Green electricity energy storage scheduling method based on big data

Abstract

The invention provides a green electricity energy storage scheduling method based on big data, and the method comprises the steps: carrying out the prediction of historical electricity utilization data and real-time natural language data through employing an electricity utilization analysis model according to the historical electricity utilization data and real-time natural language data of an electricity utilization side, obtaining an electricity consumption prediction result and an electricity consumption interference factor; acquiring feature data of a power plant connected to the power generation side, and predicting the feature data of the power plant by adopting a power supply analysis model to obtain a power generation prediction result; and determining a green electricity energy storage scheduling strategy according to the electricity prediction result, the electricity utilization interference factor, the electricity generation prediction result and the current green electricity storage amount. The green electricity energy storage scheduling method has the beneficial effect that the green electricity energy storage scheduling efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric green energy dispatching, and in particular to a green electricity energy storage dispatching method based on big data. Background Art

[0002] Due to the instability of green electricity (i.e. renewable energy), that is, the time period in which it is generated and the great impact of climate, and the fact that the electricity demand on the electricity consumption side is due to some accidental factors, the scheduling of green electricity storage will be inaccurate, resulting in electricity shortages or energy waste. The existing technology currently mainly uses cost and loss as scheduling targets, and does not consider the failure of scheduling strategies caused by the instability of green electricity and accidental factors. Summary of the invention

[0003] The main purpose of the embodiment of the present invention is to propose a green electricity energy storage scheduling method based on big data, which achieves scheduling efficiency and energy-saving management of green electricity energy storage.

[0004] One aspect of the present invention provides a green electricity storage scheduling method based on big data, characterized by comprising:

[0005] According to the historical power consumption data and the real-time natural language data of the power consumption side, the historical power consumption data and the real-time natural language data are predicted using the power consumption analysis model to obtain the power consumption prediction result and the power consumption interference factor;

[0006] Acquire characteristic data of the power plant connected to the power generation side, use the power supply analysis model to predict the characteristic data of the power plant, and obtain the power generation prediction result;

[0007] Determining a green electricity storage scheduling strategy according to the electricity prediction result, the electricity interference factor, the power generation prediction result and the current green electricity storage amount;

[0008] The training process of the power consumption analysis model includes:

[0009] Acquire time series data of the historical electricity data, and acquire historical natural language data, wherein the natural series data is used to characterize electricity consumption data affected by time, and the historical natural language data is used to characterize text data that has an occasional impact on electricity consumption;

[0010] The time series data is processed using a GRU recurrent neural network to obtain a first fusion feature, wherein the first fusion feature is used to characterize the dependency relationship between the historical electricity data and the time series data;

[0011] The time series data and the historical natural language data are processed by using an NLP language model to obtain a second fusion feature, where the second fusion feature is used to characterize a dependency relationship between the time series data and the historical natural language data;

[0012] Connecting the first fusion feature and the second fusion feature to obtain a comprehensive feature, determining the power consumption analysis model according to the comprehensive feature, and determining the power consumption disturbance factor according to the comprehensive feature;

[0013] The training process of the power analysis model includes:

[0014] Get the dataset;

[0015] Perform feature engineering on the power plant dataset to obtain the main features that affect power generation;

[0016] According to the main features and the feature data, XGBoost regression training is used for each type of power plant to obtain a corresponding basic machine learning model;

[0017] The basic machine learning model is subjected to model fusion processing to obtain the power supply analysis model.

[0018] According to the green electricity storage scheduling method based on big data, the training process of the electricity consumption analysis model includes:

[0019] Obtain time series data and historical natural language data, where time series data includes electricity consumption data, weather data, holidays and event data on the electricity user side, and historical natural language data includes social media data, news report data and user feedback;

[0020] Preprocess the time series data and historical natural language data. The preprocessing of time series data includes missing value processing, normalization and feature engineering. Feature engineering is also used to fuse features with correlation relationships. The preprocessing of historical natural language data includes text cleaning, word segmentation and word vectorization.

