A wind turbine fault analysis method, device, intelligent device and storage medium

Through the flow clustering algorithm and transfer entropy calculation, the problem of gradual fault diagnosis of wind turbines is solved, and reliable diagnosis of gradual faults of wind turbines and fault source tracking is achieved, improving the accuracy and efficiency of fault analysis.

CN118606798BActive Publication Date: 2024-11-12北京科锐特科技有限公司
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411082095.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-08
Publication Date
2024-11-12
Estimated Expiration
2044-08-08

Smart Images

  • Figure CN118606798B_ABST
    Figure CN118606798B_ABST
Patent Text Reader

Abstract

The present application relates to the field of fault detection technology, and specifically provides a method, apparatus, intelligent device and storage medium for analyzing faults of a wind turbine. It aims to solve the problem of lack of reliable diagnostic methods for gradual faults of wind turbines. To this end, the method for analyzing faults of a wind turbine provided by the present application includes: obtaining the time series operation data of the wind turbine, clustering the time series operation data based on a flow clustering algorithm to obtain multiple microclusters, determining the global stability domain based on the central core of the multiple microclusters and the number of data samples contained therein, and when the second global stability domain and the third global stability domain appear in sequence after the preset first global stability domain, it is judged that a gradual fault occurs in the wind turbine. The present application calculates the global stability domain according to the multiple microclusters obtained by flow clustering through flow clustering, and analyzes the changes in the global stability domain, thereby achieving the purpose of reliable diagnosis of gradual faults.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of fault detection, and in particular to a method, apparatus, intelligent device and storage medium for analyzing faults of a wind turbine generator set. Background Art

[0002] Wind turbines are usually exposed to outdoor environments. The variability of climate and environment has become an important factor affecting the safe operation of wind turbines. Most wind turbines are installed in remote locations, with high cabin heights and high operation and maintenance costs. Therefore, it is of great significance to monitor the condition of wind turbines. Supervisory Control And Data Acquisition (SCADA) systems have been widely used in wind farms, and fault prediction by collecting wind turbine operation data has received increasing attention. Most of the faults that occur in wind turbines are classified as gradual faults, that is, faults that gradually change over time. Gradual faults do not have particularly obvious characteristics, and there is still a lack of reliable diagnosis methods for gradual faults.

[0003] Accordingly, the art needs a new wind turbine fault analysis solution to solve the above problems. Summary of the invention

[0004] In order to overcome the above-mentioned defects, the present application is proposed to solve or at least partially solve the technical problem that it is difficult to reliably diagnose gradual faults.

[0005] In a first aspect, a method for analyzing a fault of a wind turbine is provided, the method comprising: acquiring sequential operation data of the wind turbine, the sequential operation data comprising data samples of multi-dimensional features recorded in chronological order; clustering the sequential operation data based on a flow clustering algorithm to obtain a plurality of microclusters; determining a global stability domain based on a central core of the plurality of microclusters and the number of data samples included; and determining that a gradual fault occurs in the wind turbine when a second global stability domain and a third global stability domain appear in sequence after a preset first global stability domain.

[0006] In a technical solution of the above-mentioned wind turbine fault analysis method, the method also includes: extracting sequences corresponding to features of multiple dimensions from data samples of the second global stability domain based on a preset time scale; calculating the first transfer entropy between sequences corresponding to features of any two dimensions; when the first transfer entropy is greater than or equal to a preset threshold, determining that there is a temporal causal relationship between the features of any two dimensions.

[0007] In a technical solution of the above-mentioned wind turbine fault analysis method, the method also includes: when the first transfer entropy value is less than a preset threshold, extracting a data stream sequence corresponding to features of multiple dimensions from data samples of a third global stable domain based on a preset time scale and a data stream sampling window; calculating a transfer entropy sequence between data stream sequences corresponding to features of any two dimensions; and judging the causal relationship between features of any two dimensions based on the transfer entropy sequence and a preset threshold.

[0008] In a technical solution of the above-mentioned wind turbine fault analysis method, when the second global stability domain and the third global stability domain appear in sequence after the preset first global stability domain, it is judged that the wind turbine has a gradual fault, including: calculating a first statistic between the second global stability domain and the preset first global stability domain; calculating a second statistic between the third global stability domain and the preset first global stability domain, the statistic including a mean and a variance; when both the first statistic and the second statistic are greater than a preset statistical threshold, it is judged that the wind turbine has a gradual fault.

[0009] In a technical solution of the above-mentioned wind turbine fault analysis method, the method also includes: when the first statistic or the second statistic is less than or equal to a preset statistical threshold, repeatedly executing the steps of obtaining the timing operation data of the wind turbine, clustering the timing operation data based on the flow clustering algorithm to obtain multiple micro-clusters, determining the global stability domain based on the central core of the multiple micro-clusters and the number of data samples contained therein, and when the second global stability domain and the third global stability domain appear successively after the preset first global stability domain, judging that the wind turbine has a gradual fault.

[0010] In a technical solution of the above-mentioned wind turbine fault analysis method, the timing operation data is clustered based on the flow clustering algorithm to obtain multiple micro-clusters, including: obtaining the data sample collection time and the delay equation; determining the weight corresponding to the data sample based on the delay equation and the difference between the current time and the data sample collection time; determining the cluster center and cluster radius corresponding to each data sample based on the weight, data sample and data sample collection time; and dividing the timing operation data into multiple micro-clusters based on the cluster center and cluster radius.

[0011] In a technical solution of the above-mentioned wind turbine fault analysis method, a global stability domain is determined based on the central cores of multiple microclusters and the number of data samples contained therein, including: when the core distances between the central cores of multiple stable microclusters are all less than a preset distance threshold, the multiple stable microclusters are regarded as a global stability domain, and the stable microcluster is a microcluster whose central core and the number of data samples contained therein do not change within a preset time.

