Offshore wind power output prediction and correction method system based on k-shape clustering

Abstract

The present invention discloses a method for predicting and correcting offshore wind power output based on k-shape clustering, which comprises the following steps: acquiring historical data of offshore wind turbines and performing abnormal data preprocessing; performing real-time data processing on the real-time data of the offshore wind turbines based on an LSTM network model to obtain a real-time prediction value; performing data fusion processing on the real-time prediction value and the historical abnormal data preprocessing result to obtain wind turbine operation data; performing k-shape clustering grouping processing on the wind turbine operation data to obtain the cluster center of each group; comparing the wind turbine operation data with the cluster center of each group, and correcting the data that deviates too much from the cluster center; thus, the offshore wind power cluster prediction and correction algorithm based on k-shape clustering is realized with high accuracy, flexibility and real-time performance, thereby providing strong support and guarantee for the management and operation of offshore wind power clusters.

Description

Technical Field

[0001] The present invention relates to the technical field of offshore wind power output prediction, and in particular to an offshore wind power output prediction correction method system based on k-shape clustering. Background Art

[0002] As global demand for renewable energy continues to grow, offshore wind power is becoming an increasingly popular clean energy source. However, due to natural factors such as weather and the marine environment, offshore wind power output is difficult to predict, posing challenges to the security and stability of power systems. Traditional wind speed prediction models have limitations, such as assuming a uniform and stable wind speed distribution and ignoring the distance between wind turbines. This results in low accuracy in predicting wind turbine output. Summary of the Invention

[0003] In order to overcome the shortcomings of the existing technology, the present invention provides a method system for predicting and correcting offshore wind power output based on k-shape clustering, which realizes that the offshore wind power cluster prediction and correction algorithm based on k-shape clustering has high accuracy, flexibility and real-time performance, providing strong support and guarantee for the management and operation of offshore wind power clusters.

[0004] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions:

[0005] The first aspect of the present application provides a method for correcting offshore wind power output prediction based on k-shape clustering, comprising the following steps:

[0006] S101. Obtain historical data of offshore wind turbines and perform abnormal data preprocessing;

[0007] S102, performing real-time data processing on the offshore wind turbine real-time data based on the LSTM network model to obtain a real-time prediction value;

[0008] S103, performing data fusion processing on the real-time prediction value and the historical abnormal data preprocessing result to obtain the wind turbine operation data;

[0009] S104, performing k-shape clustering on the wind turbine operating data to obtain the cluster center of each group;

[0010] S105 , comparing the wind turbine operating data with the cluster center of each group, and correcting the data that deviates too much from the cluster center.

[0011] Furthermore, abnormal data preprocessing includes missing value processing, outlier processing, duplicate value processing...

[0012] Furthermore, obtaining historical data of offshore wind turbines and performing abnormal data preprocessing includes the following steps:

[0013] Obtain historical data of offshore wind turbines;

[0014] Process missing values ​​in the historical data of offshore wind turbines;

[0015] or / and, performing outlier processing on historical data of offshore wind turbines;

[0016] or / and, performing duplicate value processing on historical data of offshore wind turbines;

[0017] The historical data of the wind turbine units that have been processed for missing values, outliers, and duplicate values ​​are aligned using a unit time scale and the same initial moment.

[0018] Furthermore, the structure of the LSTM network model includes an input layer, a hidden layer, and an output layer. The input layer receives the corresponding real-time data and transmits it to the hidden layer. The hidden layer processes the data through an activation function, and the output layer converts the processed data into a predicted value through linear calculation.

[0019] Furthermore, the real-time data of offshore wind turbines is processed based on the LSTM network model to obtain real-time prediction values, which includes the following steps:

[0020] Obtain real-time data from offshore wind turbines as input data for the LSTM network model;

[0021] The input layer receives the corresponding real-time data and transmits it to the hidden layer;

[0022] The hidden layer processes the data through the activation function;

[0023] The output layer converts the processed data into predicted values ​​through linear calculations.

[0024] Furthermore, the k-shape clustering grouping process includes random initial grouping process, intra-cluster distance calculation process, cluster center calculation process, and re-iteration process.

