Search NASA⌕ Search

SEARCH · Search NASA

Results for “filter methods”

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.

At least 631 records · Page 35

Systolic VLSI for Kalman filters

A novel two-dimensional parallel computing method for real-time Kalman filtering is presented. The mathematical formulation of a Kalman filter algorithm is rearranged to be the type of Faddeev algorithm for generalizing signal processing. The data flow mapping from the Faddeev algorithm to a two-dimensional concurrent computing structure is developed. The architecture of the resulting processor cells is regular, simple, expandable, and therefore naturally suitable for VLSI chip implementation. The computing methodology and the two-dimensional systolic arrays are useful for Kalman filter applications as well as other matrix/vector based algebraic computations.

Yeh, H.-G.↗

Scatter-Reducing Sounding Filtration Using a Genetic Algorithm and Mean Monthly Standard Deviation

Retrieval algorithms like that used by the Orbiting Carbon Observatory (OCO)-2 mission generate massive quantities of data of varying quality and reliability. A computationally efficient, simple method of labeling problematic datapoints or predicting soundings that will fail is required for basic operation, given that only 6% of the retrieved data may be operationally processed. This method automatically obtains a filter designed to reduce scatter based on a small number of input features. Most machine-learning filter construction algorithms attempt to predict error in the CO2 value. By using a surrogate goal of Mean Monthly STDEV, the goal is to reduce the retrieved CO2 scatter rather than solving the harder problem of reducing CO2 error. This lends itself to improved interpretability and performance. This software reduces the scatter of retrieved CO2 values globally based on a minimum number of input features. It can be used as a prefilter to reduce the number of soundings requested, or as a post-filter to label data quality. The use of the MMS (Mean Monthly Standard deviation) provides a much cleaner, clearer filter than the standard ABS(CO2-truth) metrics previously employed by competitor methods. The software's main strength lies in a clearer (i.e., fewer features required) filter that more efficiently reduces scatter in retrieved CO2 rather than focusing on the more complex (and easily removed) bias issues.

Mandrake, Lukas↗

An automated method of tuning an attitude estimator

Attitude determination is a major element of the operation and maintenance of a spacecraft. There are several existing methods of determining the attitude of a spacecraft. One of the most commonly used methods utilizes the Kalman filter to estimate the attitude of the spacecraft. Given an accurate model of a system and adequate observations, a Kalman filter can produce accurate estimates of the attitude. If the system model, filter parameters, or observations are inaccurate, the attitude estimates may be degraded. Therefore, it is advantageous to develop a method of automatically tuning the Kalman filter to produce the accurate estimates. In this paper, a three-axis attitude determination Kalman filter, which uses only magnetometer measurements, is developed and tested using real data. The appropriate filter parameters are found via the Process Noise Covariance Estimator (PNCE). The PNCE provides an optimal criterion for determining the best filter parameters.

Mason, Paul A. C.↗

Computer-aided detection of clustered microcalcifications in multiscale bilateral filtering regularized reconstructed digital breast tomosynthesis volume

Purpose: Develop a computer-aided detection (CADe) system for clustered microcalcifications in digital breast tomosynthesis (DBT) volume enhanced with multiscale bilateral filtering (MSBF) regularization. Methods: With Institutional Review Board approval and written informed consent, two-view DBT of 154 breasts, of which 116 had biopsy-proven microcalcification (MC) clusters and 38 were free of MCs, was imaged with a General Electric GEN2 prototype DBT system. The DBT volumes were reconstructed with MSBF-regularized simultaneous algebraic reconstruction technique (SART) that was designed to enhance MCs and reduce background noise while preserving the quality of other tissue structures. The contrast-to-noise ratio (CNR) of MCs was further improved with enhancement-modulated calcification response (EMCR) preprocessing, which combined multiscale Hessian response to enhance MCs by shape and bandpass filtering to remove the low-frequency structured background. MC candidates were then located in the EMCR volume using iterative thresholding and segmented by adaptive region growing. Two sets of potential MC objects, cluster centroid objects and MC seed objects, were generated and the CNR of each object was calculated. The number of candidates in each set was controlled based on the breast volume. Dynamic clustering around the centroid objects grouped the MC candidates to form clusters. Adaptive criteria were designed to reduce false positive (FP) clusters based on the size, CNR values and the number of MCs in the cluster, cluster shape, and cluster based maximum intensity projection. Free-response receiver operating characteristic (FROC) and jackknife alternative FROC (JAFROC) analyses were used to assess the performance and compare with that of a previous study. Results: Unpaired two-tailedt-test showed a significant increase (p < 0.0001) in the ratio of CNRs for MCs with and without MSBF regularization compared to similar ratios for FPs. For view-based detection, a sensitivity of 85% was achieved at an FP rate of 2.16 per DBT volume. For case-based detection, a sensitivity of 85% was achieved at an FP rate of 0.85 per DBT volume. JAFROC analysis showed a significant improvement in the performance of the current CADe system compared to that of our previous system (p = 0.003). Conclusions: MBSF regularized SART reconstruction enhances MCs. The enhancement in the signals, in combination with properly designed adaptive threshold criteria, effective MC feature analysis, and false positive reduction techniques, leads to a significant improvement in the detection of clustered MCs in DBT.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

