Multi-objective intelligent scheduling system for new energy bus battery quick repair station

Abstract

The application relates to the technical field of operation and maintenance management, and discloses a multi-target intelligent scheduling system for a new energy bus battery quick repair station, which comprises the following steps: continuously acquiring bus state data and dimensionally reducing and mapping the bus state data into an equivalent risk energy pool; dynamically quantifying and jointly releasing power in combination with station operation data; constructing a time-space interference repulsion index and deducing actual expected time consumption; extracting a cell voltage drop rate, comparing a global average, calculating a self-adaptive priority multiplier, and finally dimensionally reducing and constructing an Euclidean space cost functional of the multiple indexes, so that an optimal distribution matrix is obtained through optimization and solving, and a scheduling instruction is output. The application can effectively avoid scheduling deadlock, and improve maintenance flow efficiency and station comprehensive utilization rate.

Description

Technical Field

[0001] This invention relates to the field of operation and maintenance management technology, and more specifically, to a multi-objective intelligent scheduling system for quick repair stations of batteries for new energy buses. Background Technology

[0002] To meet the demands of high-load operation, the power battery packs of new energy buses are typically large and densely arrayed. During rapid battery repairs, to accommodate the extra-long vehicle body, repair bays often employ a compact, side-by-side layout. Furthermore, for overall construction considerations, multiple compact bays typically share the same high-power refrigerant piping network and heavy-duty microgrid at the bottom. After prolonged high-load operation, the uneven heat accumulated inside the massive battery pack generates significant deformation stress, causing the chassis fastening bolts and high-voltage interfaces to become tightly sealed. Before any actual repair intervention, cooling must be injected through the shared bottom piping network, and a micro-lifting mechanism must be used to slowly release this stress.

[0003] Traditional scheduling algorithms or systems typically employ lookup tables, treating maintenance time as a static constant. When the scheduling engine, in an effort to maximize overall workstation utilization, densely allocates multiple buses with severe overheating faults to adjacent, compact workstations, it instantly and significantly depletes the pressure drop of the shared coolant in that area. This drastically reduces the stress-relieving capacity of each workstation. Simultaneously, the dissipated heat easily intrudes into adjacent operating spaces, creating spatial obstruction and interference effects. This interference means that the actual maintenance time for a single bus is no longer constant but increases non-linearly due to the allocation of adjacent workstations. Because traditional scheduling systems cannot anticipate and quantify this spatiotemporal obstruction and interference effect, their scheduling expectations become severely disconnected from actual operation. Ultimately, this not only prolongs the overall waiting time for vehicles but also easily triggers under-level coolant depletion and localized allocation stagnation. Summary of the Invention

[0004] This invention provides a multi-objective intelligent scheduling system for fast repair stations of batteries in new energy buses, which solves the technical problems mentioned in the background art.

[0005] This invention provides a multi-objective intelligent scheduling system for quick repair bays of new energy bus batteries, applied to a quick repair bay array including a queuing area and multiple quick repair bays, configured to execute:

[0006] Obtain the vehicle status data and queuing time of each passenger vehicle waiting in the waiting queue in the queuing area, and map the passenger vehicle status data into an equivalent hazard energy pool; Obtain the station operation data of each of the quick repair stations and quantify the corresponding dynamic joint release power; For any of the aforementioned passenger vehicles to be repaired, the vehicle is assigned to any target workstation. Based on the dynamic joint release power of the equivalent hazardous energy pool and other surrounding workstations besides the target workstation, a spatiotemporal interference repulsion index is constructed. By combining the equivalent hazardous energy pool, the dynamic joint release power corresponding to the target workstation, and the spatiotemporal interference repulsion index, the actual expected time consumption is deduced. Extract the cell voltage drop rate of the bus to be repaired and calculate the adaptive priority multiplier by comparing it with the global mean of the waiting queue; By combining the adaptive priority multiplier, the queuing time, and the actual expected time, a Euclidean spacetime cost functional is constructed, and the optimal allocation matrix is ​​obtained by solving it. Based on the optimal allocation matrix superimposed with vehicle motion parameter inversion, a spatiotemporal arrival scheduling command is output to the bus to be repaired.

[0007] The beneficial effects of this invention include: by uniformly reducing the state data of buses to an equivalent hazardous energy pool, and constructing a spatiotemporal interference repulsion index by combining the dynamic joint release power of workstations, the degree of spatial obstruction interference between adjacent compact workstations can be quantified in real time, thereby more accurately predicting the actual expected time consumption of nonlinearity; at the same time, by introducing an adaptive priority multiplier based on the cell voltage drop rate and constructing an Euclidean spatiotemporal cost functional, the system achieves dual-objective overall optimization of vehicle queuing delay and workstation operation status while avoiding fixed preset constants, effectively avoiding the phenomenon of shared pipeline refrigerant squeeze and circulation stagnation caused by the dense allocation of multiple high-temperature fault vehicles to adjacent workstations, and effectively improving the circulation efficiency of new energy buses and the comprehensive resource utilization rate of the fast repair array. Attached Figure Description

