There are parameter variation effects that would reduce the performance of missile terminal guidance system, e.g., target maneuverability, missile autopilot time constant, turning rate time constant as well as radome slope error effects. To solve this problem this research proposed a novel neural-fuzzy missile terminal guidance law by applying three different neural network optimization algorithms alternatively in each step, such as the Gradient Descent (GD), SCG (Scaled Conjugate Gradient), and Levenberg-Marquardt (LM) methods. Moreover, the missile turning rate time constant, autopilot time delay, target maneuverability, glint and fading noises, radome slope error, missile initial heading error as well as acceleration limits were taken into consideration. On the other hand, performance comparisons with the proportional navigation (PN) method for not only the lower and higher altitudes but the lateral and head-on interceptions were also made. One can see that the miss distances, acceleration commands and engagement times by using the proposed guidance law are lower than the other methods for the encountered engagement conditions.
Lin, Jium-Ming and Lin, Cheng-Hung
"NOVEL INTELLIGENT NEURAL GUIDANCE LAW BY USING MULTI-OPTIMIZATION ALGORITHMS,"
Journal of Marine Science and Technology: Vol. 25:
1, Article 1.
Available at: https://jmstt.ntou.edu.tw/journal/vol25/iss1/1