CN117498400A - Distributed photovoltaic and energy storage data processing method and system - Google Patents
Distributed photovoltaic and energy storage data processing method and system Download PDFInfo
- Publication number
- CN117498400A CN117498400A CN202410004568.4A CN202410004568A CN117498400A CN 117498400 A CN117498400 A CN 117498400A CN 202410004568 A CN202410004568 A CN 202410004568A CN 117498400 A CN117498400 A CN 117498400A
- Authority
- CN
- China
- Prior art keywords
- data
- energy storage
- photovoltaic
- historical
- layer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000004146 energy storage Methods 0.000 title claims abstract description 156
- 238000003672 processing method Methods 0.000 title claims abstract description 7
- 238000009826 distribution Methods 0.000 claims abstract description 73
- 238000012549 training Methods 0.000 claims abstract description 71
- 238000012545 processing Methods 0.000 claims abstract description 65
- 238000005457 optimization Methods 0.000 claims abstract description 18
- 230000006870 function Effects 0.000 claims description 77
- 238000000034 method Methods 0.000 claims description 41
- 210000002569 neuron Anatomy 0.000 claims description 37
- 239000002245 particle Substances 0.000 claims description 20
- 210000005036 nerve Anatomy 0.000 claims description 17
- 238000012360 testing method Methods 0.000 claims description 17
- 230000008569 process Effects 0.000 claims description 13
- 230000004927 fusion Effects 0.000 claims description 9
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000004140 cleaning Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 238000010845 search algorithm Methods 0.000 claims description 4
- 238000009499 grossing Methods 0.000 claims description 3
- 238000003860 storage Methods 0.000 claims description 3
- 239000002699 waste material Substances 0.000 abstract description 2
- 238000010248 power generation Methods 0.000 description 28
- 239000013598 vector Substances 0.000 description 7
- 230000003213 activating effect Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000013500 data storage Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000002354 daily effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000007599 discharging Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000017525 heat dissipation Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/28—Arrangements for balancing of the load in a network by storage of energy
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2155—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06313—Resource planning in a project environment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
- H02J3/381—Dispersed generators
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/10—Power transmission or distribution systems management focussing at grid-level, e.g. load flow analysis, node profile computation, meshed network optimisation, active network management or spinning reserve management
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
- H02J2300/20—The dispersed energy generation being of renewable origin
- H02J2300/22—The renewable source being solar energy
- H02J2300/24—The renewable source being solar energy of photovoltaic origin
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Human Resources & Organizations (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Tourism & Hospitality (AREA)
- Power Engineering (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Probability & Statistics with Applications (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Educational Administration (AREA)
Abstract
The invention provides a processing method and a system for distributed photovoltaic and energy storage data, which relate to the technical field of photovoltaic data processing and comprise the following steps: acquiring historical photovoltaic data and historical meteorological data; establishing a prediction model, taking the processed historical photovoltaic data and the historical meteorological data as training data, and performing iterative training on the prediction model to obtain a trained prediction model; acquiring weather forecast data, and inputting the processed weather forecast data into a trained prediction model to obtain predicted photovoltaic data; acquiring energy storage system data, forming an initial energy storage distribution scheme set according to the predicted photovoltaic data and the energy storage system data, and optimizing the energy storage distribution scheme set by utilizing an optimization algorithm to obtain an optimal energy storage distribution scheme; and carrying out energy storage distribution on each distributed photovoltaic power station according to the optimal energy storage distribution scheme. The invention can improve the energy utilization efficiency of the photovoltaic and energy storage system and reduce the energy waste of the energy storage system, thereby reducing the energy cost.
Description
Technical Field
The invention relates to the technical field of photovoltaic data processing, in particular to a processing method and a processing system of distributed photovoltaic and energy storage data.
Background
The distributed photovoltaic and energy storage system is characterized in that the photovoltaic power generation system and the energy storage system are combined to realize energy storage and distribution control of the photovoltaic power generation system, so that reliability, economy and sustainability of the photovoltaic power generation system are improved. With the rapid development of renewable energy sources, the application of the distributed photovoltaic and energy storage system in the energy field is more and more widespread, and the demands for data processing methods and systems thereof are also increasing.
In a distributed photovoltaic and energy storage system, the photovoltaic power generation system converts solar energy into electrical energy, and the energy storage system can store excess electrical energy for release when needed. However, since the photovoltaic power generation system is affected by factors such as weather, the power generation amount of the photovoltaic power generation system has certain uncertainty, the photovoltaic power generation amount needs to be predicted so as to reasonably arrange the charge and discharge strategies of the energy storage system, and therefore the efficient utilization of the electric energy is achieved.
The invention patent with the Chinese application number of 202211384339.7 discloses an energy storage optimization method and electronic equipment of a distributed photovoltaic and energy storage system, which are combined with various contents such as the output power characteristics of the distributed photovoltaic, the service life and capacity characteristics of the energy storage system, cost factors and the like, and aims at the minimum cost, optimize the relation between daily energy storage margin and relation curves of output power and load demand, and optimize the current daily charging margin to obtain more accurate current-day energy storage energy and grid-connected power generation capacity of the energy storage system, so that the total output power of the distributed photovoltaic and energy storage system meets the stability index. The prior art mainly focuses on the cost problem to optimize the distribution of the energy storage system on the same day, does not consider the influence of distributed photovoltaic and meteorological factors on photovoltaic power generation, has less calculated amount and convenient execution, but the optimization result is not optimal, the precision is not high, and the high-efficiency utilization of electric energy cannot be realized.
Disclosure of Invention
In view of the above, the present invention provides a method and a system for processing distributed photovoltaic and energy storage data, which predicts photovoltaic data in a period of time in the future by using historical photovoltaic data, historical meteorological data and weather forecast data, and optimizes and obtains an optimal energy storage allocation scheme based on the predicted photovoltaic data and energy storage system data, so as to improve the energy utilization efficiency of the photovoltaic and energy storage system, reduce the energy waste of the energy storage system, and thus reduce the energy cost.
The technical purpose of the invention is realized as follows:
in one aspect, the invention provides a method for processing distributed photovoltaic and energy storage data, comprising the following steps:
s1, acquiring historical photovoltaic data and historical meteorological data, preprocessing the historical photovoltaic data, and carrying out interpolation processing on the historical meteorological data to obtain processed historical photovoltaic data and historical meteorological data;
s2, establishing a prediction model, taking the processed historical photovoltaic data and the historical meteorological data as training data, and performing iterative training on the prediction model to obtain a trained prediction model;
s3, acquiring weather forecast data, carrying out interpolation processing on the weather forecast data, and inputting the processed weather forecast data into a trained prediction model to obtain predicted photovoltaic data;
S4, collecting energy storage system data, forming an initial energy storage distribution scheme set according to the predicted photovoltaic data and the energy storage system data, and optimizing the energy storage distribution scheme set by utilizing an optimization algorithm to obtain an optimal energy storage distribution scheme;
and S5, carrying out energy storage distribution on each distributed photovoltaic power station according to an optimal energy storage distribution scheme.
On the basis of the above scheme, preferably, step S2 includes the steps of:
s21, clustering historical meteorological data, wherein the number of the clusters is k, and k historical meteorological sets are obtained;
s22, marking the historical meteorological set as a cluster type to which the historical meteorological set belongs, forming k types of historical meteorological data columns according to k types, aligning the historical photovoltaic data with the corresponding historical meteorological data columns according to the time stamp, and adding the historical meteorological data columns into the historical photovoltaic data to form a dataset;
s23, constructing a prediction model, and performing iterative training on the prediction model by using the data set to obtain a trained prediction model.
