In the complex and changeable mining environment, it is difficult to implement the plan. For example, during the implementation process, it is found that the production plan is not feasible or the process flow is not reasonable, the unexpected events that are not under early warning occur frequently, and the production implementation results have not reached the expected effect, etc., which is one of the key difficulties faced by mining enterprises for a long time.Therefore, a new intelligent mining paradigm of "First virtual experiment then precise implementation" is proposed.Firstly, initialize the production plan or process flow according to the basic needs; Secondly, the time-varying computation experiment is carried out for the plan scheme.That is to build a computer-based artificial system with multi-agent method. The computing experiments is carried out,with the independent deduction of each simple agent and the emergence of multi-agent interaction based on subscription awareness. So as to evaluate the rationality and reliability of the scheme and predict the possible emergency situation. According to the experimental results, the scheme is adjusted and the experiments are carried out again until the expected results are stable.Finally, the optimized scheme and process flow are used for on-site execution. At the same time, the real-time perception, simulation and correction of the execution process are carried out to achieve the planned control. So as to realize process control according to plan.Driven by the Internet of things data, this low-cost, high-efficiency, zero risk and fast-speed preview method and precise process control method are of great significance for auxiliary planning and safe-efficient production.