[0012] In a second aspect, a wind turbine fault analysis device is provided, which includes: a data acquisition module, used to acquire the time series operation data of the wind turbine, the time series operation data including data samples recorded in time sequence with multi-dimensional features; a flow clustering module, used to cluster the time series operation data based on the flow clustering algorithm to obtain multiple micro-clusters; a stable domain determination module, used to determine the global stable domain based on the central cores of the multiple micro-clusters and the number of data samples included; a fault judgment module, used to judge that the wind turbine has a gradual fault when a second global stable domain and a third global stable domain appear in sequence after a preset first global stable domain.

[0013] In a technical solution of the above-mentioned wind turbine fault analysis device, the device also includes: a first data extraction module, used to extract sequences corresponding to features of multiple dimensions from data samples of the second global stability domain based on a preset time scale; a first transfer entropy calculation module, used to calculate the first transfer entropy between sequences corresponding to features of any two dimensions; a first temporal causal judgment module, used to judge that there is a temporal causal relationship between features of any two dimensions when the first transfer entropy is greater than or equal to a preset threshold.

[0014] In a technical solution of the above-mentioned wind turbine fault analysis device, the device also includes: a second data extraction module, which is used to extract data stream sequences corresponding to features of multiple dimensions from data samples in the third global stability domain based on a preset time scale and a data stream sampling window when the first transfer entropy value is less than a preset threshold; a second transfer entropy calculation module, which is used to calculate the transfer entropy sequence between data stream sequences corresponding to features of any two dimensions; and a second time series causal judgment module, which is used to judge the causal relationship between features of any two dimensions based on the transfer entropy sequence and a preset threshold.

[0015] In a technical solution of the above-mentioned wind turbine fault analysis device, the fault judgment module includes: a first statistic calculation unit, used to calculate the first statistic between the second global stability domain and the preset first global stability domain; a second statistic calculation unit, used to calculate the second statistic between the third global stability domain and the preset first global stability domain, the statistics including the mean and the variance; a gradual fault judgment unit, used to judge that a gradual fault occurs in the wind turbine when both the first statistic and the second statistic are greater than a preset statistical threshold.

[0016] In a technical solution of the above-mentioned wind turbine fault analysis device, the device also includes: a loop module, which is used to repeatedly execute the steps of obtaining the timing operation data of the wind turbine when the first statistic or the second statistic is less than or equal to a preset statistical threshold, clustering the timing operation data based on the flow clustering algorithm to obtain multiple micro-clusters, determining the global stability domain based on the central core of the multiple micro-clusters and the number of data samples contained therein, and when the second global stability domain and the third global stability domain appear successively after the preset first global stability domain, judging that the wind turbine has a gradual fault.

[0017] In a technical solution of the above-mentioned wind turbine fault analysis device, the flow clustering module includes: an acquisition unit, used to obtain the data sample collection time and the delay equation; a weight determination unit, used to determine the weight corresponding to the data sample based on the delay equation and the difference between the current time and the data sample collection time; a parameter calculation unit, used to determine the cluster center and cluster radius corresponding to each data sample based on the weight, data sample and data sample collection time; a clustering unit, used to divide the time series operation data into multiple micro-clusters based on the cluster center and cluster radius.

[0018] In a technical solution of the above-mentioned wind turbine fault analysis device, the stable domain determination module includes: a stable domain determination unit, which is used to treat the multiple stable micro-clusters as a global stable domain when the core distances between the central cores of multiple stable micro-clusters are all less than a preset distance threshold. The stable micro-cluster is a micro-cluster with a central core and the number of data samples contained therein no longer changes within a preset time.

[0019] In a third aspect, an intelligent device is provided, comprising at least one processor; and a memory communicatively connected to the at least one processor; wherein a computer program is stored in the memory, and when the computer program is executed by the at least one processor, the wind turbine fault analysis method in the above-mentioned first aspect or any corresponding technical solution is implemented.

[0020] In a fourth aspect, a computer-readable storage medium is provided, wherein a plurality of program codes are stored therein, and the program codes are suitable for being loaded and run by a processor to execute the wind turbine fault analysis method in the above-mentioned first aspect or any corresponding technical solution thereof.

[0021] The above one or more technical solutions of the present application have at least one or more of the following beneficial effects:

[0022] In the technical solution for implementing the present application, flow clustering is performed on the time-series operation data of the wind turbine set, and the change of the global stability domain is determined based on the multiple micro-clusters obtained by the flow clustering, thereby realizing the analysis of the operation status of the wind turbine set. Since a gradual fault is a fault caused by gradual changes over time, the gradual fault can be diagnosed through the change of the global stability domain. When the second global stability domain and the third global stability domain appear in sequence after the preset first global stability domain, it is determined that the stable state of the wind turbine set has changed, that is, a gradual fault has occurred, thereby achieving the purpose of reliably diagnosing a gradual fault.

[0023] In the technical solution for implementing the present application, by extracting sequences corresponding to features of different dimensions from data samples in the second global stability domain and calculating the first transfer entropy between sequences corresponding to any two dimensional features, the purpose of analyzing the temporal causal relationship between features is achieved, the purpose of tracing the source of gradual faults is achieved, and the effect of improving the reliability and practical value of fault analysis is achieved.

[0024] In the technical solution for implementing the present application, the sequences corresponding to the features of different dimensions are processed using a preset time scale, thereby achieving the purpose of simplifying the sequences, reducing the amount of calculation and improving the efficiency of wind turbine fault analysis.

[0025] In the technical solution for implementing the present application, by comparing the first transfer entropy with a preset threshold, it is determined that there is a temporal causal relationship between two features only when the first transfer entropy is greater than or equal to the preset threshold, thereby achieving the purpose of improving the reliability of temporal causal relationship analysis.

