<p>In order to address the problems of fingerprint redundancy, omission and waste of computing resources caused by the rapid increase in the number of smart grid devices, a large-scale smart grid equipment fingerprint database construction method based on game-theoretic matching was proposed. Firstly, the discrete wavelet transform was used to extract the RF distortion features of the device signal and construct a unique fingerprint. Then, the fingerprint database subset was divided by spectral clustering, and the game matching model was designed to optimize the fingerprint allocation strategy to reduce the risk of redundancy and omission. Finally, a two-level authentication mechanism is constructed to reduce the amount of global traversal computation through local priority matching. Experimental results show that the proposed method achieves a duplicate detection success rate of 93.9% and an authentication time reduction of 24.16% in simulations based on 110 expanded device fingerprints derived from a public dataset. In a simulation scenario with 1,000 fingerprints, the authentication time reduction still reaches 22.64%, demonstrating its application potential in large-scale device fingerprint database construction and efficient authentication.</p>