An intelligent scheduling method and system for a photovoltaic storage and charging microgrid system

Abstract

The present invention provides an intelligent dispatching method and system for a photovoltaic storage and charging microgrid system, which relates to the technical field of power dispatching, including: obtaining the current photovoltaic installed capacity, the current energy storage installed capacity and the corresponding current collection time, the current weather data and the current electricity price to form a first eigenvector; according to the current photovoltaic power generation power sequence of the park, the current charging power sequence of the charging pile system and the current load power sequence of the remaining loads in the park, respectively predicting the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to the future time period to form a second eigenvector; inputting the first eigenvector and the second eigenvector into an intelligent dispatching model trained under the goal of maximizing the electricity consumption income of the park and a corresponding constraint condition to predict the predicted energy storage capacity of the energy storage system. Comprehensively considering multiple characteristic indicators, the prediction accuracy is improved, and the stability and economy of the photovoltaic storage and charging microgrid system are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power dispatching, and in particular to an intelligent dispatching method and system for a photovoltaic storage and charging microgrid system. Background Art

[0002] As the global carbon emission reduction process continues to deepen, the integration of "photovoltaic storage and charging" will be the core energy system or business model to promote further reduction of carbon emissions, and it is also an important means to promote the realization of "carbon neutrality". Photovoltaic, energy storage and charging facilities form a microgrid, which interacts intelligently with the public power grid according to demand. The application of energy storage system can also alleviate the impact of high-current charging of charging piles on the regional power grid. The integrated "photovoltaic storage and charging" charging station can not only supply green electricity to electric vehicles, but also realize auxiliary service functions such as power peak shaving and valley filling, which can effectively improve the system operation efficiency.

[0003] However, since the photovoltaic, energy storage and charging microgrid system involves the combination of multiple systems such as photovoltaic system, energy storage system, charging pile system, and photovoltaic power generation is greatly affected by factors such as weather, and the power generation power is highly random and volatile, higher requirements are placed on the complexity, safety, stability and power requirements of the intelligent scheduling of the photovoltaic, energy storage and charging system. The unreasonable scheduling of the photovoltaic, energy storage and charging microgrid will affect the safety and stability of the power grid, as well as the economy of the system. Therefore, there is an urgent need for an intelligent scheduling method that can improve the stability and economy of the photovoltaic, energy storage and charging microgrid system. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides an intelligent scheduling method for a photovoltaic storage and charging microgrid system, wherein the photovoltaic storage and charging microgrid system includes a photovoltaic system, an energy storage system and a charging pile system, including:

[0005] Step S1, respectively obtaining the current photovoltaic installed capacity of the photovoltaic system of a park, the current energy storage installed capacity of the energy storage system, and the corresponding current collection time, current weather data and current electricity price to form a first feature vector;

[0006] Step S2, based on the current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the remaining loads of the park, respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to a future period after the current collection time and form a second feature vector;

[0007] Step S3, input the first eigenvector and the second eigenvector into an intelligent scheduling model trained with the goal of maximizing the electricity consumption revenue of the park and under a corresponding constraint condition, so as to predict the predicted energy storage power of the energy storage system in the future time period after the current collection time, so as to provide intelligent scheduling for the photovoltaic storage and charging microgrid system.

[0008] Preferably, before executing step S1, a model training process is also included, including:

[0009] Step A1, respectively obtaining the historical photovoltaic installed capacity of the photovoltaic system, the historical energy storage installed capacity of the energy storage system, and the corresponding historical collection time, historical weather data, and historical electricity prices to form a historical feature vector;

[0010] Step A2, according to the acquired historical photovoltaic power generation sequence of the photovoltaic system, the historical charging power sequence of the charging pile system and the historical load power sequence of the remaining loads, respectively predict the corresponding historical predicted photovoltaic power generation power, historical predicted charging power and historical predicted load power;

[0011] Step A3, with the goal of maximizing the electricity consumption revenue of the park and under the corresponding constraints, the historical feature vector, the historical predicted photovoltaic power generation power, the historical predicted charging power and the historical predicted load power are used as input, and the energy storage capacity of the energy storage system is used as output, to train the intelligent scheduling model.

