How can you safely detect and isolate sensor faults in hazardous or inaccessible environments?

1 Fault detection and isolation

Fault detection and isolation (FDI) is the process of identifying and locating the source of abnormal behavior or malfunction in a system. FDI can be based on different approaches, such as model-based, data-driven, or hybrid methods. Model-based FDI uses mathematical models of the system and its components to compare the expected and actual outputs and diagnose the faults. Data-driven FDI relies on statistical or machine learning techniques to analyze the sensor data and detect the anomalies. Hybrid FDI combines both methods to improve the accuracy and robustness of FDI.

2 Sensor redundancy

One way to enhance the FDI capability of a control system is to use sensor redundancy, which means having more sensors than the minimum required to measure the system variables. Sensor redundancy can provide multiple sources of information that can be compared and validated to detect and isolate sensor faults. Sensor redundancy can be classified into two types: hardware and analytical. Hardware redundancy involves installing multiple physical sensors of the same type or different types to measure the same variable. Analytical redundancy uses mathematical relationships or models to generate virtual sensors that estimate the same variable as the physical sensors.

3 Sensor fusion

Sensor fusion is the process of combining and integrating the data from multiple sensors to obtain a more accurate and reliable representation of the system state. Sensor fusion can be performed at different levels, such as data, feature, or decision level. Data level fusion combines the raw sensor data to produce a single output. Feature level fusion extracts relevant features from the sensor data and fuses them to generate a feature vector. Decision level fusion merges the outputs of different FDI algorithms to produce a final diagnosis. Sensor fusion can improve the FDI performance by reducing the noise, uncertainty, and redundancy in the sensor data.

4 Fault-tolerant control

Fault-tolerant control (FTC) is the process of maintaining or restoring the desired performance and stability of a control system in the presence of faults. FTC can be achieved by using different strategies, such as passive, active, or adaptive FTC. Passive FTC uses a robust controller that can handle a range of faults without requiring any modification. Active FTC uses a reconfigurable controller that can switch or adjust its parameters according to the fault diagnosis. Adaptive FTC uses a learning controller that can update its model and parameters based on the online feedback from the system.

5 Fault isolation tools

In order to successfully implement the FDI and FTC methods, you need some helpful tools. Residuals are the differences between the actual and expected outputs of the system or its components and can be used to monitor the system behavior and detect faults. Filters can process sensor data and remove unwanted components such as noise, drift, or bias. Observers are models or estimators that can reconstruct the system state or output based on the sensor data and system dynamics. Fault indicators are variables or functions that can indicate the presence, location, and severity of faults and can be derived from residuals, filters, observers, or other sources. Utilizing these tools can improve signal quality and FDI accuracy, as well as provide additional information to isolate sensor faults.

6 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?