CN115822887A

CN115822887A - Performance evaluation and energy efficiency diagnosis method and system of wind turbine generator

Info

Publication number: CN115822887A
Application number: CN202211548233.6A
Authority: CN
Inventors: 朱俊杰; 祝金涛; 王一妹; 吕亮; 赵鹏程; 吴昊; 魏昂昂; 武青
Original assignee: Huaneng Clean Energy Research Institute; Huaneng Lancang River Hydropower Co Ltd
Current assignee: Huaneng Clean Energy Research Institute; Huaneng Lancang River Hydropower Co Ltd
Priority date: 2022-12-05
Filing date: 2022-12-05
Publication date: 2023-03-21

Abstract

The application provides a method and a system for performance evaluation and energy efficiency diagnosis of a wind turbine generator, wherein the method comprises the following steps: on the basis of the active power of a target wind turbine generator to be diagnosed, carrying out on-line evaluation on the power generation performance of the target wind turbine generator; combining the online evaluation result to construct an energy efficiency state index system of the target wind turbine generator, and determining a reference value of each energy efficiency state index; establishing a wind turbine generator efficiency diagnosis ontology knowledge base aiming at an energy efficiency state index system; and acquiring real-time operation data of the target wind turbine generator, performing energy efficiency abnormity identification according to the real-time operation data and the reference value, and performing energy efficiency fault diagnosis through the energy efficiency diagnosis ontology knowledge base based on the result of the energy efficiency abnormity identification. The method can evaluate the power generation performance of the wind turbine generator in real time, can accurately diagnose the energy efficiency fault mode and the energy efficiency fault reason, and improves the accuracy and pertinence of energy efficiency diagnosis.

Description

Performance evaluation and energy efficiency diagnosis method and system of wind turbine generator

Technical Field

The application relates to the technical field of wind power generation, in particular to a method and a system for performance evaluation and energy efficiency diagnosis of a wind turbine generator.

Background

With the development of wind power generation technology, the installed capacity of a wind turbine generator is gradually increased. However, with the operation of large-scale wind power plants, the proportion of aged and qualified wind power plants is increasing, and the normal operation of the wind power plant is greatly affected by the high failure rate and low energy efficiency operation of the wind power plants, so that the wind power plants need to be subjected to energy efficiency diagnosis so as to remove failures in time.

In the related art, when energy efficiency fault diagnosis is performed on a wind turbine generator, an expert in the field generally performs manual energy efficiency diagnosis according to expert knowledge, personal experience and the like. However, in practical application, because the wind turbine includes a large number of devices, a large number of operation data, and a complex working condition environment, there are many factors that affect the energy efficiency of the wind turbine. The diagnosis method of performing the artificial energy efficiency diagnosis only by the expert may cause the problems of deviation of the diagnosis result, low reliability of the diagnosis result, complex diagnosis process, low efficiency and the like.

Disclosure of Invention

The present application is directed to solving, at least to some extent, one of the technical problems in the related art.

Therefore, a first objective of the present application is to provide a method for evaluating performance and diagnosing energy efficiency of a wind turbine, which can evaluate power generation performance of the wind turbine in real time, accurately diagnose an energy efficiency fault mode and an energy efficiency fault reason, and improve accuracy and pertinence of energy efficiency diagnosis.

The second purpose of the present application is to provide a performance evaluation and energy efficiency diagnosis system for a wind turbine;

a third object of the present application is to propose a non-transitory computer-readable storage medium.

In order to achieve the above object, a first aspect of the present application is to provide a method for performance evaluation and energy efficiency diagnosis of a wind turbine generator, where the method includes the following steps:

the method comprises the steps of carrying out online evaluation on the power generation performance of a target wind turbine generator set based on the active power of the target wind turbine generator set to be diagnosed;

establishing an energy efficiency state index system of the target wind turbine generator set by combining the online evaluation result, and determining a reference value of each energy efficiency state index;

establishing a wind turbine generator efficiency diagnosis ontology knowledge base aiming at the energy efficiency state index system;

and acquiring real-time operation data of the target wind turbine generator, performing energy efficiency abnormity identification according to the real-time operation data and the reference value, performing energy efficiency fault diagnosis through the energy efficiency diagnosis ontology knowledge base based on the energy efficiency abnormity identification result, and determining an energy efficiency fault mode and a corresponding energy efficiency fault reason.

Optionally, in an embodiment of the present application, after performing energy efficiency fault diagnosis, the method further includes: and integrating the energy efficiency diagnosis ontology knowledge bases of the wind turbines on a centralized control center platform according to the distribution characteristics of the wind turbines in the wind power station.

Optionally, in an embodiment of the present application, after performing energy efficiency fault diagnosis, the method further includes: determining key components of each wind turbine generator in the wind power plant, and generating an energy efficiency diagnosis sub-knowledge base corresponding to each key component according to the electromechanical parameters of the wind turbine generator where each key component is located.

Optionally, in an embodiment of the present application, the online evaluation of the power generation performance of the target wind turbine generator based on the active power of the target wind turbine generator to be diagnosed includes: monitoring the active power variation of the target wind turbine generator, and calculating instantaneous efficiency at different moments according to the active power variation and a standard power curve; dividing the wind speed working condition of the target wind turbine generator, carrying out grade division on the instantaneous efficiency through quantiles in different wind speed working conditions, and calculating a power generation index according to a sliding window containing a plurality of continuous instantaneous efficiencies and the grades of the plurality of continuous instantaneous efficiencies; generating a training set of active power prediction models, performing working condition division on data in the training set according to pitch angles and yaw errors, training a corresponding active power prediction model according to the working condition of each pitch angle and yaw error, and calculating the power generation potential index of the target wind turbine generator according to the predicted value of the active power output by the active power prediction model.

Optionally, in an embodiment of the present application, determining the reference value of each energy efficiency state index includes: screening out historical operating data of the corresponding target wind turbine generator set by a sliding window method aiming at each energy efficiency state index; performing steady-state screening on the historical operating data; carrying out working condition division on the screened steady-state data through a K-means clustering algorithm; and determining the reference value through a multi-element Gaussian mixture model MGMM and a support vector regression model SVR under each working condition.

Optionally, in an embodiment of the present application, the performing condition division on the screened steady-state data through a K-means clustering algorithm includes: performing primary division on the steady-state data, and determining an initial clustering number and a cluster corresponding to each clustering point in the initial clustering number; and evaluating the similarity between each clustering point and the sample data in the corresponding cluster according to a silouette criterion, determining a final clustering number according to the similarity, and dividing the working conditions according to the final clustering number.

Optionally, in an embodiment of the present application, performing energy efficiency anomaly identification according to the real-time operation data and the reference value includes: determining an upper limit function and a lower limit function of a reference interval of each energy efficiency state index through polynomial curve fitting, and comparing whether the real-time operation data is located between the upper limit function and the lower limit function; and calculating the change rate of the characteristic parameters corresponding to the real-time operation data, and comparing whether the change rate is greater than the reference value.