[0012] Preferably, the historical collection time includes the data collection moments of the historical photovoltaic installed capacity and the historical energy storage installed capacity, and the date type and season to which the data collection moments belong.

[0013] Preferably, the objective function corresponding to the goal of maximizing the electricity revenue of the park is:

[0014] Y=p*(Q PY +Q ES -Q EC -Q EV -Q load -Q grid )

[0015] The constraints are:

[0016] P PV +P ES -P EV -P load =P grid

[0017] 0<SOC<100

[0018] Among them, Y represents the electricity income, p represents the electricity price, Q PY represents the photovoltaic power generation of the photovoltaic system, Q ES Indicates the energy storage and discharge capacity of the energy storage system, Q EC represents the energy storage charging capacity of the energy storage system, Q EV Indicates the charging capacity of the charging pile, Q load Indicates the power consumption of the remaining loads, Q grid Represents the power grid electricity of the park; P PY represents the photovoltaic power generation power of the photovoltaic system, P ES represents the energy storage power generation of the energy storage system, P EV represents the charging power of electric vehicles in the photovoltaic storage and charging microgrid system, P grid It represents the power generation power of the power grid of the park, and SOC represents the energy storage capacity of the energy storage system.

[0019] Preferably, step A1 also includes removing abnormal data from the acquired historical photovoltaic installed capacity, the historical energy storage installed capacity and the corresponding historical collection time, the historical weather data and the historical electricity price, and forming the historical feature vector based on the historical photovoltaic installed capacity, the historical energy storage installed capacity and the corresponding historical collection time, the historical weather data and the historical electricity price after the abnormal data removal processing.

[0020] Preferably, in step A1, a clustering algorithm is used to screen abnormal data, and the screened abnormal data is eliminated.

[0021] Preferably, step S2 comprises:

[0022] Step S21, based on the acquired current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system, and the current load power sequence of the remaining loads of the park, respectively predict the predicted photovoltaic power generation power, the predicted charging power, and the predicted load power corresponding to the future time period after the current collection time;

[0023] Step S22, respectively obtaining the current total number of vehicles and the current total number of people in the park, and correcting the predicted charging power according to the current total number of vehicles to obtain a corresponding corrected charging power, and correcting the predicted load power according to the current total number of people to obtain a corresponding corrected load power;

[0024] Step S23, forming the second feature vector according to the predicted photovoltaic power generation power, the corrected charging power and the corrected load power.

[0025] Preferably, in step S21, the current photovoltaic power generation sequence, the current charging power sequence and the current load power sequence are respectively input into a first time series model, a second time series model and a third time series model obtained in advance by training, so as to respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to the future time period after the current collection time.

[0026] The present invention also provides an intelligent dispatching system for a photovoltaic storage and charging microgrid system, applying the above-mentioned intelligent dispatching method, the intelligent dispatching system comprises:

[0027] A data acquisition module, used to respectively acquire the current photovoltaic installed capacity of the photovoltaic system of a park, the current energy storage installed capacity of the energy storage system, and the corresponding current collection time, current weather data and current electricity price to form a first feature vector;

[0028] A power prediction module, configured to respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to a future period after the current collection time according to the acquired current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the remaining loads of the park, and form a second feature vector;

[0029] The scheduling prediction module is connected to the data acquisition module and the power prediction module respectively, and is used to input the first feature vector and the second feature vector into an intelligent scheduling model trained with the goal of maximizing the electricity consumption benefit of the park and a corresponding constraint condition, so as to predict the predicted energy storage power of the energy storage system in the future time period after the current collection time, so as to provide intelligent scheduling for the photovoltaic storage and charging microgrid system.

[0030] Preferably, the power prediction module includes:

[0031] A prediction unit, configured to respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to the future time period after the current collection time according to the acquired current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the remaining loads of the park;

[0032] a result correction unit, connected to the prediction unit, for respectively obtaining the current total number of vehicles and the current total number of people in the park, and correcting the predicted charging power according to the current total number of vehicles to obtain a corresponding corrected charging power, and correcting the predicted load power according to the current total number of people to obtain a corresponding corrected load power;

[0033] A vector generating unit is connected to the prediction unit and the result correcting unit respectively, and is used to form the second feature vector according to the predicted photovoltaic power generation power, the corrected charging power and the corrected load power.

