Load prediction-based airport charging pile load management method and system, and medium

Abstract

The invention relates to the technical field of deep learning, in particular to an airport charging pile load management method and system based on load prediction and a medium. Acquiring a current load data set in the airport power grid; inputting the current load data into an improved particle swarm algorithm-long short-term memory neural network model, and predicting predicted load data in a future preset time period, the improved particle swarm algorithm including an adaptive Cauchy variation mechanism; determining a plurality of charging pile load management strategies of the airport according to the predicted load data; on the basis of an improved raccoon optimization algorithm, an optimal charging pile load management strategy in the multiple charging pile load management strategies is determined, and the improved raccoon optimization algorithm comprises a first-time improved search strategy; and carrying out load regulation and control on each charging pile in the airport based on the optimal charging pile load management strategy. The objective of the invention is to solve the problems of how to predict load changes of different power utilization systems in an airport in advance and how to optimize and adjust the load state of an electric vehicle charging pile system in a transformer substation.

Description

Technical Field

[0001] The present invention relates to the field of deep learning technology, and in particular to a method, system and medium for managing airport charging pile load based on load forecasting. Background Art

[0002] With the advancement of the airport's "oil-to-electricity" construction project, fuel-consuming special vehicles in airports will be gradually replaced by new energy special vehicles in the future. Charging piles and other supporting power equipment will also need to be added to the airport. Considering that the basic operation of the airport also requires a large amount of electricity to maintain, how to balance the contradiction between the charging needs of the new new energy special vehicles and the huge power consumption projects of other airport facilities, while ensuring safe electricity use. Realizing green airport construction and avoiding energy waste is a difficult problem that the airport currently needs to solve.

[0003] In the process of conceiving and implementing the present application, the inventors discovered that due to the special nature of the airport's mission, different power systems have different power usage priorities. For example, the lighting system, special area dedicated power supply system, fire protection system and other power systems in the airport with a priority of first-level load have the highest corresponding response priority. Therefore, it is necessary to predict the load changes of different power systems in the airport in advance and optimize the load status of the electric vehicle charging pile system in the substation to achieve reasonable allocation and efficient utilization of limited power resources.

[0004] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of the present invention is to provide an airport charging pile load management method based on load forecasting, aiming to solve the problem of how to predict the load changes of different power systems in the airport in advance and optimize the load state of the electric vehicle charging pile system in the substation.

[0006] To achieve the above object, the present invention provides a method for managing airport charging pile load based on load forecasting, the method comprising:

[0007] Obtain current load data sets in the airport power grid;

[0008] Inputting the current load data into an improved particle swarm algorithm-long short-term memory neural network model to predict the predicted load data within a future preset period, wherein the improved particle swarm algorithm includes an adaptive Cauchy mutation mechanism;

[0009] Determining a load management strategy for multiple charging piles at the airport based on the predicted load data;

[0010] Determine the optimal charging pile load management strategy among the plurality of charging pile load management strategies based on the improved raccoon optimization algorithm, wherein the improved raccoon optimization algorithm includes a first improved search strategy;

[0011] Based on the optimal charging pile load management strategy, load control is performed on each charging pile in the airport.

[0012] Optionally, the data type of the current load data set is a time series data set, and before the step of inputting the current load data into the improved particle swarm algorithm-long short-term memory neural network model to predict the predicted load data within a future preset period, the step further includes:

[0013] Normalizing the current load data according to preset normalization constraints;

[0014] The improved particle swarm algorithm-long short-term memory neural network model includes:

[0015] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0016] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0017]

[0018] o t =σ(Wo·[h t-1 ,x t ]+b o )

[0019] h t =o t *tanh(C t )

[0020] In the formula, x t is the input of the current time step, f t is the activation vector of the forget gate, i t is the activation vector of the input gate, is the candidate memory cell state, Ct is the memory cell state at the current time step, o t is the activation vector of the output gate, h t is the hidden state of the current time step, h t-1 is the hidden state of the previous time step, C t-1is the memory cell state of the previous time step;

[0021] Among them, σ is the sigmoid activation function, b f 、b i 、b C and b o Bias vector, W f , W C , W i and W o is the weight vector.

[0022] Optionally, the preset normalization constraint condition specifically includes:

[0023] x′ α =x α ·(max(x′ α )-min(x′ α ′)+min(x′ α ))

[0024]

[0025] x′ α =min(max(x′ α ′,-3),3)

[0026] In the formula, x α is the normalized data, x′ α is the data after outlier processing, x′ α ′ is the trimmed data, x′ α ″ is the original data, min(x′ α ) is x′ α The minimum value in the data set, max(x′ α ) is x′ α The maximum value in the data set, μ is x′ α ″, δ is the mean of x′ α The standard deviation of ″.

[0027] Optionally, an improved particle swarm algorithm including an adaptive Cauchy mutation mechanism includes:

[0028]

[0029] G best * =G best +F(xm)

[0030] Where ω(t) is the inertia weight, F(xm) is the perturbation function of the standard Cauchy cumulative distribution, coh is the overall cohesion of the particle group, and avg_P best (j) is the individual optimal value P of all particlesbest The average value of r(j) is the current global optimal value G best The distance in the jth dimension, xm(j) is the adaptive variation adjustment variable, G best * It is an adaptive Cauchy mutation operation;

[0031] Among them, ω max and ω min are the maximum and minimum inertia weights, respectively, t is the number of iterations, t max is the maximum number of iterations, is the optimal value P of the i-th individual in the t-th iteration best The value of the position in the jth dimension, is the i-th current global optimal value G in the t-th iteration best The value of the position in the jth dimension, is the average position of the i-th particle in the j-th dimension in the t-th iteration, λ is a constant, r max is the maximum distance between dimensions.