On the basis of the above-described scheme, preferably, step S21 includes:
s211, constructing a clustering network, and determining the number of neurons of an input layer and the number of neurons of an output layer of the clustering network, wherein the number of neurons of the output layer is k;
S212 sets the iteration number N, and randomly initializes the weights { ω ] of the neurons in the clustered network when the initial step n=0 1 ,ω 2 ,...,ω n };
S213 from historical Meteorological data { ζ 1 ,ξ 2 ,...,ξ M Random extraction of a data xi in } 1 ;
S214 calculating the difference;
S215 selects the weight ω with the smallest difference v (0) The corresponding neuron v is taken as an excellent neuron;
s216, repeating the steps S213-S215 to obtain excellent neurons of all historical meteorological data;
s217 updates only the weights of the excellent neurons:
;
wherein η represents a learning parameter;
s218 let n=n+1, and go to step S213;
s219 repeats steps S213-S218 until n=n, and the iteration ends, where each output layer neuron represents a cluster, and k clusters, that is, k historical meteorological sets, are obtained.
On the basis of the above-described scheme, preferably, step S23 includes:
s231, dividing the data set into a training set and a testing set according to a ratio of 7:3;
s232, respectively inputting the training sets into the prediction models for iterative training until reaching the iterative stopping condition to obtain pre-trained prediction models;
s233, predicting the test set by using a pre-trained prediction model, and sorting the prediction result:
if the data which does not meet the prediction requirement exists, the data is used as error data, the error data is replaced from the test set to the training set to update the training set, and the step S232 is carried out;
And if the test sets all meet the prediction requirements, ending the whole training to obtain a trained prediction model.
On the basis of the above-mentioned scheme, preferably, step S232 includes:
the prediction model comprises a first input layer, a second input layer, a first full-connection layer, a second full-connection layer, a splicing layer, a hiding layer and an output layer;
inputting historical photovoltaic data in a training set into a first input layer, and processing the historical photovoltaic data through a first full-connection layer to obtain photovoltaic sign X1;
inputting a historical meteorological data column in the training set into a second input layer, and processing the historical meteorological data column by a second full-connection layer to obtain meteorological features X2;
inputting the photovoltaic characteristic X1 and the meteorological characteristic X2 into a splicing layer for characteristic splicing to obtain a fusion characteristic X3;
inputting the fusion characteristic X3 into a hidden layer, wherein the hidden layer comprises m nerve unit layers which are sequentially overlapped, and carrying out layer-by-layer weighted summation on the fusion characteristic X3 from bottom to top by utilizing the m nerve unit layers to obtain a hidden characteristic X4, wherein the weight of the m-1 nerve unit layer is connected with the weight of the m nerve unit layer;
inputting the hidden characteristic X4 into an output layer, wherein the output layer is a two-layer full-connection layer, and predicting the hidden characteristic X4 by using the output layer to obtain a predicted value;
Calculating a loss function according to the predicted value and the true value of the training set:
if the loss function converges, ending the iteration;
if the loss function is not converged, the error value between the predicted value and the true value is transmitted back to the first input layer by layer through a back propagation algorithm, the network parameters of each layer are adjusted in the back propagation process, and the next iteration is carried out.
Based on the above scheme, preferably, the loss function is:
;
;
in the method, in the process of the invention,is a loss function, y represents a true value, < ->Represents the predicted value, M is the number of training sets, j represents the jth training data,y j Is the true value of the j-th training data, for example>Is the predicted value of the jth training data, lambda is the regularization coefficient, D is the number of network parameters, D represents the jth network parameter, w d Weight representing the d-th network parameter, < ->Is a smooth function +.>Is a superparameter->For controlling the degree of smoothness of the smoothing function.
On the basis of the above scheme, preferably, in step S3, interpolation processing is performed on weather forecast data, including:
step one, carrying out discrete gridding processing on weather forecast data according to the longitude and latitude of the weather forecast data, wherein each grid represents a longitude and latitude range, taking the weather forecast data in the grid as known data points, and determining the number and the positions of interpolation points according to the distribution of the known data points in the grid;
Step two, selecting a Gaussian function to describe the spatial correlation between data points;
step three, calculating the space distance between any two known data points for all the known data points, calculating the variance according to the space distance, and substituting the space distance and the variance into a Gaussian function for fitting to obtain a fitting function;
step four, determining an interpolation range from the fitting function, and estimating the length scale of the Gaussian function;
step five, for a single interpolation point, calculating the spatial distance between the interpolation point and all known data points in the interpolation range according to the position of the interpolation point, substituting the spatial distance into a Gaussian function to calculate the interpolation weight of each known data point to the interpolation point;
step six, obtaining the value of the interpolation point by using the value of the known data point and the interpolation weight through a weighted average formula;
and step seven, repeating the step five to the step six to obtain the numerical values of all interpolation points.
Based on the above scheme, preferably, the gaussian function is:
;
wherein γ (h) is a Gaussian function, σ 2 Is variance, s is a length scale, h represents a spatial distance between two points;
the weighted average formula is:
;
wherein z' (x) 0 ) Representing interpolation point x 0 Numerical value at A is the number of known data points in the interpolation range, x a Represents the a-th known data point, z (x a ) Represents x a Numerical value of beta a Represents x a For x 0 Is used for the interpolation weight of the (a);
the calculation formula of the interpolation weight is as follows:
;
wherein h (x 0 ,x a ) Represents x a And x 0 Is a spatial distance of (c).
On the basis of the above scheme, preferably, step S4 includes the steps of:
s41, collecting data of an energy storage system, wherein the data comprise charge and discharge efficiency and energy storage capacity of the energy storage system;
s42, taking the predicted photovoltaic data and the energy storage system data as decision variables, searching all paths by using a search algorithm, wherein each path is an initial energy storage distribution scheme, and obtaining an initial energy storage distribution scheme set;
s43, carrying out iterative optimization on an initial energy storage distribution scheme set by adopting a particle swarm optimization algorithm, taking a minimized voltage offset index and photovoltaic active loss as objective functions, searching an optimal solution by iteratively updating the particle swarm, and outputting the minimum values of the optimal solution and the objective functions when the iteration termination condition is met, wherein the optimal solution is the optimal energy storage distribution scheme;
wherein, the objective function is:
;
wherein T represents a target time range, T represents a time period within the target time range, c U Penalty coefficient for voltage deviation, c PV Penalty coefficient for photovoltaic active loss, U r,t For the actual voltage value of the r-th distributed photovoltaic in the t time period, U r,R For the voltage rating of the r-th distributed photovoltaic in the t time period, G dec,r,t And the photovoltaic active loss of the r-th distributed photovoltaic in the t time period is represented.
In another aspect, the present invention further provides a distributed photovoltaic and energy storage data processing system, where the system is configured to perform the method of any one of the above claims, and the system includes:
the historical data acquisition module is configured to acquire historical photovoltaic data and historical meteorological data according to preset time;
the photovoltaic data processing module is configured to preprocess historical photovoltaic data, and comprises data cleaning, missing value processing and abnormal value processing, so that processed historical photovoltaic data is obtained;
the weather forecast data acquisition module is configured to acquire corresponding weather forecast data according to the prediction time;
the weather data processing module is configured to conduct interpolation processing on weather data, wherein the weather data comprises historical weather data and weather forecast data, and fills the weather data;
the model storage module is configured to establish a prediction model, and perform iterative training on the prediction model by using the processed historical photovoltaic data and the historical meteorological data to obtain and store the trained prediction model;
The photovoltaic prediction module is configured to input the weather forecast data after interpolation processing into a prediction model after training to obtain prediction photovoltaic data;
the energy storage system management module is configured to acquire energy storage system data according to requirements and perform preliminary pretreatment on the energy storage system data;
the energy storage scheme optimizing module is configured to form an initial energy storage distribution scheme set according to the predicted photovoltaic data and the energy storage system data, and an internal optimizing algorithm is utilized to optimize the energy storage distribution scheme set so as to obtain an optimal energy storage distribution scheme;
and the energy storage distribution execution module is configured to distribute the energy storage of each distributed photovoltaic power station according to an optimal energy storage distribution scheme.