Optionally, in an embodiment of the present application, performing energy efficiency fault diagnosis through the energy efficiency diagnosis ontology knowledge base based on a result of energy efficiency anomaly identification includes: determining the fault signs extracted by the energy efficiency abnormity identification; querying a fault mode matched with the fault symptom in the wind turbine generator efficiency diagnosis body knowledge base through a SPARQL query language; and constructing a fault cause and effect graph model corresponding to the fault mode according to the attribute information in the body knowledge base, and carrying out reasoning according to the fault cause and effect graph model to obtain the fault cause of the fault mode.

In order to achieve the above object, a second aspect of the present application further provides a performance evaluation and energy efficiency diagnosis system for a wind turbine, including the following modules:

the evaluation module is used for carrying out online evaluation on the power generation performance of the target wind turbine generator set based on the active power of the target wind turbine generator set to be diagnosed;

the determining module is used for constructing an energy efficiency state index system of the target wind turbine generator set by combining the online evaluation result and determining a reference value of each energy efficiency state index;

the building module is used for building a wind turbine generator energy efficiency diagnosis ontology knowledge base aiming at the energy efficiency state index system;

and the diagnosis module is used for acquiring real-time operation data of the target wind turbine generator, performing energy efficiency abnormity identification according to the real-time operation data and the reference value, performing energy efficiency fault diagnosis through the energy efficiency diagnosis ontology knowledge base based on the result of the energy efficiency abnormity identification, and determining an energy efficiency fault mode and a corresponding energy efficiency fault reason.

The technical scheme provided by the embodiment of the application at least has the following beneficial effects: based on the structure, the working principle, the operation data and other contents of the wind turbine generator, the artificial intelligence and the big data technology, the data retrieval technology, the information mining technology and the like are combined to evaluate the operation performance of the wind turbine generator, analyze the energy efficiency level and diagnose the energy efficiency abnormity fault. The method specifically comprises the steps of carrying out real-time performance evaluation work of the wind turbine generator based on active power grading, a power generation index and power generation potential, carrying out energy efficiency state analysis on subsystems and body equipment of the wind turbine generator, establishing an energy efficiency index system of the wind turbine generator, and determining an energy efficiency state index reference value so as to locate abnormal indexes. Typical abnormal modes and fault modes causing the abnormal energy efficiency state indexes of the wind turbine system, signs, reasons, influences and processing measures of the abnormal energy efficiency state indexes are arranged, the wind turbine operation real-time data are combined, the wind turbine energy efficiency early warning and diagnosis work is carried out, and the fault early warning causing the abnormal energy efficiency of the wind turbine can be effectively realized through the diagnosis result. Therefore, the energy efficiency fault mode and the energy efficiency fault reason can be accurately diagnosed, the accuracy, timeliness and pertinence of energy efficiency diagnosis are improved, the maintenance suggestions can be provided for field operation and maintenance personnel, and the improvement of economy and reliability of the unit is facilitated.

In order to implement the foregoing embodiments, an embodiment of the third aspect of the present application further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for performance evaluation and energy efficiency diagnosis of a wind turbine generator in the foregoing embodiments.

Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.

Drawings

The foregoing and/or additional aspects and advantages of the present application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

fig. 1 is a flowchart of a method for performance evaluation and energy efficiency diagnosis of a wind turbine generator set according to an embodiment of the present application;

fig. 2 is a flowchart of an online evaluation method for power generation performance of a wind turbine generator set according to an embodiment of the present application;

FIG. 3 is a schematic diagram illustrating a method for online evaluation of power generation performance of a wind turbine generator set according to an embodiment of the present application;

fig. 4 is a schematic structural diagram of an equipment tree according to an embodiment of the present application;

fig. 5 is a flowchart of a method for determining a reference value of an energy efficiency state index according to an embodiment of the present application;

fig. 6 is a schematic flowchart of a specific method for evaluating performance and diagnosing energy efficiency of a wind turbine generator according to an embodiment of the present disclosure;

fig. 7 is a schematic structural diagram of a system for performance evaluation and energy efficiency diagnosis of a wind turbine generator set according to an embodiment of the present disclosure.

Detailed Description

Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.

The following describes a method and a system for performance evaluation and energy efficiency diagnosis of a wind turbine generator set according to an embodiment of the present invention in detail with reference to the accompanying drawings.

Fig. 1 is a flowchart of a method for evaluating performance and diagnosing energy efficiency of a wind turbine generator according to an embodiment of the present application, and as shown in fig. 1, the method includes the following steps:

and S101, carrying out online evaluation on the power generation performance of the target wind turbine generator set based on the active power of the target wind turbine generator set to be diagnosed.

Specifically, the evaluation of the power generation performance of the wind turbine is the consideration of the wind energy conversion capability of the wind turbine in the current operation state of the wind turbine, so that the gap between the wind turbine and the excellent level of the generated energy is correctly known. According to the method and the device, after the target wind turbine generator needing energy efficiency diagnosis is determined, the active power of the target wind turbine generator is monitored, and the real-time evaluation of the power generation performance of the target wind turbine generator from different angles is realized by calculating based on the active power.

It should be noted that, in the operation process of the wind turbine, the change of the active power is mainly caused by the following reasons: firstly, the change of environmental factors such as wind speed and wind direction; the second is the corresponding coping strategy of the unit coping with the environmental change, such as braking action, pitch-changing action, yawing action and the like; and thirdly, component deterioration, abnormal faults and the like in the running process of the unit.

For the reasons, in one embodiment of the application, when the power generation performance is specifically evaluated in real time, a quantile-based real-time evaluation model of the power generation performance of the wind turbine generator is provided, and in order to capture the change information of the power generation performance of the wind turbine generator, which is covered behind the instantaneous efficiency, an online evaluation model of the power generation performance based on the grading of the instantaneous efficiency is provided. And on the basis of evaluating the current power generation index, a wind turbine generator set power generation potential concept within an achievable range is also provided, the power generation performance promotion space of the current unit is judged on the basis of an active power prediction model of the current wind speed and wind direction environment, and a power generation quantity promotion quantification model in the current environment is provided.

In order to more clearly illustrate the specific implementation process of the online power generation performance evaluation performed by the present embodiment, a detailed description is given below with respect to an embodiment of a specific power generation performance evaluation method. Fig. 2 is a schematic flow diagram of a method for online assessment of the power generation performance of a wind turbine generator set provided in the embodiment of the present application, and fig. 3 is a schematic principle diagram of a method for online assessment of the power generation performance of a wind turbine generator set provided in the embodiment of the present application, as shown in fig. 2, the method includes the following steps:

step S201, monitoring the active power variation of the target wind turbine generator, and calculating instantaneous efficiency at different moments according to the active power variation and a standard power curve.