[0032] Optionally, the calculation steps in the improved particle swarm optimization algorithm-long short-term memory neural network model include:

[0033] Determining the current number of iterations in the improved particle swarm algorithm-long short-term memory neural network model;

[0034] When the current number of iterations is in a first preset iteration interval, a disturbance amount is assigned to the current global optimal value based on a disturbance function of the standard Cauchy cumulative distribution.

[0035] Optionally, after the step of determining the current number of iterations in the improved particle swarm algorithm-long short-term memory neural network model, the method further includes:

[0036] When the current number of iterations is within a second preset iteration interval, determining the current cohesion of the particle swarm as a whole, and the average distance between the individual optimal value and the group optimal value;

[0037] When the current cohesion is less than a preset cohesion threshold and the average distance is greater than a preset distance threshold, increasing the perturbation amount in the perturbation function of the standard Cauchy cumulative distribution;

[0038] When the current cohesion is greater than or equal to the preset cohesion threshold and the average distance is less than the preset distance threshold, the disturbance amount is reduced.

[0039] Optionally, the step of determining the optimal charging pile load management strategy among the plurality of charging pile load management strategies based on the improved raccoon optimization algorithm includes:

[0040] Select any charging pile load management strategy as an initial solution, and determine the initial fitness value corresponding to the initial solution;

[0041] Based on a preset neighborhood range, generating a neighbor solution set corresponding to the initial solution, wherein the neighbor solution set includes multiple neighbor solutions;

[0042] Select any neighbor solution from the neighbor solution set as a target neighbor solution, and calculate the fitness value corresponding to the target neighbor solution;

[0043] If the fitness value is greater than the initial fitness value, return to the step of selecting any neighbor solution from the neighbor solution set as the target neighbor solution and calculating the fitness value corresponding to the target neighbor solution until a target neighbor solution with a fitness value less than the initial fitness value appears;

[0044] If the fitness value is less than the initial fitness value, the target neighbor solution corresponding to the fitness value is updated to the optimal solution, and the charging pile load management strategy serving as the optimal solution is determined as the optimal charging pile load management strategy.

[0045] Optionally, the optimal charging pile load management strategy includes an objective function and constraints, wherein:

[0046] The objective function includes:

[0047] min L

[0048] min L p

[0049]

[0050] The constraints include:

[0051]

[0052] C i,R (t)∈{0,1}

[0053] L P ≥P′(t)

[0054] L P -P′(t)≥P o (t)

[0055] Where, L p is the load peak, P o (t) is the airport basic load at time t, C i,R is the charging rate of the i-th vehicle at time t, P iN is the rated charging power of the vehicle, P i(t) is the charging power of the ith vehicle, P′(t) is the grid load at time t after the electric vehicle is connected, t is the monitoring time point divided into 24 hours a day, and N is the total number of time points.

[0056] In addition, to achieve the above-mentioned purpose, the present invention also provides an airport charging pile load management system based on load forecasting, and the airport charging pile load management system based on load forecasting includes: a memory, a processor, and an airport charging pile load management program based on load forecasting stored in the memory and executable on the processor, and the airport charging pile load management program based on load forecasting, when executed by the processor, implements the steps of the airport charging pile load management method based on load forecasting as described above.

[0057] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which is stored an airport charging pile load management program based on load forecasting, and when the airport charging pile load management program based on load forecasting is executed by a processor, the steps of the airport charging pile load management method based on load forecasting as described in the claim are implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is a schematic diagram of the architecture of the hardware operating environment of the load management system for airport charging piles based on load forecasting involved in an embodiment of the present invention;

[0059] Figure 2 It is a flow chart of a first embodiment of a method for managing airport charging pile load based on load forecasting according to the present invention;

[0060] Figure 3 This is a schematic diagram of an electricity metering network for an airport terminal involved in an embodiment of the present invention;

[0061] Figure 4 It is a flow chart of a second embodiment of the method for managing airport charging pile load based on load forecasting of the present invention;

[0062] Figure 5 It is a flow chart of the third embodiment of the method for load management of airport charging piles based on load forecasting of the present invention.

[0063] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0064] The embodiment of the present invention provides an airport charging pile load management method, system and medium based on load forecasting. According to the airport load characteristics, charging pile load characteristics and constraints, with the goal of minimizing the load peak and the minimum sum of squared loads, an LTSM load forecasting model and an airport charging pile management optimization model are established. The accuracy and efficiency of the forecasting model are improved by an improved particle swarm algorithm, and the improved raccoon optimization algorithm is integrated to search for the optimal charging pile management optimization solution, so as to realize the reasonable allocation and efficient utilization of limited power resources.

[0065] In order to better understand the above technical solution, exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0066] As an implementation plan, Figure 1 This is a schematic diagram of the architecture of the hardware operating environment of the load management system for airport charging piles based on load forecasting involved in an embodiment of the present invention.

[0067] like Figure 1 As shown, the load management system for airport charging piles based on load forecasting may include: a processor 1001, such as a CPU, a memory 1005, a user interface 1003, a network interface 1004, and a communication bus 1002. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may optionally be a storage device independent of the aforementioned processor 1001.

