基于形态谱和形状特征的平面四杆机构设计方法

Synthesize Four-bar Linkage Based on the Curve's Morphological and Shape Characteristics

连杆轨迹曲线与连杆机构参量之间存在着复杂的非线性关系,用具有旋转、缩放、平移不变特性的形状谱和几何参数作为联合特征,描述了连杆轨迹曲线的结构形态特征和整体形状特征,并构建曲线形态的识别、比较方法,实现了连杆轨迹曲线描述最小的数据冗余;利用人工神经网络拟合了连杆轨迹曲线与连杆机构之间的非线性关系,获取与预期轨迹最为吻合的连杆曲线参数,实现了机构的快速综合设计。以512幅连杆曲线图像为学习库,64幅图像为工程测试样本的实验数据说明了本文方法的可行性,并分析了误差产生的原因。

There is a complicated nonlinear relation between the coupler curve of the four-bar linkage and its configuration parameters. The morphological characteristics are invariant moments by orientation, scale and position. The morphological characteristics by the coupler curve of the four-bar linkage and geometry characteristics as the united characteristics were described, and some methods to identify and compare the curves were built, the minimal data redundancy of the curves were achieved, finally the nonlinear relationship between the characteristics of curves and the parameters by using neural network was synthesized and the characteristics of the curves which were close to the anticipated target to realize the quick synthetic design of the linkage were chosen. 512 images have been used as study samples and 64 images were used as test samples. The data of the experiment explained the feasibility of the method and the cause of the error was analyzed.