A monitoring and management method and system for wind power generation

By clustering and reducing the multi-dimensional environmental data of wind power generation equipment, removing unstable data, and combining PCA and LOF algorithms for abnormal detection, the accuracy of wind power generation equipment monitoring and management in complex environments is solved, and the operation of equipment is improved Stability and safety.

CN118713309BActive Publication Date: 2024-11-26HUBEI ZHONGKENENG ENERGY TECH
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202411196535.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-29
Publication Date
2024-11-26
Estimated Expiration
2044-08-29

AI Technical Summary

Technical Problem

The prior art is difficult to achieve accurate and timely monitoring and management of wind power equipment when facing complex environments, resulting in unstable operating parameters and affecting the safety and effective management of equipment.

Method used

By obtaining multi-dimensional environmental data of wind power equipment, clustering and dimensionality reduction, eliminating unstable environmental data, and using PCA algorithm and LOF algorithm to perform abnormal detection, improving the accuracy of monitoring and management.

Benefits of technology

It realizes stable monitoring and management of wind power equipment, reduces the interference of environmental changes on monitoring, improves the accuracy of abnormal detection and the operational reliability of wind power equipment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118713309B_ABST
    Figure CN118713309B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of electric power systems, and in particular to a monitoring and management method and system for wind power generation, the method comprising: obtaining power data of multiple wind power generation equipment and multi-dimensional environmental data corresponding to the power data; wherein the dimensions include wind direction, wind force, temperature, humidity and / or equipment orientation information; clustering all multi-dimensional environmental data to obtain multiple initial clusters; obtaining dimension reduction data of each initial cluster under the optimal dimension, and obtaining an adjustment cluster; calculating the similarity of each initial cluster and the corresponding adjustment cluster; performing abnormal detection on the power data corresponding to the initial cluster whose similarity is greater than or equal to a threshold, and issuing an early warning for abnormal wind power generation equipment. That is, the scheme of the present invention can accurately monitor and manage each wind power generation equipment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of electric power systems, and more specifically, to a monitoring and management method and system for wind power generation. Background Art

[0002] As the demand for renewable energy continues to increase, wind power generation, as a clean and sustainable form of energy, has received widespread attention and application. However, the operation and management of wind farms face a series of challenges, the most important of which is to ensure a stable supply of renewable energy and ensure the safety of workers.

[0003] At present, ensuring a stable supply of renewable energy includes abnormal monitoring of wind power equipment, prediction and optimization of power generation, and diversified energy reserves. Among them, the monitoring of wind power equipment generally involves directly collecting the operating parameters of wind power equipment, such as vibration data, current and voltage data, to monitor the wind power equipment.

[0004] In the related technology, such as the patent application document with publication number CN118297282A and titled "A wind power generation information integrated management system", it discloses that the monitoring and management of wind power generation settings are achieved by setting up a status monitoring module, an ecological monitoring module, an equipment aging assessment module, and a data analysis module; the data analysis module performs a comprehensive analysis of multi-source data to identify factors affecting wind power generation efficiency and operation strategies, identify potential equipment problems in advance, and perform maintenance in a timely manner.

[0005] In the above scheme, when facing dynamic and constantly changing wind power generation scenarios, it uses a particle swarm algorithm to optimize the operating parameters of the wind turbine in real time to ensure that the wind turbine reaches the optimal operating state. However, the particle swarm algorithm itself lacks dynamic adjustment of speed, is prone to fall into a local optimal solution, and has the problem of low algorithm efficiency. In this way, when facing instantaneous environmental changes, it is difficult to optimize the operating parameters of the wind turbine in a timely and accurate manner, that is, the operating parameters of the wind turbine are unstable, and it is impossible to ensure that the wind power generation equipment can operate stably and reliably in this environment. Therefore, it is impossible to perform safe and effective monitoring and management of the wind power generation equipment.

[0006] Therefore, it is particularly important to achieve effective monitoring and management of wind power generation equipment in a complex environment. Summary of the invention

[0007] The purpose of the present invention is to propose a wind power generation monitoring and management method and system to solve the problem that the prior art cannot achieve accurate and effective monitoring and management of wind power generation equipment due to the complexity of the environment; to this end, the present invention provides solutions in the following two aspects.

