最后开发了电网故障辅助诊断系统，集成了电网故障诊断领域知识图谱和基于知识问答的辅助诊断模型。针对辅助诊断模型无法推理动态知识图谱的局限性， 本文提出了融合用户特征的动态问答策略，将运维工程师维护数据库的个人倾向 和操作记录作为辅助诊断模型的参考，进一步提高了本系统的灵活性和应用价值。
综上所述，电网故障辅助诊断系统是协助运维人员进行故障排查的有力工具。 将知识图谱技术应用到电网故障诊断领域中，可以实现故障知识的高效推理。这 种基于知识图谱的方法可以提供更加精准和高效的问题解决方案，同时也可以实 现知识共享和传递，促进电网故障诊断领域的技术创新和发展。
The reliability and stability of the national power system operation are crucial to industrial production and people's lives. However, at present, some grid fault diagnosis work still requires on-site investigation by operation and maintenance engineers. Their work experience and knowledge reserves directly determine the quality of grid fault diagnosis and maintenance. Therefore, realizing accurate and efficient grid fault diagnosis and reducing the efficiency difference caused by human factors are the difficulties that need to be broken through urgently. In recent years, artificial intelligence technology has become more and more closely related to traditional industrial fields. If artificial intelligence technology can be applied to power grid fault diagnosis, it will be possible to solve this problem, which is of great significance to the development of the power grid field.
This thesis conducts research on the auxiliary diagnosis system of power grid faults based on knowledge graph. Operation and maintenance engineers can use natural language text to input questions in the system, and use the reasoning model based on the knowledge graph to mine the internal relationship of power grid knowledge to obtain the required relevant knowledge and fault analysis results. The main work of this thesis is as follows:
Firstly, a knowledge graph in the field of power grid fault diagnosis is constructed. In order to fit the actual grid operation and maintenance scenario, this thesis comprehensively considers multiple sub-fields, and proposes a knowledge graph model layer design method that integrates grid structure and electrical faults, covering information such as grid system structure, fault types, and solutions. Support problem reasoning for operation and maintenance engineers.
Then, an auxiliary diagnosis model of power grid faults based on knowledge question answering is designed. In order to overcome the problem that the traditional question answering model is difficult to complete implicit reasoning in the absence of paths, this thesis designs a message passing gating network to aggregate the structural features of reasoning paths into the features of entities and relations. The link confidence of candidate answer nodes is predicted directly based on the head entity and relationship of the question link, and the explicit and implicit knowledge reasoning is realized at the same time.
Finally, an auxiliary diagnosis system for power grid faults is developed, which integrates the knowledge graph in the field of power grid fault diagnosis and the auxiliary diagnosis model based on knowledge question answering. Aiming at the limitation that the auxiliary diagnosis model cannot reason about the dynamic knowledge graph, this thesis proposes a dynamic question-and-answer strategy that integrates user characteristics, and uses the personal tendency and operation records of the operation and maintenance engineer to maintain the database as a reference for the auxiliary diagnosis model, which further improves the flexibility of the system and application value.
To sum up, the grid fault auxiliary diagnosis system is a powerful tool to assist operation and maintenance personnel in troubleshooting. Applying knowledge graph technology to the field of power grid fault diagnosis can realize efficient reasoning of fault knowledge. This knowledge graph-based method can provide more accurate and efficient problem solutions, and can also realize knowledge sharing and transfer, and promote technological innovation and development in the field of power grid fault diagnosis.
|Keyword||电网故障诊断 知识图谱 问答系统|
|Sub direction classification||人工智能+制造|
|planning direction of the national heavy laboratory||语音语言处理|
|Paper associated data||否|
|Files in This Item:|
|202028014628059于雅涵 (（4352KB）||学位论文||限制开放||CC BY-NC-SA|
|Recommend this item|
|Export to Endnote|
|Similar articles in Google Scholar|
|Similar articles in Baidu academic|
|Similar articles in Bing Scholar|
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.