Energy Foundation

The rapid proliferation of rooftop photovoltaics and behind-the-meter battery storage is creating systemic risks in distribution networks. This includes local network constraint violations, insufficient outage resilience, and enlarged cybersecurity attack surfaces, leaving millions of demand-side distributed energy resources (DERs) to operate without coordinated oversight. Existing solutions (dynamic operating envelopes, virtual power plants, peer-to-peer trading, and single-household energy management systems) each address only partial aspects of this challenge. Based on multi-agent reinforcement learning, this paper proposes a Decentralised Coordination Framework (DCF) that organises the energy dispatch into a three-tier hierarchical architecture. Specifically, it has a user layer executing Proximal Policy Optimisation (PPO) for local dispatch, a feeder layer in which a dynamically elected L1 leader applies MADDPG to coordinate community flexibility via a Virtual Aggregation Unit (VAU); and a cross-feeder layer where an L2 leader manages inter-community balancing and market interfaces. It integrates a directed acyclic graph (DAG)-based verifiable execution ledger, Paillier homomorphic encryption, and LLM-based anomaly detection to enhance security. Potential market participation pathway and revenue distribution mechanism are proposed to align with the Australian National Electricity Market. The DCF provides a scalable, market-ready foundation for commercial demandside DER deployment under high renewable penetration.

As the share of nuclear power plants in the global energy mix increases, ensuring the safety of nuclear power plants has become a crucial task. Existing radiation monitoring technologies face problems such as poor data reliability, difficult data fusion and low accuracy of anomaly detection. In this paper, a machine learning-based method for radiation monitoring data fusion and anomaly identification in nuclear power plants is proposed, which effectively fuses heterogeneous data from multiple sources of multiple sensors by adopting deep learning models such as self-encoder and convolutional neural network (CNN) and improves the accuracy and robustness of the monitoring data. Experimental results show that the method significantly outperforms traditional methods in terms of data fusion accuracy and anomaly detection accuracy, especially when dealing with high-dimensional and complex nonlinear data. Meanwhile, this paper also discusses the challenges faced by deep learning models in practical applications, including computational complexity, model interpretability, and real-time issues. The results validate the great potential of machine learning-based radiation monitoring methods in improving the performance of nuclear power plant safety monitoring systems and provide technical support for building smarter and more reliable monitoring systems in the future.

ABSTRACT Corrosion limits structural reliability in molten salt Concentrated Solar Power (CSP) systems. This review categorizes mechanisms into chemical, electrochemical, and mechanical modes. Chemical corrosion varies by environment: nitrate salts enable protective oxide films (e.g., Cr₂O₃, FeCr₂O₄) on Cr‐steels, while chloride systems require Ni‐based alloys with high Mo/W and strict impurity control to suppress Cr volatilization. Electrochemical corrosion is impurity‐driven in chlorides (H₂O/O₂/Cl⁻ causing film breakdown and acidification) but milder in nitrates. Notably, welded joints are vulnerable; microstructural inhomogeneity induces galvanic cells, accelerating Heat‐Affected Zone (HAZ) erosion. Mechanical corrosion involves flow delamination, thermal flaking, and stress‐corrosion, where initial Cr diffusion eventually promotes intergranular cracking. Since static testing underestimates synergistic effects, future research must prioritize dynamic coupled testing and multiscale modeling to ensure high‐temperature reliability.

Based on the geological conditions of the auxiliary transportation roadway 42307 in Wanli No. 1 Mine and the rapid tunneling technology using a combined excavation and bolting machine, it was found that the short parallel working time of excavation and support, as well as the low level of intelligence in the integrated excavation-bolting equipment, are the bottle necks limiting further improvement of the tunneling speed. Through theoretical analysis, engineering analogy, and numerical simulation, a mechanical model for roof stability analysis was established, yielding a limit unsupported span of 1.8 m at the tunnel face. A strong support strategy with reduced bolt density was proposed. The tunnel bolt support system was optimized by adopting high-performance bolts, high pre-tension force, and high system stiffness. Furthermore, countermeasures to improve tunneling speed were provided from aspects including the addition of an intelligent navigation system for the combined excavation-bolting machine, upgrading auxiliary equipment, and equipment modification. Field measurements showed that the optimized support system increased the parallel working time of excavation and support, improving the monthly tunneling advance from 840 m to 1142 m, a 36%increase. Meanwhile, the roof subsidence and convergence of the two ribs were 87 mm and 55 mm respectively, indicating good surrounding rock control.

ABSTRACT This paper introduces a new semiparametric spatial regression model with unknown forcing terms (SSRUT) in the prior information of the nonparametric function for spatial anisotropic data. The estimators are obtained by the proposed penalized least squares method which incorporates partial differential equation (PDE) penalization with mixed finite elements. In particular, the unknown coefficient functions in PDE which capture the spatial anisotropy, are chosen by our developed penalized spatial local least squares method. The corresponding implementation procedure is also developed. Extensive simulations and a real data analysis are conducted to validate the effectiveness of the proposed estimation approach.