Papers by Alexander Brodsky
Journal of Decision Systems, May 7, 2018
We propose an efficient one-stage stochastic optimisation algorithm for the problem of finding pr... more We propose an efficient one-stage stochastic optimisation algorithm for the problem of finding process controls that minimise the expectation of cost while satisfying multiple deterministic and stochastic feasibility constraints with a given high probability. The proposed algorithm is based on a series of deterministic approximations to produce a candidate solution set and on a refinement step using stochastic simulations with optimal simulation budget allocation. We conduct an experimental study for a realworld manufacturing service network, which shows that the proposed algorithm significantly outperforms four popular simulation-based stochastic optimisation algorithms.
International Journal of Computer Integrated Manufacturing, Feb 6, 2019
This paper reports on the development of Factory Optima, a web-based system that allows manufactu... more This paper reports on the development of Factory Optima, a web-based system that allows manufacturing process engineers to compose, optimise and perform trade-off analysis of manufacturing and contract service networks based on a reusable repository of performance models. Performance models formally describe process feasibility constraints and metrics of interest, such as cost, throughput and CO 2 emissions, as a function of fixed and control parameters, such as equipment and contract properties and settings. The repository contains performance models representing (1) unit manufacturing processes, (2) base contract services and (3) a composite steady-state service network. The proposed framework allows process engineers to hierarchically compose model instances of service networks, which can represent production cells, lines, factory facilities and supply chains, and perform deterministic optimisation based on mathematical programming and Pareto-optimal trade-off analysis. Factory Optima is demonstrated using a case study of a service network for a heat sink product which involves contract vendors and manufacturing activities, including cutting, shearing, Computer Numerical Control (CNC) machining with milling and drilling operations, quality inspection, finishing, and assembly.
A Simulation Query Language for Defining and Analyzing Uncertain Data
This paper proposes a new approach, and studies an algorithm to address the Maximum Diversity Pro... more This paper proposes a new approach, and studies an algorithm to address the Maximum Diversity Problem (MDP) of recommendations for composite products or services. First, the proposed approach is based on constructing and using a multi-dimensional diversity feature space, which is separate from the utility space used for utility elicitation. Second, we introduce a randomized algorithm, which is based on iterative relaxation of selections by the Greedy algorithm with an exponential probability distribution. The algorithm produces a competitive solution with respect to finding a diverse set from candidate recommendations. Finally, we conduct an experimental study to compare the efficacy and efficiency of the proposed algorithm with two broadly used diversity algorithms, as well as with the exhaustive algorithm, which we could only compute for sets of up to seven returned recommendations. The experimental results show that the proposed algorithm is highly efficient computationally and that in terms of diversity, it consistently outperforms the two competitive algorithms and converges to the optimal solutions on cases run with the exhaustive algorithm in under 100 ms.
Optimal splitting for rare-event simulation
Iie Transactions, May 1, 2012
Simulation is a popular tool for analyzing large, complex, stochastic engineering systems. When e... more Simulation is a popular tool for analyzing large, complex, stochastic engineering systems. When estimating rare-event probabilities, efficiency is a big concern, since a huge number of simulation replications may be needed in order to obtain a reasonable estimate of the rare-event probability. The idea of splitting has emerged as a promising variance reduction technique. The basic idea is to create
Regression Based Algorithm for Optimizing Top-K Selection in Simulation Query Language
ABSTRACT In this paper we propose an algorithm for optimizing simulation budget allocation while ... more ABSTRACT In this paper we propose an algorithm for optimizing simulation budget allocation while minimizing the total processing cost for top-k queries. We also implement this algorithm as part of SimQL: an extension of SQL that includes probability functions expressed through stochastic simulation.
Learning Occupancy Prediction Models with Decision-Guidance Query Language
ABSTRACT Occupancy prediction is a relatively new domain of research. It has gained momentum over... more ABSTRACT Occupancy prediction is a relatively new domain of research. It has gained momentum over the past decade. Varying approaches have been proposed to profile occupancy of buildings or space. Smart occupancy patterns, once predicted, can be effectively used in modeling energy management systems to achieve energy savings. While doing so, we also take into consideration the potential for occupant discomfort. In this paper, we propose DOPM - an occupancy prediction model built by using Decision Guidance Query Language (DGQL) framework that can optimize prediction rules governing occupancy patterns in a domain. Motive of DOPM is to perform two actions: a) Maximize energy saved in a location and b) limit inconvenience caused to its occupants in the process. This paper presents a generic DOPM model. A case study is developed for occupancy prediction on a university campus setting and the results of running the model will be presented.
