The single neuron adaptive PID control is a simple intelligence control strategy. Commonly, the neuron learning is through the supervised Hebb learning rule. In this paper, a supervised Oja's learning rule is presented. In addition, the rule is introduced into linear neuron adaptive PID control algorithm. An analysis is given that the effect of learning rate on controlling. A decimal genetic algorithm (GA) is used to optimize the neuron's learning rate. To avoid premature convergence, the GA employs multi-operator.The Digital Signal Processor (DSP, TMS320F240), as core, has been used to perform the calculation of the intelligent control algorithm with its trait.The practical and simulation results of neuro-genetic adaptive PED control show that the proposed algorithm provides more improved control performance for the brushless DC motor (BLDC) controller.
|