Linear Quadratic Optimization Research Articles

Relationship between maximum principle and dynamic programming principle for stochastic recursive optimal control problems of jump diffusions

SUMMARYThis paper is concerned with the relationship between maximum principle and dynamic programming principle for stochastic recursive optimal control problems of jump diffusions. Under the assumption that the value function is smooth, relations among the adjoint processes, the generalized Hamiltonian function, and the value function are given. A linear quadratic recursive utility portfolio optimization problem in the financial market is discussed to show the applications of the main result. Copyright © 2012 John Wiley & Sons, Ltd.

Fluence map optimization (FMO) with dose–volume constraints in IMRT using the geometric distance sorting method

A new heuristic algorithm based on the so-called geometric distance sorting technique is proposed for solving the fluence map optimization with dose–volume constraints which is one of the most essential tasks for inverse planning in IMRT. The framework of the proposed method is basically an iterative process which begins with a simple linear constrained quadratic optimization model without considering any dose–volume constraints, and then the dose constraints for the voxels violating the dose–volume constraints are gradually added into the quadratic optimization model step by step until all the dose–volume constraints are satisfied. In each iteration step, an interior point method is adopted to solve each new linear constrained quadratic programming. For choosing the proper candidate voxels for the current dose constraint adding, a so-called geometric distance defined in the transformed standard quadratic form of the fluence map optimization model was used to guide the selection of the voxels. The new geometric distance sorting technique can mostly reduce the unexpected increase of the objective function value caused inevitably by the constraint adding. It can be regarded as an upgrading to the traditional dose sorting technique. The geometry explanation for the proposed method is also given and a proposition is proved to support our heuristic idea. In addition, a smart constraint adding/deleting strategy is designed to ensure a stable iteration convergence. The new algorithm is tested on four cases including head–neck, a prostate, a lung and an oropharyngeal, and compared with the algorithm based on the traditional dose sorting technique. Experimental results showed that the proposed method is more suitable for guiding the selection of new constraints than the traditional dose sorting method, especially for the cases whose target regions are in non-convex shapes. It is a more efficient optimization technique to some extent for choosing constraints than the dose sorting method. By integrating a smart constraint adding/deleting scheme within the iteration framework, the new technique builds up an improved algorithm for solving the fluence map optimization with dose–volume constraints.

Multivariable norm optimal and parameter optimal iterative learning control: a unified formulation

This article investigates the two paradigms of norm optimal iterative learning control (NOILC) and parameter optimal iterative learning control (POILC) for multivariable (MIMO) ℓ-input, m-output linear discrete-time systems. The main result is a proof that, despite their algebraic and conceptual differences, they can be unified using linear quadratic multi-parameter optimisation techniques. In particular, whilst POILC has been naturally regarded as an approximation to NOILC, it is shown that the NOILC control law can be generated from a suitable choice of control law parameterisation and objective function in a multi-parameter MIMO POILC problem. The form of this equivalence is used to propose a new general approach to the construction of POILC problems for MIMO systems that approximates the solution of a given NOILC problem. An infinite number of such approximations exist. This great diversity is illustrated by the derivation of new convergent algorithms based on time interval and gradient partition that extend previously published work.

Optimal two-layer directive microphone array with application in near-field acoustical holography

Conventional near-field acoustical holography (NAH) is generally based on the free-field assumption, which can cause errors when interfering sources are present in practical environments. To cope with this problem, previous research suggested a combined pressure-velocity approach for NAH that provides certain advantages to rejection of interferences. This paper revisits this idea in a broader context of optimal array design and examines the feasibility of using unidirectional microphones in each channel of the array such that the robustness of inverse reconstruction is enhanced against interfering sources. As indicated in the simulation, the numerical noise in finite difference estimation of particle velocity can nullify the advantage of the well-conditioned velocity-based reconstruction. In the proposed approach, the characteristics of each array channel consisting of two microphones are tailored by taking into account not only the directivity, but also the robustness against self-noise. An objective function based on directivity index and white noise gain is exploited in a linear quadratic optimization of a two-element end-fire array. The proposed optimal array is validated in conjunction with the equivalent source model (ESM) -based NAH through numerical simulations, with an interfering source positioned behind the array. The results have shown the directive optimal array has yielded improved quality of images in comparison with conventional approaches in the presence of an interfering source and sensor noise.