Compared with the prior art, the method has the following beneficial effects:
(1) According to the invention, through preprocessing and interpolation processing of historical photovoltaic data and historical meteorological data, the accuracy and completeness of the data can be improved, the built prediction model can be used for carrying out iterative training by using the processed historical photovoltaic data and meteorological data, so that a more accurate photovoltaic power generation amount prediction result is obtained, an energy storage distribution scheme is optimized by using an optimization algorithm, an optimal energy storage distribution scheme can be found on the basis of considering photovoltaic power generation amount prediction and energy storage system data, the energy utilization efficiency of a photovoltaic and energy storage system is improved, the system can have real-time response capability by acquiring the meteorological prediction data and the real-time acquisition of the energy storage system data, the energy storage distribution scheme is timely adjusted to cope with meteorological changes and photovoltaic power generation amount fluctuation, and the stability and reliability of the system are improved;
(2) According to the method, the clustering network is built, the historical meteorological data can be automatically clustered, the number of the clustering clusters is not required to be set manually, so that influence of subjective factors on clustering results is avoided, the objectivity and accuracy of the clustering are improved, the clustering network can be continuously optimized by repeatedly updating the weights of neurons in an iterative training mode, the clustering results are more accurate, potential meteorological data features and rules can be found, in each iteration, excellent neurons are selected according to the difference value of the historical meteorological data, the weights of the excellent neurons are updated, the clustering network can perform self-adaptive learning according to the features of the data, the flexibility and the adaptability of the clustering are improved, and the method has better applicability to different types of historical meteorological data and can be flexibly applied to clustering analysis of different meteorological data because the method is based on unsupervised learning of the neural network;
(3) When the prediction model is trained, the method comprises a plurality of layers of feature processing, so that the feature information of the historical photovoltaic data and the historical meteorological data can be fully mined, the data representation capacity and the prediction precision of the model are improved, error values are transmitted back layer by using a back propagation algorithm, network parameters of each layer are adjusted, a better local optimal or even global optimal area can be achieved, weights are connected together on the two highest nerve unit layers of the hidden layer, and thus, the output of the lower layer can be associated with the top layer, and the robustness and the convergence speed of the model can be increased;
(4) According to the interpolation processing method provided by the invention, the spatial correlation among the data points is considered, the Gaussian function is fitted, the change rule of meteorological data in space can be better captured, the spatial consistency and accuracy of interpolation results are improved, in addition, according to the position of an interpolation point and the spatial distance between the interpolation point and all known data points in an interpolation range, the interpolation weight of each known data point on the interpolation point is calculated, the influence degree of different known data points on the interpolation point can be more reasonably considered, the accuracy and the interpretability of the interpolation results are improved, finally, the numerical value of the interpolation point is obtained through a weighted average formula, the numerical value and the interpolation weight of the known data points can be effectively utilized, and more accurate and reliable interpolation results are obtained;
(5) The processing system provided by the invention can integrate historical photovoltaic data, weather forecast data and energy storage system data, and can accurately predict photovoltaic power generation through a prediction model after pretreatment, interpolation processing and training, and meanwhile, the energy storage distribution scheme is optimized through the energy storage scheme optimizing module, so that the self-digestion capacity of the photovoltaic power generation system can be improved to the greatest extent, the running efficiency and the reliability of the system are further improved, and the application of the system is beneficial to optimizing the coordinated running of the photovoltaic power generation system and the energy storage system, improving the utilization efficiency of renewable energy sources and reducing the energy consumption cost, thereby having a positive pushing effect on the sustainable development of the energy system.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a predictive model according to an embodiment of the invention;
FIG. 3 is a system frame diagram of an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will clearly and fully describe the technical aspects of the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, are intended to fall within the scope of the present invention.
As shown in fig. 1, the invention provides a method for processing distributed photovoltaic and energy storage data, which comprises the following steps:
S1, acquiring historical photovoltaic data and historical meteorological data, preprocessing the historical photovoltaic data, and carrying out interpolation processing on the historical meteorological data to obtain processed historical photovoltaic data and historical meteorological data;
s2, establishing a prediction model, taking the processed historical photovoltaic data and the historical meteorological data as training data, and performing iterative training on the prediction model to obtain a trained prediction model;
s3, acquiring weather forecast data, carrying out interpolation processing on the weather forecast data, and inputting the processed weather forecast data into a trained prediction model to obtain predicted photovoltaic data;
s4, collecting energy storage system data, forming an initial energy storage distribution scheme set according to the predicted photovoltaic data and the energy storage system data, and optimizing the energy storage distribution scheme set by utilizing an optimization algorithm to obtain an optimal energy storage distribution scheme;
and S5, carrying out energy storage distribution on each distributed photovoltaic power station according to an optimal energy storage distribution scheme.
Specifically, in an embodiment of the present invention, step S1 includes:
acquiring historical photovoltaic data and historical meteorological data: historical photovoltaic data can be obtained through a monitoring system or data acquisition equipment of a photovoltaic power station, and the historical photovoltaic data comprises information such as the generated energy, the working temperature and irradiance of a photovoltaic module. Historical meteorological data can be obtained through a meteorological station, a meteorological satellite or a meteorological forecasting mechanism, and the meteorological data comprise meteorological parameters such as air temperature, wind speed, humidity, sunshine hours and the like.
Preprocessing historical photovoltaic data: data cleaning: and removing repeated data, abnormal data and error data, and ensuring the accuracy and the integrity of the data. Missing value processing: for data points with missing values, the missing values can be filled in by interpolation methods or estimated and filled in according to related data. Outlier processing: abnormal values, such as data points outside of normal ranges, are identified and processed to ensure the rationality and reliability of the data.
Interpolation processing is carried out on the historical meteorological data: the method of this interpolation processing is the same as in step S3, and a specific interpolation process is described in detail below.
Specifically, in an embodiment of the present invention, step S2 includes:
s21, clustering historical meteorological data, wherein the number of the clusters is k, and k historical meteorological sets are obtained;
s22, marking the historical meteorological set as a cluster type to which the historical meteorological set belongs, forming k types of historical meteorological data columns according to k types, aligning the historical photovoltaic data with the corresponding historical meteorological data columns according to the time stamp, and adding the historical meteorological data columns into the historical photovoltaic data to form a dataset;
s23, constructing a prediction model, and performing iterative training on the prediction model by using the data set to obtain a trained prediction model.
In this embodiment, step S21 includes:
s211, constructing a clustering network, and determining the number of neurons of an input layer and the number of neurons of an output layer of the clustering network, wherein the number of neurons of the output layer is k.
S212 sets the iteration number N, and randomly initializes the weights { ω ] of the neurons in the clustered network when the initial step n=0 1 ,ω 2 ,...,ω n }。
S213 from historical Meteorological data { ζ 1 ,ξ 2 ,...,ξ M Random extraction of a data xi in } 1 ;
S214 calculating the differenceThe method comprises the steps of carrying out a first treatment on the surface of the For selected historical meteorological data xi 1 The difference between this and the weights of each neuron in the clustered network is calculated to determine the most suitable neuron.
