CN117703690A - Wind generating set health state assessment method and system - Google Patents

Abstract

The invention discloses a health state evaluation method of a wind generating set, which comprises the steps of dividing acquired monitoring data of the wind generating set according to label quantity, analysis variables and working condition variables, calculating correlation between the analysis variables and the working condition variables, and acquiring a working condition variable set corresponding to the analysis variables according to the correlation between the analysis variables and the working condition variables; dividing the working condition variable values into a plurality of sections by adopting a box diagram method according to the working condition variables of the obtained monitoring data of the wind generating set, and taking the working condition variable values as judgment bases of different working conditions of the running state of the set; the invention adopts the equipment combination working condition analysis method to analyze the health state of the real-time monitoring data of the unit under different working condition combination conditions, and can solve the problems of discrete health state, coupling of operation working conditions, variable operation and maintenance decisions and the like of the equipment.

Description

A method and system for health status assessment of wind turbines

Technical field

The invention belongs to the field of generator status monitoring and evaluation, and specifically relates to a wind turbine health status evaluation method and system.

Background technique

At present, wind energy, as a clean energy, plays an increasingly important role in improving China's energy structure, but the accompanying issues of wind farm safety and economic benefits have gradually attracted attention. As a typical distributed complex electromechanical system, wind turbines have the typical characteristics of physical dispersion, logical integration, multi-agent, and multi-modality. Their service health status is affected by the health status of discrete unit equipment in the area, individual operating conditions, and differentiation. The comprehensive influence of multiple strongly coupled uncertain factors such as production scheduling and subjective operation and maintenance decisions presents complex coupling and aggregation characteristics in time and space. SCADA (Supervisory Control And Data Acquisition, Data Acquisition and Supervisory Control System) is a computer-based DCS and electric power automation monitoring system. It has a wide range of applications and can be used in electric power, metallurgy, petroleum, chemical industry, gas, railway and other fields. Data collection and monitoring control as well as process control and many other fields. As the most important subsystem of the energy management system (EMS system), it has the advantages of complete information, improved efficiency, correct grasp of system operating status, accelerated decision-making, and can help quickly diagnose system fault conditions. It has become an indispensable part of power dispatching. Missing tools. It plays an irreplaceable role in improving the reliability, security and economic benefits of power grid operation, reducing the dispatching burden, realizing the automation and modernization of power dispatching, and improving the efficiency and level of dispatching.

The alarm mechanism of the SCADA system commonly used in wind turbines plays an important role in the safe and stable operation of the unit. However, the current SCADA alarm mechanism relies heavily on the empirical judgment of specific monitored variables, and the alarms are often lagging and one-sided, and provide early warning of abnormalities. Weak ability. At the same time, the current analysis of wind turbine monitoring systems is mostly qualitative analysis, which can only determine normality and abnormality, and only has a single alarm across the board, but cannot provide the degradation process and trend of system monitoring variables, and cannot classify alarms caused by different abnormalities. , it is difficult to issue an alarm in time at or even before the moment when the abnormality actually occurs. Methods and systems that can conduct accurate quantitative analysis from monitoring variables to the health status of the entire machine are needed.

Contents of the invention

The purpose of the present invention is to provide a wind turbine health status assessment method and system to overcome the shortcomings of traditional wind turbine health status analysis and early warning methods, accurately analyze the wind turbine set before SCADA alarms, and realize monitoring variables and the entire machine. health status warning.

A method for health status assessment of wind turbines, including the following steps:

S1. Divide the obtained monitoring data of the wind turbine generator according to the label quantity, analysis variables and working condition variables, calculate the correlation between the analysis variables and the working condition variables, and obtain the data based on the correlation between the analysis variables and the working condition variables. The set of working condition variables corresponding to the analysis variables;

S2, use the box plot method to divide the working condition variable values into multiple intervals based on the working condition variables of the obtained monitoring data of the wind turbine unit, which will be used as the basis for judging different working conditions of the unit's operating status;

S3: Obtain different working condition combinations based on the multiple intervals divided in step S2, screen the monitoring data under the corresponding working condition combinations for each analysis variable, and determine the variable threshold center, so as to obtain the health level of the analyzed variables at each moment, and further obtain the overall The health level of the machine at different times and the health level of the entire machine every day.

