Research On The Scenario Based Simulation And The Scenarios Generation Methods Of Process Industry

Process industry is the pillar industry of China's national economy,and its optimization,management and control research depend on simulation technology.With the rise of intelligent manufacturing technology represented by big data,cloud computing and Internet of things in recent years,enterprise production presents the trend of service-oriented,flexible and green.The increase of production uncertainty brings more complex scenarios to enterprise production simulation.As a kind of stochastic simulation which describes resource uncertainty by scenarios,scenario based simulation has attracted the attention of many manufacturing enterprises,simulation researchers and simulation tool manufacturers at home and abroad.However,the current scenario based simulation can not meet the needs of complex and changeable scenario configuration and rapid decision-making in intelligent manufacturing era,and the demand of scenario data still hinders the further full application of scenario based simulation.Based on the requirements of scenario based simulation in intelligent manufacturing era,a new paradigm of scenario based simulation is proposed.At the same time,the generation methods of scenario data are proposed for the requirements supporting scenario based simulation.The production scheduling scenario simulations based on these methods are also realized.The main work and innovation are as follows:(1)In view of the new demand of industrial enterprises for scene simulation under the intelligent manufacturing production mode,a scenario based simulation paradigm for intelligent manufacturing is proposed.Compared with the traditional scenario based simulation paradigm,the new paradigm adds the support of scenario generation method to the scenario based simulation,forming a sustainable scenario data generation and decision support.(2)A static scenario sample generation method is proposed.Based on the demand of static parameters of scenario based simulation configuration and the complex coupling of process industry scenarios,a static scenario generation method based on SAE and Vine Copulas is proposed,which can effectively generate static scenario samples conforming to the distribution and correlation of variables.(3)Aiming at the demand of time series sample generation in scenario based simulation of process industry,a dynamic scenario generation method based on Wasserstein generation countermeasure network is proposed.A generation network combined with convolutional neural network is designed.The gradient vanishing problem is improved by batch gradient descent and gradient penalty term.Based on the idea of supervised learning,the problem of multi scene occurrence is solved.(4)Focusing on the workflow simulation of production scheduling in process industry,the scenario generator is applied to the workflow simulation system of process industry production scheduling as a module.Static data and dynamic data generation are applied to the function realization of the simulation system,which reflects the application value of scenario generation method.