Fault detection using cross regulation model(CRM) was an important method for swarm robots fault detection based on artificial immune model. Aiming at the limited generality of CRM in fault detection of swarm robots under various swarm behaviors, cross regulation model enhanced with weighted hamming distance (WHD-CRM) was proposed by studying the influence levels of behavioral feature under different swarm behaviors. Compared with CRM, different weights to the robot behavior characteristics of different swarm behaviors were assigned by WHD-CRM, which obtained more accurate intermediate result(affinity), and improved the detection rate of faulty robots under different swarm behaviors. Experimental results showed that compared with method using CRM, the fault detection rate of WHD-CRM based method was improved by 13%.