A datadriven robust optimization ddro methodology is proposed to deal with uncertainty in the optimization of steam system in an ethylene plant. Distributionally robust optimization closeness to reference distribution 1. Motivated by this growing availability, we propose a novel schema for utilizing data to design uncertainty sets for robust optimization using statistical hypothesis tests. The theory is supported by numerous examples and computational illustrations. Robust datadriven optimization using machine learning and. Datadriven distributionally robust polynomial optimization. Therefore, the proposed uncertainty set can flexibly capture a. Optimization under moment, robust, and datadriven models. Datadriven robust optimzation an introduction to developing. Datadriven robust optimization mathematical programming. Datadriven distributionally robust optimization for power. Jul 12, 2018 then the density contourbased uncertainty set in this work will be adaptively modified. An alternative streamline of datadriven optimization is the statistical robust optimization, which utilizes data, hypothesis tests, and likelihood to construct the statistically interpretable uncertainty set at a high confidence level 9, 10, 11.
Portfolio optimization results are strongly dependent on the model parameters. Datadriven robust optimization for steam systems in. Robust optimization is an important sub eld of optimization that deals with uncertainty in the data of optimization problems. An industrial case study of the blending system is conducted to indicate the effectiveness of the proposed method. Motivated by this growing availability, we propose a novel schema for utilizing data to design uncertainty sets for robust optimization using. Datadriven optimal transport cost selection for distributionally robust optimization jose blanchet management science and engineering stanford university 475 via ortega, suite 310 stanford, ca, 94305, usa yang kang department of statistics columbia university 1255th amsterdam ave. Jan 20, 2021 in this paper, a datadriven distributionally robust optimization dro method for rted considering automatic generation control agc is proposed. Robust datadriven vehicle routing with time windows.
Totestforpracticalperformance,weleverageauniquedatasetfroma cancer center that combines realtime locations, electronic health records, and appointments log. Data driven robust optimization 7 delage and ye 8 prov ed that the robust counterpart of the problem 2 according to u can be formulated as a semidefinite program sdp. Datadriven robust optimization 237 p2 the set u implies a probabilistic guarantee for p. Distributionally robust optimization under moment uncertainty with application to datadriven problems. Datadriven robust optimization with known marginal distributions rui gao, anton j. Distributionally robust optimization reference delage, e. Different from conventional robust optimization methods, the proposed framework incorporates distributional information to avoid over. This entails the development of a datadriven robust optimization approach with noveluncertaintysets. Datadriven robust optimization has also been applied to industries, including steam systems, biodiesel and oil production.
In this paper, a datadriven distributionally robust optimization dro method for rted considering automatic generation control agc is proposed. Datadriven robust optimization 239 gtest are based on phidivergences see also 32. A hybrid model of extractionexhausting steam turbine is developed, and its coefficients are considered as uncertain parameters. The counterpart of the proposed ddro model is formulated to promote the optimization process. They focus on the case where the uncertain parameters u. On the power of robust solutions in twostage stochastic and adaptive optimization problems, with vineet goyal, submitted to mathematics of operations research, 2009.
The main idea is to automatically build a single and global predictive model using a machine learning technique random forests, and then to use a derivativefree blackbox optimization technique mads to maximize a performance criterion. While there are many high optimization ro came from the robust control community, ro is a distinct. First, a copulabased agc signal model is developed to reflect the correlations among the agc signal, load power and renewable generation variations. Datadriven robust optimization with known marginal distributions. Polynomial optimization problems are inherently hard due to nonconvex objectives and. This paper proposes a novel scheduling approach of batch processes based on a datadriven robust mixedinteger linear programming milp model.
Robust optimization for process scheduling under uncertainty. Datadriven learning in dynamic pricing using adaptive. Momentbased uncertainty sets momentbased uncertainty sets is comparable to that of the classical. Momentbased uncertainty sets momentbased uncertainty sets is. This set is a ball around a density function estimated from data samples, i. Optimizationbased scenario reduction for datadriven two. Distributionally robust optimization under moment uncertainty operations research 000, pp. Datadriven robust optimization based on kernel learning chao shanga, xiaolin huangb, fengqi you,a asmith school of chemical and biomolecular engineering, cornell university, ithaca, new york 14853, usa binstitute of image processing and pattern recognition, shanghai jiao tong university, shanghai 200400, china abstract we propose piecewise linear kernelbased support vector clustering svc. An essential book for anyone working on optimization. May 24, 2008 robust optimization for the pooling problem.
Datadriven optimization uses observations of the random variables as direct inputs to the mathematical programming problems. In this paper we develop a method of data driven stochastic programming that avoids the arti. Before illustrating the detailed procedures, a timeline for the rollinghorizon scheduling is defined. Rm1005 new york, ny, 10027, usa karthyek murthy engineering. Abstractmotivated by datadriven decision making and sampling problems, we investigate probabilistic interpretations of robust optimization ro. We establish a connection between ro and distributionally robust stochastic programming drsp, showing that the solution to any ro problem is also a solution to a drsp problem. Datadriven distributionally robust optimization for real. Datadriven distributionally robust optimization for power system operations by yuanyuan guo a dissertation submitted in partial ful. We show that common datadriven methods, such as sample average approximation, datadriven robust optimization, and certain regularized policies, may perform poorly in this new setting. Introduction to robust optimization robust optimization. Apr 15, 2020 this chapter proposes a novel method for semi. Datadriven adaptive robust unit commitment under wind power. Then, we make distributionally robust decisions that hedge against the worstcase distributions within the ambiguity set. Dro formulation and a datadriven calibration of the underlying uncertainty set to handle.
