Short Term Gas Demand Forecasting Based on Artificial Neural Network

摘要

Teikoku Oil Co. Ltd. (TOC) and NKK Corp. established a jointrnpilot project in 1994 in order to provide pipeline application andrnevaluation of NKK's gas hydraulic simulation enginern(GASTRAN) and to co-develop a Demand Forecasting Modelrn(DFC). When the pilot project finished in March 1997, arncommercial system, called Support Operation and MonitoringrnApplication of Pipeline Simulator (SMAPS), was installed inrnTOC's operation center.rnThe DFC, which is based on an artificial neural networkrnarchitecture, has several advantages for sales forecastingrnespecially as several dozen delivery points that have differentrnsales patterns are connected to the pipeline network. The resultsrnfrom DFC can be easily used for scenarios in off-line simulationrnto predict future pipeline situations when it is attached to thernSMAPS system. It automatically assists the pipeline operator byrnreducing his workload and evaluating operation plans.