PENGENALAN JERUK SIAM BERFORMALIN BERBASIS JARINGAN SYARAF TIRUAN RADIAL BASIS FUNCTION; RECOGNITION OF FORMALINED ORANGE WITH RADIAL BASIS NEURAL NETWORK

PRASOJO, SATRIO, Sri Hartati

2016 | Skripsi | FMIPA

Orange is one of the most popular fruits in Indonesia. Orange can stand up to 1-5 months after harvest time. To maintain the freshness, sometimes unscrupulous rogue sellers do preservation with formalin to reduce losses at a low cost. Formalin is dangerous illegal food preservatives, which are carcinogens and can be resulted body's organ systems failure. Curently the detection on the formalin content of fruits and food using analytical methods, such as spectrophotometer ultra violet (UV), high performance liquid chromatography (HPLC), Gas Chromatography (GC), and the detection of liquid saffron tumerik paper. Another alternative detection method that also can be used is an electronic tongue sensor system (e-tongue) which responds to content of substate by an electrical potentials. This system then developed by adding the methods of artificial intelligence for decision support with the artificial neural network radial basis function (RBFNN). RBFNN is a method that can perform artificial intelligence pattern recognition approach which reference to the human nervous system (Fu, 1994). The system built for the purpose to distinguish inputs into three classes, formalined orange, orange without formalin, and not orange. Based on the test results, systems has achieved 83,3% of acuracy to identify data that has not been trained. Learning rate value was set at 0,03 and neuron configuration of input-hidden-output set at 6-12-1 unit with achievement of MSE 0,000957.

Kata Kunci : N