[0008] In a first aspect, the present invention provides a monitoring and management method for wind power generation, comprising:

[0009] Obtaining power data of multiple wind power generation equipment and multi-dimensional environmental data corresponding to the power data; wherein the dimensions include wind direction, temperature, humidity and / or equipment orientation information;

[0010] Cluster all multidimensional environmental data to obtain multiple initial clusters;

[0011] Obtain the dimension reduction data of each initial cluster at the optimal dimension and obtain the adjusted cluster;

[0012] Calculate the similarity of each initial cluster and the corresponding adjustment cluster; perform anomaly detection on the power data corresponding to the initial cluster whose similarity is greater than or equal to a threshold value, so as to provide an early warning of abnormal wind power generation equipment;

[0013] The optimal dimension is: using the PCA algorithm to obtain the principal components of all multidimensional environmental data; arranging and combining the principal components to obtain multiple combinations; constructing a combined objective function, and taking the combination corresponding to the maximum value of the objective function as the optimal dimension;

[0014] The objective function for: , is the maximum value of the Euclidean distance between any two projection data in the projection data of the kth initial cluster under the nth combination, K is the total number of initial clusters, is the minimum Euclidean distance between two projection data in any two initial clusters under the nth combination; is the maximum value function, is an exponential function with the natural number e as the base; the projection data is the dimension reduction data of each multidimensional environmental data under the corresponding combination.

[0015] In the above scheme, firstly, similar or identical environmental data in the multidimensional environmental data of multiple wind power generation equipment collected at different times can be classified into one category through the clustering method, and then the principal components are obtained by performing dimensionality reduction operations on all the multidimensional environmental data to obtain the best combination, and the dimensions of each initial cluster are adjusted using the best combination to make the obtained adjustment clusters more dispersed and the data in each adjustment cluster more concentrated; and then, by comparing the similarities of the clusters before and after the adjustment, the unstable environmental data in the initial cluster is eliminated to avoid the subsequent interference of abnormal detection of wind power generation equipment due to the unstable environment, thereby improving the accuracy of subsequent abnormal detection.

[0016] In one embodiment, the similarity is to obtain the class offset of any initial cluster and the class offset of the corresponding adjusted cluster, and use the modified cosine similarity method to calculate the similarity of the class offset of any initial cluster and the class offset of the corresponding adjusted cluster; wherein the class offset is the offset vector between the central data in the initial cluster or the adjusted cluster and all other remaining data.

[0017] In one embodiment, before obtaining the principal component, the method also includes performing a dimensionality increase operation on all multidimensional environmental data to obtain the increased dimensional environmental data, and using the PCA algorithm to reduce the dimensionality of all the increased dimensional environmental data to obtain the principal component.

[0018] The above scheme can ensure the linear separability of data by successively performing dimensionality increase and dimensionality reduction operations on all multidimensional environmental data.

[0019] In one embodiment, the dimension increasing operation adopts a polynomial mapping method.

[0020] In one embodiment, the method of arranging and combining the plurality of principal components is:

[0021] A set number of principal components is randomly selected from multiple principal components to form a combination, thereby obtaining multiple combinations; wherein the set number is less than or equal to the total number of principal components.

[0022] By arranging and combining a plurality of principal components, a plurality of combinations of principal components can be obtained, which facilitates subsequent screening of the best combination.

[0023] In one embodiment, the specific process of the anomaly detection is: statistically analyzing the power data corresponding to the same wind power generation equipment in the initial cluster whose similarity is greater than or equal to a threshold; and treating each statistical power data as a data point;

[0024] Calculate the local outlier factor for each data point;

[0025] When the local outlier factor of any data point is greater than the set value, the data point is an abnormal point; otherwise, the data point is a normal point; when the number of abnormal points is greater than the set threshold, the wind power generation equipment is abnormal and an early warning is issued.