[0021] According to the green electricity storage scheduling method based on big data, the time series data is processed by using a GRU recurrent neural network to obtain a first fusion feature, including:

[0022] The GRU recurrent neural network includes an input layer, a GRU layer and a fully connected layer;

[0023] The time series data is input as a three-dimensional tensor through the input layer, the long-term dependency of the time series is captured by the three-dimensional tensor through the GRU layer, and the long-term dependency is feature fused through the fully connected layer to obtain the first fused feature, which is a time series feature.

[0024] According to the green electricity storage scheduling method based on big data, the time series data and the historical natural language data are processed by an NLP language model to obtain a second fusion feature, including:

[0025] The NLP model includes an embedding layer, a GRU layer and a fully connected layer;

[0026] Obtain word vectors of the historical natural language data, and process the word vectors through an embedding layer to obtain semantic and contextual information related to electricity consumption;

[0027] Extracting semantic and contextual information through the GRU layer of the NLP model to obtain text features;

[0028] The time series features and the text features are fused through the fully connected layer of the NLP model to obtain a second fused feature.

[0029] According to the green electricity energy storage scheduling method based on big data, the first fusion feature and the second fusion feature are connected to obtain a comprehensive feature, the power consumption analysis model is determined according to the comprehensive feature, and the power consumption disturbance factor is determined according to the comprehensive feature, including:

[0030] The first fusion feature and the second fusion feature are connected to obtain a comprehensive feature, and a power consumption prediction result is obtained through an output layer;

[0031] The power consumption analysis model uses mean square error as a loss function and uses Adam or RMSprop optimizer for training, and the training process is optimized using cross validation;

[0032] The trained power consumption analysis model is analyzed using one of the SHAP value and the LIME to obtain the power consumption disturbance factor.

[0033] According to the green electricity storage scheduling method based on big data, obtaining a data set also includes:

[0034] Obtained through third-party platforms and data synthesis, where the data set includes historical power generation, meteorological data, and characteristic data of distributed green power generation equipment;

[0035] And, the acquired data set is subjected to data cleaning and data conversion processing; wherein data cleaning includes missing value processing and outlier detection, wherein missing value processing is performed by filling and deleting missing values, wherein data conversion includes identifying abnormal data using the Z-score statistical method; wherein data conversion includes time feature processing and category coding processing, wherein time feature processing converts time features into numerical types, wherein category coding processing includes converting power plant types into numerical types.

[0036] According to the green electricity storage scheduling method based on big data, the method further includes:

[0037] The data obtained from the third-party platform is padded and enhanced by data synthesis to obtain the data set, wherein the data enhancement is obtained by using random data generated by statistical distribution.

[0038] According to the green electricity storage scheduling method based on big data, obtaining a data set also includes:

[0039] The acquired data set is subjected to data cleaning and data conversion processing; wherein data cleaning includes missing value processing and outlier detection, wherein missing value processing is performed by filling and deleting missing values, wherein data conversion includes identifying abnormal data using the Z-score statistical method; wherein data conversion includes time feature processing and category coding processing, wherein time feature processing converts time features into numerical types, wherein category coding processing includes converting power plant types into numerical types.

[0040] According to the green electricity storage scheduling method based on big data, XGBoost regression training is used for each type of power plant according to the main features and the feature data to obtain a corresponding basic machine learning model, including:

[0041] According to the type of the power plant, a model that meets the main characteristics and the characteristic data is selected from basic machine learning for training;

[0042] The trained basic machine learning model is used for prediction, evaluation and hyperparameter tuning, where the hyperparameter tuning includes grid search and random search, to complete the training of the basic machine learning model for each type of power plant, where the basic machine model is selected according to the power plant type and power generation characteristics of the power plant.

[0043] According to the green power storage scheduling method based on big data, the basic machine learning model is subjected to model fusion processing to obtain the distributed green power analysis model, including:

[0044] Decision tree regression and support vector regression are used as meta-learners, and multiple trained basic machine learning models are used as features to fit the meta-learners to obtain the distributed green power analysis model.

[0045] According to the green electricity energy storage scheduling method based on big data, the green electricity energy storage scheduling strategy is determined according to the electricity prediction result, the electricity interference factor, the power generation prediction result and the current green electricity storage amount, including:

[0046] Determine the electricity demand within a preset time period according to the electricity prediction result and the electricity interference factor, and determine the total power generation of the connected power plant within the preset time period according to the power generation prediction result;

[0047] According to the electricity demand, the total power generation and the current green electricity storage, a corresponding green electricity storage scheduling strategy is selected, wherein the green electricity storage scheduling strategy includes at least one of green electricity priority, energy storage system scheduling, dynamic pricing and load management.

