Double-layer Network-based Control For Operation Of Industrial Production Processes

Integrated automation technology for industrial processes is the key to the integra-tion of industrialization and informatization and relies on the integrated control technol-ogy for the overall manufacturing process, which is based on the operation control toachieve certain integrated operation index. Thus operation control, or control for indus-trial operation, is highlight by the scholars in industry and academy.This paper aims at tracking operation objective in industrial process. By consideringthe features and requirements of network-based industrial manufacturing, a double-layernetwork-based control structure based on device layer and operation layer is proposed.Moreover, via utilizing this new structure, industrial processes, classified into two cate-gories according to linear and nonlinear description, are considered, along with real-timeoptimization, multirate sampling, package dropouts, etc. The approaches designed in thispaper tackle the problem of conventional upper-layer-openloop control and success in thedynamic regulation and operation objective tracking. What is studied in this paper, whichis the basic prospection in this problem, is novel and seldom reported in other existingstudies. The detailed content of this paper is summarized as follow.(1) By overviewing the development of the structure of industrial control structuresand analyzing the inaccuracy tracking of operation objective in the one layer control situ-ation, a two-layer control scheme is proposed. While considering the requirement of inte-gration of industrialization and informatization and “Industry4.0”, network is introducedinto industrial process control and a control structure of double-layer network-based con-trol for operation of industrial production processes is proposed.(2) Considering industrial models can be linearized of around their steady states, asetpoints compensation approach is proposed for linear systems in the situation of double-layer network control scheme. Plants at the device layer are controlled by the local regu-lation controllers, and a multirate output feedback control approach for setpoints compen-sation is proposed such that the local plants can reach the dynamically changed setpointsand the given economic objective can also be tracked via certain economic performanceindex (EPI). First, a sampled-data multivariable output feedback proportional integral (PI)controller is designed to regulate the performance of local plants. Second, the outputs and control inputs of the local plants at the device layer are sampled at operation layer sam-pling time to form the EPI. Thus the multirate problem is solved by a lifting method.Third, a situation without taking network into consideration is analyzed to obtain the the-orem to design the sepoints compensator. Then, a networked case is studied consideringunreliable data transmission described by a stochastic packet dropout model. Finally, arougher flotation process model is employed to demonstrate the efectiveness of the pro-posed method.(3) Considering more general industry model, a setpoints compensation for nonlin-ear industrial processes with disturbances based on fuzzy logic control is proposed. First,the nonlinear plants in the device layer are modeled by using Takagi-Sugeno (T-S) fuzzytechnique, and are controlled by local PI controller to guarantee the H∞performance.Then, the outputs and inputs of local plants are sampled and transited to the operationlayer to form the economic performance index (EPI), which is used to represent the per-formance of the tracking of economic objective. Furthermore, the setpoints, which aredynamically changing, are calculated via a compensator based on the error between theobjective and the EPI at each step of the operation layer. Finally, the efectiveness of theproposed method is demonstrated by a nonlinear continuous stirred tank reactor (CSTR)model.