Research Interests

  • Power & Energy Systems: Renewable energy grid integration, power system operations, microgrids, electricity market, distributed energy resources, grid resilience

  • Renewable Energy: Renewable and load forecasting, wind and solar resource assessment, wind farm design, mobile marine energy

  • Complex Networks: Multi-layer networks, power network, transportation network, network resilience

  • Energy Management: Integrated energy systems, transactive energy, building-to-grid, vehicle-to-grid, energy storage, battery power and thermal management

  • Big Data Analytics: Data-driven design, energy analytics, machine learning, deep learning

  • Multidisciplinary Design Optimization: Surrogate modeling, optimization, uncertainty quantification, probabilistic design, sensitivity analysis

  • Complex Engineered Systems: Cyber-physical systems, energy system modeling, design, and optimization

Codes (DOES Lab GitHub)

  • M3 Forecasting: The Machine Learning-based Multi-Model (M3) forecasting framework is able to generate both short-term deterministic and probabilistic forecasts.

  • QLearningForecast: This is a package that implements forecasting model dynamic selection via Q-learning.

  • OpenSolar: OpenSolar is developed to enhance the openness and accessibility of publicly available solar datasets. (Python version)

Datasets

  • DRD (Dallas repeated driving cycle dataset) (Download): A repeated driving cycle dataset generated in the Dallas area, aiming to simulate a daily commuting route and serves as a base for further vehicle energy management study.

  • AMI Load Dataset (Download): This dataset contains UTD campus load data with 13 buildings, together with 20 weather and calendar features, spanning from 01/01/2014 to 12/31/2015 with an hourly resolution.

Research Sponsors

alt text 

Current Research Projects

  • Nuclear-Renewable-Storage Digital Twin: Enhancing Design, Dispatch, and Cyber Response of Integrated Energy Systems (2021 - 2024, funded by INL LDRD/DOE)

alt text 

As high-fidelity physics model can be used to inform cost-effective and reliable operations of integrated energy systems (IES), their highly complex and non-linear natures often make it computationally prohibitive for real-time implementation. The overarching objective of this project is to develop a set of deep reinforcement learning (DRL)-based strategies to enable real-time decision-making of physics-informed IES operation. We will develop an integrated IES representation using existing high-fidelity physics models developed by INL and apply it to a Nuclear-Renewable-Storage Integrated Energy Systems (N-R-S IES).

  • Learning on Graphs for Resilience Decision-Support in Real-World Networks (2021 - 2025, funded by DOD ONR)

alt text 

This project is to develop a unified “learning on graphs” framework that enables generation of robust, scalable, and generalizable combinatorial policies over graphs, to support real-time decision-making for recovery of disrupted real-world networks. The following integrative research activities will be performed: 1) abstraction of topological descriptors of graphs; 2) integrated Artificial Intelligence (AI) framework with graph convolutional networks, reinforcement learning and topological inputs/outputs, to generate network reconfiguration and restoration policies; and 3) training and evaluation setup/process on grid resiliency problems.

  • Deep Learning-based Reliability and Resilience Enhancement of Future Navy Ships and Their Integration into Power Networks under Extreme Event (2020 - 2023, funded by DOD ONR YIP), UTD News

alt text 

This project is to (i) study the energy management and fault detection of the Navy power ship systems, by utilizing deep learning-based techniques to track the demand changes with real-time interactions, and (ii) study the future potential Navy power ship systems and their interactions with typical coastal city electric network, to enhance the reliability and resilience of both the Navy ships and the coastline power grids in the presence of natural hazards, cyber/physical attacks, or any possible contingency.

Papers:

  1. *Dabbaghjamanesh, M., *Senemmar, S. and Zhang, J., Resilient Distribution Networks Considering Mobile Marine Microgrids: A Synergistic Network Approach, IEEE Transactions on Industrial Informatics, Vol. 17, Issue 8, 2021, pp. 5742-5750. (PDF)

  2. *Jacob, R. A., *Senemar, S. and Zhang, J., Fault Diagnostics in Shipboard Power Systems using Graph Neural Networks, IEEE International Symposium on Diagnostics for Electric Machines, Power Electronics and Drives, Virtual, August 22-25, 2021.