[0048] The beneficial effects of the present invention are as follows: the electricity consumption data on the electricity consumption side is predicted based on the time series through the GRU recurrent neural network and the NLP model, while taking into account the objective influencing factors (such as weather, time period, etc.) and accidental factors (human or accidental events), so that the electricity consumption on the electricity consumption side is accurately predicted; through integrated learning, different types of renewable energy power plants are used to predict the accurate power generation of different connected power plants; in green energy storage, efficient green power scheduling is performed according to the real-time predicted power generation and demand power consumption according to the pre-scheduling priority. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0050] Figure 1 It is a schematic diagram of the green electricity storage scheduling process of an embodiment of the present invention.

[0051] Figure 2 4 is a schematic diagram of the training process of the electricity consumption analysis model according to an embodiment of the present invention.

[0052] Figure 3 1 is a schematic diagram of a preprocessing flow of training data for an electrical analysis model according to an embodiment of the present invention.

[0053] Figure 4 It is a schematic diagram of the time series feature extraction process of an embodiment of the present invention.

[0054] Figure 5 It is a schematic diagram of a process for obtaining the dependency relationship between time series data and historical natural language data according to an embodiment of the present invention.

[0055] Figure 6 It is a schematic diagram of the training process of the power supply analysis model according to an embodiment of the present invention.

[0056] Figure 7 It is a schematic diagram of the integrated learning process of an embodiment of the present invention. DETAILED DESCRIPTION

[0057] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. In the subsequent description, the use of suffixes such as "module", "component" or "unit" used to represent elements is only for the purpose of facilitating the description of the present invention, and has no specific meaning in itself. Therefore, "module", "component" or "unit" can be used in a mixed manner. "First", "second" and the like are only used for the purpose of distinguishing technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features or implicitly indicating the order of the indicated technical features. In this subsequent description, the continuous numbering of the method steps is for the convenience of review and understanding. In combination with the overall technical solution of the present invention and the logical relationship between the various steps, adjusting the implementation order between the steps will not affect the technical effect achieved by the technical solution of the present invention. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0058] refer to Figure 1 ,in Figure 1 A schematic diagram of the green power storage scheduling process of an embodiment of the present invention includes but is not limited to steps S100 to S300:

[0059] S100, based on the historical power consumption data and the real-time natural language data of the power consumption side, the historical power consumption data and the real-time natural language data are predicted using a power consumption analysis model to obtain a power consumption prediction result and a power consumption interference factor.

[0060] In some embodiments, the electricity consumption interference factors include weather, electricity consumption holidays and events, and some occasional times, such as electricity price adjustments, policy changes, economic activities, etc.

[0061] S200, acquiring characteristic data of a power plant connected to the power generation side, and predicting the characteristic data of the power plant using a power supply analysis model to obtain a power generation prediction result.

[0062] In some embodiments, the characteristic data of the power plant includes the type of power plant, historical power generation, meteorological data, equipment status data (such as the opening and failure of the generator set) and time characteristics (such as date, hour, season).

[0063] S300, determining a green electricity storage scheduling strategy according to electricity prediction results, electricity interference factors, power generation prediction results and current green electricity storage capacity.

[0064] In some embodiments, the electricity demand within a preset time period is determined based on electricity forecast results and electricity interference factors, and the total power generation of connected power plants within the preset time period is determined based on power generation forecast results; based on the electricity demand, total power generation and current green electricity storage, a corresponding green electricity energy storage scheduling strategy is selected, wherein the green electricity energy storage scheduling strategy includes at least one of green electricity priority, energy storage system scheduling, dynamic pricing and load management.

[0065] In some embodiments, reference Figure 2 The training process diagram of the power consumption analysis model shown includes but is not limited to steps S400 to S700:

[0066] S400, obtaining time series data of historical electricity data and obtaining historical natural language data, wherein the natural series data is used to represent electricity consumption data affected by time, and the historical natural language data is used to represent text data that has occasional effects on electricity consumption.

