An innovative hybrid control chart combining wavelet decomposition and support vector machine for effective outlier detection

Introduction Statistical Process Control (SPC) charts are widely used to detect process shifts, yet classical Shewhart X̄ charts often fail in the presence of outliers and non-normal data distributions. To address these shortcomings, we propose a hybrid approach that integrates discrete wavelet decomposition with support vector machine (SVM) classification for enhanced outlier detection. Methods The proposed Dynamic Wavelet-SVM X̄ Chart (DWS-X̄) combines multi-resolution wavelet feature extraction with an SVM classifier to distinguish between in-control and out-of-control states. Its performance was assessed through extensive simulations varying subgroup size, number of subgroups, and contamination levels. Key metrics included Detection Rate (DR), False Alarm Rate (FAR), and Average Run Length (ARL). Results Simulation results consistently demonstrated the superiority of the DWS-X̄ chart compared to the traditional X̄ chart. The hybrid chart achieved higher detection rates (above 96%), lower false alarm rates (below 1-6% depending on variability), and shorter ARLs (1.6-5.1). Application to real neonatal heart rate data further confirmed its enhanced sensitivity in identifying abnormal cardiac patterns. Discussion The findings indicate that the proposed DWS-X̄ chart provides more reliable monitoring under contamination and high variability, offering earlier detection of abnormal signals with fewer false alarms. This hybrid method holds promise for both industrial quality control and healthcare process monitoring.