Ternary Search Trie based Algorithms for Recognizing the Names of Power Devices

With the development of the smart grid technology and the advent of the big data, it is highly desired to develop efficient computational methods for mining the values behind of huge volumes of data on the grid. However non-unified naming of the electronic devices from the de-centralized and heterogeneous power systems makes the procedure a hard problem. In the present work, we proposed a Ternary Search Trie based algorithm for recognizing the devices from the huge volumes of transaction data in the power grid. On the base of well-defined professional vocabulary of electronic devices, the algorithm scans and segments records having names of devices using the Ternary Search Trie method. An integrative scoring schema combing semantic and word order similarity is designed to evaluate the segments against the target device names. Experimental results indicate that the proposed method achieves only 3.13% error in parsing stage, and higher recognition rate. Moreover, the algorithm lists top-n matches for professional judgments to make the work more useful and rigorous.