中文摘要：

基于核和灰度的区间灰数预测模型，初步解决了区间灰数序列预测问题，但模型中灰度预测值的确定方法存在不足，且模型不支持误差分析。提出了合成灰数灰度的定义及其性质，据此分析了模型存在的问题，并建立灰度序列的预测模型实现灰度预测，以代替原有模型中灰度预测值的确定方法，从而改进和完善了原有区间灰数预测模型。改进模型从核和灰度两个方面同时发掘区间灰数序列的内蕴信息与发展趋势，克服了原有模型存在的不足，且支持误差分析和精度检验。算例表明了改进模型的有效性和可用性。

英文摘要：

A prediction model of interval grey number based on kernel and grey degree preliminaryly solves the problem of prediction for interval grey number, while still has some model deficiencies which need to be im- proved in error analysis and the method of determining the predicted value of grey degree of interval grey num- ber. These model deficiencies are analyzed based on the definition and characteristics of compound grey number which are proposed. Furthermore, a prediction model for grey degree is built instead of the original deficient method, thus the original prediction model of interval grey number is improved and perfected. The improved model can excavate both potential information and developing trend of interval grey number list from the aspects of both kernel and grey degree, thus the original model deficiencies are overcome, and error analysis and preci- sion test are supported. Example analysis demonstrates the applicability of the improved model.