Power Data Traceability Mechanism Based on Data Processing Unit

With the digitization and intelligent transformation of the power industry, the scale of data is increasing exponentially. This surge in data circulation presents notable challenges in terms of data monitoring and management. At the same time, the potential risks of data leaks and attacks have made it increasingly difficult to identify the sources and relevant personnel involved in the event of an accident. Data traceability technology can be used to track and record the source, flow path, and processing process of data. This paper introduces a data traceability mechanism based on the Data Processing Unit (DPU). We propose a traceability model suitable for new power systems based on the W7 model, which enables comprehensive tracking of data across nine dimensions. Within our framework, the swift data processing and analytical capabilities of DPU provide strong support for establishing data traceability and accelerating the data tracing process. Additionally, we introduce a knowledge graph architecture in the field of new power systems. To orderly utilize large-scale text data in new power systems, we propose a named-entity recognition method based on the BiLSTM-CRF model to enhance the comprehensiveness of data traceability results. Finally, we validate the effectiveness of our proposed method through experimental simulations.