KAKaushallya Adhikari

Kaushallya Adhikari

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Kaushallya Adhikari has a total of 26 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Specialized State Space Battery Modeling for EKF-Based SOC Estimation in Photovoltaic Battery Storage Systems

    by Showrov Rahman, Kaushallya Adhikari

    The use of Kalman filter (KF) algorithms for state of charge (SOC) estimation in photovoltaic battery energy storage systems (PV-BESS) has received comparatively less attention than their widespread application in electric vehicles (EVs). However, unlike EV batteries, the charging behavior of a PV-BESS is significantly influenced by weather conditions. Consequently, modifying the KF algorithm to incorporate weather-related factors may lead to better accuracy. To explore this potential, we developed specialized state space models (SSMs) for the extended Kalman filter (EKF) algorithm, specifically tailored for SOC estimation in PV-BESS. The SSMs integrates key environmental variables, including ambient temperature, solar irradiance and humidity. For model validation, we built a custom-designed data acquisition (DAQ) system to collect real-world data. The design, technical specifications, and cost evaluation of the DAQ system are presented in detail. A two-level illustration approach is adopted to provide a comprehensive overview of the estimator performance. First, case studies under specific charging and discharging conditions highlight estimation errors at each time step and the corresponding root mean square error (RMSE). Second, a generalized evaluation presents the mean, median, maximum, and minimum RMSE values across different estimation scenarios, along with their corresponding confidence intervals. The results exhibits that the specialized EKF estimators outperform the baselines during charging, reducing RMSE by 3% and 4.75% for the combined and Rint models, respectively. During discharging, baseline models yield a mean RMSE of 3–4% across discharge rates.


  • Open Access

    Descriptor: University of Rhode Island Photovoltaic Battery Energy Storage SOC Tracking Dataset (URI-PBEST)

    by Showrov Rahman, Kaushallya Adhikari, Kaan Icer

    Battery energy storage systems (BESS) have emerged as a premier energy storage solution due to their modular architecture, decreasing costs, and increasing energy densities. These systems are increasingly integrated with photovoltaic (PV) arrays to mitigate energy wastage and enable sophisticated energy management. However, the operational environment and charging/discharging profiles of a PV-BESS differ significantly from other applications that also deploy BESSs, such as electric vehicles (EVs). While EV batteries typically undergo controlled, high-power charging cycles, a PV-BESS must manage the stochastic and intermittent nature of solar irradiance, leading to irregular cycling patterns and complex power dynamics. In this article, we introduce the URI-PBEST dataset, specifically developed for the State of Charge (SOC) tracking of PV-integrated battery systems. The data were collected at a high temporal resolution (1-minute intervals) during the 2024–2025 period at the University of Rhode Island (URI) Kingston Campus and are provided in a standard spreadsheet (.xlsx) format. The dataset captures distinct charging and discharging phases; specifically, the system records energy storage under highly variable weather conditions and discharge cycles under constant current conditions. The primary contribution of this dataset is its unique ability to facilitate investigations into the impact of environmental variables, such as solar irradiance, ambient temperature, and humidity, on SOC estimation of PV batteries. By capturing operational data across diverse weather profiles, including sunny, partly cloudy, and overcast days, URI-PBEST offers a realistic benchmark for analyzing how weather-related fluctuations influence the charging dynamics of PV-BESS.


  • Open Access

    Physics-Informed Voltage Data Imputation for Robust Machine-Learning-Based Battery State of Charge Estimation

    by Showrov Rahman, Kaushallya Adhikari

    Machine learning (ML) estimators used for battery state of charge (SOC) prediction are vulnerable to missing measurement data and therefore require data imputation techniques to replace unavailable observations. However, most existing data imputation methods for SOC estimation rely primarily on statistical approaches rather than incorporating battery physics and dynamic behavior. Consequently, the estimation accuracy of such estimators degrades significantly during long periods of consecutive missing data. This study addresses this research gap by proposing a physics-informed (PI) data imputation method that reconstructs missing voltage measurements using a battery equivalent circuit model. The proposed PI data imputation technique is integrated with a long short-term memory (LSTM) architecture to develop the ML estimator. To evaluate the estimation performance, the developed model is compared with two other LSTM estimators that use conventional voltage imputation methods: last observation carried forward (LOCF) and extrapolation. The estimation errors of the three estimators are evaluated using experimental data obtained from a photovoltaic battery energy storage system (PV-BESS) equipped with lithium iron phosphate (LFP) batteries. Furthermore, the robustness of the estimators is analyzed by quantifying the confidence intervals using Bayesian approximation and Monte Carlo forward passes, providing a comprehensive assessment of estimation performance. The estimators are evaluated under different missing data ratios of 25%, 50%, and 75% in order to systematically compare their performance. The results demonstrate that the proposed PI-based estimator consistently achieves superior performance across all scenarios, and the performance gap between the PI-based approach and conventional imputation methods (LOCF and extrapolation) increases as the duration of missing data becomes longer.


