Optimal Operation of District Cooling System with Multiple Energy Stations and Energy Storages

District Cooling Systems (DCS) provide centralized chilled water to multiple buildings, offering high efficiency, environmental sustainability, and flexibility in power systems. As urban DCS networks expand, formerly isolated cooling islands are becoming interconnected, leading to increased topological complexity and evolving operational modes. Optimizing DCS operation has grown more challenging due to enhanced system interconnectivity, spatial-temporal coupling of thermal energy storage, and the thermal-hydraulic characteristics of the cooling network. To address these complexities, this paper proposes an optimal power dispatch model for DCS under a time-of-use electricity tariff. By integrating a network reduction technique with a spatial-temporal decomposition approach, the model optimizes the operation of chillers, thermal energy storage units, and cooling networks while maintaining thermal-hydraulic balance. Numerical simulations on a real-world DCS validate the model’s effectiveness. Results indicate that the proposed method achieves over 5% energy savings and reduces energy costs by more than 10%, while obtaining a computationally efficient solution within five minutes.

AI-based Monitoring and Risk Prevention of Electricity Price Anomaly

Electricity markets exhibit high volatility due to the non-storability of electricity, real-time balancing needs, and network constraints. These physical characteristics often result in price anomalies such as extreme spikes or negative prices, which pose challenges to traditional econometric models. While statistical methods can capture historical trends, they typically overlook grid dynamics, spatial dependencies, and rapid shifts caused by unexpected events. Artificial intelligence (AI) offers new opportunities to monitor and prevent such anomalies. By learning nonlinear patterns and integrating multi-source data, AI can provide real-time anomaly detection and risk assessment. This study focuses on AI-based approaches to identify electricity price anomalies and understand their underlying causes, aiming to support more resilient and informed market operations.

Real-time Operation of Electric Autonomous Mobility-on-Demand System Considering Power System Regulation

Electric autonomous mobility-on-demand (EAMoD) systems are rapidly emerging worldwide. However, their concentrated depot charging may strain the power system, which in turn influences their own charging decisions. To address this risk, this paper proposes a real-time coordination framework between EAMoD fleets and the power system. First, a Markov decision process models the temporal-spatial dynamics of EAMoD fleets, including trip service, repositioning, and charging. Second, a power system regulation model imposes real-time charging constraints to prevent overload and undervoltage. To tackle the solution challenges posed by the EAMoD system’s complex action space, a piecewise linear-based approximate dynamic programming approach integrated with model predictive control is developed. Numerical experiments on the Manhattan transportation network and a 14-node power distribution system validate the effectiveness of the proposed method and assess the impact of depot charging on the power system, with and without regulation constraints.

A Power-Decoupled Three-Phase Current Source Inverter for Low-Voltage-Ride-Through in Wind Power Generation

With the widespread integration of intermittent wind energy, grid-connected wind turbines must possess the capability to maintain frequency stability during contingent events. In unbalanced grid fault scenarios, negative sequence voltage will induce double-line-frequency power oscillations, resulting in extra power loss on the interfacing systems, while also generating electromagnetic torque oscillations in Permanent Magnet Synchronous Generators (PMSGs). To tackle this issue, a novel three-phase power-decoupled current source inverter (PD-CSI) is proposed. The proposed inverter can actively divert the double-line-frequency power to a ripple storage capacitor, thereby maintaining the torque stability of the PMSG. For the PD-CSI in wind power LVRT applications, the elaboration of its circuit topology and mathematical model, the design of a simplified modulation scheme and the formulation of comprehensive control strategies are proposed in this paper. Ultimately, through simulations and Hardware-in-the-Loop (HIL) experiments, the functionality of the proposed topology to divert oscillatory power is demonstrated.

A Multi-Frequency Equivalent Harmonic Modeling Method for Digital SPWM in MMC

This work presents a multi-frequency equivalent output voltage harmonic modeling method for sinusoidal pulse width modulation (SPWM) of digitally controlled modular multilevel converter (MMC). The model utilizes double Fourier integrals and considers both symmetric and asymmetric regular sampling processes. The interaction between a multi-frequency modulation waveform and triangular during the modulation process is fully analyzed from a mathematical perspective so that the combined effects of digital sampling and submodule (SM) capacitor voltage ripple on the output voltage are summarized in time and frequency domains. The characteristics and performance of the output voltage harmonic spectrum are investigated through analytical analysis, simulations, and experiments. The verification results demonstrate that the modeling method provides an accurate estimation of output voltage harmonics for digital SPWM at various cases.