  3. *Senemar, S. and Zhang, J., Deep Learning-based Fault Detection, Classification, and Locating in Shipboard Power Systems, IEEE Electric Ship Technologies Symposium, Virtual, August 3-6, 2021.

  • Co-located Wind Farm and Hydrogen Plant Energy System Study (2021 - 2022, funded by WindSTAR (NSF IUCRC))

alt text 

This project will perform a planning & operational study on a co-located wind farm and hydrogen energy system. Integrated energy systems (IES) of this genre need to be designed considering both the financial and technical constraints to demonstrate their feasibility rather than relying on additional incentives or favoring policies. REopt, an open source tool developed by NREL, will be adopted and improved to perform the study.

  • Machine Learning-based Real-time Optimal Switchings of Reconfigurable Microgrids (2021 - 2023, funded by U.S. Army ERDC)

alt text 

This project is to explore how machine learning could be leveraged to solve the optimal switching of reconfigurable microgrids in real time under both normal and extreme operations (e.g., grid contingencies, natural disasters, and cyber/physical attacks). A learn-to-reconfiguration (L2R) methodology will be developed to learn the relationship between microgrid state and optimal reconfiguration decisions with an end-to-end manner based on state-of-the-art deep learning models.

Papers:

  1. *Dabbaghjamanesh, M. and Zhang, J., Deep Learning-based Real-time Switching of Reconfigurable Microgrids, The Eleventh Conference on Innovative Smart Grid Technologies (ISGT 2020), Washington D.C., February 17-20, 2020.

  • Solar Uncertainty Management and Mitigation for Exceptional Reliability in Grid Operations (2018 - 2021, funded by DOE EERE)

alt text 

The project is designing novel algorithms to create probabilistic solar power forecasts and automate their integration into power system operations. Adaptive reserves will dynamically adjust reserve levels conditional on meteorological and power system states. Risk-parity dispatch will be developed to produce optimal dispatch strategies by cost-weighting solar generation scenarios on forecast uncertainty. This project will test the integration of probabilistic solar forecasts into the Electric Reliability Council of Texas’ real-time operation environment through automated reserve and dispatch tools that can increase economic efficiency and improve system reliability.

Papers:

  1. *Li, B., *Feng, C. and Zhang, J., Multi-Timescale Simulation of Non-Spinning Reserve in Wholesale Electricity Markets, 2021 IEEE Green Technologies Conference (GreenTech), Virtual, April 7-9, 2021.

  2. *Feng, C., *Sun, M., Zhang, J., Doubleday, K., Hodge, B.-M. and Du, P., A Data-driven Method for Adaptive Reserve Requirements Estimation via Probabilistic Net Load Forecasting, IEEE Power & Energy Society General Meeting, Montreal, Canada, August 2-6, 2020.

  3. *Feng, C., Yang, D., Hodge, B.-M. and Zhang, J., OpenSolar: Promoting the Openness and Accessibility of Diverse Public Solar Datasets, Solar Energy, Vol. 188, 2019, pp. 1369-1379. (PDF)

  • Coordinated Ramping Product and Regulation Reserve Procurements in California Independent System Operator and Midcontinent Independent System Operator Using Multi-Scale Probabilistic Solar Power Forecasts (2018 - 2021, funded by DOE EERE)

alt text 

This project is advancing the state-of-the-art in solar forecasting technologies by developing short-term and day-ahead probabilistic solar power prediction capabilities. The proposed technology will be based on the big-data-driven, transformative IBM Watt-Sun platform, which will be driven by parallel computation-based scalable and fast data curation technology and multi-expert machine learning based model blending. The integration of validated probabilistic solar forecasts into the scheduling operations of both the Midcontinent and California Independent System Operators will be tested, via efficient and dynamic procurement of ramp product and regulation. Integration of advanced visualization of ramping events and associated alerts into their energy management systems and control room operations will also be researched and validated.

