USE OF ARTIFICIAL INTELLIGENCE METHODS IN RESPONDING TO EMERGENCIES IN THE WASTE MANAGEMENT SYSTEM

Ravnanje z odpadki zajema zbiranje, prevažanje, predelavo in odstranjevanje odpadkov, vključno s kontrolo tega ravnanja. Izvajalci si zbiranje in odvoz organizirajo glede na potrebe in razpoložljive vire. Zaradi velikega števila odjemalcev, posod, vozil in voznikov se vnaprej pripravljajo urniki, ki so pripravljeni tako, da se viri znotraj podjetja izkoriščajo optimalno. V fazi izvajanja pogosto pride do nepredvidenih dogodkov, ki onemogočijo normalno izvajanje odvozov po urniku. V diplomskem delu se posvečamo uporabi metod umetne inteligence, s katerimi želimo izboljšati odzivnost sistema izvajalca odvoza odpadkov na izredne dogodke. Pri tem se podrobno posvetimo analizi razpoložljivih podatkov in iskanju najboljšega modela predstavitve sistema ravnanja z odpadki v digitalni obliki primerni za avtomatske obdelave, v nadaljevanju pa iskanju primernega načina vnosa izrednih dogodkov v sistem v realnem času. V nadaljevanju z uporabo diskretnih optimizacijskih metod in avkcijskega algoritma določimo najoptimalnejšo spremembo sistema in obvestimo vse deležnike sistema o spremembah v seznamu opravil.Waste management involves collection, transport, processing and disposal of waste, including the management of those operations. Operators organize collection and transport according to needs and available resources. Due to large number of customers, waste bins, vehicles and drivers schedules are prepared in advance in such a manner that resources in the company are used optimally. In the execution phase unforeseen events often prevent execution of scheduled operations. In this thesis we focus on the application of artificial intelligence methods to improve responsiveness of system of waste collection to emergencies. Available data in analyzed in detail to find the best presentation model of waste management system in a digital format suitable for automatic processing. After that appropriate methods for input of emergency information in real-time system are determined. Using discrete optimization methods and specifically auction algorithm we determine the optimum system change and inform all stakeholders of changes in the list of tasks