[0068] Those skilled in the art will understand that Figure 1 The architecture of the airport charging pile load management system based on load forecasting shown in the figure does not constitute a limitation of the airport charging pile load management system based on load forecasting, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0069] like Figure 1As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and a load management program for airport charging piles based on load forecasting. Among them, the operating system is a program for managing and controlling the hardware and software resources of the load management system for airport charging piles based on load forecasting, the operation of the load management program for airport charging piles based on load forecasting, and other software or programs.

[0070] exist Figure 1 In the load management system for airport charging piles based on load forecasting shown, the user interface 1003 is mainly used to connect to the terminal and communicate data with the terminal; the network interface 1004 is mainly used for the background server and communicates data with the background server; the processor 1001 can be used to call the load management program for airport charging piles based on load forecasting stored in the memory 1005.

[0071] In this embodiment, the load management system for airport charging piles based on load forecasting includes: a memory 1005, a processor 1001, and a load management program for airport charging piles based on load forecasting stored in the memory and executable on the processor, wherein:

[0072] When the processor 1001 calls the airport charging pile load management program based on load forecasting stored in the memory 1005, the following operations are performed:

[0073] Obtain current load data sets in the airport power grid;

[0074] Inputting the current load data into an improved particle swarm algorithm-long short-term memory neural network model to predict the predicted load data within a future preset period, wherein the improved particle swarm algorithm includes an adaptive Cauchy mutation mechanism;

[0075] Determining a load management strategy for multiple charging piles at the airport based on the predicted load data;

[0076] Determine the optimal charging pile load management strategy among the plurality of charging pile load management strategies based on the improved raccoon optimization algorithm, wherein the improved raccoon optimization algorithm includes a first improved search strategy;

[0077] Based on the optimal charging pile load management strategy, load control is performed on each charging pile in the airport.

[0078] When the processor 1001 calls the airport charging pile load management program based on load forecasting stored in the memory 1005, the following operations are performed:

[0079] Normalizing the current load data according to preset normalization constraints;

[0080] When the processor 1001 calls the airport charging pile load management program based on load forecasting stored in the memory 1005, the following operations are performed:

[0081] Determining the current number of iterations in the improved particle swarm algorithm-long short-term memory neural network model;

[0082] When the current number of iterations is in a first preset iteration interval, a disturbance amount is assigned to the current global optimal value based on a disturbance function of the standard Cauchy cumulative distribution.

[0083] When the processor 1001 calls the airport charging pile load management program based on load forecasting stored in the memory 1005, the following operations are performed:

[0084] When the current number of iterations is within a second preset iteration interval, determining the current cohesion of the particle swarm as a whole, and the average distance between the individual optimal value and the group optimal value;

[0085] When the current cohesion is less than a preset cohesion threshold and the average distance is greater than a preset distance threshold, increasing the perturbation amount in the perturbation function of the standard Cauchy cumulative distribution;

[0086] When the current cohesion is greater than or equal to the preset cohesion threshold and the average distance is less than the preset distance threshold, the disturbance amount is reduced.

[0087] When the processor 1001 calls the airport charging pile load management program based on load forecasting stored in the memory 1005, the following operations are performed:

[0088] Select any charging pile load management strategy as an initial solution, and determine the initial fitness value corresponding to the initial solution;

[0089] Based on a preset neighborhood range, generating a neighbor solution set corresponding to the initial solution, wherein the neighbor solution set includes multiple neighbor solutions;

[0090] Select any neighbor solution from the neighbor solution set as a target neighbor solution, and calculate the fitness value corresponding to the target neighbor solution;

[0091] If the fitness value is greater than the initial fitness value, return to the step of selecting any neighbor solution from the neighbor solution set as the target neighbor solution and calculating the fitness value corresponding to the target neighbor solution until a target neighbor solution with a fitness value less than the initial fitness value appears;

[0092] If the fitness value is less than the initial fitness value, the target neighbor solution corresponding to the fitness value is updated to the optimal solution, and the charging pile load management strategy serving as the optimal solution is determined as the optimal charging pile load management strategy.

[0093] Based on the hardware architecture of the airport charging pile load management system based on load forecasting using the above-mentioned deep learning technology, an embodiment of the airport charging pile load management method based on load forecasting of the present invention is proposed.

[0094] Reference Figure 2 In the first embodiment, the load management method of the airport charging pile based on load forecasting includes the following steps:

[0095] Step S10, obtaining a current load data set in the airport power grid;

[0096] In this embodiment, first, an airport charging pile load management system (hereinafter referred to as the system) based on load forecasting obtains a current load data set in the airport power grid.

[0097] Optionally, the current load data set can be collected by a power grid monitoring module pre-installed in the airport, which monitors the load data in the power grid in real time and transmits the data to the background monitoring of the system.

[0098] The current load data set includes the total load required by the entire power grid of the airport at the current moment, and the regional load distribution of each area in the airport. Optionally, the load data set may be a load curve, which records the load changes before the current moment within the day.

[0099] Step S20, inputting the current load data into the improved particle swarm algorithm-long short-term memory neural network model to predict the predicted load data within a future preset period, wherein the improved particle swarm algorithm includes an adaptive Cauchy mutation mechanism;

[0100] In this embodiment, the system inputs the collected current load data set into the improved particle swarm algorithm-long short-term memory neural network model to predict the predicted load data within a preset time period in the future.

[0101] Optionally, the future preset time period may be 15 minutes after the current time.