Specifically, the present embodiment first proposes to convert active power with a fluctuation range from zero to maximum power into instantaneous efficiency with a fluctuation range around 1 based on the instantaneous efficiency of a standard power curve, where the standard power curve is a graph and a table representing the relationship between the net electric power output and the wind speed of the wind turbine under a standard atmospheric condition.

When obtaining the active power variation of the wind turbine generator, in order to accurately capture the power output capability of the wind turbine generator in consideration of the frequency of wind speed fluctuation, the present embodiment may analyze the active power variation by using the detection interval of the SCADA.

Step S202, the wind speed working conditions of the target wind turbine generator are divided, the instantaneous efficiency is graded through quantiles in different wind speed working conditions, and the power generation index is calculated according to a sliding window containing a plurality of continuous instantaneous efficiencies and the grades of the plurality of continuous instantaneous efficiencies.

Specifically, in order to capture information of unit power generation performance change covered behind instantaneous efficiency and simultaneously consider complexity and diversity of operation conditions of the wind turbine, the embodiment provides an online power generation performance evaluation model based on instantaneous efficiency grading, a quantile mode is adopted during unit instantaneous efficiency grade grading to ensure that the frequency of instantaneous efficiency appearing in each quantile interval is equal, the instantaneous efficiency before and after the instantaneous efficiency is correlated in a sliding window mode, and a power generation index is calculated through a related formula according to the step length of the sliding window and the grades of a plurality of continuous instantaneous efficiencies corresponding to the window.

Step S203, generating a training set of active power prediction models, performing working condition division on data in the training set according to the pitch angle and the yaw error, training a corresponding active power prediction model according to the working condition of each pitch angle and the yaw error, and calculating the power generation potential index of the target wind turbine generator according to the predicted value of the active power output by the active power prediction model.

Specifically, on the basis of evaluating the current power generation index, the embodiment provides the power generation potential of the wind turbine generator within the realizable range, which is different from the power output of the wind turbine generator in a specific wind speed environment, which is obtained under a strict test environment with standard power, provides an active power prediction model considering the current wind speed and wind direction environment, and determines the power generation performance promotion space of the current wind turbine generator based on the active power prediction model.

It should be noted that the above-mentioned evaluation method is executed according to the flow shown in fig. 3, and the specific implementation manner may be determined in each step according to actual needs, which is not limited herein.

And S102, constructing an energy efficiency state index system of the target wind turbine generator set by combining online evaluation results, and determining a reference value of each energy efficiency state index.

Specifically, the selection of the energy efficiency index and the construction of an index system are the basis of the whole energy efficiency analysis of the wind turbine generator, and the wind turbine relates to a plurality of subsystems in the wind energy conversion process, and each subsystem comprises a plurality of operating parameters. In the process of capturing and converting wind energy, parameter indexes related to energy transfer and loss of different subsystems are obtained through energy conversion and loss analysis. Therefore, the energy efficiency state index system must be built on a multiple subsystem basis.

In one embodiment of the application, a target wind turbine generator is divided into a plurality of subsystems, and then each subsystem is divided into a plurality of devices. As an example, a system division and an equipment analysis are performed on a wind turbine generator to be researched, and the wind turbine generator may be divided into: the wind turbine generator system comprises four subsystems, namely a wind turbine generator system transmission system, a wind capturing system, a generator system and a converter system, and the subsystems are divided to obtain an equipment tree shown in the figure 4.

Furthermore, parameter indexes representing the operation energy efficiency level of each subsystem are selected for the wind turbine generator energy transfer mechanism and the energy loss mechanism based on the field. The wind turbine generator energy efficiency state index screening work is carried out according to 'system-subsystem-equipment' based on the equipment tree, and an obtained wind turbine generator state monitoring index hierarchy table is shown in the following table 1:

TABLE 1

It should be noted that, in the embodiment of the present application, an energy efficiency state index system of the target wind turbine generator may be constructed by combining the result of the online power generation performance evaluation in the previous step. For example, in the above-described stack layer of table 1, the correlation index of the determined power generation performance may be used as an index. And when the energy efficiency indexes of all devices in all subsystems in the device layer are determined, determining the device with larger influence on the generating performance of the unit as a target device of the determined index according to the evaluation of the generating performance, and taking parameters of the target device influencing the generating performance as the index and the like. Therefore, the energy efficiency state index system is established by considering the power generation performance of the generator set, and the power generation performance of the wind turbine generator set is favorably ensured through subsequent diagnosis and relevant operation and maintenance measures.

Furthermore, after the wind turbine generator energy efficiency state index system is built, the energy efficiency state index reference value solving work is carried out.

In one embodiment of the present application, historical operating data of a unit may be analyzed to find out feature information, and a reference value and a reference interval of a feature parameter may be determined. The reference interval is used as a healthy and reasonable operation range of characteristic parameters of the wind turbine generator in the operation process under a certain working condition, if the operation data exceed the reference interval, the operation data can be judged to be abnormal, and judgment conditions are provided for the follow-up abnormal detection and energy efficiency diagnosis work. In the embodiment of the application, data in the normal operation process of the unit is screened out by adopting a sliding window method, the working condition type of the unit operation data is divided according to a working condition division method based on a K-mean clustering algorithm, and finally, a reference value and a reference interval solving method of a multi-Gaussian Mixture Model (MGMM) and a Long Short-term memory neural network algorithm (Long Short-term Memmory) are adopted to determine the reference value and the reference interval of each energy efficiency state index under each working condition.

In order to more clearly illustrate the specific implementation process of determining the reference value in this embodiment, a detailed description is given below with respect to an embodiment of a specific method for determining the reference value. Fig. 5 is a flowchart of a method for determining a reference value of an energy efficiency state index according to an embodiment of the present application, and as shown in fig. 5, the method includes the following steps:

and S301, screening out historical operation data of the corresponding target wind turbine generator set by a sliding window method according to each energy efficiency state index.

Specifically, in this embodiment, when determining the reference value and the reference interval of the energy efficiency state index of the wind turbine generator, the reference value and the reference interval of the characteristic parameter may be determined by analyzing the historical data of the operation of the wind turbine generator to find out the characteristic information. During specific implementation, historical operation data of the unit is obtained, and data in the normal operation process of the unit is screened out by adopting a sliding window method. And then, correspondingly screening out the historical operation data of each energy efficiency state index in the normal historical operation data of the unit according to each energy efficiency state index.

Step S302, steady-state screening is conducted on historical operating data.

