Genetic algorithm based optimization of Kullback Information Criterion: Improved system identification of skeletal muscle force and sEMG signals

This paper focuses on determining the sensitivity of the number of data points used in computing the Kullback Information Criterion (KIC) for the use in sensor data fusion. The primary objective of the sensor fusion is to improve the extraction of dynamic models relating Surface Electromyogrphic (sEMG) signals with the corresponding skeletal muscle force signals. The proposed approach utilizes a pre-processing of the sEMG data with a Half-Gaussian filter. System Identification techniques are employed to extract a relationship between the sEMG and the skeletal muscle force. In this paper linear and non-linear models are inferred from the fused data to describe the sEMG/force relationship. In order to optimize the number of data points for finding the optimum KIC, a Genetic Algorithm (GA) is used.