[0102] In this embodiment, in order to accurately predict load data, we use the long short-term memory network (LSTM) for deep learning. The long short-term memory network (LSTM) is a special recurrent neural network (RNN) suitable for processing and predicting time series data. The LSTM unit contains three gates: forget gate, input gate, output gate, and a memory cell. These components work together to enable LSTM to maintain lossless information transmission over a long span.

[0103] In addition, considering the complex situations such as the working area of ​​special vehicles and the location of charging piles in the airport, and the various types of airport loads, many variables and constraints, the computational complexity of solving the optimal solution is high, and an efficient optimization algorithm needs to be designed. In this embodiment, an improved particle swarm (PSO) algorithm is used to optimize the hyperparameters of LSTM, train the model, find the appropriate hyperparameters, and train the model, so as to find the appropriate hyperparameters.

[0104] It should be noted that the particle swarm optimization (PSO) algorithm is an optimization technology based on group collaboration. It is a global optimization algorithm derived from simulating the foraging behavior of bird flocks in nature. Every possibility in the world is regarded as a particle, and each particle has a different movement direction and speed, moving towards the optimal position. By updating the individual optimal position and the global optimal position, the optimal solution of the objective function is obtained, thereby achieving global optimization.

[0105] The specific expression of PSO algorithm is:

[0106] v i (t+1)=ωv i (t)+c1r1(P best,i -s i (t))+c2r2(G best -s i (t))(1)

[0107] s i (t+1)=s i (t)+v i (t+1)(2)

[0108] Formula (1) is the PSO speed update formula, which means that the speed of each particle is composed of three parts: the first part is based on its current speed (considering inertia), the second part is based on its own optimal historical position (individual learning component), and the third part is based on the optimal position of the entire group (social learning component). The combination of these three parts enables particles to explore and utilize the search space to find the global optimal solution; Formula (2) is the PSO position update formula, which means that the position of the particle is updated according to its speed. In this way, particles can move in the search space and gradually approach the optimal solution.

[0109] Furthermore, although PSO has the ability to converge to the solution of the optimization problem quickly, is applicable to various types of optimization problems and has good parallelism, its performance depends largely on the selection of parameters ω, c1, and c2. Improper parameter settings may lead to slow convergence and easy to fall into local optimal solutions. Therefore, in order to solve the various problems caused by inappropriate parameter settings in PSO, we propose an improved particle swarm algorithm that incorporates adaptive Cauchy mutation to improve the algorithm's parameter optimization ability. Specifically, the improved particle swarm algorithm containing the adaptive Cauchy mutation mechanism includes:

[0110]

[0111] G best * =G best +F(xm)

[0112] Where ω(t) is the inertia weight, F(xm) is the perturbation function of the standard Cauchy cumulative distribution, coh is the overall cohesion of the particle group, and avg_P best (j) is the individual optimal value P of all particles best The average value of r(j) is the current global optimal value G best The distance in the jth dimension, xm(j) is the adaptive variation adjustment variable, G best * It is an adaptive Cauchy mutation operation;

[0113] Among them, ω max and ω min are the maximum and minimum inertia weights, respectively, t is the number of iterations, t max is the maximum number of iterations, is the optimal value P of the i-th individual in the t-th iteration best The value of the position in the jth dimension, is the i-th current global optimal value G in the t-th iteration best The value of the position in the jth dimension, is the average position of the i-th particle in the j-th dimension in the t-th iteration, λ is a constant, r max is the maximum distance between dimensions.

[0114] It should be noted that the larger the Coh value, the smaller the cohesion, the more dispersed the particles, and the smaller the disturbance required; conversely, the smaller the Coh value, the greater the cohesion, the more aggregated the particles, and the larger the disturbance required.

[0115] Furthermore, in this embodiment, the improved particle swarm algorithm-long short-term memory neural network model includes:

[0116] f t =σ(W f ·[h t-1 ,x t ]+b f )(3)

[0117] i t =σ(W i ·[h t-1 ,x t ]+b i )(4)

[0118]

[0119] o t =σ(Wo·[h t-1 ,x t ]+b o )(7)

[0120] h t =o t *tanh(C t )(8)

[0121] In the formula, x t is the input of the current time step, f t is the activation vector of the forget gate, i t is the activation vector of the input gate, is the candidate memory cell state, Ct is the memory cell state at the current time step, o t is the activation vector of the output gate, h t is the hidden state of the current time step, h t-1 is the hidden state of the previous time step, C t-1 is the memory cell state of the previous time step;

[0122] Among them, σ is the sigmoid activation function, b f 、b i 、b C and b o Bias vector, W f , W C , W i and W o is the weight vector.

[0123] It should be noted that formula (3) is the forget gate of the model, which is responsible for deciding which information should be discarded from the cell state; formulas (4) and (5) are the input gates of the model, the former determines which new information will be stored in the cell state, and the latter creates a vector of new candidate values ​​to be added to the state update; formula (6) is the cell state update process, which specifically combines the information of the forget gate and the input gate to update the cell state; formulas (7) and (8) are output gates, the former determines which parts of the next hidden state will be output, and the latter controls the final hidden state and filters it through the tanh function.