[0034] The above technical solution has the following advantages or beneficial effects:

[0035] 1) The intelligent dispatching model trained based on the goal of maximizing electricity revenue and corresponding constraints is used to predict the energy storage power. Multiple characteristic indicators that affect the photovoltaic storage and charging microgrid system are comprehensively considered to improve the prediction accuracy, so that the dispatching plan formulated based on the predicted energy storage power is more reasonable, which can effectively improve the user experience, reduce charging costs, extend the life of energy storage batteries, improve the load characteristics of the distribution network, and effectively improve the stability and economy of the photovoltaic storage and charging microgrid system;

[0036] 2) Considering that the charging data of charging piles and the changes in other loads in the park are greatly affected by the date, and the amount of photovoltaic power generation is greatly affected by the season and weather, the data collection time is associated with characteristic indicators such as period type and season, which effectively improves the prediction accuracy of the intelligent scheduling model;

[0037] 3) Based on the current total number of vehicles and the total number of people in the park, the predicted charging power and predicted load power are adaptively corrected to make them closer to the current scenario, effectively improving the prediction accuracy of the intelligent scheduling model;

[0038] 4) The intelligent scheduling model can be applied to both large and small sample numbers, and is more flexible in the use of convolutional neural networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A schematic flow chart of an intelligent scheduling method for a photovoltaic storage and charging microgrid system in a preferred embodiment of the present invention;

[0040] Figure 2 A flow chart of a model training process in a preferred embodiment of the present invention;

[0041] Figure 3 Schematic diagram of the process of generating the second feature vector in a preferred embodiment of the present invention;

[0042] Figure 4 The figure is a schematic diagram of the structure of an intelligent dispatching system of a photovoltaic storage and charging microgrid system in a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0043] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The present invention is not limited to this embodiment, and other embodiments may also fall within the scope of the present invention as long as they conform to the gist of the present invention.

[0044] In a preferred embodiment of the present invention, based on the above problems existing in the prior art, an intelligent scheduling method for a photovoltaic storage and charging microgrid system is provided. The photovoltaic storage and charging microgrid system includes a photovoltaic system, an energy storage system and a charging pile system. Figure 1 As shown, including:

[0045] Step S1, respectively obtaining the current photovoltaic installed capacity of a photovoltaic system of a park, the current energy storage installed capacity of an energy storage system, and the corresponding current collection time, current weather data, and current electricity price to form a first feature vector;

[0046] Step S2, based on the current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the rest of the loads in the park, respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to a future period after the current collection time and form a second feature vector;

[0047] Step S3, input the first eigenvector and the second eigenvector into an intelligent scheduling model trained with the goal of maximizing the electricity revenue of the park and a corresponding constraint condition, so as to predict the predicted energy storage capacity of the energy storage system in the future period after the current collection time, so as to provide intelligent scheduling for the photovoltaic storage and charging microgrid system.

[0048] Specifically, in this embodiment, the first eigenvector and the second eigenvector can be data collected during a certain time period in the morning of the same day, such as from 10:00 to 12:00 in the morning, and the corresponding predicted future time period can be the predicted energy storage power of the corresponding time period in the afternoon or the corresponding time period of the next day. The predicted energy storage power can be a positive value or a negative value, wherein a positive value is used to represent the discharge power of the energy storage system, and a negative value is used to represent the charge power of the energy storage system.

[0049] Furthermore, the prediction of energy storage power comprehensively considers the current photovoltaic installed capacity of the photovoltaic system, the current energy storage installed capacity of the energy storage system and the corresponding current collection time, current weather data and current electricity price, while considering the predicted photovoltaic power generation power, predicted charging power and predicted load power corresponding to the future time period after the current collection time, thereby improving the prediction accuracy, thereby making the scheduling plan based on the predicted energy storage power more reasonable, which can effectively improve the user experience, reduce charging costs, extend the life of energy storage batteries, improve the load characteristics of the distribution network, and effectively improve the stability and economy of the photovoltaic storage and charging microgrid system.

