Bikash Chandra Pal’s research while affiliated with Imperial College London and other places

This paper proposes a secure and privacy-preserving framework for detecting unauthorized energy usage by leveraging consumer energy consumption data in smart grids. Addressing the limitations of traditional centralized schemes, a secure privacy-preserving federated learning framework, named Paillier-Encrypted Federated Learning-based Detection (PE-FLD), is adopted. This architecture consists of a global centre and multiple local detection centers, which interact solely with local consumer data and subsequently communicate aggregated parameters to the global center, thereby safeguarding user privacy. Furthermore, a modified Transformer Neural Network is employed for energy theft monitoring in smart meters. Experimental validation is conducted using real energy consumption data from the Irish Electricity Metering Dataset.

This article proposes port-Hamiltonian (pH) stability analysis of synchronous reference frame-phase-locked loop (SRF-PLL) and double second-order generalized integrator-PLL (DSOGI-PLL) while accounting for the overlapping converter dynamics under low-inertia and weak-grid scenarios. The main aim is to highlight the risk of PLL interactions with the converter controllers under nonideal operating conditions. The nonlinear pH models of SRF-PLL and DSOGI-PLL are used to derive analytical stability criteria, which help monitor the effect of PLL interactions on synchronization stability. The stability criteria are substantiated through MATLAB/Simulink simulations on a 400-V Converter-Grid test system. It is shown that the stability criteria derived based on time-scale separation is inexact. In comparison, the proposed criteria, accounting for converter dynamics, offer better stability predictions and match closely with the simulation results.

Integrated electrical and heating systems(IEHS) are drawing huge attention because of their cost-effective and flexible operational capabilities. Optimal dispatch of renewable units in an IEHS is challenging because of their intermittent and stochastic nature. Uncertainties in the renewable energy source (RES) forecast, electrical, and thermal loads can cause voltage and temperature violations compromising the operational security of the system. In this paper, a rule-based dynamic dispatch is proposed to handle the increasing levels of uncertain RES and ensure the system's security. A chance-constrained (CC) co-optimization algorithm considering the uncertain loads and RES is developed. In the proposed method, electric boiler power and curtailment of RES are dynamically dispatched to reduce operational costs and network losses. The developed method is tested on the IEEE 15 bus distribution system and UK Generic Distribution System (UKGDS) 95 bus test system models. The results show that the proposed method significantly reduces the voltage deviations and temperature violations compared to the deterministic case.

The state estimation of an active distribution network (ADN) in the absence of enough measurements, and with the presence of distributed energy resources (DERs) located behind-the-meter (BTM), is a significant challenge. The operational philosophy of power distribution networks is also changing from "generation following demand" to "demand following generation". Consequently, emerging demand and generation patterns are difficult to handle through standard state estimators. This paper introduces a profiling framework that characterises responsive/flexible demands and BTM DERs to represent non-conventional load power. The impact of these non-conventional load power on distribution network monitoring and the challenges of system observability are addressed using a deep learning-based forecaster. This forecaster utilises weather and other relevant input features to tackle the irregularities in demand profiles. Additionally, the framework includes proposing a strategy for time-varying smoothing parameters in forecasting-aided state estimation to address the uncertainties associated with loads and the outputs of DERs. The framework is validated on a modified IEEE-123 bus unbalanced three-phase distribution network for demonstrating improved accuracy of the estimation.

This paper proposes a collaborative learning framework based on vertical federated learning for detecting false data injection attacks in distribution networks. The proposed framework empowers entities that are responsible for a sub-network to collaboratively construct an FDIA detection model, effectively addressing issues associated with data sharing and enabling the utilization of various measurements from each sub-network. The proposed framework enables real-time collaboration between the server and the grid edge-side by allocating the two models created through the split learning approach applied to the proposed attention-based hybrid deep learning model. The grid edge-side is tasked with extracting spatial features, while the server is responsible for extracting temporal features from the data processed by the grid edge-side. The edge-side model is designed by adopting an attention module integrated into a deep learning model while the server-side model is designed based on the Bi-LSTM model. The effectiveness of the proposed framework is demonstrated on the IEEE 123 and IEEE 37 node test systems.

An increase in the share of weather-dependent generation at low voltage levels necessitates incorporating the low-voltage network in optimizing a distribution network. Optimization in a multi-voltage network requires significant computation time and effort due to many nodes operating at different voltage levels. This research proposes a decomposition and strategic optimization method to reduce the computation requirements for such large multi-voltage distribution networks. The proposed algorithm reduces the space complexity and the computation time required for solving the optimization routines of these multi-voltage distribution networks. A virtual transformer model incorporates tap-changer as a continuous variable in the semidefinite programming power flow optimization model. The zero-duality gap condition for multiple virtual transformers is proven empirically. Compared to a centralized optimization using the same power flow model, the proposed framework reduced the computation time by 96%.