As the aviation industry drives toward predictive maintenance and operational and aftermarket efficiencies, intelligent sensors have expanded in variety and scope. Meanwhile the overall defense and aerospace sensor market has hit $4 billion, with expected annual growth of about 5%, according to manufacturer TE Connectivity.
The trend toward smart — and digitally capable — sensors reflects the effort “to push intelligence away from a central computer and to move [it] out into the sensors,” said Pete Smith, manager of sensor product knowledge with TE. If the local sensors have computing smarts, they don’t have to wait for a central computer to process their output, so they can respond more quickly. Such sensors can also put the signals they are detecting into a format that’s easier for aircraft systems to use, he said, so these systems “don’t have to do [measurement unit] conversions somewhere else.”
Sensor intelligence can be expressed on multiple levels — at the sensor level but also at the level of the software that interprets sensor data. This software can use techniques such as physics-based modeling, big data analytics/statistics and machine learning.
Other trends include the move from analog to digital output sensors and the effort to introduce wireless sensors into airplanes. Digital sensors can consume less power and transmit more data, proponents say. Wireless sensors would save weight and increase reliability.
Proximity, Altimeter and Air Data Smarts
Honeywell’s Safety and Productivity Solutions business unit recently introduced integral health monitoring (IHM) proximity sensors that can monitor their own health as well as the position of various safety-critical systems.
Designed to work in the harshest environments, the analog sensors can monitor the position status of thrust reverser systems, flight controls, aircraft doors, evacuation slide door lock mechanisms and landing gear.
Sensor self-awareness increases system mean time between failure because it eliminates false positives. This increases MRO efficiency and minimizes time on ground. The pilot can distinguish between system and sensor malfunctions.
IHM sensors originally were designed to monitor the position of the thrust reverser and whether the ram air turbine door is open or closed, said Daniel Crosby, the unit’s product marketing manager for aerospace and defense, North America.
Components are prequalified to DO-160 environmental requirements. The sensors also are suited to composite structures with their higher levels of vibration and to engine accessories with more demanding thermal requirements. The three-wire, stainless steel, hermetically sealed sensor operates at temperatures ranging from minus 67 to 239 deg F.
Self-awareness is achieved not by adding a computer chip, but by adding a “third state.” The first two states represent the proximity of the metal or composite “target” to the sensor — “near” or “far.” The third state is the indication of a fault within the sensor. The sensor’s three-wire harness is tied into the plane’s control electronics in order to be able to send alerts to the pilot.
IHM is an enhancement of eddy current killed oscillator (ECKO) proximity sensing technology, in which the proximity of a target causes target “eddy currents” to increase as the target gets closer to the sensor. The ensuing loss of energy from the sensor to the target eventually “kills” the sensor’s electrically created magnetic field oscillation, triggering sensor output.
While based on ECKO technology, IHM’s fixed amplitude variable current oscillator (FAVCO) technology adds self-diagnostics and measures the current in the oscillation circuit instead of the oscillation amplitude. FAVCO measures energy required to maintain the oscillation, Crosby said.
The IHM circuit oscillation is continuous. If the sensor fails and the oscillation stops, however, the circuit changes the output to a fault rather than sending incorrect information.
Smart Drone Sensor
TE offers a smart altimeter for commercial and hobby drones, Smith said. “The intelligence embedded in these sensors helps improve the performance, reliability and accuracy of the sensor.”
The altimeter contains two sensors, a barometric pressure sensor to read the altitude based on air pressure, and a temperature sensor element to monitor the ambient temperature around the sensor. The altimeter also features a microcontroller that manages the signals from these elements and performs digital communication tasks that send sensor data to the main aircraft control systems.
The temperature sensor serves two purposes, Smith explained. “It measures the temperature of the sensor package itself and then uses that information to correct any errors in the pressure sensor reading that might occur” because of temperature changes associated with altitude changes. But it also can provide “direct temperature measurements that can be used by the aircraft for compensating other systems and sensor measurements.” So, this sensor “exhibits intelligence because it not only measures altitude, it also measures another parameter that is then applied to the altitude reading to correct for errors that might occur in uncompensated devices.”
The altimeter provides digital signal outputs. TE provides the signal in the inter-integrated circuit (I2C) format, an industry standard for digital communications. This format enables “very low-power operation for the altimeter because the sensor is active only when it’s being queried for data,” Smith said. “In between these queries, the sensor is in sleep mode, which consumes almost no power.”
UTC Aerospace Systems cites its Smart Probe, which is replacing pitot tube sensors on some airplanes. The Smart Probe measures aircraft speed, angle of attack, air pressure and altitude. It packs in “an incredible amount of intelligence and has self-diagnostics on board,” said Mauro Atalla, VP of engineering and technology for the sensors and integrated systems business. It’s able to tell whether the sensor is blocked, whether the heater is operating and whether its measurements are inconsistent, he added.
UTC views sensors as ground-floor elements in a larger structure, which it calls an intelligent ecosystem. It offers sensors that monitor parameters such as electrical current, torque, position and air pressure. “A sensor is not necessarily self-aware, but the associated software allows the system to assess the performance of the sensor,” Atalla said.
The intelligent ecosystem embraces prognostics and health management (PHM) and MRO services — packaged as FlightSense — as well as aircraft sensors, hardware and software. It involves having the right sensors in the aircraft, collecting the data from the sensors, transferring the data to a cloud- or ground-based system, and analyzing the data via OEM engineering models and other techniques to interpret it, predict what will happen to a box and make recommendations, explained Ajay Mahajan, UTC’s VP of asset management programs and strategy for the aftermarket business. UTC’s repair network, the final link in the chain, backs up the other elements.
