Sensor faults mainly include: complete failure fault, fixed deviation fault, drift deviation fault and precision descent four kinds. As shown in figure 1.
fault type sensor in Figure 1;
failure is a sudden failure of sensor measurements, has been a constant; bias fault mainly refers to a kind of fault sensor measurements is a constant and the true values, it is seen that the measurement and fault free measurement of fault are parallel;
fault diagnosis is a kind of fault refers to the sensor measurement value and the true value of the difference increases with time and biochemical;
precision decline means that the measuring ability of the sensor becomes worse and the precision becomes low. When the accuracy level decreases, the average value of the measurement does not change, but the variance of the measurement changes.
fixed bias fault and drift faults are not easy to find fault, in the process of failure occurrence is expected to cause a series of problems, the long-term can not control system function.
fault classification of
1, according to the degree of
damage fault fault refers to the general structure, large amplitude, sudden change; variation characteristics of soft fault refers to the smaller amplitude, slow change.
hard fault also known as complete fault, complete fault measurement value does not change with the actual change, always maintain a reading. This is usually a constant value is zero or the maximum reading. Fault measurement is roughly a straight line.
soft fault includes data deviation, drift, precision grade decline, etc.. Soft fault is relatively small, difficult to be found, therefore, in a sense, the soft fault harm than hard fault greater harm, its harm has gradually aroused people's attention.
2, according to the fault classification performance of
according to the presence of fault performance can be divided into intermittent fault and permanent fault.
intermittent failure is good and bad; permanent failure, can not resume normal.
3, according to the fault occurrence and development process of
mutation fault signal change rate; slow change fault signal change rate is small.
4, according to the fault classification of
according to the fault reason can be divided into fault, impact fault, open circuit fault, drift fault, short circuit fault, periodic interference, nonlinear dead zone fault.
fault bias fault: bias current or bias voltage;
fault impact fault: random interference power and ground the surge, EDM, D/A converter in
fault drift fault: temperature;
fault fault: pollution caused by bridge corrosion, short line;
fault causes periodic interference: power 50 Hz interference;
nonlinear dead time fault reason: amplifier saturation, nonlinear link, etc..
in addition, from modeling and simulation point of view, can be divided into multiplicative fault and additive fault. The bias fault with small signal on a constant or random in the original signal; for the impact of interference on the original signal superimposed on a pulse signal; the short-circuit fault signal is close to zero; the open circuit fault signal, the proximity sensor output maximum drift; fault signal at a rate offset from the original signal fault; periodic interference, the original signal is superimposed on a frequency signal.
1, based on the analytic method of mathematical model of
according to the different forms of residuals, the method based on analytical mathematical model can be further divided into: parameter estimation method, state estimation method and equivalent space method.
model-based fault diagnosis method, is one of the earliest developed diagnostic methods, but also a research and application of the most extensive diagnostic methods.
advantage is clear model mechanism, simple structure, easy to implement, easy to analyze, real-time diagnosis. It plays an important role in the field of fault diagnosis, and will be the main method of sensor fault diagnosis in the future.
