Novel Active Control Algorithm for Specific Target Reduction Using Neural Network

A novel active control algorithm based on the neural network has been proposed for seismic control of a single-storey building frame. The proposed algorithm is based on the assumption that the shape of the time history of the control force is the same as that of the shape of the time history of the earthquake. Numerical analysis of the governing equation of motion, utilizing state-space methodology, is employed to derive the responses. The earthquake time histories, compatible with the response spectrum for seismic zones three, four, and five, were generated in accordance with IS 1893: 2016 and utilized as the training dataset for the neural network. The neural network is trained offline, with target responses and ground motion data as inputs, while the output generates the required control force. The trained neural network is then incorporated into the online simulation to generate the desired control force for specific target reduction. The study results reveal that the reduction provided by the proposed control algorithm is higher than the target level. Additionally, time-step delay analysis demonstrates the reliability of the developed control algorithm.