Transfer component analysis based method and system for operating state identification of multiple wind turbines

Abstract

Provided in the present application are a transfer component analysis based method and system for operating state identification of multiple wind turbines, the method comprising: collecting historical data of SCADA systems of all wind turbines in a wind farm, and cleaning same; screening historical operating data of a reference wind turbine in a normal operating state; screening key variables that affect the operating states of the wind turbines; constructing a double hidden layer BP neural network based reference wind turbine normal behavior model; analyzing a wind turbine power residual on the basis of a confidence interval, and dividing the operating states of the reference wind turbine; integrating and analyzing data of various wind turbines except the reference turbine; calculating the wind turbine power residual, and, on the basis of the confidence interval, dividing the operating states of multiple wind turbines except the reference turbine. The system comprises a data collection module, a data cleaning module, a data screening module, a model construction module and a data analysis module, etc. The present application solves the problem of differences in data distribution across various wind turbines.

Description

Method and system for identifying operating state of multiple wind turbines based on transfer component analysis

[0001] The present application claims priority to the Chinese patent application No. 202410830302.5, filed on June 25, 2024, and entitled "Method and system for identifying operating state of multiple wind turbines based on transfer component analysis", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application belongs to the field of wind turbines, and in particular relates to a method and system for identifying operating state of multiple wind turbines based on transfer component analysis. BACKGROUND

[0003] Wind turbines often face various harsh operating environments, including extreme weather conditions, strong vibrations, and corrosive atmospheric environments. These adverse factors can cause the mechanical strength and operating performance of the components of the wind turbine to gradually decline, thereby affecting its power generation performance and increasing the frequency of faults. Therefore, effective identification and monitoring of the operating state of the wind turbine can help us discover potential problems and faults in a timely manner and take appropriate measures for maintenance and replacement, thereby ensuring the normal operation and power generation performance of the wind turbine. The evaluation of power generation performance and the fine management of wind farms have far-reaching influence and important significance.

[0004] Based on the SCADA operating data and normal behavior model of the wind turbine, the wind turbine operating state model is constructed by using multivariate state estimation technology, probability density statistics, polynomial regression fitting method, fuzzy comprehensive evaluation, neural network depth, etc. This is a common method for identifying the operating state of the wind turbine.

[0005] However, current research mainly focuses on the state division of a single turbine. However, due to the problem of data distribution difference between different wind turbines, if the trained normal behavior model of a single wind turbine is directly applied to the operating state identification of multiple wind turbines, the identification accuracy may be low. In order to improve the identification accuracy, it is necessary to recalculate the model parameters to divide the operating state of different wind turbines, but this will cause waste of computing resources. The current method cannot be effectively applied to the operating state identification of multiple wind turbines. SUMMARY

[0006] The present application aims to provide a method for identifying the operating state of multiple wind turbines based on transfer component analysis. This method constructs a model for identifying the operating state of multiple wind turbines based on transfer component analysis, which solves the problem of data distribution difference between different wind turbines, optimizes the resources and time for calculating network parameters, and thus reduces unnecessary waste of time and resources.

[0007] The application is implemented by adopting the following technical scheme:

[0008] The method for identifying the operation state of multiple wind turbines based on the transfer component analysis comprises the following steps:

[0009] Collecting historical data of SCADA systems of all wind turbines in a wind farm;

[0010] Cleaning the collected historical data of SCADA systems of all wind turbines to remove abnormal data and noise;

[0011] Selecting a turbine as a reference turbine, and screening historical operation data of the reference turbine in a normal operation state based on the cleaned historical data and a temperature index of the wind turbine;

[0012] Screening key influence variables of the operation state of the wind turbine based on the historical operation data of the reference turbine in the normal operation state by using the maximum mutual information number and the BP neural network method;

[0013] Taking the key influence variables of the operation state of the wind turbine as input, and constructing a normal behavior model of the reference turbine based on the BP double-hidden layer neural network;

[0014] Dividing the operation state of the reference turbine based on confidence interval analysis of a power residual of the wind turbine;

[0015] Based on the normal behavior model of the reference turbine and the transfer component analysis, a data distribution assimilation model of multiple wind turbines is constructed to integrate and analyze data of different wind turbines except the reference turbine;

[0016] Calculating a power residual of the wind turbine, and dividing the operation state of multiple wind turbines except the reference turbine based on a confidence interval.

