The Bayesian Inventory Problem

Abstract

Our objective is to improve the management of repair parts for newly fielded weapon systems. The initial procurement of each part is made a lead time before the system is fielded, and is based on an engineering estimate of the mean part demand rate per fielded system. There may be additional procurements before the fielding date. Once the system is fielded, demand experience accrues and is used to update the forecast of the demand rate, improving its accuracy. Inventories are managed under a periodic review, (s,S) policy: when assets fall to s, order up to S. The period is as small as one week. The issue of concern is how the expected improvement in accuracy of the demand forecast should affect the values of the inventory control parameters. Formally, we are seeking to determine optimum (s,S) parameters - they change each period - when there is Bayesian updating, periodic review and a dynamic mean with demand randomly distributed about the mean. An algorithm is programmed, sample results obtained, and conclusions drawn. Originator supplied key words include: Dynamic Programming, Bayes, Inventory Models.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
May 01, 1984
Accession Number
ADA150812

Entities

People

  • A. J. Kaplan

Tags

Communities of Interest

  • Cyber
  • Weapons Technologies

DTIC Thesaurus Topics

  • Accuracy
  • Algorithms
  • Computations
  • Computer Programming
  • Computer Programs
  • Dynamic Programming
  • Engineering
  • Equations
  • Inventory
  • Inventory Control
  • Lead Time
  • Mathematics
  • Probability
  • Random Variables
  • Statistics
  • Systems Analysis
  • Weapon Systems

Readers

  • Logistics and Supply Chain Management.
  • Statistical inference.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - DoD AI Strategy