Sistem Prediksi Curah Hujan Bulanan Menggunakan Jaringan Saraf Tiruan Backpropagation

Abstract

Rainfall has important role for human life. Rainfall information can be used in several fields including agriculture. As a benchmark for planting periods, water infiltration management, and irrigation. The resources for calculating rainfall are rainfall gauges, ground-based radars and remote sensing satellites. Wonosobo area’s rainfall type is monsoon, meaning that it has one wet period and one dry period. It has fluctuating varied rainfall every month and the availability of rainfall data is uncertain each year. As a mountainous area, Wonosobo’s agricultural sector is very dominant for their economic. Weather Observation, especially rainfall, is important because it can be used by related parties, especially in the agricultural sector. In addition, to provide rainfall data in areas with no observation stations. This study aims to design and implement a rainfall prediction system by developing the Waterfall Model Development Life Cycle (SDLC) Software and implementing backpropagation artificial neural networks (ANN). System development using the SDLC waterfall model was chosen because it is simple, easy to understand and implement. ANN backpropagation is applied in the prediction system because of its advantage that can be applied to a problem related to prediction. Testing on the system built for training and validation produces training accuracy of 93.92% with validation of 73.04%, indicating that the system can be used and has been running expectedly. The best ANN architecture was obtained on the test with input layer 3, hidden layer 12, and output 1 values, learning rate 0.5 momentum 0.9. From the SSE 0.1 target, the SSE is 0.302868.