[0026] In one embodiment, the LOF algorithm is used to perform anomaly detection on all power data in the initial cluster whose similarity is greater than or equal to a threshold, to obtain abnormal data, determine the wind power generation equipment corresponding to the abnormal data, and issue an early warning notification.

[0027] In one embodiment, before clustering, a step of normalizing all multi-dimensional environmental data is also included.

[0028] The above normalization process is to eliminate the dimensions of different dimensions to facilitate subsequent data analysis.

[0029] In a second aspect, the present invention provides a wind power generation monitoring and management system, comprising:

[0030] processor;

[0031] The memory stores computer instructions for monitoring and managing wind power generation. When the computer instructions are executed by the processor, the system executes the above-mentioned method for monitoring and managing wind power generation.

[0032] The beneficial effects of the present invention are:

[0033] The solution of the present invention monitors the changing data of the power generation environment of each wind power generation equipment, and can quickly classify different power generation environments through a clustering method, and analyze the global environmental data to obtain the best combination of dimensional adjustment, and then obtain the adjustment cluster corresponding to the initial cluster after clustering, so as to eliminate the local unstable power generation environment, reduce the interference of subsequent abnormal detection of wind power generation equipment, and improve the accuracy of monitoring of wind power generation equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0035] Figure 1 A flowchart schematically shows a method for monitoring and managing wind power generation in this embodiment;

[0036] Figure 2 The structural block diagram of a wind power generation monitoring and management system in this embodiment is schematically shown. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0038] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0039] like Figure 1 As shown, a wind power generation monitoring and management method in this embodiment includes the following steps:

[0040] In step S1, power data of multiple wind power generation equipment and multi-dimensional environmental data corresponding to the power data are obtained. Specifically, power data and environmental data of multiple wind power generation equipment in a monitoring area in a set time period are collected, wherein the environmental data includes data of multiple dimensions. Exemplarily, the dimensions include but are not limited to wind direction, temperature, humidity, wind force and / or equipment orientation information.

[0041] The set time period includes the current time and the time within a period of time before the current time. The multi-dimensional environmental data collected at different sampling times can constitute a data point, and then multiple data points of each wind power generation equipment in the set time period can be obtained.

[0042] The device orientation information may be the direction information of the wind power generation equipment installation, such as the east, west, south and north directions. Specifically, when the wind power generation equipment is oriented north, the direction is 0 degrees, and when the wind power generation equipment is oriented south, the direction is 180 degrees.

[0043] It should be noted that the environmental data of each wind power generation equipment is introduced in this embodiment because when the environment changes, the power data of some wind power generation equipment in the monitoring area may change, resulting in excessive differences in wind power generation in the same monitoring area. In the subsequent monitoring of the wind power generation equipment, the power generation in the time period corresponding to the changed environment may also change due to changes in the environment in which it is located, thereby interfering with the abnormal monitoring of each wind power generation equipment. Therefore, environmental data needs to be introduced in this embodiment.

[0044] In step S2, all multidimensional environmental data of all wind power generation equipment are clustered to obtain multiple initial clusters. Specifically, each initial cluster includes multiple data points, which can be multidimensional environmental data corresponding to multiple wind power generation equipment respectively; or multidimensional environmental data of the same wind power generation equipment at different sampling times; of course, it can also be multidimensional environmental data corresponding to multiple wind power generation equipment respectively and multidimensional environmental data of the same wind power generation equipment at different sampling times.

[0045] The clustering method in this embodiment adopts the mean shift algorithm. Specifically, when clustering, the clustering radius of mean shift clustering is 0.05 and the threshold is 0.1. The clustering radius and threshold can also be adjusted by the implementer according to the environmental differences of different environments.

[0046] Of course, K-means clustering algorithm can also be used. Since the above specific clustering method is an existing technology, it will not be introduced in detail here.

[0047] The purpose of classifying all multidimensional environmental data mentioned above is to classify the same or similar environmental data into one category, and the same or similar environmental data corresponds to wind power generation equipment. Therefore, this classification also classifies different wind power generation equipment in the same or similar environment into one category.

