Abstract: By employing an artificial immune clustering algorithm‚the elevator traffic flow in a certain building during one week was analyzed．The promotion and suppression mechanisms of artificial immune algorithm were exploited to compress the original elevator traffic flow data collected in5min‚and a distinct elevator traffic flow memory dataset was obtained．The memory dataset was then classified on the basis of two methods： the Minimal Spanning Tree and the Minimal Distance Method．The result of classification clearly reveals the real feature of elevator traffic．Therefore‚the approach addressed here may provide a powerful theoretical support for elevator group control． 

Key words: artificial immune system, clustering analysis, traffic flow, elevator