A new energy power generation power prediction system based on deep learning

Abstract

The present invention discloses a new energy power generation prediction system based on deep learning, comprising a wind power ramping framework construction module for constructing an information hierarchy framework for new energy power generation, a natural language processing module, a real-time correction module for flow information, a regional meteorological analysis module, and a fusion prediction module for performing fusion prediction of new energy power generation. The present invention not only introduces the influence of active and reactive flow power data at different nodes on wind power ramping events, but also can correct the flow power information in real time when encountering wind power ramping. After correction, it is combined with the numerical results of the new energy power generation environment after regional meteorological analysis to perform fusion prediction, forming a more comprehensive prediction model input, ensuring the accuracy and reliability of the data, and being beneficial to the prediction and evaluation of new energy power generation when facing wind power ramping events.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of new energy data processing, and particularly relates to a new energy power generation power prediction system based on deep learning. BACKGROUND

[0002] The large amplitude change of wind power generation power of wind power generation in new energy in a short time is referred to as a wind power ramping event, and the wind power ramping event can cause short-term imbalance of power generation and power consumption of an electric power system, and threatens safe operation of a power grid. The existing prediction methods of the wind power ramping event can be roughly divided into two categories of indirect prediction method and direct prediction method. The indirect prediction method refers to predicting a wind power time sequence by means of wind power prediction technology, and then detecting occurrence of the ramping event according to a prediction result and a definition of the ramping event; the direct prediction method uses historical sample data to mine a relationship between ramping characteristic quantities and regional meteorological information, and establishes a mapping from the meteorological information to the ramping characteristic quantities, without needing to perform prediction of wind power in advance.

[0003] However, in the prediction of the wind power ramping event by using the existing new energy power generation power prediction system, the method for predicting the acquired regional meteorological information is relatively single in the direct prediction method, and the influence of power flow information of the wind power grid on the wind power ramping event is not considered, and the power flow information of the current wind power grid is not fused and predicted after being corrected in real time. Therefore, in the scene of a multi-dimensional space range and a time scale, in the face of a multi-level and multi-target influence event, the prediction effect and accuracy of the new energy power generation in the face of the wind power ramping event are generally poor. SUMMARY

[0004] In view of the above problems in the existing data processing of the wind power ramping, the present application is proposed.

[0005] Therefore, one object of the present application is to provide a new energy power generation power prediction system based on deep learning, which has a multi-level and multi-target influence event judgment function in the process of fusing and predicting regional meteorological information after collecting and correcting real-time power flow information of a wind power grid in real time, and the prediction data is more rich and accurate.

[0006] To solve the above technical problems, the present application provides the following technical scheme: a new energy power generation power prediction system based on deep learning, comprising:

[0007] A wind power ramping framework modeling module is configured to construct an information level framework of new energy power generation power, divide and configure a new energy power generation power prediction level and a matching power generation power associated characteristic word, and generate an information framework of new energy related data of different levels and targets according to the new energy power generation power prediction level and the matching power generation power associated characteristic word.

[0008] The natural language processing module is configured to receive a power generation prediction request sent by the new energy power station, perform semantic analysis on the power generation prediction request according to a pre-trained natural language processing model, and obtain a corresponding information hierarchical framework and matched power generation associated characteristic words.

[0009] The power flow information real-time correction module is configured to determine power flow information of new energy power generation and information framework information of a matched level in response to the information framework of the wind power climbing framework construction module and each power generation associated characteristic word processed by the natural language processing module, analyze a comprehensive influence degree of the power flow information under a current information framework according to the power flow information and the information framework information of the matched level, determine whether to correct the power flow information according to the comprehensive influence degree of the power flow information under the current hierarchical framework, build a power flow information correction model based on the Newton-Raphson method to complete the correction, and generate a power flow information correction result.

[0010] The regional meteorological analysis module is configured to obtain regional meteorological information under given initial conditions and boundary conditions and receive the power generation associated characteristic words of the natural language processing module, perform numerical environmental prediction of new energy power generation according to the power generation associated characteristic words and in combination with the obtained regional meteorological information, and generate a new energy power generation environmental numerical result.

[0011] The fusion prediction module is configured to receive the new energy power generation environmental numerical result and the power flow information correction result, perform fusion prediction of new energy power generation power, generate a fusion prediction result of new energy power generation power, and output the fusion prediction result.