[0025] Furthermore, the wind turbine operating data is subjected to k-shape clustering and grouping processing to obtain the cluster center of each group, which includes the following steps:

[0026] For the reference time series and each other time series in the cluster, align their positions;

[0027] Select the reference time series as the benchmark and calculate a lower bound and an upper bound. The lower bound is formed by the Euclidean distance between subsequences with the same relative position, and the upper bound is obtained by merging the optimal solutions of compressing each time series to the same size.

[0028] The distance between other time series in each cluster and the reference time series is calculated, and the subsequence with the smallest distance is selected as the subsequence representing the time series.

[0029] Furthermore, comparing the wind turbine operating data with the cluster center of each group and correcting the data that deviates too much from the cluster center includes the following steps:

[0030] S1: Power prediction is performed for each wind turbine to obtain a set of [y1, y2, y3…y i ];

[0031] S2: Use the k-shape method to cluster the wind turbines in the same group again and calculate the cluster centers to obtain the cluster centers of K clusters [μk1, μk2, μk3...μk j ];

[0032] S3: Mark the points whose deviation from the cluster center reaches γ = 0.2 as outliers P, satisfying the formula:

[0033]

[0034] Where j is the cluster number, i is the fan number belonging to cluster j;

[0035] S4: Add an offset β to the predicted value of point P to make it closer to the cluster center, recorded as P1, and satisfy:

[0036]

[0037] Where j is the cluster number, i is the number of the wind turbine belonging to cluster j, and β is the offset.

[0038] The second aspect of the present application provides a k-shape clustering-based offshore wind power output prediction and correction system, comprising:

[0039] A data acquisition unit, used to acquire historical data and real-time data of offshore wind turbines;

[0040] The first processing unit is used to perform abnormal data preprocessing on the historical data of the offshore wind turbine;

[0041] The second processing unit is used to perform real-time data processing on the real-time data of the offshore wind turbine generator system based on the LSTM network model to obtain a real-time prediction value;

[0042] The third processing unit is used to perform data fusion processing on the real-time prediction value and the historical abnormal data preprocessing result to obtain the wind turbine operation data;

[0043] The fourth processing unit is used to perform k-shape clustering and grouping processing on the wind turbine operating data to obtain the cluster center of each group;

[0044] The fifth processing unit is configured to compare the wind turbine operating data with the cluster center of each group, and perform correction processing on the data that deviates too much from the cluster center.

[0045] Furthermore, the first processing unit is configured to perform abnormal data preprocessing on the historical data of the offshore wind turbine generator system, including the following steps:

[0046] Obtain historical data of offshore wind turbines;

[0047] Process missing values ​​in the historical data of offshore wind turbines;

[0048] or / and, performing outlier processing on historical data of offshore wind turbines;

[0049] or / and, performing duplicate value processing on historical data of offshore wind turbines;

[0050] The historical data of the wind turbine units that have been processed for missing values, outliers, and duplicate values ​​are aligned using a unit time scale and the same initial moment.

[0051] Furthermore, the second processing unit is configured to perform real-time data processing on the real-time data of the offshore wind turbine generator system based on the LSTM network model to obtain a real-time prediction value, including the following steps:

[0052] Obtain real-time data from offshore wind turbines as input data for the LSTM network model;

[0053] The input layer receives the corresponding real-time data and transmits it to the hidden layer;

[0054] The hidden layer processes the data through the activation function;

[0055] The output layer converts the processed data into predicted values ​​through linear calculations.

[0056] Furthermore, the fourth processing unit is configured to perform k-shape clustering on the wind turbine operating data to obtain the cluster center of each group, including the following steps:

[0057] For the reference time series and each other time series in the cluster, align their positions;

[0058] Select the reference time series as the benchmark and calculate a lower bound and an upper bound. The lower bound is formed by the Euclidean distance between subsequences with the same relative position, and the upper bound is obtained by merging the optimal solutions of compressing each time series to the same size.

[0059] The distance between other time series in each cluster and the reference time series is calculated, and the subsequence with the smallest distance is selected as the subsequence representing the time series.

