We present an alternating augmented Lagrangian method for convex optimization problems where the cost function is the sum of two terms, one that is separable in the variable blocks, and a second that is separable in the difference between consecutive variable blocks. Examples of such problems include Fused Lasso estimation, total variation denoising, and multiperiod portfolio optimization with transaction costs. In each iteration of our method, the first step involves separately optimizing over each variable block, which can be carried out in parallel. The second step is not separable in the variables, but can be carried out very efficiently. We apply the algorithm to segmentation of data based on changes in mean ( 1 mean filtering) or changes in variance ( 1 variance filtering). In a numerical example, we show that our implementation is around 10000 times faster compared with the generic optimization solver SDPT3.

Model predictive control has become an increasingly popular control strategy thanks to the ability to handle constrained systems. Obtaining the required models through system identification is often a time consuming and costly process. Applications oriented experiment design is a means of reducing this effort but is often formulated in terms of the input's spectral properties. Therefore, time domain constraints are difficult to enforce. In this contribution we combine MPC with experiment design to formulate a control problem where excitation constraints are included. The benefits are that time domain constraints are respected while the experiment design criteria are fulfilled. The method is evaluated on a numerical example.

No abstract

We present an Alternating Direction Method of Multipliers (ADMM) algorithm for solving optimization problems with an 1 regularized least-squares cost function subject to recursive equality constraints. The considered optimization problem has applications in control, for example in 1 regularized MPC. The ADMM algorithm is easy to implement, converges fast to a solution of moderate accuracy, and enables separation of the optimization problem into sub-problems that may be solved in parallel. We show that the most costly step of the proposed ADMM algorithm is equivalent to solving an LQ regulator problem with an extra linear term in the cost function, a problem that can be solved efficiently using a Riccati recursion. We apply the ADMM algorithm to an example of 1 regularized MPC. The numerical examples confirm fast convergence to moderate accuracy and a linear complexity in the MPC prediction horizon.

Abstract-This paper considers a recently proposed framework for experiment design in system identification for control. We study model based control design methods, such as Model Predictive Control, where the model is obtained by means of a prediction error system identification method. The degradation in control performance due to uncertainty in the model estimate is specified by an application cost function. The objective is to find a minimum variance input signal, to be used in system identification experiment, such that the control application specification is guaranteed with a given probability when using the estimated model in the control design. We provide insight in the potentials of this approach by finite impulse response model examples, for which it is possible to analytically solve the optimal input problem. The examples show how the control specifications directly affect the excitation conditions in the system identification experiment.