[0012] As a preferred scheme of the present application, the power flow information real-time correction module analyzes the comprehensive influence degree of the power flow information under the current information framework according to the power flow information and the information framework information of the matched level, and specifically by building a comprehensive influence degree evaluation model of the power flow information.

[0013] The power flow information and the information framework information of the matched level are input into the comprehensive influence degree evaluation model of the power flow information to evaluate the comprehensive influence degree, that is, the comprehensive influence degree of the power flow information under the current information framework is calculated according to the risk confidence of each layer information in the hierarchical framework, as follows:

[0014]

[0015] wherein, is the score of the current information framework E on the power flow information xi, and infR e,xi is an evaluation value of the e-th level data of the current information framework E, and e is the e-th level of the information framework E.

[0016] As a preferred scheme of the present application, wherein: the comprehensive influence degree evaluation model of the power flow information is evaluated by using fuzzy evaluation method according to the generation power correlation characteristics of the input power flow information, and the evaluation set of the comprehensive influence degree evaluation of the power flow information is determined, the weight of the generation power correlation characteristics of the power flow information is determined, the fuzzy comprehensive evaluation matrix is constructed, the membership function is determined through expert scoring, the fuzzy evaluation matrix is formed, the characteristic weight fuzzy vector on the comprehensive influence degree evaluation factor set of the power flow information is converted into the fuzzy vector on the evaluation set of the comprehensive influence degree evaluation of the power flow information through fuzzy transformation, and the evaluation value of the power flow information is obtained.

[0017] As a preferred scheme of the present application, wherein: the power flow information correction model is constructed based on Newton-Raphson method to complete correction, and is expressed as follows:

[0018]

[0019] Wherein, S ij represents the transmission power from node i to node j, P ij represents the active power from node i to node j, Q i represents the reactive power from node i to node j, and * represents conjugate, represents the ground admittance of node i, represents the branch admittance between node i and node j, V i represents the voltage of the i-th node, represents the vector value of the voltage of the i-th node, represents the vector value of the voltage of the j-th node.

[0020] As a preferred scheme of the present application, wherein: the power flow information real-time correction module inputs the basic parameters of the nodes, branches and generators of the new energy power generation system to form an admittance matrix;

[0021] Suppose that the system has n nodes and m PQ nodes, because there is only one balanced node, so there are n-m-1 PV nodes, for all PQ nodes and PV nodes, the active power and reactive power imbalance equation is established, and then the Jacobian matrix is solved, the correction equation is solved, the voltage of each node is added to the correction amount, and the iteration processing is returned.

[0022] As a preferred scheme of the present application, wherein: the numerical environmental prediction of the new energy power generation of the regional meteorological analysis module is analyzed and calculated by using regression algorithm to obtain the numerical environmental result of the new energy power generation, as follows:

[0023] P = a0 + a1b1 + a2b2 +... + a m bm + epsilon;

[0024] Wherein P is a prediction probability value, that is, a new energy power generation environment numerical result, alpha1 to alpha m are coefficients of regional meteorological analysis linear regression model respectively, the coefficients are obtained by training a historical data set, beta1 to beta m are feature vectors of power generation power associated features, and epsilon is an error term.

[0025] As a preferred scheme of the application, wherein: the fusion prediction module adopts a covariance cross fusion method to combine the new energy power generation environment numerical result and the power flow power information correction result together, outputs a state estimation value after fusion, and determines as a fusion prediction result of the new energy power generation power; as follows:

[0026]

[0027] Wherein, R ab is a state estimation value after fusion, n is the total number of the ith node, w n is a weight value of the node n, and P ab is a fusion estimation variance value, P n is a local estimation variance of the node n, R n is a fusion value of the node n.

[0028] As a preferred scheme of the application, wherein: when generating the fusion prediction result of the new energy power generation power and outputting, a rolling prediction method is also adopted, and a deep belief network predict function is called to predict future power data.

[0029] As a preferred scheme of the application, wherein: further comprising:

[0030] A new energy power generation power knowledge graph construction module is used to receive the power generation power associated feature words of the natural language processing module under given initial conditions and boundary conditions, associate the new energy power generation scheduling scheme according to the power generation power associated feature words, and establish a new energy power generation scheduling scheme knowledge graph.

[0031] A new energy power generation power scheduling module is used to receive the fusion prediction result of the new energy power generation power, match the new energy power generation power adjustment scheme in the new energy power generation scheduling scheme knowledge graph, and perform visual presentation output.

[0032] The application has the beneficial effects that: the application has a multi-level multi-target influence event judgment function in the process of fusing and predicting the regional meteorological information after collecting and correcting the real-time power flow information of the wind power grid in real time, and the prediction data is more rich and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0034] Figure 1 Schematic diagram of the modular structure of the system of the present invention;

[0035] Figure 2 This is an analysis flow chart of the real-time correction module of the current information in the system of the present invention;

[0036] Figure 3 It is a modular schematic diagram of the wind power ramping power correlation characteristics of the present invention. DETAILED DESCRIPTION

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of the present invention.

[0038] Reference Figures 1-3 , is an embodiment of the present invention, which provides a new energy power generation prediction system based on deep learning, including:

[0039] The wind power ramping framework construction module is used to build an information hierarchical framework for renewable energy power generation, divide and configure renewable energy power generation prediction levels and matching power generation related feature words, and generate information frameworks for renewable energy related data at different levels and targets based on the renewable energy power generation prediction levels and matching power generation related feature words;

[0040] The natural language processing module is used to receive power generation prediction requests sent by new energy power stations, perform semantic analysis on the power generation prediction requests based on the pre-trained natural language processing model, and obtain the corresponding information hierarchy framework and matching power generation related feature words;

[0041] like Figure 3 As shown, it is further explained that the wind power ramping power generation related features include wind speed measurement value, wind vector measurement value, temperature measurement value, humidity measurement value, etc.;

[0042] The tide flow information real-time correction module is configured to determine tide flow power information of new energy power generation and information framework information of a matched level in response to the information framework of the wind power climbing framework modeling module and each power generation power associated feature word processed by the natural language processing module, analyze a comprehensive influence degree of the tide flow power information under a current information framework according to the tide flow power information and the information framework information of the matched level, judge whether the tide flow power information is to be corrected according to the comprehensive influence degree of the tide flow power information under the current level framework, complete correction based on a Newton-Raphson method to build a tide flow power information correction model, and generate a tide flow power information correction result, as shown in Figure 2

[0043] The regional meteorological analysis module is configured to obtain regional meteorological information under given initial conditions and boundary conditions and receive power generation power associated feature words of the natural language processing module, generate new energy power generation environment numerical results by performing numerical environment prediction of new energy power generation according to the power generation power associated feature words and in combination with the obtained regional meteorological information.

[0044] The fusion prediction module is configured to receive the new energy power generation environment numerical results and the tide flow power information correction result, perform fusion prediction of new energy power generation power, generate a fusion prediction result of the new energy power generation power, and output the fusion prediction result.

[0045] Based on the above, the embodiment first constructs an information framework of wind power climbing event prediction based on natural language processing, performs feature processing on tide flow information of wind power generation power, matches corresponding level frameworks, analyzes a comprehensive influence degree of tide flow power information under a current information framework according to the tide flow power information and the information framework information of the matched level, and thus not only introduces influences of active and reactive tide flow power data at different nodes on wind power climbing events, but also can correct tide flow power information in real time when encountering wind power climbing, and combine the corrected tide flow power information with new energy power generation environment numerical results after regional meteorological analysis to perform fusion prediction, thereby forming a more comprehensive prediction model input, ensuring data accuracy and reliability, and being beneficial to prediction and judgment of new energy power generation in the face of wind power climbing events.

[0046] Specifically, the tide flow information real-time correction module analyzes a comprehensive influence degree of tide flow power information under a current information framework according to tide flow power information and information framework information of a matched level, and specifically performs evaluation of the comprehensive influence degree by constructing a comprehensive influence degree evaluation model of the tide flow power information.

[0047] The tide flow power information and the information framework information of the matched level are input into the comprehensive influence degree evaluation model of the tide flow power information to evaluate the comprehensive influence degree, that is, the comprehensive influence degree of the tide flow power information under the current information framework is calculated according to a risk confidence degree of each layer of information in the level framework, as follows: ​

[0048]

[0049] wherein, is the score of the current information framework E on the power flow information xi, infR e,xi is the evaluation value of the current information framework e-th level data, e is the e-th level of the information framework E.

[0050] The embodiment needs to emphasize that the comprehensive influence degree evaluation model of the power flow information is evaluated by using the fuzzy evaluation method according to the power generation power correlation characteristics of the input power flow information. Specifically, a factor set for evaluating the comprehensive influence degree of the power flow information is established, an evaluation set for evaluating the comprehensive influence degree of the power flow information is determined, the weight of the power generation power correlation characteristics of the power flow information is determined, a fuzzy comprehensive evaluation matrix is constructed, and after the membership function is determined through expert scoring, a fuzzy evaluation matrix is formed. By fuzzy transformation, each feature weight fuzzy vector on the comprehensive influence degree evaluation factor set of the power flow information is converted into a fuzzy vector on the evaluation set of the comprehensive influence degree evaluation of the power flow information, and the evaluation value of the power flow information is obtained.

[0051] The embodiment specifically constructs a power flow information correction model based on the Newton-Raphson method to complete the correction, which is represented as follows:

[0052]

[0053] wherein, S ij represents the transmission power from node i to node j, P ij represents the active power from node i to node j, Q i represents the reactive power from node i to node j, and * represents conjugate, represents the ground admittance of node i, represents the branch admittance between node i and node j, V i represents the voltage of the i-th node, represents the vector value of the voltage of the i-th node, represents the vector value of the voltage of the j-th node.

[0054] The embodiment specifically inputs the basic parameters of the nodes, branches, and generators of the new energy power generation system into the power flow information real-time correction module to form an admittance matrix;

[0055] Suppose that the system has n nodes and m PQ nodes. Since there is only one balanced node, there are n-m-1 PV nodes. For all PQ nodes and PV nodes, the unbalanced amount equations of active power and reactive power are established. Then, the Jacobian matrix is solved, the correction equation is solved, the voltage of each node is added to the correction amount, and the iteration process is continued.

[0056] The numerical environment prediction of new energy power generation of the regional weather analysis module is preferably analyzed and calculated by using a regression algorithm to obtain new energy power generation environment numerical results, as follows:

[0057] P = a0 + a1b1 + a2b2 +... + a m b m + e;

[0058] Wherein, P is a predicted probability value, i.e. the new energy power generation environment numerical result, a1 to a m are coefficients of the regional weather analysis linear regression model, the coefficients are obtained by training the historical data set, b1 to b m are feature vectors of the features associated with the power generation, and e is an error term.

[0059] Specifically, the fusion prediction module uses a covariance cross-fusion method to combine the new energy power generation environment numerical results and the power flow information correction results together, outputs the state estimation value after fusion, and determines the fusion prediction result of the new energy power generation power; as follows:

[0060]

[0061] Wherein, R ab is the state estimation value after fusion, n is the total number of the ith node, w n is the weight value of the node n, and P ab is the estimated variance value after fusion, P n is the local estimated variance of the node n, and R n is the fusion value of the node n.

[0062] Specifically, when generating the fusion prediction result of the new energy power generation power and outputting, a rolling prediction method is used, and a deep belief network predict function is called to predict future power data. Further, in the process of wind power data processing, wavelet transform can extract time domain and frequency domain information of time series data, which helps to capture the mutation of wind power, which is a key feature of wind power climbing event. Generally, a deep belief network (DBN) is used, which can construct an accurate wind power climbing event prediction model through feature adaptive selection and data wavelet decomposition.

[0063] In order to realize the corresponding new energy power prediction after the analysis and processing based on deep learning, subsequent scheduling and control operations are often required, which provides auxiliary reference for power personnel, therefore, the new energy power prediction system based on deep learning of the embodiment further comprises:

[0064] The new energy power generation power knowledge graph construction module is configured to receive the power generation power associated feature words from the natural language processing module under given initial conditions and boundary conditions, and to associate the power generation power associated feature words with the new energy power generation scheduling scheme to establish a new energy power generation scheduling scheme knowledge graph.

[0065] The new energy power generation power scheduling module is configured to receive the fusion prediction result of the new energy power generation power, match the new energy power generation power adjustment scheme in the new energy power generation scheduling scheme knowledge graph, and perform visual presentation output.

[0066] Based on the above, the present application collects real-time power flow information of wind power grid and performs real-time correction, and fuses with regional meteorological information to have multi-level and multi-target influence event judgment function. In this process, the wind power data is corrected in real time, which improves the richness and accuracy of the prediction data. Specifically, the present application relates to the following several key functions and effects, as follows:

[0067] Real-time power flow information collection: first, the wind power generation power data of the wind power grid needs to be collected in real time, which is the basis for accurate prediction.