[0026] In the technical solution for implementing the present application, the transfer entropy sequence is calculated based on the data samples of the third global stability domain, which enables further analysis of the temporal causal relationship in the temporal development, thereby achieving the effect of improving the reliability and accuracy of wind turbine fault analysis.

[0027] In the technical solution for implementing the present application, by comparing the first statistic and the second statistic with the preset statistical threshold, the reliability of the global stability domain change is verified, thereby achieving the effect of further improving the reliability of wind turbine gradual fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The disclosure of the present application will become easier to understand with reference to the accompanying drawings. It is easy for those skilled in the art to understand that these drawings are only for illustrative purposes and are not intended to limit the scope of protection of the present application. Among them:

[0029] Figure 1 This is a schematic flow chart of the main steps of a wind turbine fault analysis method according to an embodiment of the present application;

[0030] Figure 2 is a schematic flow chart of main steps of a method for analyzing a wind turbine fault according to another embodiment of the present application;

[0031] Figure 3 This is a schematic diagram of the overall steps of a wind turbine fault analysis method according to an embodiment of the present application;

[0032] Figure 4 It is a schematic diagram of the main structural block diagram of a wind turbine fault analysis device according to an embodiment of the present application;

[0033] Figure 5 It is a schematic diagram of the main structure of a smart device according to an embodiment of the present application. DETAILED DESCRIPTION

[0034] Some embodiments of the present application are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application and are not intended to limit the protection scope of the present application.

[0035] In the description of the present application, the terms "first", "second", etc. are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order 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 that includes 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. The terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected, it can be indirectly connected through an intermediate medium, it can also be the internal communication of two elements, it can be a wireless connection, or it can be a wired connection.

[0036] In addition, "module" and "processor" may include hardware, software or a combination of the two. A module may include hardware circuits, various suitable sensors, communication ports, and memories, and may also include software parts, such as program codes, or a combination of software and hardware. The processor may be a central processing unit, a microprocessor, an image processor, a digital signal processor, or any other suitable processor. The processor has data and / or signal processing functions. The processor may be implemented in software, hardware, or a combination of the two. Computer-readable storage media include any suitable media that can store program codes, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc.

[0037] In addition, if the meaning of "and / or" appears in this application, it includes three parallel solutions. Taking "A and / or B" as an example, it includes solution A, solution B, or a solution that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in this field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by this application. The term "at least one A or B" or "at least one of A and B" has a similar meaning to "A and / or B" and can include only A, only B, or A and B. The singular terms "one" and "this" can also include plural forms.

[0038] See attached Figure 1 , Figure 1 FIG. 1 is a schematic diagram of the main steps of a wind turbine fault analysis method according to an embodiment of the present application. Figure 1 As shown, the wind turbine fault analysis method in the embodiment of the present application mainly includes the following steps S101 to S104:

[0039] Step S101, obtaining time series operation data of a wind turbine generator set, wherein the time series operation data includes data samples recorded in time sequence with multi-dimensional features.

[0040] In this embodiment, the time series operation data of the wind turbine set is the continuous data generated by the operation of the wind turbine set. That is to say, as time goes by, the operation data will gradually increase. Multidimensional features refer to various features that affect the operation of the wind turbine set, including but not limited to: ambient wind speed, ambient temperature, frequency, active power, reactive power, impeller speed, generator speed, wind angle, wind direction angle, yaw position, yaw speed, grid-side A-phase current, grid-side B-phase current, grid-side C-phase current, grid-side A-phase voltage, grid-side B-phase voltage, grid-side C-phase voltage, horizontal acceleration, vertical acceleration, blade angular velocity, blade pitch angle, pitch motor temperature, hydraulic system oil temperature, hydraulic system oil pressure, cabin temperature, gearbox oil temperature, gearbox high-speed bearing temperature, generator slip ring temperature, generator stator winding temperature, generator drive side bearing temperature, converter controller temperature, converter control cabinet temperature, converter rotor side temperature, converter grid side reactor temperature, etc. For example, taking the three characteristics of impeller speed, generator speed and wind angle as examples, the time series operation data is a three-dimensional vector composed of the three characteristics of impeller speed, generator speed and wind angle at multiple sampling moments. If n three-dimensional vectors are obtained by sampling at n moments, the time series operation data includes n data samples.

[0041] In one implementation, features may be selected according to specific circumstances, such as selecting ten features with greater relevance to known faults or predicted faults of the wind turbine generator system from the features listed above to obtain time series operation data.

[0042] In one embodiment, the time series operation data may be obtained from a supervisory control and data acquisition system based on a plurality of selected features.

[0043] In one implementation, the time series data can be represented by a data stream model. The data stream model is an emerging data model, and its main feature is to focus on the real-time nature of data processing. The data stream contains a large amount of continuous data. As the process progresses, new data will continue to expand and update the model. As an example of a data stream model, suppose there is multidimensional data [X 1 , X 2 , ..., X k ], any dimensional data is in T 1 , T 2 , ..., T d Time arrival data record:

[0044] .

[0045] Step S102: clustering the time series operation data based on a flow clustering algorithm to obtain a plurality of micro-clusters.

[0046] In this embodiment, the stream clustering algorithm (CluStream) is an algorithm for processing clustering problems in data streams, which mainly includes two processes: micro clustering and macro clustering. Micro clustering processes newly arrived data in real time and stores statistical results periodically, while macro clustering uses the statistical results to obtain clustering results.

[0047] In one implementation, the process of clustering time series running data using a stream clustering algorithm is as follows: q microclusters are pre-established, and when a new data sample (data point) X arrives, it is calculated whether the distance from X to the center of each microcluster is less than or equal to a preset radius threshold, and if so, X is added to the existing microcluster closest to it, otherwise a new microcluster is created for X. After a new microcluster is created, an old microcluster can be deleted or two old microclusters can be merged according to time validity.

