The extremely high business volume of the financial industry brings unaffordable operating pressure to the back-end data system of financial companies. Recently, data-driven deep learning algorithms have achieved breakthroughs in analyzing and predicting system anomalies. However, in the case of high-dimensional data, deep learning faces the problems of long training time, lack of explainability and transferability. In this paper, we propose a model based on fuzzy integral for observing and modeling the state of the system. Firstly, the fuzzy integral algorithm has lower complexity, which is more suitable for the time-sensitive financial industry. Then, based on the fuzzy integral, the vector composed of the fuzzy measures of all the features is used to represent the state of the system. It is proved that the system constructed by this modeling method has the Markov property. Moreover, compared with deep learning, fuzzy integral-based methods are not only more computationally efficient but also explainable and transferable. Experimentally, we use the actual data of securities companies and have better results in the systematic anomaly analysis.