The analysis methods for mechanical behavior and safety in engineering still fail to meet real demands. Simply applying available artificial intelligence (AI) methods cannot fundamentally address the strict requirements on the stability and reliability of output in engineering. To tackle this issue, by simulating the thinking and decision-making process of human experts, the ‘mechanism’, represented by mechanical analysis methods, and the ‘data’, obtained after multi-source information assimilation, are integrated in real time. Centering around the mechanical models of engineering, three main methods for AI of Engineering are established, namely the multi-source data assimilation and data quality evaluation method, the mechanism-data coupling-driven AI method, and the cross-engineering synergistic analysis method. These methods are progressively implemented into the framework of AI of Engineering, forming a new generation engineering intelligent agent, and achieving a qualitative change from ‘one-way AI for engineering’ to ‘integrated AI of engineering’. AI systems developed therefrom are applied to landslide dams, slopes, and wind turbine generators to predict the performance. The applications indicate that AI of engineering is not constrained by the limited quantity, unstable quality and weak correlation of multi-source data in practice. It also addresses the limitations of mechanical methods under complex conditions and the difficulties in accurately obtaining computation parameters. AI of engineering integrates multiple functions, such as deformation source tracing, mechanical behavior prediction, risk early-warning and risk regulation, providing solid supports for engineering projects.

engineering  /  artificial intelligence  /  mechanism-data coupling-driven  /  data assimilation  /  data quality evaluation  /  numerical simulation