[0017] Optionally, the application is improved in that the historical data of SCADA systems of all wind turbines in a wind farm are collected, and the collection comprises the following steps:

[0018] The sampling period is 1 year, the data resolution is second-level or minute-level, and the collected data at least includes generator speed, main shaft speed, wind speed, blade angle, temperature of a bearing at a driving end of an intermediate shaft of a gearbox, temperature of a bearing at a non-driving end of the intermediate shaft of the gearbox, oil pressure of a distributor position of the gearbox, oil pressure of an oil pump suction port of the gearbox, temperature of a V-phase coil of a stator of the generator, temperature of a U-phase coil of the stator of the generator, and temperature of an oil pool of the gearbox.

[0019] Optionally, the application is improved in that the collected historical data of SCADA systems of all wind turbines are cleaned, and the cleaning comprises the following steps: based on wind speed and power data, abnormal power data in the operation of the turbine is screened out by using the quartile method.

[0020] An optional improvement in this application is that, when filtering historical operating data under normal operating conditions of the reference unit, temperature index is used as the criterion to obtain the normal operating condition data of the wind turbine unit.

[0021] An optional improvement to this application is that, based on historical operating data under normal operating conditions of the reference unit, key influencing variables of the wind turbine operating state are screened using the maximum mutual information number and BP neural network method, including:

[0022] The formula for calculating the maximum mutual information between different variables and power of a wind turbine is as follows:

[0023] In the formula, y is the target variable, x is the variable to be screened, p(x,y) is the joint probability distribution of x and y; p(x) and p(y) are the marginal probabilities distribution of x and y, respectively.

[0024] Using key influencing variables of wind turbine operating status as input, a normal behavior model of wind turbine based on BP double hidden layer neural network is constructed.

[0025] The key influencing variables of different orders will be formed by eliminating the last digit of the maximum mutual information coefficient, and the key influencing variables of the wind turbine operating status will be determined based on the regression error assessment index.

[0026] A multi-wind turbine operating status identification system based on migration component analysis includes:

[0027] The data collection module is used to collect historical data from the SCADA systems of all wind turbines within the wind farm.

[0028] The data cleaning module is used to clean all historical data collected from the SCADA systems of all wind turbine units, removing abnormal data and noise;

[0029] The first data filtering module is used to select a unit as a reference unit and filter the historical operating data of the reference unit under normal operating conditions based on the historical data after cleaning and the temperature index of the wind turbine unit.

[0030] The second data filtering module is used to filter key influencing variables of wind turbine operating status based on historical operating data under normal operating conditions of the reference unit, using the maximum mutual information number and BP neural network method.

[0031] The model building module is used to construct a reference unit normal behavior model based on a BP double hidden layer neural network, using key influencing variables of wind turbine operating status as input.

[0032] The first data analysis module is used to analyze the power residual of wind turbine units based on confidence intervals and to classify the operating status of reference units.

[0033] The second data analysis module is configured to construct a wind turbine operation data distribution assimilation model based on the reference wind turbine normal behavior model and the transfer component analysis, and integrate and analyze data of different wind turbines except the reference wind turbine.

[0034] The third data analysis module is configured to calculate wind turbine power residuals, and divide operation states of multiple wind turbines except the reference wind turbine based on a confidence interval.

[0035] A computer readable storage medium stores a computer program, and the computer program, when executed by a processor, implements steps of the method for identifying operation states of multiple wind turbines based on transfer component analysis.

[0036] The method and system for identifying operation states of multiple wind turbines based on transfer component analysis have at least the following beneficial technical effects:

[0037] The method and system for identifying operation states of multiple wind turbines based on transfer component analysis can effectively avoid the tedious process of repeatedly adjusting neural network parameters, greatly improve the efficiency of model training, ensure the accuracy of the model, and realize the dual optimization of model performance and computing efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0038] FIG. 1 is a flowchart of the method for identifying operation states of multiple wind turbines based on transfer component analysis.

