Automated extraction of physical parameters from experimentally obtained thermal profiles using a machine learning approach
Explore the source record for details and available documents.
99 GENERAL AND MISCELLANEOUS↗
SEARCH · Search NASA
Results for “multivariate experimental data”
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Artificial Neural Network Representation of Dynamic Criticality Excursion Experiment Data
This paper outlines time dependent considerations in applying the artificial neural network (ANN) paradigm to dynamic nuclear criticality excursion experiment data. The ANN amongst a variety of uses, can apply to function fitting of multiple input-multiple output data sets, pattern recognition, cluster analysis, and time series prediction. The ANN is a component of Machine Learning (ML) data analysis. The practical application of ML to criticality safety has been funded through the Y-12 Plant Directed Research and Demonstration (PDRD) program to support criticality safety and emergency planning efforts. The initial effort considered function fitting of criticality excursion data. Recent efforts successfully extended the ANN to criticality safety code validation. The initial ANN work applied to criticality excursions centered around multivariate function fitting of static (tabular) data taken from the French Consequences Radiologiques d'un Accident de Criticite (CRAC) and SILENE reactor campaigns. The former report (LeCorche) summarized important tabular data for the first 44 CRAC experiments. The initial ANN was trained to specific time-independent inputs, then applied to predict peak fission rate and total fissions for a wide variety of fissile solutions process conditions. This work extends that effort to apply the ANN paradigm to dynamic prediction of time-dependent criticality excursion experiment data. A dynamic criticality excursion in fissile solution represents a complex multiphysics phenomenon. Factors include among other things, coupled neutronic, multiphase fluid flow, temperature, pressure, acoustic shock, boiling heat transfer, reactivity feedback, and radiolytic gas considerations. There have been many efforts to model (over time) as coupled ordinary and (space) as partial differential equations (ODE, PDE). The ANN representation noted here does not rely on solving explicit ODE or PDEs. Rather the intent is to demonstrate that as a time series, the criticality excursion can be predicted by training specific ANN architectures to experimental input and output time series exclusively. The rationale for extending this work to the time-dependent regime is important in developing a 'criticality excursion emulator' solely on the basis of learning characteristics from experiment dynamic data. An emulator is different than a computer code simulator driven by the solution of ODEs and PDEs given select 'tuning parameters' that match outputs to experiments. A criticality excursion emulator, derived exclusively from an ANN or ML, could predict most likely excursion characteristics, trained from historic dynamic experiment data, presented with input values within the range of those experiments. The work noted shows that generic time-series ANN architectures can be satisfactorily applied to the prediction of dynamic, extremely chaotic criticality excursion power given an initial time dependent input series. In reality, an excursion emulator could be driven by user-supplied initial (t=0), time independent or constant values to generate time dependent outputs. Thus ongoing work includes developing a custom ANN such that initial dynamic conditions would be predicted. Those results then serve as inputs to a customized ANN architecture currently in development. (author)
42 ENGINEERING↗
Interactive XCMS Online: Simplifying Advanced Metabolomic Data Processing and Subsequent Statistical Analyses
XCMS Online (xcmsonline.scripps.edu) is a cloud-based informatic platform designed to process and visualize mass-spectrometry-based, untargeted metabolomic data. Initially, the platform was developed for two-group comparisons to match the independent, “control” versus “disease” experimental design. Here, we introduce an enhanced XCMS Online interface that enables users to perform dependent (paired) two-group comparisons, meta-analysis, and multigroup comparisons, with comprehensive statistical output and interactive visualization tools. Newly incorporated statistical tests cover a wide array of univariate analyses. Multigroup comparison allows for the identification of differentially expressed metabolite features across multiple classes of data while higher order meta-analysis facilitates the identification of shared metabolic patterns across multiple two-group comparisons. Given the complexity of these data sets, we have developed an interactive platform where users can monitor the statistical output of univariate (cloud plots) and multivariate (PCA plots) data analysis in real time by adjusting the threshold and range of various parameters. On the interactive cloud plot, metabolite features can be filtered out by their significance level (p-value), fold change, mass-to-charge ratio, retention time, and intensity. The variation pattern of each feature can be visualized on both extracted-ion chromatograms and box plots. The interactive principal component analysis includes scores, loadings, and scree plots that can be adjusted depending on scaling criteria. The utility of XCMS functionalities is demonstrated through the metabolomic analysis of bacterial stress response and the comparison of lymphoblastic leukemia cell lines.