Preferably, the copula nonlinear analysis method is used to calculate the nonlinear correlation between each analysis variable and each working condition variable, construct a correlation coefficient matrix between the analysis variable and the working condition variable, and set the threshold based on the strong correlation determination , identify the set of operating condition variables most relevant to each analysis variable.

Preferably, the box plot method is used to calculate each box plot characteristic value of the working condition variable monitoring data, arrange each group of data from small to large, find the number that divides the group of data into four equal parts, and divide the first and fourth The quantile, median, and third quartile are used as the endpoints of the interval division to initially divide the numerical interval of the working condition variable.

Preferably, assume that the data of an analysis variable under a certain combination of working conditions is x ₁ , x ₂ ,..., x _N , where N is the data length, then

Among them, C is the center value under the combined working conditions at that time, and M is the standard deviation under the combined working conditions at that time.

Preferably, the monitoring data of analysis variables and working condition variables at the time to be analyzed are collected, and based on the corresponding working condition threshold center determined above, the degree of deviation of the monitoring value of the analysis variable at that time from the center value is quantitatively calculated to determine the health of the analysis variable at that time. Level; analyze the variable health level and set the threshold based on normal working conditions, as the basis for monitoring variable alarms.

Preferably, after obtaining the health levels of different analysis variables at each moment, the health levels of all the analysis variables every day are integrated to obtain the health levels of the entire machine at different times, and further to obtain the health levels of the entire machine each day, and to construct changes in the health level of the entire machine. curve.

A wind turbine health status assessment system, including a variable division module, an interval division module and an evaluation module;

The variable division module divides the obtained monitoring data of wind turbines according to the label quantity, analysis variables and working condition variables, calculates the correlation between the analysis variables and the working condition variables, and calculates the correlation between the analysis variables and the working condition variables according to the correlation between the analysis variables and the working condition variables. Sexually obtain the set of working condition variables corresponding to the analysis variables;

The interval division module will use the box plot method to divide the operating condition variable values into multiple intervals based on the obtained operating condition variables of the wind turbine monitoring data, which will be used as a basis for judging different operating conditions of the unit's operating status;

The evaluation module obtains different working condition combinations based on multiple divided intervals, screens the monitoring data under the corresponding working condition combinations for each analyzed variable, and determines the variable threshold center. This way, the health level of the analyzed variables at each moment can be obtained, and further the complete machine can be obtained. The health level at different times and the health level of the entire machine every day.

Preferably, the copula nonlinear analysis method is used to calculate the nonlinear correlation between each analysis variable and each working condition variable, construct a correlation coefficient matrix between the analysis variable and the working condition variable, and set the threshold based on the strong correlation determination , identify the set of operating condition variables most relevant to each analysis variable.

Preferably, the box plot method is used to calculate each box plot characteristic value of the working condition variable monitoring data, arrange each group of data from small to large, find the number that divides the group of data into four equal parts, and divide the first and fourth The quantile, median, and third quartile are used as the endpoints of the interval division to initially divide the numerical interval of the working condition variable.

Preferably, assume that the data of an analysis variable under a certain combination of working conditions is x ₁ , x ₂ ,..., x _N , where N is the data length, then

Among them, C is the center value under the combined working conditions at that time, and M is the standard deviation under the combined working conditions at that time.

Compared with the existing technology, the present invention has the following beneficial technical effects:

The present invention provides a method for evaluating the health status of a wind turbine generator set. By dividing the acquired monitoring data of the wind turbine generator set according to the label quantity, analysis variables and working condition variables, the correlation between the analytical variables and the working condition variables is calculated. According to The correlation between the analysis variables and the working condition variables is used to obtain the set of working condition variables corresponding to the analysis variables; the working condition variables according to the obtained monitoring data of the wind turbine are divided into multiple values using the box plot method. Each interval is used as the basis for judging different working conditions of the unit's operating status; different working condition combinations are obtained according to the multiple divided intervals, the monitoring data under the corresponding working condition combinations of each analysis variable are screened, and the variable threshold center is determined, and each Analyze the health level of variables at all times to further obtain the health level of the entire machine at different times and the health level of the entire machine every day. The present invention adopts an equipment combination working condition analysis method to analyze the health status of real-time monitoring data of the unit under different working condition combinations, and can solve problems such as discrete equipment health status, coupling of operating conditions, and changeable operation and maintenance decisions.