[0039] FIG. 2 is a structural block diagram of the system for identifying operation states of multiple wind turbines based on transfer component analysis. DETAILED DESCRIPTION

[0040] In the following, only certain exemplary embodiments are simply described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the present application. Therefore, the drawings and the description are considered to be essentially exemplary rather than limiting.

[0041] In the description of the application, it should be understood that the orientation or positional relationship indicated by terms such as "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like is based on the orientation or positional relationship shown in the drawings, and is only for the purpose of facilitating the description of the application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the application.

[0042] In addition, the terms "first", "second", "third", etc. are used only for descriptive purposes and should not be construed as indicating or implying relative importance or an indicated number of technical features. Therefore, the features defined as "first", "second", etc. can explicitly or implicitly include one or more of the features. In the description of the application, the meaning of "a plurality of" is two or more, unless otherwise explicitly specified and limited.

[0043] In this application, unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting", "fixing" and the like should be understood in a broad sense, for example, it can be fixed connection, or detachable connection, or integrated; it can be mechanical connection, or electrical connection, or communication; it can be directly connected, or indirectly connected through intermediate medium, or the internal communication of two elements or the interaction relationship between two elements. For those skilled in the art, the specific meaning of the above terms in this application can be understood according to the specific circumstances.

[0044] In this application, unless otherwise explicitly specified and limited, the first feature "above" or "below" the second feature can include the direct contact of the first and second features, or the contact of the first and second features through another feature between them. Moreover, the first feature "above", "above" and "above" the second feature includes the first feature directly above and obliquely above the second feature, or only indicates that the horizontal height of the first feature is higher than that of the second feature. The first feature "below", "below" and "below" the second feature includes the first feature directly above and obliquely above the second feature, or only indicates that the horizontal height of the first feature is less than that of the second feature.

[0045] It should also be understood that the terms used in the specification of the application are only for the purpose of describing specific embodiments and do not intend to limit the application. As used in the specification and the appended claims of the application, unless otherwise clearly indicated by the context, the singular form "a", "an" and "the" is intended to include the plural form.

[0046] It should also be optionally understood that the term "and / or" used in the description and the appended claims of the present application means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0047] Various structural diagrams according to embodiments of the present disclosure are shown in the accompanying drawings. These diagrams are not drawn to scale, in which certain details are shown exaggerated in scale for purposes of clarity and understanding, and certain other details are omitted. The shapes and relative sizes of the various regions, layers, and elements illustrated in the drawings are exemplary only and can vary in actual implementation, depending on, for example, manufacturing techniques and / or technology restraints. The skilled person can design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0048] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0049] Embodiment 1

[0050] The method for identifying the operating state of multiple wind turbines based on the transfer component analysis provided in this embodiment includes:

[0051] Collecting historical data of SCADA systems of all wind turbines in a wind farm;

[0052] Cleaning the collected historical data of SCADA systems of all wind turbines to remove abnormal data and noise;

[0053] Selecting one turbine as a reference turbine, and based on the cleaned historical data and the temperature index of the wind turbine, screening historical operating data of the reference turbine in a normal operating state;

[0054] Based on the historical operating data of the reference turbine in the normal operating state, screening key influence variables of the operating state of the wind turbine through the maximum mutual information number and the BP neural network method;

[0055] Taking the key influence variables of the operating state of the wind turbine as input, constructing a normal behavior model of the reference turbine based on a BP double-hidden layer neural network;

[0056] Based on the confidence interval, analyzing the power residual of the wind turbine, and dividing the operating state of the reference turbine;

[0057] Based on the normal behavior model of the reference turbine and the transfer component analysis, constructing a multiple wind turbine operating data distribution assimilation model, and integrating and analyzing the data of different wind turbines except the reference turbine;

[0058] Calculating the power residual of the wind turbine, and based on the confidence interval, dividing the operating state of multiple wind turbines except the reference turbine.