TU-F-BRE-06: Flattening-Filter-Free Beam Quality Correction Factor Determination Using Experimental and Monte Carlo Methods

Purpose: To investigate beam quality correction factors for the flattening-filter-free (FFF) energies of the TrueBeam™ accelerator based on a dosimetry formalism for small and nonstandard fields. Methods: Three detectors - an Exradin W1 scintillator, Sun Nuclear EDGE diode, and LiF(Mg,Tl) TLD-100 chips - were investigated to determine their applicability as tools to measure quality correction factors for ionization chambers in the small and nonstandard fields of the TrueBeam™. Volume-averaging effects and energy dependence were observed in fields ranging from 1×1 to 40×40 cm{sup 2} for 6 MV and 10 MV beam energies using both FFF and flattened beam modes. Correction factors were determined for three ionization chambers: an Exradin A12 Farmer-type chamber, an Exradin A1SL scanning chamber, and an Exradin A26 reference-class microchamber. Beam quality corrections were also obtained using a benchmarked model of the TrueBeam™ created with the BEAMnrc user code of EGSnrc. Results: All three detectors demonstrated measureable energy dependence in the megavoltage range. The EDGE diode was deemed the most appropriate tool for beam quality correction factor measurements due to its low energy dependence and small size; however, alanine will be used in the future to reduce energy dependent effects even further. Measured k{sub Qmsr,Q} corrections of up to 4% were found for the 6MV FFF and 10 MV FFF beams, corresponding to a discrepancy of up to 3% compared to TG-51-determined dose. Up to a 10% k{sub Qclin,Qmsr} correction was measured for small fields referenced to a 10×10 cm{sup 2} field of the same energy. Much larger corrections were determined using the Monte Carlo model, and these discrepancies require further investigation. Conclusion: Progress has been made toward determining beam quality correction factors for the small and nonstandard fields of the TrueBeam™ accelerator. Further work must be done to ensure greater accuracy in patient treatments with this new modality.

07 ISOTOPE AND RADIATION SOURCES↗

Interior region-of-interest reconstruction using a small, nearly piecewise constant subregion

Purpose: To develop a method to reconstruct an interior region-of-interest (ROI) image with sufficient accuracy that uses differentiated backprojection (DBP) projection onto convex sets (POCS) [H. Kudo et al., ''Tiny a priori knowledge solves the interior problem in computed tomography'', Phys. Med. Biol. 53, 2207-2231 (2008)] and a tiny knowledge that there exists a nearly piecewise constant subregion. Methods: The proposed method first employs filtered backprojection to reconstruct an image on which a tiny region P with a small variation in the pixel values is identified inside the ROI. Total variation minimization [H. Yu and G. Wang, ''Compressed sensing based interior tomography'', Phys. Med. Biol. 54, 2791-2805 (2009); W. Han et al., ''A general total variation minimization theorem for compressed sensing based interior tomography'', Int. J. Biomed. Imaging 2009, Article 125871 (2009)] is then employed to obtain pixel values in the subregion P, which serve as a priori knowledge in the next step. Finally, DBP-POCS is performed to reconstruct f(x,y) inside the ROI. Clinical data and the reconstructed image obtained by an x-ray computed tomography system (SOMATOM Definition; Siemens Healthcare) were used to validate the proposed method. The detector covers an object with a diameter of {approx}500 mm. The projection data were truncated either moderately to limit the detector coverage to diameter 350 mm of the object or severely to cover diameter 199 mm. Images were reconstructed using the proposed method. Results: The proposed method provided ROI images with correct pixel values in all areas except near the edge of the ROI. The coefficient of variation, i.e., the root mean square error divided by the mean pixel values, was less than 2.0% or 4.5% with the moderate or severe truncation cases, respectively, except near the boundary of the ROI. Conclusions: The proposed method allows for reconstructing interior ROI images with sufficient accuracy with a tiny knowledge that there exists a nearly piecewise constant subregion.