37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
Multivariate network traffic analysis using clustered patterns
Traffic analysis is a core element in network operations and management for various purposes including change detection, traffic prediction, and anomaly detection. In this paper, we introduce a new approach to online traffic analysis based on a pattern-based representation for high-level summarization of the traffic measurement data. Unlike the past online analysis techniques limited to a single variable to summarize (e.g., sketch), the focus of this study is on capturing the network state from the multivariate attributes under consideration. To this end, we employ clustering with its benefit of the aggregation of multidimensional variables. The clustered result represents the state of the network with regard to the monitored variables, which can also be compared with the observed patterns from previous time windows enabling intuitive analysis. Finally, we demonstrate the proposed method with two popular use cases, one for estimating state changes and the other for identifying anomalous states, to confirm its feasibility. Our extensive experimental results with public traces and collected monitoring measurements from ESnet traffic traces show that our pattern-based approach is effective for multivariate analysis of online network traffic with visual and quantitative tools.
97 MATHEMATICS AND COMPUTING↗
MetaboDirect: an analytical pipeline for the processing of FT-ICR MS-based metabolomic data
Background: Microbiomes are now recognized as the main drivers of ecosystem function ranging from the oceans and soils to humans and bioreactors. However, a grand challenge in microbiome science is to characterize and quantify the chemical currencies of organic matter (i.e., metabolites) that microbes respond to and alter. Critical to this has been the development of Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS), which has drastically increased molecular characterization of complex organic matter samples, but challenges users with hundreds of millions of data points where readily available, user-friendly, and customizable software tools are lacking. Results: Here, we build on years of analytical experience with diverse sample types to develop MetaboDirect, an open-source, command-line-based pipeline for the analysis (e.g., chemodiversity analysis, multivariate statistics), visualization (e.g., Van Krevelen diagrams, elemental and molecular class composition plots), and presentation of direct injection high-resolution FT-ICR MS data sets after molecular formula assignment has been performed. When compared to other available FT-ICR MS software, MetaboDirect is superior in that it requires a single line of code to launch a fully automated framework for the generation and visualization of a wide range of plots, with minimal coding experience required. Among the tools evaluated, MetaboDirect is also uniquely able to automatically generate biochemical transformation networks (ab initio) based on mass differences (mass difference network-based approach) that provide an experimental assessment of metabolite connections within a given sample or a complex metabolic system, thereby providing important information about the nature of the samples and the set of microbial reactions or pathways that gave rise to them. Finally, for more experienced users, MetaboDirect allows users to customize plots, outputs, and analyses. Conclusion: Application of MetaboDirect to FT-ICR MS-based metabolomic data sets from a marine phage-bacterial infection experiment and a Sphagnum leachate microbiome incubation experiment showcase the exploration capabilities of the pipeline that will enable the research community to evaluate and interpret their data in greater depth and in less time. It will further advance our knowledge of how microbial communities influence and are influenced by the chemical makeup of the surrounding system. The source code and User’s guide of MetaboDirect are freely available through (https://github.com/Coayala/MetaboDirect) and (https://metabodirect.readthedocs.io/en/latest/), respectively.
54 ENVIRONMENTAL SCIENCES↗
MAD: Self-Supervised Masked Anomaly Detection Task for Multivariate Time Series
In this paper, we introduce Masked Anomaly Detection (MAD), a general self-supervised learning task for multivariate time series anomaly detection. With the increasing availability of sensor data from industrial systems, being able to detecting anomalies from streams of multivariate time series data is of significant importance. Given the scarcity of anomalies in real-world applications, the majority of literature has been focusing on modeling normality. The learned normal representations can empower anomaly detection as the model has learned to capture certain key underlying data regularities. A typical formulation is to learn a predictive model, i.e., use a window of time series data to predict future data values. In this paper, we propose an alternative self-supervised learning task. By randomly masking a portion of the inputs and training a model to estimate them using the remaining ones, MAD is an improvement over the traditional left-to-right next step prediction (NSP) task. Our experimental results demonstrate that MAD can achieve better anomaly detection rates over traditional NSP approaches when using exactly the same neural network (NN) base models, and can be modified to run as fast as NSP models during test time on the same hardware, thus making it an ideal upgrade for many existing NSP-based NN anomaly detection models.