[0059] In the embodiment, the historical data of the SCADA system of all wind turbines in the wind farm is collected, including: the sampling period is 1 year, the data resolution is second or minute level, and the collected data at least includes the generator speed, the main shaft speed, the wind speed, the blade angle, the temperature of the intermediate shaft driving end bearing of the gearbox, the temperature of the intermediate shaft non-driving end bearing of the gearbox, the oil pressure of the gearbox distributor position, the oil pressure of the gearbox oil pump suction port, the temperature of the generator stator V-phase coil, the temperature of the generator stator U-phase coil, and the temperature of the gearbox oil pool.

[0060] In the embodiment, the collected historical data of the SCADA system of all wind turbines is cleaned, including: through the quartile method, the power abnormal data in the operation of the unit is screened out based on the wind speed and power data.

[0061] Embodiment 2

[0062] Referring to FIG. 1, the method for identifying the operation state of multiple wind turbines based on the transfer component analysis provided in the embodiment includes the following specific steps:

[0063] Step 1: Collecting the historical data of the SCADA system of the wind turbine.

[0064] Step 2: Cleaning the historical data of the SCADA system of the wind turbine by using the bidirectional quartile method to remove abnormal data and noise.

[0065] Step 3: Screening the historical operation data of the wind turbine in the normal operation state based on the temperature index.

[0066] Step 4: Screening the key influence variables of the operation state of the wind turbine by using the maximum mutual information number and the BP neural network method.

[0067] Step 5: Taking the key influence variables of the operation state of the wind turbine as the input, constructing the normal behavior model of the wind turbine based on the BP double-hidden layer neural network.

[0068] Step 6: Dividing the operation state of a single wind turbine based on the confidence interval analysis of the power residual of the wind turbine.

[0069] Step 7: Using the data of the single unit in the divided operation state and other unit data, constructing the distribution assimilation model of the operation data of multiple wind turbines based on the transfer component analysis. The data of different wind turbines is integrated and analyzed.

[0070] Step 8: Calculating the power residual of the wind turbine, and dividing the operation state of multiple wind turbines based on the confidence interval.

[0071] The historical data of the SCADA system of the wind turbine mentioned in the step 1 generally has a sampling period of 1 year and a data resolution of second or minute level.

[0072] The step 4 specifically comprises the following steps:

[0073] Step 4-1: Calculate the maximum mutual information number between different variables of the wind turbine and power. The specific formula is as follows:

[0074] In the formula, y is the target variable, x is the variable to be screened, p(x, y) is the joint distribution probability of x and y; p(x) and p(y) are the marginal distribution probabilities of x and y, respectively.

[0075] Step 4-2: Take the key influence variables of the wind turbine operating state as input, and construct a wind turbine normal behavior model based on a BP double-hidden layer neural network.

[0076] Step 4-3: According to the maximum mutual information coefficient last elimination method, different order (n order, n-1 order, n-2 order … 3 order) potential key influence variables are formed, and based on the regression error evaluation index, the key influence variables of the wind turbine operating state are determined.

[0077] Tables 1 and 2 show the operating state division results and power residual error results after migration learning according to the method, and it can be seen that the method improves the model precision and makes the model more stable, verifying the effectiveness and applicability of the proposed model.

[0078] Table 1: Comparison of operating state division results

[0079] Table 2: Comparison of power residual error analysis indicators

[0080] Example 3

[0081] Referring to FIG. 2, the multi-wind turbine operating state recognition system based on migration component analysis provided in the embodiment comprises:

[0082] A data collection module for collecting historical data of all wind turbine SCADA systems in the wind farm;

[0083] A data cleaning module for cleaning the collected historical data of all wind turbine SCADA systems to remove abnormal data and noise;

[0084] A first data screening module for selecting a unit as a reference unit, and screening historical operating data of the reference unit in normal operating state based on the cleaned historical data and the wind turbine temperature index;

[0085] A second data screening module for screening key influence variables of the wind turbine operating state based on the historical operating data of the reference unit in normal operating state through the maximum mutual information number and the BP neural network method;

[0086] a model construction module configured to construct a reference unit normal behavior model based on a BP double-hidden layer neural network, with key influence variables of wind turbine operation states as input;

[0087] a first data analysis module configured to analyze wind turbine power residuals based on a confidence interval, and divide reference unit operation states;

[0088] a second data analysis module configured to construct a multi-wind turbine operation data distribution assimilation model based on the reference unit normal behavior model and a transfer component analysis, and integrate and analyze data of different wind turbines except the reference unit;

[0089] a third data analysis module configured to calculate wind turbine power residuals, and divide operation states of multiple wind turbines except the reference unit based on a confidence interval.