The present invention adopts an equipment combination working condition analysis method to analyze the health status of real-time monitoring data of the unit under different working condition combinations, and can solve problems such as discrete equipment health status, coupling of operating conditions, and changeable operation and maintenance decisions. Compared with the existing wind turbine monitoring and alarm mechanism, which relies on the setting premise of prior knowledge, the present invention proposes a health status evaluation and alarm mechanism based on the distribution law of the data itself, so that the wind turbine monitoring and alarm has continuity and traceability. ;

Compared with the traditional SCADA alarm mechanism, this invention prominently considers mining and extracting system status characteristics from the distribution law of the monitoring variable data itself, reducing the impact of personnel operations and operation and maintenance decisions on monitoring judgments, and has stronger versatility and stability.

Description of the drawings

Figure 1 is a flow chart of the wind turbine health status assessment method of the present invention.

Detailed ways

In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only These are some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts should fall within the scope of protection of the present invention.

It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances so that the embodiments of the invention described herein are capable of being practiced in sequences other than those illustrated or described herein. Furthermore, the terms "include" and "having" and any variations thereof are intended to cover non-exclusive inclusions, e.g., a process, method, system, product, or apparatus that encompasses a series of steps or units and need not be limited to those explicitly listed. Those steps or elements may instead include other steps or elements not expressly listed or inherent to the process, method, product or apparatus.

As shown in Figure 1, the present invention provides a method for assessing the health status of a wind turbine generator set, which includes the following steps:

S1: Divide the obtained monitoring data of the wind turbine generator according to the tag volume, analysis variables and working condition variables, calculate the correlation between the analysis variables and the working condition variables, and obtain the data based on the correlation between the analysis variables and the working condition variables. The set of working condition variables corresponding to the analysis variables;

S2, use the box plot method to divide the working condition variable values into multiple intervals based on the working condition variables of the obtained monitoring data of the wind turbine unit, which will be used as the basis for judging different working conditions of the unit's operating status;

S3: Obtain different working condition combinations based on the multiple intervals divided in step S2, screen the monitoring data under the corresponding working condition combinations for each analysis variable, and determine the variable threshold center, so as to obtain the health level of the analyzed variables at each moment, and further obtain the overall The health level of the machine at different times and the health level of the entire machine every day.

The present invention adopts an equipment combination working condition analysis method to analyze the health status of real-time monitoring data of the unit under different working condition combinations, and can solve problems such as discrete equipment health status, coupling of operating conditions, and changeable operation and maintenance decisions.

In step S1, the system monitoring data variable categories are divided: the wind turbine monitoring variables are collected, and based on the actual indicators and physical meanings of the system monitoring variables, combined with the presence or absence of monitoring data and the wind turbine operating status monitoring and analysis experience, the operating status quantities of the unit are distinguished. Working condition quantity, the monitoring value is divided into label quantity, analysis variable and working condition variable.

Identification of working condition variables based on copula nonlinear analysis: Use the copula nonlinear analysis method to calculate the nonlinear correlation between each analysis variable and each working condition variable, and construct a correlation coefficient matrix between the analysis variables and the working condition variables. Set thresholds based on strong correlation determination to identify the set of operating condition variables most relevant to each analysis variable.

_{Suppose X 1 , ​ ​ ​ ​ ​} is H(x ₁ ,x ₂ ,…,x _N ), then the Copula nonlinear correlation calculation can be obtained from the following formula,

C(u ₁ ,u ₂ ,…,u _N )=H[F ₁ ^-1 (u ₁ ),F ₂ ^-1 (u ₂ ),…,F _N ^-1 (u _N )]

Among them, C(u ₁ ,u ₂ ,…,u _N ) is the Copula function, and F ^-1 (u) represents the inverse function of F(u).

Wind turbine operating condition classification steps: For each identified operating condition variable, the box plot method is used to divide the operating condition variable values into multiple intervals as a basis for judging different operating conditions of the unit.

Division of operating condition intervals based on boxplots: Use the boxplot method to calculate the characteristic values of each boxplot of the operating condition variable monitoring data, arrange each group of data from small to large, and find that the group of data can be divided into four equal parts. The first quartile Q ₁ , the median Q ₂ , and the third quartile Q ₃ are used as the end points of the interval division, and the numerical interval of the working condition variable is initially divided.

Field-oriented adjustment of the working condition interval division: Based on the working condition interval divided in the above method, and based on expert experience, monitoring system logic and the actual working status of the unit, the working condition interval division is adjusted.

