Automatic Block Removal and Anticlogging in a Drainage Management System via a Hybrid Fuzzy Deep Learning Approach

ABSTRACT Essentially, drainage systems (DSs) frequently suffer from blockages, thus resulting in costly maintenance. Therefore, various studies have focused on detecting blockages arising in DSs; however, this has been difficult during pandemic situations. Therefore, this article aims to implement a fully automated system for real‐time blockage removal in DSs via a hybrid fuzzy deep learning approach. The sensor data are primarily gathered. During preprocessing, noise removal is performed via Spearman's rank correlation coefficient applied via the pairwise attribute noise detection algorithm (SRCC‐PANDA). Thereafter, the proposed two‐sided Gaussian‐fuzzy C‐means clustering (TSG‐FCMC) is established to group the data into three different classes. If the data are clustered into a blockage class, then the manhole status is determined via the rule‐based decision controller (RBDC). Afterward, to locate the blockage, the unmanned autonomous surface vehicle (UASV) is used. Additionally, the proposed APTx activated‐active‐learned‐recurrent neural network (AAL‐RNN) significantly categorizes the types of blockages. Thus, the proposed approach achieved a noticeable accuracy, precision and recall of 97.7%, 98.1% and 97.3%, respectively. Hence, the outcome showed that the proposed technique outperformed conventional methods with effective performance measures. The proposed work is highly recommended for large‐scale deployment, thus enhancing public health safety.