dispatch are higher than AGVs marked as vehicles requiring preventative charging and awaiting dispatch.

[0111] When there are multiple vehicles waiting to be scheduled for charging, they are sorted in ascending order based on the first predicted remaining power of each AGV. The lower the first predicted remaining power, the higher the charging priority.

[0112] In the same priority AGV queue with the same charging demand label level and similar first predicted remaining power, the AGVs are sorted in descending order based on the urgency level of the currently executed task. The higher the urgency of the task, the higher the charging priority, ensuring that the AGVs corresponding to urgent tasks complete charging first and return to the work area in a timely manner.

[0113] Based on the sorting results of charging start times and the distribution of idle time windows of each charging position, the estimated task completion time of each AGV to be charged is used as the base time for initiating charging requests. The charging urgency of each AGV and the idle time of the charging position are matched, and the charging start time and charging duration of each AGV to be charged are staggered.

[0114] The charging time slots of each AGV to be charged are allocated to different idle time windows of different charging positions. The charging start time and charging time slot of all AGVs to be charged in the fleet are staggered on the global time axis to avoid charging position congestion and queuing problems caused by multiple vehicles charging at the same time. Finally, a staggered charging plan table adapted to the dynamic operation status of the fleet is generated.

[0115] In this embodiment, the charging position occupancy time window is calculated based on the expected arrival time of the AGV to be charged and the expected charging time.

[0116] In this embodiment, after generating the off-peak charging schedule adapted to the dynamic operating status of the fleet, the method further includes:

[0117] Iterate through the AGVs to be charged corresponding to each charging time window in the off-peak charging plan table, and extract the estimated arrival time of each AGV and the determined target charging position.

[0118] Obtain the real-time charging progress and remaining charging time of the AGV at the target charging position, determine the estimated charging completion time of the AGV, and combine the estimated arrival time of the next AGV waiting to be charged at the target charging position with the difference between the estimated arrival time and the preset clearing time window to obtain the latest idle end time of the charging position.

[0119] Compare the timing relationship between the expected charging completion time and the latest idle deadline of the AGV, and make differentiated scheduling decisions based on the comparison results;

[0120] If the estimated charging completion time of the in-place charging AGV is earlier than or equal to the latest idle deadline, it is determined that the charging position can achieve seamless charging. An early clearing instruction is generated and sent to the AGV path planning module to control the in-place charging AGV to leave the charging position in advance after charging is completed, ensuring that the charging position is completely idle before the next AGV waiting to be charged arrives.

[0121] If the estimated charging completion time of the AGV in place is later than the latest idle deadline, it is determined that there is a risk of charging position time sequence crowding, and forced clearing operation is prohibited. Based on the time sequence difference, the preset charging start time of the next AGV waiting to be charged is dynamically delayed, the time window arrangement of the staggered charging schedule is adjusted simultaneously, and the pre-bound charging period and path planning constraints of the AGV are updated.

[0122] In this embodiment, the preset clearing time window is a fixed time threshold pre-calibrated by the system based on the port charging position site layout, AGV departure speed and berth switching redundancy requirements. It is used to reserve the berth departure, site reversal and safety interval buffer time after the AGV in place has finished charging, and provides a timing judgment benchmark for seamless scheduling of charging positions.

[0123] In this embodiment, by pre-verifying the charging position timing, triggering dynamic clearing, and adaptive fine-tuning of the timing, dynamic matching of the charging position occupancy timing, AGV arrival timing, and pre-bound charging resources is achieved, eliminating charging position congestion, idle workstation waste, and charging timing conflicts, and ensuring the stability and full-cycle adaptability of the fleet's off-peak charging scheduling.

[0124] In this embodiment, when the AGV path planning module plans the optimal driving path from the current position to the charging position, it performs the following steps:

[0125] Obtain the charging start time and target charging position number corresponding to the AGV to be charged in the off-peak charging plan table. Take the charging start time as the target arrival time and combine it with the preset arrival time window tolerance deviation to generate the planned arrival time window range of the AGV to be charged.