[0090] Embodiment 4

[0091] The computer readable storage medium provided in the embodiment of the present application stores a computer program, and the computer program, when executed by a processor, implements the steps of the multi-wind turbine operation state recognition method based on the transfer component analysis.

[0092] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can be in the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0093] The present application is described with reference to flowcharts and / or block diagrams of the methods, systems and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices produce a system for implementing the functions specified in one or more flows in the flowcharts and / or one or more blocks in the block diagrams.

[0094] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart or flowsheets and / or block or blocks of the block diagrams.

[0095] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart or flowsheets and / or block or blocks of the block diagrams.

[0096] The principles and main features of the present application have been shown and described above, and the advantages of the present application have been shown and described above, and it is apparent to those skilled in the art that the present application is not limited to the details of the above-described exemplary embodiments, and the present application can be realized in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered in all respects as illustrative and not restrictive, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended to encompass all changes falling within the meaning and scope of the equivalent elements of the claims. Any reference numerals in the claims should not be considered as limiting the scope of the claims to which they relate.

[0097] Furthermore, it should be understood that, although the present specification is described in terms of embodiments, not every implementation embodies only one independent technical solution, and the present specification is described in this way only for the sake of clarity, and those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that those skilled in the art can understand. The above is only to illustrate the technical idea of the present application, and cannot limit the protection scope of the present application, and any modification made on the basis of the technical solutions according to the technical idea of the present application falls within the protection scope of the claims of the present application.

Claims

1. A method for recognizing the operating state of multiple wind turbines based on migration component analysis, characterized in that, The method comprises the following steps: collecting historical data of SCADA systems of all wind turbines in a wind farm; cleaning the collected historical data of SCADA systems of all wind turbines to remove abnormal data and noise; selecting a wind turbine as a reference wind turbine, and screening historical operation data of the reference wind turbine in a normal operation state based on the cleaned historical data and a temperature index of the wind turbine; screening key influence variables of the wind turbine in an operation state based on the historical operation data of the reference wind turbine in the normal operation state by using a maximum mutual information number and a BP neural network method; constructing a normal behavior model of the reference wind turbine based on a BP double-hidden layer neural network, with the key influence variables of the wind turbine in the operation state as inputs; dividing the operation state of the reference wind turbine based on a confidence interval analysis of a power residual error of the wind turbine; constructing a multi-wind turbine operation data distribution assimilation model based on the normal behavior model of the reference wind turbine and a migration component analysis, and integrating and analyzing data of different wind turbines except the reference wind turbine; calculating a power residual error of the wind turbine, and dividing the operation state of the wind turbine based on a confidence interval.

2. The method of claim 1, wherein, The method comprises the following steps: collecting historical data of SCADA systems of all wind turbines in a wind farm, including:

3. The method of claim 1, wherein, the sampling period is 1 year, the data resolution is second-level or minute-level, and the collected data at least includes generator speed, main shaft speed, wind speed, blade angle, temperature of a bearing at a driving end of an intermediate shaft of a gearbox, temperature of a bearing at a non-driving end of the intermediate shaft of the gearbox, oil pressure of a distributor position of the gearbox, oil pressure of an oil pump suction port of the gearbox, temperature of a V-phase coil of a generator stator, temperature of a U-phase coil of the generator stator, and temperature of a gearbox oil pool.