[0050] In a preferred embodiment of the present invention, before executing step S1, Figure 2 As shown, a model training process is also included, including:

[0051] Step A1, respectively obtaining the historical photovoltaic installed capacity of the photovoltaic system, the historical energy storage installed capacity of the energy storage system, and the corresponding historical collection time, historical weather data, and historical electricity prices to form a historical feature vector;

[0052] Step A2, according to the obtained historical photovoltaic power generation sequence of the photovoltaic system, the historical charging power sequence of the charging pile system and the historical load power sequence of other loads, respectively predict the corresponding historical predicted photovoltaic power generation power, historical predicted charging power and historical predicted load power;

[0053] In step A3, with the goal of maximizing the electricity consumption benefits of the park and under the corresponding constraints, the historical feature vector, the historical predicted photovoltaic power generation power, the historical predicted charging power and the historical predicted load power are used as inputs, and the energy storage capacity of the energy storage system is used as output to train an intelligent scheduling model.

[0054] Specifically, in this embodiment, the above-mentioned historical collection time can be the time of the past year, and the above-mentioned historical feature vector is a feature vector matrix. Preferably, the historical collection time includes the data collection time of the historical photovoltaic installed capacity and the historical energy storage installed capacity, as well as the date type and season to which the data collection time belongs. Taking into account that the charging data of the charging piles and the changes in the remaining loads in the park are greatly affected by the date, a characteristic indicator is added: "Date type", which includes but is not limited to three types: working days, weekends, and statutory holidays. Taking into account that the amount of photovoltaic power generation is mainly affected by the season and weather, a characteristic indicator is added: season, and the characteristic values ​​of the seasonal characteristic indicator include but are not limited to summer and non-summer.

[0055] Further preferably, the above-mentioned intelligent scheduling model is a feedforward neural network including convolution calculation and having a deep structure, and the convolution kernel parameter sharing and the sparsity of the inter-layer connection in the hidden layer enable the convolution neural network to grid the features with a small amount of calculation, which is a mechanism model obtained by multi-layer dimensionality reduction and feature extraction of the n*n input matrix. Among them, the n*n matrix is ​​a purely numerical matrix (including time series variables), which includes the first eigenvector and the second eigenvector respectively, wherein the second eigenvector is a time series variable. The input features of the convolution neural network need to be standardized when learning using the gradient descent algorithm. Specifically, before the learning data is input into the convolution neural network, the input data needs to be normalized in the channel or time / frequency dimension. The hidden layer of the convolution neural network includes a convolution layer, a pooling layer, and a fully connected layer. The dimensionality reduction layer and the feature extraction layer are used alternately and multiple times to achieve the effect of deep learning.

[0056] Among them, dimensionality reduction is to reduce data redundancy, increase the speed of calculation and ensure reliability. The main means of dimensionality reduction include low-frequency sampling, averaging of surrounding values, etc. Feature extraction is to filter out unnecessary data through filtering tools. The repeated use of dimensionality reduction and feature extraction can ensure the retention of feature data.

[0057] As a preferred implementation, in actual application scenarios, model training may involve too much or too little data. To ensure the prediction accuracy of the training model:

[0058] When the amount of data is too small, full sampling can be used for modeling and parameter adjustment, and the parameters and models can be iterated as the number of samples increases during real-time operation.

[0059] In the case of too much data, you can take the method of sampling part of the data and use part of the sample to replace the full sample. Ensuring that the sampling rate conforms to the law of large numbers can ensure the effectiveness of the model.

[0060] In a preferred embodiment of the present invention, the objective function corresponding to the goal of maximizing the electricity revenue of the park is:

[0061] Y=p*(Q PY +Q ES -Q EC -Q EV -Q load -Q grid )

[0062] The constraints are:

[0063] P PV +P ES -P EV -P load =P grid

[0064] 0<SOC<100

[0065] Among them, Y represents the income from electricity consumption, p represents the electricity price, and Q PY Represents the photovoltaic power generation of the photovoltaic system, Q ES Indicates the energy storage discharge capacity of the energy storage system, Q EC Indicates the energy storage charging capacity of the energy storage system, Q EV Indicates the charging capacity of the charging pile, Q load Indicates the power consumption of the remaining loads, Q grid Indicates the power of the park's power grid; P PY Represents the photovoltaic power generation of the photovoltaic system, P ES Represents the energy storage power of the energy storage system, P EVrepresents the charging power of electric vehicles in the photovoltaic storage and charging microgrid system, P grid It represents the power generation power of the park's power grid, and SOC represents the energy storage capacity of the energy storage system.

