PROJECT TITLE :

A Novel Dynamic-Weighted Probabilistic Support Vector Regression-Based Ensemble for Prognostics of Time Series Data

ABSTRACT:

In this paper, a completely unique Dynamic-Weighted Probabilistic Support Vector Regression-based mostly Ensemble (DW-PSVR-ensemble) approach is proposed for prognostics of your time series data monitored on components of complicated power systems. The novelty of the proposed approach consists in i) the introduction of an indication reconstruction and grouping technique suited to time series data, ii) the utilization of a changed Radial Basis Operate (RBF) kernel for multiple time series data sets, iii) a dynamic calculation of sub-models weights for the ensemble, and iv) an aggregation methodology for uncertainty estimation. The dynamic weighting is introduced within the calculation of the sub-models' weights for every input vector, based mostly on Fuzzy Similarity Analysis (FSA). We tend to take into account a true case study involving twenty failure eventualities of a part of the Reactor Coolant Pump (RCP) of a typical nuclear Pressurized Water Reactor (PWR). Prediction results are given with the associated uncertainty quantification, under the idea of a Gaussian distribution for the expected price.