[0008] Figure 1 This is a flowchart of the multi-objective intelligent scheduling system for the fast repair station of new energy bus batteries according to the present invention. Detailed Implementation

[0009] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0010] like Figure 1 As shown, a multi-objective intelligent scheduling system for quick repair bays of new energy bus batteries is applied to a quick repair bay array that includes a queuing area and multiple quick repair bays, and is configured to execute: Obtain the vehicle status data and queuing time of each passenger vehicle waiting in the waiting queue in the queuing area, and map the passenger vehicle status data into an equivalent hazard energy pool; Obtain the station operation data of each of the quick repair stations and quantify the corresponding dynamic joint release power; For any of the aforementioned passenger vehicles to be repaired, the vehicle is assigned to any target workstation. Based on the dynamic joint release power of the equivalent hazardous energy pool and other surrounding workstations besides the target workstation, a spatiotemporal interference repulsion index is constructed. By combining the equivalent hazardous energy pool, the dynamic joint release power corresponding to the target workstation, and the spatiotemporal interference repulsion index, the actual expected time consumption is deduced. Extract the cell voltage drop rate of the bus to be repaired and calculate the adaptive priority multiplier by comparing it with the global mean of the waiting queue; By combining the adaptive priority multiplier, the queuing time, and the actual expected time, a Euclidean spacetime cost functional is constructed, and the optimal allocation matrix is ​​obtained by solving it. Based on the optimal allocation matrix superimposed with vehicle motion parameter inversion, a spatiotemporal arrival scheduling command is output to the bus to be repaired.

[0011] This system is applied to a quick repair station array containing a queuing area and M quick repair stations, where M is a positive integer. The queuing area is used to park new energy buses awaiting repair. Each quick repair station is equipped with an independent lifting mechanism, a refrigerant pipeline branch, and a sensor acquisition terminal. All stations share the main refrigerant pipeline network and the industrial control bus. The system runs on an industrial control computer or a cloud computing cluster. The underlying data acquisition terminals include an on-board battery management system, multi-parameter sensor terminals for each station, a vehicle identification device in the queuing area, and an automated guided vehicle (AGV) control system. All data acquisition terminals achieve clock synchronization via a network time protocol, with a time synchronization error not exceeding 10% of the sampling period. The sampling period is uniformly set to 1 second and can be adjusted within the range of 0.1 seconds to 60 seconds according to the accuracy requirements of the site. The system scheduling trigger mechanism adopts a combination of event triggering and timed triggering. Real-time scheduling is triggered when a passenger car to be repaired enters the queuing area, the work station completes its work and is released, or the vehicle status data shows abnormal fluctuations. When there is no event trigger, rolling optimization scheduling is performed every 30 seconds. During rolling optimization, vehicles that have already entered the site and started work are not included in the scheduling queue, and vehicles that have been assigned but have not yet entered the site can be reassigned according to the real-time working conditions.

[0012] The system collects data on the battery mass, specific heat capacity, real-time battery temperature, ambient temperature, equivalent stiffness, deformation displacement, real-time voltage, leakage current, and arrival time of each passenger vehicle waiting in the queue within the queuing area. The battery mass is the nominal total mass of the power battery pack of the bus to be repaired, in kilograms, and is taken from the vehicle's factory technical parameters; the specific heat capacity is the constant-pressure specific heat capacity of the battery cell, in joules per kilogram (Kelvin), using the fixed value nominally stated by the battery manufacturer. If the battery management system can provide dynamic specific heat capacity parameters that change with cell temperature, then the dynamic value corresponding to the real-time battery temperature is used; the real-time battery temperature is the average temperature collected by eight temperature sensors evenly distributed within the battery pack, in Kelvin. In practical applications, the number of temperature sensors is not limited to eight. When the bus to be repaired is equipped with other numbers of temperature sensors, the system automatically extracts the arithmetic mean of the temperatures of all currently effective working sensors; the ambient temperature is the average temperature collected by the array of ambient temperature sensors within the quick-repair workstation array, in Kelvin, with one sensor arranged every two workstations along the length of the workstation array; the equivalent stiffness is the electrical... The equivalent linear stiffness of the battery pack mounting structure, in Newtons per meter (N / m), is derived from the vehicle chassis structure design parameters. If the chassis structure design parameters exhibit different stiffness values ​​distributed across multiple points, the arithmetic mean of the stiffness at all mounting points is extracted as the set scalar value. Deformation displacement is the maximum displacement value collected by displacement sensors installed at the four corner points of the battery pack bottom, in meters. The four sensors are installed at the four mounting points on the bottom of the battery pack. This displacement value is an absolute deformation physical quantity measured relative to the horizontal mounting reference plane of the bus chassis as the zero-point reference system. Real-time voltage is the total terminal voltage of the power battery pack, in volts. Leakage current is the insulation leakage current of the high-voltage circuit of the battery pack, in amperes, collected by the insulation monitoring module of the on-board battery management system. Arrival time is the moment the bus awaiting repair enters the queuing area, recorded by the vehicle identification device at the queuing area entrance, in seconds.