Papers:

  1. *Li, B. and Zhang, J., A Review on the Integration of Probabilistic Solar Forecasting in Power Systems, Solar Energy, Vol. 210, 2020, pp. 68-86. (Invited Paper) (PDF)

  2. *Li, B., Zhang, J. and Hobbs, B., A Copula Enhanced Convolution for Uncertainty Aggregation, The Eleventh Conference on Innovative Smart Grid Technologies (ISGT 2020), Washington D.C., February 17-20, 2020.

  3. *Cui, M. and Zhang, J., Estimating Ramping Requirements With Solar-Friendly Flexible Ramping Product in Multi-Timescale Power System Operations, Applied Energy, Vol. 225, 2018, pp. 27-41. (PDF)

alt text 

Nuclear and renewable energy sources are important to consider in the U.S. economy's evolution because both are clean, non-carbon-emitting energy sources. Advanced nuclear-renewable hybrid energy systems (N-R HESs) composed of nuclear and renewable energy sources, industrial energy users, and energy storage systems are being evaluated for their economic benefit and technical feasibility. N-R HESs have been proposed as a technology that can generate very low-carbon, dispatchable electricity and provide very low-carbon energy to industry at a lower cost than many other options. Beyond classic energy-shifting services, N-R HESs may be able to provide a suite of services at finer time-scales to promote a safer and more reliable integration of renewable energy resources. The overarching objective of this project is to develop a multi-timescale N-R HESs operations framework to provide different types of grid products.

Papers:

  1. Li, D., Wu, J., Zhang, J. and Wang, P., Co-Design Optimization of a Combined Heat and Power Hybrid Energy System, ASME 2021 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference, Paper No. IDETC2021-71304, Virtual, August 17-20, 2021.

  2. *Rahman, J. and Zhang, J., Optimization of Nuclear-Renewable Hybrid Energy System Operation in Forward Electricity Market, 2021 IEEE Green Technologies Conference (GreenTech), Virtual, April 7-9, 2021.

  • Battery Thermal Management Systems for Electric Vehicles

alt text 

Thermal management plays a significant role in the life, performance, safety, and cost of the lithium-ion battery modules in electric vehicles. Battery thermal management systems (BTMS) aim to improve the temperature uniformity among all battery cells, and to prevent battery cells from very high temperature which may likely cause their explosion. This project is to develop a hybrid approach by combining flat heat pipe, phase change material, and air cooling through modeling, experiments, and optimization.

Papers:

  1. *Liu, Y. and Zhang, J., Electric Vehicle Battery Thermal and Cabin Climate Management Based on Model Predictive Control, Journal of Mechanical Design, Vol. 143, Issue 3, 2021, pp. 031705. (PDF)

  2. *Liu, Y. and Zhang, J., Self-adapting J-type Air-based Battery Thermal Management System via Model Predictive Control, Applied Energy, Vol. 263, 2020, pp. 114640. (PDF)

  3. *Liu, Y. and Zhang, J., Design A J-type Air-based Battery Thermal Management System through Surrogate-based Optimization, Applied Energy, Vol. 252, 2019, pp. 113426. (PDF)

  4. *Wang, X., *Liu, Y., Sun, W., Song, X. and Zhang, J., Multidisciplinary and Multifidelity Design Optimization of Electric Vehicle Battery Thermal Management System, Journal of Mechanical Design, Vol. 140, Issue 9, 2018, pp. 094501 (8 pages). (PDF)

  5. *Wang, X., *Li, M., *Liu, Y., Sun, W., Song, X. and Zhang, J., Surrogate based Multidisciplinary Design Optimization of Lithium-ion Battery Thermal Management System in Electric Vehicles, Structural and Multidisciplinary Optimization, Vol. 56, Issue 6, 2017, pp. 1555-1570. (PDF)

Past Research Projects

  • A Hybrid Approach To SCOPF Using Cross-Entropy (2018 - 2020, funded by DOE ARPA-E)

alt text 

This project will develop an innovative hybrid optimization method to solve the Security-Constrained Optimal Power Flow (SCOPF) problem formulated under the Challenge 1 of the ARPA-E Grid Optimization (GO) Competition. The base of the proposed optimization method will rely on a combination of formal mathematical optimization, metaheuristics, and the cross- entropy method, an approach that has proven to be successful in solving this kind of optimization problem, recently winning the SCOPF competition organized by the IEEE Working Group on Modern Heuristic Optimization (WGMHO).