[0060] Furthermore, the fifth processing unit is configured to compare the wind turbine operating data with the cluster center of each group and perform correction processing on the data that deviates too much from the cluster center, including the following steps:

[0061] S1: Power prediction is performed for each wind turbine to obtain a set of [y1, y2, y3…y i ];

[0062] S2: Use the k-shape method to cluster the wind turbines in the same group again and calculate the cluster centers to obtain the cluster centers of K clusters [μk1, μk2, μk3...μk j ];

[0063] S3: Mark the points whose deviation from the cluster center reaches γ = 0.2 as outliers P, satisfying the formula:

[0064]

[0065] Where j is the cluster number, i is the fan number belonging to cluster j;

[0066] S4: Add an offset β to the predicted value of point P to make it closer to the cluster center, recorded as P1, and satisfy:

[0067]

[0068] Where j is the cluster number, i is the number of the wind turbine belonging to cluster j, and β is the offset.

[0069] The beneficial effects of the present application are as follows: by obtaining the historical data of each wind turbine, using the LSTM network model for prediction, performing k-shape clustering on the historical data and the predicted data respectively, and obtaining the corresponding cluster grouping and the cluster center of each group, comparing the predicted data with the cluster center of the predicted data, obtaining the relative error between the predicted data and the predicted cluster center, and correcting the predicted data with large errors; realizing the offshore wind power cluster prediction and correction algorithm based on k-shape clustering with high accuracy, flexibility and real-time performance, providing strong support and guarantee for the management and operation of offshore wind power clusters.

[0070] The algorithm based on k-shape clustering can more accurately cluster the historical data of offshore wind farms, thereby finding more appropriate cluster groupings. By correcting the prediction results, the prediction results can be corrected more accurately, improving the prediction accuracy.

[0071] The k-shape clustering algorithm incorporates historical data from multiple wind turbines, allowing for prediction corrections based on different groups, improving both flexibility and accuracy. Due to the algorithm's high computational speed, predictions can be completed in a shorter timeframe, improving real-time performance and efficiency.

[0072] Compared with the existing cluster prediction method, the offshore wind power cluster prediction correction algorithm based on k-shape clustering enhances the correlation between multiple wind turbines, utilizes the similarity of wind turbines under the same operating conditions, and improves the overall prediction accuracy by correcting the prediction data with excessive prediction errors. It is suitable for situations with multiple wind turbines in offshore wind farms. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0074] Figure 1 Schematic diagram of the steps of the offshore wind power output prediction and correction method based on k-shape clustering of the present invention;

[0075] Figure 2 This is a schematic diagram of the k-shape clustering algorithm flow of the k-shape clustering offshore wind power output prediction and correction method of the present invention;

[0076] Figure 3 It is a schematic diagram comparing correction processing results of the offshore wind power output prediction correction method based on k-shape clustering of the present invention. DETAILED DESCRIPTION

[0077] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0078] The following describes the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0079] Example 1:

[0080] The offshore wind power output prediction correction method based on k-shape clustering includes the following steps:

[0081] S101. Obtain historical data of offshore wind turbines and perform abnormal data preprocessing;

[0082] Obtain historical data for offshore wind turbines and perform abnormal data preprocessing on the data. This preprocessing includes missing value processing, outlier processing, and duplicate value processing, and obtains historical abnormal data preprocessing results. Align the historical abnormal data preprocessing results with the previously processed historical data for the wind turbines, using a unit time scale and the same initial time scale. Optionally, align the previously processed historical data for the wind turbines using a 15-minute unit time scale and the same initial time scale.

[0083] Obtaining historical data of offshore wind turbines and performing abnormal data preprocessing includes the following steps:

[0084] Obtain historical data of offshore wind turbines;

[0085] Process missing values ​​in the historical data of offshore wind turbines;

[0086] or / and, performing outlier processing on historical data of offshore wind turbines;

[0087] or / and, performing duplicate value processing on historical data of offshore wind turbines;

[0088] The historical data of the wind turbine units that have been processed for missing values, outliers, and duplicate values ​​are aligned using a unit time scale and the same initial moment.

[0089] S102, performing real-time data processing on the offshore wind turbine real-time data based on the LSTM network model to obtain a real-time prediction value;

[0090] Real-time data from offshore wind turbines is obtained and processed using an LSTM network model to obtain real-time predictions. Specifically, the LSTM network model processes the real-time data of each wind turbine in a unit time scale to obtain real-time predictions. For example, the unit time scale can be 15 minutes. The LSTM network model processes the real-time data of each wind turbine in a unit time scale of 15 minutes to obtain real-time predictions for each wind turbine.