[0013] The difference between the real-time battery temperature and the ambient average temperature is multiplied by the battery mass and specific heat capacity to obtain the thermal energy term. The formula for calculating the thermal energy term is:

[0014] In the formula, Let be the thermodynamic energy term of the i-th bus awaiting repair at time t, in joules; The battery quality of the i-th bus awaiting repair; This refers to the specific heat capacity of the battery. Let be the real-time battery temperature of the i-th bus under repair at time t; Let t be the ambient average temperature at time t.

[0015] Multiplying the square of the deformation displacement by the equivalent stiffness and taking half the result, we obtain the mechanical energy term. The formula for calculating the mechanical energy term is:

[0016] In the formula, Let be the mechanical energy term of the i-th passenger car to be repaired at time t, in joules; Let be the equivalent stiffness of the battery pack of the i-th bus to be repaired; Let be the deformation displacement of the i-th passenger car awaiting repair at time t.

[0017] Multiply the real-time voltage by the leakage current and integrate over time from the moment of entry to the current moment to obtain the electrical energy integral term. The formula for calculating the electrical energy integral term is:

[0018] In the formula, Let be the integral term of the electrical energy of the i-th bus to be repaired at time t, in joules; Let be the real-time voltage of the i-th bus to be repaired at time τ; Let be the leakage current of the i-th passenger car awaiting repair at time τ; Let t be the arrival time of the i-th bus awaiting repair; t be the current time; and τ be the integration time variable. For discretely sampled time series data, the trapezoidal integral method is used to calculate the integral value, with the integration step size being the system's general sampling period. If data loss occurs in a single sampling period, linear interpolation of the two preceding and following valid sampling periods is used to complete the data. If continuous packet loss exceeds 3 sampling periods, the value of that integration interval is set to 0.

[0019] Adding the thermal energy term, mechanical energy term, and electrical energy integral term yields the equivalent hazardous energy pool. The formula for calculating the equivalent hazardous energy pool is:

[0020] In the formula, Let be the equivalent hazardous energy pool of the i-th bus under repair at time t, expressed in joules. It represents the total equivalent energy currently accumulated in the battery pack of the bus under repair that could potentially lead to maintenance lock-up or safety risks.

[0021] The system collects data on lifting force, lifting speed, refrigerant density, refrigerant specific heat capacity, coolant flow rate, return water temperature, and inlet water temperature for each quick-repair station. Lifting force is the real-time output lifting force of the station's lift, measured in Newtons (N); lifting speed is the real-time lifting speed of the lift, measured in meters per second (m / s); both lifting force and lifting speed are collected in real-time by the lift's servo drive; refrigerant density and specific heat capacity are standard physical property parameters of the circulating medium in the station's refrigerant piping network, measured in kilograms per cubic meter (kg / m³) and joules per kilogram (J / K), respectively. Standard values ​​are derived from the NIST Chemistry WebBook standard thermodynamic database. If the piping network is equipped with temperature and pressure sensors, dynamic values ​​corresponding to the real-time operating conditions can be used; coolant flow rate is the real-time volumetric flow rate of the refrigerant piping branch, measured in cubic meters per second, collected by a volumetric flow meter within the branch; return water temperature and inlet water temperature are the real-time temperatures of the return and inlet ends of the refrigerant piping branch, respectively, measured in Kelvin, collected by a temperature sensor within the branch. When the workstation is idle, the lifting force is 5% of the rated lifting force of the workstation, the lifting speed is the rated lifting speed of the workstation, the coolant flow rate is the rated flow rate of the branch, and the return water temperature and inlet water temperature are the rated operating temperature difference of the pipe network. Here, the lifting force is selected as five percent of the rated value instead of zero value in order to characterize the basic power of the system in standby mode.

[0022] Multiplying the lifting force by the lifting speed yields the lifting power. The formula for calculating lifting power is:

[0023] In the formula, The lifting power of the j-th quick repair station at time t is expressed in watts. Let be the lifting force of the j-th quick repair station at time t; Let be the lifting speed of the j-th quick repair station at time t.

[0024] The inlet and outlet water temperature difference is obtained by subtracting the outlet water temperature from the outlet water temperature. The refrigerant density, refrigerant specific heat capacity, and coolant flow rate are then multiplied consecutively by the inlet and outlet water temperature difference to obtain the cooling capacity release power of the pipe network. The formula for calculating the cooling capacity release power of the pipe network is:

[0025] In the formula, The cooling capacity released by the pipeline at time t for the j-th quick repair station is expressed in watts. This refers to the refrigerant density. The specific heat capacity of the refrigerant; Let be the coolant flow rate at time t for the j-th quick repair station; Let be the return water temperature of the j-th quick repair station at time t; Let be the inlet water temperature of the j-th quick-repair station at time t. During the calculation, the absolute value of the inlet and outlet water temperature difference is taken to ensure that the cooling capacity release power of the pipeline network is always non-negative.