Specifically, considering that the wind turbine generator receives frequent start-stop requirements of wind speed and power grid instructions, the operating state of the wind turbine generator can be divided into a steady state and an unsteady state. And when the running data of the unit can be stabilized within a certain range within a period of time, judging the running data to be a stable state. The fluctuation of the operation parameters is large in the unsteady operation process of the unit, and the output parameters and the input parameters of the unit are difficult to keep consistent. At this time, the operation data cannot correctly reflect the operation condition of the unit, so in order to ensure that the information mined in the historical data is accurate enough, the first work is to remove the shutdown and unsteady data.

Whether the output power of the unit is stable or not can directly reflect the unsteady and steady operation conditions of the unit. The steady state screening is carried out by judging the output power, and the specific judging method is shown as the following formula:

wherein p is _t Representing the measured value of the output power at time t, p _r Represents the true value of the output power at time t, λ represents the rate of change of the output power at time t, and ε represents the random error of the output power (ε follows a normal distribution).

According to the formula (1), the main criterion for determining the steady state and the unsteady state is whether the change rate λ is 0. The estimation of the λ value can be embodied by finding the difference Δ p between the output powers at adjacent time instants, as shown in the following equation (2):

Δp＝p _t -p _t-1 ＝λ+(ε _t -ε _t-1 ) (2)

due to epsilon _t ～N(0,σ ² ) The calculated amount Δ p is desirably λ, Δ p to N (λ,2 σ) ² ). According to the characteristics of the time series, λ can be estimated by averaging the statistical samples within a sliding window formed by a specified time period, as shown in the following equation (3):

where m is the number of samples within the sliding window formed over the prescribed time period.

In order to ensure accurate estimation results, the embodiment of the present application adopts a method of interval estimation, which is specifically shown in the following formula (4):

wherein, alpha is a given significance level,

and

two statistics at a given significance level.

By the formula(4) It can be seen that there is a confidence level of (1-. Alpha.) that the true value of λ lies within

Within the interval. Thus, if the confidence interval is

If 0 is not included, the unit is operated in an unstable state in a period of time, and data in the period of time should be removed before subsequent work is carried out.

And step S303, performing working condition division on the screened steady-state data through a K-means clustering algorithm.

In the embodiment of the application, the working condition division based on the K-means clustering algorithm comprises the following steps: the method comprises the steps of performing initial division on steady-state data, determining an initial clustering number and clusters corresponding to each clustering point in the initial clustering number, evaluating the similarity between each clustering point and sample data in the corresponding clusters according to a silouette criterion, determining a final clustering number according to the similarity, and performing working condition division according to the final clustering number.

Specifically, the K-means clustering algorithm is an efficient clustering algorithm, and selects K points from specified sample data at will, takes the K points as initial clustering centers and forms corresponding clusters, calculates the distances between other sample points and the K points, sorts the points according to the length of the distances, and then divides the points into clusters formed by the initial clustering centers closest to the points. And after all data points are divided into corresponding clusters, realizing the first clustering work. Then calculating the average value of all data in each cluster, taking the average value as a new clustering center, and continuously repeating the clustering steps until the criterion function converges, wherein the convergence function is shown as the following formula:

where E is the sum of the squared errors of all data points, b _j Data points of the i-th class, a _j Is the center of the ith cluster.

It should be noted that, for the K-means clustering algorithm, the determination of the clustering number K is very important. Too few clusters can cause all the characteristics of the sample to be reflected unfamiliar, and too many clusters can cause the data belonging to the same mode to be divided. To determine the optimal number of clusters, the embodiments of the present application use the Silhauelle criterion. After a cluster number K is preliminarily determined, the similarity between the K cluster points and other sample data in the cluster is evaluated by using the criterion, and a calculation formula of the average similarity W of the N sample points is shown as the following formula:

wherein x is _i Is the average value of the distance between the ith sample point and other points in the cluster, y _i Is the minimum of the distances between the ith sample point and the points in the other different clusters. When the average similarity W is maximum, the K value is the optimal clustering number.

In one embodiment of the application, in order to further improve the accuracy of the working condition division, a multi-step K-means clustering method can be further performed to perform the working condition division. For example, two steps are taken as an example: firstly selecting wind direction as a unique characteristic to carry out K-means clustering, then taking a data cluster formed after primary clustering as a father sample set of a secondary K-means clustering algorithm taking wind speed as the unique characteristic, marking and sorting the data cluster after secondary clustering to finally form a multi-step K-means clustering division result, wherein the principle of the multi-step K-means clustering algorithm is shown in the following chart. And after the twice clustering algorithm is completed, the range of each boundary condition of all points in each cluster is obtained and used as a typical working condition.

And step S304, determining a reference value through the multi-element Gaussian mixture model MGMM and the support vector regression model SVR under each working condition.

Specifically, the Multivariate Gaussian Mixture Model (MGMM) combines a parametric estimation method and a nonparametric estimation method, and is a probability density estimation method based on half parameters. When the number of the sub-models is large enough, the multivariate Gaussian mixture model can approach any continuous distribution with higher precision, and the probability distribution of the multivariate Gaussian mixture model is shown as the following formula:

wherein X is an L-dimensional parameter data column vector, X = [ X = ₁ x ₂ ,...,x _L ] ^T (ii) a k is the number of sub-models in the multivariate Gaussian mixture model; omega _k Is the weight coefficient of the kth sub-model, and

φ(X|θ _k ) The Gaussian probability density function of the kth sub-model is shown, and the specific calculation formula is shown in formula (8).

The multivariate Gaussian mixture model adopts an Expectation Maximization (EM) algorithm for parameter estimation, and the EM algorithm is an iterative algorithm suitable for parameter estimation of a probability model containing hidden variables. Wherein, sigma _k And mu _k Values are estimated by the EM algorithm for the covariance matrix and the mean. The objective function of the EM algorithm is:

the iterative process of the EM algorithm comprises the following steps: selecting initial values of mu and sigma, calculating the posterior probability of the corresponding kth model by the following formula (10), and finally iteratively calculating the mu and sigma in the model according to the following formulas (11), (12) and (13) until the difference between the objective functions of two adjacent iterations is less than 10 ^-5 Stopping iteration:

therefore, data under a certain working condition is screened by a multivariate Gaussian mixture model to form a sample data set, and then a reference value regression curve is determined by a regression estimation algorithm of SVR.

