(264d) Quasi-Decentralized Networked Process Control Using A State-Dependent Communication Policy

Quasi-decentralized control refers to a distributed control strategy in which most signals used for control are collected and processed locally, while some signals are transferred between the local units and controllers to adequately account for the interactions between them and minimize the propagation of process upsets. A key objective of this strategy is to enforce the desired closed-loop stability and performance objectives of the plant with minimal cross communication between the component subsystems. This is an important consideration particularly when the communication medium is a resource-constrained wireless sensor network where conserving network resources is key to prolonging the service life of the network and minimizing frequent battery replacements.

An effort to address this problem was initiated in [1] where a quasi-decentralized networked control architecture that keeps the rate of communication between the plant units to a minimum without jeopardizing closed-loop stability was developed. The main idea was to embed in the local control system of each unit a set of dynamic models that provide the local controller with estimates of the states of the neighboring units, in order to be used when state information is not transmitted over the network. The results where subsequently generalized in [2] to address the problem when full-state measurements are not available. An exact characterization of the minimum allowable communication rate was obtained in terms of the plant dynamics, the quality of models, as well as the control laws and state observers. A key feature of the communication logic used in both cases is that it is static in the sense that the communication rate is constant and can be computed off-line prior to plant operation. A constant communication rate, however, may not always be the best choice, especially in cases when plant operations are subject to unpredictable and time-varying external disturbances which, if unaccounted for, can degrade the networked closed-loop performance and may even lead to instability.

One approach to deal with this problem is to use robust control techniques to obtain an upper bound on the minimum stabilizing communication rate if an upper bound on the size of the disturbances is available. However, the resulting bound is typically conservative and may lead to an unnecessary increase in the utilization of network resources, especially when the external disturbances are not persistent. Furthermore, in cases of unpredictable disturbances, information about the size of the disturbances may not be available or easily obtained. An alternative approach is to design the networked control system in a way that allows the network scheduler to determine and adjust the necessary communication rate on-line (i.e., during plant operation) based on the state of the plant. An advantage of this dynamic (feedback-based) communication policy is that it is more robust to unpredictable disturbances and allows the plant to respond quickly in an adaptive fashion to a unit that requires immediate attention. Another key advantage of this approach is that it ultimately leads to a more efficient utilization of network resources since the communication rate is increased only when necessary to maintain closed-loop stability.

Motivated by these considerations, we present in this work a model-based quasi-decentralized networked control structure with a state-dependent communication policy for plants with interconnected processing units that exchange measurements over a shared, resource-constrained communication network. The objective is to find a strategy for establishing and terminating communication between the local control systems in a way that minimizes the utilization of network resources without jeopardizing closed-loop stability. To this end, we initially synthesize for each unit a Lyapunov-based controller that enforces closed-loop stability in the absence of communication suspensions. To reduce network resource utilization, we include within each local control system a set of dynamic models that provide estimates of the states of its neighbors in the plant when communication is suspended and measurements are not transmitted through the network. To determine when the models's states must be updated and communication re-established, the evolution of each Lyapunov function is monitored locally within each unit such that if it begins to increase at any time, the sensor suites of the neighboring units are prompted to send their data over the network to update their corresponding models. Communication is then suspended for as long as the Lyapunov functions continue to decay. The underlying idea here is to use the Lyapunov stability constraint for each unit as the basis for switching on or off the communication between a given unit and its neighbors. This formulation, which leads to a state-dependent time-varying communication rate, is flexible and can be extended in several directions which will be discussed in the presentation. Finally, the results are illustrated through an application to a chemical plant example.

References:

[1] Sun, Y. and N. H. El-Farra, ``Quasi-decentralized Model-based Networked Control of Process Systems", Comp. & Chem. Eng., 32: 2016-2029, 2008.

[2] Sun, Y. and N. H. El-Farra, ``Quasi-decentralized State Estimation and Control of Process Systems Over Communication Networks", Proceedings of the 47th IEEE Conference on Decision and Control, pp. 5468-5475, Cancun, Mexico, 2008.