[0026] Add the lifting power to the cooling capacity release power of the pipeline network to obtain the dynamic combined release power of the corresponding quick-repair station. The formula for calculating the dynamic combined release power is:

[0027] In the formula, The dynamic combined release power of the j-th quick repair station at time t is expressed in watts, representing the total power capacity of the station at the current moment that can be used to release battery pack stress and complete maintenance work.

[0028] Obtain the absolute length of the bus to be repaired, and the geometric center distance between the target workstation and other surrounding workstations. The absolute length of the bus to be repaired is the total length of the vehicle, in meters, derived from the vehicle's factory specifications. The target workstation is the quick repair workstation currently assigned in the trial calculation. Other surrounding workstations are all quick repair workstations in the quick repair workstation array that have a geometric center distance from the target workstation of less than or equal to 20 meters. 20 meters is an initial benchmark threshold set based on conventional station thermal interference attenuation tests. Maintenance personnel can customize and modify this distance threshold according to the actual effective interference radius of the pipeline pressure drop at the station. The geometric center distance is the straight-line distance between the center point of the target workstation's work area and the center points of the work areas of other surrounding workstations, in meters. The center point of the work area is the installation center point of the workstation lift, and the value is derived from the layout design parameters of the quick repair workstation array.

[0029] Subtracting the entry time from the current time gives the queue waiting time. Dividing the equivalent hazardous energy pool by the queue waiting time gives the hazardous energy accumulation rate. The formula for calculating the queue waiting time is:

[0030] The formula for calculating the rate of dangerous energy accumulation is:

[0031] In the formula, Let be the queuing time of the i-th bus awaiting repair at time t, in seconds; Let be the hazardous energy accumulation rate of the i-th bus awaiting repair at time t, expressed in watts, representing the hazardous energy accumulated by the bus per unit time. When the queuing time is less than one system general sampling period, the queuing time is directly taken as the duration of one sampling period to avoid the error of dividing by zero during the calculation.

[0032] The energy suppression ratio is obtained by dividing the hazardous energy accumulation rate by the dynamic combined release power of other surrounding workstations. The formula for calculating the energy suppression ratio is:

[0033] In the formula, When the i-th bus to be repaired is assigned to the j-th target workstation, the energy suppression ratio relative to the k-th surrounding workstation is a dimensionless parameter that characterizes the degree to which the risk accumulation of the bus to be repaired suppresses the working capacity of the surrounding workstations; k is the traversal index of the surrounding workstations, and its value range is all workstation numbers other than the target workstation j that meet the distance requirements.

[0034] Dividing the absolute length by the geometric distance between the centers yields the length proportion. Squaring this proportion gives the spatial hindrance coefficient. The formula for calculating the spatial hindrance coefficient is:

[0035] In the formula, When the i-th bus to be repaired is assigned to the j-th target work station, the spatial obstruction coefficient relative to the k-th surrounding work station is a dimensionless parameter that characterizes the spatial obstruction effect of the vehicle size relative to the work station spacing. Let be the absolute length of the i-th bus awaiting repair; Let be the geometric center distance between the j-th target workstation and the k-th surrounding workstation.

[0036] The individual interference value is obtained by multiplying the energy suppression ratio corresponding to each of the surrounding workstations by the spatial hindrance coefficient. The individual interference values ​​corresponding to all the surrounding workstations are then summed to obtain the spatiotemporal interference repulsion index. The formula for calculating the spatiotemporal interference repulsion index is:

[0037] In the formula, The spatiotemporal interference repulsion index is the index of the i-th bus to be repaired when it is assigned to the j-th target work station. It is a dimensionless parameter that comprehensively quantifies the total spatiotemporal interference resistance intensity of the bus to be repaired when it is assigned to the target work station. The larger the value, the more significant the interference effect.

[0038] Obtain the basic disassembly and assembly standard mechanical work of the bus to be repaired under non-locked conditions. The basic disassembly and assembly standard mechanical work is the mechanical work required to complete the standard disassembly and assembly operation of the battery pack of the bus to be repaired under ideal conditions of no stress lock-up and no thermal interference, and the unit is joules; for battery pack replacement operations, take the standard value of the corresponding new energy bus maintenance industry quota value; for cell-level repair operations, take 30% of the standard value of the whole pack replacement operation; when there is no industry standard, use the maintenance operation quota data provided by the vehicle manufacturer.

[0039] The total work done to overcome hazards is obtained by adding the standard mechanical work done in basic disassembly and assembly to the equivalent hazard energy pool. The formula for calculating the total work done to overcome hazards is:

[0040] In the formula, Let represent the total work done by the i-th bus to be repaired at time t, in joules, which represents the total energy required to complete the repair work on the vehicle. The basic disassembly and assembly standard mechanical work for the i-th bus to be repaired.