[0048] Furthermore, before clustering, all dimensional data need to be normalized to obtain normalized multidimensional data, wherein the normalization may be performed using a maximum normalization method.

[0049] In step S3, the dimension reduction data of each initial cluster under the optimal dimension is obtained, and an adjusted cluster is obtained. Specifically, the adjusted cluster in this embodiment includes the dimension reduction data of the corresponding initial cluster under the optimal dimension; the adjusted cluster can be considered to be obtained by adjusting the dimension of the initial cluster according to the optimal dimension.

[0050] The process of obtaining the optimal dimension in this embodiment is as follows:

[0051] The principal components of all multidimensional environmental data are obtained using the PCA (principal component analysis) algorithm, and the principal components are arranged and combined to obtain multiple combinations. The objective function of the combination is constructed, and the combination corresponding to the maximum value of the objective function is taken as the optimal dimension.

[0052] Among them, in this embodiment, all multidimensional environmental data are regarded as a group of data, and the dimensionality reduction processing is performed on the group of data to obtain the main components of the group of data, which provides a data basis for finding the best combination of all initial clusters, so that all subsequent initial clusters use the same best combination to make the obtained adjustment clusters more dispersed and the data within each adjustment cluster more concentrated.

[0053] Among them, the PCA algorithm is a data dimensionality reduction technology that converts the original data into a set of linearly related new variables, namely principal components, through orthogonal transformation. Specifically, the PCA algorithm calculates the covariance matrix of the data set; and performs eigenvalue decomposition on the covariance matrix to obtain eigenvalues ​​and eigenvectors; and sorts the eigenvalues, selects the first n (n is greater than 1) largest eigenvalues ​​and their corresponding eigenvectors; and forms a projection matrix with the selected eigenvectors in descending order of the corresponding eigenvalues; and multiplies the original data matrix by the projection matrix to obtain the reduced-dimensional data. Based on the specific process of the PCA algorithm, it can be seen that an eigenvector is actually a principal component; multiple principal components are multiple eigenvectors.

[0054] The combination in this embodiment is to randomly select a set number of principal components from multiple principal components to form a combination, thereby obtaining multiple combinations; wherein the set number is less than or equal to the total number of principal components. For example, when the total number of principal components is 5, the set number can be 1, 2, 3, 4 or 5, etc. When the set number is 2, there are 10 combinations of the principal components; when the set number is 1, there are 5 combinations of the principal components.

[0055] It should be noted that the combinations in this embodiment can be obtained by exhaustively enumerating all set numbers to obtain all combination methods; or a set number can be randomly selected to obtain a corresponding combination method.

[0056] The construction process of the objective function in this embodiment is:

[0057] Take any combination as the target combination and obtain the projection matrix of the target combination;

[0058] Multiply each initial cluster by the projection matrix to obtain projection data of each multidimensional environmental data in the initial cluster, wherein the projection data constitutes a cluster after dimensionality reduction;

[0059] Calculate the Euclidean distance between any two projected data in each initial cluster, and take the maximum value of the Euclidean distance as the intra-class maximum value of the corresponding initial cluster; randomly select two initial clusters, calculate the Euclidean distance between any two projected data in the two initial clusters, and take the minimum value of the Euclidean distance as the inter-class minimum value;

[0060] The objective function is constructed based on the inter-class minimum and the intra-class maximum of all initial clusters.

[0061] Specifically, the objective function The expression is: ; In the formula, is the maximum value within the class of the kth initial cluster under the nth combination, K is the total number of initial clusters, is the minimum Euclidean distance between any two projection data in any two initial clusters under the nth combination; is the maximum value function, is an exponential function with the natural number e as base.

[0062] The above formula Characterizes the dispersion of the data points of the kth initial cluster. The larger its value, the more dispersed the projection data distribution within the corresponding initial cluster. The smaller the value of , the smaller the overall dispersion of all initial clusters, and the better the dimensionality reduction effect corresponding to the principal component combination.

[0063] Based on the objective function constructed above, the optimal dimension of the principal component can be obtained, and the dimension reduction data of each initial cluster under the optimal dimension can be used as the corresponding adjustment cluster.