In step S3, the analysis variable threshold center is determined: filter the monitoring data of each analysis variable under the working condition combination according to the working condition combination, and calculate the center value and standard deviation of each analytical variable under each working condition as different working conditions. The basis for judging the lower variables and unit status.

Suppose the data of an analysis variable under a certain combination of working conditions are x ₁ , x ₂ ,..., x _N , where N is the data length, then

Among them, C is the center value under the combined working conditions at that time, and M is the standard deviation under the combined working conditions at that time.

Determine the health level of analysis variables: Collect the analysis variables and working condition variable monitoring data at the moment to be analyzed, and based on the corresponding working condition threshold center determined above, quantitatively calculate the deviation degree of the analysis variable monitoring value at that moment from the center value, and determine the analysis at that moment. The health level of the variable; analyze the variable health level and set the threshold based on normal working conditions as a basis for monitoring variable alarms.

Among them, D is the monitoring data, C is the central value under the combined working conditions at that moment, M is the standard deviation under the combined working conditions at that moment, Result represents the health level of the analysis variable at that moment, and its physical meaning is monitoring The degree to which the value deviates from the central value. The larger the value, the further it deviates from the central value, the worse the health condition.

Determination of the health level of the whole machine: After obtaining the health level of different analysis variables at each moment, the health levels of all the analysis variables every day are integrated to obtain the health level of the whole machine at different times, and further to obtain the health level of the whole machine every day, and construct the whole machine. Health level change curve.

Compared with the existing wind turbine monitoring and alarm mechanism, which relies on the setting premise of prior knowledge, the present invention proposes a wind turbine health status assessment mechanism based on the distribution law of the data itself, so that the wind turbine monitoring and alarm has continuity and traceability. sex;

Compared with the traditional SCADA alarm mechanism, this invention prominently considers mining and extracting system status characteristics from the distribution law of the monitoring variable data itself, reducing the impact of personnel operations and operation and maintenance decisions on monitoring judgments, and has stronger versatility and stability;

The working condition division and threshold setting of the present invention are based on algorithms and can be adjusted at any time based on the actual situation on site. It has stronger adaptability to complex and changeable unit operating conditions and operation and maintenance decisions;

The scheme of the invention is reasonable and easy to implement, and provides a flexible algorithm system, which can give full play to the advantages of data analysis and provide a good foundation for subsequent corrections and improvements.

The present invention provides a method for evaluating the health status of a wind turbine generator set. Aiming at the problems of discrete equipment health status, coupled operating conditions, and changeable operation and maintenance decisions, the present invention adopts an equipment combination working condition analysis method. Under different working condition combination conditions, Perform health status analysis on real-time monitoring data of the unit. Since the real-time monitoring data when the unit is running is often affected by changes in working conditions and presents different distributions under different working conditions, the present invention focuses on distinguishing and analyzing the distribution characteristics of the operating status variables of the unit under different working conditions, so as to Eliminate the influence of different working conditions on variable analysis. At the same time, in view of the hysteresis, discontinuity and other characteristics of the traditional alarm mechanism, the present invention proposes a quantitative unit health status evaluation method, which adopts the health level calculation method of wind turbines from variables to the entire machine, and then constructs a system from variables to complete units based on the health level. The health status change curve of the whole machine is used as the criterion for alarm of the monitoring system.

Specifically, it includes the following steps:

S1, wind turbine monitoring variable identification step: collect historical monitoring data of wind turbines, divide the historical monitoring data into label quantities, analysis variables and working condition variables, and calculate the correlation between variables to obtain the working conditions corresponding to the analysis variables. Collection of variables. The tag quantity is used to distinguish the unit number and time.

(2) Steps for classifying wind turbine operating conditions. For each working condition variable identified in step 1), the box plot method is used to divide the working condition variable value into multiple intervals as a basis for judging the different working conditions of the unit's operating status, and the working condition intervals are analyzed based on the actual situation on site. Adjustment.

(3) Quantitative analysis and alarm procedures for unit health status. Based on the different working condition combinations obtained in step 2), screen the monitoring data under the corresponding working condition combination for each analysis variable, determine the variable threshold center, and quantitatively calculate the health level of the analysis variable at each moment, and further obtain the health level and health level of the entire machine at different times. The health level of the entire machine every day.

