Hits:

Indexed by:期刊论文

Date of Publication:2010-04-01

Journal:CHINESE JOURNAL OF MECHANICAL ENGINEERING

Included Journals:SCIE、EI、CSCD、Scopus

Volume:23

Issue:2

Page Number:200-208

ISSN No.:1000-9345

Key Words:characteristics prediction; rough set; adaptive neuro-fuzzy inference system; electro-hydraulic servo valve; artificial neural networks

Abstract:Synthesis characteristics of the electro-hydraulic servo valve are key factors to determine eligibility of the hydraulic production. Testing all synthesis characteristics of the electro-hydraulic servo valve after assembling leads to high repair rate and reject rate, so accurate prediction for the synthesis characteristics in the industrial production is particular important in decreasing the repair rate and the reject rate of the product. However, the research in forecasting synthesis characteristics of the electro-hydraulic servo valve is rare. In this work, a hybrid prediction method was proposed based on rough set(RS) and adaptive neuro-fuzzy inference system(ANFIS) in order to predict synthesis characteristics of electro-hydraulic servo valve. Since the geometric factors affecting the synthesis characteristics of the electro-hydraulic servo valve are from workers' experience, the inputs of the prediction method are uncertain. RS-based attributes reduction was used as the preprocessor, and then the exact geometric factors affecting the synthesis characteristics of the electro-hydraulic servo valve were obtained. On the basis of the exact geometric factors, ANFIS was used to build the final prediction model. A typical electro-hydraulic servo valve production was used to demonstrate the proposed prediction method. The prediction results showed that the proposed prediction method was more applicable than the artificial neural networks( ANN) in predicting the synthesis characteristics of electro-hydraulic servo valve, and the proposed prediction method was a powerful tool to predict synthesis characteristics of the electro-hydraulic servo valve. Moreover, with the use of the advantages of RS and ANFIS, the highly effective forecasting framework in this study can also be applied to other problems involving synthesis characteristics forecasting.