DEMAND FORECASTING SCENARIO GENERATION FOR REGIME SWITCHING PROCESSES PRODUCTION PLANNING
Publication Date: 2017-Aug-08
The IP.com Prior Art Database
The issue that is solved in this invention is the generation of short-term and mid-term scenarios of customer demand based on regime switching manufacturing processes. These prediction scenarios based on historical data and associated with probabilities can be used for operational purposes (production scheduling with commitments of equipments in advance) or for contractual purposes (in advance purchase of raw materials and energy to process for producing the final product).
Current existing solutions for forecasting customer demand are mainly based on different types of statistical techniques like time series, pattern recognition, neural networks, etc. They usually provide one scenario with confidence interval and are not specifically adapted to regime switching customer behaviors.
The invention lays out an automatic process for defining the action plan to supply regime customers from the historical raw data to the generation of scenarios and ultimately the proposal for an action plan limiting the risk.
The novelty lies in two aspects. First, the use of regime switching techniques (Markov Chains) usually used for price forecasting or economic cycles predictions applied to industrial processes. Second, based on the set of scenarios generated, the action plan is the result of the limitation to the risk exposure.
The methodology is decomposed into three steps: