The problem of distance protection misoperation has been a research priority recently because it was a major contributing factor in blackouts around the world such the 2003 US-Canada blackout and the 2015 Turkish blackout. In this paper, we propose a method that uses local relay data to enhance the security of zone 3 distance protection for transmission lines, even during wide area cascading events. Three k-nearest neighbor classifiers are trained using the details coefficients of discrete wavelet transform of the aerial mode currents without using any voltage signals altogether. The first classifier determines whether a fault has occurred. The second and third classifiers collectively determine whether this fault is within the zone 3 reach of the relay. If the second and third classifiers determine that the fault is outside the zone 3 reach of the relay, then a blocking signal will be sent to the relay to prevent operation even if the impedance falls within the operating characteristic of the relay. Alternatively, if the second and third classifiers determine that the fault is within zone 3 reach of the relay, a permissive trip signal will be sent to the relay only if the fault has not been cleared.