60 APPLIED LIFE SCIENCES↗

Plasma discharge self-cleaning filtration system

The present invention is directed to a novel method for cleaning a filter surface using a plasma discharge self-cleaning filtration system. The method involves utilizing plasma discharges to induce short electric pulses of nanoseconds duration at high voltages. These electrical pulses generate strong Shockwaves that disintegrate and dislodge particulate matter located on the surface of the filter.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Frequency-Domain Deconvolution for Flight Dynamics Applications

A deconvolution method is presented for estimating input data from measured output data and a model of the dynamic process involved. The method uses an optimal Wiener filter for separating the measured data into signal and noise components, and a high-accuracy Fourier transform for inverting the model dynamics in the frequency domain. The method is an extension of optimal Fourier smoothing, and uses a technique to enhance the contrast between the signal and noise spectra in designing the Wiener filter. The deconvolution method was applied to simulation and flight test data for the purposes of removing unwanted distortions introduced by signal-conditioning filters and sensor dynamics, and for reconstructing turbulence inputs from measured sensor data. Results indicated hat the method performs well given good signal-to-noise levels and accurate models of the dynamic process.

Grauer, Jared A.↗

SU-E-J-02: 4D Digital Tomosynthesis Based On Algebraic Image Reconstruction and Total-Variation Minimization for the Improvement of Image Quality

Purpose: In this paper, we implemented the four-dimensional (4D) digital tomosynthesis (DTS) imaging based on algebraic image reconstruction technique and total-variation minimization method in order to compensate the undersampled projection data and improve the image quality. Methods: The projection data were acquired as supposed the cone-beam computed tomography system in linear accelerator by the Monte Carlo simulation and the in-house 4D digital phantom generation program. We performed 4D DTS based upon simultaneous algebraic reconstruction technique (SART) among the iterative image reconstruction technique and total-variation minimization method (TVMM). To verify the effectiveness of this reconstruction algorithm, we performed systematic simulation studies to investigate the imaging performance. Results: The 4D DTS algorithm based upon the SART and TVMM seems to give better results than that based upon the existing method, or filtered-backprojection. Conclusion: The advanced image reconstruction algorithm for the 4D DTS would be useful to validate each intra-fraction motion during radiation therapy. In addition, it will be possible to give advantage to real-time imaging for the adaptive radiation therapy. This research was supported by Leading Foreign Research Institute Recruitment Program (Grant No.2009-00420) and Basic Atomic Energy Research Institute (BAERI); (Grant No. 2009-0078390) through the National Research Foundation of Korea(NRF) funded by the Ministry of Science, ICT and Future Planning (MSIP)

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Safety of Percutaneous Hepatic Perfusion with Melphalan in Patients with Unresectable Liver Metastases from Ocular Melanoma Using the Delcath Systems’ Second-Generation Hemofiltration System: A Prospective Non-randomized Phase II Trial

Purpose: To investigate the safety and toxicity of percutaneous hepatic perfusion with melphalan (M-PHP) with the Delcath Systems’ second-generation (GEN 2) filter and compare the outcomes with historical data from studies using the first-generation filter. Materials and Methods: A prospective, single-arm, single-center phase II study was carried out including 35 patients with unresectable, histologically confirmed liver metastases from ocular melanoma between February 2014 and June 2017. Main exclusion criteria were extrahepatic disease and age > 75 years. M-PHP was performed with melphalan 3 mg/kg (maximum dose 220 mg). Safety and toxicity were assessed according to the Common Terminology Criteria for Adverse Events version 4.03. Results: A total of 67 M-PHPs were performed in 35 patients (median 2 procedures). Although hematologic grade 3/4 events were seen in the majority of patients (thrombocytopenia 54.5%, leukopenia 75.6%, neutropenia 66.7%, anemia (only grade 3) 18.1%), these were all well manageable or self-limiting. Of the non-hematologic grade 3 events (n = 14), febrile neutropenia (n = 3), pulmonary emboli (n = 2) and post-procedural hemorrhage (n = 2) were most common. A case of sepsis with bacterial pharyngitis was the only non-hematologic grade 4 event. Prior therapy for liver metastases was found to be a predictor of late grade 3/4 neutropenia with an odds ratio of 5.5 (95% CI 1.4–21.7). Conclusions: M-PHP using the GEN 2 filter has an acceptable safety and toxicity profile, and seems to reduce hematologic toxicity when compared to M-PHP with a first-generation filter. Prior therapy of liver metastases is a possible predictive factor in developing grade 3/4 hematologic toxicity.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Solid particulate mass and number from ducted fuel injection in an optically accessible diesel engine in skip-fired operation

