Decision Support to Identification of Road Infrastructure Segments With Poor Conditions

Abstract

Planning of maintenance activities (an integral approach) on ever growing system such as urban road infrastructure requires decision makers to take into account large number of different data and usually conflicted criteria, derived from several aspects of analysed problem. This systematic approach to planning is based on the main three steps: identification of alternatives, their validation and selection for inclusion into a maintenance plan. Most of authors are dealing with the last two steps however, focus of this research is on the first step. This step is a critical phase of planning because it is important to identify road infrastructure segments that need to be improved. Meaning, those with poor condition. The condition of a segment must be expressed as simple as possible (by unique Condition Assessment Value) and by usage of few but most relevant criteria (such as level of service, safety, time period passed form the last renovation and stability). Another important issue is provision of relevant, relatively quick and consistent expert assessment of segment condition according to above mentioned criteria and for large number of segment. To insure such assessment with high quality of segments identification this paper propose an decision support concept/expert system based on combination of trained and tested Artificial Neural Networks (ANN) and Simple Additive Weighting (SAW) method. The concept is validated on urban road infrastructure of city centre of Split, Croatia and it proved to be useful for determining a set of road infrastructure segments where maintenance activities should be undertaken.

Full document