一类随机曲线的控制生成

提出一类形状具有较好可控性和可预见性的随机曲线生成方法，用于描述不具有分形特征的随机现象，从确定的自由曲线出发，对自由曲线的每一型值点给予一个随机权系统，随机地改变型值点的线性组合关系，使自由曲线向随机曲线转变，应用自适应线性神经元网络对自由曲线的调配函数进行训练，由自适应神经网络的自适应学习机理控制随机权系数的取值，并通过控制神经网络的可调参数影响随机权第数的神经网络输出，进一步控制随机曲线的整体或局部形状，对神经网络可控参数的三种不同的设置方式，可把自由曲线转变为规则的、非规则的或具有分形特征的随机曲线，该文所提方法生成的随机曲线，具有较好的可控性，且方法简明、易于实现。

Modelling and Controlling of A Random Curve

In this paper, a new method of modeling and controlling random curve is presented for simulating random phenomena of non fractal character. From determinate free curve, every control point is given a random weight coefficient. The free curve is changed to random curve by modifying the linear combination among control points.Due to the mixed function of free curve is trained by applying Adaline (Adaptive Linear Neuron), the numerical value of random weight coefficient can be controlled by adaptive learning mechanism of Adaline. Further, the whole or part of random curve can be controlled with controlling adjustable parameter of Adaline influenced random coefficient. Three ways of setting the adjustable parameters are put forward to change the free curve into regular, irregular or fractal characteristic random curve respectively. The examples verify that with our method the generated random curve is more controllable and easy to be realized.