A novel approach for sustainable construction using Bi-LSTM-CNN based predictive analytics on ceramic waste powder for concrete

The enormous volumes of waste produced by the building sector, particularly by the ceramics industry, which releases large amounts of CWP-wates of ceramic, through polishing and cutting procedures, causes serious environmental issues. Traditional methods of concrete production increase pollutions and create a severe burden on waste capacity. CWP usage in concrete mixtures as a responsible option, but however, accurate prediction is a one of the major problems. Current predictive models commonly create issues regarding accuracy, particularly by capturing the difficult, interactions among the mechanical features of concrete based CWP content. These drawbacks highlight the urgent need for an advanced predictive analytics model that can able to handle both environmental issues related to CWP concrete's manufacture also accurately predict the material's qualities. By considering this as a background, this study presents an effective novel hybrid solution for sustainable construction, which combines Deep convolutional particle filtering techniques (CNN-PF) with attention based Bi-LSM based predictive analytics on CWP for concrete. The proposed model combines the benefits of CNN-PF to extract important features in input data like composition details or microstructural images. Where the particle filtering improves the model's efficacy by handling abnormalities in the data. Additionally, attention based Bi-LSTM helps to capture long term dependencies in sequential data to improve prediction accuracy. The experimental investigation is performed with 54 different concrete mixes with two existing articles, in different Molds, using CWP. The outcomes highlight that our proposed model marks its remarkable outcomes than the existing models by achieving a notable result in both model-based prediction and Mold based prediction.