Keywords

clustering, rules, states, trajectories, wastewater

Start Date

1-7-2008 12:00 AM

Description

In this work the Trajectories’ Mining between subprocesses is presented in regards to the dynamics of a Wastewater Treatment Plant (WWTP). This kind of Environmental Systems involves a high complexity inherent to its own characteristics. Intelligent Environmental Decision Support Systems (IEDSS) can improve the decision making process, assisting decision makers in the evaluation of alternatives and improving management and control of Environmental Systems in general and for this particular application of WWTP. Our line of work is based on the development of methodologies of Artificial Intelligence and Statistics to solve problems of Knowledge Discovery of Data as is shown by Fayyad [1996]. In this work, knowledge extraction is approached with a methodology named Clustering Based on Rules by States (ClBRxE) formally presented by Gibert and Rodr´ıguez [2007]. Basically, consists of analyzing a process that can be divided in S = {e1 : : : eE} states or subprocesses. After dividing variables based on the subprocesses to which they refer, knowledge discovered is integrated from each subprocess into a unique model of global operation of the phenomenon. Once the data of the WWTP has been analyzed, it was empirically verified that probability of a transition between two consecutive states depends on not only on the previous state but on the whole sequence of previous states.

Share

COinS
 
Jul 1st, 12:00 AM

Trajectories’ Mining Between Subprocesses in a Wastewater Treatment Plant

In this work the Trajectories’ Mining between subprocesses is presented in regards to the dynamics of a Wastewater Treatment Plant (WWTP). This kind of Environmental Systems involves a high complexity inherent to its own characteristics. Intelligent Environmental Decision Support Systems (IEDSS) can improve the decision making process, assisting decision makers in the evaluation of alternatives and improving management and control of Environmental Systems in general and for this particular application of WWTP. Our line of work is based on the development of methodologies of Artificial Intelligence and Statistics to solve problems of Knowledge Discovery of Data as is shown by Fayyad [1996]. In this work, knowledge extraction is approached with a methodology named Clustering Based on Rules by States (ClBRxE) formally presented by Gibert and Rodr´ıguez [2007]. Basically, consists of analyzing a process that can be divided in S = {e1 : : : eE} states or subprocesses. After dividing variables based on the subprocesses to which they refer, knowledge discovered is integrated from each subprocess into a unique model of global operation of the phenomenon. Once the data of the WWTP has been analyzed, it was empirically verified that probability of a transition between two consecutive states depends on not only on the previous state but on the whole sequence of previous states.