Wherein, SVR regression can be understood as: given a sample set { (x) _i ,y _i )}(i＝1,2,…,M),x _i Is the ith input vector, y _i For the ith output scalar, M is the total number of samples. The function g (x) maps the input vector into an l-dimensional space and constructs a hyperplane f (x) = W in this space ^T g (x) + b, W is a weight vector of dimension l, and b is a bias term. Distance of all sample points to hyperplane | y _i -f(x _i ) The method is characterized in that | ≦ epsilon (epsilon is more than or equal to 0), epsilon is precision and is a regression maximum allowable error, and due to the existence of errors, when the optimal hyperplane construction problem is converted into a convex quadratic optimization solution, a penalty factor C and relaxation variables xi and xi should be introduced ^* The specific expression is shown as (14):

further, by means of a non-negative Lagrange multiplier alpha _i And alpha _i ^* Converting the solution of equation (14) into a solution of equation (15) below:

after solving the formula (15), the optimal hyperplane can be obtained as shown in the following formula:

it is understood that equation (16) is also an expression of a regression function, the kernel function k (x) introducing the Mercer condition _i X) instead of g ^T (x _i ) g (x), so that there is no need to solve g (x), the expression of the regression function is finally derived as shown in the following equation (17), where α _i 、α _i ^* And b can be obtained from sample training iterations:

a commonly used Kernel function for SVR is the Gaussian Kernel function (Gaussian Kernel), defined as shown in equation (18) below:

wherein σ is a kernel parameter.

The Mean Absolute Error (MAE) is used as the evaluation index of regression in the final prediction result. The calculation formula of MAE is shown in formula (19).

Wherein n is the number of samples,

indicates the actual value of the i-th correlation index,

and (4) a predicted value of the ith correlation index is shown.

Thereby, the reference value determination based on the MGMM model and the SVR model is realized.

And S103, establishing a wind turbine generator energy efficiency diagnosis ontology knowledge base for the energy efficiency state index system.

It should be noted that a large amount of expert knowledge and unit equipment information are needed in the wind turbine energy efficiency diagnosis and maintenance work, and most of the power generation units store the information in an unstructured text form, which is difficult to implement the automatic query reasoning diagnosis work of faults, and knowledge is not combined with the actual situation on site, so that the update iteration of knowledge is poor. Therefore, the unstructured knowledge is expressed in a structured mode through an ontology modeling mode, and the conceptual and normalized ontology constraint can well avoid language ambiguity. And conveniently constructing a wind turbine key equipment energy efficiency fault diagnosis knowledge base based on the sorted ontology knowledge base to provide a basis for subsequent diagnosis reasoning.

The ontology is an expression form of structured conceptual knowledge, and mainly comprises classes, attributes and individuals.

In specific implementation, proper description languages need to be selected in the ontology construction process to better describe the concept and relationship of knowledge. In an embodiment of the application, an OWL DL grammar which is excellent in language expression capability and computer automation reasoning capability is adopted, a fault knowledge body of the wind turbine generator is constructed by means of project software and a project-based seven-step method, and the method comprises the following specific steps:

1) An object is determined. The body is constructed in the field of wind turbine generator fault diagnosis and overhaul and maintenance.

2) Existing ontologies in the field are collected and analyzed for availability.

3) The important terms in the body are sorted out. And extracting important concepts and attribute relations in a related fan fault knowledge table obtained based on an FMEA analysis method.

4) Classes and class hierarchies are determined. The hierarchical structure of parent class and subclass class is combed out from three major classes of running state of the fan, system equipment and overhaul and maintenance, and the specific division is shown in table 2.

TABLE 2

5) Attributes and relationships of classes are determined. According to the actual situation of a fault knowledge field body of a wind turbine generator, object attributes are divided into 7 types including state, state and historical frequency, wherein the 7 types include state, occurrence and the like, three data attributes including prior probability, effect intensity and historical frequency are set simultaneously, and the prior probability, the effect intensity and the historical frequency of the equipment fault cause are determined by means of the data attributes. The object attributes are used for establishing association relations among classes, between classes and individuals and between individuals, and the data attributes are used for representing numerical attributes of the individuals.

6) Adding the individual. The abstract class is composed of concrete individuals, all the individuals are divided and distributed into various classes according to concepts, and the relationship among individual elements is constructed after the division is completed.

7) And (5) detecting consistency. And determining that the constructed wind turbine generator fault knowledge ontology has no logic conflict.

It should be noted that, when the energy efficiency diagnosis ontology knowledge base of the wind turbine generator is constructed, the knowledge base can be confirmed in the above steps

And determining the energy efficiency state index system as reference, namely ensuring that the energy efficiency diagnosis ontology knowledge base is constructed to contain knowledge related to each index in the energy efficiency state index system, and realizing construction of the energy efficiency diagnosis ontology knowledge base of the wind turbine generator set aiming at the energy efficiency state index system. Therefore, the comprehensiveness of constructing the energy efficiency diagnosis ontology knowledge base is improved, and the constructed knowledge base can meet the energy efficiency diagnosis serving as the energy efficiency state index.

And step S104, acquiring real-time operation data of the target wind turbine generator, performing energy efficiency abnormity identification according to the real-time operation data and a reference value, performing energy efficiency fault diagnosis through an energy efficiency diagnosis ontology knowledge base based on the result of the energy efficiency abnormity identification, and determining an energy efficiency fault mode and a corresponding energy efficiency fault reason.

Specifically, according to the method, historical operating data and real-time monitoring data of the wind turbine generator under each operating condition are combined, an abnormal mode identification model based on artificial intelligence and a data mining technology is built, and abnormal modes of each system (such as a generator, an encoder, a gear box, a frequency converter and the like) in the operating process of the wind turbine generator are identified. And moreover, a key equipment fault diagnosis inference model based on the energy efficiency diagnosis ontology knowledge base is constructed, real-time operation data of the unit is combined with the energy efficiency diagnosis knowledge base, and energy efficiency fault inference diagnosis work of 'abnormal state identification-fault mode positioning-fault reason diagnosis-maintenance strategy proposition' is realized.

In specific implementation, after real-time operation data of the target wind turbine generator are acquired through modes such as data acquisition by an SCADA system, energy efficiency abnormity identification is carried out according to the real-time operation data and a reference value.

As one possible implementation, the present application implements anomaly pattern recognition through threshold-type anomaly detection and trend-type anomaly detection. That is, in this embodiment, the energy efficiency abnormality recognition based on the real-time operation data and the reference value includes: determining an upper limit function and a lower limit function of a reference interval of each energy efficiency state index through polynomial curve fitting, and comparing whether real-time operation data is located between the upper limit function and the lower limit function; and calculating the change rate of the characteristic parameters corresponding to the real-time operation data, and comparing whether the change rate is greater than the reference value.

Specifically, in threshold anomaly detection, for the start-up and shutdown conditions, in order to obtain a functional relationship between the characteristic parameters and the boundary conditions, an approximate image of a curve y = f (x) needs to be drawn through a set of historical data points (x, y) of normal start-up. The goal of curve fitting is to find regularity from a set of seemingly cluttered data, i.e., to construct a curve that approximately reflects the overall trend of all data points to eliminate local fluctuations in the set of data. In the embodiment, in consideration of the characteristics of the actual operating parameters, a polynomial curve fitting is selected to improve the fitting accuracy, and the polynomial function form is shown in the following formula (20):

wherein y is a characteristic parameter, x is a boundary condition, and w _i Is a polynomial coefficient.