[0041] Dividing the total work done to overcome the obstacle by the dynamic combined release power corresponding to the target workstation yields the baseline expected processing time. The formula for calculating the baseline expected processing time is as follows:

[0042] In the formula, The baseline expected processing time, in seconds, is the time it takes for the i-th bus to be assigned to the j-th target workstation when it is tested. It represents the baseline time required for the target workstation to complete the repair of the vehicle without considering interference from surrounding workstations. When the dynamic combined release power is less than 1e-3 watts, the baseline expected processing time is set to a preset maximum value of 3600 seconds to avoid division by zero errors during calculation. This preset maximum value is not limited to 3600 seconds; the system allows for adjustments to the upper limit of the processing time based on the longest historical repair record under extreme lock-up conditions for different bus models.

[0043] The hysteresis amplification factor is obtained by exponentiation using the natural constant as the base and the spatiotemporal interference repulsion index as the exponent. The formula for calculating the hysteresis amplification factor is:

[0044] In the formula, is the hindrance amplification factor when the i-th passenger car to be repaired is assigned to the j-th target work station. It is a dimensionless parameter used to amplify the time increase caused by surrounding interference. Its nonlinear characteristics match the law that the stronger the interference in actual operation, the more significant the unexpected increase in time consumption. exp is an exponential function with the natural constant e as the base.

[0045] Multiply the baseline expected processing time by the hysteresis amplification factor to obtain the actual expected processing time. The formula for calculating the actual expected processing time is:

[0046] In the formula, The actual expected time, in seconds, is the time taken to assign the i-th bus to the j-th target work station. It is a repair time estimate that takes into account the work station's operational capacity, the vehicle's dangerous condition, and the interference from surrounding work stations.

[0047] The system obtains the cell voltage drop rate of the bus awaiting repair and the cell voltage drop rates of all other buses in the waiting queue, and also obtains the arrival time of each bus in the waiting queue. The cell voltage drop rate is the rate of change of the average voltage of all individual cells in the battery pack of the bus awaiting repair, measured in volts per second. It is calculated using the first-order backward difference method, based on the average cell voltage data from three consecutive sampling periods. The calculation formula is as follows:

[0048] In the formula, The system uses a common sampling period. Before calculation, a 5-point moving average filter is used to denoise the cell voltage data, eliminating the influence of sampling noise on the calculation results. The total number of passenger cars waiting for repair in the queue is denoted as... , where n is a positive integer, and n is the index of the queued vehicles, ranging from 1 to 1. .

[0049] Subtracting the arrival times of each individual bus from the current time yields the queuing time for the buses awaiting repair and the queuing time for each of the remaining buses. The formula for calculating the queuing time is:

[0050] In the formula, Let be the waiting time in seconds for the nth bus in the queue at time t. Let n be the arrival time of the nth passenger bus awaiting repair. When the queuing time is less than one system general sampling period, the queuing time is directly taken as the duration of one sampling period.

[0051] The individual crisis inflation rate is obtained by multiplying the absolute value of the rate of voltage drop in the battery cells of the bus awaiting repair by the queuing time. The formula for calculating the individual crisis inflation rate is:

[0052] In the formula, Let be the individual crisis expansion degree of the i-th bus waiting for repair at time t, in volts, representing the degree of battery failure crisis of a single bus waiting for repair as the waiting time increases. Let be the absolute value of the cell voltage drop rate of the i-th bus under repair at time t.

[0053] The average crisis mean is calculated by multiplying the absolute values ​​of the cell voltage drop rates of all passenger cars waiting for repair in the queue by their respective waiting times, summing these sums, and then dividing by the total number of passenger cars in the queue.

[0054] In the formula, The average crisis level of the waiting queue at time t, expressed in volts, represents the average crisis level of all vehicles in the waiting queue and serves as a benchmark for priority adjustment. Let be the absolute value of the cell voltage drop rate of the nth bus under repair at time t.

[0055] The adaptive priority multiplier is obtained by adding one to the quotient of the individual crisis inflation rate divided by the average crisis mean. The formula for calculating the adaptive priority multiplier is:

[0056] In the formula, Let be the adaptive priority multiplier for the i-th bus waiting for repair at time t. is a dimensionless parameter with a minimum value of 1. When the vehicle's crisis level is higher than the average level in the queue, the multiplier is greater than 1, and the priority is increased accordingly, achieving adaptive priority adjustment based on the severity of the vehicle's fault. When the average crisis level is less than 1e-9 volts, the adaptive priority multiplier is directly set to 1 to avoid division by zero errors during calculation. When there is only one bus waiting for repair in the queue, the adaptive priority multiplier is directly set to 2.

