Abstract:Accurate and rapid recognition of defects is essential for structural integrity and health monitoring of in-service device using eddy current (EC) non-destructive testing. This paper introduces a novel model-free method that includes three main modules: a signal pre-processing module, a classifier module and an optimisation module. In the signal pre-processing module, a kind of two-stage differential structure is proposed to suppress the lift-off fluctuation that could contaminate the EC signal. In the classifier module, multi-class support vector machine (SVM) based on one-against-one strategy is utilised for its good accuracy. In the optimisation module, the optimal parameters of classifier are obtained by an improved particle swarm optimisation (IPSO) algorithm. The proposed IPSO technique can improve convergence performance of the primary PSO through the following strategies: nonlinear processing of inertia weight, introductions of the black hole and simulated annealing model with extremum disturbance. The good generalisation ability of the IPSO-SVM model has been validated through adding additional specimen into the testing set. Experiments show that the proposed algorithm can achieve higher recognition accuracy and efficiency than other well-known classifiers and the superiorities are more obvious with less training set, which contributes to online application.

An Intelligent Model Validation Method Based On Ecoc Svm

Research on SVM integration for intelligent sensor systen

A robust reliability prediction method using Weighted Least Square Support Vector Machine equipped with Chaos Modified Particle Swarm Optimization and Online Correcting Strategy

Support Vector Machines-Based Method for Machining Error Prediction Modeling

M-estimator Based Support Vector Machine and Its Application

An Improved Error-Correcting Output Coding Framework with Kernel-Based Decoding

A Cuckoo Search-Support Vector Machine Model For Predicting Dynamic Measurement Errors Of Sensors

A New Validation Metric for Models with Correlated Responses Using Limited Experimental and Simulation Data

SVM-Based Cost-sensitive Classification Algorithm with Error Cost and Class-dependent Reject Cost

Modeling of CNC Machine Tool Thermal Errors Based on LS-SVM within Bayesian Evidence Framework

Research on the Assessment for Air Environment Quality Based on Support Vector Machine

An Improved Pso-Svm Model For Online Recognition Defects In Eddy Current Testing

Support Vector Machine-Based Multi-Model Predictive Control

An Intelligent Validation System of Simulation Model

Synthetic Error Modeling for NC Machine Tools Based on Intelligent Technology

Key Problems in Validation of Intelligent Models

Hybrid SVM algorithm oriented to classifying imbalanced datasets

An Evidence Theory-Based Validation Method for Models with Multivariate Outputs and Uncertainty

A Model Validation Method Based on the Orthogonal Polynomial Transformation and Area Metric

A New Methodology For Simulation Validation And Model Modification

The Validation of A Microscopic Simulation Model: A Methodological Case Study