Pro-active Risk Management Model of gas transmission network Using data mining and Markov decision process

The aim of the research is predicting critical risks of the gas transmision network and choosing optimal corrective action in optimal time and cost. Risks were predicted with data-mining algorithms based on the CRISP methodology. K-Means, Kohnen, Two Step algorithm and Neural Network Algorithms, C.5 Tree, Nearest Neighbor and Support Vector have been used for clsutering and classification. Markov decision process is also used to select the optimal control action.Decision making problem is based on back down induction in stochastic dynamic programming in the limited time of modeling and simulation and sensitivity analysis and model validation. Based on results, in 97.56% of the agreed data, learning was created and the accuracy and validity of the data mining model was estimated at 86.92%. Also, 13 risks have been identified as critical, and the simulation results show a 92% improvement rate in the cost and 77% in the control action implementation time.