[0057] Define the underlying allocation decision variables and the virtual start time for the trial calculation. The underlying allocation decision variables are Boolean variables, denoted as... The value can be 0 or 1; a value of 1 means that the i-th bus awaiting repair will be assigned to the j-th quick repair station, and a value of 0 means that the assignment will not be performed. The virtual start time is denoted as... , where is the time in seconds when the i-th bus to be repaired starts maintenance work at the j-th workstation during the trial allocation. The value of the virtual start time must simultaneously meet two constraints: first, it is not earlier than the absolute current real time; second, it is not earlier than the expected completion time of the vehicles already assigned to the target workstation. The expected completion time of the vehicles already assigned to the target workstation is the sum of the virtual start time and the actual expected time of the vehicle, ensuring that there is no conflict in the workstation operation sequence. When the underlying program performs the trial calculation, the specific value of the virtual start time is directly taken as the larger value between the absolute current real time and the expected completion time of the vehicles already assigned to the target workstation.

[0058] The estimated queuing delay is obtained by subtracting the entry time from the virtual start time. The formula for calculating the estimated queuing delay is:

[0059] In the formula, The estimated queuing delay, in seconds, is the total waiting time from the arrival of the i-th bus to the start of work when it is assigned to the j-th work station.

[0060] The queuing cost area term is obtained by continuously multiplying the underlying allocation decision variable, the adaptive priority multiplier, and the expected queuing delay, and then summing the results over both the waiting queue and the express repair station array, and finally squaring the sum. The formula for calculating the queuing cost area term is as follows:

[0061] In the formula, Let X be the queuing cost area term corresponding to allocation scheme X, in square seconds, which quantifies the total queuing cost brought about by the allocation scheme. The higher the priority of the vehicle, the more significant the increase in cost caused by queuing delay. X is a matrix composed of all underlying allocation decision variables.

[0062] Multiply the underlying allocation decision variable by the actual expected time, and then sum the results by doubly summing the sums of all waiting cars in the waiting queue and all quick repair stations in the quick repair station array, and finally square the sums to obtain the operation cost area term. The formula for calculating the operation cost area term is:

[0063] In the formula, The area term representing the operating cost corresponding to allocation scheme X is expressed in square seconds. It quantifies the total operating cost of the workstations brought about by the allocation scheme and characterizes the total operating time load of the workstation array.

[0064] The queuing cost area term is added to the operation cost area term, and then the square root is taken to construct the Euclidean spacetime cost functional. The formula for calculating the Euclidean spacetime cost functional is as follows:

[0065] In the formula, Let Euclidean spacetime cost functional, in seconds, be used to assign scheme X. This functional unifies the two optimization objectives of queuing delay and workstation operation into a single cost index using Euclidean distance, achieving comprehensive optimization of both objectives. For peak-hour operation scenarios, a weighting coefficient of 1.2 can be applied before the queuing cost area term; for off-peak operation scenarios, a weighting coefficient of 1.2 can be applied before the operation cost area term, adapting to different operational needs.

[0066] The optimal allocation matrix is ​​the set matrix of underlying allocation decision variables that minimizes the Euclidean spacetime cost functional. Integer programming branch and bound is used, with underlying allocation decision variables being 0-1 Boolean variables. The branching rule prioritizes the vehicle with the largest priority multiplier. The pruning rule removes a branch when its lower bound is greater than the cost functional of the currently found optimal solution. The convergence condition is traversing all feasible branches or reaching 1000 iterations, and taking the feasible solution with the smallest cost functional during the traversal as the optimal allocation matrix. Hard constraints in the solution process include: a single bus can only be assigned to one fast repair station; a single fast repair station can only assign one bus at a time; and there are no scheduling conflicts between stations. When the number of vehicles in the waiting queue exceeds the total number of stations M, only the vehicles with the highest priority multipliers are allocated; the remaining vehicles remain in the waiting queue and are included in the next scheduling cycle. In the optimal allocation matrix, the optimal value of the underlying allocation decision variable is denoted as... The corresponding optimal virtual start time is denoted as .

[0067] The system obtains the absolute current real-time, the exact tracking trajectory length from the queuing area to the target workstation, and the real-time speed of the automatic guidance traction. The absolute current real-time is the real-time clock time used by the system to perform scheduling calculations, denoted as... The unit is seconds; the exact tracking trajectory length is the shortest collision-free path length from the queuing area entrance to the center point of the target workstation's work area, in meters. Path planning uses A*. The algorithm is based on the generation of a grid map of the quick repair workstation array, with a grid resolution of 0.1 meters; For the optimal allocation matrix that satisfies The index of the target workstation; the real-time speed of the automated guided vehicle is the rated speed of the automated guided vehicle towing the passenger car to be repaired, in meters per second, and the speed is reduced to 50% of the rated speed on turning sections.

[0068] Extract the optimal allocation Boolean value and its corresponding optimal virtual start time from the optimal allocation matrix. The optimal allocation Boolean value is the value in the optimal allocation matrix corresponding to the passenger car to be repaired and each workstation. The optimal virtual start time is the corresponding of .

[0069] The waiting time difference is obtained by subtracting the absolute current real time from the optimal virtual start time. This waiting time difference is then multiplied by the optimal allocation Boolean value and summed over all quick-repair stations to obtain the target waiting time.

