Decision on Fault Diagnosis of Liquid Propellant Rocket Engine Based on Deep Analytic Hierarchy Process

Fault diagnosis is critical for liquid-propellant rocket engines (LREs). Although there are mature single diagnostic algorithms for both slow-varying faults triggered by abnormal slow-varying parameters and fast-varying faults triggered by abnormal fast-varying parameters, these algorithms differ in parameters and principles, resulting in a lack of effective interaction. Consequently, a comprehensive decision-making method is urgently needed for the LRE system to achieve final decisions. To address this issue, this paper proposes a comprehensive decision-making method based on the Deep Analytic Hierarchy Process (DAHP). Firstly, two diagnostic algorithms are selected for each type of fault: the thermodynamic model algorithm A and the multi-parameter adaptive threshold algorithm B for slow-varying faults, as well as the vibration analysis algorithm C and the displacement analysis algorithm D for rapid-varying faults. Then, DAHP is utilized to comprehensively consider the individual and overall performance of the algorithms, applying prior knowledge from offline evaluations to calculate two layers of weights. The first layer weights are computed based on the fault detection rate, accuracy, false alarm rate, and missed alarm rate of each algorithm. The second layer weights are determined by integrating the overall fault coverage and detection speed of the slow-varying and rapid-varying algorithms. Finally, expert decision-making experience in the field of fault diagnosis, combined with practical engineering rules, is used to determine the final decision. Case studies using historical test data demonstrate the feasibility and effectiveness of the proposed deep analytic hierarchy process. This method not only enhances the accuracy and reliability of fault diagnosis but also provides more scientific decision support for the health management of liquid propellant rocket engines.