(465e) Multi-Objective Control-Relevant Demand Modeling for Supply Chain Management

The development of control-oriented decision policies for inventory management in supply chains has received considerable interest in recent years, and demand modeling to supply forecasts for these policies is an important component of an effective solution to this problem. Previous work in our laboratory has developed tactical decision policies based on Model Predictive Control [1,2] which have been demonstrated to be highly effective in supply chains problems associated with semiconductor manufacturing and similar forms of high volume, long througput time manufacturing processes. In these policies, however, the demand forecast signal is treated as exogeneous to the algorithm. Drawing from the problem of control-relevant identification [3], we present an approach for demand modeling based on data that relies on a control-relevant prefilter that tailors the emphasis of the fit to the intended purpose of the model, which is to provide forecast signals for these MPC-based decision policies. Analysis developed as part of this work shows that from a spectral/frequency-domain standpoint, forecast signals for supply chain management need be accurate only over an intermediate frequency range, as dictated by controller performance requirements. The implications of this result are signficant, as being able to integrate demand modeling with the subsequent inventory control problem offers the opportunity to obtain reduced-order models that exhibit superior performance, with potentially lower user effort relative to traditional "open-loop" methods. A systematic approach for generating these prefilters is presented, and the benefits resulting from their use are demonstrated on two representative production/inventory system case studies. The formulation is multi-objective in nature in that it that allows the user to emphasize minimizing inventory variance, minimizing starts change variance, or their combination.

[1] Braun, M.W., D.E. Rivera, M.E. Flores, W.M. Carlyle, and K.G. Kempf, "A Model Predictive Control Approach for Robust Management of Multi-product,multi-echelon demand networks," Annual Reviews in Control, Vol. 27, Issue 2, 229, 2003.

[2] Schwartz, J.D., W. Wang, and D.E. Rivera, "Simulation-based optimization of Model Predictive Control policies for inventory management in supply chains," Automatica, 42(8), 1311-1320, 2006.

[3] Rivera, D.E., J.F. Pollard, and C.E. Garcia, "Control-relevant prefiltering: a systematic design approach and case study," IEEE Transactions on Autom. Control, 37, 964, 1992.