[0068] Real-time data correction: the collected data will be corrected in real time to ensure the accuracy and reliability of the data.

[0069] Regional meteorological information fusion: the corrected wind power data is fused with regional meteorological information to form a more comprehensive prediction model input.

[0070] Multi-level and multi-target influence event judgment: in this process, the invention has the function of judging different levels and different target influence events, which helps to analyze and predict the change of wind power from multiple angles.

[0071] Richness and accuracy of prediction data: through the above steps, the final prediction data not only has rich content, but also has high accuracy, which provides strong support for the scheduling and operation of the power grid.

[0072] This data-driven and meteorological information fusion-based prediction method can better cope with the volatility and uncertainty of wind power, and provides important decision basis for the stable operation and optimal scheduling of the power grid. Through real-time correction and multi-level judgment, the present application can effectively improve the accuracy of wind power prediction and reduce the risk of power grid operation.

[0073] In the above-described embodiments, all or part can be implemented by software, hardware, firmware, or any combination thereof. When implemented by software, all or part can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded on a computer, all or part generates a flow or function according to the present application. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium.

[0074] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. Also, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or more embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the present specification and the features of the different embodiments or examples without contradiction.

[0075] In addition, the terms "first", "second", etc. are used only for the purpose of description and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly specified.

[0076] Any process or method descriptions in flow charts or described elsewhere herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing specific logic functions or other processes. Also, the preferred embodiments of the present application can include additional implementation with respect to the order in which steps are performed, including a substantially simultaneous performance of the functions according to the functionality involved, or a performance of the functions in reverse order, depending upon the functionality involved.

[0077] The logic and / or steps represented in flow charts or otherwise described herein, for example, can be considered as a list of executable instructions for implementing the logic function, which can be specifically embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor, or other system that can fetch the instructions from the instruction execution system, apparatus or device and execute the instructions, or in conjunction with these instructions execution system, apparatus or device.

[0078] It should be understood that each part of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above 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. All or part of the steps of the above-mentioned embodiment method can be completed by a program instructing the relevant hardware, which can be stored in a computer readable storage medium, and the program includes one or a combination of the steps of the method embodiment when executed.

[0079] In addition, each functional unit in each embodiment of the present application can be integrated in one processing module, or each unit can be physically present separately, or two or more units can be integrated in one module. The above-mentioned integrated module can be realized in the form of hardware or in the form of a software functional module. The above-mentioned integrated module, if realized in the form of a software functional module and sold or used as an independent product, can also be stored in a computer readable storage medium. The storage medium can be a read-only memory, a magnetic disk or an optical disk, etc.

