【摘要】 在每一步预测过程中都选取使预测误差最小的预测系数及其个数 ,而不是采取用固定的系数个数来求预测系数的方法进行预测 ,这是对线性预测算法的改进。每次预测的结果在下一次预测时都作为已知值 ,再确定合适的预测系数个数及预测系数 ,这样与用固定的预测系数进行预测相比可以得到更好的结果。预测值与实测值的偏差 ,不同的地区可能显示出不同的规律性。找出此规律性后 ,对预测值进行校正 ,校正后的值与实测值基本一致更多还原

【Abstract】 Linear prediction algorithm can be improved by selecting those prediction coefficients and their numbers at each prediction step which minimize the prediction errors, rather than use constant number of coefficients. Better prediction was reached by using previous prediction results in the determination for the number of coefficients and suitable prediction coefficients in subsequent prediction steps. Although different area may show different error law, such error law can be used to correct the prediction results. The corrected results show little difference with the observed values.更多还原