Abstract: To address the high manual verification workload in power-warehouse management and the difficulty of conventional weight rounding in balancing quantity recognition and abnormal rejection, an intelligent weighing method based on weight feature matching is proposed. A weighing model including drift, noise, and dynamic disturbance is established. Stable-platform extraction, standardized distance, and standardized residual are used for material category, quantity, and anomaly recognition. MATLAB simulations show that the category-recognition accuracy for ten material classes is 95.11%, while the quantity, abnormal-placement, and overall recognition accuracies are 99.732%, 94.708%, and 97.220%, respectively. The method can support weighing-based metering in power warehouses.