[0064] Furthermore, in order to prevent the data after dimensionality reduction from being unable to obtain dispersed classified data, this embodiment also includes a step of performing dimensionality increase operations on all multidimensional environmental data to obtain increased dimensionality environmental data before obtaining the principal components, and then reducing the dimensionality of all increased dimensional environmental data to obtain the principal components.

[0065] When obtaining the principal components, the number of principal components may be equal to the dimension after dimensionality increase.

[0066] The dimension-raising method may adopt a polynomial mapping method. Since the polynomial mapping method is a prior art, the specific process will not be described in detail here.

[0067] In this embodiment, the PCA algorithm is used to perform dimensionality reduction operations on all the increased-dimensional environmental data to obtain principal components (eigenvectors), and the obtained principal components are arranged and combined to obtain combinations between different principal components, and then the objective function is constructed to obtain the best combination, and the best combination is used as the best dimension of all initial clusters.

[0068] The purpose of first increasing the dimension and then reducing the dimension of all multidimensional environmental data in the above steps is to make the data after the dimensionality increase linearly separable, so as to obtain dispersed classified data.

[0069] Of course, as another implementation method, each initial cluster may be subjected to a dimensionality increase operation to obtain a cluster after dimensionality increase, and then a dimensionality reduction operation may be performed on all the dimensionality increase environmental data in all the clusters after dimensionality increase to obtain the principal component.

[0070] Taking the clusters obtained after dimensionality increase as an example, the objective function of the combination is constructed, and the best combination is obtained. The objective function is constructed based on the minimum value between classes and the maximum value within classes in the clusters after dimensionality increase corresponding to each initial cluster.

[0071] It should be noted that the present invention takes into account the problem that only selecting the principal component with the largest eigenvalue cannot characterize whether its data distribution conforms to the distribution of the initial cluster, and thus chooses to retain all principal components and arrange and combine all principal components to find the best combination of principal components; this is different from the method of obtaining the principal component with the largest eigenvalue as the projection coordinate in the existing method.

[0072] Furthermore, this embodiment also considers the problem that there may be intersections between the projection data of any two clusters after dimension reduction under the same combination, that is, when there is an intersection between the spaces formed by the projection data of each cluster after dimension increase and corresponding to the dimension reduction, it proves that the dimension reduction effect of the combination is poor; otherwise, the dimension reduction effect of the combination is good. Therefore, before calculating the objective function, it is necessary to perform a preliminary screening step of the combination, and the specific method is as follows:

[0073] Get two spatial bodies formed by the projection data of any two dimensionality-upgraded corresponding dimensionality-reduced data, and use the separating axis theorem to determine whether any two dimensionality-upgraded clusters intersect; if they intersect, discard the combination, otherwise, retain the corresponding combination. The specific steps of the separating axis theorem are: determine the separating axis; then project the two spatial bodies onto the separating axis, if the projections overlap, the two spatial bodies intersect.

[0074] The space body in this embodiment includes a polygon or a convex polyhedron.

[0075] In addition, if all combinations are discarded, it proves that the parameter settings of the initial cluster during dimensionality increase are inappropriate. At this time, the parameters in the dimensionality increase process need to be adjusted to ensure that there are no overlaps.

[0076] The reason for obtaining the optimal dimension is to prevent the data after dimensionality reduction from being unable to obtain dispersed classification data, so that the multi-dimensional environmental data can be linearly separable.

[0077] In step S4, the similarity of each initial cluster and the corresponding adjustment cluster is calculated; anomaly detection is performed on the power data corresponding to the initial cluster whose similarity is greater than or equal to a threshold, and an early warning is issued for abnormal wind power generation equipment.

[0078] The similarity acquisition process in this embodiment is as follows:

[0079] The class offset of any initial cluster and the class offset of the corresponding adjusted cluster are obtained, and the similarity of the class offset of any initial cluster and the class offset of the corresponding adjusted cluster is calculated.

