Modeling and problem solving of building defects using point clouds and enhanced case-based reasoning

Many factors, including improper maintenance and material aging, may lead to the occurrence of defects during the operation of the various functions of buildings. Building defect information is normally stored in a discrete and unstructured way, and for this reason, building a case-based reasoning framework regarding building defects to enhance the level of building maintenance management has become an important field in the related research. At present, there is limited research available on the integration of geometric data models that are built by means of scanning and multi-attribute selection strategies. This study proposes an integrated information management framework for superficial defects in buildings, which is compatible with a point clouds model as a central data source. It features the attributes of defects used in multi-criteria decision analysis. A CBR (case-based reasoning) approach that considers case-based distance is used to enhance the performance of similarity calculations and case retrieval. A case-based distance model is utilized for the data processing stage and concentrates on a smaller case set that contains best alternatives. The potential benefit offered by this approach is that more efficient results can be obtained from classified cases during the retrieval phase process. A comparison of a CBR query with ungrouped sample data is performed to establish patterns to verify the effectiveness of the calculation method of determining case similarity, which is supported by the pre-processing of classified information about the building defects. The analytical results show that the proposed method performs well in solving the multi attribute classification of building defects and avoiding ambiguous answers retrieved from unrelated subsets. This approach might be capable of investigating the practical problems involved in building maintenance in the AEC domains.