Swing Control of 2D Overhead Crane Using Proportional Integral Derivative Neural Network

There are several factors that contributed to swings in crane system such as payload hoisting and non-zero initial condition. Payload hoisting increases the moment of inertia, which can amplify swing amplitude due to the greater energy needed to stop and reverse motion while non-zero initial conditions introduce an initial angular momentum, potentially leading to irregular swinging patterns and complicating control algorithms. This paper proposes an Adaptive Proportional Derivative like Neural Network (APIDLNN) controller for a swing control of overhead crane under the effects of payload hoisting and non-zero initial condition. The adaptive mechanism of the neural network has the capability in predicting the control parameters in real-time to eliminate the payload swing under these effects. Additionally, Genetic Algorithms (GA) are introduced for optimizing APIDLNN initial weights and enhancing the control efficiency. For the controller stability, the learning rate of APIDLNN is designed based on a discrete-type Lyapunov function. Simulation results demonstrate the proposed algorithm effectiveness in reducing the payload swing under the effects of payload hoisting and non-zero initial condition. The APIDLNN performances are verified by reductions of 93.08% and 64.71% in the overall and residual swing responses as compared to the comparative method.