Papers:

  1. *Rahman, J., *Feng, C. and Zhang, J., A Learning-Augmented Approach for AC Optimal Power Flow, International Journal of Electrical Power and Energy Systems, Vol. 130, 2021, pp. 106908. (PDF)

  2. *Rahman, J., *Feng, C. and Zhang, J., Machine Learning-Aided Security Constrained Optimal Power Flow, IEEE Power & Energy Society General Meeting, Montreal, Canada, August 2-6, 2020.

  • Providing Ramping Service with Wind to Enhance Power System Operational Flexibility (2016 - 2019, funded by DOE EERE)

alt text 

With increasing wind power penetration in the electricity grid, system operators are recognizing the need for additional flexibility, and some are implementing new ramping products as a type of ancillary service. However, wind is generally thought of as causing the need for ramping services, not as being a potential source for the service. In this project, a multi- timescale unit commitment and economic dispatch model is developed to consider the wind power ramping product (WPRP). Designed as positive characteristics of WPRs, the WPRP is then integrated into the multi-timescale dispatch model that considers new objective functions, ramping capacity limits, active power limits, and flexible ramping requirements.

Papers:

  1. *Li, B., Sedzro, K., Fang, X., Hodge, B.-M. and Zhang, J., A Clustering-Based Scenario Generation Framework for Power Market Simulation with Wind Integration, Journal of Renewable and Sustainable Energy, Vol. 12, 2020, pp. 036301. (PDF)

  2. *Cui, M., Krishnan, V., Hodge, B.-M. and Zhang, J., A Copula-Based Conditional Probabilistic Forecast Model for Wind Power Ramps, IEEE Transactions on Smart Grid, Vol. 10, Issue 4, 2019, pp. 3870-3882. (PDF)

  3. *Cui, M., Zhang, J., Wang, Q., Krishnan, V. and Hodge, B.-M., A Data-Driven Methodology for Probabilistic Wind Power Ramp Forecasting, IEEE Transactions on Smart Grid, Vol. 10, Issue 2, 2019, pp. 1326-1338. (PDF)

  4. *Cui, M., *Feng, C., *Wang, Z. and Zhang, J., Statistical Representation of Wind Power Ramps Using a Generalized Gaussian Mixture Model, IEEE Transactions on Sustainable Energy, Vol. 9, Issue 1, 2018, pp. 261-272. (PDF)

  5. *Cui, M., Zhang, J., Wu, H. and Hodge, B.-M., Wind-Friendly Flexible Ramping Product Design in Multi-Timescale Power System Operations, IEEE Transactions on Sustainable Energy, Vol. 8, Issue 3, 2017, pp. 1064-1075. (PDF)

  • WindView: An Open Platform for Wind Energy Forecast Visualization (2016 - 2019, funded by DOE EERE)

alt text 

This project is to create an open situational awareness and decision support platform “WindView”, that provides grid operators with knowledge on the state and performance of their power system, with an emphasis on wind energy. The focus will be on utilizing advanced visualization to display pertinent information, extracted through computational techniques, from wind power forecasts for high-wind penetration systems

Papers:

  1. *Sun, M., *Feng, C. and Zhang, J., Conditional Aggregated Probabilistic Wind Power Forecasting Based on Spatio-temporal Correlation, Applied Energy, Vol. 256, 2019, pp. 113842. (PDF)

  2. *Sun, M., *Feng, C., Chartan, E., Hodge, B.-M. and Zhang, J., A Two-Step Short-Term Probabilistic Wind Forecasting Methodology Based on Predictive Distribution Optimization, Applied Energy, Vol. 238, 2019, pp. 1497-1505. (PDF)