Ducted fuel injection (DFI) is a new combustion strategy that has been shown to significantly attenuate soot formation in diesel engines. While previous studies have used optical diagnostics and optical filter smoke number methods to show that DFI reduces in-cylinder soot formation and engine-out soot emissions, respectively, this is the first study to measure solid particle number (PN) emissions in addition to particle mass (PM). Furthermore, this study quantitatively evaluates the use of transient particle instruments for measuring particles from skip-fired operation in an optical single cylinder research engine (SCRE). Engine-out PN was measured using an engine exhaust particle sizer following a catalytic stripper, and PM was measured using a photoacoustic analyzer. The study improves on earlier preliminary emissions studies by clearly showing that DFI reduces overall PM by 76%–79% and PN for particles larger than 23 nm by 77% relative to conventional diesel combustion at a 1200-rpm, 13.3-bar gross indicated mean effective pressure operating condition. The degree of engine-out PM reduction with DFI was similar across both particulate measurement instruments used in the work. Through the use of bimodal distribution fitting, DFI was also shown to reduce the geometric mean diameter of accumulation mode particles by 26%, similar to the effects of increased injection pressure in conventional diesel combustion systems. This work clearly shows the significant solid particulate matter reductions enabled by DFI while also demonstrating that engine-out PN can be accurately measured from an optical SCRE operating in a skip-fired mode. Based on these results, it is believed that DFI has the potential to enable fuel savings when implemented in multi-cylinder engines, both by lowering the required frequency of active diesel particulate filter regeneration, and by reducing the backpressure imposed by exhaust filtration systems.

33 ADVANCED PROPULSION SYSTEMS↗

Towards Efficient Convolutional Neural Networks Through Low-Error Filter Saliency Estimation

Filter saliency based channel pruning is a state-of-the-art method for deep convolutional neural network compression and acceleration. This channel pruning method ranks the importance of individual filter by estimating its impact of each filter’s removal on the training loss, and then remove the least important filters and fine-tune the remnant network. In this work, we propose a systematic channel pruning method that significantly reduces the estimation error of filter saliency. Different from existing approaches, our method largely reduces the magnitude of parameters in a network by introducing alternating direction method of multipliers (ADMM) into the pre-training procedure. Therefore, the estimation of filter saliency based on Taylor expansion is significantly improved. Extensive experiments with various benchmark network architectures and datasets demonstrate that the proposed method has a much improved unimportant filter selection capability and outperform state-of-the-art channel pruning method.

Wang, Zi↗

Optimization of leaf margins for lung stereotactic body radiotherapy using a flattening filter-free beam

