| Wind power acts as the offensive force of China’s energy transformation and lowcarbon development.The unit capacity of wind turbine is gradually increasing,so that the maintenance costs of wind turbine are constantly increasing.The main transmission system of wind turbine acts as the energy transmission channel and load transfer path.Due to its variable working conditions and complex and harsh service environment,which results in a high frequency of mechanical failure of the main transmission system,the study of the health state assessment of the wind turbine main transmission system is an urgent issue at present.This subject is supported by the National Natural Science Foundation “SCADA data-based dynamic behavior and state assessment of main transmission system of wind turbine under multiple working condition”.In view of the high failure rate of the main transmission system of wind turbine,the complex working conditions and the inadequacy of the current classifying the working conditions,and the lack of methods on the condition assessment of the main transmission system,the main transmission system of wind turbine is taken as the main research object.And using the data collected by the SCADA system of wind turbine,the evaluation index and quantification algorithm of the main transmission system operation status are proposed by mining and analyzing the relationship between the SCADA data status parameters and the main transmission system operation status.The wind turbine main transmission system condition identification model is constructed for the requirements of working conditions in complex service environments.And evaluate the online operating status of the wind turbine main transmission system under different operating conditions to achieve safe and reliable operation of the wind turbine.The main research work includes:(1)The development status of the global wind power industry and the current status of research on the condition assessment and working conditions classification of wind turbines are described.In view of the high failure rate of the main transmission system of wind turbine,the complex working conditions and the inadequacy of the current method of classifying the working conditions,and the lack of research on the condition assessment of the main transmission system,the general research idea is proposed to firstly use the clustering algorithm to subdivide the working conditions,then establish the prediction model of the main transmission system condition evaluation parameters,and finally apply the fuzzy comprehensive evaluation algorithm to evaluate the operating health status of the main transmission system.(2)The structure of wind turbine main transmission system,SCADA system and monitoring parameters are introduced.The abnormal and missing values in SCADA data are processed and normalized to remove the effect of magnitude.The characteristic parameters of main transmission system are selected by using Maximal Information Coefficient(MIC)correlation analysis method to prepare data for multi-conditions classification and condition assessment.(3)In view of the complex operating conditions of the main transmission system and the inadequacy of the current working condition classification method,a main transmission system condition classification method based on the operating characteristics of wind turbine and K-Means clustering algorithm is proposed.The traditional method of classifying conditions is according to the operating characteristics of wind turbine,the working conditions are divided into four stages: start-up stage,maximum wind energy tracking stage,constant speed stage and constant power stage.However,the maximum wind energy tracking phase data after division is relatively large,and its active power and rotational speed are changing with the wind speed,which makes the working conditions more complicated.In this paper,the unsupervised K-Means clustering algorithm is applied to subdivide the maximum wind energy tracking stage again to complete the division of the main transmission system working conditions,which solves the problems of uneven division of conditions and unknown distribution characteristics caused by the traditional method of dividing conditions.(4)In view of the deterioration of the performance of the wind turbine during operation,the evaluation of the condition of the wind turbine by simply using the condition monitoring results and setting a fixed threshold,which can cause the discrepancy between the evaluation results and the actual operating condition.The SDAE-BiLSTM model is used to predict the condition evaluation parameters of the wind turbine main transmission system.The Auto Encoder network is used to extract the implied features from SCADA data,so as to realize the compression and dimensionality reduction of the data features.The reduced dimensional data is used as the input to the Bi-directional Long Short-Term Memory(BiLSTM)model for the prediction of the main transmission system state index parameters.The final prediction results are compared with those of RNN,LSTM,BiLSTM and PCA-BiLSTM methods,and it is concluded that the SDAE-BiLSTM model has better prediction effect when processing SCADA data after condition classification.(5)In order to comprehensively consider the influence of the three condition evaluation indexes on the final evaluation results and to classify the health status of the main transmission system,the fuzzy evaluation method is used to evaluate the health status of the main transmission system.By using a fuzzy comprehensive evaluation algorithm,the residuals of the gearing high-speed shaft bearing temperature,active power and torque feedback predicted by the SDAE-BiLSTM model are used as indicators for the fuzzy comprehensive evaluation to establish the main transmission system health status assessment model.It is verified by abnormal data that the method can detect the decline trend of the main transmission system health status in advance and realize the early warning of main transmission system failure. |
PDF Full Text Request