  3. *Feng, C., *Sun, M., *Cui, M., Chartan, E., Hodge, B.-M. and Zhang, J., Characterizing Forecastability of Wind Sites in the United States, Renewable Energy, Vol 133, 2019, pp. 1352-1365. (PDF)

  4. *Feng, C., *Cui, M., Hodge, B.-M. and Zhang, J., A Data-Driven Multi-Model Methodology with Deep Feature Selection for Short-Term Wind Forecasting, Applied Energy, Vol. 190, 2017, pp. 1245-1257. (PDF)

  5. *Feng, C., Chartan, E., Hodge, B.-M. and Zhang, J., Characterizing Time Series Data Diversity for Wind Forecasting, The 4th IEEE/ACM International Conference on Big Data Computing, Applications and Technologies (BDCAT 2017), Austin, TX, December 5-8, 2017. (Best Student Paper Award)

  • Data-Driven Hierarchical Load Forecasting with Distributed Energy Resources (funded by Oncor)

alt text 

This project is developing load forecasting models (especially under extreme weather conditions) of different forecasting horizons at different levels of aggregation, by taking account of behind-the-meter (BTM) PV.

Papers:

  1. *Feng, C. and Zhang, J., Assessment of Aggregation Strategies for Machine Learning based Short-Term Load Forecasting, Electric Power Systems Research, Vol. 184, 2020, pp. 106304. (PDF)

  2. *Sun, M., *Feng, C. and Zhang, J., Factoring Behind-the-Meter Solar into Load Forecasting: Case Studies under Extreme Weather, The Eleventh Conference on Innovative Smart Grid Technologies (ISGT 2020), Washington D.C., February 17-20, 2020.

  3. *Feng, C. and Zhang, J., Short-Term Load Forecasting With Different Aggregation Strategies, ASME 2018 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference, Paper No. DETC2018-86084, Quebec City, Canada, August 26-29, 2018.

  • Stochastic Security-Constrained Unit Commitment and Economic Dispatch (funded by DOE EERE)

alt text 

A stochastic unit commitment and economic dispatch model that considers stochastic variables at multiple operational timescales is developed. It involves four distinct stages: stochastic day-ahead security-constrained unit commitment, stochastic real-time security-constrained unit commitment, stochastic real-time security-constrained economic dispatch, and deterministic automatic generation control.

Papers:

  1. Wu, H., Krad, I., Florita, A., Hodge, B.-M., Ibanez, E., Zhang, J. and Ela, E., Stochastic Multi-timescale Power System Operations with Variable Wind Generation, IEEE Transactions on Power Systems, Vol. 32, Issue 5, 2017, pp. 3325-3337. (PDF)

  • Ramp Forecasting and Detection in Wind Power, Solar Power, and Netload

alt text 

An optimized swinging door algorithm (OpSDA) is developed to improve the state of the art in ramping detection. The swinging door algorithm (SDA) is utilized to segregate power data through a piecewise linear approximation. A dynamic programming algorithm is performed to merge adjacent segments with the same ramp changing direction.

Papers:

  1. *Cui, M., Zhang, J., *Feng, C., Florita, A., Sun, Y. and Hodge, B.-M., Characterizing and Analyzing Ramping Events in Wind Power, Solar Power, Load, and Netload, Renewable Energy, Vol. 111, 2017, pp. 227-244. (PDF)

  2. *Cui, M., Zhang, J., Florita, A., Hodge, B.-M., Ke, D. and Sun, Y., An Optimized Swinging Door Algorithm for Identifying Wind Ramping Events, IEEE Transactions on Sustainable Energy, Vol. 7, Issue 1, 2016, pp. 150-162. (PDF)

  3. *Cui, M., Ke, D., Sun, Y., Gan, D., Zhang, J. and Hodge, B.-M., Wind Power Ramp Event Forecasting Using a Stochastic Scenario Generation Method, IEEE Transactions on Sustainable Energy, Vol. 6, Issue 2, 2015, pp. 422-433. (PDF)

  • Watt-sun: A Multi-scale, Multi-Model, Machine-Learning Solar Forecasting Technology

alt text 

Solar forecasting is a challenging task, since solar power generation is present in both the transmission and the distribution sides of the grid. A multi-scale, multi-model, machine-learning solar forecasting technology is in development. This project is (1) developing a scalable, cost-effective technology platform for improved solar forecasting by using big-data information processing technologies; and (2) developing a suite of generally applicable, value-based metrics for deterministic and probabilistic solar forecasting for a comprehensive set of scenarios (different time horizons, applications, etc.), which can assess the economic and reliability impact of improved solar forecasting.