Includes the following steps:

1) Wind turbine monitoring variable identification steps. Collect wind turbine SCADA system monitoring data, divide the monitoring variable categories into label quantities, analysis variables and working condition variables, calculate the correlation between variables, and obtain the set of working condition variables corresponding to the analysis variables.

1.1) Classification of system monitoring data variables. Collect wind turbine monitoring variables. Based on the actual indicators and physical meanings of the system monitoring variables, combined with the presence or absence of monitoring data and the wind turbine operating status monitoring and analysis experience, the operating status quantity and working condition quantity of the unit are distinguished, and the monitoring values are divided into three Parts: label quantities, analysis variables, working condition variables.

1.2) Identification of working condition variables based on copula nonlinear analysis. The copula nonlinear analysis method is used to calculate the nonlinear correlation between each analysis variable and each working condition variable, construct the correlation coefficient matrix between the analysis variable and the working condition variable, set the threshold based on the strong correlation, identify and The set of operating condition variables most relevant to each analysis variable.

2) Steps for classifying wind turbine operating conditions. For each working condition variable identified in 1.2), the box plot method is used to divide the working condition variable value into multiple intervals as a basis for judging different working conditions of the unit's operating status, and the working condition intervals are adjusted according to the actual situation on site .

2.1) Division of working condition intervals based on box plot. The box plot method is used to calculate each box plot characteristic value of the working condition variable monitoring data, and the characteristic value is used as the end point of the interval division to initially divide the numerical interval of the working condition variable.

2.2) On-site working condition interval division and adjustment. Based on the working condition intervals divided in 2.1), the division of working condition intervals shall be adjusted based on expert experience, monitoring system logic and the actual working conditions of the unit.

3) Quantitative analysis and alarm procedures for unit health status. Screen the monitoring data under the corresponding working condition combination of each analysis variable according to different working condition combinations, determine the variable threshold center, quantitatively calculate the health level of the analyzed variables at each moment, and further obtain the health level of the whole machine at different times and the health level of the whole machine every day. health level.

3.1) Determine the threshold center of the analysis variable. Screen the monitoring data of each analysis variable under the working condition combination according to the working condition combination, and calculate the central value and standard deviation of each analytical variable under each working condition as the basis for judging the variables and unit status under different working conditions.

3.2) Analyze variable health level determination. Collect the monitoring data of analysis variables and working condition variables at the time that needs to be analyzed, and based on the corresponding working condition threshold center determined in 3.1), quantitatively calculate the deviation degree of the monitoring value of the analysis variable at that time from the center value, and determine the health level of the analysis variable at that time; Analyze variable health levels and set thresholds based on normal working conditions as the basis for monitoring variable alarms.

3.3) The health level of the entire machine is determined. After obtaining the health levels of different analysis variables at each moment, the health levels of all analysis variables for each day are combined to obtain the health levels of the entire machine at different times. The health level of the entire machine is further obtained for each day, and the health level change curve of the entire machine is constructed.

The present invention adopts an equipment combination working condition analysis method to analyze the health status of real-time monitoring data of the unit under different working condition combinations, and can solve problems such as discrete equipment health status, coupling of operating conditions, and changeable operation and maintenance decisions. Compared with the existing wind turbine monitoring and alarm mechanism, which relies on the setting premise of prior knowledge, the present invention proposes a health status evaluation and alarm mechanism based on the distribution law of the data itself, so that the wind turbine monitoring and alarm has continuity and traceability. ;

Compared with the traditional SCADA alarm mechanism, this invention prominently considers mining and extracting system status characteristics from the distribution law of the monitoring variable data itself, reducing the impact of personnel operations and operation and maintenance decisions on monitoring judgments, and has stronger versatility and stability;

The working condition division and threshold setting of the present invention are based on algorithms and can be adjusted at any time based on the actual situation on site. It has stronger adaptability to complex and changeable unit operating conditions and operation and maintenance decisions;

The scheme of the invention is reasonable and easy to implement, and provides a flexible algorithm system, which can give full play to the advantages of data analysis and provide a good foundation for subsequent corrections and improvements.

