Groutability Estimation of Grouting Processes with Microfine Cements Using an Evolutionary Instance-Based Learning Approach

In the construction industry, estimating groutability is a crucial task in the planning phase of a grouting project. Hence, establishing an effective groutability prediction model that is simple to implement and can deliver quick responses with high accuracy is a practical need of construction engineers. In this research, a novel instance-based learning approach—Evolutionary Fuzzy k-Nearest Neighbor Inference Model (EFKNIM)—is proposed to forecast the result of grouting activities that employ microfine cement grouts. In the proposed model, the fuzzy k-nearest neighbor algorithm is used to classify grouting activities into two classes: success and failure. Meanwhile, the differential evolution optimization approach is deployed to select the most appropriate tuning parameters of the fuzzy k-nearest neighbor algorithm, namely the neighboring size (k) and the fuzzy strength (m). This integrated framework allows the EFKNIM to operate autonomously without human prior knowledge or tedious processes for parameter setting. An experiment using 240 in situ data samples demonstrates that the newly established groutability prediction model can outperform other benchmark approaches including the k-nearest neighbor, fuzzy k-nearest neighbor, logistic regression and artificial neural network algorithms.