Post-Storm Tide Power Grid Fault Recovery Method Based on Coordinated Dispatching of MESS-DV

Coastal urban distribution networks face significant risks from storm-tide-induced disruptions, leading to large-scale outages and socioeconomic losses. Existing research lacks comprehensive frameworks integrating interdependent infrastructure systems for post-storm recovery. This study proposes a resilience-oriented methodology leveraging coordinated scheduling of mobile energy storage systems (MESS) and drainage vehicles (DV). The approach incorporates multi-coupling characteristics among storm-tide risk zones, power grids, transportation, and drainage systems through an uncertainty-aware simulation environment. It models active/passive drainage interactions in low-lying areas with water-immersion-based safety constraints and employs event-driven multi-agent reinforcement learning (MARL) for real-time adaptive coordination. Experimental results show active drainage reduces total restoration time by ~6 hours compared to passive strategies, achieving full power resumption within 7 hours across all scenarios without post-restoration blackouts. This highlights the value of cross-domain collaboration and AI-driven methods in enhancing climate-resilient urban infrastructure.

Enhancing the Self-Healing Capability of Cyber-Physical Distribution Systems: A UAV-5G Hybrid Wired/Wireless Communication Method

Unmanned aerial vehicle (UAV)-assisted wireless communication is emerging as a pivotal technique to facilitate the self-healing of cyber-physical distribution systems (CPDSs) after disasters. However, owing to the limited resources of UAVs, wireless communication is unable to cover all the disconnected distribution terminals, which impedes the automatic load pick-up. To tackle this problem, this letter proposes a novel UAV-5G hybrid wired/wireless communication method based on hybrid communication networks and UAV-5G base stations. Numerical studies verify that the proposed method efficiently optimizes UAV utilization for the timely reconnection of intact terminals, further enhancing the self-healing capability of CPDSs.

Powering the Skies: How Power Systems Enable the Low-Altitude Economy

The low-altitude economy, driven by unmanned aerial vehicle (UAV) and electric vertical takeoff and landing aircraft (eVTOL), faces significant energy challenges that require power system support. This talk explores the role of modern power infrastructure in enabling scalable and safe operations for these emerging technologies. Key focus areas include optimizing grid architectures to manage dynamic loads from electric aircraft, integrating renewable energy sources for sustainable charging networks, and improving grid flexibility to accommodate urban-rural hybrid energy demands. By analyzing technical barriers and potential solutions, this talk highlights the critical connection between advanced power systems and the growth of low-altitude applications, emphasizing their collective impact on economic efficiency and environmental sustainability.

Harmonic Regulation and Optimization for Medium-Voltage Modular Multilevel Converters with PSC-PWM

Modular multilevel converters (MMCs) play an important role in modern power systems. This presentation mainly focuses on how to regulate and optimize the harmonics of the MMC. For the underlying modulation stage, the first work comprehensively reveals the benefits of using different carrier arrangements between different phase legs of the MMC. Two optimal working modes are proposed for high-frequency harmonic minimization of the line-to-line voltage or common mode voltage. Furthermore, flexible trade-offs between them can also be achieved by solving the proposed optimization problem. On the other hand, for the upper-level control stage, the second work proposes an easy-to-implement multi-harmonic injection (MHI) method for MMC, which can effectively reduce the peak arm current with minimally increased cost in capacitor voltage ripples. By lowering the peak arm current, the proposed method can allow MMCs to handle larger output currents using the same current-rated switches, enhancing the system’s overall power-handling capability. The simulation and experimental results verify the effectiveness of the proposed modulation and control methods.

A Quasi-Z-source based fault-tolerant PV micro-inverter: design and control

Sustainable power production of low-carbon PV systems is pivotal for alleviating the energy crisis. However, frequent failures of photovoltaic (PV) panels and DC-link capacitors of inverters fundamentally compromise the reliability of PV systems. To address these issues, a quasi-Z-source-based dual-input PV inverter is proposed first. This inverter can achieve active power decoupling (APD), differential PV module currents operation, and imbedded shoot-through function, thereby enhancing both fault-tolerant operation and reliability. Notably, the proposed topology uses only one additional switch while achieving multiple complex functions. The talk begins with an in-depth exposition of the topology design, operation modes, and the modulation strategy. Subsequently, state equations for this nonlinear system are established and the multi-functional controller is designed.

One Quarter Each (on Average) Ensures Proportionality

We consider the problem of fair allocation of m indivisible items to a group of n agents with subsidy (money). Our work mainly focuses on the allocation of chores but most of our results extend to the allocation of goods as well. We consider the case when agents have (general) additive cost functions. Assuming that the maximum cost of an item to an agent can be compensated by one dollar, we show that a total of n/4 dollars of subsidy suffices to ensure a proportional allocation. Moreover, we show that n/4 is tight in the sense that there exists an instance with n agents for which every proportional allocation requires a total subsidy of at least n/4. We also consider the weighted case and show that a total subsidy of (n-1)/2 suffices to ensure a weighted proportional allocation.

