SENSOR LOCATION OPTIMIZATION FOR EFFICIENT STRUCTURAL HEALTH MONITORING THROUGH INTELLIGENT DESIGN

Abstract

Vibration based Structural health monitoring (SHM) is an emerging field in Civil engineering, which evaluates the structure’s integrity and performance. It facilitates the early identification of damage/deterioration and reduces maintenance costs. Due to technological development in research nowadays, sensors play a critical role in SHM, by offering consistent and reliable measurements of structural dynamic parameters, such as displacement, temperature, strain, and vibration, thereby enabling timely detection of potential issues, targeted maintenance, and repairs. However, choosing the correct sensor locations on a structure can be ambiguous, hence, the development of an effective sensor optimization technique is essential so that it can reduce the maintenance cost. This involves identifying the most relevant parameters to monitor and determining the optimal number and location of sensors. Too many sensors can result in excessive data and processing requirements, leading to increased costs and reduced reliability. Conversely, too few sensors can result in incomplete data and reduced accuracy, leading to missed or delayed detection of potential problems. Hence, sensor optimization is necessary to ensure optimal data acquisition and efficient monitoring. The algorithm proposed in this study for sensor optimization employs Particle Swarm Optimization (PSO), Bayesian Optimization (BO), and Optuna to determine the optimal sensor locations for monitoring a structure. To achieve this, the algorithm employs modal assurance criteria (MAC), Fisher Information Matrix (FIM), and both MAC and FIM as objective functions for each of the optimization algorithms in addressing the Optimal Sensor Location Problem (OSLP). The algorithm aims to provide an efficient and effective method for determining the best sensor locations for monitoring the structure’s health and identifying system parameters, such as frequencies, mode shapes, and damping ratios. The algorithm aims to provide a reliable method for identifying system parameters such as mode shapes, frequencies, and damping ratios by finding the best sensor locations. By leveraging the strengths of PSO, BO, and Optuna, the algorithm optimizes the objective function to identify the best sensor locations for the structure and also gives minimum and maximum number of sensor locations for identifying the system parameters. Based on the optimized results obtained from utilizing the optimization techniques in the OSLP, it can be concluded that the optimal sensor locations can guarantee improved linear independence of the MSV and are validated with experimental modal parameters. Furthermore, the proposed algorithm can provide a reliable and efficient method for determining the best sensor locations to monitor a structure. Such findings can greatly benefit the field of structural health monitoring and contribute to enhanced safety and reliability of engineering structures.