A fault occurring in a section of a power grid causes the corresponding protective relays to trip the related circuit breakers and cut off the fault current, which results in one or more areas of power outage. The efficiency of fault diagnosis programs will be greatly promoted if the faulty section is restricted within the outage area. Conversely, the outrage area will not be identifiable if the control center fails to receive alarm messages from all circuit breakers, and fault diagnosis will need to be conducted for all fault sections. Furthermore, incorrect diagnosis results may be obtained because of missing alarm messages. This work applies graph theory to partition one or more outrage areas according to whether graph vertices are a cut set of the graph. A multi-objective optimization method based on skyline query is applied to sort outrage areas according to their correlative character of protection, which alleviates the need to conduct fault diagnosis for all sections. The sorted results are then provided to a fault diagnosis program. The developed diagnosis method can therefore increase the speed of diagnosis programs, and can insure that fault sections are not omitted owing to missing alarm information. Petri net is applied as the fault diagnosis tool as an example. Simulations involving several fault examples demonstrate the benefits of the proposed fault diagnosis method.

To address the personalized knowledge demands in converter transformer operation and maintenance (O&M) and the accuracy limitations of general large models in specialized Q&A, this paper proposes and implements a professional Q&A system integrating semantic understanding, knowledge reasoning, and generative AI. The system is supported by a converter transformer O&M knowledge graph, with deep semantic parsing performed by open-source large models during query processing. The knowledge retrieval phase employs a dynamic CYPHER query generation algorithm to automatically construct multi-hop relational schema matching statements, enabling precise structured knowledge retrieval. The results are then transformed into natural language responses through retrieval-enhanced generation technology. Experimental validation demonstrates that the system achieves significantly higher response accuracy in converter transformer O&M scenarios compared to general large models, providing an efficient and precise technical solution for intelligent Q&A in vertical domains.

To address issues in the training of distribution network O&M (Operation and Maintenance) personnel-such as the lack of real-scenario support, insufficient practical operation, and incomplete evaluation function, a work orderdriven active operation and maintenance training system is designed for intelligent distribution network operation. Firstly, the development background and research status of work order-driven systems is investigated in distribution networks. Secondly, the composition structure of the practical training system, including the Web main station terminal, mobile practical training APP terminal, and multi-system interaction interfaces is proposed, while extracting the core structural elements of typical work orders. Then, it constructs a multi-dimensional work order execution evaluation model covering 10 key indicators (e.g, information processing, fault diagnosis, and process standardization), then proposed the WS-TA and ML-RFR models to optimize the indicator weights and scoring rules. Finally, through illustrative experiments, it statistically compares the scores of O&M personnel’ s various capabilities before and after the system application, verifying the system’s effectiveness. The proposed system can provide technical support for the efficient training of distribution network O&M personnel.

Abstract Cable lines are now being widely used. According to statistics data, cable dampness is the main factor affecting the operation of cable lines at present. If the cables are damp, evaluate the degree of damp and replace the cables that are severely damp in sequence. In this paper, the physical and chemical properties, mechanical properties and dielectric properties of the cable are used to evaluate the status of the cable. By comparing the length and carrying capacity of cables in operation, the degree of damp was sorted by skyline algorithm, and the most serious cables were replaced first.

With the increase of voltage level and capacity of modern power system, three-level converter, as the main circuit of active power filter, meets the requirement of improving compensation capacity. However, the number of power devices of three-level converter is twice as large as that of two-level converter, and the probability of power device failure is greater, which affects the normal operation of the system. Aiming at the fault of three-level active power filter, this paper studies the fault characteristics of different power components of the main circuit T-type three-level converter when open-circuit fault occurs, applies the hybrid theory to the fault diagnosis of the converter, establishes the normal model and the fault model of different power components when open-circuit fault occurs, and realizes the location of the fault components according to the residual generated by the predicted output and the actual output of the model, the effectiveness of this method is verified by simulation.

Nowadays, the penetration of distributed power in distribution network increases the complexity of power supply restoration. For the recovery of the traditional method in the field of distributed power supply system restore deficiency, the improving measures are put forward based on the many kinds of operation constraints, proposes the mathematical model of multi-objective optimization. In addition to conventional constraints in the mathematical model, will also be the network structure and branch power flow constraints into consideration. In view of the model, using the rough set theory, a multi-objective optimization problem into a single objective problem, and on this basis, considering the characteristic of radial network and the characteristics of the different load, by using genetic algorithm to reconstruct network, finally through the corresponding example recovery strategy is feasible and efficient.

The traditional three-stage current protection method is no longer applicable because the access of distributed power sources has altered the structure and operation of the traditional radial distribution network. In order to adapt the active distribution network protection scheme, this article firstly analyzes the topology of the negative sequence equivalent network in the case of asymmetric faults in a distribution network containing inverter type distributed power sources. Then reasonably deriving the negative sequence voltage and current components at both ends of the protected line based on Kirchhoff's law. Based on the obtained negative sequence electrical quantity characteristics, impedance-based active distribution network pilot protection is proposed as a novel approach. This method can precisely and immediately identify asymmetric faults within the area. The difference in fault eigenvalues is significant.

Application of big data techniques in power system will contribute to the sustainable development of power industry companies and the establishment of strong smart grid. This article introduces a universal framework of electric power big data platform, based on the analysis of the relationships among the big data, cloud computing and smart grid. Then key techniques of electric power big data is discussed in four aspects, including big data management techniques, big data analysing techniques, big data processing techniques and big data visualization techniques. Finally, the article presents three typical application examples of electric power big data techniques which are new and renewable energy integration, wind turbine condition monitoring and assessment and data base integrative backup for electric power enterprises.

This study deals with the idea that comprehensive knowledge representation should be established for fault diagnosis. Sufficient grid fault information including the network topology and protection knowledge are used with a diagnostic algorithm. In this way, the fault diagnosis programme not only facilitates accurate judgment of fault sections for which many kinds of information are available but also optimises knowledge to simplify the fault diagnosis method. Petri nets are used for logical reasoning on the basis of knowledge representation, which can be used to judge fault elements accurately even when the protective relays and circuit breakers malfunction. It was proved through experimentation here that this method meets the requirements of real-world diagnosis. The programme can be used as an interface to the self-healing mechanism of a smart grid. This study also posits that the smart grids should be constructed on the basis of knowledge representation for every subsystem.

The corona characteristic of HVDC tube bus is the main factor taken into account for choosing tube bus of HVDC converter substation. The intension of corona could affect the electromagnetic environment of converter station and generate audible noise. For the HVDC transmission line has long distance, it is inevitable for converter station to built in high humidity area. It is necessary for studying the corona characteristics of converter station bus and choosing the best type of the bus to consider the humidity. By establishing the calculation models of electric field strength along the tube bus surface, the paper computes the influence of humidity on corona inception electric field, high and electrode spacing of tube bus. And gives the high and electrode spacing, at which the tube bus in ±660kV HVDC converter substation will not generate corona.