INTELLIGENT RECOGNITION ALGORITHM FOR HUMAN DYNAMIC PARAMETERS OF WALKING LOAD

With the development trend of modern architectural structure "lighter and larger span", the structural design is prevailed by the vibration serviceability limit state rather than the ultimate state (i.e. load-carrying capacity of the structural elements). It can be predicted that the human-induced structural vibration serviceability will become more and more common. Accurate prediction of human-induced structural vibration response (acceleration response) and structural vibration characteristics (frequency and damping) is a prerequisite for evaluating structural vibration serviceability. The accuracy of human-induced structural vibration response is closely related to the human-induced load model. At present, the human-induced load proposed by various design codes is mostly based on certainty, ignoring the difference of human body, i.e. the randomness of human-induced load. To consider the randomness of human-induced loads and improve the calculation accuracy of structural vibration response for comfortable vibration analysis, an intelligent algorithm is developed based on walking experiments, theoretical studies (using an inverted pendulum model to simulate the entire walking process and perturbation method to establish a walking load theoretical model), and various intelligent algorithms (genetic algorithm, grey wolf algorithm, bat algorithm, cuckoo search algorithm, biogeography-based algorithm, and ant-lion optimizer) to identify the dynamic parameters (stiffness k_leg, length l₀, roller radius R, mass m, and initial velocity v₀) of walking loads. Compared with the experimental data (walking load of 25 participants), the intelligent recognition algorithm has the characteristics of high recognition accuracy and fast calculation efficiency.