Driving the demand for analytics is the flood of data generated by modern aircraft. Whereas legacy single-aisle aircraft produce about 100 mb of data per two- to four-hour flight, modern aircraft such as the 787 or the A350 produce 10 times that much per flight, Atalla said.
FlightSense uses a PHM system dubbed Ascentia, with physics-based modeling, statistical analysis and machine learning. “Predictive maintenance is the whole intent,” Mahajan said.
To start, the right sensors can collect the right data, Mahajan said. But you also need to do something with the data. That’s where physics-based testing and analysis come in, enabling the system to make maintenance recommendations, he said.
Physics-based modeling relates to the OEM’s knowledge of a product’s design and potential failure modes, Atalla explained. It also involves understanding what variables need to be measured, as they will change over time, serving as leading indicators of particular failure modes.
An alternative approach involves the use of big data analytics and statistics — as well as machine learning — to search for correlations across a large number of variables in order to identify potential failure modes, he said.
For many situations an understanding of the physics enables you “to analyze a much smaller quantity of data,” the data set that you know to be relevant to a particular failure mode, Atalla said. With the physics-based approach, the OEM may be able to identify critical things to measure over time because changes in them are likely to “indicate an early degradation in performance that eventually will lead to a failure.”
However, there are other failure modes that may not involve physics-based models, such as interactions across systems, particular routes an airline operates or particular maintenance procedures at certain airlines or airports. “So we also use big data/statistics and machine learning to capture all these other variables and potential failure modes,” Atalla said. This approach builds knowledge over time as these modes were not necessarily anticipated during a product’s design.
The two techniques can be combined. Algorithms can be designed with physics-based modeling. These algorithms can be trained and refined using field data via machine learning, he said.
UTC also offers an aircraft interface device (AID) that collects aircraft data and transmits it to the ground via Wi-Fi, cellular and satcom links. The box aggregates the data and feeds it into the Ascentia PHM platform and the Ascentia customer portal. The PHM software analyzes the data, predicts what is going to happen with an LRU and recommends what to do about it.
UTC said that Ascentia can reduce potential flight delays by 30% and unscheduled removals by 20% for its components. Some 777 customers, for example, avoided costly unscheduled maintenance on power electronics cooling systems because the sensors, together with the PHM software, assessed problems like leakage and provided predictive recommendations, Mahajan said.
Data Hunger Equals Sensor Opportunities
Customersareaskingforhigheraccuracy,lowertotal cost of ownership and more insight into what systems are actually “experiencing,” observed Sean Gough, TE Connectivity’s director of product management.
Gough anticipates sensors that will inform operators, for example, “what loads are actually experienced by the aero structure during a flight cycle,” so that “we [can] gain insight that would inform engineering for structural design of next-gen airframes.” He also expects that in a decade sensor content on aircraft will be significantly higher and that sensors will be networked.
How far could this go? “Networked smart polymers and paints may one day sense temperature and pres- sure,” Gough predicts. This would generate a “plethora of real-time data that [would] allow optimizations, such as morphing wings that change shape as fuel burns and weight decreases, leading to much better fuel economy.”
Planes already transmit a host of data for mission-crit- ical functions, Gough said. This trend will be enhanced and probably accelerated by the cloud, with digitization as an enabler.
But “adoption of smart sensing for mission-critical appli- cations really depends on the ability to certify open ASICs [application-specific integrated circuits],” he said. This means that the “architecture is known, controlled and under- stood” and that the software is open-source. “This is a costly undertaking currently and aircraft manufacturers will need to agree on standards that facilitate sensor development.”
Another challenge from a certification perspective is software in the loop, he said. Once these challenges are solved, however, the “benefits are huge from cost and performance standpoints. “Accuracy and resolution will provide better and more actionable data.”
The evolution toward predictive maintenance — driven by burgeoning data sets — presents opportunities for Honeywell’s Safety and Productivity Solutions business unit, according to Daniel Crosby, the unit’s product marketing manager for aerospace and defense, North America. For one thing, “a lot of sensors in ‘connected aircraft’ do not have the same robustness requirements as DO-160,” he said.
Honeywell is looking at monitoring the forces exerted to apply brakes “to better predict when they are going to fail.” Another opportunity might be measuring the im- pacts of landings on a landing gear over time “to under- stand how much life is left in the system.”
Digital vs. Analog
Overall there is a move toward sensors that produce digi- tal rather than analog outputs, observed Pete Smith, man- ager of sensor product knowledge for TE Connectivity. For one thing, digital sensors can operate from much less power, so they can reduce energy consumption, he said.
What’s more, “you can transmit a lot more data and information over a digital signal” than you can with an analog signal, he said. So additional data about the in- tegrity and status of the sensor is much easier to commu- nicate. Moreover, it can be a two-way communication. “We can actually tell sensors to operate differently under certain conditions.”
Another thing to watch for is the appearance of wireless sensors inside airplanes. UTC Aerospace Systems has been involved in government/industry efforts to allocate frequencies within the 4.2-4.4 GHz band for this pur- pose, said Mauro Atalla, VP of engineering and tech- nology for the sensors and integrated systems business. Wireless sensors would save weight and increase reli- ability by reducing the amount of wiring and the number of connectors, respectively. Such sensors are rare today because the bands where commercial devices oper- ate suffer from interference. AVS