With the growing demand for electricity in people’s daily production and life, the density and complexity of the distribution network is also increasing, and line faults in the distribution network will affect the normal power supply of the distribution network system, thus affecting people’s normal production and life, and causing huge economic losses. This paper collects the data of distribution network cable lines, and determines whether the distribution network is discharging hidden danger by analyzing the relationship between the discharge current and the number of discharges. Use the double-ended traveling wave method to accurately locate the cable faults and construct the distribution network cable hidden danger early warning model construction. Aiming at the problem of complex fault situations in the field operation of power equipment, the reinforcement learning method is proposed to further optimize fault location. The fastest response time, the slowest response time and the average response time of the proposed early warning method in this paper are 0.3569s, 0.7259s and 0.5328s, respectively, which are better than the comparison method. In fault localization, the five differential currents are basically stabilized near 0 at the beginning stage of wave recording, and at about 190-205ms , mutated, mutated close to 7000A, and it is judged that the B-phase and the mid-point area of the high and low valves have occurred a After the field test, the actual fault is consistent with the analysis results.
