A charging pile and power grid coordinated intelligent scheduling method and system

Abstract

The application discloses a kind of charging pile and power grid collaborative intelligent scheduling method and system, belong to new energy automobile technical field, this method includes: the charging pile in station is divided into main pile and slave pile, in slave pile in constant current stage positioning mutant pile, calculate the charging power after mutation of mutant pile, based on charging parameter determine the urgency value of all slave piles in charging stage;With the distribution power of each slave pile as optimization variable, with the upper limit of power of charging station issued by power grid as constraint, solve based on particle swarm optimization algorithm, wherein, the individual factor and social factor of each dimension in particle swarm optimization algorithm are independently controlled by the urgency value of corresponding slave pile;The charging power instruction corresponding to each slave pile in the optimal distribution scheme obtained by solving particle swarm optimization algorithm is issued to slave pile by main pile for execution, and the collaborative scheduling of charging pile and power grid can be improved by the application.

Description

Technical Field

[0001] This invention belongs to the field of new energy vehicle technology, specifically relating to an intelligent scheduling method and system for the coordinated operation of charging piles and power grids. Background Technology

[0002] In centralized charging stations, multiple charging piles share one or more distribution network access lines. After the charging station is connected to the grid as a controllable load, its maximum allowable charging power is no longer determined solely by the distribution transformer capacity or fixed demand contracts. Instead, it is dynamically issued by the grid dispatch center based on the real-time operating status of the grid, renewable energy consumption demand, and regional load balancing strategies. The grid dispatch center periodically or event-drivenly issues power limit commands to the charging station through a communication interface. The charging station must then allocate power among its charging piles under the constraints of these commands to ensure that the total power of the charging station does not exceed the power limit issued by the grid. When multiple new energy vehicles are simultaneously charging at high power, the total power demand of the charging station may exceed the power limit issued by the grid within a short period. If the charging station fails to respond to the grid commands in a timely manner and adjust its power accordingly, it will lead to regional distribution network overload, grid frequency deviation, or trigger punitive assessments by the grid side.

[0003] To address the aforementioned issues, existing technologies generally employ a master-slave architecture for group control. This involves selecting a master charging station as the scheduling controller, collecting real-time data on the operational status of each slave station via an intra-station communication bus, and executing a power allocation algorithm according to a fixed scheduling cycle to distribute the maximum power allocated by the grid among the slave stations. Common allocation strategies include equal distribution and priority round-robin. However, new energy vehicles undergo a transition from a constant current phase to a constant voltage phase during charging. When a charging station enters the constant voltage phase, its battery's acceptable power drops sharply. Existing scheduling algorithms only make allocation decisions based on the current state and cannot predict the upcoming phase transition. This causes the allocation scheme to fail after the sudden phase change, requiring readjustment only in the next scheduling cycle, resulting in idle available capacity allocated by the grid. Furthermore, when multiple charging stations undergo constant current / constant voltage phase transitions within a similar timeframe, the system still locks the allocation quota for that station based on the original high power demand, while the actual power consumed by the vehicle is significantly reduced. This difference creates power fragmentation within the system. When multiple new energy vehicles enter the constant voltage phase around the same time, power fragments locked by charging piles but not actually utilized accumulate rapidly. Due to the lag in scheduling algorithms, these power fragments cannot be recovered and redistributed in a timely manner. This results in charging stations being fully occupied within the power limit issued by the grid, while the actual output power is far below the grid's allowed power limit. On the one hand, this reduces the effective power utilization rate of charging stations; on the other hand, it prevents charging stations from accurately tracking the power scheduling instructions issued by the grid, weakening the coordinated scheduling capability between charging piles and the grid, and affecting the grid's precise control over the charging station load and the effective absorption of renewable energy. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides an intelligent scheduling method and system for the coordinated operation of charging piles and the power grid, thereby resolving the issues present in the background art.

[0005] To achieve the aforementioned objectives, this invention proposes an intelligent scheduling method for the coordinated operation of charging piles and the power grid, comprising:

[0006] The charging piles in the station are divided into main piles and slave piles. The main piles collect the charging parameters of each slave pile, and the charging stage of the slave pile is determined based on the charging parameters. The charging stage includes a constant voltage stage and a constant current stage.

[0007] Locate the sudden change pile among the slave piles in the constant current stage, calculate the charging power of the sudden change pile after the sudden change, and determine the urgency value of the charging stage of all slave piles based on the charging parameters.

[0008] The power allocation of each slave pile is used as the optimization variable, and the upper limit of the charging station power issued by the power grid is used as the constraint. The solution is based on the particle swarm algorithm. In the particle swarm algorithm, the individual factors and social factors of each dimension are independently controlled by the urgency value of the corresponding slave pile.

[0009] During the iteration of the particle swarm optimization algorithm, the particles are divided into active particles and stagnant particles based on their search state. Fitness evaluations of different precision are performed on active particles and stagnant particles respectively. The fitness value is determined based on the current state evaluation value and the mutation correction evaluation value. The mutation correction evaluation value is calculated based on the charging power after the mutation of the mutation stake.

[0010] The master pile sends the charging power commands corresponding to each slave pile in the optimal allocation scheme obtained by the particle swarm optimization algorithm to the slave piles for execution.

[0011] This invention also provides an intelligent scheduling system for the coordination of charging piles and the power grid. This system is used to implement the methods described above, and includes:

[0012] The status awareness module divides the charging piles in the station into main piles and slave piles. The main piles collect the charging parameters of each slave pile and determine the charging stage of the slave piles based on the charging parameters. The charging stage includes a constant voltage stage and a constant current stage.

[0013] The mutation prediction module locates the mutation pile among the slave piles in the constant current stage, calculates the charging power after the mutation of the mutation pile, and determines the urgency value of the charging stage of all slave piles based on the charging parameters.

[0014] The collaborative optimization module uses the allocated power of each slave pile as the optimization variable and the upper limit of the charging station power issued by the power grid as the constraint. It is solved based on the particle swarm optimization algorithm. In the particle swarm optimization algorithm, the individual factors and social factors of each dimension are independently controlled by the urgency value of the corresponding slave pile.

[0015] The dynamic precision calculation module divides particles into active and stagnant particles based on their search state during the particle swarm algorithm iteration process. It performs fitness evaluations of different precision on active and stagnant particles respectively. The fitness value is determined based on the current state evaluation value and the mutation correction evaluation value. The mutation correction evaluation value is based on the charging power calculation after the mutation of the mutation pile. The master pile sends the charging power command corresponding to each slave pile in the optimal allocation scheme obtained by the particle swarm algorithm to the slave piles for execution.

