Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.

'The cardinal sin in control is to believe that the plant is given' Karl Astrom. Astrom, a towering figure of control theory and practice and awardee of the 1993 IEEE Medal of Honor for his work on adaptive control, provides this assessment of the obstinate part of realizing a feedback controller. And yet we are exhorted to rely on solely-data-driven methods of control design skipping the modeling and plant identification phases entirely. What is going on? Whom should we trust? How do we reconcile the implied ease (or indeed avoidance) of modeling with the steely focus on robustness of the control and the capacity of feedback to accommodate uncertainty? This paper seeks to investigate this subject with the objective of appreciating not whom to trust but what are the circumstances where the direct paradigm of control design from any lightly qualified data set provides a sensible way forward. Here is a clue: It depends on the confidence of your contentions about the plant system, the detailed data themselves and your appetite for failure. The paper attempts to segue repeatedly between the broad philosophical context and hard engineering examples. To instantiate ideas and add detail to vagaries, we incorporate a number of examples, each validated by their demonstrated commercial and industrial viability and each terminating with ein Blickwinkel or perspective in German. As David Wallace-Wells poses, before investing hundreds of billions of dollars we really ought to ask what is the trillion-dollar problem which might potentially be solved. By sticking to industrially proven technologies, we hope to delineate what works suitably well in practice or was judged worthy of the risk.

A common loop from gas processing is studied from a modeling and control perspective with an emphasis on the simplicity and control-orientation of the model components which comprise the model: pipes, branches, compressor, valves, tanks etc. The process is to take recent component model classes and to lift them to be more realistic and representative of a practical compressor operational area, notably taking into account compressor operation curves. The control objective is to improve transient response of the overall system by developing two-input controllers to manage fluctuations in either compressor suction pressure or discharge pressure in response to supply and delivery changes. The control mechanism is the use of a recycle loop capable of transporting gas from the discharge side to the suction side - this is the faster manipulated variable - together with the slower dynamics of the compressor speed, which regulates overall flow. The return of the recycle valve to closed during steady-state operation is included into the design.

Power grid, communications, computer and product reticulation networks are frequently layered or subdivided by design. The OSI seven-layer computer network model and the electrical grid division into generation, transmission, distribution and associated markets are cases in point. The layering divides responsibilities and can be driven by operational, commercial, regulatory and privacy concerns. From a control context, a layer, or part of a layer, in a network isolates the authority to manage, i.e. control, a dynamic system with connections into unknown parts of the network. The topology of these connections is fully prescribed but the interconnecting signals are largely unavailable, through lack of sensing and even prohibition. Accordingly, one is driven to simultaneous input and state estimation methods. This is the province of this paper, guided by the structure of these network problems. We study a class of algorithms for this joint task, which if the system has transmission zeros outside the unit circle leads to an unstable and unimplementable estimator. Two modifications to the algorithm to ameliorate this problem were recently proposed involving (a) replacing the troublesome subsystem with its outer factor from its inner-outer factorization or (b) using the Kalman filter for a high-variance white signal model for the unknown inputs. The outer factor has only stable transmission zeros and so is stably invertible. The Kalman filter is stable by design. Here, we establish the connections between the original estimation problem for state and input signal and the outputs/estimates from the algorithm applied solely to the outer factor. We further show that both fixes coincide. Copyright (C) 2024 The Authors. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)

Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.

Power grid, communications, computer and product reticulation networks are frequently layered or subdivided by design. The layering divides responsibilities and can be driven by operational, commercial, regulatory and privacy concerns. From a control context, a layer, or part of a layer, in a network isolates the authority to manage, i.e. control, a dynamic system with connections into unknown parts of the network. The topology of these connections is fully prescribed but the interconnecting signals, currents in the case of power grids and bandwidths in communications, are largely unavailable, through lack of sensing and even prohibition. Accordingly, one is driven to simultaneous input and state estimation methods. We study a class of algorithms for this joint task, which has the unfortunate issue of inverting a subsystem, which if it has unstable transmission zeros leads to an unstable and unimplementable estimator. Two modifications to the algorithm to ameliorate this problem were recently proposed involving replacing the troublesome subsystem with its outer factor from its inner-outer factorization or using a high-variance white signal model for the unknown inputs. Here, we establish the connections between the original estimation problem for state and input signal and the estimates from the algorithm applied solely to the outer factor. It is demonstrated that the state of the outer factor and that of the original system asymptotically coincide and that the estimate of the input signal to the outer factor has asymptotically stationary second-order statistics which are in one-to-one correspondence with those of the input signal to the original system, when this signal is itself stationary. Thus, the simultaneous input and state estimation algorithm applied just to the outer factor yields an unbiased state estimate for control and the statistics of the interface signals.

