nder the context of rapid development in intelligent water transport,the safety status monitoring of the entire life-cycle of high-pile wharves has become a core issue in ensuring the reliability of port infrastructure.However,abnormal monitoring data caused by complex environmental conditions severely constrain accurate assessment and prediction of the status of high-pile wharves.To address the frequent occurrence of high-frequency noise and transient distortion in monitoring data from high-pile wharves,as well as the incompatibility of conventional denoising methods with non-stationary signal characteristics,a time-frequency joint denoising method integrating mean filtering and wavelet decomposition is proposed.A multi-indicator evaluation model prioritizing correlation coefficient and signal-to-noise ratio is established.Through comparative method analysis and feature parameter optimization,the optimal parameter combination is selected and validated from two dimensions:data quality improvement and prediction accuracy enhancement.Research shows that the time-frequency joint denoising method based on mean filtering wavelet decomposition effectively balances signal detail preservation and trend smoothing requirements while suppressing random noise and improving signal-to-noise ratio.The correlation between the denoised data and the original signal is significantly better than that of a single filtering method.The research results provide a solution that balances efficiency and accuracy for the processing and prediction of monitoring data for high-pile wharves.