Use Of Barrier Functions Research Articles

Safety Certification for Stochastic Systems via Neural Barrier Functions

Providing non-trivial certificates of safety for non-linear stochastic systems is an important open problem. One promising solution to address this problem is the use of barrier functions. Barrier functions are functions whose composition with the system forms a Martingale and enable the computation of the probability that the system stays within a safe set over a finite time horizon. However, existing approaches to find barrier functions generally restrict the search to a small class of functions, often leading to conservatism. To address this problem, in this letter, we parameterize barrier functions as neural networks and show that bound propagation techniques and linear programming can be successfully employed to find Neural Barrier Functions. Further, we develop a branch-and-bound scheme based on linear relaxations that improves the scalability of the proposed framework. On several case studies we show that our approach scales to neural networks of hundreds of neurons and multiple hidden layers and often produces certificates of safety that are tighter than state-of-the-art methods.

Control Barrier Functions in Sampled-Data Systems

This paper presents conditions for ensuring forward invariance of safe sets under sampled-data system dynamics with piecewise-constant controllers and fixed time-steps. First, we introduce two different metrics to compare the conservativeness of sufficient conditions on forward invariance under piecewise-constant controllers. Then, we propose three approaches for guaranteeing forward invariance, two motivated by continuous-time barrier functions, and one motivated by discrete-time barrier functions. All proposed conditions are control affine, and thus can be incorporated into quadratic programs for control synthesis. We show that the proposed conditions are less conservative than those in earlier studies, and show via simulation how this enables the use of barrier functions that are impossible to implement with the desired time-step using existing methods.

Port-Hamiltonian based Optimal Power Flow algorithm for multi-terminal DC networks

In this paper an algorithm for solving the Optimal Power Flow problem for multi-terminal DC networks based on the gradient method is proposed. The aim is seeking the optimal point subject to voltage, current and power constraints. The algorithm is described by a continuous-time port-Hamiltonian model, and the inequality constrains are included by the use of barrier functions. The dynamics of the algorithm is studied and stability conditions are obtained. Finally, the method is used for the offshore wind integration grid in the North Sea and the interconnection with the network dynamics is tested by means of numerical simulations.

Adaptive truncation of matrix decompositions and efficient estimation of NMR relaxation distributions

The two most successful methods of estimating the distribution of nuclear magnetic resonance relaxation times from two dimensional data are data compression followed by application of the Butler–Reeds–Dawson algorithm, and a primal-dual interior point method using preconditioned conjugate gradient. Both of these methods have previously been presented using a truncated singular value decomposition of matrices representing the exponential kernel. In this paper it is shown that other matrix factorizations are applicable to each of these algorithms, and that these illustrate the different fundamental principles behind the operation of the algorithms. These are the rank-revealing QR (RRQR) factorization and the LDL factorization with diagonal pivoting, also known as the Bunch–Kaufman–Parlett factorization. It is shown that both algorithms can be improved by adaptation of the truncation as the optimization process progresses, improving the accuracy as the optimal value is approached. A variation on the interior method viz, the use of barrier function instead of the primal-dual approach, is found to offer considerable improvement in terms of speed and reliability. A third type of algorithm, related to the algorithm known as Fast iterative shrinkage-thresholding algorithm, is applied to the problem. This method can be efficiently formulated without the use of a matrix decomposition.

Use of barrier functions for optimal correction of improper problems of linear programming of the 1st kind

Possibilities are investigated of the combined use of interior and exterior penalty functions for discovering the generalized (approximate) solutions of improper problems of linear programming of the 1st kind. Diagrams of algorithms, theorems of convergence, results of numerical experiments are presented.

A NEW REAL-TIME APPROACH FOR NONLINEAR MODEL PREDICTIVE CONTROL

A new formulation of continuous-time nonlinear model predictive control is developed in which the parameters defining a piecewise-constant input trajectory are adapted continuously in real time. Stability is then proven when the parameter minimization evolves in the same timescale as the process dynamics. Pointwise constraints are incorporated by use of barrier functions.