[0066] Specifically, in this embodiment, the dispatch control of the photovoltaic storage and charging microgrid system is aimed at economic efficiency as a whole, mainly involving photovoltaic power generation costs, energy storage charging and discharging benefits, and grid charging costs. Based on this, the above objective function is constructed. Among them, the product of the electricity price and the photovoltaic power generation represents the photovoltaic power generation cost, the product of the difference between the energy storage charging power and the charging power of the charging pile and the power consumption of the remaining loads and the electricity price represents the energy storage charging and discharging benefits, and the product of the electricity price and the grid power represents the grid charging cost.

[0067] To be more specific, the photovoltaic storage and charging microgrid system needs to achieve power balance, and the energy storage capacity of the energy storage system needs to be between the minimum and maximum values. Here, the minimum value is defined as SOC=0, and the maximum value is defined as SOC=100. Based on this, the above two constraints are constructed.

[0068] In a preferred embodiment of the present invention, step A1 also includes removing abnormal data from the acquired historical photovoltaic installed capacity, historical energy storage installed capacity and the corresponding historical collection time, historical weather data and historical electricity price, and forming a historical feature vector based on the historical photovoltaic installed capacity, historical energy storage installed capacity and the corresponding historical collection time, historical weather data and historical electricity price after the abnormal data removal.

[0069] In a preferred embodiment of the present invention, in step A1, a clustering algorithm is used to screen abnormal data, and the screened abnormal data is eliminated.

[0070] Specifically, in this embodiment, the clustering process of the clustering algorithm is as follows:

[0071] 1) Randomly select k objects, each of which initially represents the center of a cluster;

[0072] 2) For each remaining object, assign it to the nearest cluster based on its distance from the center of each cluster;

[0073] 3) Recalculate the average value of each cluster and update it as the new cluster center;

[0074] 4) Repeat 2 and 3 until the criterion function converges.

[0075] Each of the above data is processed using the above clustering process to filter out abnormal data, thereby ensuring the correctness of the training data and further improving the accuracy of the trained intelligent scheduling model.

[0076] In a preferred embodiment of the present invention, Figure 3As shown, step S2 includes:

[0077] Step S21, based on the current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the rest of the loads in the park, respectively predict the predicted photovoltaic power generation power, predicted charging power and predicted load power corresponding to the future time period after the current collection time;

[0078] Step S22, respectively obtaining the current total number of vehicles and the current total number of people in the park, and correcting the predicted charging power according to the current total number of vehicles to obtain the corresponding corrected charging power, and correcting the predicted load power according to the current total number of people to obtain the corresponding corrected load power;

[0079] Step S23, forming a second eigenvector according to the predicted photovoltaic power generation power, the corrected charging power and the corrected load power.

[0080] In a preferred embodiment of the present invention, in step S21, the current photovoltaic power generation sequence, the current charging power sequence and the current load power sequence are respectively input into a first time series model, a second time series model and a third time series model obtained in advance, so as to respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to the future time period after the current collection time.

[0081] Specifically, in this embodiment, the most similar daily data of the closest 60 days can be selected from the historical data according to the date type, weather, and season as input data, and the photovoltaic power generation power can be predicted using the first time series model. Preferably, the photovoltaic power generation power prediction can be realized by predicting the power generation data of the next day on the previous day. The charging power of the charging pile can be predicted using the second time series model based on the charging power historical data of the charging pile and the date type data, and preferably, the power prediction of the next day is completed after the previous day. The load power can be predicted using the third time series model based on the load historical data of the park and the date type data, and preferably, the power prediction of the next day is completed after the previous day.

[0082] Specifically, on the day of the energy storage power forecast, due to weather, date and other reasons, there may be too many or too few vehicles or people entering and leaving the park on that day. In order to ensure the accuracy of the data, it is preferred to correct the predicted charging power according to the total number of vehicles in the park and the predicted load power according to the current total number of people on the day of the energy storage power forecast. Specifically, the current total number of vehicles can be compared with the total number of vehicles on the day with the highest similarity in the historical data. Preferably, the predicted charging power can be corrected according to the ratio of the current total number of vehicles to the total number of vehicles on the day with the highest similarity. The method of correcting the predicted load power according to the current total number of people is similar and so on, which will not be repeated here.