[0048] Step S103: determining a global stable region based on the central cores of the multiple microclusters and the number of data samples included.

[0049] In this embodiment, the global stable domain represents the stable state of the wind turbine operation. When the central cores of multiple microclusters and the data samples contained therein no longer change within a certain period of time, and the center distance between any two microclusters in the multiple microclusters is less than a preset distance threshold, the multiple microclusters can be considered to form a global stable domain.

[0050] Step S104: when the second global stability region and the third global stability region appear in sequence after the preset first global stability region, it is determined that a gradual fault occurs in the wind turbine generator set.

[0051] In this embodiment, the preset first global stability domain represents the global stability domain reached in the initial operation state of the wind turbine generator set. The default initial operation state is the stable state of the wind turbine generator set in non-fault operation. Since a gradual fault is a fault that gradually changes over time, when a gradual fault occurs, the stable state (global stability domain) will change at least twice. Therefore, when the second global stability domain and the third global stability domain appear in sequence after the preset first global stability domain, it is determined that a gradual fault occurs in the wind turbine generator set, wherein the preset first global stability domain, the second global stability domain, and the third global stability domain are different stability domains.

[0052] Based on the method described in steps S101 to S104 above, flow clustering is performed on the time-series operation data of the wind turbine generator set, and the change of the global stability domain is determined based on the multiple micro-clusters obtained by the flow clustering, thereby realizing the analysis of the operating status of the wind turbine generator set. Since a gradual fault is a fault caused by gradual changes over time, the gradual fault can be diagnosed through the change of the global stability domain. When the second global stability domain and the third global stability domain appear in sequence after the preset first global stability domain, it is determined that the stable state of the wind turbine generator set has changed, that is, a gradual fault has occurred, thereby achieving the purpose of reliably diagnosing the gradual fault.

[0053] In an application scenario according to an embodiment of the present application, wind energy, as a renewable energy source, has received great attention from the industry and academia in the past two decades. Because of its large-scale development prospects and the advantages of clean and pollution-free, wind power generation has been widely used worldwide. The large-scale application of wind power generation has brought great opportunities for the development of the energy market economy, but also brought more severe challenges to the reliability and stability of wind turbines. First, the variability of climate and environment, as an important factor affecting the safe operation of wind turbines, increases the probability of failure of wind turbines. Secondly, since most wind turbines are installed in remote locations and the cabin height is high, the operation and maintenance costs of wind turbines are higher than other traditional power generation methods. Condition monitoring of wind turbines, timely fault warning in the early stage of failure, and reducing the impact of failures can improve the operating reliability of the unit and reduce operation and maintenance costs.

[0054] In recent years, data-driven wind turbine fault prediction methods have received increasing attention. In the field of wind power, there have been a lot of studies on wind turbine fault detection, diagnosis and early warning based on SCADA data analysis. At present, most of the faults that occur in wind turbines are classified as gradual faults, but the definition of gradual faults is relatively vague, and there is a lack of analysis of their changing characteristics. In addition, since the signs of gradual faults are not obvious, it is difficult to perform causal feature analysis on the process containing the fault, that is, it is impossible to track the source of the fault.

[0055] In view of the above problems, this embodiment provides a method for analyzing wind turbine faults. Figure 2 FIG. 1 is a schematic flow chart of the main steps of a method for analyzing a wind turbine fault according to another embodiment of the present application. Figure 2 As shown, the wind turbine fault analysis method in the embodiment of the present application mainly includes the following steps S201 to S210:

[0056] Step S201, obtaining time series operation data of a wind turbine generator set, wherein the time series operation data includes data samples recorded in time sequence with multi-dimensional features.

[0057] In this embodiment, the detailed description of step S201 can be found in Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0058] Step S202: clustering the time series operation data based on a flow clustering algorithm to obtain a plurality of micro-clusters.

[0059] In this embodiment, the detailed description of step S202 can be found in Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0060] In an optional implementation, the above step S202 includes the following steps S2021 to S2024:

[0061] Step S2021, obtaining the data sample collection time and the delay equation.

[0062] In this embodiment, the data sample collection moment, that is, the data sample collection time, can be obtained from the SCADA system, and the delay equation is used to assign a weight to the data sample that changes over time, so that the information of the new data can be more fully applied to the calculation.

[0063] As an example, the time delay equation f(t) is as follows:

[0064]

[0065] Among them, λ represents the delay parameter, and λ can be selected according to demand. The larger the λ value is, the smaller the importance (weight) of the historical data sample is, and the greater the importance of the current data sample is.

[0066] Step S2022: determine the weight corresponding to the data sample based on the time delay equation and the difference between the current time and the data sample collection time.

[0067] In this embodiment, the weight w can be calculated using the following formula:

[0068]

[0069] in, represents the collection time of the jth data in the i-th feature sequence ( ), t represents the current time, n represents the total number of features, m represents the total number of time, f(t) represents the delay equation, and μ represents the first preset threshold parameter.

[0070] Step S2023, determining the cluster center and cluster radius corresponding to each data sample based on the weight, data sample and the time when the data sample was collected.

[0071] In this embodiment, the cluster center c and the cluster radius r are calculated based on the weight w in step S2022 using the following formulas:

[0072]

[0073]

[0074] in, represents the jth data in the i-th feature sequence, dist() represents the Euclidean distance, is the second preset threshold parameter.

[0075] Step S2024: Divide the time series operation data into a plurality of micro-clusters based on the cluster center and the cluster radius.

[0076] In this embodiment, according to the cluster center c and cluster radius r calculated in step S2023, X is added to the existing micro-cluster within the cluster radius r and closest to it. If X cannot be added to the existing micro-cluster, the corresponding weight w is compared with the preset weight threshold Compare, if w ≥ , then X is taken as a new micro-cluster and the above steps are repeated to cluster the new data points.