[0126] The current position of the AGV is matched to the current node of the road network topology as the starting point of the path, and the charging berth node corresponding to the target charging position is used as the ending point of the path. Based on the road network connectivity, all passable connecting paths from the starting point to the ending point are traversed to construct each candidate path at the current moment.

[0127] The real-time traffic congestion index and average speed of each path segment of the candidate path are obtained at the current moment. Combined with the second power consumption model of the AGV in the unloaded driving state, the segmented driving power consumption and total driving time of each candidate path in each path segment are calculated respectively.

[0128] From all candidate routes, select candidate routes whose total travel time falls within the planned arrival time window to generate a preliminary route set;

[0129] If the total travel time of all candidate routes is not within the planned arrival time window, the charging start time of the AGV will be adjusted backward by a preset step size, and the adjusted charging start time will be updated synchronously to the off-peak charging schedule.

[0130] Based on the initial path set, the total power consumption of each candidate path in the initial path set is calculated. It is obtained by summing the segmented driving power consumption of each path segment, and the candidate path with the lowest total power consumption is selected as the first preferred path.

[0131] Among them, the road condition data of the unpassable path section in front of the AGV is monitored in real time according to the preset replanning frequency, and the updated real-time traffic congestion index is continuously obtained.

[0132] If the traffic congestion index of any unvisited segment in the current optimal planned route increases beyond the preset congestion deterioration threshold, and continuing along the current optimal route will fail to meet the planned arrival time window constraint, then the online route replanning mechanism will be triggered.

[0133] Starting from the current position of the AGV as updated in real time and ending at the allocated charging position, the candidate paths are retraced and a new optimal planning path is solved. The updated optimal path instruction is then sent to the AGV for execution, thereby achieving dynamic adaptive optimization of the charging path.

[0134] In this embodiment, the real-time average traffic speed of each path segment is obtained by aggregating and statistically analyzing the real-time driving speed data of the fleet AGV terminal. By filtering and smoothing the driving speeds of multiple vehicles within the same path segment and removing abnormal start-stop data, the real-time average traffic speed of the road segment is obtained. Based on the deviation ratio between the standard free-flow speed of the road segment and the real-time average traffic speed, the real-time traffic congestion index of each path segment is quantitatively calculated, realizing the full-area, dynamic, and refined perception of the traffic status of the port area road network, and providing real-time traffic data support for charging path power consumption prediction, duration verification, and online replanning.

[0135] In this embodiment, a battery health management module is also included, used for:

[0136] Real-time collection of battery operation data for each AGV, including charge and discharge capacity of each charge and discharge cycle, temperature rise curve during charging, time-series change trend of battery internal resistance, and cumulative number of charge and discharge cycles.

[0137] The collected battery operation data was used to calculate the health evaluation sub-indicators for each AGV battery across various dimensions, including capacity retention rate, internal resistance growth rate, abnormal temperature rise rate index, voltage consistency deviation, and charge / discharge efficiency decay rate. The comprehensive health score was then calculated using the following weighted fusion model:

[0138]

[0139] In the formula, Assess the overall health of the battery; This is a normalized sub-index of capacity retention. This is a normalized sub-index for charge / discharge efficiency. This is a normalized sub-index of the internal resistance growth rate; This is a normalized sub-index for the abnormal rate of temperature rise; This is a sub-index for normalized voltage consistency deviation; , , , , As the basic weights for each dimension; , , The internal resistance, temperature rise, and voltage deviation penalty coefficients are used; the capacity and charge / discharge efficiency are positive gain terms, and the higher the index value, the higher the health score. , , This represents the degree of degradation. Combined with the penalty coefficient, which amplifies the impact of degradation, even slight battery degradation can significantly lower the overall health score.

