The application of traditional feature extraction methods in subway mechanical fault diagnosis has attracted the attention of researchers. Based on the 5G intelligent sensor network signal processing theory, this paper constructs a fault diagnosis model for subway machinery and equipment and analyzes the effectiveness of the information redundancy calculation method proposed in this paper by using the on-site vibration data of subway units. The model is obtained by analyzing the entropy characteristics of the vector vibration signal in each bearing section of the subway machinery and equipment. The entropy value quantitatively reflects the vibration complexity of the rotor in the section from different angles and solves the problem of data quantitative analysis. In the simulation process, based on the 5G intelligent sensor network signal and the fuzzy mean clustering information fusion method, in the process of identifying the fault state of the subway unit, a relatively optimistic identification result was obtained. The experimental results show that, no matter whether the Gaussian kernel or the polynomial kernel is selected, the number of kernel principal components whose cumulative contribution rate is greater than 0.85 decreases with the increase of the kernel parameter, and the fault identification support rate is 85%, that is, the diagnosis results of 3 sets of sample data are nonrolling element faults, which significantly improves the performance of the bearing.