[0067] In some embodiments, reference Figure 3 The schematic diagram of the preprocessing process of the training data of the electrical analysis model shown includes but is not limited to steps S410 to S430:

[0068] S410, acquiring time series data and historical natural language data, wherein the time series data includes power consumption data, weather data, holidays and event data on the power user side, and the historical natural language data includes social media data, news report data and user feedback;

[0069] S420, preprocessing the time series data and historical natural language data, wherein the preprocessing of the time series data includes missing value processing, normalization and feature engineering, wherein the feature engineering is also used to fuse features with correlation relationships; wherein the preprocessing of the historical natural language data includes text cleaning, word segmentation processing and word vectorization processing.

[0070] In some embodiments, electricity consumption data includes the collection of historical electricity consumption data, which is usually recorded by hours or minutes; weather data includes temperature, humidity, precipitation, and wind speed, which are obtained by calling the meteorological API interface on the electricity consumption side (target area); among them, holiday and event data include public holidays and specific activities that affect electricity demand.

[0071] In some embodiments, natural language data includes: social media data, which can capture social media posts, comments, etc. related to electricity consumption through crawler technology; news reports, collecting news articles about electricity demand, policy changes, economic activities, etc.; user feedback, user evaluation and feedback information on electricity services.

[0072] In some embodiments, time series data preprocessing includes: missing value processing, using interpolation or filling methods to process missing data; normalization, normalizing the data to the range of [0, 1], this method is used to improve the model convergence speed; feature engineering, extracting features (such as time series features, lag features, etc.), and merging weather data with electricity consumption data.

[0073] In some embodiments, natural language data preprocessing includes: text cleaning, removing special characters, punctuation marks, and stop words; word segmentation, dividing the text into words; word vectorization, using pre-trained word vectors (such as Word2Vec, GloVe) or models such as BERT to convert text into vector representations.

[0074] S500, using a GRU recurrent neural network to process the time series data to obtain a first fusion feature, wherein the first fusion feature is used to characterize the dependency relationship between the historical electricity data and the time series data.

[0075] In some embodiments, reference Figure 4 The schematic diagram of the time series feature extraction process shown in FIG. 1 includes but is not limited to steps S510 to S520:

[0076] S510, constructing a GRU recurrent neural network, wherein the GRU recurrent neural network includes an input layer, a GRU layer, and a fully connected layer;

[0077] S520, input the time series data as a three-dimensional tensor through the input layer, capture the long-term dependency of the time series of the three-dimensional tensor through the GRU layer, and perform feature fusion on the long-term dependency through the fully connected layer to obtain a first fused feature, which is a time series feature.

[0078] S600, using an NLP language model to process the time series data and the historical natural language data to obtain a second fusion feature, where the second fusion feature is used to characterize the dependency relationship between the time series data and the historical natural language data.

[0079] In some embodiments, reference Figure 5 The schematic diagram of the process of obtaining the dependency relationship between time series data and historical natural language data shown in FIG. 1 includes but is not limited to steps S610 to S630:

[0080] S610, constructing an NLP model, wherein the NLP model includes an embedding layer, a GRU layer, and a fully connected layer;

[0081] S620, obtaining word vectors of historical natural language data, and processing the word vectors through an embedding layer to obtain semantic and context information related to electricity consumption;

[0082] S630, extracting semantic and context information through the GRU layer of the NLP model to obtain text features;

[0083] S640: Fusing the time series features and the text features through a fully connected layer of the NLP model to obtain a second fused feature.

[0084] S700, connecting the first fusion feature and the second fusion feature to obtain a comprehensive feature, determining a power consumption analysis model according to the comprehensive feature, and determining a power consumption disturbance factor according to the comprehensive feature.

[0085] In some embodiments, a comprehensive feature is obtained by connecting the first fusion feature and the second fusion feature, and a power consumption prediction result is obtained through the output layer; the power consumption analysis model uses mean square error as the loss function, and is trained using Adam or RMSprop optimizer, and the training process is optimized using cross validation; the trained power consumption analysis model is analyzed using one of the SHAP value and LIME to obtain a power consumption disturbance factor.

