Wavelet support vector machine-based prediction for emptying time of mobile pipeline

For the mobile pipeline, if fuel oil is single and the terrain where the pipeline located is changeless, the pipeline length, air compressor discharge pressure and air displacement will be main factors affecting pipeline emptying operation time. But these factors follow a complex nonlinear relationship of impacts on the emptying time. Experimental data for emptying process of mobile pipelines is taken as the basis to analyze the law for changes in the emptying time with influencing factors. As shown from the analysis, the factors follow a better correlation with the emptying time, and the method of support vector regression can be used to predict the emptying time. In the practice and prediction of sample data, Morlet wavelet kernel-based support vector regression method is used and its prediction effects are compared with those of Gaussian kernel-based support vector regression, which indicates that the wavelet kernel can provide better prediction effects. In addition, a pipeline gas-cap emptying time prediction formula is given.