| In order to solve the real-time monitoring and fault diagnosis problems of the huge key equipment, continuous operating equipment and high-security equipment in fields, development of on-line oil monitoring system is becoming very important in the area of condition monitoring. Based on the analysis of oil monitoring technology background systemically, the dissertation introduces development of on-line monitoring instrument and its engineering application, outlines the necessity of this instrument and decides research contents.Wear particles generated from the friction couples are important information carrier of wear condition, which directly reflect wear degree and wear mode. Based on the relations of wear particles, wear faults and wear mode, main characteristics of wear particles size, origin, shape and granularity are analyzed. On the view of strategies and functions of monitoring, an on-line oil monitoring system is designed and key technologies for development of monitoring system are presented. The inductance and optic-fiber transducer based laboratory are designed and tested. The inductance transducer can classify wear of large ferrous and nonferrous debris and the optic-fiber transducer can be used for inspection of oil contamination. In combination with inductance transducer and optic-fiber transducer, it centralizes two measurements method's virtues, extends detection area and validly improves reliability of monitoring system. It can meet industrial monitoring requirement to some extent. The following results are obtained through above research.1) The on-line oil monitoring parameters are selected and wear particle fundamental characteristics are defined by analyzing developments, contents of oil monitoring technology and its relation to tribo-system fault.2) The inductance transducer used to identify large wear particle is developed. The analytical theory of wear particle and detection windings are expatiated and inner magnetic distribution of inductance transducer is discussed, the geometry parameters are optimized by using Genetic Algorithm, and the finite element analysis method is used in the designing.3) The optical fiber sensor used hi oil contamination measurement is also developed. Some key techniques such as selection of light source and photoelectric cell, designs of measurement set and flow cell are analyzed. A new orthogonal optics has been presented and virtually improved the reliability of test data. The detection principle is discussed and a new inversion algorithm of constrained function method is introduced, which used to solve the particle distributions. Some problems of calibration, sample washing and so on are investigated in the course of designing optic-fibermeasurement system.4) A test rig, which simulates the mechanical lubrication system, can provide circling lubricant and change the wear particle parameters. A primary transducer experiment designed by the author consists of stability, repeatability. The problems and deficiencies are discussed, the corresponding measure is also presented. Auto-Regressive (AR) model and Neural Network time series mathematical models are used, it effectively identifies the different oil condition collected from the test data from laboratory and fields.5) A Neural Network fusion model is set up, which combines on-line data with off-line data. Through analysis the data of steering propellers in tugboat, the model are trained and validated. Results showed that different kinds of data through fused in the model can improve the veracity of condition monitoring6) Based on the concept of Virtual Instrument technology, the instrument driver program and panel application program are designed. This monitoring system has the function of collecting data and general analysis.7) The on-line measurement efficiency and accuracy are precondition of developing an on-line wear debris monitoring, the relation between detection precision and error caused by system, the compensate means...
|
PDF Full Text Request