[0086] In some embodiments, reference Figure 6 The training process diagram of the power analysis model shown includes but is not limited to steps S800 to S1000:

[0087] S800, obtaining a data set, wherein the data set includes a power plant type, historical power generation, meteorological data, and time characteristics of power generation equipment status data of the power plant, and performing feature engineering processing on the power plant data set to obtain main features that affect power generation.

[0088] In some embodiments, the acquired data set is also included in data cleaning and data conversion processing; wherein data cleaning includes missing value processing and outlier detection, wherein missing value processing is performed by filling and deleting missing values, wherein data conversion includes using Z-score statistical methods to identify abnormal data; wherein data conversion includes time feature processing and category coding processing, wherein time feature processing converts time features into numerical types, wherein category coding processing includes converting power plant types into numerical types.

[0089] In some embodiments, data obtained by the third-party platform is padded and enhanced using data synthesis to obtain a data set, wherein the data enhancement is obtained using random data generated by statistical distribution, wherein the third-party platform is a power company, meteorological service provider or institution that provides power plants, wherein the random data is generated based on statistical distribution (such as normal distribution, uniform distribution, etc.). For example, new power generation data can be generated based on the distribution of historical power generation, and it can be understood that the data synthesis is only used for model training.

[0090] In some embodiments, reference Figure 7 The schematic diagram of the integrated learning process shown includes but is not limited to steps S810 to S820:

[0091] S810, selecting a model that meets the main characteristics and characteristic data from basic machine learning according to the type of power plant for training.

[0092] S820, predict, evaluate and perform hyperparameter tuning on the trained basic machine learning model, wherein the hyperparameter tuning includes grid search and random search, to complete the training of the basic machine learning model for each type of power plant, wherein the basic machine model is selected according to the power plant type and power generation characteristics of the power plant.

[0093] In some embodiments, the power generation characteristics are the characteristics of different green energy sources, including power generation time, seasonal changes, wind speed changes, water volume, etc.

[0094] In some embodiments, wind power generation is integrated with multiple decision trees through random forests, which is suitable for processing complex nonlinear relationships, and uses support vector machines to classify or regress by constructing hyperplanes, which is suitable for small samples and high-dimensional data, and through long short-term memory networks (LSTMs) suitable for processing time series data, which can capture the temporal characteristics of wind speed changes.

[0095] In some embodiments, tidal power generation uses the K nearest neighbor algorithm (KNN), that is, prediction is made by finding the nearest neighbor samples, which is suitable for small data sets. Tidal power generation can use a decision tree, considering its periodicity, for classification and regression problems, and finally ensemble learning uses AdaBoost.

[0096] Through the above embodiments, the characteristics of different green energy sources can be fully utilized to improve the accuracy of power generation prediction.

[0097] In some embodiments, it is also possible to use the distribution weights of different green energy sources. For example, a city in a mountainous area is mainly powered by wind power, supplemented by hydropower. Therefore, weights can be assigned to different sub-learners, that is, the sub-learner for wind power supply has a higher weight, while the weight of hydropower is lower.

[0098] S900, based on the main features and feature data, uses XGBoost regression training for each type of power plant to obtain the corresponding basic machine learning model.

[0099] In some embodiments, decision tree regression and support vector regression are used as meta-learners, and multiple trained basic machine learning models are used as features to fit the meta-learner to obtain a distributed green power analysis model.

[0100] S1000 performs model fusion processing on the basic machine learning model to obtain a power supply analysis model.

[0101] The beneficial effects of the present invention are as follows: the electricity consumption data on the electricity consumption side is predicted based on the time series through the GRU recurrent neural network and the NLP model, while taking into account the objective influencing factors (such as weather, time period, etc.) and accidental factors (human or accidental events), so that the electricity consumption on the electricity consumption side is accurately predicted; through integrated learning, different types of renewable energy power plants are used to predict the accurate power generation of different connected power plants; in green energy storage, efficient green power scheduling is performed according to the real-time predicted power generation and demand power consumption according to the pre-scheduling priority.