[0124] Furthermore, the input current load data needs to be normalized before input, and because the normalization process is very sensitive to outliers, outlier processing is also introduced in this embodiment as a normalization constraint condition. Specifically, the normalization constraint condition specifically includes:

[0125] x′ α =x α ·(max(x′ α )-min(x″ α )+min(x′ α )) (9)

[0126]

[0127] x′ α =min(max(x″ α ,-3),3) (11)

[0128] In the formula, x α is the normalized data, x′ α is the data after outlier processing, x″ α is the trimmed data, x″′ α is the original data, min(x′ α ) is x′ α The minimum value in the data set, max(x′ α ) is x′ α The maximum value in the data set, μ is x″′ α The mean of α The standard deviation of .

[0129] It should be noted that formula (9) represents the reversibility of the normalization process, which can restore the data to the original scale; formula (10) represents the Z-score normalization process; formula (11) represents the data clipping process; formulas (10) and (11) are outlier processing in data preprocessing. Since Min-Max normalization is very sensitive to outliers, outlier processing is a very important step.

[0130] Step S30, determining a load management strategy for multiple charging piles at the airport according to the predicted load data;

[0131] In this embodiment, after the predicted load data is obtained, multiple charging pile load management strategies are determined according to the predicted load data.

[0132] Optionally, when the predicted load data is greater than the current load data obtained in real time, it means that the airport power grid load is large, and the charging pile load management strategy can be formulated based on the power supply safety in the airport, including but not limited to: reducing the charging power of the charging piles and reducing the number of charging piles charging at the same time;

[0133] Optionally, when the predicted load data obtained is less than the current load data obtained in real time, it means that the airport power grid load is small. The obtained charging pile load management strategy can be formulated based on the charging efficiency, which may include but is not limited to: setting the charging power of the currently charging charging pile to the preset maximum safe charging power.

[0134] Optionally, when the obtained predicted load data is equal to the current load data acquired in real time, the current charging pile load management strategy is maintained.

[0135] It is understandable that the load management strategy on the charging pile can be determined or generated in real time based on the predicted load data. Therefore, the charging pile used in this embodiment is an intelligent charging pile that can perform load regulation and management.

[0136] For example, refer to Figure 3 The schematic diagram of the power metering network of the terminal building in the airport is shown. The terminal building switch station mainly includes 5 different low-voltage substations, which are divided into A, B, C, D, and E according to the priority. When the predicted load data is large, the power is supplied to the low-voltage substation with a higher priority first, and the electric vehicle charging pile system with a lower power supply priority is subject to power restriction. Otherwise, the power supply is normal.

[0137] Step S40, determining the optimal charging pile load management strategy among the plurality of charging pile load management strategies based on the improved raccoon optimization algorithm, wherein the improved raccoon optimization algorithm includes a first improved search strategy;

[0138] In this embodiment, since there are multiple charging pile load management strategies based on the predicted load data, in order to obtain an optimal charging pile load management strategy, we use an improved raccoon optimization algorithm including a first improved search strategy to perform optimal selection.

[0139] The basic idea of ​​the Coati Optimization Algorithm (COA) is to simulate two natural behaviors of coatis: (1) attacking and hunting iguanas; (2) escaping from predators. The Coati Optimization Algorithm can simulate the raccoon's intelligent strategies in foraging and defending against predators, making it show excellent adaptability and diversity in solving complex problems.

[0140] In addition, COA has the ability to efficiently handle various optimization problems in various sciences and complex high-dimensional problems, and can effectively explore the potential solution space in the search space, so as to find the global optimal solution. Compared with similar intelligent optimization algorithms, COA balances the resource utilization capabilities of global search and local search during the search process, which enables it to converge at a high speed and provide appropriate values ​​for decision variables in optimization tasks.

[0141] The load management optimization problem requires multiple different solutions, and the raccoon's foraging-escaping from predators behavior can be regarded as multiple survival solutions. Therefore, in this embodiment, the raccoon optimization algorithm is applied to the load management optimization problem. By learning from the raccoon's behavioral strategy, the charging efficiency of special vehicles at the airport is improved and the risk of load overload is minimized.

[0142] Exemplarily, the calculation steps of the raccoon optimization algorithm are as follows:

[0143] (1) First, we need to randomly initialize the population in the optimization space, specifically:

[0144]

[0145] where x i,j For individuals, To find the optimal lower bound, To find the optimal upper bound, r is a random number between [0,1].

[0146] (2) Next, we model the first stage of updating the raccoon population in the search space: we simulate the strategy of raccoons attacking iguanas. In this strategy, a group of raccoons climb a tree to scare an iguana. The other raccoons wait under the tree until the iguana falls to the ground. After the iguana falls to the ground, the raccoons attack and kill it. This strategy causes the raccoons to move to different positions in the search space. Assuming that the position of the best member of the population is the position of the iguana, the mathematical expression of the position is:

[0147]

[0148] in, is the new position of the i-th raccoon in the j-th dimension, which is a random number between [0,1], G jis the position of the iguana in the jth dimension, that is, the position of the best member, I is a number randomly selected from the set {1,2}; N is the number of raccoons, and m is the number of policy variables.

[0149] After the iguana falls to the ground, it is placed at a random position in the search space. Based on this random position, the raccoon on the ground moves in the search space as follows:

[0150]

[0151] In the formula, is the position of the iguana on the ground in the jth dimension; is the objective function value after the iguana falls to the ground in the jth dimension, F i,j is the objective function value of the i-th raccoon in the j-th dimension.

[0152] If the updated individual is better, update the current individual; otherwise keep it as it is. The specific expression is:

[0153]

[0154] In the formula, F i P1 is the objective function value of the i-th raccoon in the new position, F i is the objective function value of the i-th raccoon at its previous position.