[0140] The overall health score of each AGV battery is compared with the preset grading threshold to generate a health level label for each AGV battery, including four levels: healthy battery, sub-healthy battery, deteriorated battery, and retired battery. When the overall health score of each AGV battery is lower than the lower limit of the preset grading threshold, a battery replacement warning signal is generated and output to remind the operation and maintenance personnel to replace and maintain the battery.

[0141] At the same time, the health level tags of each AGV are fed back to the AGV operation scheduling module and the charging prediction scheduling module, and the corresponding task allocation strategy and charging strategy are configured according to the health level tags.

[0142] Among them, healthy batteries are suitable for all working conditions; sub-healthy batteries are suitable for short and medium distance operations; deteriorated batteries are only suitable for short distance light load operations, and the estimated working time of their assigned tasks does not exceed the preset time threshold; retired batteries stop being assigned new tasks and generate a battery replacement warning signal.

[0143] Healthy batteries are charged according to standard procedures; sub-healthy batteries have their charging current rate reduced and charging time extended; degraded batteries are charged and discharged in a shallow charge-discharge strategy, with their maximum SOC (State of Charge) lowered to 85%-90% of the nominal value, while the remaining capacity threshold for triggering forced charging is raised to 115%-120% of the nominal value.

[0144] In this embodiment, by collecting real-time battery operation data of each AGV, the battery status is quantitatively evaluated, and a weighted fusion model with a penalty coefficient is introduced to calculate the comprehensive health score. This avoids high-loss heavy-load operations and full-charge-discharge conditions, enabling targeted maintenance of degraded batteries, extending battery life, and completing the full life cycle collaborative management of fleet battery status, power scheduling, and operation scheduling. This effectively slows down the overall battery performance degradation process of the fleet and significantly improves the service life and operational safety of the battery throughout its entire life cycle.

[0145] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A dynamic power scheduling and charging management system for the entire lifecycle of AGV fleets in ports, characterized in that, include: The AGV operation prediction module is used to collect the real-time operating status and power parameters of each AGV. Based on the real-time remaining power and historical power consumption characteristics of each AGV, combined with the operation status identifier, it predicts the available operation time of each AGV. The task scheduling module is used to obtain the real-time task queue, and allocate the corresponding transportation task to each AGV based on the available operation time corresponding to each AGV and the estimated operation time of each transportation task in the real-time task queue. The charging prediction and scheduling module is used to obtain the task destination coordinates and estimated travel distance of each AGV according to the assigned transportation task, combine the current remaining power of each AGV and the available working time, predict the low power AGVs in the fleet that have charging needs, and generate a staggered charging plan based on priority rules. The AGV path planning module is used to allocate corresponding charging positions to each AGV to be charged according to the off-peak charging schedule. It plans the optimal driving route from the current location to the charging position by combining the current traffic congestion of each path segment and the remaining power constraints of the AGV. At the same time, after the AGV completes the charging operation, it plans the optimal return route of the AGV from the charging position back to the work area by combining the real-time task queue of the port and the site road conditions.

2. The dynamic power scheduling and charging management system for the entire lifecycle of a port AGV fleet as described in claim 1, characterized in that, The power parameters include remaining power, voltage, current, temperature, and battery health indicators. The real-time operating status includes the AGV's real-time position coordinates, driving speed, current acceleration, and operation status indicators, including heavy-load transportation status, no-load driving status, waiting for instructions status, and charging status.

3. The dynamic power scheduling and charging management system for the entire lifecycle of port AGV fleets as described in claim 2, characterized in that, The specific process for predicting the available operating time of each AGV includes: Based on the historical operating data of each AGV, the historical average power consumption rate of each AGV under heavy-load transportation and empty-load driving conditions is extracted. Based on the ratio sequence of the actual power consumption of each AGV in each transportation task within a preset historical period to the theoretical power consumption calculated based on the basic power consumption model, the mechanical wear deviation coefficient corresponding to each AGV is determined. Based on the mechanical wear deviation coefficient, the historical average power consumption rate under heavy-load transportation and empty-load driving states is corrected to obtain the first power consumption model of each AGV under heavy-load transportation state and the second power consumption model under empty-load driving state. Obtain the current remaining power and current operation status identifier of each AGV, match the corresponding first power consumption model or second power consumption model according to the current operation status identifier of the AGV, and calculate the maximum available operation time for the AGV to continue to perform the transportation task under the current remaining power based on the matching result.

