Microseismic signals induced by mining activities often have low signal-to-noise ratios, and traditional picking methods are easily affected by noise, making accurate identification of P-wave arrivals difficult. To address this problem, this study proposes an adaptive denoising algorithm based on wavelet-threshold-enhanced Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and develops an automatic P-wave arrival picking method incorporating fractal box dimension features, along with a corresponding accuracy evaluation framework. The raw microseismic signals are decomposed using the improved CEEMDAN method, with high-frequency intrinsic mode functions (IMFs) processed by wavelet-threshold denoising and low- and mid-frequency IMFs retained for reconstruction, effectively suppressing background noise and enhancing signal clarity. Fractal box dimension is applied to characterize waveform complexity over short and long-time windows, and by introducing fractal derivatives and short-long window differences, abrupt changes in local-to-global complexity at P-wave arrivals are revealed. Energy mutation features are extracted using the short-term/long-term average (STA/LTA) energy ratio, and noise segments are standardized via Z-score processing. A multi-feature weighted fusion scoring function is constructed to achieve robust identification of P-wave arrivals. Evaluation metrics, including picking error, mean absolute error, and success rate, are used to comprehensively assess the method’s performance in terms of temporal deviation, statistical consistency, and robustness. Case studies using microseismic data from a mining site show that the proposed method can accurately identify P-wave arrivals under different signal-to-noise conditions, with automatic picking results highly consistent with manual labels, mean errors within the sampling interval (2–4 ms), and a picking success rate exceeding 95%. The method provides a reliable tool for seismic source localization and dynamic hazard prediction in mining microseismic monitoring.

Microseismic monitoring is a crucial technical approach for deep underground engineering and the early warning of dynamic disasters. However, the high-precision automatic picking of microseismic P-wave first arrivals under strong non-stationary background noise remains a critical challenge in this field. To address the limitations of conventional time-frequency denoising methods, in which hard thresholding may oversmooth the onset edges of first arrivals and consequently cause STA/LTA or global AIC algorithms to be trapped by noise-induced local extrema, this paper proposes a novel automatic P-wave picking method. This method relies on the synchrosqueezing transform (SST) and local box-counting dimension for denoising, integrating permutation entropy with the Akaike Information Criterion (AIC). During the denoising stage, a dual-discrimination mechanism combines SST-based time-frequency component selection with local box-counting dimension analysis. Initially, candidate high-energy time-frequency points are selected from the SST representation using adaptive thresholds, while the remaining components are attenuated to reduce diffuse background noise. Subsequently, the local box-counting dimension of the continuous time-frequency plane is calculated to characterize the differences in the time-frequency fractal structure of the signal. Based on this, an asymmetric fractal gain operator is constructed to suppress noise disturbances while providing directional energy compensation for the first-arrival jump edges, thereby mitigating the smoothing distortion of weak waveforms during the denoising process. In the picking stage, the permutation entropy and variance features of the local waveforms are extracted to formulate a joint feature function for locating candidate onset regions. This is combined with a sliding AIC to perform a localized fine search, overcoming the vulnerability of traditional global searches to noise extrema interference and achieving high-precision localization of P-wave first arrivals. Verification using real mine data demonstrates that the proposed method significantly enhances signal quality under complex background noise conditions while effectively preserving the initial onset characteristics of the seismic phases. Under simulated input SNRs of 20, 10, and 0 dB, the proposed method achieves SNR gains of 17.22, 15.98, and 10.54 dB, respectively, corresponding to a mean gain of 14.58. The method also yields the lowest RMSE among the compared denoising methods at input SNRs of 10 and 0 dB. The research results indicate that this method can provide reliable technical support for the accurate picking and source localization of microseismic events in highly noisy environments.