06.1 - Flight Dynamics and Control (Control & Modelling)
NEURAL NETWORK BASED SLIDING MODE CONTROL FOR HYPERSONIC LONGITUDINAL CONTROL
H.J. Kim¹, J.H. Shin¹; ¹Chungbuk national University, South Korea
This paper applies and evaluates a neural-network-based tuning method that adjusts the design parameters of Sliding Mode Control (SMC) online to improve simultaneous velocity V and altitude h tracking in hypersonic longitudinal dynamics. The proposed method preserves the SMC structure. It makes a neural network tuner adjust the sliding surface parameters of the velocity and altitude channels over time. The neural network tuner receives the command, current state, and sliding surface observed during control as inputs. It outputs λ and is updated online during simulation. The performance is evaluated by comparing a conventional SMC with fixed λ and the proposed neural-network-aided NN-SMC under the same flight condition. The comparison uses the time responses of the velocity and altitude tracking errors and their RMSE/MAE values. It also presents the control input RMS and control energy to analyze both tracking performance and control characteristics.
DOI: 10.71945/icas2026_0872