  • Open Access

    Multi-Step Kalman Filter With Sparse Measurements

    by Showrov Rahman, Kaushallya Adhikari

    Kalman filters (KFs) often operate under intermittent measurements, either due to unintentional conditions (e.g., sensor failure) or due to strategic modeling decisions (e.g., reducing computational burden by using fewer measurements). As measurements become sparse, the error in state prediction accumulates, making the predicted state increasingly uncertain. This paper develops a mathematical framework for a multi-step KF that analytically characterizes error propagation in the presence of sparse measurements. Closed-form expressions for an L-step KF estimator are derived, where the estimator’s state-space model is developed from a first-order Gauss–Markov process that experiences (L − 1) consecutive missing measurements. The derived closed-form expressions in the prediction stage enable direct quantification of the mean squared error (MSE) of the state prediction under missing measurements. The proposed framework is validated using experimental data to estimate the state of health (SOH) of lithium-ion batteries. For verification, Monte Carlo simulation trials are conducted to compute the empirical MSE under different periodic sparse measurement (PSM) patterns. The results demonstrate that, with a sufficiently large number of trials (5000), the empirical MSE closely matches the MSE obtained from the analytical prediction error expression. These findings demonstrate that the proposed framework is effective in providing forward state prediction for systems experiencing measurement outages or strategically designed intermittent sensing.


  • Local-CGFC: A Local Cumulant Generating Function Classification Rule

    by Bo Tang, Steven Kay, Kaushallya Adhikari

    A classification rule based on the cumulant generating function of the training data, called the Cumulant Generating Function Classifier (CGFC), has been recently proposed, and has shown promising performance in terms of improved classification accuracy and robustness against noises. This paper first presents a new information-theoretical explanation of CGFC which indeed makes a classification by minimizing sample mutual information. The original CGFC is a type of global model, and a new variant, called Local-CGFC, is further introduced in this paper to achieve a local classification rule. Experimental studies on real-life datasets demonstrate the effectiveness of the proposed classifier and further illustrate its great potential for a number of real-world applications.


  • An Analytical Implementation of the Rosenblatt Transformation

    by Steven Kay, Kaushallya Adhikari, Kaan Icer

    A new approach to the analytical implementation of the Rosenblatt transformation is described. It leverages the properties of the empirical probability density function, which is the standard estimate of an unknown density. As such its utility is to applications where training data is available for the unknown density. These applications include data-driven algorithms for detection/classification and other statistical signal processing problems where the underlying probabilistic description of the data is unknown. As an illustration, an application to anomaly detection is described in detail using Gaussian and radar datasets.


  • Introducing Information Metrics for Statistical Signal Processing [Lecture Notes]

    by Steven Kay, Kaushallya Adhikari

    Statistical signal processing involves extracting information from random observations, a task that relies heavily on metrics that quantify the information present in the data. These metrics are crucial for practitioners and researchers in fields such as radar, sonar, and communications. Over time, several key information metrics have been introduced, often using different names. This tutorial describes essential information metrics—the Kullback–Leibler divergence (KLD), Fisher information (FI), and mutual information (MI)—exploring their relationships and demonstrating their relevance to estimation and detection problems. The article illustrates these concepts using the example of a constant-amplitude signal in white Gaussian noise (WGN) and discusses the asymptotic behavior of these metrics in more general scenarios. Additionally, it presents the minimum discrimination information (MDI) theorem, which states that the probability density function (PDF) minimizing the KLD under a mean constraint belongs to the linear exponential family. The tutorial concludes with an exploration of the conditionality principle, revealing the insightful relationship among the KLD, signal-to-noise ratio (SNR), and MI and its implications for improved detection in a multipath environment.


  • Open Access

    Convolutional Neural Networks for Direction of Arrival Estimation Compared to Classical Estimators and Bounds

    by Christopher J. Bell, Kaushallya Adhikari, Andrew Brown

    Recently, there has been a proliferation of applied machine learning (ML) research, including the use of convolutional neural networks (CNNs) for direction of arrival (DoA) estimation. With the increasing amount of research in this area, it is important to balance the performance and computational costs of CNNs with classical methods of DoA estimation such as Multiple Signal Classification (MUSIC). We outline the performance of both methods of DoA estimation for single-source and two-source cases for multiple array conditions. The results are also compared to the Cramer-Rao lower bound (CRLB) and conventional beamforming. For each source case, CNNs were trained for a perfect uniform line array (ULA) and tested against data from a perfect ULA, perturbed ULAs, ULAs with missing sensors, and ULAs with muffled sensors. We show that for the single-source case, the CNNs do not offer any performance improvement relative to MUSIC at low signal-to-noise ratio (SNR). For the two-source cases, the CNNs perform better than MUSIC but only at low SNR. For the remaining array cases, the CNNs outperform MUSIC. These results indicate that the performance improvements from CNNs are highest for situations where there is signal model to data mismatch (imperfect information). This work also illustrates that the CNN estimators developed in this work exceed the CRLB and are biased estimators caused by the lack of unbiased constraint in the loss function during training of the CNNs.


  • Open Access

    A Nonparametric Data-Driven Classifier Based on the Cumulant Generating Function

    by Steven Kay, Kaushallya Adhikari, Bo Tang

    We introduce a nonparametric data-driven classifier that harnesses the statistical properties of the data through the cumulant generating function of the training data. Its implementation is straightforward, requiring only a single tuning parameter. Moreover, it ensures global solutions due to inherent convex optimization. The classifier is explainable, where unexpected or poor results can be interpreted and ameliorated. We derive the properties of the classification statistic, offering insightful observations. We apply the classifier to real-world datasets. The simulation results demonstrate the efficacy of the proposed classifier in signal classification, even in scenarios with mismatched training and testing datasets. Moreover, the results demonstrate that the CGFC has lower computational complexity compared to neural networks.