Damage assessment in structures using artificial neural network working and a hybrid stochastic optimization

Abstract

Artificial neural network (ANN) has been commonly used to deal with many problems. However, since this algorithm applies backpropagation algorithms based on gradient descent (GD) technique to look for the best solution, the network may face major risks of being entrapped in local minima. To overcome those drawbacks of ANN, in this work, we propose a novel ANN working parallel with metaheuristic algorithms (MAs) to train the network. The core idea is that first, (1) GD is applied to increase the convergence speed. (2) If the network is stuck in local minima, the capacity of the global search technique of MAs is employed. (3) After escaping from local minima, the GD technique is applied again. This process is applied until the target is achieved. Additionally, to increase the efficiency of the global search capacity, a hybrid of particle swarm optimization and genetic algorithm (PSOGA) is employed. The effectiveness of ANNPSOGA is assessed using both numerical models and measurement. The results demonstrate that ANNPSOGA provides higher accuracy than traditional ANN, PSO, and other hybrid ANNs (even a higher level of noise is employed) and also considerably decreases calculational cost compared with PSO.

Damage Detection in Structural Health Monitoring Using an Integrated ANNIRSA Approach

“…The efficacy of an ANN predominantly hinges on whether the network resides within the most favorable local minima. Hoa et al [12] posited that the strategic selection of an initial starting point merely serves to assist the network in navigating around local minima within a subset of rudimentary problems. Consequently, this methodology has been adopted by numerous researchers to address the local minima challenges encountered by ANNs [16,17].…”

“…One of the most common approaches to solving the limitations of gradient-based optimization methods is using metaheuristic algorithms, including evolutionary algorithms and swarm intelligence, which have been widely investigated to obtain generalized feedforward neural networks (FNNs) for specific problems due to the limitations of gradient-based optimization methods [9]. Additionally, the application of particle swarm optimization (PSO), genetic algorithms (GAs) or hybrid algorithms has been shown to effectively address local minimum problems in the training processes of neural networks [10][11][12]. Furthermore, the susceptibility of algorithms like grey wolf optimization (GWO) to local optima and slow convergence has been noted, resulting in degraded performance [13].…”

A promising approach using Fibonacci sequence-based optimization algorithms and advanced computing

“…or human-induced impacts (overload, collision, etc.) 1 – 5 . In addition, bridges also have their own vibration patterns that possibly cause amplified vibrations when the natural frequencies of the bridges coincide with those of moving vehicles.…”

Optimization of Wire Arc Additive Manufacturing Process Parameters for Low‐Carbon Steel and Properties Prediction by Support Vector Regression Model

“…[27][28][29][30] In summary, the optimization of the process parameters is crucial to achieving the desired performance of the WAAM parts. ML-based approaches, such as ANN, [31,32] DT, [17] Fibonacci sequence-based optimization, [33] Shrimp and Goby association search algorithm, [34] and SVM, [23] have emerged as powerful tools for optimizing the process parameters and property prediction in WAAM compared to traditional methods. The significance of this research lies in the optimization of a multitude of process parameters, which surpasses the limited scope of previous studies that have examined fewer variables.…”