4. The dynamic power scheduling and charging management system for the entire lifecycle of port AGV fleets as described in claim 3, characterized in that, The determination of the mechanical wear deviation coefficient for each AGV also includes: The ratio of the actual power consumption of each AGV in each transportation task within a preset historical period to the theoretical power consumption calculated based on the basic power consumption model is obtained, and an individual power consumption ratio sequence for each AGV is generated. Remove abnormal ratios from the individual power consumption ratio sequence that deviate from the midpoint of the sequence by more than a preset deviation threshold to obtain a corrected ratio sequence; Calculate the statistical median of the correction ratio sequence and use it as the initial mechanical wear deviation coefficient for the AGV; After the AGV completes a preset number of transport tasks, the newly generated ratio is included in the correction ratio sequence. The mechanical wear deviation coefficient of the AGV is updated by performing a sliding weighted average calculation, which assigns the first weight to the ratio within the preset time window before the current time and the second weight to the ratio before the preset time window.

5. The dynamic power scheduling and charging management system for the entire lifecycle of a port AGV fleet as described in claim 1, characterized in that, The charging prediction and scheduling module includes: The charging demand prediction unit is used to calculate the first predicted remaining power of each AGV when it finishes the currently assigned transportation task and travels to the task destination, based on the task destination coordinates and estimated travel distance of each AGV and the available working time of each AGV. The charging demand determination unit is used to compare the first predicted remaining power with the preset forced charging threshold and the preventive charging threshold, and to classify and mark each AGV in the fleet to generate a charging demand tag queue. The off-peak charging unit is used to generate an off-peak charging plan table based on priority rules, according to the charging demand tag queue and the estimated task completion time of each AGV to be charged as the charging request initiation time benchmark. The task data interaction unit is used to receive the task endpoint coordinates, estimated travel distance, and estimated task completion time of the transportation task assigned to each AGV. At the same time, it feeds back the generated off-peak charging plan and the occupancy time window of each charging position to the task scheduling module.

6. The dynamic power scheduling and charging management system for the entire lifecycle of a port AGV fleet as described in claim 5, characterized in that, When the task scheduling module assigns the corresponding transportation tasks, it also includes: Before each round of task allocation, the idle time window distribution and number of idle charging positions of each charging position within the current and future preset time windows are obtained based on the task data interaction unit. Combining the real-time remaining power, operation status indicators and corresponding power consumption models of each candidate AGV, the estimated total power consumption of each candidate AGV in performing the proposed transportation task is estimated, and the second predicted remaining power after the candidate AGV completes the proposed task is calculated based on the estimated total power consumption. The second predicted remaining power is compared with the preset preventive charging threshold. If the second predicted remaining power is lower than the preset preventive charging threshold, the current transportation task to be assigned to the candidate AGV is stopped.

