As a result of the rapid expansion of photovoltaic systems, raising efficiency and managing ‎maintenance became the PV systems' main factors. After that comes the cost and the time of ‎repair immediately. This research provides an artificial neural network (ANN) to classify the ‎system's type of failure. Three types of failure have been studied: line-to-line fault with a ‎small voltage difference, a line-to-line fault with a large voltage difference, and ground fault. ‎In addition to the fourth normal operation case, no failure is applied. The ANN employs five ‎input data: power, voltage, current, temperature, and solar radiation. The output is a number ‎from (0 to 3), each number denotes a specific type of failure: number '0' denotes the normal ‎operation, number '1' denotes a line to line fault with a small voltage difference, number '2' ‎denotes a ground fault, and number '3' denotes a line to line fault with a large voltage ‎difference. Samples of collected data are used to train the ANN, with MATLAB Software ‎Package, to model and simulate the system. Then, the proposed ANN is tested. Its ability to ‎detect and classify the type of failure in the system is validated at a satisfactory success rate. ‎The research's focus was on the discovery of a failure in the PV system, Not only the ‎existence of a failure but also the discovery of the type of failure that occurred; this helps in ‎speeding up the solution of the problem, speeding maintenance, and reducing the loss of ‎power.‎