As the need for clean energy continues to increase, the utilization of renewable energy is becoming increasingly widespread. Solar power plants (PLTS) are among the most widely used in Indonesia. The reliability of grid-connected PLTS is also a concern in order to maintain the quality of distributed energy. One phenomenon that occurs is the islanding condition, in which the PLTS continues to supply electrical energy to the load. Due to the limitations and operating patterns of the inverter, power quality decreases during islanding conditions, resulting in increased harmonics, fluctuating frequency, and over/under voltage. Islanding condition detection is especially necessary in solar power systems with grid-follow inverters. This study presents a method for detecting islanding conditions that utilizes wavelet transform as part of a multi-layer perceptron (MLP)-based neural network process. Simulation results show that the detection process is 98.11% accurate with a detection time of 0.006254 seconds per sample.