S215 selects the weight ω with the smallest difference v (0) The corresponding neuron v is taken as an excellent neuron;
s216, repeating the steps S213-S215 to obtain excellent neurons of all historical meteorological data;
s217 updates only the weights of the excellent neurons:
;
wherein η represents a learning parameter;
s218 let n=n+1, and go to step S213;
s219 repeats steps S213-S218 until n=n, and the iteration ends, where each output layer neuron represents a cluster, and k clusters, that is, k historical meteorological sets, are obtained.
In this embodiment, a self-organizing map network is used as a clustering network, and the input layer neurons and the output layer neurons are organized into a simple neighborhood structure. Each neuron is associated with one reference vector (weight vector), and each data point is "mapped" to a neuron with the "nearest" reference vector. During the operation of the algorithm, each data point serves as a training sample, guiding the reference vector to move toward the value of the sample data. The vectors associated with neurons, called weights, change during learning and tend to input eigenvalues of the data distribution.
In this embodiment, in order to converge as soon as possible and prevent over-fitting, the value of N is taken to be 10. And k is 4 corresponding to different weather types, and the k corresponds to rainy days, cloudy days and sunny days respectively.
In this embodiment, step S22 includes:
marking cluster categories of historical meteorological data sets: each historical meteorological data point is assigned to a cluster category based on the results of the previous cluster network. Each historical meteorological data point is marked as a clustering category to which the historical meteorological data point belongs, so that a category label of each data point is obtained.
Forming k types of historical meteorological data columns: and dividing the historical meteorological data set into k classes according to the number k of the clusters, wherein each class corresponds to one cluster class. Then, the historical meteorological data in each category is arranged into a row to form k types of historical meteorological data rows.
Aligning the historical photovoltaic data with the historical meteorological data columns: the time stamps of the historical photovoltaic data are aligned with the time stamps of the historical meteorological data.
Adding the historical meteorological data column to historical photovoltaic data: and adding the formed k types of historical meteorological data columns into the historical photovoltaic data, and aligning according to the time stamp, so that each historical photovoltaic data point corresponds to the data in the k types of historical meteorological data columns. This forms a complete data set in which each historical photovoltaic data point contains k types of historical meteorological data corresponding thereto.
In this embodiment, step S23 includes:
s231, dividing the data set into a training set and a testing set according to a ratio of 7:3;
s232, respectively inputting the training sets into the prediction models for iterative training until reaching the iterative stopping condition to obtain pre-trained prediction models;
specifically, step S232 includes:
the prediction model comprises a first input layer, a second input layer, a first full-connection layer, a second full-connection layer, a splicing layer, a hiding layer and an output layer;
inputting historical photovoltaic data in a training set into a first input layer, and processing the historical photovoltaic data through a first full-connection layer to obtain photovoltaic sign X1;
inputting a historical meteorological data column in the training set into a second input layer, and processing the historical meteorological data column by a second full-connection layer to obtain meteorological features X2;
inputting the photovoltaic characteristic X1 and the meteorological characteristic X2 into a splicing layer for characteristic splicing to obtain a fusion characteristic X3;
inputting the fusion characteristic X3 into a hidden layer, wherein the hidden layer comprises m nerve unit layers which are sequentially overlapped, and carrying out layer-by-layer weighted summation on the fusion characteristic X3 from bottom to top by utilizing the m nerve unit layers to obtain a hidden characteristic X4, wherein the weight of the m-1 nerve unit layer is connected with the weight of the m nerve unit layer;
inputting the hidden characteristic X4 into an output layer, wherein the output layer is a two-layer full-connection layer, and predicting the hidden characteristic X4 by using the output layer to obtain a predicted value;
Calculating a loss function according to the predicted value and the true value of the training set:
if the loss function converges, ending the iteration;
if the loss function is not converged, the error value between the predicted value and the true value is transmitted back to the first input layer by layer through a back propagation algorithm, the network parameters of each layer are adjusted in the back propagation process, and the next iteration is carried out.
S233, predicting the test set by using a pre-trained prediction model, and sorting the prediction result:
if the data which does not meet the prediction requirement exists, the data is used as error data, the error data is replaced from the test set to the training set to update the training set, and the step S232 is carried out;
and if the test sets all meet the prediction requirements, ending the whole training to obtain a trained prediction model.
Referring to fig. 2, step S23 is described in an embodiment:
the whole data set is divided into a training set and a test set according to a ratio of 7:3. The training set is used to train parameters of the model, while the test set is used to evaluate the performance and accuracy of the model.
And using historical photovoltaic data and meteorological data in the training set as input, and obtaining a pre-trained prediction model through a series of processing and calculation. Specifically, a neural network model is used that includes a first input layer, a second input layer, a first fully-connected layer, a second fully-connected layer, a splice layer, a hidden layer, and an output layer, each network layer including an activation function that employs a ReLU function. The model receives historical photovoltaic data and meteorological data as input, and finally obtains a predicted value through multi-layer processing and feature extraction.
Specifically, the first input layer and the second input layer respectively perform reshape operation on historical photovoltaic data and historical meteorological data, the data are adjusted to be the size of suitable model identification, then the photovoltaic characteristic X1 and the meteorological characteristic X2 are extracted through the first fully-connected layer, the second fully-connected layer and corresponding activating functions, then the splicing layer fuses the X1 and the X2 through the conclusive operation to obtain a fused characteristic X3, the hidden layer is a structure comprising m nerve unit layers, each nerve unit layer is provided with activating functions, after the fused characteristic X3 is input into the hidden layer, weighting summation is performed from bottom to top, the activating functions are applied, then the result is transmitted to the next layer, the output of each nerve unit layer is the result after the activating functions are processed, the nonlinear transformation of the input data can be understood, the output can be used as the input of the next nerve unit, the output is sequentially transmitted to the top layer, and each layer network parameter of the hidden layer can be initialized through an unsupervised learning method, so that better local optimal solution, even global optimal solution can be found in the training process. By pre-training layer by layer greedy, network parameters may better capture the distribution characteristics of the input data. At the highest two levels of the hidden layer, the weights are connected together, which means that the output of the lower layer can be associated with the top layer. This way of connecting helps to combine low-level features with high-level features, thereby better expressing complex features of the data. The output of the hidden layer is the hidden characteristic X4 with rich expression, then the X4 is input into the output layer, namely the two layers of full-connection layers, and the predicted value is obtained after the activation function is applied.
In the training process, a loss function between a predicted value and a true value is continuously calculated, and a back propagation algorithm is utilized to adjust network parameters of each layer so as to improve the identification performance of the network and minimize the loss function. This process will continue until the loss function converges or a preset iteration stop condition is reached.
The loss function is:
;
;
in the method, in the process of the invention,is a loss function, y represents a true value, < ->Represents the predicted value, M is the number of training sets, j represents the jth training data, y j Is the true value of the j-th training data, for example>Is the predicted value of the jth training data, lambda is the regularization coefficient, D is the number of network parameters, D represents the jth network parameter, w d Weight representing the d-th network parameter, < ->Is a smooth function +.>Is a superparameter->For controlling the degree of smoothness of the smoothing function.
And predicting the test set by using a pre-trained prediction model, and sorting the prediction result. If there is data which does not meet the prediction requirements, the data is used as error data, and the error data is moved from the test set to the training set to update the training set. Then, the model training and optimization of step S232 is performed again until the data in the test set all meet the prediction requirements. In this embodiment, the prediction requirement is evaluated by using a mean square error, and if the mean square error is less than 0.5, the prediction requirement is met.
