Domain knowledge acquisition from solid waste management models using rough sets.

摘要

Solid waste management is a branch of environmental engineering which preoccupies itself with the collection, treatment and disposal of wastes within the boundaries of a municipality. Activities on this scale involve political, operational and environmental issues which combine to form a complex system. Furthermore, as each system tends to be unique, a generalized lack of domain knowledge is to be expected.; In this thesis a search space of optimized environmental engineering models is generated from a list of scenarios, and Rough Sets used to extract relevant domain knowledge from it. The scenarios are generated on the basis of unknown factors within the systems, including possible system design choices or the outcome of future events.; The scenarios are stored within a database management system which is then queried to form information tables that represent the behavior of the system under different conditions. These information tables are then processed using a Rough Set algorithm which generates deterministic and non-deterministic rules used to acquire system knowledge.; The proposed approach is applied to the Hamilton-Wentworth solid waste management case, for which a database of scenarios was generated based on a basic Linear Programming model. The information tables generated dealt with the use of different transfer stations and disposal methods, and the consequence on the system cost. The results determined that certain transfer stations have a greater ability to meet disposal method quotas and that waste volume variations do not affect the optimization of the waste management system.; The proposed method is a viable way to obtain background about very large scale systems. However, because of the overhead required from the system, it is preferable to use it for very complex problems.