The new-generation artificial intelligence technologies, represented by deep learning, provided new opportunities for the digital and intelligent operation of new power systems. To deepen the understanding of the application of artificial intelligence in complex power flow calculation problems, the review and prospect of research in relevant fields were presented. Based on the current development status of the new-generation artificial intelligence technologies and grounded in power flow calculation with various scenarios, the traditional methods were summarized and the research progress of artificial intelligence techniques in power flow calculation was reviewed. The urgent challenges were analyzed and future research directions were envisioned to provide references for further applications of artificial intelligence technologies in the field of power flow calculation.

The inherent intermittency and uncertainty of renewable energy sources had a challenge to operation decisions of the system. To solve this problem, a robust unit commitment model with multi-energy coupled system considering gas-heat network dynamics was proposed. The mathematical expressions that characterize the dynamic characteristics of gas network and thermal network were established, which were incorporated into the robust unit commitment optimization model of multi-energy coupled system. A multi-dimensional uncertainty set from the perspectives of interval, time, and space to achieve flexible adjustment of wind power absorption boundaries was established. At the same time, concentrating solar power was used to replace the output of some thermal power units to further improve the utilization rate of renewable energy. The column-and-constraint generation algorithm was employed to transform the established min-max-min structure optimization model into a mixed-integer linear programming master-subproblem form for optimization, improving the solution speed of the model. The effectiveness of the proposed model and method was verified on 6-6-8 and 118-20-16 electricity-gas-heat systems, with results indicating that the dynamic characteristics of gas and heat networks can improve the economy of system operation and the utilization rate of renewable energy.

Aiming at the problem that large-scale offshore wind farms were connected to the grid through the frequency decoupling control of modular multilevel matrix converter(M³C), which caused the inertia level of offshore low frequency wind power system to decrease, a coordinated inertia response control strategy for offshore low frequency wind power system based on adaptive virtual inertia of M³C was proposed. The adaptive virtual inertia control strategy of M³C was proposed, which used the change information of the capacitor voltage of M³C submodule to adjust the size of the virtual inertia time constant. The linear coupling method between the frequency of the grid frequency side of M³C and the frequency of the low frequency side of M³C was proposed, so that the offshore wind farm could respond to the change of the system frequency through the frequency of the low frequency side, which provided the inertia support of the system together with M³C. The simulation results showed that the proposed adaptive control strategy could enhance the inertia response capability of M³C, avoid the system frequency second fall, and improve the frequency stability of the system.

To effectively identify the key links in complex power system, a critical line identification method considering source fault state and secondary fault risk was proposed. In the source fault phase, weighted power flow entropy, weighted grid entropy, and weighted power flow impact entropy—three indicators for evaluating the system state—were proposed to characterize the source fault level. Load loss risk and grid loss risk—two indicators for characterizing the secondary fault risk—were combined to evaluate the development of power system faults. The objective entropy weight method determined each metric's weight, producing composite indicators for critical line identification. Simulations were conducted on the IEEE 118-bus test system. The results demonstrated that attacks on the critical lines identified by this method led to the largest decline in system network efficiency, which was consistently lower compared to other methods, confirming the validity of the proposed approach.

Traditional distribution network line loss calculation methods can not simultaneously consider the causal logic of the physical model and the computational accuracy advantage of the data model, and are unable to accurately calculate the active distribution network line loss under the access of different subjects. Therefore, this paper proposed a hybrid data-mechanism driven active distribution network line loss calculation method based on topology gating. Firstly, the correlation analysis of active distribution network line loss features after different subjects' access was performed based on random matrix theory, and the line loss features with the strongest correlation with line loss rate were obtained. Then, based on the mechanism model and data model of traditional line loss calculation, topological gating was used to fuse the causal logic of the mechanism model with the prediction accuracy of the data model to establish a hybrid data-mechanism driven line loss calculation model based on topological gating. The active distribution network in a region with distributed power, energy storage, and electric vehicle charging pile access was taken as an example for analysis, and evaluation indexes such as average absolute error and root mean square error were used to assess the calculation accuracy of the model. The example results showed that the method proposed in this paper had high calculation accuracy and could be well adapted to the calculation of line loss of active distribution network after the access of different subjects.

Aiming at the problems of privacy protection, computational efficiency, and predictive accuracy in existing forecasting methods for electric vehicle charging stations, a cloud-edge collaborative and graph neural network based load forecasting approach was proposed. A privacy preserving module based on embedding is developed in the cloud, which reconstructs the input data through embedding transformation to prevent potential privacy leakage risks. A method for generating representation with graph structure based on clustering is proposed to provide additional spatiotemporal information and achieve more accurate forecasting. Personalized graph neural network forecasting models are designed for clients based on cloud's graph structure representation, enabling collaborative training of electric vehicle charging stations in different regions while protecting privacy. Experimental results on the Perth dataset demonstrate that the model outperforms benchmark methods in predictive accuracy and that the cloud-edge collaborative framework proposed in this study significantly enhances the performance of graph neural network algorithms in the task of load forecasting for electric vehicle charging stations.