Specifically, in an embodiment of the present invention, in step S3, interpolation processing is performed on weather forecast data, including:
step one, carrying out discrete gridding processing on weather forecast data according to the longitude and latitude of the weather forecast data, wherein each grid represents a longitude and latitude range, taking the weather forecast data in the grid as known data points, and determining the number and the positions of interpolation points according to the distribution of the known data points in the grid;
step two, selecting a Gaussian function to describe the spatial correlation between data points;
step three, calculating the space distance between any two known data points for all the known data points, calculating the variance according to the space distance, and substituting the space distance and the variance into a Gaussian function for fitting to obtain a fitting function;
step four, determining an interpolation range from the fitting function, and estimating the length scale of the Gaussian function;
step five, for a single interpolation point, calculating the spatial distance between the interpolation point and all known data points in the interpolation range according to the position of the interpolation point, substituting the spatial distance into a Gaussian function to calculate the interpolation weight of each known data point to the interpolation point;
step six, obtaining the value of the interpolation point by using the value of the known data point and the interpolation weight through a weighted average formula;
And step seven, repeating the step five to the step six to obtain the numerical values of all interpolation points.
In this embodiment, the weather forecast data is first subjected to discrete gridding processing according to the longitude and latitude of the weather forecast data, and the map is divided into a plurality of grids, each grid representing a longitude and latitude range. Then, the weather forecast data in each grid is used as known data points, and the number and the positions of interpolation points are determined according to the known data point distribution in the grid. Specifically, the intersection point of the grid, the grid center point, may be selected as the point where interpolation is required.
A gaussian function is then chosen to describe the spatial correlation between the data points. By fitting the spatial distances of existing data points, the parameters of the gaussian function can be estimated.
The gaussian function is:
;
wherein γ (h) is a Gaussian function, σ 2 Is the variance, s is the length scale, h represents the spatial distance between two points.
The parameters of the gaussian function are estimated from the fitted function, and the interpolation range can be estimated from the fitted function by observation. Specifically, the interpolation range indicates that the spatial correlation between points is high within this range, and the spatial correlation gradually decreases beyond this range. The interpolation range can be estimated by observing the attenuation of the gaussian function. In general, the interpolation range may be defined as the distance corresponding when the gaussian function value drops to 50% of the original variance. I.e. find a distance h such that γ (h) =0.5×σ 2 . This distance h can be used as an estimate of the interpolation range.
For a location where interpolation is desired, it is first necessary to calculate the spatial distance between it and a known data point and calculate the interpolation weight using a gaussian function. Data points that are closer will have a greater impact on the interpolation results.
And carrying out weighted average on the interpolation points by using the numerical values of the known data points and the corresponding interpolation weights to obtain numerical value estimation on the interpolation points. The accuracy of the interpolation results is affected by the accuracy of the gaussian fitting and the rationality of the interpolation weight calculation.
The weighted average formula is:
;
wherein z' (x) 0 ) Representing interpolation point x 0 Numerical value at A is the number of known data points in the interpolation range, x a Represents the a-th known data point, z (x a ) Represents x a Numerical value of beta a Represents x a For x 0 Is used for the interpolation weight of the (a);
the calculation formula of the interpolation weight is as follows:
;
wherein h (x 0 ,x a ) Represents x a And x 0 Is a spatial distance of (c).
After interpolation is carried out on the weather forecast data, the weather forecast data is input into a trained prediction model, and predicted photovoltaic data is obtained.
Specifically, the present embodiment uses past data and environmental factors to infer possible future conditions, and the historical photovoltaic data includes trends and laws of changes in photovoltaic power generation over a period of time. Through analysis and processing of the historical data, key features and modes can be extracted, such as periodic changes every day, every week or every quarter, the influence of illumination intensity and temperature changes on photovoltaic power generation, and the like. These features and patterns can help the predictive model understand the relationship between photovoltaic power generation and factors such as time, season, weather, etc.
Meteorological characteristics include various meteorological factors such as temperature, illumination intensity, wind speed, etc. These factors have a direct or indirect effect on photovoltaic power generation. For example, the illumination intensity and temperature have a significant effect on the power generation efficiency of the photovoltaic panel, and wind speed may also affect the heat dissipation of the photovoltaic panel, and so on. Thus, the meteorological features may provide possible weather conditions over a period of time in the future, thereby helping the predictive model to predict future photovoltaic power generation.
Historical photovoltaic data and meteorological features provide past experience and environmental factors, and through processing and analyzing the data, the prediction model can learn the relation between photovoltaic power generation capacity and factors such as time, weather and the like, and accordingly predict future photovoltaic data. When the model learns these relationships during the training process, it can use the learned knowledge to infer future photovoltaic data as future meteorological features are received.
Specifically, the weather forecast data after interpolation processing is input into a trained prediction model, characteristics of the weather forecast data are extracted by utilizing a second input layer and a second full-connection layer, weather forecast characteristics are obtained, then hidden relations between the photovoltaic data learned by the model and the weather data are correspondingly matched with the photovoltaic characteristics of the stored historical photovoltaic data according to the weather forecast data, the photovoltaic characteristics and the weather forecast characteristics are called and input into a splicing layer together, and a predicted value is obtained according to the splicing layer, the hidden layer and an output layer, namely the predicted photovoltaic data.
Specifically, in an embodiment of the present invention, step S4 includes the following steps:
s41, collecting data of an energy storage system, wherein the data comprise charge and discharge efficiency and energy storage capacity of the energy storage system;
s42, taking the predicted photovoltaic data and the energy storage system data as decision variables, searching all paths by using a search algorithm, wherein each path is an initial energy storage distribution scheme, and obtaining an initial energy storage distribution scheme set;
in order to obtain more energy storage distribution schemes as possible, a random search algorithm is adopted to search paths, and the specific process is as follows:
1. and determining the range of the decision variable, and taking the energy storage capacity as an upper value limit and the predicted photovoltaic data as a lower value limit.
2. The random number seed is set to 100.
3. And generating a plurality of random energy storage distribution schemes by using a random number generator according to the range of the decision variable, and ensuring that each scheme meets constraint conditions, namely that the energy storage distributed to the distributed photovoltaic power station does not exceed the energy storage capacity of an energy storage system of the distributed photovoltaic power station.
4. Repeating 3, and generating a random energy storage distribution scheme of the random number seed number so as to ensure that a diversified scheme set is obtained.
5. The repeated schemes are deleted, and the remaining schemes form an initial set of energy storage allocation schemes.
S43, carrying out iterative optimization on an initial energy storage distribution scheme set by adopting a particle swarm optimization algorithm, taking a minimized voltage offset index and photovoltaic active loss as objective functions, searching an optimal solution by iteratively updating the particle swarm, and outputting the minimum values of the optimal solution and the objective functions when the iteration termination condition is met, wherein the optimal solution is the optimal energy storage distribution scheme;
wherein, the objective function is:
;
wherein T represents a target time range, T represents a time period within the target time range, c U Penalty coefficient for voltage deviation, c PV Penalty coefficient for photovoltaic active loss, U r,t For the actual voltage value of the r-th distributed photovoltaic in the t time period, U r,R For the voltage rating of the r-th distributed photovoltaic in the t time period, G dec,r,t And the photovoltaic active loss of the r-th distributed photovoltaic in the t time period is represented.
Specifically, the particle swarm optimization algorithm is a heuristic optimization algorithm, which is used for searching an optimal solution in a multidimensional space, and in the embodiment, the specific optimization process is as follows:
1) Initializing a particle swarm: first, a population of particles needs to be initialized, each particle representing one possible energy storage allocation scheme. Each particle has a position vector (representing the energy storage distribution scheme) and a velocity vector.