[0077] In one embodiment, microclusters may be periodically inspected to remove unimportant microclusters. For example, the inspection period T p for:

[0078]

[0079] Among them, λ is the delay parameter, is the preset weight threshold.

[0080] Weight is less than The micro-clusters (considered to be unimportant micro-clusters) will be removed and added to the outlier micro-cluster group. For the outlier micro-clusters, it can be further determined whether the weight is less than :

[0081]

[0082] Among them, t c , t 0 The current time and the outlier microcluster generation time are T p is the inspection period. When inspecting all outlier microclusters, t 0 Set as the data sampling period.

[0083] For weights less than The outlier micro-clusters are frozen to obtain frozen micro-clusters, wherein the frozen micro-clusters can be activated when accessed by users, and the clustering operation is performed again to ensure the integrity of the data.

[0084] Step S203: determining a global stable region based on the central cores of the multiple microclusters and the number of data samples included.

[0085] In this embodiment, if the number of data samples of a micro-cluster N i and the central core (microcluster center) c i If it does not change any more, the microcluster is considered to be a stable microcluster (local stable domain), denoted as LS i =[c i , N i ],[c i , N i ] indicates that it contains the central core c i And the number of samples N i A collection of .

[0086] If there are d local stability regions LS 1 LS 2 ,…,LS d The central cores are [c 1 , …, c d ], and the distance between any two central cores is dist(c i , c j ) are all less than the first preset value δ max , then the global stability region GS is defined as LS 1 LS 2 ,…,LS d Collection of [LS 1 , L.S. 2 ,…,LS d ].

[0087] Step S204: when the second global stability region and the third global stability region appear in sequence after the preset first global stability region, it is determined that a gradual fault occurs in the wind turbine generator set.

[0088] In this embodiment, if a new local stable region LS appears i , the central core c i and the adjacent global stability region GS i The distances of the d central cores contained in the d-th core are all greater than the second preset value δ min , then a new global stable region GS is generated i+1 Compared with the initial state, that is, the preset first global stable region GS 1 , if two new global stable regions GS appear in succession 2 GS 3 This means that the operating status of the wind turbine has migrated twice, and it is judged that a gradual fault may have occurred.

[0089] In an optional implementation, the above step S204 may further include the following steps a1 to a3:

[0090] Step a1, calculating a first statistic between the second global stable region and a preset first global stable region.

[0091] Step a2, calculating a second statistic between the third global stable region and the preset first global stable region, the statistic including a mean and a variance.

[0092] Step a3: when both the first statistic and the second statistic are greater than a preset statistical threshold, it is determined that a gradual fault occurs in the wind turbine generator set.

[0093] For the above steps a1 to a3, that is, to compare the two global stability regions GS 2 GS 3 With the preset first stable region GS 1 Whether the mean μ and variance σ between them are greater than the corresponding statistical threshold and If both the mean and the variance exceed the corresponding statistical threshold, it can be determined that a gradual fault occurs. Otherwise, the above steps S201 to S204 are repeated to continue accumulating a new global stable domain.

[0094] In the technical solution of implementing the present application, by comparing the first statistic and the second statistic with the preset statistical threshold, the purpose of verifying the reliability of the global stability domain change is achieved, thereby achieving the effect of improving the reliability of wind turbine gradual fault diagnosis.

[0095] Step S205: extracting sequences corresponding to features of multiple dimensions from data samples in the second global stable domain based on a preset time scale.

[0096] In this embodiment, the (time) sequence corresponding to the feature of any dimension is a sequence composed of data of the feature of any dimension at multiple moments. Assume that the data samples of the features of the two dimensions of wind direction angle and opposite wind angle in the second global stable domain are respectively , .in, represents the sequence corresponding to the wind direction angle features extracted from the second global stable domain, Indicates the wind direction angle at the Nth 2 Data at a certain moment; represents the sequence corresponding to the wind angle feature extracted from the second global stable domain, Indicates the wind angle at the Nth 2 Data at a certain moment. Preset time scale Used to , The length N 2 Compress to .

[0097] As an example, take the sequence For example, the following formula can be used to process it to obtain :

[0098]

[0099] in, is the time scale, ≥1 and is a positive integer, Represents the data at the mth moment in the second global stable domain.

[0100] Step S206, calculating the first transfer entropy between sequences corresponding to features of any two dimensions.

[0101] In this embodiment, the transfer entropy measures the influence of one variable on another variable in the system or the intensity of information transmission. By calculating the transfer entropy, the causal relationship and interaction between different variables in a complex system can be analyzed. and The transfer entropy between It can be calculated using the following formula:

[0102]

[0103] in, Representation sequence The sequence obtained after time scale processing; Representation sequence The sequence obtained after time scale processing; Representation sequence The sequence obtained after time scale processing is Right now The time series of the next sampling window; p[X,Y] represents the joint probability distribution of X and Y; p[X|Y] represents the probability of X occurring under the condition that Y occurs; H[X] represents the entropy of X; H[X,Y] represents the joint information content of X and Y; log() represents the logarithmic function.

[0104] By , Substitute into the calculation formula of the transfer entropy in step S206 to obtain the first transfer entropy .

[0105] Step S207: when the first transfer entropy is greater than or equal to a preset threshold, it is determined that there is a temporal causal relationship between the features of any two dimensions.

[0106] In this embodiment, when the first transfer entropy Greater than or equal to a preset threshold (such as significance level ), it is considered that there is a temporal causal relationship between features X and Y. Specifically, if , indicating that the appearance of Y leads to the appearance of X; if , indicating that Y has no effect on X, that is, there is no temporal causal relationship.