Claims (10)

1. The method for evaluating the health state of the wind generating set is characterized by comprising the following steps of:

s1, dividing the obtained monitoring data of the wind generating set according to the label quantity, the analysis variable and the working condition variable, calculating the correlation between the analysis variable and the working condition variable, and obtaining a working condition variable set corresponding to the analysis variable according to the correlation between the analysis variable and the working condition variable;

s2, dividing the working condition variable values into a plurality of sections by adopting a box diagram method according to the working condition variable of the obtained monitoring data of the wind generating set, and taking the working condition variable values as judgment bases of different working conditions of the running state of the set;

s3, obtaining different working condition combinations according to the intervals divided in the step S2, screening monitoring data of each analysis variable under the corresponding working condition combination, determining a variable threshold center, and obtaining the health grade of the analysis variable at each moment, and further obtaining the health grade of the whole machine at different moments and the health grade of the whole machine at each day.

2. The method for evaluating the health status of a wind turbine generator system according to claim 1, wherein a copula nonlinear analysis method is adopted, nonlinear correlations between each analysis variable and each working condition variable are calculated, a correlation coefficient matrix between the analysis variable and the working condition variable is constructed, a threshold is set according to strong correlation judgment, and a working condition variable set most correlated with each analysis variable is identified.

3. The method for evaluating the health status of a wind generating set according to claim 1, wherein the characteristic values of each box diagram of the condition variable monitoring data are calculated by adopting a box diagram method, each group of data is arranged from small to large, the number of quarters of the group of data is found, the first quartile, the median and the third quartile are used as endpoints of interval division, and the condition variable value interval is initially divided.

4. The method for evaluating the health status of a wind turbine generator system according to claim 1, wherein the data of an analysis variable under a certain combined condition is defined as x ₁ ,x ₂ ,…,x _N Where N is the data length, then

Wherein C is the central value of the combined working condition at the moment, and M is the standard deviation of the combined working condition at the moment.

5. The method for evaluating the health status of a wind generating set according to claim 1, wherein the analyzing variable and the working condition variable monitoring data of the moment to be analyzed are collected, the deviation degree from the analyzing variable monitoring value to the central value of the moment is quantitatively calculated according to the determined corresponding working condition threshold center, and the health grade of the analyzing variable of the moment is determined; and setting a threshold according to the normal working condition analysis variable health grade, and taking the threshold as a basis index for monitoring variable alarming.

6. The method for evaluating the health status of a wind generating set according to claim 1, wherein after the health grade of different analysis variables at each moment is obtained, the health grade of each day of all analysis variables is fused to obtain the health grade of the whole machine at different moments, the health grade of each day of the whole machine is further obtained, and a change curve of the health grade of the whole machine is constructed.

7. The system for evaluating the health state of the wind generating set is characterized by comprising a variable dividing module, an interval dividing module and an evaluating module;

the variable dividing module divides the obtained monitoring data of the wind generating set according to the label quantity, the analysis variable and the working condition variable, calculates the correlation between the analysis variable and the working condition variable, and obtains a working condition variable set corresponding to the analysis variable according to the correlation between the analysis variable and the working condition variable;

the interval dividing module divides the working condition variable values into a plurality of intervals by adopting a box diagram method according to the working condition variable of the obtained monitoring data of the wind generating set, and the working condition variable values are used as judging basis of different working conditions of the running state of the set;

the evaluation module obtains different working condition combinations according to the divided intervals, screens the monitoring data of each analysis variable under the corresponding working condition combination, and determines a variable threshold center, so that the health grade of the analysis variable at each moment can be obtained, and the health grade of the whole machine at different moments and the health grade of the whole machine at each day are further obtained.

8. The system for evaluating the health status of a wind turbine generator system according to claim 7, wherein a copula nonlinear analysis method is adopted to calculate nonlinear correlations between each analysis variable and each working condition variable, a correlation coefficient matrix between the analysis variable and the working condition variable is constructed, a threshold is set according to the strong correlation determination, and a working condition variable set most correlated with each analysis variable is identified.

9. The system for evaluating the health status of a wind turbine generator system according to claim 7, wherein the characteristic values of each box diagram of the condition variable monitoring data are calculated by adopting a box diagram method, each group of data is arranged from small to large, the number of quarters of the group of data is found, the first quartile, the median and the third quartile are used as endpoints of interval division, and the condition variable value interval is initially divided.

10. A system for evaluating the health of a wind turbine according to claim 7, wherein an analytical variable is providedThe data under a certain combined working condition is x ₁ ,x ₂ ,…,x _N Where N is the data length, then

Wherein C is the central value of the combined working condition at the moment, and M is the standard deviation of the combined working condition at the moment.