Figure 1: Different types of transmitter.

Transmitter faults mainly include complete fault, bias fault, shifting fault, and precision degradation fault. This is shown in figure 2.

Complete fault refers to the sudden failure of transmitter measurement, and the measured value has been constant. Bias fault is that the measured value of the transmitter differs from the real value by a certain constant. From figure 2, we can see that the measurement with fault and the measurement without fault are parallel.

Shifting failure is that the difference between the measured value and the real value of the transmitter changes with the increase of time. Precision degradation fault refers to the deterioration of the transmitter's measurement ability and the decrease of accuracy. When the accuracy level is lowered, the mean of the measurement does not change, but the variance of the measurement changes.

Bias fault and shifting fault are not easy to find, but they will cause a series of unexpected problems in the process so that the control system can not play a normal role for a long time.

Figure 2: Diagrams of complete fault, shifting fault, bias fault and precision degradation fault.

1. Transmitter Fault Type

1.1 Transmitter Fault Degree

According to the degree of faults, there are hard faults and soft faults. Hard faults with large amplitude are suddenly caused by structural failure. And soft faults are caused by slow variation of characteristics with small amplitude.

Hard faults are also called complete faults, in which the measured value always maintains a certain number. Usually, this constant value is zero or the maximum value. The function of these measurements is roughly a horizontal line.

Soft faults are that deviation failure, drift failure and accuracy decline, etc. The faults are relatively difficult to be found. Therefore, in a sense, the harm of soft faults is greater than that of hard faults. So soft faults have gradually attracted people more.

1.2 Transmitter Fault Existing Time

From the perspective of existing time, it could be an intermittent failure and permanent failure. Intermittent failure is a malfunction of a transmitter that occurs at intervals, while permanent failure is a complete failure of the device, from which recovery is impossible.

1.3 Transmitter Fault Speed

According to the speed of faults, the transmitter faults could be divided into catastrophic failure and slow failure. The sign of catastrophic failure is changing quickly, while that of slow failure is in opposite ways.

1.4 Transmitter Fault Reasons

Table 1: Reasons and signal of several transmitter failures, including aviation failure, impact failure, open circuit failure, drift failure, short circuit failure, periodic failure, and dead zone failure.

In addition, from the perspective of modeling and simulation, it can be divided into multiplicative faults and additive faults.

2. Transmitter Fault Diagnosis Methods

Fault diagnosis methods are divided from different perspectives. In this article, we will introduce the method based on the analytical mathematical model and that independent of the mathematical model.

Figure 3: An analytical mathematical model.

2.1 Diagnosis Method Based on the Analytical Mathematical Model

According to the different forms of residual errors, this type of method is further divided into parameter estimation, state estimation, and equivalent space. The model-based fault diagnosis method is one of those earliest developed ones, which is also the most widely studied and applied.

The advantages are the model with clear mechanism, simple structure, easy to implement and analyze, and real-time. It plays an important role in the field of fault diagnosis and the development of transmitter fault diagnosis.

On the contrary, the disadvantages are complex calculation and system. There would be modeling errors and the insufficient adaptability of the model. The lack of reliability is prone to false and missed positives. Based on the model, the method is of the robustness from external disturbance and the insensitivity of system noise and disturbance.

Till now, the researches of this diagnosis method are still mainly focused on linear systems, which is of great significance to the study of general fault diagnosis for nonlinear systems. Meanwhile, the robustness also features a high research value.

Table 2: Pros and cons of model-based diagnosis methods, including parameter estimation, state estimation, and equivalent space.

2.2 Diagnosis Method Independent of the Mathematical Model

As the control system is increasingly complex, it is harder to establish an accurate analytical mathematical model in practice. And when there is a modeling error, the method based on the model will give false or missed positives. As a result, the fault diagnosis method independent of the model has been highly valued.

For an approach without relying on the mathematical model, the pros are not requiring an accurate model of the object and well adaptable. But it has a complex structure that is difficult to implement.

Fault diagnosis methods which are independent of the system model are composed of data-driven, knowledge-based and discrete event based method.

* Data driven diagnosis method

Type: Signal processing
Statistical

Signal processing methods:
Absolute value test and trend test;
Fault detection using Kullback information criterion;
Method based on adaptive sliding Lattice filter;
Method based on signal modal estimation correlation analysis;
Wavelet analysis;
Information fusion method, etc.

* Knowledge-based diagnosis method


Type: Symptom-based
Qualitative model based

* Discrete event based diagnosis method
A new type of fault diagnosis method.
Basic rule: The state of the discrete event model reflects both the normal and the failure state of the system.

In short, with the progress of theoretical research and the improvement of technology, the study of transmitter fault diagnosis tends to be more practical, so that it can solve some problems we will encounter in practice.

Related Info