Furthermore, in this embodiment, the upper and lower limit function expressions of the characteristic parameter reference interval are determined according to the fitted sample residual distribution. For the rising symptom, if the characteristic parameter exceeds the upper limit, the symptom is considered to be abnormal; for a dip symptom, if the characteristic parameter exceeds the lower limit, it is considered abnormal.

In addition, for normal operation conditions, the result is obtained according to the reference value of the characteristic parameter, the reference operation curve of the characteristic parameter under each condition is fitted, and the reference interval is determined by the same method as that of the unsteady state condition.

The method and the device perform rate anomaly detection besides threshold anomaly detection. The threshold value abnormal detection and the rate abnormal detection are independent and make up for each other, so that the reliability of the abnormal detection is greatly improved. Here, the rate anomaly detection is to detect the rate of change of the characteristic parameter in a predetermined time, as shown in the following expression (21).

Where δ is the rate of change of the parameter, y _t+Δt Is the value of the parameter at a later time, y _t Is the value of the parameter at the previous moment, and Δ t is the time difference between the next moment and the previous moment.

Regarding the rising type symptom, if the rising rate of the characteristic parameter exceeds a reference line, the characteristic parameter is considered to be abnormal; for a declining symptom, if the decline rate of the characteristic parameter exceeds the baseline, the symptom is considered abnormal. The reference line is obtained by selecting the upper and lower limits with higher confidence as the upper and lower limits of the reference according to the statistical analysis of the normal sample.

Further, constructing a diagnosis inference model.

In an embodiment of the application, the energy efficiency fault diagnosis is performed through the energy efficiency diagnosis ontology knowledge base based on the result of energy efficiency anomaly identification, and the method includes the following steps. Firstly, determining energy efficiency abnormity and identifying extracted fault symptoms. And then, inquiring a fault mode matched with the fault symptom in a wind turbine generator energy efficiency diagnosis ontology knowledge base through an SPARQL inquiry language. And finally, constructing a fault cause and effect graph model corresponding to the fault mode according to the attribute information in the ontology knowledge base, and reasoning according to the fault cause and effect graph model to obtain the fault cause of the fault mode.

Specifically, the energy efficiency diagnosis reasoning process of the embodiment of the application comprises a forward diagnosis step and a reverse diagnosis step, wherein the forward diagnosis step is to inquire a corresponding fault mode according to the extracted fault symptoms in a matching and searching mode. The reverse diagnosis is to determine the fault reason causing the fault mode by combining reasoning and searching according to the determined fault mode.

In specific implementation, after the extraction of the fault symptoms is completed according to the threshold value abnormality detection and rate abnormality detection model, the diagnosis reasoning process of forward diagnosis is completed according to the extracted fault symptoms. Through rule-based correlation query, the query of a certain knowledge unit in the ontology knowledge base is completed by means of SPARQL query language under a matching and searching mechanism, and the work of fault mode identification and fault reason query is completed. Different from the complicated SQL sentences of the traditional database, the SPARQL language can realize the direct query of subjects, predicates and objects in specific knowledge expression form RDF triples in the knowledge map, and can position a fault mode under the common SELECT sentences:

SELECTX？Y WHERE{？X rdfs:subClassOf:？Y.}。

furthermore, after the specific fault mode is located, reasoning work of the fault occurrence reason for reverse diagnosis is carried out. The embodiment of the application provides a fault cause diagnosis and inference method based on a causal graph model by combining fault tree analysis and a conditional probability formula, and the probability sequence of each cause causing the fault can be calculated by means of the causal graph model. According to the method, a fault cause and effect graph model corresponding to a fault mode is constructed according to attribute information in an ontology knowledge base, wherein the attribute information comprises object attributes and data attributes, the object attributes are used for establishing association relations among different classes, between classes and individuals and between different individuals, and the data attributes are used for representing numerical values of the individuals. As an example, the object attribute includes 7 pieces of information such as "composition", "containing", "occurrence", "cause is", "influence is", "symptom is", and "maintenance measure is". The data attribute includes 3 pieces of information such as "prior probability is", "causal strength is", and "history number is".

Specifically, the failure cause diagnosis reasoning link based on the causal graph model refers to the calculation of a required value according to a knowledge unit meeting the condition. The searched fault reasons can be preliminarily screened, quantitative index fault reasons which can be directly eliminated are removed, and a fault reason alternative set is preliminarily obtained. In the troubleshooting work of the final fault cause, combining a cause and effect strength assignment based on expert knowledge and a prior probability assignment method based on historical occurrence times of the fault cause, obtaining all minimum cut set Boolean expressions and non-intersection final cut set expressions according to a cause and effect model diagram, calculating the occurrence probability of each fault cause when a fault mode occurs, and troubleshooting the fault cause from high to low according to a probability sequence.

Therefore, the wind turbine generator efficiency early warning and diagnosis work can be carried out according to the diagnosis result, the diagnosis result can effectively realize the fault early warning causing the abnormal generator efficiency, the overhaul and maintenance suggestions are provided for the improvement of field operation and maintenance personnel, for example, a corresponding solution is provided for the diagnosed fault reason, and the economic efficiency and the reliability of the generator are improved.

In one embodiment of the application, intelligent operation and maintenance can be performed on key components of the wind turbine generator. In this embodiment, the energy efficiency diagnosis ontology knowledge bases of the plurality of wind turbines may be integrally set on the centralized control center platform according to the distribution characteristics of the plurality of wind turbines in the wind farm. And determining key components of each wind turbine generator in the wind power plant, and generating an energy efficiency diagnosis sub-knowledge base corresponding to each key component according to the electromechanical parameters of the wind turbine generator where each key component is located.

Specifically, aiming at the station distribution characteristics of the main machine equipment of the wind power station, the energy efficiency diagnosis and diagnosis knowledge base of the key equipment of the on-line wind turbine generator set can be integrated on the existing centralized control center system platform. And migrating the key component equipment energy efficiency diagnosis model to a target unit by utilizing a unit-station-centralized control three-level framework, and matching according to specific mechanical and electrical parameters of the unit to generate a corresponding knowledge base. Therefore, more targeted diagnosis can be performed through the energy efficiency diagnosis sub-knowledge base corresponding to the key equipment in the unit according to the specific unit condition, so that the range of energy efficiency fault diagnosis troubleshooting can be reduced, and the accuracy of energy efficiency fault diagnosis is improved.