[0107] In one embodiment, the significance level may be calculated by summing the mean and variance of the transfer entropy.

[0108] Step S208, when the first transfer entropy value is less than a preset threshold, extracting a data stream sequence corresponding to features of multiple dimensions from data samples in the third global stable domain based on a preset time scale and a data stream sampling window.

[0109] The main difficulties of the transfer entropy algorithm in dealing with gradual faults of wind turbines are as follows: the online use of transfer entropy has a heavy computational burden and it is difficult to track fault changes in a timely manner; since the number of fault samples will increase with the increase in data volume, calculating global data will cause a huge computational burden; the validity of the calculation results cannot be determined in real time, and the significance level of the algorithm will change over time.

[0110] In this embodiment, if , then based on the preset time scale And the data stream sampling window SW D Extract the data stream sequence corresponding to the features of multiple dimensions from the data samples of the third global stable domain. It should be noted that the preset time scale It can be set as required each time a sequence is extracted. Data stream sampling window SW D Also called a sliding window, it is used to divide the data stream into multiple data blocks. Since the data stream is continuously updated over time, SW D The data contained in it changes all the time, that is to say, the sampling process of the data stream is a process in which the data is constantly updated within a fixed window, and the window length n can be adjusted according to the timing characteristics of the variable changes.

[0111] Assume that the time unit of data stream sampling is T S , sampling window SW D At time t i and t i+1 The time interval between two consecutive samplings is , t i+1 The data contained in the time is compared with t i The number of samples added within a time is called the data update amount D u , , the unit is the number of samples. Assume that the data stream sampling window SW DThe length of is n, then the sequence of multiple sampling windows (data stream sequence) can be expressed as: .

[0112] In one embodiment, the sampling window SW D Length and data update amount D u It can be set according to specific needs, and this embodiment does not limit this.

[0113] In a real-time mode, if the data samples in the third global stable domain are small, the transfer entropy calculation is temporarily not performed until the data samples in the third global stable domain accumulate to a number greater than or equal to the number of data samples in the second global stable domain, and then the transfer entropy is calculated.

[0114] Step S209, calculating the transfer entropy sequence between the data stream sequences corresponding to the features of any two dimensions.

[0115] In this embodiment, the data stream sequence corresponding to the features of any two dimensions in step S208 is substituted into the data stream sequence in step S206 to transfer entropy. The calculation formula is used to obtain the transfer entropy sequence DDS(T) under multiple sampling windows:

[0116]

[0117] Step S210, determining the causal relationship between features of any two dimensions based on the transfer entropy sequence and a preset threshold.

[0118] In this embodiment, for In the case of the third stable domain, the record values ​​in the transfer entropy sequence DDS(T) obtained based on T=2 are compared with the preset thresholds (such as ) are compared. When any item in the transfer entropy sequence DDS(T) is greater than or equal to the preset threshold, it is determined that there is a temporal causal relationship between the features (sequences) of any two dimensions, that is, the fault sequence of feature Y causes the fault sequence of feature X to change.

[0119] It can be understood that DDS(T) represents the dynamic flow transfer entropy, T represents the unit of transfer entropy update, and is related to the sampling window SW D The time interval between two consecutive samplings Equal to, usually greater than the time unit T of the data stream sampling S . The data update amount D u From the calculation formula, we can know that when D u >1, T>T S .

[0120] In one embodiment, within a certain range, the time scale is changed If the corresponding transfer entropy is approximately equal to zero, it means that sequence Y has no effect on X at this scale; if there is , so that the value of transfer entropy is greater than zero, indicating that sequence Y has an impact on X; the larger the transfer entropy value, the more it can capture the real delay information between sequences at this scale.

[0121] In this embodiment, by calculating the transfer entropy sequence DDS(T) under multiple sampling windows, that is, the dynamic flow transfer entropy, the purpose of online calculation is achieved by replacing the original static data with a dynamic time series, and the effect of timely tracking the latest changes of the fault point is achieved. By changing the time scale parameter, the purpose of extracting multi-scale information between variables, reducing the amount of calculation and improving the efficiency of online calculation can be achieved. The use of a density-based flow clustering algorithm achieves the purpose of accurately clustering the time series operation data of the wind turbine, and achieves the effect of improving the accuracy of fault feature extraction. By calculating the dynamic flow transfer entropy, real-time tracking of the real-time changes in the causal relationship between features during the operation of the wind turbine is achieved, and finally the purpose of causal feature analysis of the gradual fault process and tracking the source of the fault is achieved.

[0122] In an optional embodiment, see the attached Figure 3 , Figure 3 FIG. 1 is a schematic diagram of the overall steps of a wind turbine fault analysis method according to an embodiment of the present application. Figure 3 As shown, the wind turbine fault analysis method in the embodiment of the present application mainly includes the following steps S301 to S307:

[0123] Step S301, input the timing operation data of the wind turbine. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0124] Step S302: cluster the relevant process variable pairs using the flow clustering algorithm. The relevant process variables are the time series operation data of the wind turbines containing multi-dimensional features. For details, see Figure 2 Step S202 of the illustrated embodiment will not be described in detail here.

[0125] Step S303, determining whether two new global stable domains appear successively. The two new global stable domains appear successively means that the second global stable domain and the third global stable domain appear successively after the preset first global stable domain (or the global stable domain corresponding to a normal operating state).

[0126] Step S304: If two new global stable domains do not appear consecutively, then continue to accumulate new stable domains.

[0127] Step S305: If two new global stable domains appear successively, it is determined to be a fault (gradual fault).

[0128] Step S306: Use the dynamic flow transfer entropy algorithm to calculate the transfer entropy between the fault features in sequence. The specific calculation method of the transfer entropy can be found in Figure 2 The step S206 of the illustrated embodiment will not be described in detail here. The transfer entropy calculated in step S306 can be used to perform causal relationship analysis on the fault characteristics.

