.

Time and thoughts

Home Blogs TILs Papers Projects Writings

Development Of A Knowledge Based Expert System For Landslide Prediction

Parameswaran and I discussed a lot on this project. Some of the specifics about ANN and the direction of code was suggested by me while the details on the specific attributes were provided by Parameswaran. This was for the Master’s Thesis that he had done in NIT, Calicut in May, 2011. This was implemented in Python.

Abstract

In Kerala with the onset of monsoons, landslides occur in many highland regions of the state. The resulting damage is vast and often it claims many lives. A prediction system with sufficient reliability if available would be of great assistance in minimizing the hazard caused by these landslides. There has been considerable development and research in site specific landslide prediction systems using Geographic information systems, Global positioning systems, Probabilistic approaches etc; but a generic system for predicting landslides has not yet evolved due to its volatile nature and inherent complexity. Hence developing a Knowledge Based Expert system for predicting Landslides is planned in this project work.

An expert system is a computer program that simulates the judgment and behavior of a human or an organization that has expert knowledge and experience in a particular field. Expert systems are necessitated by the limitations associated with conventional human decision-making processes which includes scarcity of human experts, inability of humans to comprehend large amount of data quickly and limited working memory. The present expert system is designed as a web based application so that it can be accessed through ubiquitous web browsers and the need for installing the package on a desktop system can be averted.

The proposed Web based Expert system consists of 4 modules such as input, analysis, output and feedback. The input module accepts data from the user and passes it to the analysis module. The analysis module then runs the heuristic model & the physical model and arrives at a conclusion regarding the stability of the area. The combined inference concerning the stability of the area is passed to the output module which is then displayed. Results like what rainfall can trigger the landslide in the specified area and during which part of the year is the landslide possibility maximum are also included. The user is also provided an option to enter the feedback based on what actually happens at the site which helps the system in its self learning.