Agent Based Soft Computing Approach for Component Fault Detection and Isolation of CNC X-Axis Drive System

A novel agent based soft computing approach is proposed for fault detection and isolation (FDI) systems for industrial plants, in particular a highly nonlinear CNC X-axis drive system’s component fault detection. The fuzzy-neuro architecture utilizes fuzzy clustering to build a nominal model, several fuzzy agents with local expertise, a fuzzy moderator for estimation of fault location, and finally several neuro-based (RBF) agents to estimate fault size. To illustrate the merits of the proposed method, it is applied to diagnosis of component faults of a CNC X-axis drive system amid significant noise levels. Simulation results demonstrate that the resulting FDI system is able to properly locate the fault types under all test conditions, and is sensitive to faults sizes as small as 0.5%.