Analysis and optimization in smart manufacturing based on a reusable knowledge base for process performance models
In this paper, we propose an architectural design and software framework for fast development of ... more In this paper, we propose an architectural design and software framework for fast development of descriptive, diagnostic, predictive, and prescriptive analytics solutions for dynamic production processes. The proposed architecture and framework will support the storage of modular, extensible, and reusable Knowledge Base (KB) of process performance models. The approach requires the development of automatic methods that can translate the high-level models in the reusable KB into low-level specialized models required by a variety of underlying analysis tools, including data manipulation, optimization, statistical learning, estimation, and simulation. We also propose an organization and key structure for the reusable KB, composed of atomic and composite process performance models and domain-specific dashboards. Furthermore, we illustrate the use of the proposed architecture and framework by performing diagnostic tasks on a composite performance model.
Modular Modeling and Optimization of Temporal Manufacturing Processes with Inventories
Smart manufacturing requires streamlining operations and optimizing processes at a global and loc... more Smart manufacturing requires streamlining operations and optimizing processes at a global and local level. This paper considers temporal manufacturing processes that involve physical or virtual inventories of products, parts and materials that move through a network of subprocesses. The inventory levels vary with time and are a function of the configuration settings of the machines involved in the process. These environments require analysis, e.g., answering what-if questions, and optimization to determine optimal operating settings for the entire process. To address this problem, the paper proposes modular process components that can represent these manufacturing environments at various levels of granularity for performing what-if analysis and decision optimization queries. These components are extensible and reusable against which optimization and what-if questions can be posed. Additionally, the paper describes the steps to translate these complex components and optimization queries into a formal mathematical programming model, which is then solved by a commercial optimization solver.
Toward Smart Manufacturing Using Decision Guidance Analytics | NIST
COD: An adaptive utility learning method for composite recommendations
Advanced Information Management and Service, Nov 1, 2010
ABSTRACT This paper studies and proposes a method for learning the user's preferences and... more ABSTRACT This paper studies and proposes a method for learning the user's preferences and propsing recommendations on composite bundles of products and services. The user preferences are learned using a regression analysis on the historical purchase information. These learned preferences are used to infer a utility axis in the multi-dimensional utility space. Subsequently, the standard utility axes are adaptivelly adjusted towards the inferred utility axis to generate initial utility axes for the utility elicitation process. The amount by which the axes are adjusted is proportional to the confidence degree. An experimental study is conducted on real data which shows that the proposed method significantly outperforms the standard utility elicitation method in terms of precision of the recommendation set.
A Regression Dependent Iterative Algorithm for Optimizing Top-K Selection in Simulation Query Language
International Journal of Decision Support System Technology, Jul 1, 2012
In this paper the authors propose an extension of the algorithm General Optimal Regression Budget... more In this paper the authors propose an extension of the algorithm General Optimal Regression Budget Allocation ScHeme (GORBASH) for iteratively optimizing simulation budget allocation while minimizing the total processing cost for top-k queries. They also implement this algorithm as part of SimQL: an extension of SQL that includes probability functions expressed through stochastic simulation.
Battle Management System (BMS): An Optimization for Military Decision Makers
ABSTRACT Military commanders are continuously faced with the problem of choosing between multiple... more ABSTRACT Military commanders are continuously faced with the problem of choosing between multiple courses of action (COA) based on current intelligence. The Battle Management System (BMS) is an event-based simulation that optimizes each Blue force move based on the anticipated loss by the Red force from that move, tempered by the anticipated loss to Blue's own force. Specifically, the Blue (Offensive) commander must choose to send forces along alternative routes to the objective given his intelligence about Red (Defense) force disposition. Each combination of forces positioned on East and West routes is referred to as a Course of Action (COA). He then has opportunities to engage or continue to advance without engagement at several event points along the way which occur when his sensors and weapons come into range of Red. Similarly, Red can choose to engage or not engage each time the advancing Blue force enters an engagement envelop for a different weapon system. Blue advances until the forces are in physical contact (Range = 0).
Modeling and Optimization of Virtual Networks in Multi-AS Environment
Journal of Decision Systems, Apr 3, 2015
Smart manufacturing (SM) systems have to optimise manufacturing activities at the machine, unit o... more Smart manufacturing (SM) systems have to optimise manufacturing activities at the machine, unit or entire manufacturing facility level as well as adapting the manufacturing process on the fly as required by a variety of conditions (e.g. machine breakdowns and/or slowdowns) and unexpected variations in demands. This paper provides a framework for autonomic smart manufacturing (ASM) that relies on a variety of models for its support: (1) a process model to represent machines, parst inventories and the flow of parts through machines in a discrete manufacturing floor; (2) a predictive queueing network model to support the analysis and planning phases of ASM; and (3) optimisation models to support the planning phase of ASM. In essence, ASM is an integrated decision support system for smart manufacturing that combines models of different nature in a seamless manner. As shown here, these models can be used to predict manufacturing time and the energy consumed by the manufacturing process, as well as for finding the machine settings that minimise the energy consumed or the manufacturing time subject to a variety of constraints using non-linear optimisation models. The diversity of models used affords different levels of integration and granularity in the decision-making process. An example of a car manufacturing process is used throughout the paper to explain the concepts and models introduced here.