[0080] The class offset of the initial cluster is the offset vector between the central data in the initial cluster and all other data; since the offset vector is the sum of the vectors of the central point and each point, it can represent the change of the overall information in the initial cluster. The calculation method of the class offset of the adjusted cluster is the same as that of the class offset of the initial cluster, which will not be repeated here.

[0081] Among them, the method for calculating similarity can adopt improved cosine similarity, cosine similarity, etc.

[0082] In another embodiment, the similarity in this embodiment is the difference between the cluster centers of the initial cluster and the corresponding cluster.

[0083] When the similarity is small, it means that the power generation environment corresponding to the current classification result has changed greatly. In this embodiment, similar power generation environments are screened by setting a threshold. That is, when the similarity is greater than or equal to the threshold, it means that the power generation environments of the wind power generation equipment in the current initial cluster are similar; otherwise, the power generation environment has changed greatly, and the initial cluster is eliminated.

[0084] The above threshold is set to 0.1, and can of course be adjusted by the implementer according to the specific implementation scenario.

[0085] In this embodiment, abnormality detection is performed on the power data corresponding to the initial clusters whose similarity is greater than or equal to the threshold value to realize the monitoring and management of wind power generation equipment. Existing abnormal data detection methods include local outlier factor (LOF) algorithm, box plot, SVM and other abnormal data detection methods. Taking the LOF algorithm as an example, the specific process of abnormality detection of power generation data is as follows:

[0086] Firstly, the power data corresponding to the same wind power generation equipment in the initial cluster whose similarity is greater than or equal to the threshold are counted; and each of the counted power data is taken as a data point.

[0087] The data of the same wind power generation equipment counted above may be insufficient. In this case, data under the same or similar environment within a historical time period can be selected for expansion to ensure sufficient data for abnormal detection of wind power generation data.

[0088] The specific process of screening is as follows: the radius of the circular range centered on any data point is the neighborhood radius, and any data point within the circular range whose number of data points is greater than the minimum sample number threshold is taken as a data point. The neighborhood radius and the minimum sample number threshold set above can be set according to actual conditions. In this embodiment, the neighborhood radius is 5, and the minimum sample number threshold is 3.

[0089] The selection of the core points mentioned above can improve the accuracy of anomaly detection and simplify the complexity of anomaly detection.

[0090] Secondly, the local outlier factor of each data point is calculated. The local outlier factor in this embodiment for: ;in, is the kth local reachable density of data point o, is the kth distance neighborhood of the current data point p, is the kth local reachable density of the current data point p. It should be noted that The larger the ratio is, the smaller the density of point p is, the smaller the density of point p is, and the more likely point p is an outlier. The smaller the ratio is, the larger the density of point p is, the larger the density of point p is, and the more likely point p is a normal point.

[0091] Then, when the local outlier factor of any data point is greater than the set value, the data point is an abnormal point; when the local outlier factor of any data point is less than or equal to the set value, the data point is a normal point, and then all abnormal points are obtained. When the number of abnormal points is greater than the set threshold, the abnormal result of the wind power generation equipment is obtained.

[0092] The above setting value can be 1, and can also be set according to actual conditions. The set threshold can be half or one quarter of the total number of power data; that is, when there are power data at multiple sampling moments, and the number of power data is greater than the set threshold, it proves that the corresponding wind power generation equipment is abnormal, and an early warning is issued to remind maintenance personnel to handle it; otherwise, no early warning is issued.

[0093] Of course, as another implementation method, it is also possible to directly perform anomaly detection on all power data in the initial cluster to obtain abnormal data, determine the wind power generation equipment corresponding to the abnormal data, and issue an early warning notification.

[0094] The solution of the present invention can analyze the environmental data of multiple wind power generation equipment, thereby screening out wind power generation equipment in a relatively stable environment, and then perform abnormal detection on the power data of wind power generation equipment when the environment is relatively stable, eliminating the influence of power data in an unstable environment, and can accurately monitor wind power generation equipment to achieve effective management.

[0095] The present invention also provides a monitoring and management system for wind power generation. Figure 2 As shown, the monitoring and management system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a wind power generation monitoring and management method according to the present invention is implemented.

