The double negation structure is a special structure of "expressing positive meaning through two negations", in which the two negations have an important impact on the semantic analysis and emotional classification in natural language processing. Taking "¬ ¬ P==>P " as the standard, this paper makes an ergodic study on the "negation word + negation word" structures in modern Chinese; According to the formal features and semantic expressions, the double negation structure is divided into three categories, 25 sub-categories and 132 common structures or constructions. Then, based on the theories of factuality, negation focus, descriptive truth-functional negation and non-truth-functional negation, this paper investigates the double negation structure, summarizes the three conditions for the establishment of the double negation structure, and designs the program of automatically recognizing the double negation structure. The accuracy rate of the recognition of the double negation structure is 98.80%, the recall rate is 98.90%, and the F1 value is 98.85%. Meanwhile, the program also obtains 8640 sentences which 99% contains double negation structure from 96281 sentences, which provides corpus for the subsequent statistical based deep learning model.