97 MATHEMATICS AND COMPUTING↗
Flexible classification with spatial quantile comparison and novel statistical features applied to spent nuclear fuel analysis
Multivariate classification algorithms are a common tool in contemporary data science that have begun to be applied to nuclear forensic analyses. Using a multivariate generalization of quantile–quantile plots for comparing unknown statistical distributions, a new model-free multivariate classifier called the Quantile–Quantile Comparator has been developed and tested on the analysis of simulated irradiated nuclear fuel. A method to estimate the sample-specific probability of having made an incorrect classification decision by analyzing the distribution of quantile comparisons has been developed, and methods have been developed to analyze the partition of the known data library used to analyze unknown data. By analyzing this partition, the analyst may identify classes that are likely to be confused, identify test data that are inconsistent with the data library and return a “None-of-the-Above” classification decision, and optimize the class boundaries. In this paper, the development of these capabilities is described, and a case study using real experimental assay data collected from nuclear fuel extracted from the newly-released In conclusion, Spent Fuel Isotopic Composition Database to demonstrate the efficacy of these capabilities is presented.
11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
A web-based geovisual analytics platform for identifying potential contributors to culvert sedimentation
Sediment accumulation at culverts involves large-scale and interlinked environmental processes that are difficult to address with experimental or physical modeling methods. This article presents an alternative data-driven investigation for shedding insights into these processes. Accordingly, a web-based geovisual analytics application, the IowaDOT platform, was developed, which allows users to explore the complex processes associated with the sediment deposition at culverts. The platform provides systematic procedures for (1) collecting and integrating analytical variables into a single dataset, (2) quantifying the degree of culvert sedimentation using time series of aerial images, (3) identifying drivers that contribute to culvert sedimentation processes from a variety of culvert structural and upstream landscape characteristics using a tree-based feature selection algorithm, and (4) facilitating the understanding of complex spatial and relational patterns of culvert sedimentation processes using multivariate geovisualizations supported by a self-organizing map (SOM). As the outcomes of this study, these patterns identify culvert sedimentation-prone regions in Iowa and quantify empirical relationships between the drivers and culvert sedimentation degrees. Finally, a simple evaluation of the platform was performed to assess the usefulness and user satisfaction of the tool by professional users, and positive feedbacks are received.
54 ENVIRONMENTAL SCIENCES↗
Full data acquisition in Kelvin Probe Force Microscopy: Mapping dynamic electric phenomena in real space
Kelvin probe force microscopy (KPFM) has provided deep insights into the role local electronic, ionic and electrochemical processes play on the global functionality of materials and devices, even down to the atomic scale. Conventional KPFM utilizes heterodyne detection and bias feedback to measure the contact potential difference (CPD) between tip and sample. This measurement paradigm, however, permits only partial recovery of the information encoded in bias- and time-dependent electrostatic interactions between the tip and sample and effectively down-samples the cantilever response to a single measurement of CPD per pixel. This level of detail is insufficient for electroactive materials, devices, or solid-liquid interfaces, where non-linear dielectrics are present or spurious electrostatic events are possible. Here, we simulate and experimentally validate a novel approach for spatially resolved KPFM capable of a full information transfer of the dynamic electric processes occurring between tip and sample. General acquisition mode, or G-Mode, adopts a big data approach utilising high speed detection, compression, and storage of the raw cantilever deflection signal in its entirety at high sampling rates (> 4 MHz), providing a permanent record of the tip trajectory. We develop a range of methodologies for analysing the resultant large multidimensional datasets involving classical, physics-based and information-based approaches. Physics-based analysis of G-Mode KPFM data recovers the parabolic bias dependence of the electrostatic force for each cycle of the excitation voltage, leading to a multidimensional dataset containing spatial and temporal dependence of the CPD and capacitance channels. We use multivariate statistical methods to reduce data volume and separate the complex multidimensional data sets into statistically significant components that can then be mapped onto separate physical mechanisms. Overall, G-Mode KPFM offers a new paradigm to study dynamic electric phenomena in electroactive interfaces as well as offer a promising approach to extend KPFM to solid-liquid interfaces.
46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
Search for Light Pseudoscalar Bosons, Pair-Produced in Higgs Boson Decays in the Four-Electron Final State in Proton-Proton Collisions at $\sqrt{s}=13$ TeV
A search for pairs of light neutral pseudoscalar bosons (𝐴) resulting from the decay of a Higgs boson is performed. The search is conducted using LHC proton-proton collision data at $\sqrt{s}=13$ TeV, collected with the CMS detector in 2016–2018 and corresponding to an integrated luminosity of 138 fb −1 . The 𝐴 boson decays into a highly collimated electron-positron pair. A novel multivariate algorithm using tracks and calorimeter information is developed to identify these distinctive signatures, and events are selected with two such merged electron-positron pairs. No significant excess above the standard model background predictions is observed. Upper limits on the branching fraction for 𝐻 → 𝐴𝐴 → 4𝑒 are set at 95% confidence level, for masses between 10 and 100 MeV and proper decay lengths below 100 μm, reaching branching fraction sensitivities as low as 10 −5 . This is the first search for Higgs boson decays to four electrons via light pseudoscalars at the LHC. It significantly improves the experimental sensitivity to axionlike particles with masses below 100 MeV.