[0016] The beneficial effects of this invention are as follows:

[0017] This invention determines the urgency value of each charging stage by locating a sudden change charging station among slave charging stations in the constant current stage and calculating the charging power after the sudden change. Then, using the allocated power of each slave charging station as the optimization variable and the upper limit of charging station power issued by the power grid as a constraint, a particle swarm optimization (PSO) algorithm is used to solve the problem. The individual and social factors of each dimension are independently controlled by the urgency value of the corresponding slave charging station. This allows the PSO algorithm to classify particles into active and stagnant particles based on their search state during iteration, and to perform fitness evaluations of different precision on active and stagnant particles respectively. Finally, the master charging station issues the charging power command corresponding to each slave charging station in the optimal allocation scheme obtained by the PSO algorithm to the slave charging stations for execution. Notably, this invention solves the problems in existing technologies where scheduling algorithms cannot predict upcoming stage transitions, leading to allocation scheme failures, idle available capacity resulting in power fragmentation, and weakened collaborative scheduling capabilities between charging stations and the power grid. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the steps of an intelligent scheduling method for the coordination of charging piles and power grids according to the present invention.

[0019] Figure 2 This is a diagram showing the overall architecture of the device of the present invention;

[0020] Figure 3 The diagram shows a comparison between the improved particle swarm optimization algorithm of this invention and the traditional particle swarm optimization algorithm.

[0021] Figure 4 This is a schematic diagram of the structure of an intelligent scheduling system for the coordination of charging piles and power grids according to the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0023] It is understood that the terms "first," "second," etc., used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, without departing from the scope of this application, a first script may be referred to as a second script, and similarly, a second script may be referred to as a first script.

[0024] like Figure 1 As shown, an intelligent scheduling method for the coordinated operation of charging piles and the power grid includes:

[0025] S1: The charging piles in the station are divided into main piles and slave piles. The main piles collect the charging parameters of each slave pile and determine the charging stage of the slave piles based on the charging parameters. The charging stage includes a constant voltage stage and a constant current stage.

[0026] like Figure 2 As shown, the charging parameters include the main charging pile acquiring the output power of each slave charging pile through the station's communication bus, and obtaining the state of charge (SOC) value, battery capacity, and maximum power acceptance of the vehicle connected to each slave charging pile. The charging stages include a constant current stage and a constant voltage stage. The constant current / constant voltage conversion threshold reported by the battery management system of the vehicle connected to each slave charging pile is obtained. The constant current / constant voltage conversion threshold is the critical point of SOC value when the battery switches from the constant current charging stage to the constant voltage charging stage. It is usually preset by the battery manufacturer based on the battery characteristics and ranges between 80% and 95%. The charging stage of the slave charging pile is determined based on the relationship between the SOC value and the constant current / constant voltage conversion threshold.

[0027] S2: Locate the sudden change pile among the slave piles in the constant current stage, calculate the charging power after the sudden change of the sudden change pile, and determine the urgency value of the charging stage of all slave piles based on the charging parameters.

[0028] During the constant current phase, some slave charging piles have reached a state of charge (SOC) close to the constant current / constant voltage transition threshold, potentially leading to a phase transition within the current scheduling cycle. When a phase transition occurs, the battery management system (BMS) drastically reduces the maximum acceptable power from the rated power of the constant current phase to a lower level of the constant voltage phase. If the scheduling scheme still allocates power to these slave charging piles according to the rated power of the constant current phase, the excess power after the transition cannot be absorbed by the battery, resulting in a waste of total capacity. Therefore, it is necessary to identify these slave charging piles that are about to undergo a phase transition in advance and mark them as abrupt transition points, calculating the acceptable post-transition charging power.

[0029] The urgency value reflects the efficient utilization of high power allocation by each slave station. Slave stations in the constant current phase with sufficient remaining charge capacity are allowed to accept higher power by the battery management system; allocating more power can be effectively absorbed by the battery and accelerate the charging process, resulting in a higher urgency value. Slave stations in the constant voltage phase have their power acceptance limited to a lower level by the battery management system; even if more power is allocated, it cannot be absorbed by the battery, resulting in a lower urgency value. The urgency value is used to control the search strategy in subsequent particle swarm optimization algorithms, ensuring that the algorithm prioritizes allocating sufficient capacity to slave stations that can efficiently utilize power.

[0030] The specific method for calculating the urgency value in this embodiment is as follows: After completing the stage calibration, the power acceptance ratio is calculated based on the ratio of the current maximum power received by the vehicle charging at each charging pile to its rated power. The power acceptance ratio reflects the proportion of power that the battery management system currently allows to absorb from that pile relative to the rated power. Simultaneously, the uncharged ratio is calculated based on the ratio of the difference between the target state of charge and the current state of charge of the vehicle connected to each charging pile to the battery capacity. The uncharged ratio reflects the proportion of the battery capacity that still needs to be charged at that pile. Based on this, the product of the power acceptance ratio and the uncharged ratio is used as the urgency value for each charging pile. When the charging pile is in the constant current stage and the remaining uncharged amount is sufficient, the power accepted by the battery management system is close to the rated power, the power acceptance ratio is close to 1, and the uncharged ratio is also relatively large. Therefore, the urgency value is high, indicating that allocating high power to that charging pile at this time can effectively shorten the charging time. Conversely, when the slave pile has entered the constant voltage stage, the battery management system has limited the power accepted to a low level, and the power acceptance ratio approaches 0. Even if the remaining charge is not fully charged, the urgency value also approaches 0, indicating that the extra power allocated to this slave pile cannot be actually absorbed by the battery and should be allocated to other slave piles that can be used efficiently.

[0031] S3: Using the power allocation of each slave pile as the optimization variable and the upper limit of the charging station power issued by the power grid as the constraint, the solution is based on the particle swarm optimization algorithm. In the particle swarm optimization algorithm, the individual factors and social factors of each dimension are independently controlled by the urgency value of the corresponding slave pile.

[0032] A position vector for the particle swarm optimization algorithm is established using the power allocation of each slave pile as the optimization variable, with each dimension corresponding to the power allocation value of one slave pile. In the velocity update formula, individual factors and social factors are transformed into dimension-dependent variables, dynamically adjusted through urgency values. Dimensions with high urgency values ​​use larger social factors and smaller individual factors to rapidly converge to the global optimum. Dimensions with low urgency values ​​use smaller social factors and larger individual factors to maintain search diversity. A hierarchical constraint correction is applied to the particle position vector using the total charging station capacity allocated by the power grid as a constraint, ensuring that the allocation scheme always meets the capacity constraint.