In summary, the performance evaluation and energy efficiency diagnosis method for the wind turbine generator according to the embodiment of the present application is based on the structure, the working principle, the operation data, and other contents of the wind turbine generator, and adopts artificial intelligence in combination with a big data technology, a data retrieval technology, an information mining technology, and the like, so as to perform wind turbine generator operation performance evaluation, energy efficiency level analysis, and energy efficiency abnormality fault diagnosis. The method specifically comprises the steps of carrying out real-time performance evaluation work of the wind turbine generator based on active power grading, a power generation index and power generation potential, carrying out energy efficiency state analysis on subsystems and body equipment of the wind turbine generator, establishing a wind turbine generator energy efficiency index system, determining an energy efficiency state index reference value and positioning abnormal indexes. Typical abnormal modes and fault modes causing the abnormal energy efficiency state indexes of the wind turbine system, signs, reasons, influences and processing measures of the abnormal energy efficiency state indexes are arranged, the wind turbine operation real-time data are combined, the wind turbine energy efficiency early warning and diagnosis work is carried out, and the fault early warning causing the abnormal energy efficiency of the wind turbine can be effectively realized through the diagnosis result. Therefore, the method can accurately diagnose the energy efficiency fault mode and the energy efficiency fault reason, improves the accuracy, timeliness and pertinence of energy efficiency diagnosis, can realize the improvement of maintenance suggestions for field operation and maintenance personnel, and is favorable for improving the economy and reliability of the unit.

In order to more clearly describe the implementation flow of the method for evaluating the performance and diagnosing the energy efficiency of the wind turbine generator according to the embodiment of the present application, a specific embodiment of evaluating the performance and diagnosing the energy efficiency of the wind turbine generator is described in detail below. Fig. 6 is a schematic flow chart of a specific method for evaluating performance and diagnosing energy efficiency of a wind turbine generator according to an embodiment of the present application, and as shown in fig. 6, the method includes:

and S401, carrying out on-line evaluation on the power generation performance of the wind turbine generator.

Specifically, the step is to perform model analysis, and mainly comprises the following steps of: the system comprises an active power grading model, a power generation level evaluation model and a power generation capacity promotion quantification model.

And S402, constructing an energy efficiency index system of the wind turbine generator.

Specifically, this step forms a unit energy efficiency analysis framework, and mainly includes: an energy efficiency index system is formed through analysis of the wind turbine generator system, and the energy efficiency index system can be used for analyzing related abnormal modes and related faults.

And step S403, constructing an energy efficiency diagnosis knowledge base.

Specifically, the step is to construct an energy efficiency diagnosis knowledge system, and mainly to construct a strategy to form an energy efficiency diagnosis knowledge base based on the knowledge base of the ontology

In step S404, determination of the reference value and abnormality detection are performed.

Specifically, the energy efficiency diagnosis data system is constructed in the step, and the abnormality detection is mainly performed according to the reference value, and then the diagnosis and inference of the knowledge base are performed by means of the energy efficiency diagnosis knowledge base constructed in the step S403.

And step S405, developing an energy efficiency diagnosis and maintenance decision system.

It should be noted that, for specific implementation manners of each step in the method, reference may be made to the description in the foregoing embodiments, and details are not described here.

In order to implement the foregoing embodiments, the present application further provides a system for evaluating performance and diagnosing energy efficiency of a wind turbine generator, and fig. 7 is a schematic structural diagram of the system for evaluating performance and diagnosing energy efficiency of a wind turbine generator according to the embodiments of the present application, and as shown in fig. 7, the system includes an evaluation module 100, a determination module 200, a construction module 300, and a diagnosis module 400.

The evaluation module 100 is configured to perform online evaluation on the power generation performance of the target wind turbine generator based on the active power of the target wind turbine generator to be diagnosed.

And the determining module 200 is configured to construct an energy efficiency state index system of the target wind turbine generator set according to the online evaluation result, and determine a reference value of each energy efficiency state index.

The building module 300 is used for building a wind turbine generator efficiency diagnosis ontology knowledge base for an energy efficiency state index system.

The diagnosis module 400 is configured to obtain real-time operation data of the target wind turbine, perform energy efficiency anomaly recognition according to the real-time operation data and the reference value, perform energy efficiency fault diagnosis through the energy efficiency diagnosis ontology knowledge base based on a result of the energy efficiency anomaly recognition, and determine an energy efficiency fault mode and a corresponding energy efficiency fault reason.

Optionally, in an embodiment of the present application, the diagnostic module 400 is further configured to: and according to the distribution characteristics of a plurality of wind turbines in the wind power station, integrating and arranging the energy efficiency diagnosis ontology knowledge base of the plurality of wind turbines on a centralized control center platform.

Optionally, in an embodiment of the present application, the diagnostic module 400 is further configured to: the method comprises the steps of determining key components of each wind turbine generator in a wind power plant, and generating an energy efficiency diagnosis sub-knowledge base corresponding to each key component according to electromechanical parameters of the wind turbine generator where each key component is located.

Optionally, in an embodiment of the present application, the evaluation module 100 is specifically configured to: monitoring the active power variation of the target wind turbine generator set, and calculating instantaneous efficiency at different moments according to the active power variation and a standard power curve; dividing the wind speed working condition of a target wind turbine generator, carrying out grade division on instantaneous efficiency through quantiles in different wind speed working conditions, and calculating a power generation index according to a sliding window containing a plurality of continuous instantaneous efficiencies and the grades of the plurality of continuous instantaneous efficiencies; generating a training set of active power prediction models, carrying out working condition division on data in the training set according to the pitch angle and the yaw error, training a corresponding active power prediction model according to the working condition of each pitch angle and the yaw error, and calculating the power generation potential index of the target wind turbine generator according to the predicted value of the active power output by the active power prediction model.

Optionally, in an embodiment of the present application, the determining module 200 is specifically configured to: screening out corresponding historical operating data of the target wind turbine generator set by a sliding window method aiming at each energy efficiency state index; performing steady-state screening on historical operating data; carrying out working condition division on the screened steady-state data through a K-means clustering algorithm; and determining a reference value through a multi-element Gaussian mixture model MGMM and a support vector regression model SVR under each working condition.

Optionally, in an embodiment of the present application, the determining module 200 is specifically configured to: performing primary division on the steady-state data, and determining an initial clustering number and a cluster corresponding to each clustering point in the initial clustering number; and evaluating the similarity between each clustering point and the sample data in the corresponding cluster according to the Silhouette criterion, determining a final clustering number according to the similarity, and dividing the working conditions according to the final clustering number.

Optionally, in an embodiment of the present application, the diagnosis module 400 is specifically configured to: determining an upper limit function and a lower limit function of a reference interval of each energy efficiency state index through polynomial curve fitting, and comparing whether real-time operation data is located between the upper limit function and the lower limit function; and calculating the change rate of the characteristic parameters corresponding to the real-time operation data, and comparing whether the change rate is greater than a reference value.

