In　this　paper,　we　propose　a　centralized　adaptive　inventory　control　model　for　a　multi-level　multi-cycle　supply　chain　consisting　of　one　supplier　and　one　retailer　with　non-stationary　random　demand.　In　our　approach,　the　fuzzy　exponential　smoothing　method　adopted　to　forecast　the　future　demand,　and　the　EOQ　(Economic　Order　Quantity)　model　determines　the　ordered　quantity.　Besides,　a　reinforcement　learning　algorithm　is　developed　to　evaluate　the　effects　of　safety　factor.　Our　objective　is　to　satisfy　a　given　target　service　level　predefined　for　the　retailer.　Two　types　of　demand　process　patterns,　known　and　unknown　demand　distribution,　are　considered.　Moreover,　the　bullwhip　effect　generated　while　processing　of　demand　information　is　provided.　The　results　show　that　the　proposed　control　method　can　improve　the　service　level　and　reduce　the　bullwhip　effect　to　some　extent.