最佳一致逼近算法在模拟温补晶振中的应用

目前国内外频率控制领域中最常用的近似方法是最小二乘法，但是实际需要的是使频率偏差的绝对值最小而不是频率偏差的平方和最小．而最佳一致逼近算法正是能够求解出使频率偏差的绝对值最小的极大极小解的一种方法。将最佳一致逼近算法首次应用于模拟温补晶振补偿网络参数的计算中，并针对一个具体的例子，通过比较利用最佳一致逼近算法前后的数据，可以看出理论上在利用了最佳一致逼近算法后和最小二乘法相比补偿精度提高11．74％．并进行了验证实验。由于算法本质特性，从统计规律来看，该算法会提高产品的成品率，对于生产实践具有指导意义。

The Best Approximation Algorithm Applied on ATCXO

In the worldwide frequency control field, the least square method is the common approximation method. The least absolute value of frequency deviation not the least square sum of it is eager to get in practice. The best approximations algorithm is just the method which can solve the min-max solution of the least value of frequency deviation. Because of the reason above, the best approximations algorithm is introduced into calculating the parameters of ATCXO compensation network for the first time. In a practi- cal example, comparing with the method of least squares, the compensation precision can be improved about 11.74% when using the best approximations algorithm. According to the statistical laws, the best approximations algorithm can improve the rate of product and instruct the production.