Purpose: The authors sought to determine the optimal collimator leaf margins which minimize normal tissue dose while achieving high conformity and to evaluate differences between the use of a flattening filter-free (FFF) beam and a flattening-filtered (FF) beam. Methods: Sixteen lung cancer patients scheduled for stereotactic body radiotherapy underwent treatment planning for a 7 MV FFF and a 6 MV FF beams to the planning target volume (PTV) with a range of leaf margins (−3 to 3 mm). Forty grays per four fractions were prescribed as a PTV D95. For PTV, the heterogeneity index (HI), conformity index, modified gradient index (GI), defined as the 50% isodose volume divided by target volume, maximum dose (Dmax), and mean dose (Dmean) were calculated. Mean lung dose (MLD), V20 Gy, and V5 Gy for the lung (defined as the volumes of lung receiving at least 20 and 5 Gy), mean heart dose, and Dmax to the spinal cord were measured as doses to organs at risk (OARs). Paired t-tests were used for statistical analysis. Results: HI was inversely related to changes in leaf margin. Conformity index and modified GI initially decreased as leaf margin width increased. After reaching a minimum, the two values then increased as leaf margin increased (“V” shape). The optimal leaf margins for conformity index and modified GI were −1.1 ± 0.3 mm (mean ± 1 SD) and −0.2 ± 0.9 mm, respectively, for 7 MV FFF compared to −1.0 ± 0.4 and −0.3 ± 0.9 mm, respectively, for 6 MV FF. Dmax and Dmean for 7 MV FFF were higher than those for 6 MV FF by 3.6% and 1.7%, respectively. There was a positive correlation between the ratios of HI, Dmax, and Dmean for 7 MV FFF to those for 6 MV FF and PTV size (R = 0.767, 0.809, and 0.643, respectively). The differences in MLD, V20 Gy, and V5 Gy for lung between FFF and FF beams were negligible. The optimal leaf margins for MLD, V20 Gy, and V5 Gy for lung were −0.9 ± 0.6, −1.1 ± 0.8, and −2.1 ± 1.2 mm, respectively, for 7 MV FFF compared to −0.9 ± 0.6, −1.1 ± 0.8, and −2.2 ± 1.3 mm, respectively, for 6 MV FF. With the heart inside the radiation field, the mean heart dose showed a V-shaped relationship with leaf margins. The optimal leaf margins were −1.0 ± 0.6 mm for both beams. Dmax to the spinal cord showed no clear trend for changes in leaf margin. Conclusions: The differences in doses to OARs between FFF and FF beams were negligible. Conformity index, modified GI, MLD, lung V20 Gy, lung V5 Gy, and mean heart dose showed a V-shaped relationship with leaf margins. There were no significant differences in optimal leaf margins to minimize these parameters between both FFF and FF beams. The authors’ results suggest that a leaf margin of −1 mm achieves high conformity and minimizes doses to OARs for both FFF and FF beams.

60 APPLIED LIFE SCIENCES↗

Ion chromatographic determination of anions collected on filters at altitudes between 9.6 and 13.7 kilometers

The investigation of atmospheric pollution to determine the anion-containing particulates in the atmosphere at altitudes between 9.6 and 13.7 km is discussed. Air samples collected on cellulose fiber discs impregnated with dibutoxyethylphthalate require very sensitive methods of analysis. It is concluded that ion chromatography is suited for the determination of anions collected on the filters. Methods to control contamination and interfering side reactions are described.

Otterson, D. A.↗

The Twentieth Century Reanalysis Project

The Twentieth Century Reanalysis (20CR) project is an international effort to produce a comprehensive global atmospheric circulation dataset spanning the twentieth century, assimilating only surface pressure reports and using observed monthly sea-surface temperature and sea-ice distributions as boundary conditions. It is chiefly motivated by a need to provide an observational dataset with quantified uncertainties for validations of climate model simulations of the twentieth century onall time-scales,with emphasisonthe statistics of daily weather. It uses an Ensemble Kalman Filter data assimilation method with background ‘first guess’ fields supplied by an ensemble of forecasts from a global numerical weather prediction model. This directly yields a global analysis every 6 hours as the most likely state of the atmosphere, and also an uncertainty estimate of that analysis. The 20CR dataset provides the first estimates of global tropospheric variability, and of the dataset’s time-varying quality, from 1871 to the present at 6-hourly temporal and 2° spatial resolutions. Intercomparisons with independent radiosonde data indicate that the reanalyses are generally of high quality. The quality in the extratropical Northern Hemisphere throughout the century is similar to that of current three-day operational NWP forecasts. Intercomparisons over the second half-century of these surface-based reanalyses with other reanalyses that also make use of upper-air and satellite data are equally encouraging. It is anticipated that the 20CR dataset will be a valuable resource to the climate research community for both model validations and diagnostic studies. Some surprising results are already evident. For instance, the long-term trends of indices representing the North Atlantic Oscillation, the tropical Pacific Walker Circulation, and the Pacific–North American pattern are weak or non-existent over the full period of record. The long-term trends of zonally averaged precipitation minus evaporation also differ in character from those in climate model simulations of the twentieth century.