[0155] (3) The second stage of the process of updating the raccoon’s position is based on the raccoon’s natural behavior when encountering and fleeing from a predator. When a predator attacks a raccoon, the raccoon will flee from its location. The raccoon’s actions on this strategy result in it being in a safe position close to its current location. This demonstrates the COA algorithm’s ability to exploit local search.

[0156] Suppose that a random position is generated near the position of each raccoon:

[0157]

[0158] in, is the local lower bound of the j-th decision variable, is the local upper bound of the j-th decision variable, t is the number of iterations, and T is the maximum number of iterations.

[0159]

[0160] In the formula, is the new position of the i-th raccoon in the j-th dimension.

[0161] Similarly, if the updated individual is better, update the current individual, otherwise keep it as it is.

[0162]

[0163] Furthermore, as a heuristic search method, the Raccoon Optimization Algorithm has many advantages in solving a variety of engineering optimization problems. However, the single use of the Raccoon Optimization Algorithm is not enough to fully exert its performance potential in the airport load management problem. Therefore, in order to better solve the airport load management problem, this patent introduces the first improvement search strategy (FirstImprovement) to improve the Raccoon Optimization Algorithm. The first improvement search method is an effective local search strategy that can help the optimization algorithm move immediately after finding the first improved solution, thereby speeding up the search and reducing unnecessary calculations to better meet actual needs.

[0164] Step S50: Based on the optimal charging pile load management strategy, load control is performed on each charging pile in the airport.

[0165] In this embodiment, after the optimal charging pile load management strategy is obtained, the load of each charging pile in the airport is controlled based on the optimal charging pile load management strategy.

[0166] Furthermore, in this embodiment, the optimal charging pile load management strategy includes an objective function and constraints. Specifically, the objective function includes:

[0167] lm

[0168] min L p

[0169] Indicates the load peak value L p minimize;

[0170]

[0171] It means that the sum of squares of the grid load P′(t) at time t after the electric car is connected is the smallest;

[0172] Specifically, the constraints include:

[0173]

[0174] C i,R (t)∈{0,1} (14)

[0176] L P ≥P′(t) (15)

[0178] L P -P′(t)≥P o (t) (16)

[0180] Where, L p is the load peak, P o (t) is the airport basic load at time t, C i,R is the charging rate of the i-th vehicle at time t, P iN is the rated charging power of the vehicle, P i (t) is the charging power of the ith vehicle, P′(t) is the grid load at time t after the electric vehicle is connected, t is the monitoring time point divided into 24 hours a day, and N is the total number of time points.

[0181] It should be noted that formula (12) indicates that the peak load is not less than the load constraint at all time points throughout the day; formulas (13) and (14) are user charging rate constraints. Formula (15) indicates that the peak load is not greater than the peak load constraint obtained by the first stage optimization; formula (16) indicates that the load of the charging pile at any time will not affect the use of other facilities in the airport.

[0182] In the technical solution provided in this embodiment, according to the airport load characteristics, charging pile load characteristics and constraints, with the goal of minimizing the load peak and the minimum sum of squared loads, an LTSM load forecasting model and an airport charging pile management optimization model are established. The accuracy and efficiency of the forecasting model are improved by an improved particle swarm algorithm, and the improved raccoon optimization algorithm is integrated to search for the optimal charging pile management optimization solution, so as to achieve the rational allocation and efficient utilization of limited power resources.

[0183] Second embodiment

[0184] Reference Figure 4 As an optional embodiment, based on the first embodiment, in this embodiment, the calculation steps of the improved particle swarm algorithm of the above-mentioned adaptive Cauchy mutation mechanism can be as follows:

[0185] Step S201, determining the current number of iterations in the improved particle swarm algorithm-long short-term memory neural network model;

[0186] Step S202: when the current number of iterations is within a first preset iteration interval, a disturbance amount is assigned to the current global optimal value based on a disturbance function of the standard Cauchy cumulative distribution.

[0187] In this calculation step, the current iteration number in the improved particle swarm optimization-long short-term memory neural network model is first determined. If the current iteration number is in the first preset iteration interval of the initial iteration stage, the current global optimal value (G best ) will generate a larger disturbance through the perturbation function of the standard Cauchy cumulative distribution and assign the current global optimal value to increase the diversity of the particle swarm, and combined with a higher initial inertia weight, enable particles to explore a wider solution space.

[0188] That is: G best * =G best +F(xm);

[0189] Optionally, the first preset iteration interval may be [0,50];

[0190] Optionally, in this embodiment, after step S201, the following steps are further included:

[0191] Step S203, when the current number of iterations is in the second preset iteration interval, determining the current cohesion of the particle swarm as a whole, and the average distance between the individual optimal value and the group optimal value;

[0192] Step S204, when the current cohesion is less than a preset cohesion threshold and the average distance is greater than a preset distance threshold, increasing the perturbation amount in the perturbation function of the standard Cauchy cumulative distribution;

[0193] Step S205: when the current cohesion is greater than or equal to the preset cohesion threshold and the average distance is less than the preset distance threshold, reduce the disturbance amount.

[0194] If the current iteration number is in the second preset iteration interval of the late iteration stage, in order to prevent the particle swarm in the late iteration from falling into the local optimal solution, the disturbance amount is adaptively increased or decreased according to the current cohesion of the particle swarm and the average distance between the individual optimal value and the group optimal value, thereby automatically adjusting the variation amplitude.

