Fuzzy pattern recognition technique for crack propagation on earplate connection of guyed mast under wind load

Abstract

A 2-stage damage identification technique is presented for identifying crack lengths on cable-tower connections of guyed masts. It is based on fuzzy pattern recognition theory and uses the cable force and the strain at the key locations of the earplate connection as input parameters. The methodology is developed with a numerical model under stochastic wind excitation and verified by numerical simulation and experiment. First, the relationships of cable force, strain on the connection, and crack length are explored. The wind-induced responses of guyed mast are analyzed under different wind speed and directions to form a database of cable forces. The responses of a typical connection with different crack lengths are simulated to investigate the strain sensitivity to crack propagation. Its results are used to form a second-stage strain database. The first stage of the fuzzy identification consists in identifying the interval of cable forces on the basis of cable force measurements. Then, a subjection degree function is formed between the measured strain and the developed strain database to judge the damage state of the connection by identifying the existence of a crack and estimating its length. The results of a numerical case study and further experiment are presented and show the potential and practicability of the technique to estimate the right range of crack length on earplates. Therefore, the proposed technique is practical and feasible to predict the damage state of earplate and can provide valuable prewarning information for structural safety.