[0033] S4: During the iteration of the particle swarm optimization algorithm, the particles are divided into active particles and stagnant particles based on their search state. Fitness evaluations of different precision are performed on active particles and stagnant particles respectively. The fitness value is determined based on the current state evaluation value and the mutation correction evaluation value. The mutation correction evaluation value is calculated based on the charging power after the mutation of the mutation stake.

[0034] This embodiment proposes a dual fitness evaluation method. However, if dual evaluation is performed on all particles, the overall computational load will far exceed the capacity of the embedded processor when the number of slaves is large. This could lead to the inability to complete the solution within the scheduling cycle or the need to replace the processor with a high-performance one, significantly increasing hardware costs. Therefore, it is necessary to divide particles into active and stagnant particles based on their position changes, because the search states of different particles differ during the iteration process of the particle swarm optimization algorithm. Some particles have undergone significant position changes and are still exploring new regions in the search space. The fitness of these particles at their new positions may be better than their historical best values, so a complete dual fitness evaluation is needed to accurately determine this. Other particles have undergone minimal position changes and have converged to near their historical best positions. Their fitness values ​​will not change significantly, and repeatedly performing a complete dual fitness evaluation on these particles would result in a large amount of redundant computation.

[0035] Therefore, a full dual fitness evaluation is performed on active particles, while a simplified evaluation is performed on stagnant particles. This allows the algorithm to maintain full evaluation accuracy in the early stages of the search, and to reduce the computational power requirement of the master stump by reducing redundant calculations in the later stages of the search. This significantly reduces the computational load without significantly reducing the search quality, enabling the embedded processor of the master stump to complete the solution within the scheduling cycle and avoiding an increase in hardware costs.

[0036] S5: The master pile sends the charging power command corresponding to each slave pile in the optimal allocation scheme obtained by the particle swarm algorithm to the slave piles for execution.

[0037] The particle swarm optimization algorithm outputs the position vector of the globally optimal particle as the optimal allocation scheme after the termination condition is met. This scheme includes the power allocation values ​​for all slave piles, where the allocation values ​​for non-mutation piles have reserved space for the overflow amount received from mutation piles. The master pile sends the charging power command corresponding to each slave pile in the optimal allocation scheme to the slave piles through the station communication bus, and each slave pile charges according to the allocated charging power.

[0038] In this embodiment, locating the sudden change pile among the piles in the constant current stage and calculating the charging power of the sudden change pile after the sudden change includes:

[0039] For slave piles in the constant current stage, based on the difference between the state of charge value and the constant current and constant voltage conversion threshold, as well as the growth rate of the state of charge value, slave piles that reach the conversion threshold within the scheduling cycle will be marked as abruptly changed piles.

[0040] At two adjacent sampling time points, the state of charge (SOC) value of the vehicle connected to the charging pile is acquired. The difference in SOC value between the two sampling time points is calculated, and divided by the interval between the sampling time points to obtain the SOC growth rate. The SOC growth rate reflects how quickly the battery's SOC rises at the current charging power. Subtracting the current SOC value from the constant current / constant voltage conversion threshold gives the remaining SOC increase. Dividing this increase by the SOC growth rate gives the estimated time required to reach the conversion threshold.

[0041] The estimated time required to reach the transition threshold is compared with the end time of the current scheduling cycle. If the estimated time is less than or equal to the end time of the scheduling cycle, the slave is determined to reach the constant current / constant voltage transition threshold within this scheduling cycle and is marked as a sudden transition slave. If the estimated time is longer than the scheduling cycle time, the slave will not undergo a phase transition within this cycle and is not marked as a sudden transition slave.

[0042] Based on the battery capacity and the preset attenuation constant, the upper limit of the power that the sudden charging pile can accept after the stage transition is calculated according to the constant voltage stage exponential attenuation formula, and is used as the charging power after the sudden change.

[0043] For slave charging piles marked as "abrupt charging piles," the battery capacity of the connected vehicles is obtained. Based on the constant-voltage charging curve characteristics provided by the battery manufacturer, a preset decay constant is obtained. This decay constant reflects the rate at which the charging power decreases over time during the constant-voltage phase. The charging power after the abrupt change... The specific calculation is performed using the following formula: ,in, This refers to the rated power during the constant current phase. This represents the remaining duration within the current scheduling cycle after the phase transition occurs, specifically determined based on the scheduling cycle time and the expected time of the phase transition. This is a preset decay constant. This formula reflects the exponential decay of charging power over time during the constant voltage phase. When switching from the constant current phase to the constant voltage phase within the current scheduling cycle, the battery management system immediately reduces the received power from the rated power of the constant current phase to the initial power of the constant voltage phase. During the constant voltage phase, as the state of charge continues to rise, the battery management system further limits the received power, causing it to decrease according to an exponential decay law.

[0044] In this embodiment, the individual and social factors of each dimension in the particle swarm optimization algorithm are independently controlled by the urgency value of the corresponding stake, including:

[0045] In the particle swarm optimization algorithm, the position vector of each particle contains the same number of dimensions as the number of slaves, where the d-th dimension of the position vector represents the charging power allocated to the d-th slave.

[0046] In each iteration, during the velocity update of the d-th dimension of the particle, the individual factor and the social factor are independently determined by the urgency value of the corresponding stake. The social factor increases with the urgency value of the corresponding stake to accelerate the convergence of the corresponding dimension to the global optimum, while the individual factor decreases with the urgency value of the corresponding stake to maintain search diversity in the corresponding dimension.

[0047] Standard particle swarm optimization (PSO) uses a uniform individual and social factor across all dimensions, resulting in all slave charging stations being searched using the same convergence strategy. This homogenization ignores the differences in charging processes among different slave charging stations in this scenario. Slave charging stations in the constant voltage phase and those in the constant current phase are treated equally and pushed towards the global optimum. As a result, the algorithm may distribute the limited total capacity evenly, causing slave charging stations in the constant current phase to not receive enough power to accelerate charging, while slave charging stations in the constant voltage phase still occupy a large amount of capacity but cannot effectively absorb it due to the power limitations of the battery management system, ultimately leading to capacity waste and a decrease in overall charging efficiency.

