AN SHM APPROACH USING MACHINE LEARNING AND STATISTICAL INDICATORS EXTRACTED FROM RAW DYNAMIC MEASUREMENTS

Abstract

STRUCTURAL HEALTH MONITORING USING RAW DYNAMIC MEASUREMENTS IS THE SUBJECT OF SEVERAL STUDIES FOR THE IDENTIFICATION OF STRUCTURAL MODIFICATIONS OR, MORE SPECIFICALLY, FOR DAMAGE ASSESSMENT. TRADITIONAL DAMAGE DETECTION METHODS ASSOCIATE STRUCTURAL MODAL DEVIATIONS TO DAMAGE. NEVERTHELESS, THE PROCESS USED TO DETERMINE THE MODAL CHARACTERISTICS CAN INFLUENCE THE RESULT OF DAMAGE IDENTIFICATION METHODS, WHICH COULD LEAD TO A LOSS OF INFORMATION AND INTRODUCE ADDITIONAL UNCERTAINTIES. THUS, TECHNIQUES COMBINING MACHINE LEARNING AND STATISTICAL ANALYSIS APPLIED DIRECTLY TO RAW MEASUREMENTS ARE BEING DISCUSSED IN RECENT RESEARCHES. THE PURPOSE OF THIS PAPER IS TO COMPARE TWO COMPUTATIONAL INTELLIGENCE ALGORITHMS TO IDENTIFY STRUCTURAL CHANGES USING STATISTICAL PARAMETERS OBTAINED FROM RAW DYNAMIC DATA. THE ALGORITHMS, BASED ON ARTIFICIAL NEURAL NETWORKS AND SUPPORT VECTOR MACHINES, ARE EVALUATED THROUGH NUMERICAL SIMULATIONS USING A SIMPLY SUPPORTED BEAM MODEL AND THROUGH EXPERIMENTAL TESTS PERFORMED ON A SIMILAR BEAM STRUCTURE AND ALSO ON A RAILWAY BRIDGE, IN
FRANCE. FOR ALL CASES, DIFFERENT DAMAGE SCENARIOS WERE CONSIDERED. THE RESULTS OBTAINED ENCOURAGE THE DEVELOPMENT OF COMPUTATIONAL TOOLS USING STATISTICAL INDICATORS OF ACCELERATION MEASUREMENTS FOR STRUCTURAL ALTERATION ASSESSMENT.