时间序列预报技术在预报中占有重要的位置。本项目根据实际需要在国内外首次提出不等间隔时间序列的预报问题。研究不等间隔序列取样应遵循的规律、不等间隔序列可预报性判断的方法、不等间隔序列的神经网络建模预报方法等。必将丰富时间序列预报的理论和方法，扩大它在工程、经济、社会诸领域应用的深度和广度，创造可观的经济效益和社会效益。

In practice there exist many various nonuniform time series, but it has not been found that the problem predicting nonuniform time series has been discussed inside and abroad. First, the rule which should be followed when sampling nonuniformly is studied, and a new nonuniformly-sampling rule is presented. The ANN based time series auto-regression modeling method is re-studied, traditional ANN model's input pattern being revised greatly. The problem on generalization capability of an ANN predicting model is also studied preliminarily. A genetic algorithm based time series regression modeling idea is introduced, and its applications to modeling system error in batch machining and to modeling wear process in aero-engines are studied, showing satisfactory results. In addition to above, some teniques on noise depressing and feature extration, which are always related to specific predictions, are studied. In sum, the studies of this project get to or exceeed the planned objective completely, presenting new theories and methods of intelligentization, etc., for time series prediction