[0048] To address the aforementioned issues, in this embodiment, the position vector of each particle in the particle swarm optimization algorithm contains the same dimension as the number of slaves. The d-th dimension represents the charging power allocated to the d-th slave. In each iteration, the velocity update formula for the d-th dimension of particle i is improved as follows: .in, To represent the updated velocity of the i-th particle in the d-th dimension, The inertial weight determines the degree to which a particle inherits its current velocity. Let be the current velocity of the i-th particle in the d-th dimension. and A random number between 0 and 1. Let i be the historical best position of the i-th particle in the d-th dimension. Let be the position of the globally optimal particle in the d-th dimension. For the current position of particle i in the d-th dimension before the update, the individual factor in the d-th dimension is... social factors They are respectively: ,in and These are the preset upper and lower boundary values ​​for individual factors. and The upper and lower boundary values ​​are preset for the social factors. Let be the urgency value of the d-th follower. The upper boundary value of the individual factor can be set to 2.5, and the lower boundary value can be set to 0.5. The upper boundary value of the social factor can also be set to 2.5, and the lower boundary value can be set to 0.5. This formula introduces preset factor boundary values ​​so that in extreme cases, such as when a follower is at the end of the constant pressure stage and its urgency is 0, its social factor is at its minimum boundary value. This ensures that when particles in this dimension update their velocity, they retain the reference weight for the global optimal solution, preventing the search trajectory in this dimension from degenerating into a local walk that only relies on the individual's historical best, thereby avoiding the algorithm from getting stuck in local convergence and ensuring the collaborative optimization capability in the multi-dimensional solution space.

[0049] Based on the improved formula, when the urgency value is high, the social factor increases while the individual factor decreases. In the speed update formula, the increase in the social factor enhances the attraction of the globally optimal particle's position in the d-th dimension to the current particle's position, while the decrease in the individual factor weakens the influence of the particle's own historical optimal value. This causes the particle to be more inclined to move in the direction of the globally optimal solution during the iteration process, making the corresponding dimension quickly approach the globally optimal value. In the scenario of charging pile power allocation, slave piles with high urgency values ​​are in the constant current stage and have a high power acceptance ratio, which can effectively absorb the allocated power. Through the accelerated convergence strategy, the algorithm will prioritize exploring schemes to allocate more power to these slave piles during the search process, so that their power allocation value quickly stabilizes at a level that maximizes the overall charging efficiency.

[0050] When the urgency value is low, the individual factor increases while the social factor decreases. In the velocity update formula, the increase in the individual factor enhances the attraction of the particle's own historical best value to its current position, while the decrease in the social factor weakens the constraint of the global optimal solution. This results in particles maintaining strong independence during iteration, not being overly pulled by the global optimal solution, but instead exploring more extensively within the search space. In the scenario of charging pile power allocation, slave piles with low urgency values ​​are in the constant voltage stage, with a low power acceptance ratio, and cannot effectively absorb even if more power is allocated. By maintaining search diversity, the algorithm does not prematurely fix the power allocation of such slave piles to a certain value, but continues to try different allocation schemes, freeing up more capacity space for other high-urgency slave piles. Through the above improvements, slave piles in the constant current stage receive sufficient power allocation, while the power allocation of slave piles in the constant voltage stage is moderately reduced, thus improving the overall capacity utilization rate.

[0051] In this embodiment, particles are divided into active particles and stagnant particles based on their search state, including:

[0052] After each iteration of position update is completed, the sum of the absolute values ​​of the differences in each dimension between the position vector of each particle and its historical best position vector is calculated and defined as the displacement. If the displacement is greater than the preset displacement threshold, the particle is marked as an active particle; otherwise, the particle is marked as a stationary particle.

[0053] After each round of the particle swarm optimization algorithm updates the position vector, it calculates the sum of the absolute values ​​of the differences between each particle's position vector and its historical best position vector in each dimension. This value is denoted as the displacement. This calculation is performed by comparing the current position with the historical best position dimension by dimension, ensuring a comprehensive reflection of the overall movement of the particles. The calculated displacement is then compared with a preset displacement threshold. If the displacement is greater than the threshold, the particle is considered an active particle, indicating that its position has changed significantly and it is still actively searching for new solution space regions. If the displacement is less than or equal to the threshold, the particle is marked as a stagnant particle, indicating that its position has changed little and it may have converged to a local optimum, with little change in its fitness value in subsequent iterations.

[0054] In this embodiment, fitness evaluations of different precision are performed on active particles and stagnant particles, including:

[0055] Calculate the current state evaluation value and mutation correction evaluation value for active particles. Obtain the fitness value by weighting the current state evaluation value and mutation correction evaluation value. Store the ratio of the fitness value to the current state evaluation value as the correction ratio of the particle. When an active particle becomes a stagnant particle, calculate the current state evaluation value for the stagnant particle. Use the product of the current state evaluation value and the correction ratio as the fitness value of the stagnant particle.

[0056] For active particles, based on the current maximum power capacity of each slave pile, the total power of the power allocation scheme represented by that particle is calculated to obtain the current state assessment value. Then, the maximum power capacity of the aberrant pile is replaced with its post-abrupt charging power, and overflow redistribution correction is performed on the scheme. The corrected total power of the entire station is calculated to obtain the aberration correction assessment value. Finally, the fitness value is obtained by weighted averaging of the current state assessment value and the aberration correction assessment value. This calculation method ensures that the fitness calculation considers both the current capacity utilization capability and the rationality of power allocation after the phase transition.

[0057] If an active particle becomes a stagnant particle, the correction ratio stored from the previous iteration is retrieved. This correction ratio is the ratio of the fitness value calculated in the previous round to the current state evaluation value calculated in the previous round. Then, the current state evaluation value of the stagnant particle is calculated, and this current state evaluation value is multiplied by the correction ratio to calculate the stagnant particle's overall fitness value. In subsequent iterations, if the particle remains stagnant, the stored correction ratio is directly retrieved, and the current state evaluation value calculated for the corresponding round is multiplied by this correction ratio to obtain the fitness value for that round. Based on this step, the fitness of stagnant particles is evaluated using historical data ratio correction instead of mutation correction. This calculation method prevents drastic jumps in the fitness value of stagnant particles, thus continuously and smoothly guiding the population's optimization direction in the solution space. It also significantly reduces repetitive and complex calculations, lowers the computational burden on the embedded processor, and improves the scheduling response rate.

