Abstract: The correctness of the protection setting is of great importance to make full use of the relay protection system, but current setting comparison still uses manual methods, which requires heavy workload, and consumes long time, however the correctness of the results cannot be guaranteed. For this reason, the naming characteristics of the setting value name were sorted out, and a neural network-based intelligent comparison method for the setting value names of relay protection was proposed. Firstly, the text preprocessing was performed. Secondly, the preprocessed setting value text was vectorized. Finally, the bi-directional long short-term memory (Bi-LSTM) neural network was utilized to calculate the similarity of semantic feature vectors of the setting value name. Results of computing example show that the intelligent comparison method of setting value name based on neural network can effectively complete the matching of setting value list and running fixed value name, and comparing with fuzzy matching the neural network possesses higher accuracy and faster speed in handling setting value name matching problem.