An edge-filter FBG interrogation approach based on tunable Fabry-Perot filter for strain measurement of planetary gearbox

Abstract

• An improved TFP-based edge-filter FBG interrogation approach is proposed. • A feedback channel is adopted to control the creep of the piezoelectric actuator. • The TFP is driven by both the coarse voltage and fine voltage. • The FBG interrogation approach is applied to the strain measurement of ring gear. Fiber Bragg gratings (FBGs), with smaller size, are suitable to measure the tooth root strain of ring gear which is an ideal signal for fault diagnosis of planetary gearboxes. In order to identify the fault features in the strain signal, a high-speed and high-resolution FBG interrogator is needed. Compared with wavelength-scanning interrogation approach, the edge-filter approach can achieve such a performance more easily and economically. However, conventional edge-filter approaches based on fixed edge filters have low applicability. The adoption of a tunable Fabry-Perot filter (TFP) increases the flexibility of the edge-filter system, but the creep characteristics of the piezoelectric actuator in TFP may cause interrogation errors. This paper proposes an improved TFP-based edge-filter FBG interrogation approach with creep control ability. First, the interrogation principle of the proposed approach is analyzed by comparing with the conventional edge-filter approaches. The creep control method is presented by employing a feedback channel and two kinds of TFP driving voltage. Next, a FBG interrogator based on the proposed approach is established. The creep control performance, static interrogation performance and dynamic interrogation performance are all tested. The results show that the interrogator has the wavelength repeatability of ~ 0.1 pm at 1 kHz and ~ 0.2 pm at 5 kHz respectively. Finally, the proposed edge-filter approach is applied to measure the tooth root strain of ring gear in a planetary gearbox. The measured strain signal has a high quality, and the tooth fault feature can be identified obviously.