Research On Performance Degradation Evaluation Method Of Power Plant Equipment Based On Data Driving

Thermal power generation is the main mode of power generation in our country,and the health status of power plant equipment and system is an important factor affecting the economy and safety of power plant.The performance degradation evaluation of power plant equipment is of great significance to guide predictive maintenance.With the rapid development of power plant information system,massive process data can be stored and utilized.In this paper,thermal process operation data are used to track and evaluate the performance degradation process of power plant equipment by using data-driven model to mine the characteristic changes of equipment.The main research contents are as follows:Firstly,the data model of the thermal process is established.Aiming at the over-fitting problem of single model,a hybrid ensemble learning modeling method based on secondary learning is proposed,which improves the accuracy and generalization ability of the model.The performance of the model is verified by a simulation example,and the prediction model of ammonia injection quantity of SCR denitrification system is established.Secondly,in order to obtain the performance degradation index series of catalysts in SCR system,the prediction model of ammonia injection quantity in each month is established by multi-period and multi-model modeling method.The ammonia injection quantity is obtained by inputting a unified standard working condition,which is used as the activity monitoring index of the catalyst in each period of time.On this basis,the degradation trend of the catalyst is further studied.Aiming at the unconventional factors in the time series of thermal process data,an ARMA-GM(1,1)combination forecasting model based on wavelet decomposition is proposed in this paper,which separates the trend items and random items from the data,improving the prediction accuracy,and the ammonia injection quantity of SCR system in the next few months are predicted.Then,for the continuous operation data of thermal process,the process of equipment performance degradation is analyzed from the point of view of concept drift.The offline model of the health state is established,and the real-time running data are input into the model for testing.The statistical index R~2,which measures the performance of the model,is constructed to monitor the degradation degree of the equipment,and the degradation trend of the equipment is predicted.The simulation example and operation data of SCR system are used to verify the model,which proves the feasibility of the method.Finally,based on the risk theory,the comprehensive risk assessment of the operation state of power plant equipment is carried out from two aspects:the current degradation degree and the predicted degradation degree of power plant equipment.Aiming at the complexity of thermal system and the uncertainty caused by human factors,a fuzzy inference model combined with expert knowledge is proposed,and the fuzzy relationship between comprehensive risk index and equipment deterioration degree is established,which improves the reliability of the evaluation results.