2) Calculating the fitness: for each particle, its corresponding objective function value, namely the voltage offset index and the photovoltaic active loss, is calculated. These values will serve as fitness of the particles.
3) Updating the individual best position: for each particle, updating the individual optimal position according to the current fitness value, namely recording the optimal energy storage distribution scheme found so far.
4) Updating the global optimal position: and (3) in the whole particle swarm, the particle with the best fitness value is found, and the position of the particle is taken as the global optimal position, namely the optimal energy storage distribution scheme found so far is recorded.
5) Updating particle velocity and position: and updating the speed and the position of each particle according to the updating rule of the particle swarm optimization algorithm so as to move towards the direction of the individual optimal position and the global optimal position.
6) Repeating the iteration: repeating the steps 3) to 5) until the iteration termination condition is satisfied. The iteration termination condition is that the maximum number of iterations is reached or the objective function value converges to a certain threshold.
7) Outputting the optimal solution and the minimum value of the objective function: when the iteration termination condition is met, the energy storage distribution scheme corresponding to the global optimal position is output as an optimal solution, and the corresponding objective function value is output as a minimum value.
After the optimal energy storage distribution scheme is obtained, system scheduling and control are carried out according to the optimal energy storage distribution scheme, so that the energy storage system is ensured to distribute energy storage to each distributed photovoltaic power station according to the optimal scheme. And starting an energy storage system, and carrying out energy storage distribution on each distributed photovoltaic power station according to a distribution scheme.
In addition, referring to fig. 3, the present invention further provides a distributed photovoltaic and energy storage data processing system, where the system is configured to perform the method of any one of the above claims, and the system includes:
the historical data acquisition module is configured to acquire historical photovoltaic data and historical meteorological data according to preset time;
the module is responsible for acquiring historical data from a photovoltaic power station and a meteorological station, including information such as photovoltaic power generation capacity, irradiance, temperature, wind speed and the like. These data may be obtained in real time by sensors, monitoring devices, weather stations, etc., or may be imported from a history.
The photovoltaic data processing module is configured to preprocess historical photovoltaic data, and comprises data cleaning, missing value processing and abnormal value processing, so that processed historical photovoltaic data is obtained; to ensure accuracy and integrity of the data. The processed historical photovoltaic data will be used to build a predictive model and optimize the energy storage distribution scheme.
The weather forecast data acquisition module is configured to acquire corresponding weather forecast data according to the prediction time; including weather information over a period of time in the future, such as temperature, irradiance, wind speed, etc., over hours or days in the future.
And the weather data processing module is configured to perform interpolation processing on weather data, wherein the weather data comprises historical weather data and weather forecast data, and integrate and fill the historical weather data and the weather forecast data to acquire complete weather information. These data will be used for the prediction of photovoltaic power generation.
The model storage module is configured to establish a prediction model, and perform iterative training on the prediction model by using the processed historical photovoltaic data and the historical meteorological data to obtain and store the trained prediction model;
the photovoltaic prediction module is configured to input the weather forecast data after interpolation processing into a prediction model after training to obtain prediction photovoltaic data;
the energy storage system management module is configured to acquire energy storage system data according to requirements and perform preliminary pretreatment on the energy storage system data; so as to obtain the information of the real-time running state, the charging and discharging power and the like of the energy storage system.
The energy storage scheme optimizing module is configured to form an initial energy storage distribution scheme set according to the predicted photovoltaic data and the energy storage system data, and an internal optimizing algorithm is utilized to optimize the energy storage distribution scheme set so as to obtain an optimal energy storage distribution scheme; this module is used to optimize the distribution of stored energy to maximize system performance and efficiency.
And the energy storage distribution execution module is configured to distribute the energy storage of each distributed photovoltaic power station according to an optimal energy storage distribution scheme.
The processing system provided by the invention can integrate historical photovoltaic data, weather forecast data and energy storage system data, and can accurately predict photovoltaic power generation through a prediction model after pretreatment, interpolation processing and training, and meanwhile, the energy storage distribution scheme is optimized through the energy storage scheme optimizing module, so that the self-digestion capacity of the photovoltaic power generation system can be improved to the greatest extent, the running efficiency and the reliability of the system are further improved, and the application of the system is beneficial to optimizing the coordinated running of the photovoltaic power generation system and the energy storage system, improving the utilization efficiency of renewable energy sources and reducing the energy consumption cost, thereby having a positive pushing effect on the sustainable development of the energy system.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (10)
1. The processing method of the distributed photovoltaic and energy storage data is characterized by comprising the following steps of:
S1, acquiring historical photovoltaic data and historical meteorological data, preprocessing the historical photovoltaic data, and carrying out interpolation processing on the historical meteorological data to obtain processed historical photovoltaic data and historical meteorological data;
s2, establishing a prediction model, taking the processed historical photovoltaic data and the historical meteorological data as training data, and performing iterative training on the prediction model to obtain a trained prediction model;
s3, acquiring weather forecast data, carrying out interpolation processing on the weather forecast data, and inputting the processed weather forecast data into a trained prediction model to obtain predicted photovoltaic data;
s4, collecting energy storage system data, forming an initial energy storage distribution scheme set according to the predicted photovoltaic data and the energy storage system data, and optimizing the energy storage distribution scheme set by utilizing an optimization algorithm to obtain an optimal energy storage distribution scheme;
and S5, carrying out energy storage distribution on each distributed photovoltaic power station according to an optimal energy storage distribution scheme.
2. The method for processing distributed photovoltaic and energy storage data according to claim 1, wherein step S2 comprises the steps of:
s21, clustering historical meteorological data, wherein the number of the clusters is k, and k historical meteorological sets are obtained;
S22, marking the historical meteorological set as a cluster type to which the historical meteorological set belongs, forming k types of historical meteorological data columns according to k types, aligning the historical photovoltaic data with the corresponding historical meteorological data columns according to the time stamp, and adding the historical meteorological data columns into the historical photovoltaic data to form a dataset;
s23, constructing a prediction model, and performing iterative training on the prediction model by using the data set to obtain a trained prediction model.
3. The method for processing distributed photovoltaic and energy storage data according to claim 2, wherein step S21 comprises:
s211, constructing a clustering network, and determining the number of neurons of an input layer and the number of neurons of an output layer of the clustering network, wherein the number of neurons of the output layer is k;
s212 sets the iteration number N, and randomly initializes the weights { ω ] of the neurons in the clustered network when the initial step n=0 1 ,ω 2 ,...,ω n };
S213 from historical Meteorological data { ζ 1 ,ξ 2 ,...,ξ M Random extraction of a data xi in } 1 ;
S214 calculating the difference;
S215 selects the weight ω with the smallest difference v (0) The corresponding neuron v is taken as an excellent neuron;
s216, repeating the steps S213-S215 to obtain excellent neurons of all historical meteorological data;
s217 updates only the weights of the excellent neurons:
;
Wherein η represents a learning parameter;
s218 let n=n+1, and go to step S213;
s219 repeats steps S213-S218 until n=n, and the iteration ends, where each output layer neuron represents a cluster, and k clusters, that is, k historical meteorological sets, are obtained.
4. The method of claim 2, wherein step S23 includes:
s231, dividing the data set into a training set and a testing set according to a ratio of 7:3;
s232, respectively inputting the training sets into the prediction models for iterative training until reaching the iterative stopping condition to obtain pre-trained prediction models;
s233, predicting the test set by using a pre-trained prediction model, and sorting the prediction result:
if the data which does not meet the prediction requirement exists, the data is used as error data, the error data is replaced from the test set to the training set to update the training set, and the step S232 is carried out;
and if the test sets all meet the prediction requirements, ending the whole training to obtain a trained prediction model.