[0070] In the formula, The target waiting time for the i-th bus awaiting repair is expressed in seconds. This represents the duration the bus needs to wait in the queuing area, ensuring that the arrival time of the vehicle at the workstation matches the optimal start time. When all workstations are occupied and no workstation can be assigned, the optimal allocation Boolean value is all 0, and the target waiting time is set to 30 seconds, which is the time interval for the next scheduling trigger. The allocation scheme will be re-solved in the next scheduling cycle.

[0071] Dividing the exact tracking trajectory length by the real-time speed of the automatic guidance traction yields the travel time. The process of deriving the time variable using the quotient of trajectory length and real-time speed is the specific implementation method of vehicle motion parameter inversion; the formula for calculating travel time is:

[0072] In the formula, The travel time of the i-th bus to be repaired is expressed in seconds, representing the time required for the vehicle to travel from the queuing area to the target work station. To determine the exact length of the tracking trajectory; This is the real-time speed for automatic guidance and traction. During the calculation, the time taken on a straight section is the straight path length divided by the rated driving speed, the time taken on a turning section is the turning path length divided by 50% of the rated driving speed, and the total travel time is the sum of the time taken on the straight section and the time taken on the turning section.

[0073] The absolute current real-time, the waiting time for the scheduling target, and the travel time are added together to generate a spatiotemporal arrival scheduling command, which is then output to the front end. The calculation formula for the spatiotemporal arrival scheduling command is:

[0074] In the formula, The time of arrival for the i-th bus awaiting repair is specified in seconds. The system sends a dispatch instruction containing this dispatch time, the corresponding target workstation number, and the travel route information to the front-end execution terminal via HTTP protocol. The front-end execution terminal includes the automated guided vehicle control system, the on-site dispatch display screen, and the on-board dispatch terminal. After the instruction is sent, the system receives a feedback signal from the execution terminal. If no execution confirmation is received within 10 seconds, the instruction is resent, thus achieving closed-loop execution of the dispatch instruction.

[0075] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A multi-objective intelligent scheduling system for quick repair bays of new energy bus batteries, applied to a quick repair bay array including a queuing area and multiple quick repair bays, characterized in that... Configured for execution: Obtain the vehicle status data and queuing time of each passenger vehicle waiting in the waiting queue in the queuing area, and map the passenger vehicle status data into an equivalent hazard energy pool; Obtain the station operation data of each of the quick repair stations and quantify the corresponding dynamic joint release power; For any of the aforementioned passenger vehicles to be repaired, the vehicle is assigned to any target workstation. Based on the dynamic joint release power of the equivalent hazardous energy pool and other surrounding workstations besides the target workstation, a spatiotemporal interference repulsion index is constructed. By combining the equivalent hazardous energy pool, the dynamic joint release power corresponding to the target workstation, and the spatiotemporal interference repulsion index, the actual expected time consumption is deduced. Extract the cell voltage drop rate of the bus to be repaired and calculate the adaptive priority multiplier by comparing it with the global mean of the waiting queue; By combining the adaptive priority multiplier, the queuing time, and the actual expected time, a Euclidean spacetime cost functional is constructed, and the optimal allocation matrix is ​​obtained by solving it. Based on the optimal allocation matrix superimposed with vehicle motion parameter inversion, a spatiotemporal arrival scheduling command is output to the bus to be repaired.

2. The multi-objective intelligent scheduling system for quick repair stations of new energy bus batteries according to claim 1, characterized in that, Mapping the bus status data to an equivalent hazard energy pool includes: The battery mass, specific heat capacity, real-time battery temperature, ambient temperature, equivalent stiffness, deformation displacement, real-time voltage, leakage current, and arrival time of the bus to be repaired are collected. The difference between the real-time battery temperature and the ambient average temperature is multiplied by the battery mass and the specific heat capacity to obtain the thermal energy term. Multiply the square of the deformation displacement by the equivalent stiffness and take half to obtain the mechanical energy term; Multiply the real-time voltage by the leakage current and integrate over time from the moment of entry to the current moment to obtain the electrical energy integral term. The equivalent hazardous energy pool is obtained by adding the thermal energy term, the mechanical energy term, and the electrical energy integral term.

3. The multi-objective intelligent scheduling system for rapid repair stations of new energy bus batteries according to claim 2, characterized in that, The corresponding dynamic joint release power is quantified, including: Collect the lifting force, lifting speed, refrigerant density, refrigerant specific heat capacity, coolant flow rate, return water temperature, and inlet water temperature of the corresponding quick repair station; Multiply the lifting force by the lifting speed to obtain the lifting power. The inlet and outlet water temperature difference is obtained by subtracting the outlet water temperature from the outlet water temperature. The refrigerant density, the refrigerant specific heat capacity, the coolant flow rate, and the inlet and outlet water temperature difference are continuously multiplied to obtain the cooling capacity release power of the pipeline network. The lifting power is added to the cooling capacity release power of the pipeline network to obtain the dynamic combined release power of the corresponding quick repair station.

