Xiaohan Fang

Non-intrusive load monitoring (NILM), offers a privacy-preserving, real-time, and cost-effective approach to perceive power consumption patterns of appliances. The performance of NILM heavily relies on the quality of features extracted from load profiles. The shapelets method captures interpretable features that represent the unique characteristics of loads, proving highly relevant to NILM applications. In this work, an improved shapelets method for event-based NILM is proposed. First, a multi-length shapelets are creatively adopted for load identification, guaranteeing performance with lowfrequency or limited samples. Additionally, dynamic shapelets are obtained using a generative adversarial networks (GAN) instead of traditional search methods, ensuring efficient and distinctive shapelets generation across diverse appliances. Then, these generated shapelets are converted to distance features via shapelet transform and fed into classifier with squeeze-andexcitation network to enhance the accuracy of load identification. Experimental results on COOLL and BLUED datasets demonstrate that the proposed GAN-based multi-length dynamic shapelets method significantly optimizes the shapelet generation process and enhances the effectiveness of load identification in NILM.

The state of charge of lithium-ion batteries is a crucial parameter in battery management systems. Particle swarm optimization algorithm boasts advantages such as fast iteration speed and low computational complexity. However, a notable drawback of Particle swarm optimization is its tendency towards premature convergence. Therefore, this paper presents a novel improved Particle swarm optimization. Firstly, Lévy flight is used to generate random particles and exhibit wandering characteristics, thus effectively enhancing population diversity and avoiding the trapping of Particle swarm optimization in local optima. Secondly, the integration of dual chaos theory with the Golden Sine Algorithm optimizes the search performance of the particle swarm, with separate reconstructions of the fitness function and optimizations of the velocity update. Thirdly, the bias-correction exponentially weighted moving average method is further introduced to reduce the impact of noise and errors. It assigns reasonable weights to observation data at different time instances, enabling effective monitoring and propagation of varying errors. Ultimately, when data is used at 0°C, the root mean square error is 0.6844% and 0.4385%, respectively. The experimental results provide compelling evidence that they meet the operational requirements of battery management systems.

The inconsistency in state-of-charge (SOC) for electric vehicle batteries will cause component damage and lifespan reduction of batteries. Meanwhile, the consistency in the state-of-health (SOH) also negatively influences the consensus of SOC. To ensure the consensuses of SOC and SOH simultaneously, this paper introduces an innovative distributed optimal Kalman consensus filter (KCF) approach to battery management systems. In addition, at the stage where sensors transmit information to each other, a new event-triggering mechanism (ETM) based on dynamic information is proposed to reduce communication overhead effectively. Theoretical analysis verifies the optimality of the algorithm. By numerical simulations, the proposed event-triggered distributed optimal KCF (ET-DOKCF) method can improve the performance of SOC-SOH estimation and save communication resources.

At the microgrid secondary control level, distributed event triggering is widely studied to improve system efficiency, but the general event triggering schemes are continuous state domain. The majority of control algorithms used in current control systems, particularly system-level control algorithms, are implemented using digital controllers and digital communication in discrete time. For this reason, this paper proposes a event-triggered control strategy based on sampled data at the secondary control level for frequency recovery and power sharing in islanded microgrids. Unlike existing event-triggered control schemes, the control algorithm proposed in this paper is based on sampled data. Therefore, only at the instant of sample moment can an event be initiated or a controller updated, and the communication transfer between agents at each sampling time is no longer required, which reduces the communication and computation cost and effectively avoids Zeno behavior. Finally, a simulation model is built in MATLAB/Simulink environment to verify the effectiveness of the proposed strategy.