Pipe flow models are developed with a focus on their eventual use for feedback control design at the process control level, as opposed to the unit level, in gas processing facilities. Accordingly, linearized facility-scale models are generated to describe pressures, mass flows, and temperatures based on sets of nonlinear partial differential equations from fluid dynamics and thermodynamics together with constraints associated with their interconnection. As part of the treatment, the divergence of these simplified models from physics is assessed since robustness to these errors will be an objective for the eventual control system. The approach commences with a thorough analysis of pipe flow models and then proceeds to study their automated interconnection into network models, which subsume the algebraic constraints of bond graph or standard fluid modeling. The models are validated and their errors are quantified by referring them to operational data from a commercial gas compressor test facility. For linear time-invariant models, the interconnection method to generate network models is shown to coincide with automation of Mason's gain formula. These pipe network models based on engineering data are the first part of the development of general facility process control tools.

The paper provides a new approach to the determination of a single state value for stochastic output feedback problems using paradigms from Model Predictive Control, particularly the distinction between open-loop and closed-loop control and between deterministic optimal control and stochastic optimal control. The State Selection Algorithm is presented and relies on given dynamics and constraints, a nominal deterministic state-feedback controller, and a sampling based method to select the best state value, based on optimizing a prescribed finite-horizon performance function, over the available candidates provided by a particle filter. The cost function is minimized over the horizon with controls determined by the nominal controller and the selected states. So, the minimization is performed not over the selection of the control other than through the choice of state value to use. The algorithm applies generally to nonlinear stochastic systems and relies on Monte Carlo sampling and averaging. However, in linear quadratic polyhedrally constrained cases the technique reduces to a quadratic program for the state value. The algorithm is evaluated in a set of computational examples, which illustrate its efficacy and limitations. Numerical aspects and the opportunity for parallelization are discussed. The examples demonstrate the algorithm operating, in closed-loop with its attendant particle filter, over the long horizon.

This work deals with the problem of simultaneous regulation and model parameter estimation in adaptive model predictive control. We propose an adaptive model predictive control and conditions which guarantee a persistently exciting closed loop sequence by only looking forward in time into the receding prediction horizon. Earlier works needed to look backwards and preserve prior regressor data. Instead, we present a procedure for the offline generation of a persistently exciting reference trajectory perturbing the equilibrium. With the new approach we demonstrate exponential convergence of nonlinear systems under the influence of the adaptive model predictive control combined with a recursive least squares identifier with forgetting factor despite bounded noise. The results are, at this stage, local in state and parameter-estimate space.

Input estimation is a signal processing technique associated with deconvolution of measured signals after filtering through a known dynamic system. Kitanidis and others extended this to the simultaneous estimation of the input signal and the state of the intervening system. This is normally posed as a special least-squares estimation problem with unbiasedness. The approach has application in signal analysis and in control. Despite the connection to optimal estimation, the standard algorithms are not necessarily stable, leading to a number of recent papers which present sufficient conditions for stability. In this paper we complete these stability results in two ways in the time-invariant case: for the square case, where the number of measurements equals the number of unknown inputs, we establish exactly the location of the algorithm poles; for the non-square case, we show that the best sufficient conditions are also necessary. We then draw on our previous results interpreting these algorithms, when stable, as singular Kalman filters to advocate a direct, guaranteed stable implementation via Kalman filtering. This has the advantage of clarity and flexibility in addition to stability. En route, we decipher the existing algorithms in terms of system inversion and successive singular filtering. The stability results are extended to the time-varying case directly to recover the earlier sufficient conditions for stability via the Riccati difference equation.

A Maximum Likelihood recursive state estimator is derived for non-linear state–space models. The estimator iteratively combines a particle filter to generate the predicted/filtered state densities and the Expectation Maximization algorithm to compute the maximum likelihood filtered state estimate. Algorithms for maximum likelihood state filtering, prediction and smoothing are derived. The convergence properties of these algorithms, which are inherited from the Expectation Maximization algorithm and the particle filter, are examined in two examples. For nonlinear state–space systems with linear measurements and additive Gaussian noises, it is shown that the filtering and prediction algorithms reduce to gradient-free optimization in a form of a fixed-point iteration. It is also shown that, with randomized reinitialization, which is feasible because of the simplicity of the algorithm, these methods are able to converge to the Maximum Likelihood Estimate (MLE) of multimodal, truncated and skewed densities, as well as those of disjoint support.

This work provides linear control-oriented state space models of gas flow through standard equipment, such as valves, compressors, manifolds and non-trivial pipe geometries. After presenting the catalog of components, we show how to interconnect them: we briefly recall the matrix methodology derived elsewhere, accompanied by a Matlab example for a gas loop; secondly for the same example we provide an alternative approach using Matlab's connect function. All the aforementioned models are suited to model-based MIMO control design tools as they are in state-space form and they can be physically parametrized. Conservation of mass is the source of the algebraic constraints in and is a property subsumed in these new models. This is characterized by model transfer function properties at frequency zero. Each of the component models is shown to posses this property.