5. The method of claim 4, wherein step S232 includes:
the prediction model comprises a first input layer, a second input layer, a first full-connection layer, a second full-connection layer, a splicing layer, a hiding layer and an output layer;
Inputting historical photovoltaic data in a training set into a first input layer, and processing the historical photovoltaic data through a first full-connection layer to obtain photovoltaic sign X1;
inputting a historical meteorological data column in the training set into a second input layer, and processing the historical meteorological data column by a second full-connection layer to obtain meteorological features X2;
inputting the photovoltaic characteristic X1 and the meteorological characteristic X2 into a splicing layer for characteristic splicing to obtain a fusion characteristic X3;
inputting the fusion characteristic X3 into a hidden layer, wherein the hidden layer comprises m nerve unit layers which are sequentially overlapped, and carrying out layer-by-layer weighted summation on the fusion characteristic X3 from bottom to top by utilizing the m nerve unit layers to obtain a hidden characteristic X4, wherein the weight of the m-1 nerve unit layer is connected with the weight of the m nerve unit layer;
inputting the hidden characteristic X4 into an output layer, wherein the output layer is a two-layer full-connection layer, and predicting the hidden characteristic X4 by using the output layer to obtain a predicted value;
calculating a loss function according to the predicted value and the true value of the training set:
if the loss function converges, ending the iteration;
if the loss function is not converged, the error value between the predicted value and the true value is transmitted back to the first input layer by layer through a back propagation algorithm, the network parameters of each layer are adjusted in the back propagation process, and the next iteration is carried out.
6. The method of claim 5, wherein the loss function is:
;
;
in the method, in the process of the invention,is a loss function, y represents a true value, < ->Represents the predicted value, M is the number of training sets, j represents the jth training data, y j Is the true value of the j-th training data, for example>Is the predicted value of the jth training data, lambda is the regularization coefficient, D is the number of network parameters, D represents the jth network parameter, w d Weight representing the d-th network parameter, < ->Is a smooth function +.>Is a superparameter->For controlling the degree of smoothness of the smoothing function.
7. The method for processing distributed photovoltaic and energy storage data according to claim 1, wherein in step S3, interpolation processing is performed on weather forecast data, including:
step one, carrying out discrete gridding processing on weather forecast data according to the longitude and latitude of the weather forecast data, wherein each grid represents a longitude and latitude range, taking the weather forecast data in the grid as known data points, and determining the number and the positions of interpolation points according to the distribution of the known data points in the grid;
step two, selecting a Gaussian function to describe the spatial correlation between data points;
step three, calculating the space distance between any two known data points for all the known data points, calculating the variance according to the space distance, and substituting the space distance and the variance into a Gaussian function for fitting to obtain a fitting function;
Step four, determining an interpolation range from the fitting function, and estimating the length scale of the Gaussian function;
step five, for a single interpolation point, calculating the spatial distance between the interpolation point and all known data points in the interpolation range according to the position of the interpolation point, substituting the spatial distance into a Gaussian function to calculate the interpolation weight of each known data point to the interpolation point;
step six, obtaining the value of the interpolation point by using the value of the known data point and the interpolation weight through a weighted average formula;
and step seven, repeating the step five to the step six to obtain the numerical values of all interpolation points.
8. The method of claim 7, wherein the gaussian function is:
;
wherein γ (h) is a Gaussian function, σ 2 Is variance, s is a length scale, h represents a spatial distance between two points;
the weighted average formula is:
;
wherein z' (x) 0 ) Representing interpolation point x 0 Numerical value at A is the number of known data points in the interpolation range, x a Represents the a-th known data point, z (x a ) Represents x a Numerical value of beta a Represents x a For x 0 Is used for the interpolation weight of the (a);
the calculation formula of the interpolation weight is as follows:
;
wherein h (x 0 ,x a ) Represents x a And x 0 Is a spatial distance of (c).
9. The method for processing distributed photovoltaic and energy storage data according to claim 1, wherein step S4 comprises the steps of:
s41, collecting data of an energy storage system, wherein the data comprise charge and discharge efficiency and energy storage capacity of the energy storage system;
s42, taking the predicted photovoltaic data and the energy storage system data as decision variables, searching all paths by using a search algorithm, wherein each path is an initial energy storage distribution scheme, and obtaining an initial energy storage distribution scheme set;
s43, carrying out iterative optimization on an initial energy storage distribution scheme set by adopting a particle swarm optimization algorithm, taking a minimized voltage offset index and photovoltaic active loss as objective functions, searching an optimal solution by iteratively updating the particle swarm, and outputting the minimum values of the optimal solution and the objective functions when the iteration termination condition is met, wherein the optimal solution is the optimal energy storage distribution scheme;
wherein, the objective function is:
;
wherein T represents a target time range, T represents a time period within the target time range, c U Penalty coefficient for voltage deviation, c PV Penalty coefficient for photovoltaic active loss, U r,t For the actual voltage value of the r-th distributed photovoltaic in the t time period, U r,R For the voltage rating of the r-th distributed photovoltaic in the t time period, G dec,r,t And the photovoltaic active loss of the r-th distributed photovoltaic in the t time period is represented.