[0091] The structure of the LSTM network model includes an input layer, a hidden layer, and an output layer. The input layer receives the corresponding real-time data and transmits it to the hidden layer. The hidden layer processes the data through an activation function, and the output layer converts the processed data into a predicted value through linear calculation.

[0092] Real-time data processing of offshore wind turbines based on the LSTM network model to obtain real-time prediction values ​​includes the following steps:

[0093] Obtain real-time data from offshore wind turbines as input data for the LSTM network model;

[0094] The input layer receives the corresponding real-time data and transmits it to the hidden layer;

[0095] The hidden layer processes the data through the activation function;

[0096] The output layer converts the processed data into predicted values ​​through linear calculations.

[0097] S103, performing data fusion processing on the real-time prediction value and the historical abnormal data preprocessing result to obtain the wind turbine operation data;

[0098] Historical data from offshore wind turbines is obtained and preprocessed for abnormalities. This includes missing value processing, outlier processing, and duplicate value processing, to obtain historical abnormal data preprocessing results. Real-time data from offshore wind turbines is obtained and processed using an LSTM network model to obtain real-time predictions. The real-time predictions are then fused with the historical abnormal data preprocessing results to obtain wind turbine operating data.

[0099] S104, performing k-shape clustering on the wind turbine operating data to obtain the cluster center of each group;

[0100] By fusing the real-time prediction value with the historical abnormal data preprocessing results, the wind turbine operation data is obtained. Figure 2 As shown, the wind turbine operating data is subjected to k-shape clustering grouping processing to obtain the cluster center of each group. The k-shape clustering grouping processing includes random initial grouping processing, intra-cluster distance calculation processing, cluster center calculation processing, and re-iteration processing.

[0101] Perform k-shape clustering on wind turbine operating data to obtain the cluster center of each group, including the following steps:

[0102] For the reference time series and each other time series in the cluster, align their positions;

[0103] Select the reference time series as the benchmark and calculate a lower bound and an upper bound. The lower bound is formed by the Euclidean distance between subsequences with the same relative position, and the upper bound is obtained by merging the optimal solutions of compressing each time series to the same size.

[0104] The distance between other time series in each cluster and the reference time series is calculated, and the subsequence with the smallest distance is selected as the subsequence representing the time series.

[0105] It should be noted that the random initial grouping is to randomly initialize the grouping of the calculated wind turbines, and randomly assign all wind turbines to each group while determining the number of clusters. The intra-cluster distance calculation mainly uses the subsequence boundary distance (SBD) for wind turbines in the same group. It is a method for calculating the similarity of time series. It quickly calculates the distance between two time series by quickly calculating the lower and upper boundaries of all subsequences of the time series, greatly improving the calculation speed, and can effectively deal with problems such as noise and outliers in the calculation of time series similarity. The cluster center calculation is based on the above intra-cluster distance calculation results, and all subsequences are averaged to obtain the center of the cluster. Re-iteration refers to the redistribution of time series data by comparing the similarity between the time series data within each cluster to obtain new cluster divisions, and repeating the intra-cluster distance calculation and cluster center calculation.

[0106] For example, by fusing real-time predictions with preprocessed historical anomaly data, wind turbine operating data is obtained. K-shape shape-based clustering is then used to generate clusters with a number of clusters between 3 and 5, and the absolute SBD value range is between 7.5% and 15%. The number of regrouping iterations is greater than or equal to 20 and less than or equal to 24. K-shape clustering is a clustering process that combines historical and predicted data. The number of clusters is determined by the elbow rule.

[0107] S105, comparing the wind turbine operating data with the cluster center of each group, and correcting the data that deviates too much from the cluster center;

[0108] By performing k-shape clustering on wind turbine operating data, the cluster center of each group is determined. By comparing the wind turbine operating data with the cluster center of each group, data that deviates excessively from the cluster center is corrected, resulting in a corrected wind turbine operating data result. For example, the cluster results obtained by k-shape clustering and the cluster centers are used to correct the predicted data. This involves comparing the predicted wind turbine data with the predicted data from the cluster center. Curves that deviate more than 26% from the cluster center require correction, and the corrected values ​​must have a deviation of no more than 9%.

