This paper presents a novel method for diagnosing mechanical faults in circuit breakers, leveraging multi-source mechanical signal data fusion. By integrating the operating mechanism’s current signal, contact stroke signal, and mechanical vibration signal, we enhance diagnostic accuracy and range. Initially, an experimental platform is constructed to simulate various mechanical failures in circuit breakers, measuring single-source signals under each fault condition. The fuzzy C-means clustering algorithm is employed to select feature parameters, and multi-source features are filtered and reorganized using feature box plots to derive the optimal diagnostic eigenvector. A fault diagnosis model is then established based on the eigenvector composition, utilizing a GA-BP genetic neural network and a grid-optimized support vector machine (SVC) algorithm to accurately identify mechanical faults. Finally, a comprehensive fault diagnosis software system is developed using C#, HTML, and JavaScript. The system achieves a fault diagnosis accuracy of 98.46%, validating the effectiveness of the proposed diagnostic method and the performance of the classification software.