Critical sensor applications: Diagnostics or redundancy?
Smart systems and devices are proliferating in today’s “smart age.” The core of most of these smart systems are based on sensors and sensor engineering. Sensors are central to creating intelligent devices capable of gathering an array of complex data from divergent sources and translating them into measurable outputs that provide important insights about their functioning. These insights form the basis of decisions and actions that need to be taken.
Sensor measurements, engineering
Sensors are essentially transducers that gather inputs from a number of environmental phenomena like light, heat, motion, moisture, pressure, vibration, presence of chemicals and many others. Sensor engineering:
- Is the science of integrating an ensemble, enabling sensors to act and transform into effective solutions.
- Delivers the ability to make better and more accurate decisions where it is applied.
- Touches many critical areas of life, including an expanding list of medical electronics, wearables, vehicular safety, industrial equipment and toys.
Innovation has kept pace with the developments in this field over the decades, and a few key trends in this area can be observed today. Understanding these trends and the factors shaping the field will help with understand what should be done when a sensor fails or intermittently malfunctions. Sensor designs for critical application take two approaches: Sensor redundancy or sensor diagnostics. Each has been widely emulated in various industries.
Three sensor design trends
As our world becomes more connected, sensors are multiplying across many applications and innovation continues. According to industry experts, key trends include:
- Miniaturization: Many applications require hosting multiple sensors within a small area without impacting sensor performance. They also often have a low-power operating requirement. This can be achieved through the process of miniaturization with nanotechnology, among other means.
- Smart sensors: On-board processing capabilities interpret data and send out processed data or information, rather than raw data. The advancement in the sheer volume of data these sensors capture has made the shift to digital a crucial requirement for maximizing their performance. Intelligent sensors can not only capture sensing data, but also interpret the data meaningfully for a variety of applications.
- Sensor fusion: Similar to the smart sensor trend, development of the Industrial Internet of Things (IoT) and the increasing expectation of inter-connectedness have been the catalyst for multi-sensor integration. The need to capture multiple measurement types in extremely small packages is pushing the development of multi-sensing elements.
These three trends often overlap along with a need for higher performance at lower costs.
Sensor intelligence applied
With the growing relevance of sensor engineering, it is important to critically evaluate the different sensor engineering applications. Examples follow of sensor use patterns, impacts and outcomes, and the decisions and actions based on sensor data collected and analyzed in different industries and applications.
In certain cases, using redundant sensors can improve precision and reliability.
Machine presses require a very high degree of accuracy in applying and holding force on an object. One way to measure this force is via load cells. Multiple load cells can be used, and the results can be averaged out. Doing so helps to nullify any small differences in load readings between the cells, which can occur at different times and in varied environmental conditions. An additional benefit of the averaging process is noise signals on individual load cell outputs also would be reduced. Another benefit of using multiple sensors to measure one process parameter is even if one sensor fails, the system can still function, though it may not give equally accurate readings.
How does one manage gathering inputs from multiple redundant sensors and processing them to achieve the benefits mentioned? In this case, the motion controller must be programmed to do checking and averaging load cell data before feedback is provided to the control loop.
In the airline industry, sensors play a critical role in ensuring the safety of airborne planes. In a well-known instance, a leading global manufacturer opted to use diagnostics from one angle-of-attack (AOA) sensor in the automated security system instead of using two or even three such sensors to build in redundancy and fail-safe measures. In more than one instance, the AOA sensor recorded/interpreted erroneous data, which in turn led to faulty decisions taken by the automated security system regarding the plane’s maneuvers in flight mode.
In aircraft that use two or three AOA sensors, there have been instances when data from each sensor has significantly disagreed, pointing to an anomaly in at least one reading. In such cases, however, in-built warning systems provide early signals about such disagreements to the pilots, who can then decide to turn off the automatic inputs. Redundancy appears to be a critical necessity, in such applications.
Medical and healthcare applications use sensors in many different ways. To cite one of many use cases, doctors use glucose sensors to collect glucose concentration (GC) information from patients. These sensors often are limited by sensor errors, communication interruptions and noise. Thus, relying entirely on sensor data may not be the correct course of action. Clinics compensate by applying established mathematical models or computing power. Data scientists, using advanced computing models, can perform various levels of data analytics to make accurate predictions using model-based estimates (MBE). Some medical devices work a little differently from other industries as the scope to use an MBE is significant, decreasing the reliance on sensor diagnostics in some applications. The field is evolving, and there is scope for increasing application of sensors in healthcare.
Automobile measurement applications have been successfully solved since the 1980s with the help of Hall sensor integrated circuits (ICs). A common, relatively simple use case example is a Hall sensor with switching output to detect whether or not a car door is fully closed.
With massive advancements in automobile technology, position reliability requirements have increased over the last few years. If a sensor incorrectly detects accelerator pedal position, a potentially dangerous driving situation could result. To enhance safety, an option is using two sensors instead of one for the same measurement. The output signals of both sensors then can be compared, and error deviations can be detected. The on-board electronics systems in automobiles are capable of responding accordingly.
Multi-sensor application may increase weight and require more installation space, but those issues can be addressed by increasing the integration density of the sensors. A method commonly used in the semiconductor industry involves integrating several complementary metal-oxide-semiconductor (CMOS) chips in one package. However, this is a novel approach in Hall effect sensors.
Consider the application
Sensor engineering is a growing and evolving field with high potential and advancements are taking place continuously. It may not be wise to discard diagnostics or redundancy as both have their uses. Apply judgment depending on the application and results required for each case.
Shashidhara Dongre is global head, smart products and services practice at L&T Technology Services, a CFE Media content partner. Edited by Mark T. Hoske, content manager, Control Engineering, CFE Media, firstname.lastname@example.org.
KEYWORDS: Sensor design, sensor redundancy, sensor diagnostics
Sensor design trends include miniaturization, smart sensors, sensor fusion.
Different applications sensor redundancy, diagnostics
Sensor integration may help.
Look at new sensor capabilities before incorporating sensors into an automation application in a traditional way. Design changes could help.