Proceedings of the 23rd International Conference on Enterprise Information Systems
A major deficiency in the manufacturing ecosystem today is the lack of cloud-based infrastructure... more A major deficiency in the manufacturing ecosystem today is the lack of cloud-based infrastructure that supports the combined decision making and optimization of product design, process design, and supply chain, as opposed to hard wired solutions within silos today. The reported work makes a step toward bridging this deficiency by developing a software framework, prototype and a case study for SPOT-a decision guidance system for simultaneous optimization and trade-off analysis of combined service and product networks, capable to express the combined product, process and supply chain design. SPOT allows users to express, as data input, a hierarchical assembly and composition virtual products and services, i.e., having fixed and control parameters that can be optimized. Virtual services produce a flow of virtual products, such as raw materials, parts of finished products. Like the virtual services, they are associated with analytic models that express customerfacing performance metrics and feasibility constraints, which are used for optimization. The uniqueness of our approach in SPOT is the use of modular simulation-like model for product and service networks, yet optimization quality and computational time of the best available mathematical programming solvers, which is achieved by symbolic computation of simulation code to generate lower-level mathematical programming models.
Proceedings of the 10th International Conference on Operations Research and Enterprise Systems
National and local economies are strongly dependent on stable power systems. While the problem of... more National and local economies are strongly dependent on stable power systems. While the problem of power system resilience in the face of natural disasters and terrorist attacks has been extensively studied from the systems engineering perspective, a major unsolved problem remains in the need for preventive solutions against the collapse of power systems. These solutions must ensure the most economically efficient operation of power systems, within the bounds of any remaining power capacity. Transferring power usage rights from the lowest-loss to the highest-loss entities would result in significant reduction of the combined loss. The existing power systems do not take this fact into account. To address this need, we envision a paradigm shift toward three-step system for (1) a cooperation power market where power usage rights can be transferred among participating entities, (2) decision guidance to recommend market asks and bids to each entity, and (3) optimization that, given the market clearance, will recommend precise operational controls for each entity's microgrid. The key challenge to address is the design of this three-step market system that will guarantee important properties including Pareto-optimality, individual rationality, and fairness, as well as privacy, security, pseudo-anonymity and non-repudiation.
Proceedings of the 22nd International Conference on Enterprise Information Systems, 2020
Current approaches to infrastructure investment either (1) model the problem in high-level financ... more Current approaches to infrastructure investment either (1) model the problem in high-level financial terms, but do not accurately express the underlying system behavior and non-financial performance indicators, or (2) are hard-wired to infrastructure silos, and do not take into account the complex interaction across these silos. This paper proposes to bridge the gap by modeling interrelated infrastructures as a hierarchical service network operating over a time horizon, as well as an extensible repository of infrastructure-specific component models. The paper reports on formal modeling, the development and an initial experimental study of InfraSmart, a decision guidance system for investment in interdependent infrastructure service networks. An initial step in this direction was made in developing a general financial optimization model by 370
2017 IEEE International Conference on Big Data (Big Data), 2017
In this paper we report on the development of a software framework and system for composition, op... more In this paper we report on the development of a software framework and system for composition, optimization and trade-off analysis of manufacturing and contract service networks based on a reusable repository of performance models. Performance models formally describe process feasibility constraints and metrics of interest, such as cost, throughput and CO2 emissions, as a function of fixed and control parameters, such as equipment and contract properties and settings. The repository contains performance models for (1) unit manufacturing processes, (2) base contract services, and (3) a composite steady-state service network. The proposed framework allows process engineers to (1) hierarchically compose model instances of service networks, which can represent production cells, lines, factory facilities and supply chains, and (2) perform deterministic optimization based on mathematical programming and Pareto-optimal trade-off analysis. We case study the framework on a service network for a heat sink product which involves contract vendors and manufacturers, unit manufacturing process services including cutting/shearing and Computer Numerical Control (CNC) machining with milling and drilling steps, quality inspection, finishing and assembly.
Proceedings of the 10th International Conference on Operations Research and Enterprise Systems, 2021
This paper focuses on making optimal investment and operational recommendations for a Hybrid Rene... more This paper focuses on making optimal investment and operational recommendations for a Hybrid Renewable Energy System (HRES). For this purpose we develop a modular composite analytic performance model for HRES investment, which is based on an extensible library of atomic component models, including renewable sources such as solar and wind, power storage, power contracts, and programmable customer loads' switches. The performance model formally expresses feasibility constraints and key performance indicators, including total tost of ownership, environment impact, and infrastructure resilience, as a function of investment and operational decision variables. Based on the performance model, we design and develop a decision guidance system to enable actionable investment recommendations that optimize key performance indicators subject to the operational constraints associated with the network. Finally, we demonstrate the model in a case study based on a real world example for a municipal electric utility.
Uploads
Papers by Alexander Brodsky