Prediksi Respons Struktur Jembatan Beton Prategang Berdasarkan Spektrum Gempa Indonesia dengan Metode Jaringan Saraf Tiruan

The design of prestressed concrete bridge on high-risk seismic zones should consider the responses characteristic of the bridge structure. The structural responses such as displacement, velocity, and acceleration which are based on Indonesia's Seismic Hazard Map, could be predicted by using Artificial Neural Network (ANN) method. In this research, the bridge model was adopted from the real prestressed concrete bridge located in Kecamatan Tapung, Kabupaten Kampar to analyze the bridge structural response and evaluate the potential of ANN to predict the results. The bridge structure was modeled in finite element software, then the response spectrum analysis was performed based on RSNI 2833-201X. For the seismic locations, 34 capital cities and 5 other big cities in Indonesia were chosen. Then by adding 3 soil conditions, 117 data sets were produced. Based on the analysis results, the largest structural response was observed in Palu City with 0,0738 m of displacement, 1,8406 m/sec of velocity, and 24,7666 m/sec2 of acceleration. For the ANN analysis, 102 data sets were used in training phase and the remaining 15 data sets were used in testing phase. After the training and testing process, the ANN model has the potency to predict bridge structural response at 93 to 97% of prediction rate and the calculated Mean-Squared Errors (MSE) is as low as 0,00019. This indicates that the ANN model adopted in this research is capable of predicting the structural response of bridge with high accuracy.