Hayrapetyan, A. [Yerevan Physics Institute]↗
Low-level arsenic exposure via drinking water consumption and female fecundity - A preliminary investigation
High level arsenic exposure is associated with reproductive toxicity in experimental and observational studies; however, few data exist to assess risks at low levels. Even less data are available to evaluate the impact of low level arsenic exposure on human fecundity. Our aim in this pilot study was a preliminary evaluation of associations between low level drinking water arsenic contamination and female fecundity. This retrospective study was conducted among women previously recruited to a hospital-based case-control study of spontaneous pregnancy loss in Timiá¹£ County, Romania. Women (n=94) with planned pregnancies of 5–20 weeks gestation completed a comprehensive physician-administered study questionnaire and reported the number of menstrual cycles attempting to conceive as the time to pregnancy (TTP). Drinking water samples were collected from residential drinking water sources and we determined arsenic levels using hydride generation-atomic absorption spectrometry (HG-AAS). Multivariable Cox-proportional hazards regression with Efron approximation was employed to evaluate TTP as a function of drinking water arsenic concentrations among planned pregnancies, adjusted for covariates. There was no main effect for drinking water arsenic exposure, yet the conditional probability for pregnancy was modestly lower among arsenic exposed women with longer TTPs, relative to women with shorter TTPs, and relative to unexposed women. For example, 1 µg/L average drinking water arsenic conferred 5%, 8%, and 10% lower likelihoods for pregnancy in the 6th, 9th, and 12th cycles, respectively (P=0.01). While preliminary, our results suggest that low level arsenic contamination in residential drinking water sources may further impair fecundity among women with longer waiting times; however, this hypothesis requires confirmation by a future, more definitive study. - Highlights: • We assessed low level drinking water arsenic as a predictor of fecundability. • Arsenic did not affect time to pregnancy among women conceiving quickly. • Arsenic increased time to pregnancy among women taking longer to conceive. • Low level drinking water arsenic may adversely impact women with lower fecundity.
54 ENVIRONMENTAL SCIENCES↗
Evidence for top quark production in nucleus-nucleus collisions
We report droplets of quark-gluon plasma (QGP), an exotic state of strongly interacting quantum chromodynamics (QCD) matter, are routinely produced in heavy nuclei high-energy collisions. Although the experimental signatures marked a paradigm shift away from expectations of a weakly coupled QGP, a challenge remains as to how the locally deconfined state with a lifetime of a few fm can be resolved. The only colored particle that decays mostly within the QGP is the top quark. Here we demonstrate, for the first time, that top quark decay products are identified, irrespective of whether interacting with the medium (bottom quarks) or not (leptonically decaying W bosons). Using 1.7±0.1 nb -1 of lead-lead (A = 208) collision data recorded by the CMS experiment at a nucleon-nucleon center-of-mass energy of 5.02 TeV, we report evidence of top quark pair ($t\bar{t}$) production. Dilepton final states are selected, and the cross section ($σ_{t\bar{t}}$) is measured from a likelihood fit to a multivariate discriminator using lepton kinematic variables. The $σ_{t\bar{t}}$ measurement is additionally performed considering the jets originating from the hadronization of bottom quarks, which improve the sensitivity to the $t\bar{t}$ signal process. After background subtraction and analysis corrections, the measured $σ_{t\bar{t}}$ is 2.56 ± 0.82(tot) and 2.02 ± 0.69(tot)μb in the two cases, respectively, consistent with predictions from perturbative QCD.