Bumpless transfer for switched linear systems

This article deals with bumpless transfer for discrete-time switched linear systems. A bumpless transfer controller that is activated at every switching time for reducing the transient behavior is presented. The bumpless transfer control design is based on the finite horizon solution of a linear quadratic optimization problem. Dwell time conditions for assessing asymptotic stability of the closed-loop switched system are provided.

Trajectory Planning and Optimized Adaptive Control for a Class of Wheeled Inverted Pendulum Vehicle Models

In this paper, we investigate optimized adaptive control and trajectory generation for a class of wheeled inverted pendulum (WIP) models of vehicle systems. Aiming at shaping the controlled vehicle dynamics to be of minimized motion tracking errors as well as angular accelerations, we employ the linear quadratic regulation optimization technique to obtain an optimal reference model. Adaptive control has then been developed using variable structure method to ensure the reference model to be exactly matched in a finite-time horizon, even in the presence of various internal and external uncertainties. The minimized yaw and tilt angular accelerations help to enhance the vehicle rider's comfort. In addition, due to the underactuated mechanism of WIP, the vehicle forward velocity dynamics cannot be controlled separately from the pendulum tilt angle dynamics. Inspired by the control strategy of human drivers, who usually manipulate the tilt angle to control the forward velocity, we design a neural-network-based adaptive generator of implicit control trajectory (AGICT) of the tilt angle which indirectly "controls" the forward velocity such that it tracks the desired velocity asymptotically. The stability and optimal tracking performance have been rigorously established by theoretic analysis. In addition, simulation studies have been carried out to demonstrate the efficiency of the developed AGICT and optimized adaptive controller.

The aggregate capacity of virtual resources – linear models

This article addresses resource modelling with a specific interest on capacity aggregation. The capacity of a resource with regard to d types of tasks is modelled by a simplex and by a cube in the d-dimensional space. The aggregate capacity model of a virtual resource, that is, of a pool of r resources, is the d-polytope provided by the Minkowski sum of the simplices and cubes modelling the resources. The parametric identification of the d-planes supporting the facets of the polytope is established for r = 2 and linear simplifications are provided in the (r, d) general case. Formulated as linear inequalities, the aggregate capacity models fit in with linear programming and quadratic programming optimization techniques.

Maximum Principle for Stochastic Recursive Optimal Control Problems Involving Impulse Controls

We consider a stochastic recursive optimal control problem in which the control variable has two components: the regular control and the impulse control. The control variable does not enter the diffusion coefficient, and the domain of the regular controls is not necessarily convex. We establish necessary optimality conditions, of the Pontryagin maximum principle type, for this stochastic optimal control problem. Sufficient optimality conditions are also given. The optimal control is obtained for an example of linear quadratic optimization problem to illustrate the applications of the theoretical results.

Newton–Picard-Based Preconditioning for Linear-Quadratic Optimization Problems with Time-Periodic Parabolic PDE Constraints

We develop and investigate two preconditioners for a basic linear iterative splitting method for the numerical solution of linear-quadratic optimization problems with time-periodic parabolic PDE constraints. The resulting real-valued linear system to be solved is symmetric indefinite. We propose all-at-once symmetric indefinite preconditioners based on a Newton–Picard approach which divides the variable space into slow and fast modes. The division is performed either classically with eigenspace methods or with a novel two-grid approach. We prove mesh-independent convergence for the classical Newton–Picard preconditioner, present a complexity analysis, and show numerical results for the classical and the two-grid preconditioners. Moreover, the preconditioners compare favorably with existing symmetric positive definite Schur complement preconditioners in a Krylov method context.