Accurately predicting the output and ramp rate of photovoltaic(PV)systems during an eclipse is of significant importance for alleviating peak-shaving pressure and maintaining the balance of power consumption. A method for constructing an intra-day solar radiation intensity model was proposed, which characterized the daily characteristics of solar radiation intensity at a certain location. Subsequently, models for the shading percentage of the sun during total, annular, and partial eclipses were respectively established to update the intra-day solar radiation intensity model. Next, the PV generation model considering the impact of temperature on efficiency was introduced, along with the ramp rate and calculation models for PV units.In the case study section, the annular eclipse event that occurred on June 21, 2020, in China was used as an example to retrospectively analyze the output of a 100 kW PV station, thereby validating the effectiveness of the proposed method. Furthermore, the study explored the impact of different percentages of eclipse obscuration on the PV ramp rate, indicating that eclipses with greater obscuration corresponded to higher PV ramp rates. The PV ramp rate under a total solar eclipse was found to even reach more than 4 times the normal situation, and dispatch personnel were required to attach great importance to it.

The grid-connected capacity of new energy sources with low emissions, low inertia, and high volatility was rapidly increased, significantly altering the dynamic characteristics of the power system, such as frequency response. This posed great challenges to safety and stability analysis and control. While the high proportion of wind, solar, and storage created operational challenges, opportunities for coordinated control among them were also presented. The mechanism by which energy storage improved stability was explored, considering the characteristics of different types of distributed resources and the control requirements in emergency situations. The characteristics of various controllable resources were analyzed to fully utilize the fast response characteristics of energy storage. These resources were classified based on support time, and the charging and discharging power of the energy storage system was determined, alongside adjustments to control strategies. A hierarchical coordinated operation and control strategy for controllable resources, based on the rapid adjustment of energy storage power, was proposed. The effectiveness of the proposed strategy was verified using a system with multiple types of units as an example.

With the deepening implementation of the Belt and Road Initiative(BRI), artificial intelligence(AI)has been increasingly applied in several key fields, including risk assessment and prediction, public opinion sentiment analysis and hot topic mining, transportation logistics and trade optimization, environmental protection, and sustainable development, as well as cultural communication and education. Artificial intelligence, through technologies such as deep learning, machine learning, and natural language processing, has significantly improved risk prediction capabilities and management accuracy, optimized logistics network efficiency, facilitated green development, and enhanced cross-cultural communication. This study systematically reviewed the research progress of artificial intelligence in various key fields of the Belt and Road in recent years, focusing on specific achievements and innovations in practical applications. The limitations of current research, such as low data acquisition quality, insufficient interpretability of artificial intelligence models, and difficulties in cross-domain collaboration and application transformation were analyzed. The future research directions were proposed.

To study the optimal task allocation strategy and path planning problem of ground vehicles and unmanned aerial vehicles(UAVs)in the inspection process, an improved adaptive large neighborhood search(IALNS)algorithm—a two-stage hybrid heuristic algorithm—was proposed. In the first stage, a clustering algorithm was used to divide the target nodes according to the different demand levels and distances of the nodes to be inspected. In the second stage, a hybrid heuristic algorithm was used to solve the route scheduling problem. Six new local optimization operators were added, and a node redistribution strategy was introduced. The cooperative hybrid route with the minimum cost for vehicles and UAVs was obtained after iterations. The proposed algorithm solutions and other algorithm solutions were tested and comparatively analyzed, and the experimental data showed that the IALNS algorithm had significant advantages in solving the vehicles-UAVs cooperative inspection problem.

To address the limitations in semantic context utilization and spatio-temporal feature modeling in existing anomaly detection methods, a video anomaly detection method based on video caption augmentation and dual-stream feature fusion was proposed. Video captions were automatically extracted and encoded using the contrastive language-image pre-training(CLIP)model to serve as auxiliary semantic context information for anomaly detection. A spatio-temporal adaptive embedding module was introduced to capture subtle temporal variations and complex spatial structures within videos, enabling effective spatio-temporal feature fusion. A cross-modal alignment module was further designed to deeply integrate contextual semantic features with spatio-temporal visual features, allowing more accurate capture of joint spatio-temporal-semantic representations of anomalous events. Experimental results showed that the method achieved area under the curve A_UC scores of 97.54% on the ShanghaiTech dataset and 90.54% on the CUHK Avenue dataset. The results confirmed the performance and robustness of the method across multiple public video anomaly detection datasets, providing an effective solution for this critical task.

To optimize the application of HLC composite steel piles in foundation pit support and solve the problem of unclear soil pressure sharing between H-shaped steel and steel sheet piles in the existing design, this study adopted a combined method of theoretical analysis, on-site measurement, and numerical simulation. The calculation formulas for the moment of inertia of HLC composite steel piles in ideal and actual states were derived. The calculation methods for the section stiffness reduction coefficient and the synergy coefficient were proposed. The variation law of the stiffness reduction coefficient of HLC composite steel piles along the depth was analyzed to characterize the working performance of the overall structure. The results showed that the earth pressure sharing ratio of each supporting pile did not conform to the theoretical stiffness distribution law, with that of a single steel sheet pile reaching up to 10%. The actual supporting stiffness should take into account the interaction between the supporting units. This study could provide a theoretical reference for the design of cantilever HLC foundation pit support structures.

In order to study the durability of steel slag fine aggregate concrete(SSC)under the action of dry-wet cycle of salt solution, the optimal volume fraction of steel slag and fly ash in concrete was determined by mechanical property tests, and the dry and wet durability test of salt solution was carried out on concrete samples. The effects of the volume fraction of steel slag and fly ash on the appearance, quality, dynamic elastic modulus, and compressive strength of concrete were considered comprehensively. The results showed that when the volume fraction of steel slag was 30%, the compressive strength, flexural strength, and dynamic elastic modulus of SSC were better than those of ordinary concrete. The mechanical properties of SSC were further improved after 10% volume fraction of fly ash was added. After 100 dry-wet cycles of salt solution, the mass loss rate, compressive strength loss rate, and dynamic elastic modulus loss rate of steel slag with 30% volume fraction were smaller than those of ordinary concrete. The addition of 10% volume fraction fly ash could delay the deterioration of SSC durability, and the durability of SSC in dry and wet cycling of single salt solution was better than that of double salt solution. The damage evolution equation was established with the dynamic elastic modulus damage as the index, which could describe the deterioration of SSC properties under the dry-wet cycle of salt solution.

The damage pattern of the specimen and the basic physical parameters of the soil fill were determined and analyzed through the conventional triaxial test on the compartment reinforced soil after soaking in acid and alkali solutions. Combined with the stress-strain curve obtained from the triaxial test, the stress-strain response model of the compartment reinforced body was verified, and then the changes in stress-strain response and apparent cohesion of the compartment reinforced soil with equivalent diameters of 0.34 m and 0.45 m under acid and alkali environments were investigated and analyzed by using the solidification and drainage triaxial test. The research results showed that triaxial test obtained compartment reinforced soil Duncan-Zhang model parameters applied to the bias stress-axial strain, body strain-axial strain relationship curve with the test curve almost match, that was, the derivation of the formula calculations in line with the test data; stress-strain response analysis, PET than PP compartment reinforced soil acid and alkaline environment, and the PET material two sizes of the compartment reinforced soil structure of the acidic environment more sensitive to alkaline, equivalent diameter of 0.45 m than 0.34 m to acid and alkali response, easy to produce shear damage; two diameters of PET compartments reinforced soil in the acid environment, the apparent cohesion decreased by 9.2% and 7.8%, respectively, and the alkali environment was less affected. While the PP material diameter of 0.34 m was not affected by acid and alkali, only the diameter of 0.45 m was reduced by 13.3% in the alkali environment, and was not affected by the acid environment.

In response to the problem of unclear patterns of the influence of induced materials on the microwave heating performance of asphalt mixtures, coarse basalt and middle basalt asphalt mixture, coarse steel slag and middle steel slag asphalt mixture, coarse basalt and middle steel slag asphalt mixture, and coarse steel slag and middle basalt asphalt mixture with different dosages of steel fibers had been designed and prepared. The influence of steel slag and steel fiber on the heating rate and uniformity of asphalt mixture was studied by microwave heating test. In addition, rutting test, low-temperature bending test, water-soaked Marshall test, volumetric stability test and four-point bending fatigue test were used to evaluate the pavement performance of steel slag asphalt mixtures. The results found that steel slag significantly improved the microwave heating efficiency of asphalt mixtures, but reduced the heating uniformity. The effect of steel fiber on the microwave heating efficiency of asphalt mixtures was not obvious, and the heating uniformity showed the phenomenon of first improvement and then deterioration with the increase of steel fiber dosage. The incorporation of steel slag could improve the high and low temperature performance as well as fatigue resistance of asphalt mixtures, but it adversely affected the volume stability and water stability of the mixtures. This study preliminarily confirmed the feasibility of steel slag used as induced-heating self-healing asphalt pavement aggregate, showing promising application potential.

To investigate the damage that thickness defects caused to the initial support structure, this study adopted an actual engineering background, considered the specific locations and extents of thickness defects, and conducted a statistical analysis of their distribution characteristics within the tunnel. Finite element numerical simulation was employed to deeply analyze the mechanical characteristics of the initial support structure under different types of thickness defects. Additionally, the CRITIC(criteria importance through intercriteria correlation)method was used to comprehensively evaluate various risk indicators(such as the span and location of defects)affecting the safety of the initial support structure, determining the degree of influence of each indicator on structural safety. Research indicated that either located at the vault position or spanning more than 25 degrees exerted the most significant impact on the safety of the primary support structure. The research results could provide references and insights for the prevention and control of similar tunnel diseases.