Optionally, in an embodiment of the present application, the diagnosis module 400 is specifically configured to: determining fault signs extracted by energy efficiency abnormity identification; inquiring a fault mode matched with the fault symptom in a wind turbine generator energy efficiency diagnosis body knowledge base through an SPARQL inquiry language; and constructing a fault cause and effect graph model corresponding to the fault mode according to the attribute information in the ontology knowledge base, and reasoning according to the fault cause and effect graph model to obtain the fault cause of the fault mode.

It should be noted that the explanation of the embodiment of the method for evaluating performance and diagnosing energy efficiency of a wind turbine generator is also applicable to the system of the embodiment, and is not repeated here

In summary, the performance evaluation and energy efficiency diagnosis system for the wind turbine generator according to the embodiment of the present application is based on the structure, the working principle, the operation data, and other contents of the wind turbine generator, and adopts artificial intelligence to combine with a big data technology, a data retrieval technology, an information mining technology, and the like, so as to perform wind turbine generator operation performance evaluation, energy efficiency level analysis, and energy efficiency abnormality fault diagnosis. The method specifically comprises the steps of carrying out real-time performance evaluation work of the wind turbine generator based on active power grading, a power generation index and power generation potential, carrying out energy efficiency state analysis on subsystems and body equipment of the wind turbine generator, establishing an energy efficiency index system of the wind turbine generator, and determining an energy efficiency state index reference value so as to locate abnormal indexes. Typical abnormal modes and fault modes causing the abnormal energy efficiency state indexes of the wind turbine system, signs, reasons, influences and processing measures of the abnormal energy efficiency state indexes are arranged, the wind turbine operation real-time data are combined, the wind turbine energy efficiency early warning and diagnosis work is carried out, and the fault early warning causing the abnormal energy efficiency of the wind turbine can be effectively realized through the diagnosis result. Therefore, the system can accurately diagnose the energy efficiency fault mode and the energy efficiency fault reason, improves the accuracy, timeliness and pertinence of energy efficiency diagnosis, can realize the improvement of maintenance suggestions for field operation and maintenance personnel, and is favorable for improving the economy and reliability of the unit.

In order to implement the foregoing embodiments, the present application further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method for performance evaluation and energy efficiency diagnosis of a wind turbine generator as described in any one of the foregoing embodiments.

In the description herein, reference to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the application. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.

Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present application, "plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.

Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing steps of a custom logic function or process, and alternate implementations are included within the scope of the preferred embodiment of the present application in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present application.

The logic and/or steps represented in the flowcharts or otherwise described herein, e.g., an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.

It should be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.

It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, which may be stored in a computer readable storage medium, and when the program is executed, the program includes one or a combination of the steps of the method embodiments.

In addition, functional units in the embodiments of the present application may be integrated into one processing module, or each unit may exist alone physically, or two or more units are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a stand-alone product, may also be stored in a computer readable storage medium.

The storage medium mentioned above may be a read-only memory, a magnetic or optical disk, etc. Although embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present application, and that variations, modifications, substitutions and alterations may be made to the above embodiments by those of ordinary skill in the art within the scope of the present application.

Claims

1. A performance evaluation and energy efficiency diagnosis method for a wind turbine generator is characterized by comprising the following steps:

on the basis of the active power of a target wind turbine generator to be diagnosed, carrying out on-line evaluation on the power generation performance of the target wind turbine generator;

combining the online evaluation result to construct an energy efficiency state index system of the target wind turbine generator set, and determining a reference value of each energy efficiency state index;

2. The method of claim 1, further comprising, after said performing energy efficiency fault diagnostics:

and integrating the energy efficiency diagnosis ontology knowledge bases of the wind turbines on a centralized control center platform according to the distribution characteristics of the wind turbines in the wind power station.

3. The method of claim 1, further comprising, after said performing energy efficiency fault diagnostics:

determining key components of each wind turbine generator in the wind power plant, and generating an energy efficiency diagnosis sub-knowledge base corresponding to each key component according to the electromechanical parameters of the wind turbine generator where each key component is located.

4. The method according to claim 1, wherein the on-line evaluation of the power generation performance of the target wind turbine based on the active power of the target wind turbine to be diagnosed comprises:

monitoring the active power variation of the target wind turbine generator, and calculating instantaneous efficiency at different moments according to the active power variation and a standard power curve;

dividing the wind speed working condition of the target wind turbine generator, carrying out grade division on the instantaneous efficiency through quantiles in different wind speed working conditions, and calculating a power generation index according to a sliding window containing a plurality of continuous instantaneous efficiencies and the grades of the plurality of continuous instantaneous efficiencies;

generating a training set of active power prediction models, dividing working conditions of data in the training set according to pitch angles and yaw errors, training a corresponding active power prediction model according to each pitch angle and yaw error working condition, and calculating a power generation potential index of the target wind generation set according to a predicted value of active power output by the active power prediction model.

5. The method according to claim 1, wherein the determining the reference value for each energy efficiency state indicator comprises:

screening out historical operating data of the corresponding target wind turbine generator set by a sliding window method aiming at each energy efficiency state index;

performing steady-state screening on the historical operating data;

performing working condition division on the screened steady-state data through a K-means clustering algorithm;

and determining the reference value through a multi-element Gaussian mixture model MGMM and a support vector regression model SVR under each working condition.

6. The method of claim 5, wherein the partitioning the screened steady-state data by a K-means clustering algorithm comprises:

performing primary division on the steady-state data, and determining an initial clustering number and a cluster corresponding to each clustering point in the initial clustering number;

and evaluating the similarity between each clustering point and the sample data in the corresponding cluster according to a silouette criterion, determining a final clustering number according to the similarity, and dividing the working conditions according to the final clustering number.

7. The method according to claim 1, wherein the performing energy efficiency anomaly identification according to the real-time operation data and the reference value comprises:

determining an upper limit function and a lower limit function of a reference interval of each energy efficiency state index through polynomial curve fitting, and comparing whether the real-time operation data is located between the upper limit function and the lower limit function;

and calculating the change rate of the characteristic parameters corresponding to the real-time operation data, and comparing whether the change rate is greater than the reference value.

8. The method according to claim 1, wherein the energy efficiency fault diagnosis is performed through the energy efficiency diagnosis ontology knowledge base based on the result of energy efficiency anomaly identification, and comprises the following steps:

determining fault signs extracted by the energy efficiency abnormity identification;

querying a fault mode matched with the fault symptom in the wind turbine generator efficiency diagnosis body knowledge base through a SPARQL query language;

and constructing a fault cause and effect graph model corresponding to the fault mode according to the attribute information in the body knowledge base, and carrying out reasoning according to the fault cause and effect graph model to obtain the fault cause of the fault mode.

9. A performance evaluation and energy efficiency diagnosis system for a wind turbine generator is characterized by comprising:

10. A non-transitory computer-readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the method for performance evaluation and energy efficiency diagnosis of a wind turbine generator as set forth in any one of claims 1-8.