[0129] Step S307, obtaining the fault causal relationship.

[0130] For specific implementation methods of the above steps, please refer to similar step descriptions in other embodiments of the present application.

[0131] It should be pointed out that although the various steps in the above embodiments are described in a specific order, those skilled in the art can understand that in order to achieve the effect of the present application, different steps do not have to be executed in such an order, and they can be executed simultaneously (in parallel) or in other orders. These adjusted schemes are equivalent to the technical schemes described in this application, and therefore will also fall within the scope of protection of this application.

[0132] Another aspect of the present application also provides a wind turbine fault analysis device, such as Figure 4 As shown, the device includes: a data acquisition module 401, used to acquire the time series operation data of the wind turbine generator set, the time series operation data includes data samples recorded in time sequence with multi-dimensional features; a flow clustering module 402, used to cluster the time series operation data based on the flow clustering algorithm to obtain multiple micro-clusters; a stability domain determination module 403, used to determine the global stability domain based on the central cores of the multiple micro-clusters and the number of data samples contained; a fault judgment module 404, used to judge that the wind turbine generator set has a gradual fault when the second global stability domain and the third global stability domain appear in sequence after the preset first global stability domain.

[0133] In a technical solution of the above-mentioned wind turbine fault analysis device, the device also includes: a first data extraction module, used to extract sequences corresponding to features of multiple dimensions from data samples of the second global stability domain based on a preset time scale; a first transfer entropy calculation module, used to calculate the first transfer entropy between sequences corresponding to features of any two dimensions; a first temporal causal judgment module, used to judge that there is a temporal causal relationship between features of any two dimensions when the first transfer entropy is greater than or equal to a preset threshold.

[0134] In a technical solution of the above-mentioned wind turbine fault analysis device, the device also includes: a second data extraction module, which is used to extract data stream sequences corresponding to features of multiple dimensions from data samples in the third global stability domain based on a preset time scale and a data stream sampling window when the first transfer entropy value is less than a preset threshold; a second transfer entropy calculation module, which is used to calculate the transfer entropy sequence between data stream sequences corresponding to features of any two dimensions; and a second time series causal judgment module, which is used to judge the causal relationship between features of any two dimensions based on the transfer entropy sequence and a preset threshold.

[0135] In a technical solution of the above-mentioned wind turbine fault analysis device, the fault judgment module includes: a first statistic calculation unit, used to calculate the first statistic between the second global stability domain and the preset first global stability domain; a second statistic calculation unit, used to calculate the second statistic between the third global stability domain and the preset first global stability domain, the statistics including the mean and the variance; a gradual fault judgment unit, used to judge that a gradual fault occurs in the wind turbine when both the first statistic and the second statistic are greater than a preset statistical threshold.

[0136] In a technical solution of the above-mentioned wind turbine fault analysis device, the device also includes: a loop module, which is used to repeatedly execute the steps of obtaining the timing operation data of the wind turbine when the first statistic or the second statistic is less than or equal to a preset statistical threshold, clustering the timing operation data based on the flow clustering algorithm to obtain multiple micro-clusters, determining the global stability domain based on the central core of the multiple micro-clusters and the number of data samples contained therein, and when the second global stability domain and the third global stability domain appear successively after the preset first global stability domain, judging that the wind turbine has a gradual fault.

[0137] In a technical solution of the above-mentioned wind turbine fault analysis device, the flow clustering module includes: an acquisition unit, used to obtain the data sample collection time and the delay equation; a weight determination unit, used to determine the weight corresponding to the data sample based on the delay equation and the difference between the current time and the data sample collection time; a parameter calculation unit, used to determine the cluster center and cluster radius corresponding to each data sample based on the weight, data sample and data sample collection time; a clustering unit, used to divide the time series operation data into multiple micro-clusters based on the cluster center and cluster radius.

[0138] In a technical solution of the above-mentioned wind turbine fault analysis device, the stable domain determination module includes: a stable domain determination unit, which is used to treat the multiple stable micro-clusters as a global stable domain when the core distances between the central cores of multiple stable micro-clusters are all less than a preset distance threshold. The stable micro-cluster is a micro-cluster with a central core and the number of data samples contained therein no longer changes within a preset time.

[0139] It is understood by those skilled in the art that all or part of the processes in the method for implementing the above-mentioned embodiment of the present application can also be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of each of the above-mentioned method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. Computer-readable storage media may include: any entity or device, medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory, random access memory, electrical carrier signal, telecommunication signal and software distribution medium, etc. that can carry computer program code.

[0140] Another aspect of the present application also provides a computer-readable storage medium.

[0141] In an embodiment of a computer-readable storage medium according to the present application, the computer-readable storage medium may be configured to store a program for executing the wind turbine fault analysis method of the above method embodiment, and the program may be loaded and run by a processor to implement the above wind turbine fault analysis method. For ease of explanation, only the parts related to the embodiment of the present application are shown. For specific technical details not disclosed, please refer to the method part of the embodiment of the present application. The computer-readable storage medium may be a storage device formed by various electronic devices. Optionally, the computer-readable storage medium in the embodiment of the present application is a non-temporary computer-readable storage medium.

[0142] Another aspect of the present application also provides a smart device.

[0143] In an embodiment of an intelligent device according to the present application, the intelligent device may include at least one processor; and a memory connected to the at least one processor in communication; wherein a computer program is stored in the memory, and when the computer program is executed by the at least one processor, the method described in any of the above embodiments is implemented. Figure 5 , Figure 5 FIG. 4 exemplarily shows that the memory 11 and the processor 12 are communicatively connected via a bus.