[0109] like Figure 3 As shown, the wind turbine operating data is compared with the cluster center of each group, and the correction processing of the data that deviates too much from the cluster center includes the following steps:

[0110] S1: Power prediction is performed for each wind turbine to obtain a set of [y1, y2, y3…y i ];

[0111] S2: Use the k-shape method to cluster the wind turbines in the same group again and calculate the cluster centers to obtain the cluster centers of K clusters [μk1, μk2, μk3...μk j ];

[0112] S3: Mark the points whose deviation from the cluster center reaches γ = 0.2 as outliers P, satisfying the formula:

[0113]

[0114] Where j is the cluster number, i is the fan number belonging to cluster j;

[0115] S4: Add an offset β to the predicted value of point P to make it closer to the cluster center, recorded as P1, and satisfy:

[0116]

[0117] Where j is the cluster number, i is the number of the wind turbine belonging to cluster j, and β is the offset.

[0118] The above is the offshore wind power output prediction and correction method based on k-shape clustering provided in the embodiment of this application, and the following is the offshore wind power output prediction and correction system based on k-shape clustering provided in the embodiment of this application.

[0119] The offshore wind power output prediction and correction system based on k-shape clustering includes:

[0120] A data acquisition unit, used to acquire historical data and real-time data of offshore wind turbines;

[0121] The first processing unit is used to perform abnormal data preprocessing on the historical data of the offshore wind turbine;

[0122] The second processing unit is used to perform real-time data processing on the real-time data of the offshore wind turbine generator system based on the LSTM network model to obtain a real-time prediction value;

[0123] The third processing unit is used to perform data fusion processing on the real-time prediction value and the historical abnormal data preprocessing result to obtain the wind turbine operation data;

[0124] The fourth processing unit is used to perform k-shape clustering and grouping processing on the wind turbine operating data to obtain the cluster center of each group;

[0125] The fifth processing unit is configured to compare the wind turbine operating data with the cluster center of each group, and perform correction processing on the data that deviates too much from the cluster center.

[0126] The first processing unit is configured to perform abnormal data preprocessing on the historical data of the offshore wind turbines. The preprocessed historical abnormal data is then used to align the historical data of the wind turbines, which has undergone missing value processing, outlier processing, and duplicate value processing, with a unit time scale and a time scale at the same initial moment. Optionally, the wind turbine data, which has undergone missing value processing, outlier processing, and duplicate value processing, is aligned with a 15-minute unit time scale and a time scale at the same initial moment.

[0127] The abnormal data preprocessing of offshore wind turbine historical data includes the following steps:

[0128] Obtain historical data of offshore wind turbines;

[0129] Process missing values ​​in the historical data of offshore wind turbines;

[0130] or / and, performing outlier processing on historical data of offshore wind turbines;

[0131] or / and, performing duplicate value processing on historical data of offshore wind turbines;

[0132] The historical data of the wind turbine units that have been processed for missing values, outliers, and duplicate values ​​are aligned using a unit time scale and the same initial moment.

[0133] The second processing unit is configured to process the real-time data of the offshore wind turbines based on the LSTM network model to obtain real-time prediction values. The real-time data of each wind turbine is processed based on the LSTM network model at a unit time scale to obtain real-time prediction values. For example, the unit time scale may be 15 minutes. The real-time data of each wind turbine is processed based on the LSTM network model at a unit time scale of 15 minutes to obtain real-time prediction values ​​for each wind turbine.

[0134] The structure of the LSTM network model includes an input layer, a hidden layer, and an output layer. The input layer receives the corresponding real-time data and transmits it to the hidden layer. The hidden layer processes the data through an activation function, and the output layer converts the processed data into a predicted value through linear calculation.

[0135] Real-time data processing of offshore wind turbines based on the LSTM network model to obtain real-time prediction values ​​includes the following steps:

[0136] Obtain real-time data from offshore wind turbines as input data for the LSTM network model;

[0137] The input layer receives the corresponding real-time data and transmits it to the hidden layer;

[0138] The hidden layer processes the data through the activation function;

[0139] The output layer converts the processed data into predicted values ​​through linear calculations.

[0140] The fourth processing unit is configured to perform k-shape clustering on the wind turbine operating data to obtain the cluster center of each group. The k-shape clustering process includes random initial grouping, intra-cluster distance calculation, cluster center calculation, and re-iteration.

[0141] Perform k-shape clustering on wind turbine operating data to obtain the cluster center of each group, including the following steps:

[0142] For the reference time series and each other time series in the cluster, align their positions;

[0143] Select the reference time series as the benchmark and calculate a lower bound and an upper bound. The lower bound is formed by the Euclidean distance between subsequences with the same relative position, and the upper bound is obtained by merging the optimal solutions of compressing each time series to the same size.