[0058] In this embodiment, the number of consecutive stagnant rounds for each stagnant particle is recorded. When the number of consecutive stagnant rounds reaches the preset upper limit of consecutive stagnant rounds, the current state evaluation and mutation correction evaluation are forcibly performed on the stagnant particle, the correction ratio is updated, and the number of consecutive stagnant rounds is reset.

[0059] After particle marking is completed in each iteration, the consecutive stagnation count of particles marked as stagnant is incremented by 1. After the count update, it is checked whether the consecutive stagnation count of each stagnant particle has reached the preset upper limit for consecutive stagnation rounds. The upper limit for consecutive stagnation rounds is determined based on the total number of iterations in the particle swarm optimization algorithm. For example, when the total number of iterations is 100, the upper limit for consecutive stagnation rounds can be set to 5 to 10. When the consecutive stagnation count of a stagnant particle reaches the upper limit, a complete fitness evaluation is forcibly performed on that stagnant particle. The complete fitness evaluation includes a current state evaluation and a mutation correction evaluation, specifically consistent with the evaluation method for active particles, thereby obtaining a new fitness value. After the complete evaluation is completed, in the next round of calculation, the correction ratio is recalculated using the fitness value calculated above and the new current state evaluation value, and the consecutive stagnation count of the stagnant particle is reset to zero, restarting the counting process.

[0060] During the iterative process of the particle swarm optimization algorithm, the global optimum position is continuously updated as the iteration progresses, and the relative relationship between stagnant particles and the global optimum position also changes accordingly. This can lead to a difference between the performance of stagnant particles in the mutation correction evaluation and the initial calculation of the correction ratio. Periodically forcing a full evaluation and updating the correction ratio allows the simplified fitness value of stagnant particles to more accurately approximate the full evaluation result, reducing computational cost while maintaining the reliability of the fitness evaluation.

[0061] In this embodiment, after each iteration, the proportion of stagnant particles in the population is counted. If the proportion of stagnant particles exceeds a preset stagnant proportion threshold, then the particles with the worst fitness values ​​among the stagnant particles are selected and reactivated as active particles.

[0062] After each iteration, the number of stagnant particles in the population is counted, and the ratio of stagnant particles to the total number of particles in the population is calculated to obtain the stagnant particle percentage. This percentage is compared to a preset stagnant particle percentage threshold. The stagnant particle percentage threshold is determined based on the population size. For example, when the total number of particles in the population is 50, the threshold can be set to 0.7, meaning that reactivation is triggered when more than 70% of the particles in the population are in a stagnant state. If the stagnant particle percentage does not exceed the threshold, no reactivation operation is performed, and the process proceeds directly to the next iteration. If the stagnant particle percentage exceeds the threshold, all stagnant particles are sorted from lowest to highest fitness value, and the particles with the worst fitness values ​​are selected as the particles to be reactivated. The number of particles to be reactivated is determined based on the population size. For example, when the total number of particles in the population is 50, the number of particles reactivated each time can be set to 3 to 5.

[0063] For each particle to be reactivated, a new position vector is generated by superimposing random offsets within a preset perturbation range onto each dimension of the current globally optimal particle's position vector, using the current global optimal particle's position vector as a reference. The preset perturbation range is determined based on the rated power of the corresponding stake for each dimension; for example, the perturbation range can be set to ±20% of the rated power. The random offsets for each dimension are generated independently within the perturbation range, ensuring that the perturbation directions and amplitudes for different dimensions are different. In particular, after generating the new position vector, the individual historical optimal position of the reactivated particle is forcibly updated synchronously to the new position vector, and its historical fitness record is cleared to prevent the particle from being bogged down by the gravitational pull of the original local optimal solution in the next iteration and falling into stagnation again.

[0064] Through the aforementioned reactivation mechanism, when a large number of particles in the population converge to their historical optimal positions and cease searching, the stagnant particles with the worst fitness are repositioned to new positions near the global optimum and rejoin the search. These reactivated particles start from the vicinity of the global optimum and explore the unexplored regions around it along different perturbation directions, potentially discovering allocation schemes better than the current global optimum. Simultaneously, since the new positions of the reactivated particles are generated based on the global optimum position rather than being completely randomized, their starting positions are already in relatively favorable regions of the search space, enabling them to converge to valuable solutions within a fewer iterations and avoiding the large amount of invalid searching caused by completely random initialization.

[0065] In this embodiment, the current state evaluation and mutation correction evaluation are performed on active particles, including:

[0066] Based on the current power limit of each slave pile, calculate the total power of the entire station under the allocation scheme represented by the particle, obtain the current evaluation value, replace the power limit of the abruptly changed pile with the charging power after the abrupt change, perform overflow redistribution correction on the allocation scheme represented by the particle, calculate the total power of the entire station based on the corrected allocation scheme and the replaced power limit, and obtain the corrected evaluation value.

[0067] In the standard particle swarm optimization algorithm, fitness evaluation is based solely on the current maximum power capacity of each slave pile, calculating the total power of the entire station under the current state for the allocation scheme represented by the particle. This evaluation method has limitations: when a slave pile is about to transition from the constant current stage to the constant voltage stage within the current scheduling cycle, the battery management system will drastically reduce the maximum power capacity to be accepted from the rated power of the constant current stage to a lower level of the constant voltage stage. If the allocation scheme currently allocates charging power close to the rated power to this slave pile, then after the stage transition occurs, this power will exceed the actual maximum capacity that the slave pile can accept. The excess power cannot be absorbed by the slave pile, nor is it allocated to other slave piles, resulting in a waste of total capacity. Fitness evaluation that only focuses on the current state cannot anticipate this situation. The algorithm tends to select the scheme with the largest current total power of the entire station, ignoring the potential power overflow and capacity waste that may occur after the stage transition.

[0068] To address the aforementioned issues, after each iteration, two adaptive evaluations are performed on each particle. During these evaluations, the allocated power corresponding to each slave in the allocation scheme represented by the particle is compared to its current maximum accepting power. The smaller of these two values ​​is taken as the actual effective power of that slave. The actual effective powers of all slaves are then summed to obtain the total power of the entire station, which is used as the current evaluation value, reflecting the capacity utilization level of the allocation scheme at the current moment. The second evaluation replaces the maximum accepting power of each mutated slave with a lower post-mutation charging power. At this point, the mutated slave cannot absorb the excess power from its original allocated power, resulting in power overflow. Subsequently, overflow redistribution correction is performed on the allocation scheme represented by the particle, simulating the redistribution of this overflow power to non-mutated slaves with remaining accepting capacity. Finally, following the same calculation rules, the updated allocated power of each slave in the corrected allocation scheme is compared with the replaced maximum accepting power, and the smaller value is summed to calculate the total power of the entire station, which is used as the corrected evaluation value. This evaluation value reflects the capacity utilization level that the allocation scheme can maintain after overflow redistribution correction following the phase transition.

