A Regression Algorithm for the Smart Prognosis of a Reversed Polarity Fault in a Photovoltaic Generator
As a critical tool that facilitates control strategy design, performance analysis and overall systems integration, dynamical engine models play important roles in developing advanced powertrain and vehicle technologies. Methodologies for effective engine modeling and strategy calibration are in high demand to meet stringent performance specifications under time/cost constraints. Recently, we explored the use of support vector machine (SVM) for engine modeling and identified several challenging issues in capitalizing this powerful tool for powertrain applications . In this paper, we exploited the regressor structure of the SVM to separate the auto-regression (AR) from the moving average (MA) in an attempt to build a concise engine model with reduced computational effort. The new structure allows us to use different kernel functions for the AR and MA to characterize their roles, thereby providing more flexibility in the model structure. The linear programming SVM-ARMA2K is developed and then successfully applied to identify a representative dynamical engine model. A simulation study demonstrates the potential and practicability of the proposed approach.