[0144] The distance between other time series in each cluster and the reference time series is calculated, and the subsequence with the smallest distance is selected as the subsequence representing the time series.

[0145] For example, by fusing real-time predictions with preprocessed historical anomaly data, wind turbine operating data is obtained. K-shape shape-based clustering is then used to generate clusters with a number of clusters between 3 and 5, and the absolute SBD value range is between 7.5% and 15%. The number of regrouping iterations is greater than or equal to 20 and less than or equal to 24. K-shape clustering is a clustering process that combines historical and predicted data. The number of clusters is determined by the elbow rule.

[0146] The fifth processing unit is configured to compare the wind turbine operating data with the cluster center of each group and perform correction processing on data that deviates excessively from the cluster center. By comparing the wind turbine operating data with the cluster center of each group, correction processing is performed on data that deviates excessively from the cluster center, thereby obtaining a correction result for the wind turbine operating data. For example, the clustering results obtained by k-shape clustering and the cluster center are used to correct the predicted data. That is, the predicted data of the wind turbine group is compared with the predicted data of the cluster center. Curves that deviate more than 26% from the cluster center are identified as curves that require correction, and the deviation of the corrected value does not exceed 9%.

[0147] Comparing the wind turbine operating data with the cluster center of each group and correcting the data that deviates too much from the cluster center includes the following steps:

[0148] S1: Power prediction is performed for each wind turbine to obtain a set of [y1, y2, y3…y i ];

[0149] S2: Use the k-shape method to cluster the wind turbines in the same group again and calculate the cluster centers to obtain the cluster centers of K clusters [μk1, μk2, μk3...μk j ];

[0150] S3: Mark the points whose deviation from the cluster center reaches γ = 0.2 as outliers P, satisfying the formula:

[0151]

[0152] Where j is the cluster number, i is the fan number belonging to cluster j;

[0153] S4: Add an offset β to the predicted value of point P to make it closer to the cluster center, recorded as P1, and satisfy:

[0154]

[0155] Where j is the cluster number, i is the number of the wind turbine belonging to cluster j, and β is the offset.

[0156] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0157] The terms "first", "second" and "third" etc. in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can, for example, be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0158] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0159] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0160] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0161] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0162] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (English full name: Read-Only Memory, English abbreviation: ROM), random access memory (English full name: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store program code.

[0163] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for correcting offshore wind power output prediction based on k-shape clustering, characterized in that: The following steps are involved: S101. Obtain historical data of offshore wind turbines and perform abnormal data preprocessing; S102, performing real-time data processing on the offshore wind turbine real-time data based on the LSTM network model to obtain a real-time prediction value; S103, performing data fusion processing on the real-time prediction value and the historical abnormal data preprocessing result to obtain the wind turbine operation data; S104, performing k-shape clustering on the wind turbine operating data to obtain the cluster center of each group; comprising the following steps: For the reference time series and each other time series in the cluster, align their positions; Select the reference time series as the benchmark and calculate a lower bound and an upper bound. The lower bound is formed by the Euclidean distance between subsequences with the same relative position, and the upper bound is obtained by merging the optimal solutions of compressing each time series to the same size. Calculate the distance between other time series in each cluster and the reference time series, and select the subsequence with the smallest distance as the subsequence representing the time series; S105 , comparing the wind turbine operating data with the cluster center of each group, and correcting the data that deviates too much from the cluster center.

2. The offshore wind power output prediction and correction method based on k-shape clustering according to claim 1 is characterized in that: The acquisition of historical data of offshore wind turbines and preprocessing of abnormal data includes the following steps: Obtain historical data of offshore wind turbines; Process missing values ​​in the historical data of offshore wind turbines; or / and, performing outlier processing on historical data of offshore wind turbines; or / and, performing duplicate value processing on historical data of offshore wind turbines; The historical data of offshore wind turbines that have been processed for missing values, outliers, and duplicate values ​​are aligned using a unit time scale and the same initial moment.

