Shaotang Cai

The application of federated learning in Virtual Power Plants (VPPs) addresses the data silo issue between VPPs and enhances their ability to cope with nonlinear and stochastic scheduling characteristics, which enables VPPs better accommodate distributed energy resources and flexible loads while participating in frequency regulation services. However, although existing federated learning methods strive to solve privacy protection issues, the plaintext transmission of gradients still exposes sensitive data to the threat of curious power control centers and external inference attacks. Therefore, a privacy-protected horizontal federated reinforcement learning approach for multi-VPP optimal scheduling is proposed in this paper. Firstly, a cost-based global optimization scheduling model for multiple VPPs is constructed, modeling the internal scheduling process of VPPs as a Markov decision process. Then, an improved secure horizontal federated multi-VPP collaborative training method is presented, and local models are trained using the Deep Transformer Q-Network algorithm, with local differential privacy and CKKS homomorphic encryption implemented to ensure privacy protection. Finally, a case study is conducted using frequency regulation ancillary service market data and the IEEE-39 bus system structure. Simulation results show that the proposed approach outperforms similar algorithms, achieving high levels of privacy protection and economic operation for VPPs.

In recent years, Electric Vehicles (EVs) have be-come widely used due to their environmental friendliness. How-ever, the increase in essential supporting Electric Vehicle Supply Equipments (EVSEs) has led to a large and diverse cryptograph-ic demand for charging aggregators. The growing demand for computational power cannot be met by local cryptographic com-puting clusters, while cryptographic computing services on the cloud bring risks such as cross-VM (cross-Virtual-Machine) at-tacks. Meeting the requirements of security and timeliness simul-taneously poses challenges for cryptographic service architecture and scheduling algorithms. To address these challenges, a heter-ogeneous task processing architecture is proposed to integrate local and cloud computing resources. Next, a task queue is con-structed to transform the dynamic scheduling problem of multi-ple cryptographic tasks into a queue-jumping sequential deci-sion-making problem. By modeling security requirements as con-straints, the problem is modeled as a Constrained Markov Deci-sion Process (CMDP) formulation. A Deep Reinforcement Learn-ing (DRL) method with recurrent networks and a safe action mechanism are proposed to solve the formulation, which utilizes the temporal characteristics of cryptographic tasks to achieve better scheduling performance under constraints. Finally, com-pared with the state-of-the-art method, our proposed approach is validated through simulation experiments, demonstrating supe-rior security and enhanced processing efficiency across various scenarios.

Geographically distributed data centers (DCs) have emerged as significant energy consumers, which has led to the integration of renewable energy sources (RES) into DC power provisioning systems. However, the intermittent nature of RES and the randomness of user requests can cause significant fluctuations in DC operating power. It can be detrimental to the operation of IT equipment and lead to instability in the power grid. In this paper, aiming for tightly coupled interconnection scenarios with multi-data centers in varying regions, a multi-data center tie-line power smoothing method based on demand response is proposed. By modulating the power load of server clusters with workload scheduling, we establish a control model combined with intra-DC temporal task migration and inter-DC spatial task migration to deal with high-frequency power fluctuations. The uninterruptible power supply (UPS) battery control model is established to tackle low-frequency fluctuations. Furthermore, we design the two-stage heuristic power regulation algorithm to achieve the best practice of smoothing effect by real-time tracking of power targets after two-layer filtering. Finally, this paper performs a detailed performance simulation evaluation based on tracking data from a real DC and wind and photovoltaic (PV) new energy generation data, using four interconnected DC parks of different sizes across different regions as examples. The simulation results demonstrate that the proposed method effectively smoothing the multi-data center's tie-line power. Additionally, inter-DC temporal task migration serves as a viable solution to overcome the limitations of task migration response within a single DC, reducing the frequency of UPS battery bank charges and discharges, which in turn prolongs their service life. This approach facilitates the utilization of RES while maintaining power quality, and it also aids in reducing the escalating operation and maintenance expenses of DCs.

