Hyunjae Kim

Quantitative analysis on the urban flood mitigation effect by the extensive green roof system

Extensive green-roof systems are expected to have a synergetic effect in mitigating urban runoff, decreasing temperature and supplying water to a building. Mitigation of runoff through rainwater retention requires the effective design of a green-roof catchment. This study identified how to improve building runoff mitigation through quantitative analysis of an extensive green-roof system. Quantitative analysis of green-roof runoff characteristics indicated that the extensive green roof has a high water-retaining capacity response to rainfall of less than 20 mm/h. As the rainfall intensity increased, the water-retaining capacity decreased. The catchment efficiency of an extensive green roof ranged from 0.44 to 0.52, indicating reduced runoff comparing with efficiency of 0.9 for a concrete roof. Therefore, extensive green roofs are an effective storm water best-management practice and the proposed parameters can be applied to an algorithm for rainwater-harvesting tank design.

Google unveils a glimpse of allergic rhinitis in the real world

Google Trends (GT) is a Web‐based surveillance tool used to explore the searching trends of specific queries on Google. Recent studies have suggested the utility of GT in predicting outbreaks of influenza and other diseases. However, this utility has not been thoroughly evaluated for allergic diseases. Therefore, we investigated the utility of GT for predicting the epidemiology of allergic rhinitis. In the USA, GT for allergic rhinitis showed repetitive seasonality that peaked in late April and early May and then rapidly decreased, and a second small peak occurred in September. These trends are highly correlated with the searching trends for other queries such as ‘pollen count’, antihistamines such as loratadine and cetirizine (all r > 0.88 and all P < 0.001), and even the total pollen count collected from 21 pollen counters across the USA (r = 0.928, P < 0.001). Google Trends for allergic rhinitis was similar to the monthly changes in rhinitis symptoms according to the US National Health and Nutrition Examination Survey III, sales for Claritin® and all over‐the‐counter antihistamines, and the number of monthly page views of ‘claritin.com’. In conclusion, GT closely reflects the real‐world epidemiology of allergic rhinitis in the USA and could potentially be used as a monitoring tool for allergic rhinitis.

A degenerated equivalent circuit model and hybrid prediction for state-of-health (SOH) of PEM fuel cell

The 2014 IEEE PHM data challenge problem deals with the state-of-health (SOH) of proton exchange membrane fuel cell (PEMFC) given two degradation data sets: (i) a reference data set (FC1) operated under constant current is fully given until 991 h and (ii) a test data set (FC2) operated under rippled current is partially given until 550h. The proposed research aims at predicting the SOH (or EIS spectra) of PEM fuel cell after 550h for FC2. First, a full scale equivalent circuit model (ECM) with 10 parameters is developed to describe the electrochemical physics of PEMFC more realistically. The model reduction is suggested because of limited data. Since some parameters remain nearly unchanged due to irrelevance to degradation, it is reasonable to use the degenerated 4-parameter ECM while fixing the other parameters at their means. Despite the model reduction, the degradation pattern is clearly observed through the degenerated 4-parameter ECM. Then the coefficients of the four parameters are estimated by building linear regression models between the parameters and voltage. Since the voltage change after 550h is not provided for FC2, the voltage degradation model is developed by modeling both reversible and irreversible degradation processes. This research also proposes a hybrid prognostic approach to the SOH (or EIS spectra) prediction. The voltage degradation model and the degenerated 4-parameter ECM are first developed based on the observation of the physical phenomenon. They are then trained for the purpose of the SOH prediction with the training EIS data sets (FC1 and FC2). It is demonstrated that this hybrid SOH prediction offers highly accurate prediction of the SOH (or EIS spectra) at t = 666, 830, and 1016h. Moreover, possible error sources are also discussed to further improve the prediction accuracy in future.

An Online-Applicable Model for Predicting Health Degradation of PEM Fuel Cells With Root Cause Analysis

This paper proposes a new prognostic method for the health state of proton exchange membrane (PEM) fuel cells. The method is designed to predict the state-of-health (SOH) of PEMs and provide root cause analysis of the predicted health degradation. In this method, an equivalent circuit model (ECM) is built to emulate the impedance spectrum of PEM fuel cells. Because the key degradation parameters in the ECM cannot be measured in situ, this method instead estimates the parameters indirectly using the output voltage. The estimation is based on the linear relationship between the key ECM parameters and the output voltage. Using the constructed ECM and the estimated parameters, an impedance spectrum at the current moment is produced. The historical voltage evolution is then extrapolated using linear and exponential models that represent the irreversible and reversible phenomena, respectively. The models are used to predict future ECM parameters and, eventually, the impedance spectrum at any moment in the future. Through these steps, the proposed method provides an online estimation of the current SOH and predicts the level of future degradation. The primary novel feature of the proposed method is its ability to diagnose the root causes of potential degradation using data from nondisruptive online monitoring.

Liver Lesion Detection from Weakly-Labeled Multi-phase CT Volumes with a Grouped Single Shot MultiBox Detector

We present a focal liver lesion detection model leveraged by custom-designed multi-phase computed tomography (CT) volumes, which reflects real-world clinical lesion detection practice using a Single Shot MultiBox Detector (SSD). We show that grouped convolutions effectively harness richer information of the multi-phase data for the object detection model, while a naive application of SSD suffers from a generalization gap. We trained and evaluated the modified SSD model and recently proposed variants with our CT dataset of 64 subjects by five-fold cross validation. Our model achieved a 53.3% average precision score and ran in under three seconds per volume, outperforming the original model and state-of-the-art variants. Results show that the one-stage object detection model is a practical solution, which runs in near real-time and can learn an unbiased feature representation from a large-volume real-world detection dataset, which requires less tedious and time consuming construction of the weak phase-level bounding box labels.

Online thermal state estimation of high power lithium-ion battery

This paper concerns about thermal management issues of Li-ion battery for optimal and safe operation of the battery. For the thermal management, it is important to estimate implicit thermal states such as internal temperature and heat generation rate of the battery. This research presents an online thermal state estimation technique for a pouch-type Li-ion polymer battery. The thermal states are represented by a state space model based on an equivalent circuit model. A computational thermal dynamic analysis for Li-ion battery is used for identifying the model parameters. A state estimation technique is developed based on a state observer and energy balance equation. This technique is validated with a constant and dynamic current discharge experiment on a 5Ah pouch type Li-ion battery.