An Improved Particle Swarm Optimization Algorithm for the Resolution of Superimposed Motor Unit Action Potentials

This paper resolves the difficult superimpositions of Motor Unit Action Potentials (MUAPs) measured from Electromyographic Recordings using Improved Particle Swarm Optimization (IPSO) algorithm. The contraction of muscles depend on the activity of motor neurons (MNs) each giving rise to Motor Unit Action Potentials (MUAPs). Since Motor Unit (MU) firings are asynchronous, MUAPs inevitably overlap from time to time over the course of recording, making them difficult to locate. Resolution of these superimposed potentials into individual constituents is a combinatorial optimization problem which requires determining of their individual MUAPs as well as their relative time shifts. PSO is a new random computational method for tackling optimization functions. However, it is easily trapped into local optimum when solving the difficult superimposition, which makes the performance of PSO greatly reduced. To overcome this shortcoming, this paper proposes Improved Particle Swarm Optimization (IPSO) algorithm, which uses the third particle, guides the current particles’ velocity updating rule and reduces the probability of trapping into the local optimization. Thus the proposed method highly reduces the time taken to resolve difficult superimposition and gives 99.8% accuracy in simulation study. Numerical results shows that the proposed algorithm has great performance in accuracy and the results are compared with conventional particle swarm optimization.