Collaborative Recommender Systems for Building Automation: A Hybrid ANN-FOA Approach

The rapid evolution of energy and sustainability regulations pertaining to the constructed environment is a direct result of the pressing worldwide environmental crises. A renewed focus on resource efficiency is being fostered within the construction industry by innovative ideas like “Nearly Zero Energy Building” (nZEB). That is why it is necessary to evaluate the building's performance thoroughly during its entire lifecycle. According to the data, there is usually a significant discrepancy between the predicted performance in the design stage and the real performance in operation. The reason behind this is that errors might occur at any point in a building's lifespan. Feature extraction, preprocessing, and training models all rely on correct sequencing. The purpose of preprocessing is to improve the reliability of a dataset by removing inaccurate or missing data, such as outliers. In order to extract features, GLCM and the Wavelet Scattering Transform are being utilised. For training hybrid ANN-FOA models, exact control over the attributes is essential. This approach appears to be rather novel when compared to the cutting-edge FOA and ANN algorithms. With a precision of 97.35 percent, the results demonstrated a significant improvement.