Under this framework, the objective and constraint functions are only assumed to belong to certain sets in function space the socalled \uncertainty sets. We introduce a simple uncertainty set for arrival rates based on some structural properties of optimal queueing and routing solutions. Robust optimization methodology and applications 457 k is or is not polyhedral, we see that the optimal value inpiy is equal to the one in the solvable dual problem max. We note the recent progress in datadriven distributionally robust optimization dro, where it has been shown that the worstcase expectation of an uncertain cost over all distributions that are within a xed wasserstein distance from a discrete reference distribution can often be computed e ciently via convex optimization esfahani. Jan 07, 2019 this paper proposes a novel datadriven adaptive robust optimization aro framework for the unit commitment uc problem integrating wind power into smart grids. Selim mankai, france ipag lab, ipag business school. Abstract a novel datadriven stochastic robust optimization ddsro framework is proposed for optimization under uncertainty leveraging labeled multiclass uncertainty data. Datadriven robust optimization based on kernel learning.
Hanasusanto imperial college london london sw7 2az, uk g. In the first part, we propose a novel schema for utilizing data to design uncertainty sets for robust optimization using hypothesis testing. This paper presents a framework that was developed to achieve datadriven robust optimization of processes. We propose a twostage, online signal control strategy for dynamic networks using a linear decision rule ldr approach and a distributionally robust optimization dro technique. Optimization, in robust and constrained optimization. Datadriven distributionally robust optimization for vehicle. We consider robust optimization for polynomial optimization problems where the uncertainty set is a set of candidate probability density functions. Delage and ye 2010 develop datadriven distributionally robust optimization. A distributional interpretation of robust optimization. Datadriven and robust optimization approaches to call centers, with x. The primary goal of this preface is to provide the reader with a.
Datadriven adaptive robust unit commitment under wind. The approach is flexible and widely applicable, and robust optimization problems built from our new sets are. This book is devoted to robust optimization a speci. Datadriven robust milp model for scheduling of multipurpose batch. Likelihood robust optimization for datadriven problems. The key idea of robust saa is to approximate 1 by a particular datadriven, distributionally robust optimization.
Focusing on one of the centers infusion units roughly 90 daily. Finally, we investigate the application of datadriven approach with risk aversion and robust optimization approach to solve sta. We then propose a novel framework for selecting a datadriven policy from a given policy class. Feb 25, 2017 the last decade witnessed an explosion in the availability of data for operations research applications. The last decade witnessed an explosion in the availability of data for operations research applications. Processes free fulltext datadriven robust optimization. Datadriven distributionally robust optimization for vehicle balancing of mobilityondemand systems authors. In the first offline stage, a distributionally robust optimization dro problem is.
Robust optimization, a concept as familiar as it is misunderstood, is clarified in this chapter. This inspired us to develop a new strategy to dynamically utilize the datadriven robust optimization models in section 4 for the petrochemical scheduling. Datadriven decision making under uncertainty integrating. By leveraging a dirichlet process mixture model, a datadriven uncertainty set for wind power forecast errors is constructed as a union of several basic uncertainty sets. Smalldata, largescale linear optimization with uncertain. Finite adaptability in datadriven robust optimization for production scheduling. To prevent overfitting, we propose a distributionally robust optimization model that uses a wasserstein distancebased ambiguity set to characterize ambiguous distributions that are close to. The approach consists of solving a robust optimization problem that is constructed from sample paths of the underlying stochastic process. The first offline stage formulates a ldr that maps realtime.
Machine learning methods including dirichlet process mixture model and maximum likelihood estimation are. The application of robust optimization in power systems. Datadriven optimization, distributionally robust optimization, large deviations theory. Datadriven robust optimization for crude oil blending. Optimization under moment, robust, and datadriven models of. Motivated by datadriven decision making and sampling problems, we investigate probabilistic interpretations of robust optimization ro. How does robust optimization deal with uncertainty.
Uncertainty data in large datasets are often collected from various conditions, which are encoded by class labels. The approach is flexible and widely applicable, and robust optimization problems built from our new data driven sets are computationally tractable, both. Pdf datadriven distributionally robust optimization for. May 08, 2020 a novel datadriven robust model to optimize crude oil blending is then developed on the basis of the derived datadriven uncertainty set. Those losses include emissions, global warming, and automobile accidents. Computational results with real data of a call center show that a simple robust optimization approach can be more e. The uncertain parameters of the turbine model are derived from the semiempirical model and historical data. The goal is to make a decision that is feasible no matter. A bayesian nonparametric modelthe dirichlet process mixture modelis adopted and combined with a variational inference algorithm to extract the information embedded within uncertainty data. We propose a new datadriven approach for addressing multistage stochastic linear optimization problems with unknown distributions. Pdf distributionally robust optimization under moment. While automobiles provide value to society such as transportation and pleasure of driving, automobiles are producing significant amounts of losses.
1103 955 4 1857 532 967 1247 1773 834 97 68 1359 609 1543 1738 183 1078 1188 292 140 1626 1306 482 194 518 1101