[0102] An embodiment of the present invention further provides an electronic device, the electronic device comprising a processor and a memory;

[0103] The memory stores a program;

[0104] The processor executes the program to execute the aforementioned green electricity energy storage scheduling method based on big data; the electronic device has the function of carrying and running the software system for green electricity energy storage scheduling based on big data provided by an embodiment of the present invention, for example, a personal computer, a minicomputer, a main frame, a workstation, a network or distributed computing environment, a separate or integrated computer platform, or communicates with a charged particle tool or other imaging device, etc.

[0105] An embodiment of the present invention further provides a computer-readable storage medium, wherein the storage medium stores a program, and the program is executed by a processor to implement the green electricity storage scheduling method based on big data as described above.

[0106] In some selectable embodiments, the function / operation mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the function / operation involved, the two boxes shown in succession can actually be executed substantially simultaneously or the boxes can sometimes be executed in reverse order. In addition, the embodiment presented and described in the flow chart of the present invention is provided by way of example, for the purpose of providing a more comprehensive understanding of technology. The disclosed method is not limited to the operation and logic flow presented herein. Selectable embodiments are expected, wherein the order of various operations is changed and the sub-operation of a part for which is described as a larger operation is performed independently.

[0107] The embodiment of the present invention also discloses a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device can read the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the aforementioned green power storage scheduling method based on big data.

[0108] In addition, although the present invention is described in the context of functional modules, it should be understood that, unless otherwise specified, one or more of the functions and / or features described may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the present invention. More specifically, in view of the properties, functions, and internal relationships of the various functional modules in the device disclosed herein, the actual implementation of the module will be understood within the conventional skills of the engineer. Therefore, those skilled in the art can implement the present invention set forth in the claims without excessive experimentation using ordinary techniques. It is also understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents.

[0109] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0110] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0111] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0112] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0113] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0114] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

[0115] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the described embodiments. Those skilled in the art may make various equivalent modifications or substitutions without violating the spirit of the present invention. These equivalent modifications or substitutions are all included in the scope defined by the claims of this application.

Claims

1. A green electricity storage scheduling method based on big data, characterized in that: include: According to the historical power consumption data and the real-time natural language data of the power consumption side, the historical power consumption data and the real-time natural language data are predicted using the power consumption analysis model to obtain the power consumption prediction result and the power consumption interference factor; Acquire characteristic data of the power plant connected to the power generation side, use the power supply analysis model to predict the characteristic data of the power plant, and obtain the power generation prediction result; Determining a green electricity storage scheduling strategy according to the electricity prediction result, the electricity interference factor, the power generation prediction result and the current green electricity storage amount; The training process of the power consumption analysis model includes: Acquire time series data of the historical electricity data, and acquire historical natural language data, wherein the natural series data is used to characterize electricity consumption data affected by time, and the historical natural language data is used to characterize text data that has an occasional impact on electricity consumption; The time series data is processed using a GRU recurrent neural network to obtain a first fusion feature, wherein the first fusion feature is used to characterize the dependency relationship between the historical electricity data and the time series data; The time series data and the historical natural language data are processed by using an NLP language model to obtain a second fusion feature, where the second fusion feature is used to characterize a dependency relationship between the time series data and the historical natural language data; Connecting the first fusion feature and the second fusion feature to obtain a comprehensive feature, determining the power consumption analysis model according to the comprehensive feature, and determining the power consumption disturbance factor according to the comprehensive feature; The training process of the power analysis model includes: Get the dataset; Perform feature engineering on the power plant dataset to obtain the main features that affect power generation; According to the main features and the feature data, XGBoost regression training is used for each type of power plant to obtain a corresponding basic machine learning model; The basic machine learning model is subjected to model fusion processing to obtain the power supply analysis model.

2. The green electricity storage scheduling method based on big data according to claim 1 is characterized in that: The training process of the power consumption analysis model includes: Obtain time series data and historical natural language data, where time series data includes electricity consumption data, weather data, holidays and event data on the electricity user side, and historical natural language data includes social media data, news report data and user feedback; Preprocess the time series data and historical natural language data. The preprocessing of time series data includes missing value processing, normalization and feature engineering. Feature engineering is also used to fuse features with correlation relationships. The preprocessing of historical natural language data includes text cleaning, word segmentation and word vectorization.

