<p>With the development of information technology, fault diagnosis models based on data-driven technology are playing an increasingly important role in the chemical industry. However, the lack of interpretability in the decision-making process has severely restricted the application of data-driven fault diagnosis models. To address these issues, this paper proposes an interpretable Takagi-Sugeno-Kang(TSK) fuzzy system based on information granules. Here, the antecedent of the TSK is composed of information granules that are used to describe the distribution characteristics of the data, while the consequent is a nonlinear model that aims to explore the hidden relationships between samples and each fault category. The proposed TSK system, which leverages the descriptive capability of information granules and the interpretability of fuzzy rules, is well-suited to addressing the deficiencies of data-driven fault diagnosis models. Furthermore, the gradient descent algorithm is utilized to optimize the system parameters, thereby facilitating the development of a high-performance and interpretable fuzzy fault diagnosis system through extensive data. The effectiveness of this fuzzy system is verified by the Tennessee Eastman(TE) process dataset.</p>