[0083] Preferably, a real-time personnel and vehicle monitoring and data collection device may be configured in the park to obtain the current total number of vehicles and the current total number of people in the park.

[0084] The present invention also provides an intelligent dispatching system for a photovoltaic storage and charging microgrid system, which applies the above-mentioned intelligent dispatching method, such as Figure 4 As shown, the intelligent scheduling system includes:

[0085] Data acquisition module 1, used to respectively acquire the current photovoltaic installed capacity of a photovoltaic system of a park, the current energy storage installed capacity of an energy storage system, and the corresponding current collection time, current weather data, and current electricity price, and form a first feature vector;

[0086] The power prediction module 2 is used to predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to a future period after the current collection time according to the current photovoltaic power generation power sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the other loads of the park, and form a second feature vector;

[0087] The scheduling prediction module 3 is connected to the data acquisition module 1 and the power prediction module 2 respectively, and is used to input the first eigenvector and the second eigenvector into an intelligent scheduling model trained with the goal of maximizing the electricity consumption benefit of the park and a corresponding constraint condition, so as to predict the predicted energy storage power of the energy storage system in the future time period after the current collection time, so as to provide intelligent scheduling for the photovoltaic storage and charging microgrid system.

[0088] In a preferred embodiment of the present invention, the power prediction module 2 includes:

[0089] The prediction unit 21 is used to predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to the future time period after the current collection time according to the current photovoltaic power generation power sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the remaining loads of the park;

[0090] The result correction unit 22 is connected to the prediction unit 21, and is used to respectively obtain the current total number of vehicles and the current total number of people in the park, and correct the predicted charging power according to the current total number of vehicles to obtain the corresponding corrected charging power, and correct the predicted load power according to the current total number of people to obtain the corresponding corrected load power;

[0091] The vector generating unit 23 is connected to the prediction unit 21 and the result correcting unit 22 respectively, and is used to form a second feature vector according to the predicted photovoltaic power generation power, the corrected charging power and the corrected load power.

[0092] The above description is only a preferred embodiment of the present invention, and does not limit the implementation mode and protection scope of the present invention. For those skilled in the art, it should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the contents of this specification and illustrations should be included in the protection scope of the present invention.

Claims

1. An intelligent scheduling method for a photovoltaic storage and charging microgrid system, wherein the photovoltaic storage and charging microgrid system includes a photovoltaic system, an energy storage system and a charging pile system, characterized in that: include: Step S1, respectively obtaining the current photovoltaic installed capacity of the photovoltaic system of a park, the current energy storage installed capacity of the energy storage system, and the corresponding current collection time, current weather data and current electricity price to form a first feature vector; Step S2, based on the current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the remaining loads of the park, respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to a future period after the current collection time and form a second feature vector; The step S2 comprises: Step S21, based on the acquired current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system, and the current load power sequence of the remaining loads of the park, respectively predict the predicted photovoltaic power generation power, the predicted charging power, and the predicted load power corresponding to the future time period after the current collection time; In the step S21, the current photovoltaic power generation sequence, the current charging power sequence and the current load power sequence are respectively input into a first time series model, a second time series model and a third time series model obtained by pre-training, so as to respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to the future time period after the current acquisition time; Step S22, respectively obtaining the current total number of vehicles and the current total number of people in the park, and correcting the predicted charging power according to the current total number of vehicles to obtain a corresponding corrected charging power, and correcting the predicted load power according to the current total number of people to obtain a corresponding corrected load power; Step S23, forming the second characteristic vector according to the predicted photovoltaic power generation power, the corrected charging power and the corrected load power; Step S3, inputting the first feature vector and the second feature vector into an intelligent scheduling model trained with the goal of maximizing the electricity revenue of the park and a corresponding constraint condition, so as to predict the predicted energy storage power of the energy storage system in the future period after the current collection time, so as to provide intelligent scheduling for the photovoltaic storage and charging microgrid system; Before executing step S1, a model training process is also included, including: Step A1, respectively obtaining the historical photovoltaic installed capacity of the photovoltaic system, the historical energy storage installed capacity of the energy storage system, and the corresponding historical collection time, historical weather data, and historical electricity prices, and forming a historical feature vector; the historical collection time includes the data collection time of the historical photovoltaic installed capacity and the historical energy storage installed capacity, and the date type and season to which the data collection time belongs; Step A2, according to the acquired historical photovoltaic power generation sequence of the photovoltaic system, the historical charging power sequence of the charging pile system and the historical load power sequence of the remaining loads, respectively predict the corresponding historical predicted photovoltaic power generation power, historical predicted charging power and historical predicted load power; Step A3, taking maximizing the electricity revenue of the park as the goal and under the corresponding constraints, taking the historical feature vector, the historical predicted photovoltaic power generation power, the historical predicted charging power and the historical predicted load power as input, and taking the energy storage capacity of the energy storage system as output, to train and obtain the intelligent scheduling model; The objective function corresponding to the goal of maximizing the electricity revenue of the park is: Y=p*(Q PY +Q ES -Q EC -Q EV -Q load -Q grid ) The constraints are: P PV +P ES -P EV -P load =P grid 0 <SOC<100 Among them, Y represents the electricity income, p represents the electricity price, Q PY represents the photovoltaic power generation of the photovoltaic system, Q ES Indicates the energy storage and discharge capacity of the energy storage system, Q EC represents the energy storage charging capacity of the energy storage system, Q EV Indicates the charging capacity of the charging pile, Q load Indicates the power consumption of the remaining loads, Q grid Represents the power grid electricity of the park; P PY represents the photovoltaic power generation power of the photovoltaic system, P ES represents the energy storage power generation of the energy storage system, P EV represents the charging power of electric vehicles in the photovoltaic storage and charging microgrid system, P grid It represents the power generation power of the power grid of the park, and SOC represents the energy storage capacity of the energy storage system.

