一种新的机电设备状态趋势智能混合预测模型  

NEW INTELLIGENT HYBRID PREDICTION MODEL FOR CONDITION TREND OF ELECTROMECHANICAL EQUIPMENT

作　　者：胡桥[1] 何正嘉[2] 訾艳阳[1] 雷亚国[1] 刘京科[1] 

机构地区：[1]西安交通大学机械工程学院,西安710049 [2]机械制造系统工程国家重点实验室,西安710049

出　　处：《机械强度》2005年第4期425-431,共7页Journal of Mechanical Strength

基　　金：国家自然科学基金重点资助项目(50335030;50175087;50305012);国家重点基础研究发展计划(973计划)(2005CB724106);高校博士点基金资助项目(20040698026);西安交通大学科学研究基金资助项目~~

摘　　要：针对机电设备运行状态受多因素影响,变化趋势复杂,难以用单一预测方法进行有效预测的问题,提出一种新的基于改进灰色系统—支持向量机—神经模糊系统的智能混合预测模型。该模型首先利用改进灰色系统弱化数据序列波动性、支持向量机处理小样本和模糊神经系统处理非线性模糊信息的优点,分别进行趋势预测,然后通过改进遗传算法对这三者的预测结果进行自适应加权组合。将该模型应用于信号随机波动性较强、趋势变化复杂的标准算例和某机组振动趋势的预测中,研究结果表明,该模型的预测性能均优于上述三种单一预测方法。Due to the fluctuation and complexity of electromechanical equipment operation condition affected by various factors, it is difficult to use a single prediction method to accurately describe its moving trend. So a new hybrid prediction model based on improved grey system, support vector machine （SVM） and neuro-fuzzy system is proposed. In this model, the fluctuation of the data sequence is weakened by the improved grey system, the SVM can deal with small samples and neuro-fuzzy system is capable of processing non-linear fuzzy information. The hybrid prediction model combines these advantages and its prediction result is an adaptive combination of these single method＇ s via improved genetic algorithms. This model was applied to the trend prediction of a fluctuant and complicated benchmark data and a vibration trend signal from machine sets. Testing results show that the prediction performance of this model outperforms any one of the three prediction methods.

关 键 词：改进灰色系统 支持向量机 神经模糊系统 智能混合预测 

分 类 号：TH17[机械工程—机械制造及自动化] TP18[自动化与计算机技术—控制理论与控制工程]