New Insights into Optimal Control of Nonlinear Dynamic Econometric Models: Application of a Heuristic Approach

Optimal control of dynamic econometric models has a wide variety of applications including economic policy relevant issues. There are several algorithms extending the basic case of a linear-quadratic optimization and taking nonlinearity and stochastics into account, but being still limited in a variety of ways, e.g., symmetry of the objective function and identical data frequencies of control variables. To overcome these problems, an alternative approach based on heuristics is suggested. To this end, we apply a 'classical' algorithm (OPTCON) and a heuristic approach (Differential Evolution) to three different econometric models and compare their performance. In this paper we consider scenarios of symmetric and asymmetric quadratic objective functions. Results provide a strong support for the heuristic approach encouraging its further application to optimum control problems.

Condition Number Theorems in Linear-Quadratic Optimization

The condition number of a given mathematical problem is often related to the reciprocal of its distance from ill-conditioning. Such a property is proved here in the infinite-dimensional setting for linear-quadratic convex optimization of two types: linearly constrained convex quadratic problems, and minimum norm least squares solutions. A uniform version of such theorem is obtained in both cases for suitably equi-bounded classes of optimization problems. An application to the conditioning of a Ritz method is presented. For least squares problems it is shown that the semi-Fredholm property of the operators involved determines the validity of a condition number theorem.

Stabilization Theory and Policy: 50 Years after the Phillips Curve

This paper discusses the impact of A. W. H. Phillips' seminal contributions to stabilization policy on subsequent developments in that field. We review the policy rules introduced by Phillips and show how these relate to the optimal rules emerging from linear–quadratic optimization problems. The consequences of rational expectations for the design of optimal stabilization policy are discussed. These challenges arose from the role of ‘forward‐looking’ variables, through issues such as the ‘Lucas Critique’ and the ‘time consistency’ of policy. The paper also discusses some of the contemporary aspects of stabilization policy, thereby highlighting the durability of Phillips' contributions.

Relative Choice Models for Income Drawdown in a Defined Contribution Pension Scheme

This paper extends a target-based model of income drawdown developed in Gerrard et al. (Insurance: Mathematics and Economics 35: 321–342 [2006]) (GHV) for the distribution phase of a defined contribution pension scheme. The optimal investment strategy of the pension fund and the optimal drawdown are found using linear-quadratic optimization, which minimizes the deviation of the fund and the drawdown from prescribed targets. The GHV model is modified by nondimensionalizing the loss function, so that there is a relative choice between outcomes. Using this model, three classes of target are studied. Endogenous deterministic targets are suggested from the form of the optimal controls, while exogenous deterministic targets can be stated without knowledge of the optimization problem. The third class of stochastic targets is similar to recent annuity products, which incorporate investment risk. Each scheme represents a trade-off between investment risk and return, and this is illustrated by numerical simulation with reference to a canonical example. A particularly attractive form of income drawdown is given by an implied rate of return target. This yields a reasonable investment strategy and a robust consumption profile with age. In addition, it can be easily explained to pension scheme members.

LQ관측기를 사용한 새로운 누적방지 기법

In order to overcome some problems of existing anti-windup methods, this paper defines LQ (Linear Quadratic) observer and proposes a new anti-windup method using the LQ observer. LQ observer is derived by linear quadratic optimization in order to calculate controller states, which make the controller outputs equal to the plant inputs. And we propose an algorithm so that it can be implemented by a digital controller easily. The relationship between the design parameters and the anti-windup performance is shown via some numerical examples, which cover the cases with the anti-windup method using LQ observer designed and the case without it. Finally, the anti-windup performance of the proposed method is exemplified via comparison with the existing model-based conditioning scheme method[4].