71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
Leveraging large language models to address data scarcity in machine learning for graphene synthesis
Machine learning in experimental materials science faces significant challenges due to the scarcity of data, which are costly and time-consuming to generate, particularly when relying on in-house experiments. Literature data mining offers a potential solution but introduces issues like mixed data quality, inconsistent formats, and non-uniform reporting of synthesis parameters, resulting in partially missing and heterogeneous features across the dataset. Here, we propose data imputation and feature engineering methods that employ pre-trained large language models (LLMs) to enhance machine learning performance on scarce, heterogeneous datasets, demonstrated on graphene CVD synthesis data and the ML-HydPARK hydrogen storage dataset. GPT models perform data imputation via tailored prompting and semantic normalization of inconsistently reported features through embeddings, for example, to harmonize the complex nomenclature of CVD substrates. Beyond yielding more diverse and richer feature representations than traditional methods such as K-nearest neighbors (KNN) and Multivariate Imputation by Chained Equations (MICE), LLM-based data imputation is evaluated against dataset characteristics and prompting strategies. We vary the level of autonomy granted to the LLM, from generic prompting that leverages pre-trained knowledge for autonomous data generation to data-informed prompting that constrains outputs using target-specific information, and demonstrate which level of autonomy yields superior imputation performance across datasets and feature types. The proposed data engineering methods markedly improve downstream performance; for example, in graphene layer number classification using a support vector machine (SVM), binary accuracy increases from 39% to 65% and ternary accuracy from 52% to 72%. Fine-tuning experiments on both datasets show that combining our proposed LLM-based data imputation and feature encoding methods with numerical machine learning predictors outperforms standalone fine-tuned LLM predictors in data-scarce settings. The proposed strategies emphasize data enhancement techniques rather than refining learning architectures or regularizing loss functions, offering a broadly applicable framework for improving machine learning performance on scarce, inhomogeneous datasets.
Chemical vapor deposition↗
Nonlinear Decoupling Control with ANFIS-based Unmodeled Dynamics Compensation for A Class of Complex Industrial Processes
Complex industrial processes are multivariable and generally exhibit strong coupling among their control loops with heavy nonlinear nature that make it very difficult to obtain an accurate model. As a result, the conventional and data driven control methods are difficulty to apply. Using a twin-tank level control system as an example, a novel multivariable decupling control algorithm with adaptive neural fuzzy inference system (ANFIS) based unmodeled dynamics (UD) compensation is proposed in this paper for a class of complex industrial processes. At first, a nonlinear multivariable decupling controller with UD compensation is introduced. Different from the general method, the decomposition estimation algorithm by ANFIS is used to estimate the UD, and the desired estimating and decoupling control effects are achieved. Secondly, the proposed method does not require the complex switching mechanism which has been commonly used in the literature. This simplifies the obtained decoupling algorithm and its realization. Thirdly, based on some new lemmas and theorems, the conditions on the stability and convergence of the closed-loop system are analyzed to show the uniform boundedness of all the variables. Finally, experimental tests on a heavily coupled nonlinear twin-tank system are carried out to demonstrate the effectiveness and the practicability of the proposed method.
Complex industrial process, ANFIS, multivariable a↗
Design of experiments and data analysis challenges in calibration for forensics applications
Forensic science aims to infer characteristics of source terms using measured observables. Our focus is on statistical design of experiments and data analysis challenges arising in nuclear forensics. More specifically, we focus on inferring aspects of experimental conditions (of a process to produce product Pu oxide powder), such as temperature, nitric acid concentration, and Pu concentration, using measured features of the product Pu oxide powder. The measured features, Y, include trace chemical concentrations and particle morphology such as particle size and shape of the produced Pu oxide power particles. Making inferences about the nature of inputs X that were used to create nuclear materials having particular characteristics, Y, is an inverse problem. Therefore, statistical analysis can be used to identify the best set (or sets) of Xs for a new set of observed responses Y. One can fit a model (or models) such as Υ = f(Χ) + error , for each of the responses, based on a calibration experiment and then “invert” to solve for the best set of Xs for a new set of Ys. This perspectives paper uses archived experimental data to consider aspects of data collection and experiment design for the calibration data to maximize the quality of the predicted Ys in the forward models; that is, we assume that well-estimated forward models are effective in the inverse problem. In addition, we consider how to identify a best solution for the inferred X, and evaluate the quality of the result and its robustness to a variety of initial assumptions, and different correlation structures between the responses. In addition, we also briefly review recent advances in metrology issues related to characterizing particle morphology measurements used in the response vector, Y.