[0069] Based on the above introduction, the specific formula for calculating the fitness value of a particle is as follows: ,in This is the current assessment value. To correct the evaluation value, This refers to the mutation weight. In this embodiment, the mutation weight is determined as follows: when there are no mutation charging piles in the charging station, the mutation weight is zero. When mutation charging piles exist, the value with the smallest difference between the state of charge (SOC) value and the conversion threshold among all mutation charging piles is selected. The shortest arrival time is calculated based on this difference and the SOC value growth rate. The mutation weight is: ,in For the shortest arrival time, The duration of the scheduling cycle. The formula sets a preset upper limit for the weight, which increases the weight of the mutation as the distance between the mutation pile and the stage transition increases.

[0070] Using the fitness formula described above, during the algorithm iteration process, if two particles have similar current state evaluation values, but one particle has a significantly higher corrected evaluation value after a mutation, then that particle has a better overall fitness value and is more likely to be selected as the globally optimal particle. This means that the allocation scheme represented by this particle not only has a high capacity utilization level in the current state, but also that after a phase transition, the overflowing power can be effectively absorbed by other slaves, and the overall capacity utilization level will not drop significantly. Conversely, if a particle has the highest current state evaluation value, but after a mutation, a large amount of power overflows and cannot be absorbed by other slaves, its corrected evaluation value is low, and its overall fitness value will be lowered, preventing it from being selected as the optimal scheme by the algorithm. In this way, the algorithm is guided to converge in a direction that balances current capacity utilization and capacity maintenance after mutation during the search process. The final output allocation scheme can maintain a high total power of the entire station before and after a phase transition, avoiding a sharp drop in capacity utilization caused by a phase mutation.

[0071] In this embodiment, the allocation scheme represented by the particle performs overflow redistribution correction, including:

[0072] Calculate the difference between the charging power allocated to the mutation pile in the allocation scheme and the charging power after the mutation. When the difference is positive, record the difference as the overflow amount. When the difference is zero or negative, record the overflow amount as zero. Correct the charging power allocated to the mutation pile in the allocation scheme to the charging power after the mutation. Sum the overflow amounts of all mutation piles into the total overflow amount.

[0073] In the mutation correction assessment, the upper limit of the power accepted by each mutation point is first replaced with the charging power after the mutation. After the replacement, the charging power allocated to each mutation point in the allocation scheme is checked one by one to see if it exceeds the charging power after the mutation. For each mutation point, the charging power allocated to that mutation point in the allocation scheme is subtracted from the charging power after the mutation to obtain the difference. When the difference is positive, it means that the charging power allocated to that mutation point in the allocation scheme exceeds the upper limit of the power accepted by the battery management system after the stage transition, and the excess cannot be absorbed by the battery. This difference is recorded as the overflow of that mutation point. When the difference is zero, it means that the allocated charging power is exactly equal to the charging power after the mutation, and there is no overflow. The overflow is recorded as zero. When the difference is negative, it means that the allocated charging power is lower than the charging power after the mutation. Similarly, there is no overflow. The overflow is recorded as zero.

[0074] After determining the overflow amount, the charging power allocated to the mutation pile in the allocation scheme is corrected to the charging power after the mutation, so that the allocated value of the mutation pile is consistent with the actual acceptable power after the phase transition. The above operation is performed sequentially for all mutation piles, and the overflow amounts of each mutation pile are summed to obtain the total overflow amount. The total overflow amount represents the power capacity released by all mutation piles due to the decrease in the upper limit of the acceptable power after the phase transition occurs.

[0075] The difference between the upper limit of the power accepted by the non-mutation pile and the charging power already allocated in the allocation scheme is calculated as the remaining margin. The total overflow is allocated to each non-mutation pile according to the ratio of the remaining margin of each non-mutation pile to the total remaining margin of all non-mutation piles. The corrected charging power of each non-mutation pile is the original allocated charging power plus the allocated overflow amount, and does not exceed the upper limit of the power accepted. The allocation scheme after overflow redistribution correction is obtained.

[0076] After obtaining the total overflow, it is allocated to each non-mutation charging station. For each non-mutation charging station, the difference between the maximum power it can accept and the already allocated charging power in the allocation scheme is calculated. This difference represents the additional power space that the non-mutation charging station can still accept under the current allocation scheme, denoted as the remaining margin. When the remaining margin is zero, it indicates that the allocated charging power of the non-mutation charging station has reached its maximum power acceptance limit and cannot accept any more power. When the remaining margin is positive, it indicates that the non-mutation charging station still has room to accept additional power. The remaining margins of all non-mutation charging stations are summed to obtain the total remaining margin.

[0077] For each non-mutation charging station, calculate the proportion of its remaining margin to the total remaining margin, and use this proportion as the allocation ratio for that non-mutation charging station. Multiply the total overflow by the allocation ratio to obtain the overflow amount that the non-mutation charging station should receive. Add the allocated overflow amount to the original allocated charging power of the non-mutation charging station to obtain the corrected charging power. If the corrected charging power exceeds the maximum acceptable power of the non-mutation charging station, then the corrected charging power is set as the maximum acceptable power.

[0078] During the iterative process of the particle swarm optimization algorithm, the particle's position vector is updated using a velocity update formula. The velocity update formula is a purely mathematical calculation without any physical constraints. Therefore, the updated position vector may have a large velocity update value in one dimension, causing the charging power in that dimension to exceed the corresponding slave station's maximum or rated power. Alternatively, a large negative velocity update value may cause the charging power in that dimension to become negative. Furthermore, even if the charging power in each dimension is within its legal range, the sum of the charging power in all dimensions may still exceed the total capacity of the charging station. If the position vectors violating constraints are not corrected, infeasible allocation schemes will participate in the fitness evaluation, causing the fitness value to not reflect the actual feasible charging effect and misleading the algorithm's search direction. Therefore, it is necessary to perform constraint correction on the particle's position vector after each iteration's position update and before the fitness evaluation to ensure that all allocation schemes participating in the fitness evaluation are within the feasible region.

