Quantum neural networks based Lyapunov stability and adaptive learning rates for identification of nonlinear systems

This paper presents an identification model based on quantum neural network for engineering systems. Quantum neural network (QNN) is a superior strategy to improve the computational efficiency for conventional neural network structures due to their unprecedented computation capabilities. The structure of identification model consists of multi-layer QNN in which qubit neurons are used for data processing. The identification model stability is ensured by introducing a learning algorithm based on Lyapunov theorem for online tuning of the QNN. Furthermore, the convergence and stability of the identification structure are accelerated with developing adaptive learning rates based on Lyapunov theory. The effectiveness of the developed identification model is confirmed with introducing it for two engineering processes and comparing its modeling results with other structures. Simulation results reflect the high superior performance of the developed model compared with other approaches.