Papers:

  1. *Cui, M., Zhang, J., Hodge, B.-M., Lu, S. and Hamann, H. F., A Methodology for Quantifying Reliability Benefits from Improved Solar Power Forecasting in Multi-Timescale Power System Operations, IEEE Transactions on Smart Grid, Vol. 9, Issue 6, 2018, pp. 6897-6908. (PDF)

  2. Zhang, J., Hodge, B.-M., Lu, S., Hamann, H. F., Lehman, B., Simmons, J., Campos, E., Banunarayanan,V., Black, J. and Tedesco, J., Baseline and Target Values for Regional and Point PV Power Forecasts: Toward Improved Solar Forecasting, Solar Energy, Vol. 122, 2015, pp. 804-819. (PDF)

  3. Zhang, J., Florita, A., Hodge, B.-M., Lu, S., Hamann, H. F., Banunarayanan, V. and Brockway, A., A Suite of Metrics for Assessing the Performance of Solar Power Forecasting, Solar Energy, Vol. 111, 2015, pp. 157-175. (PDF)

  • Wind Forecast Improvement Project (WFIP)

alt text 

Short-term wind power forecasting plays an important role in grid operations for balancing supply and demand in the electric power system. The improved forecasting system consists of an ensemble of high-resolution rapidly updated numerical weather prediction models. Statistical analysis results show that the improved forecasting systems is more accurate than the current short-term wind power forecasts used by ERCOT. The economic analysis of using the improved forecasts are quantified, showing reductions in both unit commitment costs and reserve requirement costs.

Papers:

  1. Zhang, J., *Cui, M., Hodge, B.-M., Florita, A. and Freedman, J., Ramp Forecasting Performance from Improved Short-Term Wind Power Forecasting Over Multiple Spatial and Temporal Scales, Energy, Vol. 122, 2017, pp. 528-541. (PDF)

  2. Freedman, J. et al., The Wind Forecast Improvement Project (WFIP): A Public/Private Partnership for Improving Short Term Wind Energy Forecasts and Quantifying the Benefits of Utility Operations - the Southern Study Area, DOE EERE Wind Energy Technologies Office, April 2014.

  • Wind Farm Design and Optimization

alt text 

A new methodology, the Unrestricted Wind Farm Layout Optimization (UWFLO), that addresses critical aspects of optimal wind farm planning is developed. This methodology simultaneously determines the optimum farm layout and the appropriate selection of turbines (in terms of their rotor diameters) that maximizes the net power generation. A standard analytical wake model has been used to account for the velocity deficits in the wakes created by individual turbines. The wind farm power generation model is validated against data from a wind tunnel experiment on a scaled down wind farm. The complex nonlinear optimization problem presented by the wind farm model is effectively solved using constrained Particle Swarm Optimization (PSO).

Papers:

  1. Chowdhury, S., Zhang, J., Messac, A. and Castillo, L., Optimizing the Arrangement and the Selection of Turbines for Wind Farms Subject to Varying Wind Conditions, Renewable Energy, Vol. 52, 2013, pp. 273-282. (PDF)

  2. Zhang, J., Chowdhury, S., Messac, A. and Castillo, L., A Response Surface-Based Cost Model for Wind Farm Design, Energy Policy, Vol. 42, 2012, pp. 538-550. (PDF)

  3. Chowdhury, S., Zhang, J., Messac, A. and Castillo, L., Unrestricted Wind Farm Layout Optimization (UWFLO): Investigating Key Factors Influencing the Maximum Power Generation, Renewable Energy, Vol. 38, Issue 1, 2012, pp. 16-30. (PDF) (Best Paper Award)