[0080] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of various changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A new energy power generation prediction system based on deep learning, characterized in that: include: The wind power ramping framework construction module is used to build an information hierarchical framework for renewable energy power generation, divide and configure renewable energy power generation prediction levels and matching power generation related feature words, and generate information frameworks for renewable energy related data at different levels and targets based on the renewable energy power generation prediction levels and matching power generation related feature words; A natural language processing module is used to receive a power generation prediction request sent by a new energy power station, perform semantic analysis on the power generation prediction request based on a pre-trained natural language processing model, and obtain a corresponding information hierarchy framework and matching power generation-related feature words; The real-time correction module of the current information is used to respond to the information framework of the wind power ramping framework construction module and the various power generation related feature words processed by the natural language processing module, determine the current power information of the new energy power generation and the information framework information of the matching level, and analyze the comprehensive impact of the current power information under the current information framework based on the current power information and the information framework information of the matching level. Specifically, a comprehensive impact assessment model of the current power information is constructed; the current power information and the information framework information of the matching level are input into the comprehensive impact assessment model of the current power information to perform a comprehensive impact assessment, that is, according to the risk confidence of each layer of information in the hierarchical framework, the comprehensive impact of the current power information under the current information framework is calculated as follows: in, is the score of the current information framework E on the power flow information xi, infR e,xi is the evaluation value of the data at level e of the current information framework, where e is the e-th level of information framework E; According to the comprehensive influence of the power flow information under the current hierarchical framework, it is judged whether the power flow information needs to be corrected. A power flow information correction model is constructed based on the Newton-Raphson method to complete the correction, and a power flow information correction result is generated. The power flow information correction model is constructed based on the Newton-Raphson method to complete the correction, which is expressed as follows: Among them, S ij Expressed as the transmission power from node i to node j, P ij represents the active power from node i to node j, Q i represents the reactive power from node i to node j, * represents conjugate, represents the ground admittance of node i, represents the branch admittance between node i and node j, V i represents the voltage of the i-th node, represents the vector value of the voltage at the i-th node, represents the vector value of the voltage at the jth node; The real-time power flow information correction module inputs the basic parameters of the nodes, branches, and generators of the renewable energy power generation system to form an admittance matrix. Assuming that the system has a total of n nodes and m PQ nodes, and because there is only one balancing node, there are a total of nm-1 PV nodes. For all PQ nodes and PV nodes, an unbalanced equation for active power and reactive power is established. Then, based on the Jacobian matrix, after solving the correction equation, the voltage of each node is added with the correction value, and the process is returned to continue the iterative process. A regional meteorological analysis module is used to obtain regional meteorological information under given initial and boundary conditions and receive power generation-related feature words from the natural language processing module. Based on the power generation-related feature words and the obtained regional meteorological information, the module performs a numerical prediction of the environment for new energy power generation and generates a numerical result of the new energy power generation environment. The fusion prediction module is used to receive the numerical results of the new energy power generation environment and the correction results of the current power information, perform fusion prediction of the new energy power generation power, generate and output the fusion prediction results of the new energy power generation power.

2. A new energy power generation prediction system based on deep learning according to claim 1, characterized in that: The comprehensive impact assessment model of the flow power information adopts a fuzzy evaluation method to perform assessment based on the power generation correlation characteristics of the input flow power information. Specifically, a factor set for the comprehensive impact assessment of the flow power information is established, an evaluation set for the comprehensive impact assessment of the flow power information is determined, and weights of the power generation correlation characteristics of the flow power information are determined. A fuzzy comprehensive evaluation matrix is ​​constructed and the membership function is determined by expert scoring to form a fuzzy evaluation matrix. The fuzzy transformation is used to convert the fuzzy vectors of the characteristic weights on the comprehensive impact assessment factor set of the flow power information into fuzzy vectors on the comprehensive impact assessment evaluation set of the flow power information to obtain an assessment value of the flow power information.

3. The deep learning-based new energy power generation prediction system according to claim 1, characterized in that: The numerical environment prediction of new energy power generation in the regional meteorological analysis module uses a regression algorithm to analyze and calculate the numerical results of the new energy power generation environment, as follows: P=α0+α1β1+α2β2+...+α m b m +e; Where P is the predicted probability value, that is, the numerical result of the new energy power generation environment, α1 to α m are the coefficients of the regional meteorological analysis linear regression model, which are obtained by training the historical data set, β1 to β m is the eigenvector of the characteristics associated with the generated power, and ε is the error term.

4. The deep learning-based new energy power generation prediction system according to claim 1, characterized in that: The fusion prediction module performs fusion prediction of new energy power generation by combining the numerical results of the new energy power generation environment and the correction results of the power flow information using the covariance cross fusion method, outputs the estimated value of the fusion state, and determines it as the fusion prediction result of the new energy power generation; as follows: Among them, R ab is the estimated value of the state after the fusion state, n is the total number of nodes i, w n is the weight value of node n, and P ab is the estimated variance value after fusion, P n is the local estimated variance of node n, R n is the fusion value of node n.

5. The deep learning-based new energy power generation prediction system according to claim 1, characterized in that: When generating and outputting the fusion prediction results of renewable energy power generation, it also includes using the rolling prediction method and calling the deep belief network predict function to predict future power data.

6. The deep learning-based new energy power generation prediction system according to claim 1, characterized in that: Also includes: A new energy power generation knowledge graph construction module is used to receive the power generation related feature words from the natural language processing module under given initial conditions and boundary conditions, and associate the power generation related feature words with the new energy power generation scheduling plan to establish a new energy power generation scheduling plan knowledge graph; The new energy power generation scheduling module is used to receive the fusion prediction results of the new energy power generation, match the new energy power generation adjustment plan in the new energy power generation scheduling plan knowledge graph, and perform visual presentation and output.

Images