Research On Optimization Of Operation And Maintenance Based On Multidimensional Time Series Data

With the development of detection equipment,the measurement data in various fields have shown multi-dimensional,timing and other characteristics.Compared with other data,time series data has the characteristics of continuity,randomness and periodicity.These characteristics of time series data indicate the feasibility of data prediction and the difficulty in prediction.Through the analysis and prediction of time series data,we can mine the potential value hidden in the data,avoid the phenomenon of data waste,improve related work efficiency,and provide guidance and basis for future work through the predicted data.In order to avoid risks and reduce operation and mainte-nance costs,the prediction and analysis of multidimensional time series data are widely used in many fields.With the continuous development of HVDC transmission,its safety and stability requirements are becoming higher and higher.The rectification and in-verter work of the converter station is an important part of HVDC transmission.In the converter station,the cooling system is very important for the stable operation of the thyristor converter valve.At present,there is little research and analysis on the multidi-mensional time series data of the valve cooling system of the converter station.There-fore,this paper aims at the application of multi-dimensional time series data in opera-tion and maintenance optimization.The following works are completed based on the actual data of the valve cooling system of the converter station:(1)Data preprocessing.On the basis of collecting and sorting historical data,the identification and correction of abnormal data and the filling of missing data are real-ized.In terms of data filling,the nearest neighbor filling algorithm based on the time series sliding window idea is adopted.In the data filling,the time and space character-istics between the data are fully considered,and the accuracy of missing data filling is improved.(2)Multi-dimensional time series data prediction.By analyzing the characteristics of time series data,comparing and analyzing several different model algorithms for time series data prediction,the long-term and short-term memory network is used to realize the prediction of the time series data of the valve cooling system of the converter station.(3)Calculation of cooling capacity and failure prediction of valve cooling system.Based on the actual data of the valve cooling system of the converter station,the func-tion and structure of the valve cooling system are analyzed,and the cooling capacity of the valve cooling system is divided into the external cooling water cooling capacity and the internal cooling water cooling capacity.Based on the prediction data,the analysis and judgment of the external cooling capacity and the calculation of the internal cooling capacity are realized.At the same time,the related process of valve cooling system failure prediction is analyzed.