[0079] Specifically, dimension values ​​less than zero are set to zero, dimension values ​​exceeding the upper limit of the power received by the corresponding slave pile are set to the upper limit of the power received, and dimension values ​​exceeding the rated power of the corresponding slave pile are set to the rated power.

[0080] Next, the sum of each dimension of the location vector is calculated. When the sum of each dimension exceeds the total capacity of the charging station, the excess between the sum of each dimension and the total capacity is calculated, and the reversal urgency value of the corresponding slave pile for each dimension is calculated. The reversal urgency value is 1 minus the urgency value. The larger the reversal urgency value, the lower the power utilization capacity of the corresponding slave pile. Then, the allocation weight of each dimension is obtained. The allocation weight is the ratio of the reversal urgency value of each dimension to the sum of the reversal urgency values ​​of all dimensions. The excess is multiplied by the allocation weight of each dimension to obtain the deduction amount for each dimension. The charging power of each dimension is subtracted from the deduction amount. If the charging power of a certain dimension is less than zero after deduction, the charging power is set to zero. After the deduction, the sum of each dimension is recalculated. If the sum of each dimension still exceeds the total capacity of the charging station, the dimensions whose charging power has been set to zero are removed. Only the dimensions with remaining charging power greater than zero are used as the basis for recalculating the allocation weight, and the above deduction process is repeated until the sum of each dimension does not exceed the total capacity of the charging station. In this way, dimensions with high urgency values ​​correspond to smaller reversal urgency values, smaller allocated weights, and bear less deductions, while dimensions with low urgency values ​​correspond to larger reversal urgency values, larger allocated weights, and bear greater deductions.

[0081] Specifically, the master charging station continuously monitors the status changes of each slave charging station between scheduling cycles. When any of the following events are detected: a new vehicle completes a charging handshake at a slave charging station, a vehicle connected to a slave charging station disconnects from the charging dock, or the output power of a slave charging station decreases by more than a preset drop threshold within a single sampling interval, an additional particle swarm optimization algorithm is immediately triggered. The additional triggering algorithm uses a reduced number of iterations and uses the globally optimal particle position from the previous solution as the initial position of some particles, thereby achieving an immediate response to sudden events between scheduling cycles.

[0082] like Figure 3 As shown, in the case of a sudden transition from constant current to constant voltage among multiple slave charging piles, the traditional algorithm, due to its lack of prediction of the transition, causes the allocation scheme to fail after the sudden change, resulting in a large amount of idle power. This leads to a significant drop in the total charging power of the entire station and scheduling lag. In contrast, this invention, by locating the sudden change pile in advance and calculating the charging power after the change, combined with the overflow redistribution correction mechanism, can quickly and accurately transfer the excess capacity released by the sudden change to other slave piles with high urgency. This ensures that the total power of the entire station always closely and stably matches the upper limit of the power issued by the power grid, greatly improving the effective power utilization rate of the charging station.

[0083] like Figure 4 As shown, the present invention also provides an intelligent scheduling system for the coordination of charging piles and the power grid. This system is used to implement the methods described above, and includes:

[0084] The status awareness module divides the charging piles in the station into main piles and slave piles. The main piles collect the charging parameters of each slave pile and determine the charging stage of the slave piles based on the charging parameters. The charging stage includes a constant voltage stage and a constant current stage.

[0085] The mutation prediction module locates the mutation pile among the slave piles in the constant current stage, calculates the charging power after the mutation of the mutation pile, and determines the urgency value of the charging stage of all slave piles based on the charging parameters.

[0086] The collaborative optimization module uses the allocated power of each slave pile as the optimization variable and the upper limit of the charging station power issued by the power grid as the constraint. It is solved based on the particle swarm optimization algorithm. In the particle swarm optimization algorithm, the individual factors and social factors of each dimension are independently controlled by the urgency value of the corresponding slave pile.

[0087] The dynamic precision calculation module divides particles into active and stagnant particles based on their search state during the particle swarm algorithm iteration process. It performs fitness evaluations of different precision on active and stagnant particles respectively. The fitness value is determined based on the current state evaluation value and the mutation correction evaluation value. The mutation correction evaluation value is based on the charging power calculation after the mutation of the mutation pile. The master pile sends the charging power command corresponding to each slave pile in the optimal allocation scheme obtained by the particle swarm algorithm to the slave piles for execution.

[0088] It should be noted that the various preset thresholds (e.g., constant current / constant voltage conversion threshold, displacement threshold, stagnation ratio threshold, sudden drop threshold, etc.) and algorithm control parameters (e.g., population size, total number of iterations, upper and lower boundary values ​​of individual factors and social factors, decay constant, etc. in the particle swarm optimization algorithm) involved in the specific embodiments of this invention are all parameters that can be reasonably determined and adjusted by those skilled in the art based on their understanding of the core inventive concept of this invention, according to the specific application scenario of the charging station, the hardware specifications of the battery manufacturer, the actual needs of power grid dispatching, and historical operating data, through a limited number of conventional experiments, simulation tests, or empirical values. The selection of the above-mentioned specific values, the calibration of parameters, and the replacement of conventional mathematical processing methods do not require creative effort and can achieve the technical effects described in this invention, and should not be used as a reason to limit the scope of protection of this invention.

[0089] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0090] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

[0091] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A smart scheduling method for the coordinated operation of charging piles and the power grid, characterized in that, include: The charging piles in the station are divided into main piles and slave piles. The main piles collect the charging parameters of each slave pile, and the charging stage of the slave pile is determined based on the charging parameters. The charging stage includes a constant voltage stage and a constant current stage. In the constant current stage, the sudden change piles are located among the slave piles. For slave piles in the constant current stage, slave piles that reach the conversion threshold within the scheduling cycle are marked as sudden change piles based on the difference between the state of charge value and the constant current constant voltage conversion threshold and the growth rate of the state of charge value. The charging power after the sudden change of the sudden change pile is calculated. Based on the charging parameters, the urgency value of the charging stage of all slave piles is determined. The power acceptance ratio is calculated based on the ratio of the current upper limit of the power received by the vehicle charged by each slave pile to the rated power. The waiting ratio is calculated based on the ratio of the difference between the target state of charge value and the current state of charge value of the vehicle connected to each slave pile to the battery capacity. The product of the power acceptance ratio and the waiting ratio is used as the urgency value of each slave pile. The power allocation of each slave pile is used as the optimization variable, and the upper limit of the charging station power issued by the power grid is used as the constraint. The solution is based on the particle swarm algorithm. In the particle swarm algorithm, the individual factors and social factors of each dimension are independently controlled by the urgency value of the corresponding slave pile. During the iteration of the particle swarm optimization algorithm, particles are divided into active particles and stagnant particles based on their search state. Fitness evaluations of different precision are performed on active particles and stagnant particles respectively. The fitness value is determined based on the current state evaluation value and the mutation correction evaluation value. Specifically, based on the current upper limit of the received power of each slave pile, the allocated power of the slave pile in the allocation scheme represented by each particle is compared with its current upper limit of received power. The smaller value between the two is taken as the actual effective power of the slave pile. The actual effective power of all slave piles is accumulated to obtain the total power of the entire station. The total power of the entire station is used as the current state evaluation value. The mutation correction evaluation value is calculated based on the charging power of the mutated pile after the mutation. The master pile sends the charging power commands corresponding to each slave pile in the optimal allocation scheme obtained by the particle swarm optimization algorithm to the slave piles for execution.

