Track association has been recognized as one of the major problems in distributed multiple-radar multiple-target tracking systems. Tracks formed on the same targets by multiple radars may differ due to misalignment or bias error. So the traditional features, such as target position and speed, cannot be reliably used for track-to-track association. The information of relative positions among targets, however, is found to contain special characters and can be extracted as target topology vector of reference dynamically. Applying fuzzy pattern recognition to analyze the similarity of vectors or patterns can set up the criterion of association and provide prior information for alignment. Simulation results show that the proposed algorithm is effective to solve the association problem under the condition of inter-radar bias and robust to the random errors.