Meteorology & Atmospheric Sciences↗

Perforation of the IVC: Rule Rather Than Exception After Longer Indwelling Times for the Guenther Tulip and Celect Retrievable Filters

Purpose: This study was designed to assess the incidence, magnitude, and impact upon retrievability of vena caval perforation by Guenther Tulip and Celect conical inferior vena cava (IVC) filters on computed tomographic (CT) imaging. Methods: Guenther Tulip and Celect IVC filters placed between July 2007 and May 2009 were identified from medical records. Of 272 IVC filters placed, 50 (23 Guenther Tulip, 46%; 27 Celect, 54%) were retrospectively assessed on follow-up abdominal CT scans performed for reasons unrelated to the filter. Computed tomography scans were examined for evidence of filter perforation through the vena caval wall, tilt, or pericaval tissue injury. Procedure records were reviewed to determine whether IVC filter retrieval was attempted and successful. Results: Perforation of at least one filter component through the IVC was observed in 43 of 50 (86%) filters on CT scans obtained between 1 and 880 days after filter placement. All filters imaged after 71 days showed some degree of vena caval perforation, often as a progressive process. Filter tilt was seen in 20 of 50 (40%) filters, and all tilted filters also demonstrated vena caval perforation. Transjugular removal was attempted in 12 of 50 (24%) filters and was successful in 11 of 12 (92%). Conclusions: Longer indwelling times usually result in vena caval perforation by retrievable Guenther Tulip and Celect IVC filters. Although infrequently reported in the literature, clinical sequelae from IVC filter components breaching the vena cava can be significant. We advocate filter retrieval as early as clinically indicated and increased attention to the appearance of IVC filters on all follow-up imaging studies.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Analyzing LF/VLF Lightning Waveforms to Estimate D-region Electron Density Profiles

Lightning waveforms in the low frequency (LF; 30-300 kHz) and the very low frequency (VLF; 3-30 kHz) bands can be exploited to produce data-driven ionospheric D-region electron density profile (EDP) estimates with significantly higher spatial and temporal coverage than previously available. The lightning waveforms used in this paper are signals detected in the LF/VLF of negative cloud-to-ground lightning by the Earth Networks Total Lightning Detection Network. Each waveform contains a ground wave and a time-delayed ionospheric reflection. The time delay between the ground wave and ionospheric reflection has previously been used to estimate a single specular reflection altitude, where LF/VLF emissions are reflected by the ionosphere. Here, we expand upon previous methods to include filtering and spectral analysis, and account for oblique propagation to produce higher-order estimates for reflection altitudes and corresponding electron densities. Once estimated, reflection altitudes and corresponding electron densities can be used to derive parameters β and h’, which define an EDP for the D-region. In this study, the lightning waveform (LW) analysis is demonstrated using a single representative 24-hour dataset over the Southeast United States, and then extended to a total of 10 separate datasets with varying locations and ionospheric conditions. The LW-derived D-region EDPs are in agreement with predictions made by the Faraday International Reference Ionosphere model, and the LW EDPs β and h’ values are consistent with previous LF/VLF-derived estimates.

D-region ionosphere↗

A systematic feature extraction and selection framework for data-driven whole-building automated fault detection and diagnostics in commercial buildings

In data-driven automated fault detection and diagnostics (AFDD) modeling for building energy systems, feature engineering is a critical process of extracting information from high-dimensional and noisy sensor measurement and turning it into informative and representative inputs or features for data-driven modeling. However, few studies specifically discuss the feature engineering, especially the interactions between feature extraction and feature selection in whole-building AFDD. We developed a systematic feature extraction and selection framework for whole-building AFDD. In this framework, features are aggressively extracted from raw sensor data using statistical feature extraction techniques with various window sizes and statistics. With many features extracted, a hybrid feature selection algorithm that combines the filter and wrapper method then selects the best feature set. The framework considers diversity in the duration of fault behavior among fault types in whole-building AFDD, thus achieving high model generalization. We implemented our developed framework in a virtual testbed calibrated with measured data from Oak Ridge National Laboratory's Flexible Research Platform designed to mimic the operation of a typical small commercial building. The AFDD model is trained by the simulation data generated from the virtual testbed. The results show that (1) the developed framework improves the generalization of the AFDD model by 10.7% compared with literature-reported feature extraction and selection methods and (2) features with diverse window sizes and statistics are selected, providing insight into physical systems beyond the current understanding of buildings and faults and improving the detection and diagnostics of multiple fault types.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