Gas-station Predictive Maintenance Planning with Hybrid Model of Fuzzy Neutral Network and PCA

With the emergence of predictive maintenance in 1980, radical changes took place in maintenance planning. Predictive maintenance depends on the prediction of facilities failure which are used at present. By predicting the failures correctly in future, we can decrease the cost of maintenance to a great extent. This approach involves using multiple techniques including artificial intelligence, that is, neural network, and “fuzzy sets” theory. The cost of maintenance is high for the activities of Iran’s National Gas Company, therefore, the application of predictive maintenance can be economical. Gas is distributed through underground pipes which Corrosion destroys; thus, the cost inflicted on the society’s capital is incalculable. In this study predicting the failure was based on the combined model of fuzzy neural network and principle component analysis. The results indicated that this model can decrease the inspection costs by 25%.