The current condition of global energy utilization is that 40% is consumed by residential, this value is higher than industrial and commercial groups. Overcoming this problem can be done through energy conservation management specifically for household customers. The initial process of energy conservation is monitoring the use of electrical energy loads that are being used. Monitoring the type of use of electrical energy loads that have low-cost features is Non-Intrusive Load Monitoring (NILM). The method that can be used to monitor electrical energy loads with NILM is a combination of Fast Fourier Transform (FFT)-Artificial Neural Network (ANN). The success rate of recognizing this type of electrical load depends on the size of the epoch during the ANN training process. Based on testing the success value of being able to achieve a value of 100% if using epoch 10000, it is different if using epoch 500 the success is only up to 30%. The results of the calculation process using the confusion matrix have an accuracy of 0.5876 or 58.76%, while the F1 value is 0.6928 or 69.28%.