3. The offshore wind power output prediction and correction method based on k-shape clustering according to claim 1 is characterized in that: The real-time data processing of the offshore wind turbine generator system based on the LSTM network model to obtain the real-time prediction value includes the following steps: Obtain real-time data from offshore wind turbines as input data for the LSTM network model; The input layer receives the corresponding real-time data and transmits it to the hidden layer; The hidden layer processes the data through the activation function; The output layer converts the processed data into predicted values ​​through linear calculations.

4. The offshore wind power output prediction and correction method based on k-shape clustering according to claim 1, characterized in that: Comparing the wind turbine operating data with the cluster center of each group and correcting the data that deviates too much from the cluster center includes the following steps: S1: Power prediction for each wind turbine is performed to obtain a set of 、 、 … ; S2: Use the k-shape method to cluster the wind turbines in the same group again and calculate the cluster centers to obtain the cluster centers of K clusters. 、 、 … ; S3: The proportion of deviation from the cluster center reaches The points are marked as outliers , satisfying the formula: in, is the cluster number, For The fan number of the cluster; S4: Yes The predicted value of the point plus the offset , so that it is closer to the cluster center and is recorded as , and satisfy: in, is the cluster number, For The fan number of the cluster, is the offset.

5. Offshore wind power output prediction and correction system based on k-shape clustering, characterized by: include: A data acquisition unit, used to acquire historical data and real-time data of offshore wind turbines; The first processing unit is used to perform abnormal data preprocessing on the historical data of the offshore wind turbine; The second processing unit is used to perform real-time data processing on the real-time data of the offshore wind turbine generator system based on the LSTM network model to obtain a real-time prediction value; The third processing unit is used to perform data fusion processing on the real-time prediction value and the historical abnormal data preprocessing result to obtain the wind turbine operation data; The fourth processing unit is used to perform k-shape clustering and grouping processing on the wind turbine operating data to obtain the cluster center of each group; The fourth processing unit performs k-shape clustering and grouping processing on the wind turbine operating data to obtain the cluster center of each group, including the following steps: For the reference time series and each other time series in the cluster, align their positions; Select the reference time series as the benchmark and calculate a lower bound and an upper bound. The lower bound is formed by the Euclidean distance between subsequences with the same relative position, and the upper bound is obtained by merging the optimal solutions of compressing each time series to the same size. Calculate the distance between other time series in each cluster and the reference time series, and select the subsequence with the smallest distance as the subsequence representing the time series; The fifth processing unit is configured to compare the wind turbine operating data with the cluster center of each group, and perform correction processing on the data that deviates too much from the cluster center.

6. The offshore wind power output prediction and correction system based on k-shape clustering according to claim 5 is characterized in that: The first processing unit preprocesses abnormal data of the historical data of the offshore wind turbine generator system, including the following steps: Obtain historical data of offshore wind turbines; Process missing values ​​in the historical data of offshore wind turbines; or / and, performing outlier processing on historical data of offshore wind turbines; or / and, performing duplicate value processing on historical data of offshore wind turbines; The historical data of offshore wind turbines that have been processed for missing values, outliers, and duplicate values ​​are aligned using a unit time scale and the same initial moment.

7. The offshore wind power output prediction and correction system based on k-shape clustering according to claim 5, characterized in that: The second processing unit performs real-time data processing on the real-time data of the offshore wind turbine generator system based on the LSTM network model to obtain a real-time prediction value, including the following steps: Obtain real-time data from offshore wind turbines as input data for the LSTM network model; The input layer receives the corresponding real-time data and transmits it to the hidden layer; The hidden layer processes the data through the activation function; The output layer converts the processed data into predicted values ​​through linear calculations.

8. The offshore wind power output prediction and correction system based on k-shape clustering according to claim 5, characterized in that: The fifth processing unit is configured to compare the wind turbine operating data with the cluster center of each group and perform correction processing on the data that deviates too much from the cluster center, including the following steps: S1: Power prediction for each wind turbine is performed to obtain a set of 、 、 … ; S2: Use the k-shape method to cluster the wind turbines in the same group again and calculate the cluster centers to obtain the cluster centers of K clusters. 、 、 … ; S3: The proportion of deviation from the cluster center reaches The points are marked as outliers , satisfying the formula: in, is the cluster number, For The fan number of the cluster; S4: Yes The predicted value of the point plus the offset , so that it is closer to the cluster center and is recorded as , and satisfy: in, is the cluster number, For The fan number of the cluster, is the offset.

Images