无人机全包线模糊T-S建模

Design of fuzzy T-S model of unmanned aerial vehicle in full envelope

针对无人机大包线一体化飞行控制要求,提出全包线模糊T-S(Takagi-Sugeno)建模方法.该方法根据非仿射系统的局部线性化原理,将模糊T-S建模转化为仅对模糊规则中隶属度函数的中心和宽度的优化过程,优化的代价函数为模糊T-S模型对无人机全包线稳定性和操纵性的逼近误差的加权值.基于敏感度逐步扩展前件变量的模糊集以实现全局优化,确定模糊规则的数量和隶属度函数的初值.采用对正则因子启发式调整的Levenberg-Marquardt算法进行快速的局部优化.算例表明,建模算法收敛迅速,所建立的模糊T-S模型采用少量模糊规则实现了对无人机全包线稳定性和操纵性的高精度逼近,适用于无人机全包线一体化控制.

The method of constructing fuzzy T-S model in full envelope was designed for the integrated flight control of unmanned aerial vehicle （UAV） in entire envelope. Based on the local linearization of non-af- fine system, the design of fuzzy T-S model was transformed to the optimization of the centers and width of membership functions, with the weighted approximation errors of the stability and handing performances of UAV as cost function. The global optimization was conducted via extension of fuzzy sets of premise variables according to the sensitivity, providing the number of fuzzy rules and initial value of membership functions. Fast local optimization was performed employing Levenberg-Marquardt algorithm with the heuristic modification of regular factor. The example shows the algorithm converges rapidly, and that the fuzzy T-S model constructed realizes high-precision approximation of stability and handing performances of UAV in full envelope with less fuzzy rules, which is suitable for the integrated control of UAV in full envelope.