Abstract: Electrohydraulic servo mechanism (EHS) is an important actuator in aerospace equipment, and its maintenance of a healthy state plays a crucial role in the safe and stable flight of aerospace equipment. However, the existing health assessment models face two major problems: First, the large amount of history text information generated during the operation of the EHS fail to be utilized. Second is the lack of a model that fuses the acquired text information with the test data for multi-source information fusion and thus realizes the health state assessment. To solve these problems, an evidential reasoning rule health state assessment model fusing multi-source information (ER-MIF) is proposed. First, the extraction of biographical text information is realized using a large-scale natural language processing model (BERT). Second, the EHS operation mechanism is analyzed and the EHS health assessment index system is constructed. Then, the maintenance level and operation time influencing factors are proposed and input transformations are performed on the health indicator monitoring data to construct a health state assessment model that integrates multi-source information (ER-MIF). Furthermore, the model parameters are optimized to determine the optimal model parameters and improve the assessment accuracy. Finally, the validity of the proposed model is verified through the evaluation case and comparative study of a certain electro-hydraulic servo mechanism.

Key words: evidential reasoning rule, health assessment, natural language processing, lectro-hydraulic servomechanism, diagnostic reasoning