10. A distributed photovoltaic and energy storage data processing system for performing the method of any of claims 1-9, the system comprising:
the historical data acquisition module is configured to acquire historical photovoltaic data and historical meteorological data according to preset time;
the photovoltaic data processing module is configured to preprocess historical photovoltaic data, and comprises data cleaning, missing value processing and abnormal value processing, so that processed historical photovoltaic data is obtained;
the weather forecast data acquisition module is configured to acquire corresponding weather forecast data according to the prediction time;
the weather data processing module is configured to conduct interpolation processing on weather data, wherein the weather data comprises historical weather data and weather forecast data, and fills the weather data;
the model storage module is configured to establish a prediction model, and perform iterative training on the prediction model by using the processed historical photovoltaic data and the historical meteorological data to obtain and store the trained prediction model;
the photovoltaic prediction module is configured to input the weather forecast data after interpolation processing into a prediction model after training to obtain prediction photovoltaic data;
The energy storage system management module is configured to acquire energy storage system data according to requirements and perform preliminary pretreatment on the energy storage system data;
the energy storage scheme optimizing module is configured to form an initial energy storage distribution scheme set according to the predicted photovoltaic data and the energy storage system data, and an internal optimizing algorithm is utilized to optimize the energy storage distribution scheme set so as to obtain an optimal energy storage distribution scheme;
and the energy storage distribution execution module is configured to distribute the energy storage of each distributed photovoltaic power station according to an optimal energy storage distribution scheme.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410004568.4A CN117498400B (en) | 2024-01-03 | 2024-01-03 | Distributed photovoltaic and energy storage data processing method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410004568.4A CN117498400B (en) | 2024-01-03 | 2024-01-03 | Distributed photovoltaic and energy storage data processing method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117498400A true CN117498400A (en) | 2024-02-02 |
CN117498400B CN117498400B (en) | 2024-03-15 |
Family
ID=89680499
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410004568.4A Active CN117498400B (en) | 2024-01-03 | 2024-01-03 | Distributed photovoltaic and energy storage data processing method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117498400B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117791687A (en) * | 2024-02-28 | 2024-03-29 | 长峡数字能源科技(湖北)有限公司 | Energy management method of photovoltaic energy storage system |
CN118017564A (en) * | 2024-04-09 | 2024-05-10 | 深圳触觉智能科技有限公司 | Energy storage method based on open source hong Meng system |
CN118052337A (en) * | 2024-04-16 | 2024-05-17 | 广东工业大学 | Photovoltaic power station energy storage capacity optimization method and system based on benefit prediction |
CN118277796A (en) * | 2024-06-03 | 2024-07-02 | 中国水利水电第九工程局有限公司 | Information output method, device and equipment of photovoltaic module and storage medium |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105162149A (en) * | 2015-09-07 | 2015-12-16 | 中国电力科学研究院 | Fuzzy adaptive control based method for tracking output of power generation plan of light storage system |
CN106228274A (en) * | 2016-08-03 | 2016-12-14 | 河海大学常州校区 | Photovoltaic power station power generation amount Forecasting Methodology based on SOM Neural Network Data clustering recognition |
US20190020197A1 (en) * | 2017-07-12 | 2019-01-17 | Oci Company Ltd. | Method and system for managing charging/discharging of electric energy based on prediction for photovoltaic power and load |
CN112366753A (en) * | 2020-11-27 | 2021-02-12 | 国家电网有限公司 | Light-storage combined operation economic optimal control method |
CN114118596A (en) * | 2021-11-30 | 2022-03-01 | 华能清能通榆电力有限公司 | Photovoltaic power generation capacity prediction method and device |
CN114678903A (en) * | 2022-04-22 | 2022-06-28 | 云南电力试验研究院(集团)有限公司 | Photovoltaic power station AVC control method based on particle swarm optimization |
CN115907213A (en) * | 2022-12-19 | 2023-04-04 | 国家电网有限公司 | Cloud-terminal hierarchical architecture-based group control and group regulation strategy considering equipment health degree |
CN116738339A (en) * | 2023-06-09 | 2023-09-12 | 北京航空航天大学 | Multi-classification deep learning recognition detection method for small-sample electric signals |
-
2024
- 2024-01-03 CN CN202410004568.4A patent/CN117498400B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105162149A (en) * | 2015-09-07 | 2015-12-16 | 中国电力科学研究院 | Fuzzy adaptive control based method for tracking output of power generation plan of light storage system |
CN106228274A (en) * | 2016-08-03 | 2016-12-14 | 河海大学常州校区 | Photovoltaic power station power generation amount Forecasting Methodology based on SOM Neural Network Data clustering recognition |
US20190020197A1 (en) * | 2017-07-12 | 2019-01-17 | Oci Company Ltd. | Method and system for managing charging/discharging of electric energy based on prediction for photovoltaic power and load |
CN112366753A (en) * | 2020-11-27 | 2021-02-12 | 国家电网有限公司 | Light-storage combined operation economic optimal control method |
CN114118596A (en) * | 2021-11-30 | 2022-03-01 | 华能清能通榆电力有限公司 | Photovoltaic power generation capacity prediction method and device |
CN114678903A (en) * | 2022-04-22 | 2022-06-28 | 云南电力试验研究院(集团)有限公司 | Photovoltaic power station AVC control method based on particle swarm optimization |
CN115907213A (en) * | 2022-12-19 | 2023-04-04 | 国家电网有限公司 | Cloud-terminal hierarchical architecture-based group control and group regulation strategy considering equipment health degree |
CN116738339A (en) * | 2023-06-09 | 2023-09-12 | 北京航空航天大学 | Multi-classification deep learning recognition detection method for small-sample electric signals |
Non-Patent Citations (3)
Title |
---|
HONGPENG ZHANG等: "A Short-Term Photovoltaic Power Output Prediction for Virtual Plant Peak Regulation Based on K-means Clustering and Improved BP Neural Network", 《2021 11TH INTERNATIONAL CONFERENCE ON POWER, ENERGY AND ELECTRICAL ENGINEERING (CPEEE)》, 25 March 2021 (2021-03-25) * |
丁玉凤: "粒子群优化算法及其在电力***经济运行中的应用", 《中国硕士学位论文全文数据库 工程科技Ⅱ辑》, 15 May 2006 (2006-05-15) * |
李世昌: "用向前神经元网络进行短期负荷预测的方法", 福建电力与电工, no. 02, 30 June 1996 (1996-06-30) * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117791687A (en) * | 2024-02-28 | 2024-03-29 | 长峡数字能源科技(湖北)有限公司 | Energy management method of photovoltaic energy storage system |
CN117791687B (en) * | 2024-02-28 | 2024-05-14 | 长峡数字能源科技(湖北)有限公司 | Energy management method of photovoltaic energy storage system |
CN118017564A (en) * | 2024-04-09 | 2024-05-10 | 深圳触觉智能科技有限公司 | Energy storage method based on open source hong Meng system |
CN118052337A (en) * | 2024-04-16 | 2024-05-17 | 广东工业大学 | Photovoltaic power station energy storage capacity optimization method and system based on benefit prediction |
CN118277796A (en) * | 2024-06-03 | 2024-07-02 | 中国水利水电第九工程局有限公司 | Information output method, device and equipment of photovoltaic module and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN117498400B (en) | 2024-03-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN117498400B (en) | Distributed photovoltaic and energy storage data processing method and system | |
Aslam et al. | A survey on deep learning methods for power load and renewable energy forecasting in smart microgrids | |
CN112614009B (en) | Power grid energy management method and system based on deep expectation Q-learning | |
Wang et al. | Short-term wind speed forecasting based on information of neighboring wind farms | |
JP5888640B2 (en) | Photovoltaic power generation prediction apparatus, solar power generation prediction method, and solar power generation prediction program | |
Huang et al. | One‐day‐ahead hourly forecasting for photovoltaic power generation using an intelligent method with weather‐based forecasting models | |
Kalogirou et al. | Artificial intelligence techniques in solar energy applications | |
CN112116144B (en) | Regional power distribution network short-term load prediction method | |
CN112215428B (en) | Photovoltaic power generation power prediction method and system based on error correction and fuzzy logic | |
Wang et al. | A regional pretraining-classification-selection forecasting system for wind power point forecasting and interval forecasting | |
Eseye et al. | Short-term wind power forecasting using a double-stage hierarchical hybrid GA-ANN approach | |
CN111311001B (en) | Bi-LSTM network short-term load prediction method based on DBSCAN algorithm and feature selection | |
CN116826710A (en) | Peak clipping strategy recommendation method and device based on load prediction and storage medium | |
CN115912502A (en) | Intelligent power grid operation optimization method and device | |
Suresh et al. | Probabilistic LSTM-Autoencoder based hour-ahead solar power forecasting model for intra-day electricity market participation: A Polish case study | |
Das et al. | Optimized support vector regression-based model for solar power generation forecasting on the basis of online weather reports | |
CN116544934B (en) | Power scheduling method and system based on power load prediction | |
Choudhary et al. | Soft computing applications in wind speed and power prediction for wind energy | |
Arshad et al. | Wind power prediction using genetic programming based ensemble of artificial neural networks (GPeANN) | |
CN117034762A (en) | Composite model lithium battery life prediction method based on multi-algorithm weighted sum | |
Jafri et al. | The role of artificial intelligence in solar harvesting, storage, and conversion | |
CN115481788A (en) | Load prediction method and system for phase change energy storage system | |
Chinnadurrai et al. | Energy Management of a Microgrid based on LSTM Deep Learning Prediction Model | |
Padhan et al. | Power forecasting with minimal loss using lstm and pv model | |
EL KHAOUAT¹ et al. | A systematic literature review on prediction models in microgrids |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |