Integrating artificial intelligence (AI) into photovoltaic (PV) systems has become a revolutionary approach to improving the efficiency, reliability, and predictability of solar power generation. In this paper, we explore the impact of AI technology on PV power generation systems and its applications from a global perspective. Central to the discussion are the pivotal applications of AI in maximum power point tracking (MPPT), power forecasting, and fault detection within the PV system. On the one hand, the integration with AI technology enables the optimization and improvement of the operational efficiency of PV systems. On the other hand, new challenges have been observed, mainly in the areas of data processing and model management. Moreover, advances in AI technology and hardware upgrades will lead to the rapid global popularization of new energy sources such as solar energy, which is expected to replace traditional energy sources. Finally, we describe forward-looking solutions including transfer learning, few-shot learning, and edge computing, as well as the state of the art.

The intermodal express network of high-speed railways and highways can fully utilize the flexibility of highways and the advantages of high-speed railways, such as low cost, high efficiency, and low carbon emission. This paper studies the hub location and flow assignment problem in the intermodal express network of high-speed railways and highways, which can not only increase the transportation efficiency but also provide door-to-door service. Considering the characteristics of multiple modes, flow balance, carbon emission, capacity constraints, and time constraints in the intermodal express network, a mixed-integer linear programming model is proposed with the objective of minimizing the total cost by determining the hub locations, allocations, mode selections, and flow assignments. Owing to the NP-hard computational complexity, an improved genetic algorithm with local search is designed by combining the genetic operators and two optimization strategies to solve the problem effectively. Lastly, numerical experiments are conducted to validate the feasibility of the model and the effectiveness of the algorithm.

This paper investigates the effective management of distributed Renewable Energy (RE) sources in multi-user Park-level Integrated Energy Systems (PIESs), aiming to coordinate power supply, cooling, and other forms of energy through a Virtual Power Plant (VPP) while enhancing environmental and economic benefits in regulation markets. Beyond the uncertain influence of solar PhotoVoltaic (PV) power on VPP operation, the challenge exists in establishing carbon trading and Integrated Demand Response (IDR) mechanisms to enable internal regulation of VPP. To address the above problem, a data-driven two-stage Distributed Robust Optimization (2-DRO) model is proposed. The overall goal is to minimize the operating costs of VPP by maintaining a supply-demand balance in PIESs along with protecting the electricity for production tasks. A case study of an actual PIES in eastern China reveals that our model is more practical and cost-effective. It integrates the deployment of both electricity and cooling energy under the mechanism of IDR and carbon emissions, realising flexible scheduling for multi-scenario operation of PIES.

This paper introduces the sum-of-squares 𝑘-facility location problem (SOS-𝑘-FLP), a generalization that unifies the classical 𝑘-means problem and the sum-of-squares facility location problem. An instance of the SOS-𝑘-FLP is specified by a set $\chi⊂ ℝ^𝑑$ of 𝑛 clients, a uniform opening cost 𝑓(> 0) for any potential facility location in $ℝ^𝑑$ , and an integer 𝑘(≥ 1). The goal is to select some facility locations as a set of centers $S⊂ ℝ^𝑑$ with $|S| ≤ 𝑘$ and to partition $\chi$ into 𝑘 clusters, thereby minimizing the total cost, which which comprises both the center opening cost and the squared distances from each client to its assigned facility. We propose a (9 +√91 + 𝜀)-approximation algorithm for the SOS-𝑘-FLP, leveraging local search and scaling techniques. Empirical evaluations show that integrating the proposed algorithm with an iterated version of Lloyd’s method yields a practical heuristic that is both efficient and effective.