PEMODELAN SISTEM MULTISTATE DENGAN JARINGAN PETRI PROBABILISTIK

• Main Author: Mudjihartono, Paulus
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/3093
• Institution: Institut Teknologi Bandung

<b>ABSTRACT:</b><br> <br /> This thesis develops software prototype that enables user to build, generate and manipulate model in term of decision support systems. The target model is created by using Petri net. In Petri net view, the modeled system consists of several system variables. These system variables, in particular discrete time, keeps one certain condition. The system state domain is then defined as combination of these variable conditions. Therefore, multistate is used to call this system.</p> <br /> Petri net modeling is performed by using two things, ie: value of system variable, and relation of system variables. Since the relation is not exact, Petri net relates system variables under its components in probabilistic way. <br /> Petri net components are place, transition, output function and input function. Place represents system variable by its tokens and condition. Transition, output function and input function are assembled to build relations among system variables. Execution of net is performed by moving tokens from one place to another via relation.</p> Token mechanism is a value transfers which reducing token value of place by its multiple output and adding those by its multiple input. <br /> Probabilistic aspect comes up in calculating the correlation of data. This data correlation is used to generate model under Petri net characteristics. <br /> Execution yields fact that target model maintains the trend of original data. Some generated data fluctuate too large and have opposite trend for some time, but the global trend of data is still maintained.