Systems and methods for detecting and identifying arcing based on numerical analysis

Method and system allowing more accurate detection and identification of unwanted arcing include novel processing of signal voltage representing recovered power-line current. In one implementation, arc-faults are detected based on numerical analysis where individual cycles of line voltage and current are observed and data collected during each cycle is processed to estimate likelihood of presence of arc-event within each individual cycle based on pre-defined number of arc-events occurring within pre-defined number of contiguous cycles. In another implementation, fast transient current spikes detection can be done by: computing difference values between consecutive line-current samples collected over a cycle, average of differences, and peak-to-peak value of line-current; comparing each difference value to average of difference; comparing each difference value to peak-to-peak value; and, based on calculation of composite of two comparisons, using thresholds to determine if arcing is present within processed cycle.