Suitability of tunable Fabry-Perot spectrometers for condition monitoring purposes of gear oils in offshore wind turbines.

摘要

The offshoring of wind turbines is an important factor for increasing the regenerative part of future energy generation. The turbines are getting larger and are often inaccessible for preventive maintenance during long periods of the year. Measuring the oil condition gives important information on gear health and helps to protect machinery from damage. Benefits of such measurement systems are cost savings due to an enhanced service life of the lubricant, increased lead time for maintenance and oil changes that are conducted rather by oil condition than by fixed scheduled intervals. An infrared transmission spectrum of the oil sample provides important information of its condition. Such a measurement is usually performed at a high spectral resolution of 1 cm~(-1) covering a broad spectral range of 4000 to 500 cm~(-1). Continuous online monitoring by FTIR for series-application is expensive and bulky equipment is required. Miniaturized, tunable Fabry-Perot spectrometers (FPS) have been demonstrated to perform well in gas analysis applications and have the potential for the construction of a cost effective infrared oil sensor. In this paper the suitability of miniaturized, tunable FPSs for condition monitoring of wind turbines is analyzed. Two commercial FPSs LFP-3041L and LFP-3950L from InfraTec are analyzed for their applicability to this measurement problem. A third notional FP spectrometer based on a tandem setup of two tunable etalons is presented and its performance for the detection of oil aging is evaluated. Based on a representative data set of used oils, spectral bandwidth and resolution of the spectra are limited to the properties of the FP based mini spectrometers. The original data set, taken by a FTIR with a resolution of 1 cm~(-1) covers the spectral range from 4000 to 500 cm~(-1), the second dataset covers the range from 3333 to 2439 cm~(-1), data set three covers the range from 2564 to 2083 cm~(-1) corresponding to InfraTec LFP 3041L and LFP-3950L respectively. Data set four covers the range 1818 to 909 cm~(-1) corresponding to the simulated Fabry-Perot Tandem Spectrometer (FPTS) approach. Out of these downscaled spectra a partial least square regression (PLS) was done to predict important oil quality parameters like total acid number (TAN) and additive depletion. Prediction quality of these parameters is used to evaluate the usability of the respective mini spectrometer to the oil monitoring application.