摘要
在自然科学中，常常在相关的２个量之间并不存在严格的函数关系．利用最小二乘法原理可以确定其拟合方程及其最优系数．然而对于较复杂的非线性问题，有时并不奏效．根据曲线拟合算法原理和遗传算法的特征，提出了利用遗传算法进行曲线拟合的６项步骤．并以电容式湿敏传感器为例介绍了算法的应用．结果表明，拟合方程与测试数据相比，各对应点的湿度误差小于０．０９％．拟合过程只需方便地利用目标函数值，从而扩大了遗传算法曲线拟合的应用范围．
There is no rigorous function relation between two correlative parameters in natural science. On the basis of the least Squar method, the fitting equation and its optimal cooficient can be defined. However, for rather complex nonlinear problems, this may not work. According to the principle of curve fitting algorithm and the specific property of genetic algorithm (GA), six steps of curve fitting witb GA are presented. A case of capcitive humidity sensor is the application of GA. The results show that the humidity error between corresponding points are less than 0. 09% for the fitting equation compared with testing data. The fitting process only uses values of objective function convenietly so that applied area of GA curve fitting is greatly.
出处
《安徽机电学院学报》
2000年第3期1-5,共5页
Journal of Anhui Institute of Mechanical and Electrical Engineering
基金
安徽省教育厅自然科学基金资助项目!(98JL035)
关键词
遗传算法
曲线拟合
优化
传感器
genetic algorithm
curve fitting
optimization
