Advanced Delta Robot Control Using Adaptive Neural PID with Recurrent Fuzzy Neural Network Modeling

 This study presents an adaptive control method for a 3-degree-of-freedom parallel Delta robot using a neuron PID controller combined with a recurrent fuzzy neural network (RFNN) identifier. Due to the nonlinear kinematics and dynamics, coupling, and load changes of the Delta robot, traditional PID controllers often do not ensure optimal control quality, thereby requiring adaptive intelligent control techniques [1], [2]. In the proposed model, the PID is represented as a linear neuron capable of self-updating the parameters Kp​, Ki​, Kdd based on the Jacobian information estimated online by the RFNN identifier, inheriting the backpropagation learning principle in the recurrent neuro-fuzzy system [3]. MATLAB simulation results show that the response time is improved, the steady-state error is eliminated, and the system maintains its stability when the load changes, which is consistent with previous studies on neural network-based adaptive PID for nonlinear systems [4]–[6]. This method contributes to affirming the effectiveness of combining PID neuron and RFNN for precise control of parallel robots.