Robotic liquid tension identification with particle swarm optimized neural network

Sensorless tension control of belt-driven robotic arm, based on dual-motor synchronous automation system, requires one obtains the instantaneous magnitude of tension difference, between right hand side and left hand side of the belt that driving the robot arm. This in turn depends on the instantaneous speed interaction with the processed unit, through the internal friction of high viscous liquid contact. In this paper, we present the nonlinear liquid friction identification on the base of stator current with present and its previous two values. The fundamental equation of dual motor systems for liquid friction control, principles of mass-flow systems and the novel method of finite tension difference using particle swarm optimization trained neural network are presented. A three-layer feed-forward neural network is optimized. The simulation based on the limited factory experiment test data shows the system with Swarm Intelligence method is practical for thick adhesive application and processing automation on industry assembling lines.