舰船装备维修费预测的数据预处理技术研究

Data Pre-Processing Techniques for Maintenance Cost Prediction of Ship Equipment

在预测舰船装备维修费时,收集足够多的、准确的费用数据是关键。根据舰船装备维修费的影响因素,将舰船装备维修费预测的相关原始数据分为基础数据、使用数据、维修数据和宏观经济数据,确定了相应的数据收集渠道。分析了收集到的原始数据可能存在噪声、缺失或异常数据、冲击扰动数据、不一致数据的问题,针对性地提供了相应的数据预处理技术,经过处理后的数据可以为舰船装备维修费预测提供良好基础,进一步提高预测的有效性。

To collect data precisely and adequately is key concern to the maintenance cost prediction of ship equipment.This paper incorporates pre-processing techniques to deal with the cost prediction-related data.As a wide range of data will affect the prediction,data are to be categorized into different groups,such as basic data,operational data,maintenance data and macroeconomic data.However,by what means data acquisition can be achieved efficiently should be confirmed.The original data from different sources will likely to be of noise,default or abnormity and discrepancy.Various pre-processes are needed to handle these data respectively.The pre-processed data may provide a good basis for the maintenance cost prediction and improving the efficiency.