In islanded microgrids, the state-of-the-art distributed controller will suffer from control errors under the interference of time-varying communication delay. Towards this end, a distributed average consensus estimator is proposed to track the dynamic signal averages regardless of time-varying delay. On this basis, a distributed voltage controller resisting time-varying delay interference is proposed to precisely achieve average voltage recovery and the proportional reactive power sharing, and to constrain the voltage over-limit by introducing an additional signal, which is not possible with state-of-the-art distributed controllers. Also, the inaccurate estimation of the delay upper bound will lead to the improper parameter design, and then the controller is at risk of losing dynamic performance or instability. For this problem, the stochastic network calculus (SNC) is used to derive the tight upper bound of time-varying delay, and then the delayed discrete matrix is constructed to solve multi-delay transcendental equation, by which a parameter design algorithm is presented. Finally, the hardware-in-the-loop (HIL) experimental platform is built, and the performance of the proposed controller is tested jointly with OPNET. Compared to the state-of-the-art controller, the proposed controller has the advantages of high control accuracy, superior stability, fast convergence and no voltage crossing under the time-varying delay.

With the rapid development of the power cryptographic service, a single cryptographic computing resource, whether virtual cipher machine (VCM) or physical cipher machine (PCM), is hard to meet the security and efficiency requirements at the same time. Therefore, based on the PCM+VCM hybrid cryptographic computing platform, this brief proposes a novel multitask scheduling algorithm to satisfy the various security requirements and improve the computational efficiency of the power cryptographic service system (PCSS). Firstly, a scheduling architecture and a model for cryptographic service tasks considering security and efficiency are designed. Then, we propose a clustering mapping scheduling (CMS) algorithm based on weighted deep embedding clustering (WDEC) to implement scene partitioning and matching for cryptographic tasks. Moreover, we conduct case studies in a simulation platform that we build with S-9 PCMs and S-1 VCMs. Simulation results reveal that the trade-off between security and efficiency of task sequence is affected by the division of task execution scenarios.

In order to effectively reduce the redundant information transmission in the network, a data fusion algorithm based on extreme learning machine optimized by bat algorithm for mobile heterogeneous wireless sensor networks is proposed. In this paper, the data fusion process of mobile heterogeneous wireless sensor networks is mainly studied, and regards the nodes of wireless sensor networks as neurons in the neural network of extreme learning machines. The neural network of the extreme learning machine extracts the sensory data collected by mobile heterogeneous wireless sensor network and combines the collected sensor data with the clustering route to greatly reduce the amount of network data sent to the sink node. Aiming at the problem that the extreme learning machine randomly generates the input layer weight and the hidden layer threshold before training, the output result is unstable, affecting the data fusion efficiency and the long delay, a new method of data fusion for mobile heterogeneous wireless sensor networks based on extreme learning machine optimized by bat algorithm is proposed. Simulation experiments are carried out from two aspects: mobile heterogeneous wireless sensor networks and heterogeneous mobile heterogeneous wireless sensor networks. The simulation results show that compared with the traditional SEP algorithm, BP neural network algorithm and ELM algorithm, the proposed BAT-ELM-based data fusion algorithm can effectively reduce network traffic, save network energy, improve network work efficiency, and significantly prolong network’s lifetime.

Advance metering and reliable communication are two crucial technologies in Internet of Things, especially in the industrial application – smart grid, with which electric power system become more intelligent and easier to be controlled. And then the system can provide better quality electric power service for users, great efficiently utilize renewable energy, and decrease carbon emission. Integrated the multi-sensor fusion technology, self-organization technology and information processing technology in cloud, Cyber Physical Systems (CPS) are proposed and can apply in smart grid. Based on Time Sensitive Networking (TSN) and distributed agent technologies, this paper proposes a novel multi-QoS information flows control algorithm (MIFCA) applying in cyber power physical systems. With the algorithm, multiple electrical information data from CPS terminals can be controlled accurately, and achieve the effectively exchanging in this special IoTs. Theory analysis proves that the algorithm is convergence and validity. Using the simplified topology from "CERTS Micro-grid System" as simulation case, proposed algorithm's performance is evaluated in different traffic load levels. Bandwidth utilization ratios in light and high load levels are analyzed. The experiment results show that the proposed algorithm makes resource used more balanceable, and congestion can be avoided effectively.