Health monitoring of steel structures using Cuckoo Search algorithm-based ANN

The application of various computational methods in Structural Health Monitoring (SHM) of structures are gaining importance. One of the methodologies used is Artificial Neural Network (ANN). Though ANN can handle complex nonlinear functions that are intractable using traditional methods, the gradient descent nature of back propagation algorithms in ANN traps the solution in local minima, which prohibit it from finding the best solution. This can be resolved using various evolution algorithm combining with Artificial Neural Network, such as Cuckoo Search (CS) Algorithm, Genetic Algorithm (GA), Particle Swarm Optimisation (PSO) etc. This study presents a strategy SHM using a flexible combination of ANN and CS algorithm. This nature-inspired metaheuristic algorithm can train the parameters of ANN and can reduce the variation between the actual and predicted output. The strategy is applied on a steel truss bridge and a lattice offshore platform structure and its robustness is compared. Different damage scenarios were also considered. The study demonstrated the superiority of ANN-CS approach over ANN alone. Though ANN predicts the damage with the help of adequate vibrational data, ANN combined with Cuckoo Search algorithm shows good improvement in predicting the damage with lesser training data set.