This study addresses the challenges of limited fault samples, noise interference, and low accuracy in existing fault diagnosis methods for three-phase inverters under real acquisition conditions. To increase the number of samples, Wavelet Packet Decomposition (WPD) denoising and a Conditional Variational Auto-Encoder (CVAE) are used for sample enhancement based on the existing faulty samples. The resulting dataset is then normalized, pre-processed, and used to train an improved deep residual network (SE-ResNet18) fault diagnosis model with a channel attention mechanism. Results show that the augmented fault samples improve the diagnosis accuracy compared with the original samples. Furthermore, the SE-ResNet18 model achieves higher fault diagnosis accuracy with fewer iterations and faster convergence, indicating its effectiveness in accurately diagnosing inverter open-circuit faults across various sample situations.
Fiber optic gyro belongs to highly reliable and long-life components, which cannot be realized by traditional reliability assessment methods due to the difficulty of obtaining failure data; the Wiener process model can better model the degradation process of the device, thus realizing the reliability assessment based on the performance degradation. However, the performance degradation of fiber optic gyro exhibits nonlinear characteristics, and there is significant variability in degradation patterns among individual units within the same batch. Traditional Wiener process modeling fails to account for these two critical features. In this paper, a reliability assessment method based on the nonlinear random effect Wiener process is proposed. The nonlinear relationships are first transformed into linear forms through time-scale transformation, while the drift coefficients of the Wiener process are randomized to construct a more comprehensive stochastic degradation model. Subsequently, the Gibbs sampling method is introduced to achieve precise parameter estimation and model resolution. The proposed methodology is then applied to zero-bias performance degradation data from fiber optic gyros for reliability evaluation, generating corresponding reliability curves. The experiments show that the Akaike Information Criterion (AIC) value of the model in this paper is significantly reduced by 28.7% compared with the traditional method, indicating that the model achieves a better balance between complexity and goodness-of-fit. Therefore, the developed methodology provides a more accurate representation of the nonlinear degradation characteristics in fiber optic gyro, thereby significantly enhancing the credibility of the assessment outcomes.
11 August 2025 in Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science
The reliability of position control systems is critical for the operation of complex equipment, but their intricate operating environments and diverse fault modes pose significant challenges for fault diagnosis. This paper proposes a fault diagnosis method based on a Multimodal Wavelet Fuzzy Neural Network (MWFNN). First, by analyzing the fault mechanisms of position control systems, time-domain features are extracted from key system signals, and the 10 most relevant features are selected through Pearson correlation analysis as inputs to the model. Subsequently, an MWFNN model is constructed by integrating the characteristics of fuzzy neural networks and wavelet neural networks. This model utilizes fuzzy logic to handle uncertainties in input data while introducing a hybrid wavelet neural network to extract multi-scale features. Experiments are conducted using simulated data for validation, and the results demonstrate that the proposed MWFNN model achieves an average diagnostic accuracy of 96.72% in 10-fold cross-validation, significantly outperforming traditional machine learning and neural network models. This method effectively enhances the fault diagnosis accuracy of position control systems, exhibits high robustness and interpretability, and provides a novel approach for intelligent fault diagnosis in complex position control systems.