4. The method of claim 1, wherein, cleaning the collected historical data of SCADA systems of all wind turbines, including:

5. The method of claim 1, wherein, screening abnormal power data in operation of the wind turbine based on wind speed and power data by using a quartile method. The maximum mutual information number between different variables of a wind turbine and power is calculated, and the specific formula is as follows: When screening historical operation data of the reference wind turbine in a normal operation state, the temperature index is used as a judgment basis to obtain wind turbine normal operation state data. Screening key influence variables of the wind turbine in an operation state based on historical operation data of the reference wind turbine in a normal operation state by using a maximum mutual information number and a BP neural network method, including: In the formula, y is a target variable, x is a variable to be screened, p(x, y) is a joint distribution probability of x and y; p(x) and p(y) are marginal distribution probabilities of x and y, respectively; 6. A system for identifying the operating state of a plurality of wind turbines based on a migration component analysis, characterized in that constructing a normal behavior model of the wind turbine based on a BP double-hidden layer neural network, with the key influence variables of the wind turbine in the operation state as inputs; composing different order potential key influence variables by using a maximum mutual information coefficient last elimination method, and determining the key influence variables of the wind turbine in the operation state based on a regression error evaluation index. The method comprises the following steps: a data collection module is configured to collect historical data of SCADA systems of all wind turbines in a wind farm; a data cleaning module is configured to clean the collected historical data of SCADA systems of all wind turbines to remove abnormal data and noise; a first data screening module is configured to select a wind turbine as a reference wind turbine, and screen historical operation data of the reference wind turbine in a normal operation state based on cleaned historical data and a temperature index of the wind turbine; The second data screening module is configured to screen key influence variables of the wind turbine operating state based on historical operating data of the reference unit in a normal operating state by using a maximum mutual information number and a BP neural network method. The model construction module is configured to construct a reference unit normal behavior model based on a BP double-hidden layer neural network by taking the key influence variables of the wind turbine operating state as input. The first data analysis module is configured to analyze wind turbine power residuals based on a confidence interval to divide the operating state of the reference unit. The second data analysis module is configured to construct a multi-wind turbine operating data distribution assimilation model based on the reference unit normal behavior model and a transfer component analysis to integrate and analyze data of different wind turbines except the reference unit. The third data analysis module is configured to calculate wind turbine power residuals and divide the operating state of multiple wind turbines except the reference unit based on a confidence interval.

7. The system of claim 6, wherein, In the data collection module, historical data of SCADA systems of all wind turbines in a wind farm are collected, including: The sampling period is 1 year, the data resolution is second-level or minute-level, and the collected data at least includes generator speed, main shaft speed, wind speed, blade angle, gear box intermediate shaft driving end bearing temperature, gear box intermediate shaft non-driving end bearing temperature, gear box distributor position oil pressure, gear box oil pump suction oil pressure, generator stator V-phase coil temperature, generator stator U-phase coil temperature, and gear box oil pool temperature.

8. The system for operating condition recognition of multiple wind turbines based on shift component analysis according to claim 6, characterized in that, In the data cleaning module, the collected historical data of SCADA systems of all wind turbines are cleaned, including: abnormal power data in the operation of the unit are screened out based on wind speed and power data by using a quartile method.

9. The system for operating condition recognition of multiple wind turbines based on shift component analysis according to claim 6, characterized in that, In the second data screening module, key influence variables of the wind turbine operating state are screened based on historical operating data of the reference unit in a normal operating state by using a maximum mutual information number and a BP neural network method, including: The maximum mutual information number between different variables of the wind turbine and the power is calculated, and the specific formula is as follows: In the formula, y is a target variable, x is a variable to be screened, p(x, y) is a joint distribution probability of x and y; p(x) and p(y) are marginal distribution probabilities of x and y, respectively; The wind turbine normal behavior model based on the BP double-hidden layer neural network is constructed by taking the key influence variables of the wind turbine operating state as input. Different order potential key influence variables are formed by using a maximum mutual information coefficient last elimination method, and the key influence variables of the wind turbine operating state are determined based on a regression error evaluation index.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program, when executed by the processor, implements the steps of the multi-wind turbine operating state recognition method based on the transfer component analysis in any one of claims 1 to 5.

Images