4. The multi-objective intelligent scheduling system for fast repair stations of new energy bus batteries according to claim 3, characterized in that, Constructing a spatiotemporal interference repulsion index includes: Obtain the absolute length of the bus to be repaired, and the geometric center distance between the target workstation and other surrounding workstations; Subtracting the entry time from the current time yields the queuing time; dividing the equivalent hazardous energy pool by the queuing time yields the hazardous energy accumulation rate. The energy suppression ratio is obtained by dividing the dangerous energy accumulation rate by the dynamic combined release power corresponding to the other surrounding workstations. Divide the absolute length by the geometric center distance to obtain the length ratio, and square the length ratio to obtain the spatial hindrance coefficient. The individual interference value is obtained by multiplying the energy suppression ratio corresponding to each of the surrounding other workstations by the spatial obstruction coefficient. The individual interference values ​​corresponding to all the surrounding other workstations are summed to obtain the spatiotemporal interference repulsion index.

5. The multi-objective intelligent scheduling system for fast repair stations of new energy bus batteries according to claim 4, characterized in that, The actual expected time derived from the calculation includes: Obtain the basic disassembly and assembly standard mechanical work of the bus to be repaired in an unlocked state; The total work done to overcome the hazards is obtained by adding the basic standard mechanical work of disassembly and assembly to the equivalent hazard energy pool. Divide the total amount of work done to overcome the obstacles by the dynamic combined release power corresponding to the target workstation to obtain the baseline expected processing time. Using the natural constant as the base and the spatiotemporal interference repulsion index as the exponent, a power operation is performed to obtain the retardation amplification factor; The actual expected processing time is obtained by multiplying the baseline expected processing time by the hysteresis amplification factor.

6. The multi-objective intelligent scheduling system for fast repair stations of new energy bus batteries according to claim 5, characterized in that, Extract the cell voltage drop rate of the bus to be repaired, and calculate the adaptive priority multiplier by comparing it with the global mean of the waiting queue, including: The battery cell voltage drop rate of the bus to be repaired and the battery cell voltage drop rate of the other buses to be repaired in the waiting queue are obtained, and the arrival time of each bus to be repaired in the waiting queue is obtained. Subtract each of the respective entry times from the current time to obtain the queuing waiting time of the bus to be repaired and the queuing waiting time of the other buses to be repaired. The absolute value of the rate of voltage drop of the battery cells of the bus to be repaired is multiplied by the queuing time to obtain the individual crisis inflation degree. The average crisis mean is obtained by multiplying the absolute value of the cell voltage drop rate of all passenger cars waiting for repair in the waiting queue by their respective waiting time, summing the results, and then dividing by the total number of passenger cars waiting for repair in the waiting queue. The adaptive priority multiplier is obtained by adding one to the quotient obtained by dividing the individual crisis inflation rate by the average crisis mean.

7. The multi-objective intelligent scheduling system for fast repair stations of new energy bus batteries according to claim 6, characterized in that, Construct a Euclidean spacetime cost functional and solve for the optimal allocation matrix, including: Set the underlying allocation decision variables and the virtual start time for trial calculations; Subtracting the entry time from the virtual start time yields the estimated queuing delay. The layer allocation decision variable, the adaptive priority multiplier, and the expected queuing delay are multiplied continuously, and the queuing cost area term is obtained by double summing all the passenger cars waiting for repair in the waiting queue and all the quick repair stations in the quick repair station array and then taking the square. Multiply the underlying allocation decision variable by the actual expected time, and then sum the results of the double accumulation of all passenger cars waiting for repair in the waiting queue and all quick repair stations in the quick repair station array, and then square the sum to obtain the operating cost area term. The queuing cost area term is added to the operation cost area term, and then the square root is taken to construct the Euclidean spacetime cost functional. Solve for the set matrix of the underlying allocation decision variables that minimizes the Euclidean spacetime cost functional, and use it as the optimal allocation matrix.

8. The multi-objective intelligent scheduling system for fast repair stations of new energy bus batteries according to claim 7, characterized in that, Based on the optimal allocation matrix superimposed with vehicle motion parameter inversion, a spatiotemporal arrival scheduling instruction is output to the bus to be repaired, including: Obtain the absolute current real time, the exact tracking trajectory length from the queuing area to the target workstation, and the real-time speed of the automatic guidance traction; Extract the optimal allocation Boolean value and the corresponding optimal virtual start time from the optimal allocation matrix; The waiting time difference is obtained by subtracting the absolute current real time from the optimal virtual start time. The waiting time difference is multiplied by the optimal allocation Boolean value and summed for all the quick repair stations to obtain the scheduling target waiting time. Divide the exact tracking trajectory length by the real-time speed of the automatic guidance and traction to obtain the travel time. The absolute current real time, the waiting time for the scheduling target, and the travel time are added together to generate the spatiotemporal entry scheduling command and output to the front end.

Images