This project belongs to the field of chemical science and technology. Focusing on the needs of high-quality development of the petrochemical industry, this project has developed a multi-scale characteristic characterization and intelligent modeling method for the refining production process based on the essence of material transformation through the deep integration of new generation information technologies such as industrial big data and artificial intelligence with manufacturing processes., and developed a multi-time-scale resource optimization decision-making method for the maximization of the entire process value chain based on system behavior decision-making. Starting from process performance control, a multi-objective optimization and performance evaluation method that integrates knowledge and models has been developed, and a system with independent intellectual property rights integrating full-process simulation, intelligent production planning decision-making, device operation optimization and operating performance evaluation has been developed. Intelligent manufacturing software and systems have realized intelligent collaborative optimization of the entire process and multi-level in the large-scale refining industry. The innovation points are as follows: (1) Important breakthroughs have been made in refining process mechanism modeling and full-process real-time simulation technology. A multi-scale (molecular rarr; reaction network rarr; reactor model rarr; process model) intelligent modeling method for complex reaction processes based on laboratory reaction dynamics and data-driven characterization of key properties is proposed; a full-process integrated modeling method based on simplified model structure that integrates knowledge of process mechanisms and characteristics and convergence of heterogeneous models based on constraint relaxation matching of key components is proposed; a full-process simulation and real-time simulation system for refining the entire process has been developed, which can accurately describe the characteristics of crude oil and the operating parameters of each unit on the yields and properties of various middle distillates, refined oils and other products online. The average absolute error of prediction of key indicators is less than 1%. (2) Innovatively developed intelligent optimization decision-making technology for production planning that integrates the operating characteristics of the device. A large-scale production planning model construction technology based on automatic transfer of physical properties and device operating characteristics and sliding proxy model dimension reduction method is proposed; a multi-time scale resource optimization decision method for maximizing the entire process value chain is proposed, and intelligent production planning decision-making technology and systems are developed to automatically transfer material flow, energy flow and physical property data online, realizing automatic optimal allocation of multi-time scale resources and adaptive optimization of production models. Compared with the original production scheduling method, the production scheduling accuracy of major products such as gasoline, aviation fuel, and diesel has increased by 10%. (3) Innovatively developed refining unit operation optimization and performance evaluation technologies that integrate knowledge and models. Several types of high-dimensional and multi-objective optimization algorithms based on intelligent computing have been proposed, and refining unit operation optimization technologies and systems have been developed for multiple optimization goals such as safety, environmental protection, quality and efficiency and under complex constraints, forming a combined optimization model that can customize yield, quality, efficiency, and energy consumption for 11 sets of main refining units, including atmospheric and vacuum reforming, catalytic cracking, and delayed coking. Performance evaluation methods and technologies based on model and sensitivity analysis are proposed, which enables online evaluation of system operating performance by customized raw material characteristics and working conditions, and solves the problem that empirical methods cannot quantitatively evaluate the operating status of industrial processes and accurately trace non-optimal process parameters. Ten national invention patents were formed, including 6 authorizations, 6 registered computer software works, and 20 published papers. Since 2016, the project results have been successfully applied in Jiujiang Petrochemical's ten-million-ton refining unit, with light oil yields increasing by 2.8 percentage points and processing loss rates decreasing by more than 0.04 percentage points; in Zhenhai Refining and Chemical Co., Ltd., the 23 million tons/year refining and chemical integration unit was successfully applied, the yield of high-attached products increased by 0.2 percentage points, and the processing loss rate decreased by more than 0.01 percentage points. From 2016 to 2018, the cumulative new output value was 1.972 billion yuan, profits were 1.578 billion yuan, and taxes were 394 million yuan in three years. The industrial application of the above technologies is of great significance to improving the level of intelligent manufacturing in my country's refining industry and promoting the high-quality development of my country's manufacturing industry.