We study a gas network flow regulation control problem showing the closed-loop consequences of using interconnected component models, which have been designed to preserve a variant of mass flow conservation without the inclusion of algebraic constraints into the dynamics. These are candidate \textit{control-oriented} models because they are linear state-space systems. This leads to a study of mass conservation in flow models and the inheritance of conservation at the network level when present at each component. Conservation is expressed as a transfer function property at DC. This property then is shown to imply the existence of integrators and other DC structure of the network model, which has important consequences for the subsequent control design. An example based on an industrial system is used to explore the facility of moving from modeling to automated interconnection or components to model reduction to digital controller design and performance evaluation. Throughout, the focus is on the teasing out of control orientation in modeling. The example shows a strong connection between the modeling and the controller design.

Specific low-bitrate coding strategies are examined through their effect on LQ control performance. By limiting the subject to these methods, we are able to identify principles underlying coding for control; a subject of significant recent interest but few tangible results. In particular, we consider coding the quantized output signal deploying period-two codes of differing delay-versus-accuracy tradeoff. The quantification of coding performance is via the LQ control cost. The feedback control system comprises the coder-decoder in the path between the output and the state estimator, which is followed by linear state-variable feedback, as is optimal in the memoryless case. The quantizer is treated as the functional composition of an infinitely-long linear staircase function and a saturation. This permits the analysis to subdivide into estimator computations, seemingly independent of the control performance criterion, and an escape time evaluation, which ties the control back into the choice of quantizer saturation bound. An example is studied which illustrates the role of the control objective in determining the efficacy of coding using these schemes. The results mesh well with those observed in signal coding. However, the introduction of a realization-based escape time is a novelty departing significantly from mean square computations.

We extend results of the recursive-least-squares-with-forgetting-factor identifier for single-input-single-output systems to the multiple-output case by, under the assumption of persistence of excitation, deriving the corresponding minimized objective function and by showing exponential convergence of the estimation error.

A new formulation of Stochastic Model Predictive Output Feedback Control is presented and analyzed as a translation of Stochastic Optimal Output Feedback Control into a receding horizon setting. This requires lifting the design into a framework involving propagation of the conditional state density, the information state, via the Bayesian Filter and solution of the Stochastic Dynamic Programming Equation for an optimal feedback policy, both stages of which are computationally challenging in the general, nonlinear setup. The upside is that the clearance of three bottleneck aspects of Model Predictive Control is connate to the optimality: output feedback is incorporated naturally; dual regulation and probing of the control signal is inherent; closed-loop performance relative to infinite-horizon optimal control is guaranteed. While the methods are numerically formidable, our aim is to develop an approach to Stochastic Model Predictive Control with guarantees and, from there, to seek a less onerous approximation. To this end, we discuss in particular the class of Partially Observable Markov Decision Processes, to which our results extend seamlessly, and demonstrate applicability with an example in healthcare decision making, where duality and associated optimality in the control signal are required for satisfactory closed-loop behavior.

We present control-oriented models for transient dynamics of isothermal one-dimensional gas flow through multiple pipes in series and intersecting pipe geometries. These composite models subsume algebraic constraints that would otherwise appear due to boundary conditions, so that our linear state-space models are well-suited for model-based control design for gas flow in pipe networks with non-trivial geometries.

The SARS-CoV-2 (COVID-19) pandemic has placed unprecedented demands on entire health systems and driven them to their capacity, so that health care professionals have been confronted with the difficult problem of ensuring appropriate staffing and resources to a high number of critically ill patients. In light of such high-demand circumstances, we describe an open web-accessible simulation-based decision support tool for a better use of finite hospital resources. The aim is to explore risk and reward under differing assumptions with a model that diverges from most existing models which focus on epidemic curves and related demand of ward and intensive care beds in general. While maintaining intuitive use, our tool allows randomized "what-if" scenarios which are key for real-time experimentation and analysis of current decisions' down-stream effects on required but finite resources over self-selected time horizons. While the implementation is for COVID-19, the approach generalizes to other diseases and high-demand circumstances.

This work deals with the problem of simultaneous regulation and model parameter estimation in adaptive model predictive control. We propose an adaptive model predictive control and conditions which guarantee a persistently exciting closed loop sequence by only looking forward in time into the receding prediction horizon. Earlier works needed to look backwards and preserve prior regressor data. Instead, we present a procedure for the offline generation of a persistently exciting reference trajectory perturbing the equilibrium. With the new approach we demonstrate exponential convergence of nonlinear systems under the influence of the adaptive model predictive control combined with a recursive least squares identifier with forgetting factor despite bounded noise. The results are, at this stage, local in state and parameter-estimate space.