7. The dynamic power scheduling and charging management system for the entire lifecycle of a port AGV fleet as described in claim 6, characterized in that, The generation of the off-peak charging schedule based on priority rules specifically includes: The system obtains the tag level, current remaining power, urgency level of the task being performed, and charging position occupancy time window fed back by the task data interaction unit for each AGV to be charged in the charging demand tag queue, and determines the order of charging start time for each AGV to be charged according to priority rules. The priority rules include: Based on charging demand tags, AGVs marked as vehicles requiring mandatory charging and awaiting dispatch are higher than AGVs marked as vehicles requiring preventative charging and awaiting dispatch. When there are multiple vehicles waiting to be scheduled for forced charging at the same time, they are sorted in ascending order based on the first predicted remaining power of each AGV. In the same priority AGV queue with the same charging demand label level and similar first predicted remaining power, they are sorted in descending order based on the urgency level of the currently executing task. Based on the sorting results of charging start times and the distribution of idle time windows for each charging position, the estimated task completion time of each AGV to be charged is used as the base time for initiating charging requests, and the charging urgency of each AGV is matched with the idle time period of the charging position. The charging time slots of each AGV to be charged are allocated to different idle time windows of different charging positions. The charging start time and charging time slot of all AGVs to be charged in the fleet are staggered on the global time axis to generate a staggered charging plan that adapts to the dynamic operation status of the fleet.

8. The dynamic power scheduling and charging management system for the entire lifecycle of a port AGV fleet as described in claim 7, characterized in that, After generating the off-peak charging schedule adapted to the dynamic operating status of the fleet, the following is also included: Iterate through the AGVs to be charged corresponding to each charging time window in the off-peak charging plan table, and extract the estimated arrival time of each AGV and the determined target charging position. Obtain the real-time charging progress and remaining charging time of the AGV at the target charging position, determine the estimated charging completion time of the AGV, and combine the estimated arrival time of the next AGV waiting to be charged at the target charging position with the difference between the estimated arrival time and the preset clearing time window to obtain the latest idle end time of the charging position. The timing relationship between the expected charging completion time and the latest idle deadline of the AGV is compared, and differentiated scheduling is determined based on the comparison results.

9. The dynamic power scheduling and charging management system for the entire lifecycle of a port AGV fleet as described in claim 1, characterized in that, When the AGV path planning module plans the optimal travel path from the current location to the charging station, it performs the following steps: Obtain the charging start time and target charging position number corresponding to the AGV to be charged in the off-peak charging plan table. Take the charging start time as the target arrival time and combine it with the preset arrival time window tolerance deviation to generate the planned arrival time window range of the AGV to be charged. The current position of the AGV is matched to the current node of the road network topology as the starting point of the path, and the charging berth node corresponding to the target charging position is used as the ending point of the path. Based on the road network connectivity, all passable connecting paths from the starting point to the ending point are traversed to construct each candidate path at the current moment. The real-time traffic congestion index and average speed of each path segment of the candidate path are obtained at the current moment. Combined with the second power consumption model of the AGV in the unloaded driving state, the segmented driving power consumption and total driving time of each candidate path in each path segment are calculated respectively. From all candidate routes, select candidate routes whose total travel time falls within the planned arrival time window to generate a preliminary route set; If the total travel time of all candidate routes is not within the planned arrival time window, the charging start time of the AGV will be adjusted backward based on the preset step size, and the adjusted charging start time will be updated synchronously to the off-peak charging plan table. Based on the initial path set, calculate the total power consumption of each candidate path in the initial path set, and select the candidate path with the lowest total power consumption as the first preferred path.

10. The dynamic scheduling and charging management system for the entire lifecycle of AGV fleets in ports as described in claim 1, characterized in that, It also includes a battery health management module, used for: Real-time collection of battery operation data for each AGV, including charge and discharge capacity of each charge and discharge cycle, temperature rise curve during charging, time-series change trend of battery internal resistance, and cumulative number of charge and discharge cycles. The collected battery operation data is used to calculate the health evaluation sub-indicators of each AGV battery and to calculate the comprehensive health score; The overall health score of each AGV battery is compared with the preset grading threshold to generate a health level label for each AGV battery. When the overall health score of each AGV battery is lower than the preset lower limit of the grading threshold, a battery replacement warning signal is generated and output. At the same time, the health level tags of each AGV are fed back to the AGV operation scheduling module and the charging prediction scheduling module, and the corresponding task allocation strategy and charging strategy are configured according to the health level tags.

Images