3. The green electricity storage scheduling method based on big data according to claim 2 is characterized in that: The GRU recurrent neural network is used to process the time series data to obtain a first fusion feature, including: The GRU recurrent neural network includes an input layer, a GRU layer and a fully connected layer; The time series data is input as a three-dimensional tensor through the input layer, the long-term dependency of the time series is captured by the three-dimensional tensor through the GRU layer, and the long-term dependency is feature fused through the fully connected layer to obtain the first fused feature, which is a time series feature.

4. The green electricity storage scheduling method based on big data according to claim 3 is characterized in that: The adopting of the NLP language model to process the time series data and the historical natural language data to obtain the second fusion feature includes: The NLP model includes an embedding layer, a GRU layer and a fully connected layer; Obtain word vectors of the historical natural language data, and process the word vectors through an embedding layer to obtain semantic and contextual information related to electricity consumption; Extracting semantic and contextual information through the GRU layer of the NLP model to obtain text features; The time series features and the text features are fused through the fully connected layer of the NLP model to obtain a second fused feature.

5. The green electricity storage scheduling method based on big data according to claim 4 is characterized in that: The step of connecting the first fusion feature and the second fusion feature to obtain a comprehensive feature, determining the power consumption analysis model according to the comprehensive feature, and determining the power consumption disturbance factor according to the comprehensive feature includes: The first fusion feature and the second fusion feature are connected to obtain a comprehensive feature, and a power consumption prediction result is obtained through an output layer; The power consumption analysis model uses mean square error as a loss function and uses Adam or RMSprop optimizer for training, and the training process is optimized using cross validation; The trained power consumption analysis model is analyzed using one of the SHAP value and the LIME to obtain the power consumption disturbance factor.

6. The green electricity storage scheduling method based on big data according to claim 1 is characterized in that: The acquiring of the data set further includes: Obtained through third-party platforms and data synthesis, where the data set includes historical power generation, meteorological data, and characteristic data of distributed green power generation equipment; And, the acquired data set is subjected to data cleaning and data conversion processing; wherein data cleaning includes missing value processing and outlier detection, wherein missing value processing is performed by filling and deleting missing values, wherein data conversion includes identifying abnormal data using the Z-score statistical method; wherein data conversion includes time feature processing and category coding processing, wherein time feature processing converts time features into numerical types, wherein category coding processing includes converting power plant types into numerical types.

7. The green electricity storage scheduling method based on big data according to claim 6 is characterized in that: The method further comprises: The data obtained from the third-party platform is padded and enhanced by data synthesis to obtain the data set, wherein the data enhancement is obtained by using random data generated by statistical distribution.

8. The green electricity storage scheduling method based on big data according to claim 6 is characterized in that: According to the main features and the feature data, XGBoost regression training is used for each type of power plant to obtain a corresponding basic machine learning model, including: According to the type of the power plant, a model that meets the main characteristics and the characteristic data is selected from basic machine learning for training; The trained basic machine learning model is used for prediction, evaluation and hyperparameter tuning, where the hyperparameter tuning includes grid search and random search, to complete the training of the basic machine learning model for each type of power plant, where the basic machine model is selected according to the power plant type and power generation characteristics of the power plant.

9. The green electricity storage scheduling method based on big data according to claim 1 is characterized in that: The step of performing model fusion processing on the basic machine learning model to obtain the distributed green power analysis model includes: Decision tree regression and support vector regression are used as meta-learners, and multiple trained basic machine learning models are used as features to fit the meta-learners to obtain the distributed green power analysis model.

10. The green electricity storage scheduling method based on big data according to claim 1, characterized in that: The determining of a green electricity storage scheduling strategy according to the electricity prediction result, the electricity interference factor, the power generation prediction result and the current green electricity storage amount includes: Determine the electricity demand within a preset time period according to the electricity prediction result and the electricity interference factor, and determine the total power generation of the connected power plant within the preset time period according to the power generation prediction result; According to the electricity demand, the total power generation and the current green electricity storage, a corresponding green electricity storage scheduling strategy is selected, wherein the green electricity storage scheduling strategy includes at least one of green electricity priority, energy storage system scheduling, dynamic pricing and load management.

Images