Artificial Neural Network Levenberg-Marquardt Based Algorithm for Compressive Strength Estimation of Concrete Mixed with Magnetic Salty Water

Water quality and content significantly influence the mechanical properties of concrete. In light of the global water shortage, the utilization of seawater for concrete production has garnered considerable interest within the industry. Furthermore, sufficient compressive strength must be ensured for such an alternative. Magnetized seawater can be effectively employed for plain concrete production, thereby representing a sustainable construction material. In this investigation of Caspian Seawater, the magnetic parameters demonstrate a substantial impact on the concrete mixture's compressive strength. To accomplish this, the internal angles of hydrogen and oxygen atoms are initially altered during the magnetic treatment of saline water. In total, 364 concrete specimens were prepared for testing, with the critical water-to-cement ratio ranging from 0.45 to 0.55. Simultaneously, the magnetic field intensity (MFI) and water circulation time varied between 0.2-1.2 Tesla and 5-65 minutes, respectively. An Artificial Neural Network combined with the Levenberg–Marquardt algorithm (ANN-LM) was employed to develop a novel hybrid model for evaluating the compressive strength samples. The innovative ANN-LM hybrid model was subsequently utilized in a sensitivity analysis to determine the fundamental parameters' influence on the compressive strength of concrete specimens.