[0195] Specifically, if the current cohesion is less than the preset cohesion threshold and the average distance is greater than the preset distance threshold, the disturbance amount is increased; if the current cohesion is greater than or equal to the preset cohesion threshold and the average distance is less than the preset distance threshold, the disturbance amount is reduced.

[0196] Among them, the cohesion threshold and the preset distance threshold are set by the tester according to the specific situation.

[0197] In the technical solution provided in this embodiment, at the beginning of the iteration, a larger disturbance amount is given to the current global optimal value through the disturbance function of the standard Cauchy cumulative distribution to increase the diversity of the particle swarm, and combined with a higher initial inertia weight, the particles can explore a wider solution space. In the late iteration, in order to prevent the particle swarm in the late iteration from falling into the local optimal solution, the disturbance amount is adaptively increased or decreased according to the current cohesion of the particle swarm and the average distance between the individual optimal value and the group optimal value, thereby automatically adjusting the variation range.

[0198] Third embodiment

[0199] As an optional embodiment, based on the first embodiment, in this embodiment, the first improved search is a method widely used in local search strategies. Its basic principle is that during the search process, once a neighbor solution that is better than the current solution is found, this solution is immediately adopted as the new current solution and the search process is continued, rather than continuing to explore other neighbors. This strategy aims to quickly move to a better solution and reduce the time to calculate neighbor solutions.

[0200] Reference Figure 5 , the specific determination steps include:

[0201] Step S301, selecting any charging pile load management strategy as an initial solution, and determining an initial fitness value corresponding to the initial solution;

[0202] Step S302, generating a neighbor solution set corresponding to the initial solution based on a preset neighborhood range, wherein the neighbor solution set includes multiple neighbor solutions;

[0203] Step S303, selecting any neighbor solution in the neighbor solution set as a target neighbor solution, and calculating a fitness value corresponding to the target neighbor solution;

[0204] Step S304, if the fitness value is greater than the initial fitness value, return to the step of selecting any neighbor solution from the neighbor solution set as a target neighbor solution and calculating the fitness value corresponding to the target neighbor solution until a target neighbor solution with a fitness value less than the initial fitness value appears;

[0205] Step S305: If the fitness value is less than the initial fitness value, the target neighbor solution corresponding to the fitness value is updated to the optimal solution, and the charging pile load management strategy serving as the optimal solution is determined as the optimal charging pile load management strategy.

[0206] Here is an example of logic based on the steps above:

[0207] Assume there is a current solution x, and the optimization goal is to minimize an objective function f(x). The first improvement search step can be mathematically expressed as follows:

[0208] (1) Initialization

[0209] x←x0

[0210] Choose an initial solution as a starting point, where x0 is the initial solution.

[0211] (2) Result iteration

[0212] By making small modifications to the result x, a neighbor solution set N(x) is generated from x. Each neighbor solution x′∈N(x) is examined. When f(x′)<f(x) is found, x←x′ is executed and the loop is exited. That is, once the first neighbor solution that improves the objective function value is found, this solution is accepted.

[0213] In the technical solution provided in this embodiment, the particle swarm algorithm is improved and adaptive Cauchy mutation is incorporated to improve the parameter optimization capability of the algorithm.

[0214] In addition, it can be understood by a person skilled in the art that all or part of the processes in the method for implementing the above embodiment can be completed by instructing the relevant hardware through a computer program. The computer program includes program instructions, and the computer program can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the load management system for airport charging piles based on load forecasting to implement the process steps of the embodiment of the above method.

[0215] Therefore, the present invention also provides a computer-readable storage medium, which stores an airport charging pile load management program based on load forecasting. When the airport charging pile load management program based on load forecasting is executed by a processor, the various steps of the airport charging pile load management method based on load forecasting as described in the above embodiment are implemented.

[0216] The computer-readable storage medium may be a USB flash drive, a mobile hard disk, a read-only memory (ROM), a magnetic disk, or an optical disk, etc., which may store program codes.

[0217] It should be noted that since the storage medium provided in the embodiment of the present application is the storage medium used to implement the method of the embodiment of the present application, based on the method introduced in the embodiment of the present application, the person skilled in the art can understand the specific structure and deformation of the storage medium, so it is not repeated here. All storage media used in the method of the embodiment of the present application belong to the scope of protection of this application.

[0218] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0219] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0220] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0221] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0222] It should be noted that in the claims, any reference signs placed between parentheses shall not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claim. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention may be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names.

[0223] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0224] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for managing airport charging pile load based on load forecasting, characterized in that: The method comprises the following steps: Obtain current load data sets in the airport power grid; Inputting the current load data into an improved particle swarm algorithm-long short-term memory neural network model to predict the predicted load data within a future preset period, wherein the improved particle swarm algorithm includes an adaptive Cauchy mutation mechanism; Determining a load management strategy for multiple charging piles at the airport based on the predicted load data; Determine the optimal charging pile load management strategy among the plurality of charging pile load management strategies based on the improved raccoon optimization algorithm, wherein the improved raccoon optimization algorithm includes a first improved search strategy; Based on the optimal charging pile load management strategy, load control is performed on each charging pile in the airport.