  4. Messac, A., Chowdhury, S. and Zhang, J., Characterizing and Mitigating the Wind Resource-based Uncertainty in Farm Performance, Journal of Turbulence (special issue on Turbulence and Wind Energy), Vol. 13, Issue 13, 2012, pp. 1-26. (PDF)

  • Wind Resource Assessment

alt text 

A smooth multivariate wind distribution model is developed to capture the coupled variation of wind speed, wind direction, and air density. The Multivariate and Multimodal Wind distribution (MMWD) model is successfully applied to model (i) the distribution of wind speed (univariate); (ii) the distribution of wind speed and wind direction (bivariate); and (iii) the distribution of wind speed, wind direction, and air density (multivariate).

Papers:

  1. Zhang, J., Chowdhury, S., Messac, A. and Hodge, B.-M., A Hybrid Measure-Correlate-Predict Method for Long-Term Wind Condition Assessment, Energy Conversion and Management, Vol. 87, 2014, pp. 697-710. (PDF)

  2. Zhang, J., Chowdhury, S. and Messac, A., A Comprehensive Measure of the Energy Resource: Wind Power Potential (WPP), Energy Conversion and Management, Vol. 86, 2014, pp. 388-398. (PDF)

  3. Zhang, J., Chowdhury, S., Messac, A. and Castillo, L., A Multivariate and Multimodal Wind Distribution Model, Renewable Energy, Vol. 51, 2013, pp. 436-447. (PDF)

  • Surrogate Modeling and Applications to Complex System Design

alt text 

We developed a high fidelity surrogate modeling technique that we call the Adaptive Hybrid Functions (AHF). The AHF formulates a reliable Crowding Distance-Based Trust Region (CD-TR), and adaptively combines the favorable characteristics of different surrogate models. The weight of each contributing surrogate model is determined based on the local measure of accuracy for that surrogate model in the pertinent trust region. Such an approach is intended to exploit the advantages of each component surrogate.

Papers:

  1. Zhang, J., Chowdhury, S., Messac, A., Zhang, J.Q. and Castillo, L., Adaptive Hybrid Surrogate Modeling for Complex Systems, AIAA Journal, Vol. 51, Issue 3, 2013, pp. 643-656. (PDF)

  2. Zhang, J., Chowdhury, S. and Messac, A., An Adaptive Hybrid Surrogate Model, Structural and Multidisciplinary Optimization, Vol. 46, Issue 2, 2012, pp. 223-238. (PDF)

  • Domain Segmentation based on Uncertainty in the Surrogate (DSUS)

alt text 

In the practice of engineering, where predictive models are pervasively used, the knowledge of the level of modeling error at any region inside the design space is uniquely helpful for design exploration and model improvement. The lack of methods that can explore the spatial variation of surrogate error levels in a wide variety of surrogates (i.e., model-independent methods) leaves an important gap in our capacity of design domain exploration. We develop a novel framework, called Domain Segmentation based on Uncertainty in the Surrogate (DSUS), to segregate the design domain based on the level of local errors.

Papers:

  1. Zhang, J., Chowdhury, S., Mehmani, A. and Messac, A., Characterizing Uncertainty Attributable to Surrogate Models, Journal of Mechanical Design, Vol. 136, Issue 3, 2014, pp. 031004. (PDF)

  • Mixed-Discrete Particle Swarm Optimization

alt text 

A modification of the Particle Swarm Optimization (PSO) algorithm was developed, which can adequately address system constraints while dealing with mixed-discrete variables. Continuous search (particle motion), as in conventional PSO, was implemented as the primary search strategy; subsequently, the discrete variables were updated using a deterministic nearest-feasible-vertex criterion. This approach is expected to alleviate the undesirable difference in the rates of evolution of discrete and continuous variables.

Papers:

  1. Chowdhury, S., Tong, W., Messac, A. and Zhang, J., A Mixed-Discrete Particle Swarm Optimization Algorithm with Explicit Diversity-Preservation, Structural and Multidisciplinary Optimization, Vol. 47, Issue 3, 2013, pp. 367-388. (PDF)