Revisiting Proportional Allocation with Subsidy: Simplification and Improvements

In this paper, we revisit the problem of fair allocation with subsidy. We first consider the allocation of m indivisible chores to n agents with additive cost functions. Under the assumption that the maximum cost of an item can be compensated by one dollar, Wu et al. (WINE 2023) showed that a total of n/4 dollars suffices to guarantee a proportional allocation, by rounding fractional allocations. Their subsidy guarantee is optimal when n is even. For odd n, there is still a small gap between the upper and lower bounds for the total subsidy. In this paper, we propose a much simpler algorithm for the problem, which does not require rounding fractional allocations, and achieves an optimal subsidy guarantee for all values of n. We further show that our algorithm and analysis framework can be extended to the mixture of (subjective) goods and chores, achieving the optimal subsidy guarantee.

HyperFM: Fact-Centric Multimodal Fusion for Link Prediction over Hyper-Relational Knowledge Graphs

With the ubiquity of hyper-relational facts in modern Knowledge Graphs (KGs), existing link prediction techniques mostly focus on learning the sophisticated relationships among multiple entities and relations contained in a fact, while ignoring the multimodal information, which often provides additional clues to boost link prediction performance. Nevertheless, traditional multimodal fusion approaches, which are mainly designed for triple facts under either entity-centric or relation-guided fusion schemes, fail to integrate multimodal information with the rich context of the hyper-relational fact consisting of multiple entities and relations. Against this background, we propose HyperFM, a Hyper-relational Fact-centric Multimodal Fusion technique. It effectively captures the intricate interactions between different data modalities while accommodating the hyper-relational structure of the KG in a fact-centric manner via a customized Hypergraph Transformer. We evaluate HyperFM against a sizeable collection of baselines in link prediction tasks on two real-world KG datasets. Results show that HyperFM consistently achieves the best performance, yielding an average improvement of 6.0-6.8% over the best-performing baselines on the two datasets. Moreover, a series of ablation studies systematically validate our fact-centric fusion scheme.

CasFT: Future Trend Modeling for Information Popularity Prediction with Dynamic Cues-Driven Diffusion Models

The rapid spread of diverse information on online social platforms has prompted both academia and industry to realize the importance of predicting content popularity, which could benefit a wide range of applications, such as recommendation systems and strategic decision-making. Recent works mainly focused on extracting spatiotemporal patterns inherent in the information diffusion process within a given observation period so as to predict its popularity over a future period of time. However, these works often overlook the future popularity trend, as future popularity could either increase exponentially or stagnate, introducing uncertainties to the prediction performance. Additionally, how to transfer the preceding-term dynamics learned from the observed diffusion process into future-term trends remains an unexplored challenge. Against this background, we propose CasFT, which leverages observed information Cascades and dynamic cues extracted via neural ODEs as conditions to guide the generation of Future popularity-increasing Trends through a diffusion model. These generated trends are then combined with the spatiotemporal patterns in the observed information cascade to make the final popularity prediction. Extensive experiments conducted on three real-world datasets demonstrate that CasFT significantly improves the prediction accuracy compared to state-of-the-art approaches.

Topological Learning for Spatio-Temporal Prediction

In spatio-temporal prediction tasks, topological learning plays a crucial role in capturing the complex spatial-temporal relation between nodes. Topological information can be classified into two types: Group-level and Individual-level. Group-level topological information, derived from sensor clusters, reflects the spatial arrangement of sensors and provides a topology-distance relation for spatio-temporal prediction. Individual-level topological information, provided by users, reflects the historical interaction between users and items, offering a topology-preference relation for spatio-temporal prediction. In this talk, I will first introduce what topological learning is and why it is important for spatio-temporal prediction. Then, I will propose specific topological learning strategies to better capture two types of topological relations: (1) Group-level, with an application in traffic speed prediction, I propose the Dual Cross-Scale Transformer to capture topology-free, hierarchical, and dynamic topological relations between roads. (2) Individual-level, with an application in POI recommendation, I propose the Preference Evolution Transformer to capture the asymmetric and dynamic topological relations between users, as well as the general and unique topological relations between users and POIs. Finally, I will conclude the talk and look ahead to the future of applying topological learning in more applications.

Robust Time Series Forecasting Under Distribution Shift

Time series forecasting is a common task where models learn temporal and spatial information from historical observations to predict future values. However, distribution shift frequently occurs in time series, causing significant performance degradation in forecasting models. To address this issue, a common approach is to decouple time series into invariant and variant patterns, learning them separately and then combining them for forecasting. Since distribution shift is pervasive in time series, it needs to be properly addressed across various tasks such as offline forecasting, Variable Subset Forecasting (VSF), and online forecasting. This presentation will share effective strategies for handling distribution shift in these different forecasting scenarios.