2. The method according to claim 1, characterized in that The data type of the current load data set is a time series data set. Before the step of inputting the current load data into the improved particle swarm algorithm-long short-term memory neural network model to predict the predicted load data within a future preset period, the step further includes: Normalizing the current load data according to preset normalization constraints; The improved particle swarm algorithm-long short-term memory neural network model includes: f t =σ(W f ·[h t-1 ,x t ]+b f ) i t =σ(W i ·[h t-1 ,x t ]+b i ) O t =σ(Wo·[h t-1 ,x t ]+b o ) h t =o t *tanh(C t ) In the formula, x t is the input of the current time step, f t is the activation vector of the forget gate, i t is the activation vector of the input gate, is the candidate memory cell state, Ct is the memory cell state at the current time step, o t is the activation vector of the output gate, h t is the hidden state of the current time step, h t-1 is the hidden state of the previous time step, C t-1 is the memory cell state of the previous time step; Among them, σ is the sigmoid activation function, b f , b i , b C and b o Bias vector, W f , W C , W i and W o is the weight vector.

3. The method according to claim 2, characterized in that The preset normalization constraint conditions specifically include: x′ α =x α ·(max(x′ α )-min(x′ α ′)+min(x′ α )) x′ α =min(max(x′ α ′,-3),3) In the formula, x α is the normalized data, x′ α is the data after outlier processing, x′ α ′ is the trimmed data, x′ α ″ is the original data, min(x′ α ) is x′ α The minimum value in the data set, max(x′ α ) is x′ α The maximum value in the data set, μ is x′ α ″, δ is the mean of x′ α The standard deviation of ″.

4. The method according to claim 1, characterized in that The improved particle swarm algorithm with adaptive Cauchy mutation mechanism includes: G best * =G best +F(xm) Where ω(t) is the inertia weight, F(xm) is the perturbation function of the standard Cauchy cumulative distribution, coh is the overall cohesion of the particle group, and avg_P best (j) is the individual optimal value P of all particles best The average value of, r(j) is the current global optimal value G best The distance in the jth dimension, xm(j) is the adaptive variation adjustment variable, G best * It is an adaptive Cauchy mutation operation; Among them, ω max and ω min are the maximum and minimum inertia weights, respectively, t is the number of iterations, t max is the maximum number of iterations, is the optimal value P of the i-th individual in the t-th iteration best The value of the position in the jth dimension, is the i-th current global optimal value G in the t-th iteration best The value of the position in the jth dimension, is the average position of the i-th particle in the j-th dimension in the t-th iteration, λ is a constant, r max is the maximum distance between dimensions.

5. The method according to claim 4, characterized in that The calculation steps in the improved particle swarm algorithm-long short-term memory neural network model include: Determining the current number of iterations in the improved particle swarm algorithm-long short-term memory neural network model; When the current number of iterations is in a first preset iteration interval, a disturbance amount is assigned to the current global optimal value based on a disturbance function of the standard Cauchy cumulative distribution.

6. The method according to claim 5, characterized in that After the step of determining the current number of iterations in the improved particle swarm algorithm-long short-term memory neural network model, the method further includes: When the current number of iterations is within a second preset iteration interval, determining the current cohesion of the particle swarm as a whole, and the average distance between the individual optimal value and the group optimal value; When the current cohesion is less than a preset cohesion threshold and the average distance is greater than a preset distance threshold, increasing the perturbation amount in the perturbation function of the standard Cauchy cumulative distribution; When the current cohesion is greater than or equal to the preset cohesion threshold and the average distance is less than the preset distance threshold, the disturbance amount is reduced.

7. The method according to claim 1, characterized in that The step of determining the optimal charging pile load management strategy among the plurality of charging pile load management strategies based on the improved raccoon optimization algorithm comprises: Select any charging pile load management strategy as an initial solution, and determine the initial fitness value corresponding to the initial solution; Based on a preset neighborhood range, generating a neighbor solution set corresponding to the initial solution, wherein the neighbor solution set includes multiple neighbor solutions; Select any neighbor solution from the neighbor solution set as a target neighbor solution, and calculate the fitness value corresponding to the target neighbor solution; If the fitness value is greater than the initial fitness value, return to the step of selecting any neighbor solution from the neighbor solution set as the target neighbor solution and calculating the fitness value corresponding to the target neighbor solution until a target neighbor solution with a fitness value less than the initial fitness value appears; If the fitness value is less than the initial fitness value, the target neighbor solution corresponding to the fitness value is updated to the optimal solution, and the charging pile load management strategy serving as the optimal solution is determined as the optimal charging pile load management strategy.

8. The method according to claim 1, characterized in that The optimal charging pile load management strategy includes an objective function and constraints, wherein: The objective function includes: min L p The constraints include: C i,R (t)∈{0,1} L P ≥P′(t) L P -P′(t)≥P o (t) Where, L p is the load peak, P o (t) is the airport basic load at time t, C i,R is the charging rate of the i-th vehicle at time t, P iN is the rated charging power of the vehicle, P i (t) is the charging power of the ith vehicle, P′(t) is the grid load at time t after the electric vehicle is connected, t is the monitoring time point divided into 24 hours a day, and N is the total number of time points.

9. An airport charging pile load management system based on load forecasting, characterized in that: The load management system for airport charging piles based on load forecasting comprises: a memory, a processor and a load management program for airport charging piles based on load forecasting stored in the memory and executable on the processor. When the load management program for airport charging piles based on load forecasting is executed by the processor, the steps of the load management method for airport charging piles based on load forecasting are implemented as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores an airport charging pile load management program based on load forecasting, and when the airport charging pile load management program based on load forecasting is executed by a processor, the steps of the airport charging pile load management method based on load forecasting as described in any one of claims 1 to 8 are implemented.

Images