自适应B样条模糊神经网络控制器的设计

The Design of Adaptive B-Spline Fuzzy Neural Network Controller

B样条具有最小局部支撑和易于实现的优点。文章利用多变量B样条网络在运算表达式上与模糊神经网络结构之间的对等关系，并通过对其权值的训练，设计出自适应B样条模糊神经网络控制器。应用于具有严重非线性摩擦力影响的速度跟踪系统的仿真实验表明，所设计的控制器完全等价于模糊神经网络控制器，同时在计算量和实现上具有明显的优势。

B-spline has the advantages of local compactly support and easy realization. The paper proposes the adaptive B-spline fuzzy neural network controller by use of the equivalent relationship between the calculation of multi-variable B-spline networks and the construetion of fuzzy neural networks, and through the training of its weights. The simulation experiment applied to the velocity tracking system with high non-linear friction effect shows that the controller proposed is entirely equivalent to the fuzzy neural network controller, and at the same time, it has much less calculation and easier realization.