Moving force identification: A time subdomain genetic algorithm approach

For maintenance and assessment of existing bridges and design of new bridges, it is important to know not only the static axle loads of the vehicles using the bridge but to also have a good estimate of the dynamic force due to road roughness and vehicle-bridge interaction. Direct measurement of such forces is difficult, subjected to bias or limited in the locations that can be measured. In recent years, several methods have been proposed to indirectly identify moving loads from the measured dynamic response of structures. However, many of these methods are sensitive to noise and the dynamic component of the identified force tends to be inaccurate. A novel method for indirectly identifying the force time history of unknown loads moving across a bridge is presented in this paper. Using the dynamic response of the structure, a method is proposed by dividing the time history of interest into small time subdomains which can be quickly and accurately identified using a modified genetic algorithm (GA) strategy. The force is identified progressively with the identified force from each subdomain used to simulate the structural response so as to determine the initial conditions for the following subdomain. To demonstrate the accuracy and effectiveness of the new approach, a numerical study of moving force is presented. A very good approximation of the force is obtained even when the acceleration response measurements are contaminated with 10% noise.