[0096] The monitoring management system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface, whose configuration and functions are known in the art and thus will not be described in detail here.

[0097] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.

[0098] In the description of this specification, “plurality” means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0099] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A monitoring and management method for wind power generation, characterized in that: include: Acquire power data of multiple wind power generation equipment and multi-dimensional environmental data corresponding to the power data; The dimensions include wind direction, wind force, temperature, humidity and / or device orientation information; Cluster all multidimensional environmental data to obtain multiple initial clusters; Obtain the dimension reduction data of each initial cluster at the optimal dimension and obtain the adjusted cluster; Calculate the similarity of each initial cluster and the corresponding adjustment cluster, wherein the similarity is to obtain the class offset of any initial cluster and the class offset of the corresponding adjustment cluster, and use the modified cosine similarity method to calculate the similarity of the class offset of any initial cluster and the class offset of the corresponding adjustment cluster; wherein the class offset is the offset vector between the central data in the initial cluster or the adjustment cluster and all other remaining data; perform anomaly detection on the power data corresponding to the initial cluster whose similarity is greater than or equal to a threshold value, so as to provide an early warning of abnormal wind power generation equipment; The optimal dimension is: using the PCA algorithm to obtain the principal components of all multidimensional environmental data; arranging and combining the principal components to obtain multiple combinations; constructing a combined objective function; taking the combination corresponding to the maximum value of the objective function as the optimal dimension; the objective function for: , is the maximum value of the Euclidean distance between any two projection data in the projection data of the kth initial cluster under the nth combination, K is the total number of initial clusters, is the minimum Euclidean distance between two projection data in any two initial clusters under the nth combination; is the maximum value function, is an exponential function with the natural number e as the base; the projection data is the dimension reduction data of each multidimensional environmental data under the corresponding combination.

2. A wind power generation monitoring and management method according to claim 1, characterized in that: Before obtaining the principal component, the method also includes performing a dimensionality increase operation on all multidimensional environmental data to obtain the increased dimensional environmental data, and adopting a PCA algorithm to reduce the dimensionality of all the increased dimensional environmental data to obtain the principal component.

3. A wind power generation monitoring and management method according to claim 2, characterized in that: The dimension-raising operation adopts a polynomial mapping method.

4. A wind power generation monitoring and management method according to claim 1, characterized in that: The method of arranging and combining multiple principal components is as follows: A set number of principal components is randomly selected from multiple principal components to form a combination, thereby obtaining multiple combinations; wherein the set number is less than or equal to the total number of principal components.

5. A monitoring and management method for wind power generation according to claim 1, characterized in that: The specific process of the anomaly detection is as follows: Count the power data corresponding to the same wind power generation equipment in the initial cluster whose similarity is greater than or equal to the threshold; and regard each statistical power data as a data point; Calculate the local outlier factor for each data point; When the local outlier factor of any data point is greater than the set value, the data point is an abnormal point; otherwise, the data point is a normal point; when the number of abnormal points is greater than the set threshold, the wind power generation equipment is abnormal and an early warning is issued.

6. A wind power generation monitoring and management method according to claim 1, characterized in that: The specific process of the anomaly detection is: using the LOF algorithm to perform anomaly detection on all power data in the initial cluster whose similarity is greater than or equal to the threshold, obtain abnormal data, determine the wind power generation equipment corresponding to the abnormal data, and issue an early warning notification.

7. A wind power generation monitoring and management method according to claim 1, characterized in that: Before clustering, a step of normalizing all multidimensional environmental data is also included.

8. A monitoring and management system for wind power generation, characterized in that: include: processor; A memory storing computer instructions for monitoring and managing wind power generation. When the computer instructions are executed by the processor, the system executes a method for monitoring and managing wind power generation according to any one of claims 1 to 7.

Citation Information

Patent Citations

  • Integrated management system for informatization of wind power generation

    CN118297282A

  • Principal component analysis based wind power plant group transient voltage clustering recognition method

    CN107482683A

  • Emergency environment event safety early warning method based on big data analysis

    CN117150283A