EVALUATION OF GIMBAL ATTITUDE CONTROL OF A SPACE ROCKET SYSTEM USING CNN-PID

Abstract

This study evaluates the effectiveness of a Convolutional Neural Network-based Proportional-Integral-Derivative (CNN-PID) control system for gimbal attitude management in rocket systems, focusing on enhancing performance during critical flight phases. The primary purpose is to improve the responsiveness, accuracy, stability, and fuel efficiency of gimbal control compared to traditional PID methods. The problem addressed is the limitations of conventional gimbal control, particularly in dynamic environments, which can result in significant control errors and inefficient fuel usage. To achieve these objectives, a comprehensive simulation of the CNN-PID controller was developed and tested against standard PID control under varying flight conditions. The CNN model was trained on historical flight data to improve its predictive capabilities, while the PID parameters were optimized to minimize control error. The evaluation metrics included response time, control error, stability index, and fuel consumption rates. The CNN-PID system demonstrated a response time improvement of 35%, achieving a stabilization time of 2 seconds compared to 3.1 seconds for traditional methods. Furthermore, the control error was reduced by 40%, with an average error of 0.5° compared to 0.83° in conventional PID control. In terms of stability, the CNN-PID controller maintained a stability index of 0.9, indicating robust performance during dynamic flight scenarios. Fuel efficiency analysis showed a reduction in fuel consumption by 25% during operations, reflecting a significant enhancement in operational cost-effectiveness. This research supports the adoption of advanced control systems in aerospace applications, emphasizing the need for policies that encourage the integration of machine learning techniques to optimize rocket performance. The findings highlight the potential for CNN-PID control to revolutionize gimbal attitude management, paving the way for future advancements in space exploration technologies.