36 MATERIALS SCIENCE↗
An open source fast fluid dynamics model for data center thermal management
Although computational fluid dynamics (CFD) has been widely adopted to improve data center thermal management, the high computational demand limits its applications, such as multivariate optimal design and operation. Fast fluid dynamics (FFD), which has been applied for fast airflow simulation, shows great potential. However, few research applied FFD for optimal design and operation of data center thermal management. This research improves the FFD model for data centers and conducts a comprehensive evaluation and demonstration. First, the FFD model is improved by solving the advection and diffusion equations together using an upwind scheme instead of a semi-Lagrangian advection solver in the conventional FFD model. Second, new features for data centers are added, such as a pressure correction method to simulate plenum airflow and dynamic boundary conditions for IT racks. The new FFD model is first validated with two indoor environment cases and the results show that the new FFD model has slightly better overall prediction accuracy and faster speed compared to the conventional FFD model. It is also observed that both FFD models achieve acceptable accuracy, except for a few localized disparities with experimental data, which might be due to simplified handling of turbulence viscosity near the boundaries. Furthermore, validation with a real data center shows that the FFD model achieves a similar level of accuracy as CFD when compared to the experimental measurements with some level of uncertainties. It is then demonstrated for data center optimal design and operation, which saves 53.4–58.8% of annual energy while still meeting the thermal requirements. In conclusion, with a much faster speed and comparable accuracy compared to CFD, the FFD model parallelized on a graphics processing unit is promising for practical model-based data center early design and operation.
32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
Active Learning-driven Quantitative Synthesis-Structure-Property Relations for Improving Performance and Revealing Active Sites of Nitrogen-Doped Carbon for the Hydrogen Evolution Reaction
While quantitative structure-properties relations (QSPRs) have been developed successfully in multiple fields, catalyst synthesis affects structure and in turn performance, making simple QSPRs inadequate. Furthermore, catalysts often have multiple active sites preventing one from obtaining insights into structure-property relations. Here, we develop a data-driven quantitative synthesis-structure-property relation (QS2PRs) methodology to elucidate correlations between catalyst synthesis conditions, structural properties as well as observed performance and to provide fundamental insights into active sites and a systematic way to optimize practical catalysts. Here, we demonstrate the approach to the synthesis of nitrogen-doped catalysts (NDC) made via pyrolysis for the performance of the electrochemical hydrogen evolution reaction (HER), quantified by the onset potential and the current density. We determine crystallinity, nitrogen species type and fraction, surface area, and pore structure of the NDC’s using XRD, XPS, and BET characterization. We demonstrated that an active learning-based optimization combined with various elementary machine learning tools (regression, principal component analysis, partial least squares) can efficiently identify optimum pyrolysis conditions to tune structural characteristics and performance with concomitant savings in materials and experimental time. Unlike previous reports on the importance of pyridinic or graphitic nitrogen, we discover that the electrochemical performance is not driven by a single catalyst property; rather, it arises from a multivariate influence of nitrogen dopants, pore structure and disorder in the NDC materials. Identification of active sites can help mechanistic understanding and further catalyst improvement.
42 ENGINEERING↗
Robust Predictive Control for Modular Solid-State Transformer With Reduced DC Link and Parameter Mismatch
This paper presents the analysis and implementation of a predictive control method for dc-link regulation and voltage balance in a cascaded modular reduced dc-link solid-state transformer (SST). Passive components like bulky dc links limit the power density of power converters, especially medium-voltage (MV) SST. Reduced dc-link or low-inertia converters can dramatically reduce the size, cost, and weight by tolerating larger dc-link ripples and improve the reliability with electrolytic capacitor-less dc link. However, a small dc link leads to tight coupling between the input and the output stages, which is a challenge for control design. In stacked low-inertia converters (SLIC), the low-inertia converter modules are stacked for MV applications, resulting in coupling between the modules and making the control more challenging. A new model predictive control method which can achieve deadbeat regulation on the dc link without weighting factors has been proposed to address this novel problem. This paper focuses on analyzing the condition of the low-inertia dc link up to 80% ripple, the robustness of the control under parameter mismatches, high-order terms, and important implementation issues such as model-based sampling and computation delay compensation. Significantly, the high-order terms are introduced because of the large dc-link ripple. These high-order terms are unique to the SLIC and negligible in conventional high-inertia converters. A discrete-time large-signal model is built to capture the dc-link’s nonlinear dynamics, and the eigenvalues of a small-signal Jacobian matrix are analyzed with Floquet theory to evaluate stability, using the modular soft-switching solid-state transformer (M-S4T) as an example of the SLIC. Simulation and experimental results of an MVDC M-S4T verify the analysis and the predictive control method. Finally, the general application of the predictive control to low-inertia converters is compared against a conventional PI controller using a reduced dc-link active-front-end (AFE) rectifier as an example.
14 SOLAR ENERGY↗