MODELING OF NITROGEN IR.H.DJUANDA RESERVOIR USING NEURAL NETWORK METHOD

• Main Author: Dorty Hadi, Christine
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/43730
• Institution: Institut Teknologi Bandung

Ir.H.Djuanda Reservoir or better known as Jatiluhur Reservoir is the first multipurpose reservoir that was completed in 1957 and completed in 1967 with the aim of meeting the needs of raw water, irrigation needs, hydroelectric power (PLTA), and additional functions for tourism and fisheries. Jatiluhur Reservoir receives a lot of waste load which can reduce its water quality. The waste comes from activities that take place inside the Jatiluhur reservoir and from outside the reservoir itself. Therefore, in controlling reservoir water quality modeling is needed that can provide an overview of nutrients in the Jatiluhur reservoir water body. Especially nitrogen nutrients. This modeling uses the method of artificial neural network backpropagation scheme by modeling nitrogen. The model is used to predict nitrogen elements in 4 water quality stations, namely station A (Bojong station, Sodong station, Jamaras station), Kerenceng station, Baras Barat station and station B (Cilalawi station, Karamba station, PDAM station, Karamba station). Each station is modeled with 4 scenarios. Input parameters used are floating net cages (KJA), volume, inlet discharge, solar irradiation time, NH3, NO2, NO3. Scenario 1 uses all inputs, while Scenarios 2, 3 and 4 use 4 inputs. The modeling results show satisfactory results based on the correlation coefficient and the value of root mean square error (RMSE). The model shows a correlation coefficient of 0.90 - 0.96. The model is then selected with the best scenario then trained and tested again by removing the KJA element to see the effect of the KJA input on the results of the coding. The results show the effect of KJA on the station being modeled is not the same. The influence is seen from the performance of the model. Station A has no effect on the KJA input because the input used comes from the Parung Kalong station where the station is still influenced by the Citarum river flow and the Cirata reservoir. West Baras Station, KJA input influences the nitrogen element that is modeled because the West West station is close to zoning III KJA and there is KJA that has been operating since 2013. Concentration in Baras Barat has been mixed with nutrients derived from dissolved fish food and metabolic products and fish excretion. The concentrations of NH3, NO2, NO3 Kerenceng stations have an influence on the KJA input and volume because the concentrations that enter the Kerenceng station have passed through Bojong station, Sodong station, Jamaras station where all three stations have a number of KJA operating so that the nutrients produced are mixed with elements nitrogen produced from the rest of the feed and the results of metabolism. KJA does not significantly affect NO3 and NO2 concentrations at station B (Cilalawi station, Karamba station, PDAM station, Taroko station) while NH3 shows a significant change in model performance because the Cilalawi station is an inlet of the Cilalawi river that carries quite a lot of nutrients from both domestic and waste products. industry.