2. The method according to claim 1, characterized in that, Locate the sudden change pile among the piles in the constant current stage, and calculate the charging power of the sudden change pile after the sudden change, including: Based on the battery capacity and the preset attenuation constant, the upper limit of the power that the sudden charging pile can accept after the stage transition is calculated according to the constant voltage stage exponential attenuation formula, and is used as the charging power after the sudden change.

3. The method according to claim 1, characterized in that, In the particle swarm optimization algorithm, the individual and social factors of each dimension are independently controlled by the urgency value of the corresponding stake, including: In the particle swarm optimization algorithm, the position vector of each particle contains the same number of dimensions as the number of slaves, where the d-th dimension of the position vector represents the charging power allocated to the d-th slave. In each iteration, during the velocity update of the d-th dimension of the particle, the individual factor and the social factor are independently determined by the urgency value of the corresponding stake. The social factor increases with the urgency value of the corresponding stake to accelerate the convergence of the corresponding dimension to the global optimum, while the individual factor decreases with the urgency value of the corresponding stake to maintain search diversity in the corresponding dimension.

4. The method according to claim 1, characterized in that, Based on the search state of the particles, they are divided into active particles and stagnant particles, including: After each iteration of position update is completed, the sum of the absolute values ​​of the differences in each dimension between the position vector of each particle and its historical best position vector is calculated and defined as the displacement. If the displacement is greater than the preset displacement threshold, the particle is marked as an active particle; otherwise, the particle is marked as a stationary particle.

5. The method according to claim 4, characterized in that, Perform fitness assessments with different levels of precision for active and stagnant particles, including: Calculate the current state evaluation value and mutation correction evaluation value for active particles. Obtain the fitness value by weighting the current state evaluation value and mutation correction evaluation value. Store the ratio of the fitness value to the current state evaluation value as the correction ratio of the particle. When an active particle becomes a stagnant particle, calculate the current state evaluation value for the stagnant particle. Use the product of the current state evaluation value and the correction ratio as the fitness value of the stagnant particle.

6. The method according to claim 4, characterized in that, Record the number of consecutive stagnant rounds for each stagnant particle. When the number of consecutive stagnant rounds reaches the preset upper limit, force the stagnant particle to perform a current state evaluation and mutation correction evaluation, update the correction ratio, and reset the number of consecutive stagnant rounds.

7. The method according to claim 4, characterized in that, After each iteration, the proportion of stagnant particles in the population is counted. If the proportion of stagnant particles exceeds the preset stagnant proportion threshold, the particles with the worst fitness values ​​are selected from the stagnant particles and reactivated as active particles.

8. The method according to claim 5, characterized in that, Perform current state assessment and mutation correction assessment on active particles, including: Based on the current power limit of each slave pile, calculate the total power of the entire station under the allocation scheme represented by the particle, obtain the current evaluation value, replace the power limit of the abruptly changed pile with the charging power after the abrupt change, perform overflow redistribution correction on the allocation scheme represented by the particle, calculate the total power of the entire station based on the corrected allocation scheme and the replaced power limit, and obtain the corrected evaluation value.

9. The method according to claim 8, characterized in that, Perform overflow redistribution correction on the allocation scheme represented by the particles, including: Calculate the difference between the charging power allocated to the mutation pile in the allocation scheme and the charging power after the mutation. When the difference is positive, the difference is recorded as the overflow amount. When the difference is zero or negative, the overflow amount is recorded as zero. Correct the charging power allocated to the mutation pile in the allocation scheme to the charging power after the mutation. Sum the overflow amounts of all mutation piles into the total overflow amount. The difference between the upper limit of the power accepted by the non-mutation pile and the charging power already allocated in the allocation scheme is calculated as the remaining margin. The total overflow is allocated to each non-mutation pile according to the ratio of the remaining margin of each non-mutation pile to the total remaining margin of all non-mutation piles. The corrected charging power of each non-mutation pile is the original allocated charging power plus the allocated overflow amount, and does not exceed the upper limit of the power accepted. The allocation scheme after overflow redistribution correction is obtained.

10. An intelligent scheduling system for the coordinated operation of charging piles and the power grid, used to implement the method as described in any one of claims 1-9, characterized in that, The system includes: The status awareness module divides the charging piles in the station into main piles and slave piles. The main piles collect the charging parameters of each slave pile and determine the charging stage of the slave piles based on the charging parameters. The charging stage includes a constant voltage stage and a constant current stage. The mutation prediction module locates the mutation pile among the slave piles in the constant current stage, calculates the charging power after the mutation of the mutation pile, and determines the urgency value of the charging stage of all slave piles based on the charging parameters. The collaborative optimization module uses the allocated power of each slave pile as the optimization variable and the upper limit of the charging station power issued by the power grid as the constraint. It is solved based on the particle swarm optimization algorithm. In the particle swarm optimization algorithm, the individual factors and social factors of each dimension are independently controlled by the urgency value of the corresponding slave pile. The dynamic precision calculation module divides particles into active and stagnant particles based on their search state during the particle swarm algorithm iteration process. It performs fitness evaluations of different precision on active and stagnant particles respectively. The fitness value is determined based on the current state evaluation value and the mutation correction evaluation value. The mutation correction evaluation value is based on the charging power calculation after the mutation of the mutation pile. The master pile sends the charging power command corresponding to each slave pile in the optimal allocation scheme obtained by the particle swarm algorithm to the slave piles for execution.

Images