Abstract: This paper introduces a visualization-based elevator fault diagnosis method. Continuous wavelet transform and ridge extraction algorithms are applied to elevator fault diagnosis, analyzing signals during elevator operation to help engineers identify the possible time and components causing the fault, thus resolving the elevator malfunction. Keywords: Elevator fault diagnosis; Continuous wavelet transform; Ridge extraction; LabVIEW Introduction: Elevators often experience anomalies during operation, leading to non-stationary signals such as impacts. Engineers need to analyze the acceleration signals during elevator operation to identify the possible time and components causing the fault, ensuring smooth elevator operation. How to provide these possible fault time points and component information for engineers' reference is the problem this paper aims to solve. 1. Elevator Faults and Vibration Signals During elevator operation, external vibrations and noise reflect the internal working state of the elevator. Due to wear, defects, cracks, loosening, and changes in the gaps and positions of mating or contact surfaces of certain components, local impacts and sliding friction phenomena occur during operation. These phenomena are often included in their vibration signals. Increased elevator vibration and noise are always caused by a fault, as there is no such thing as increased vibration without a fault. By identifying the characteristics and changes in the vibration of each elevator component, vibration fault diagnosis can be performed. Therefore, detecting these impact components from the elevator vibration signal becomes a powerful method for identifying elevator component faults and is an important aspect of mechanical fault diagnosis. To detect the excitation frequency from the vibration signal acquired on-site to determine the vibration source, we use frequency domain analysis to perform power spectrum analysis on the collected signal. In our proposed method, we first perform continuous wavelet transform on the collected acceleration signals in various directions of the elevator, placing the signal in a two-dimensional time-frequency space. Then, we apply a ridge extraction algorithm from digital image processing to find the local maxima curves of the wavelet coefficient modulus, providing this information as potential fault points for engineers to further investigate. 2. Modulus Maximal and Isolated Singularities in Wavelet Transform Transient signals or points of rapid signal change often contain important fault information, which is characteristic information of the signal. Discontinuous abrupt changes are called isolated singularities, and wavelet transform has excellent detection capabilities for the location and singularity of these signal abrupt changes. Continuous wavelet transform possesses scale continuity and time-shift invariance, exhibiting not only good localization properties and approximation, but also fully demonstrating the superiority of wavelet analysis. It is effective for analyzing various transient and non-stationary signals. For details, please click: Visualization-Based Elevator Fault Diagnosis Method