| This study has designed an on-line detection system for particles of lubricating oil, systematically applying the principle of electromagnetic induction to detect the size and concentration of magnetic particles in real-time and predict the running trend of equipment to forecast failures. Sensors tend to work better for magnetic particles of more than 75 um. The whole system include hardwares and softwares:The hardwares include high-frequency excitation source circuit, analog conditioning circuit and digital processing circuit. Excitation source circuit outputs stable high-frequency signal source which is used on primary coil through hardware circuit. The analog circuit consists of pre-amplification circuit, band-pass filter circuit, measurement circuit on true RMS and automatic zeroing amplifier circuit, The analog conditioning circuit enables the signal output to be within 0 to 3 V, and thus high SNR digital signals are received. The microprocessors of digital circuit is STM32F103VET6, which includes functional modules of sampling frequency selection and restoration. Then draw and make the PCB board, to complete the hardwares of the whole detection system.The softwares have accomplished the functions of analog signal acquisition and preprocessing, as well as digital signal preprocessing, LCD display and storage. The data processing mainly includes locally weighted regression scatterplot smoothing (LOWESS) and extremum conversion. The former removes the noises in the signals, further improving the detector performance; the latter evaluates the peak of the signals and directly displays the particles in the lubricating oil to predict the equipment running status.Eventually the detector performance of the online detection system is tested through two contest experiments. Experiments on monitoring the actual particles in lubricating oil have shown that the accuracy of the system can meet the engineering requirements of monitoring particles in lubricating oil in wind turbine gearboxes. |
PDF Full Text Request