Genetic algorithms applied to estimate 6-parameters model which define analytical function to simulate the motor unity force from experimental measures

Abstract

An individual twitch of motor unity MU force can be simulated by 6-parameters model, then a train of repeated pulses evokes a tetanic contraction. However, is a hard issue to estimate the parameters for one twitch, and the way to put many of them, such that can reproduce a shape tailored a muscle contraction generated by a set of stimulation pulses. In this work, genetic algorithms are applied to estimate nine parameters from experimental measures, where six are used to define a 6-parameters model, and the other three are used to generate a train of pulses, which simulate muscle contraction. The measured data by dynamometer were used with genetic algorithms to estimate the nine parameters, after that, their information was compared with data measured by electromyograph. The results show that calculated parameters, like as latency time, contraction time, repetition time, half contraction, intermediate time of relaxation force and number of pulses, can be applied to generate a set of successive twitches to simulate the muscle contraction, with an error less than 5%. Besides, the simulation results prove that force level depends directly on repetition frequency, number of pulses and amplitude of stimulation signal. finally, in this work the genetic algorithms worked like an excellent tool to optimize and validate a theoretical model with the experimental data, a despite the process last more than 40 h for each training, the method was friendly to apply and to reconfigure many options before to find the best solution.