The invention relates to an industry business adaptive detection method based on enterprise electric power big data. The method comprises the following steps: step 1, cleaning and sorting enterprise power consumption information; step 2, constructing a complex network model and selecting related industries; step 3, separating different power utilization behavior states by using a hidden Markov model; and step 4, constructing an industry business index and carrying out self-adaptive detection. The method has an advanced theory. A complex network model and a hidden Markov model are organically combined. The complex network model analyzes and compares all global variables. Effectiveness of selecting relevant variables in the hidden Markov model is improved. Dynamic incidence relations and transfer characteristics between different industries are described. The result of the complex network model is input into the hidden Markov model. The influence of external weather and holiday factors is considered and the accuracy of predicting the industry production business condition is improved.