For the complex industrial operational control process that is difficult to establish the mathematical model, by combining system identification with deep learning, the digital twin model for operational control process is established with operation indices and related variables as input and process control system setpoints as output, based on the available industrial big data in terms of process control system setpoints, operation indices and related variables, and characteristics of operational control process. An intelligent method for control of process control system setpoints based on cloud-edge collaboration is proposed. The method consists of the cloud-digital twin model for operational control process and the edge-intelligent model for control of process control system setpoints, and the self-tuning mechanism. The structure and functions of intelligent system for industrial operational control based on end-edge-cloud collaboration are proposed by combining industrial internet and industrial process control systems. An intelligent system for operational control of industrial process has been developed by the proposed intelligent method for control of control system setpoints. The intelligent system has been successfully applied in the key equipment of mineral processing-high pressure roller mill. The intelligent system operates safely, reliably and optimally, achieving significant effects in energy saving and emission reduction.