[0144] In some embodiments of the present application, the smart device may further include at least one sensor, which is used to sense information. The sensor is communicatively connected to any type of processor mentioned in the present application. Optionally, the smart device described in the present application may be, but is not limited to, a mobile phone, a tablet computer, a desktop, a laptop, a handheld computer, a notebook computer, a vehicle-mounted device, an ultra-mobile personal computer (UMPC), etc., and the embodiments of the present application are not limited to this.

[0145] So far, the technical solution of the present application has been described in conjunction with an embodiment shown in the accompanying drawings, but it is easy for those skilled in the art to understand that the protection scope of the present application is obviously not limited to these specific embodiments. Without departing from the principles of the present application, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present application.

Claims

1. A wind turbine fault analysis method, It is characterized in that The method comprises: Acquire time-series operation data of the wind turbine generator set, wherein the time-series operation data includes data samples recorded in time sequence with multi-dimensional features; Clustering the time series operation data based on a flow clustering algorithm to obtain a plurality of micro-clusters; Determine the global stability domain based on the central cores of multiple microclusters and the number of data samples included; When the second global stability domain and the third global stability domain appear in sequence after the preset first global stability domain, it is determined that a gradual fault occurs in the wind turbine generator set; The method of determining a global stable domain based on the central cores of a plurality of microclusters and the number of data samples contained therein comprises: when the core distances between the central cores of a plurality of stable microclusters are all less than a preset distance threshold, the plurality of stable microclusters are regarded as a global stable domain, wherein the stable microcluster is a microcluster whose central core and the number of data samples contained therein do not change within a preset time.

2. The wind turbine fault analysis method according to claim 1, It is characterized in that The method further comprises: Extracting sequences corresponding to features of multiple dimensions from data samples in the second global stable domain based on a preset time scale; Calculate the first transfer entropy between the sequences corresponding to the features of any two dimensions; When the first transfer entropy is greater than or equal to a preset threshold, it is determined that there is a temporal causal relationship between the features of any two dimensions.

3. The wind turbine fault analysis method according to claim 2, It is characterized in that The method further comprises: When the first transfer entropy value is less than a preset threshold, extracting a data stream sequence corresponding to features of multiple dimensions from data samples in the third global stable domain based on a preset time scale and a data stream sampling window; Calculate the transfer entropy sequence between the data stream sequences corresponding to the features of any two dimensions; The causal relationship between the features of any two dimensions is determined based on the transfer entropy sequence and a preset threshold.

4. The wind turbine fault analysis method according to claim 1, It is characterized in that When the second global stability domain and the third global stability domain appear in sequence after the preset first global stability domain, it is determined that the wind turbine generator set has a gradual fault, including: Calculating a first statistic between the second global stable domain and a preset first global stable domain; Calculating a second statistic between the third global stability domain and the preset first global stability domain, wherein the statistic includes a mean and a variance; When both the first statistic and the second statistic are greater than a preset statistical threshold, it is determined that a gradual fault occurs in the wind turbine generator set.

5. The wind turbine fault analysis method according to claim 4, It is characterized in that The method further comprises: When the first statistic or the second statistic is less than or equal to a preset statistical threshold, the steps of obtaining the time series operation data of the wind turbine generator set, clustering the time series operation data based on the flow clustering algorithm to obtain multiple microclusters, determining the global stability domain based on the central cores of the multiple microclusters and the number of data samples contained therein, and when the second global stability domain and the third global stability domain appear in sequence after the preset first global stability domain, judging that the wind turbine generator set has a gradual fault are repeatedly executed.

6. The wind turbine fault analysis method according to claim 1, It is characterized in that The clustering of the time series operation data based on the flow clustering algorithm to obtain a plurality of micro-clusters includes: Obtain data sample collection time and delay equation; Determine the weight corresponding to the data sample based on the time delay equation and the difference between the current time and the data sample collection time; Determine the cluster center and cluster radius corresponding to each data sample based on the weight, the data sample and the time when the data sample was collected; The time series operation data is divided into a plurality of micro-clusters based on the cluster center and the cluster radius.

7. A wind turbine fault analysis device, It is characterized in that The device comprises: A data acquisition module, used to acquire the time-series operation data of the wind turbine generator set, wherein the time-series operation data includes data samples recorded in time sequence with multi-dimensional features; A flow clustering module, used for clustering the time series operation data based on a flow clustering algorithm to obtain a plurality of micro-clusters; A stable domain determination module, used for determining a global stable domain based on the central cores of multiple microclusters and the number of data samples included; A fault judgment module, used for judging that a gradual fault occurs in the wind turbine generator set when a second global stability domain and a third global stability domain appear in sequence after a preset first global stability domain; The method of determining a global stable domain based on the central cores of a plurality of microclusters and the number of data samples contained therein comprises: when the core distances between the central cores of a plurality of stable microclusters are all less than a preset distance threshold, the plurality of stable microclusters are regarded as a global stable domain, wherein the stable microcluster is a microcluster whose central core and the number of data samples contained therein do not change within a preset time.

8. A smart device, It is characterized in that include: at least one processor; and, a memory communicatively coupled to the at least one processor; Wherein, a computer program is stored in the memory, and when the computer program is executed by the at least one processor, the wind turbine fault analysis method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium having a plurality of program codes stored therein, It is characterized in that The program code is suitable for being loaded and run by a processor to execute the wind turbine fault analysis method according to any one of claims 1 to 6.

Citation Information

Patent Citations

  • Method for diagnosing HVAC (heating, ventilation and air conditioning) system gradual failure based on deep learning

    CN109213127A

  • Early-stage fault early-warning method for synchronous motor stator

    CN110763997A

  • GIS-based multi-source partial discharge signal data stream clustering separation method

    CN114386499A

  • Device fault diagnosis method and apparatus, and electronic device and storage medium

    WO2023093177A1