Research On Data Denoising And Prediction Based On Remote Monitoring Platform Of Marine Shafting

In order to simulate the running status of the shaft system during the course of the ship sailing,many colleges and universities have established the experiment platform.The experimental platform is generally composed of shaft bench,industrial computer monitoring system.But most of the experimental platform is real-time monitoring,and data recorded in the IPC next to the shaft bench,which did not achieve remote monitoring function,and does not meet the remote monitoring trends.The recorded data are generally the original data collected by the sensor,and the sensor in the collection process will be subject to various interference,therefore,the collected data with a certain noise,which also caused data display not very smooth in the real-time monitoring process.In order to solve these problems,this paper focus on the three aspects of remote monitoring,data denoising and real-time forecasting:(1)In view of the problem of the exiting software of the shafting experiment platform can not carry on the remote monitoring problem,the overall design of the remote monitoring experiment platform is carried out.The design part focuses on the implementation details of data analysis,storage scheme,real-time display and historical data display and download,and finally implements the remote monitoring platform based on ASP.NET MVC framework.(2)In view of the problem of noise in the historical data collected by the shafting experimental platform,the wavelet transform method was used to de-noise.Due to the selection of the wavelet basis,decomposition scale and the threshold function will have a great influence on the denoising effect.According to the Hubert's index theory and smoothness curve,this paper determines the best decomposition scale is 3.By using 8 commonly used wavelet basis and soft and hard threshold function on the monitoring data,the result shows that,coif2 wavelet basis and soft threshold function are more suitable for denosing.After the wavelet transform parameters are confirmed,the method of integrating the wavelet denoising program into the remote platform is proposed.(3)This paper introduces the theory of support vector machine,and proposes a real-time prediction method based on the wavelet and support vector machine for real-time prediction of shaft power.In the support vector machine training,the choice of parameters has great influence on the prediction results.This paper using grid search method to find the best parameter of support vector regression,and finally confirm the best parameters are g equals 0.0039781 and C equals 4.89842.After the parameters are confirmed,the prediction model is determined.After using the model to predict,the prediction result shows that this method is feasible.In order to illustrate the advantages of the method,this paper compares this method with the gray prediction method,and the result shows that,this method has better predictive validity.