Optimised speed control in state space for PMSM direct drives

Nowadays, more and more applications use direct drive actuators. This leads to a more flexible structure represented by a two-mass model. Furthermore, the parameter variations create an oscillating system and need a new definition of controller synthesis to maintain unchanged performance. Firstly, a common pole placement is designed, followed by three methods based on linear quadratic optimisation: the first is optimised with a small number of parameter, the second takes the energy into account and the third combines this principle with the pole placement to achieve all the goals required by an industrial application. Finally, all controllers are compared with results obtained on an experimental test bench. All controllers reach the requirements but the last is quickly tuned and has positive results, thus providing an interesting method for engineers.

Linear quadratic gaussian optimization approach for signal-to-noise ratio constrained control over network

The research area of control over networks has attracted great interest in recent years. Inserted in this research area is the study of control feedback limitations imposed by the presence of a communication channel. In this paper we analyze the fundamental limitations in control feedback stabilizability imposed by a class of Signal-to-Noise Ratio (SNR) constrained communication channels. We solve the SNR constrained control over network problem as a linear quadratic Gaussian (LQG) optimization with loop transfer recovery (LTR). If the communication channel is located on the feedback path then the LTR is said to be performed at the output. Vice versa, if the communication channel is on the control path, then the recovery is said to be performed at the input. In the present paper we address both cases, namely the LQG optimization with LTR at the output and the LQG optimization with LTR at the input to solve an LTI SNR constrained problem. We then explore the link between these two solutions.

Revisited LQ output-feedback control: minimax controller for a set of initial states

A minimax linear quadratic (LQ) output-feedback controller is introduced which minimises the maximal value of the performance index over all initial states belonging to some set separated out a priori. If the set is an ellipsoid or a polygon, such controllers are synthesised in terms of linear matrix inequalities (LMIs). In particular case when a size of this set tends to zero tightening to a point, the minimax LQ controller approaches the optimal LQ output-feedback controller for the given initial state, while in another extreme case when this size tends to infinity, we have the worst-case LQ output-feedback controller. Numerical results for an inverted pendulum are presented.

On the stratification of a class of specially structured matrices

We consider specially structured matrices representing optimization problems with quadratic objective functions and (finitely many) affine linear equality constraints in an n-dimensional Euclidean space. The class of all such matrices will be subdivided into subsets [‘strata’], reflecting the features of the underlying optimization problems. From a differential-topological point of view, this subdivision turns out to be very satisfactory: Our strata are smooth manifolds, constituting a so-called Whitney Regular Stratification, and their dimensions can be explicitly determined. We indicate how, due to Thom's Transversality Theory, this setting leads to some fundamental results on smooth one-parameter families of linear-quadratic optimization problems with (finitely many) equality and inequality constraints.

The order of singularity of pulse optimal controls in degenerate linear-quadratic optimization problems with time delay

In the paper, we study a problem of minimization of degenerate integral quadratic functional of special delayed structure (a subintegral function contains items concentrated at the moment and items with delay in phase coordinates) on trajectories of a linear system of differential equations with time delay in phase coordinates considered on a finite period of time. Sufficient conditions for the existence of optimal program control in space of generalized functions of the lth order of singularity are established and an optimal control is obtained. A special feature of the problem is that the optimal program contains pulse components concentrated at the ends of the control gap.

LQ optimal control for 2D Roesser models of finite extent

This paper investigates several aspects of linear–quadratic (LQ) optimal control for Roesser models over a two-dimensional (2D) signal-index set of finite extent. First, we consider the characterisation and computation of open-loop control laws when constraints on the system semi-states are imposed at both the south-west and north-east boundaries of the frame (i.e. signal-index set) of interest; by virtue of the quarter-plane causal structure of the Roesser model, the south-west and north-east boundary conditions are analogous to initial conditions and terminal constraints, respectively. A necessary and sufficient characterisation of optimality is obtained and explicitly computable formulae are derived to characterise the corresponding control inputs and performance index under reasonable assumptions on the problem data. In the second part of the paper, the problem of optimal LQ control via semi-state feedback is considered. A 2D Riccati-like difference equation is introduced to characterise, in a sufficient sense, a solution to this problem.