2. The intelligent scheduling method according to claim 1, characterized in that: The step A1 also includes removing abnormal data from the acquired historical photovoltaic installed capacity, the historical energy storage installed capacity and the corresponding historical collection time, the historical weather data and the historical electricity price, and forming the historical feature vector based on the historical photovoltaic installed capacity, the historical energy storage installed capacity and the corresponding historical collection time, the historical weather data and the historical electricity price after the abnormal data removal process.

3. The intelligent scheduling method according to claim 2, characterized in that: In the step A1, a clustering algorithm is used to screen abnormal data, and the screened abnormal data is removed.

4. An intelligent dispatching system for a photovoltaic storage and charging microgrid system, characterized in that: The intelligent scheduling method according to any one of claims 1 to 3 is applied, wherein the intelligent scheduling system comprises: A data acquisition module, used to respectively acquire the current photovoltaic installed capacity of the photovoltaic system of a park, the current energy storage installed capacity of the energy storage system, and the corresponding current collection time, current weather data and current electricity price to form a first feature vector; A power prediction module, configured to respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to a future period after the current collection time according to the acquired current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the remaining loads of the park, and form a second feature vector; The scheduling prediction module is connected to the data acquisition module and the power prediction module respectively, and is used to input the first feature vector and the second feature vector into an intelligent scheduling model trained with the goal of maximizing the electricity consumption benefit of the park and a corresponding constraint condition, so as to predict the predicted energy storage power of the energy storage system in the future time period after the current collection time, so as to provide intelligent scheduling for the photovoltaic storage and charging microgrid system.

5. The intelligent dispatching system according to claim 4, characterized in that: The power prediction module comprises: A prediction unit, configured to respectively predict the predicted photovoltaic power generation power, the predicted charging power and the predicted load power corresponding to the future time period after the current collection time according to the acquired current photovoltaic power generation sequence of the photovoltaic system of the park, the current charging power sequence of the charging pile system and the current load power sequence of the remaining loads of the park; a result correction unit, connected to the prediction unit, for respectively obtaining the current total number of vehicles and the current total number of people in the park, and correcting the predicted charging power according to the current total number of vehicles to obtain a corresponding corrected charging power, and correcting the predicted load power according to the current total number of people to obtain a corresponding corrected load power; A vector generating unit is connected to the prediction unit and the result correcting unit respectively, and is used to form the second feature vector according to the predicted photovoltaic power generation power, the corrected charging power and the corrected load power.

Images