Evaluation methodology for inertial measurement units (IMU) in structural health monitoring (SHM) applications

As part of the fleet management and maintenance of military aircraft, basic usage spectrum (BUS) updates are performed regularly to capture fleet usage and operating details and compare with the design usage. Traditionally,... [ view full abstract ]

As part of the fleet management and maintenance of military aircraft, basic usage spectrum (BUS) updates are performed regularly to capture fleet usage and operating details and compare with the design usage. Traditionally, these BUS updates were performed using a combination of post flight checklists, pilot questionnaires and pilot interviews. As an alternative, the National Research Council (NRC) developed a low-cost, standalone sensor system to measure and record the relevant flight data for the purpose of providing a BUS update. The core of this sensor system is a commercial off-the-shelf (COTS) inertial measurement unit (IMU) composed of a GPS, pressure sensor, 3-axis accelerometer, gyroscope and magnetometer.

This paper describes the selection and evaluation process of the IMU for use in developing the sensor system. A large number of COTS IMUs, including aerospace and hobby grade systems, are available in the marketplace, each with different costs and capabilities. A systematic approach was implemented to evaluate each IMU. The initial evaluation was based on published specifications, with a secondary extensive evaluation based on physical performance through the use of a NRC developed benchmarking protocol. This benchmarking process involved a series of static and dynamic tests to verify sensor functionality, accuracy and reliability in different environments. Due to the low-cost, open source and standalone requirement of this particular application, most dedicated aerospace IMU platforms were not viable. Hobby grade IMU systems, while low-cost were not always viable due to lack of hardware and software support. In cases where they were viable, questions remained regarding the accuracy of their measurements, and it was this concern that drove the development of our benchmarking test procedure.

Three IMU platforms were evaluated: a baseline reference IMU, MotionNode (from Motion Workshop) known for its high accuracy and consistency, and two IMUs in consideration for the sensor system, X-Monkey (from Ryan Mechatronics) and Ultimate IMU (from SparkFun). The reference IMU (MotionNode) was not a suitable candidate for the system as it needed to be connected via cable to its control node. To establish sensor accuracy in a vibration-free environment and verify basic functionality of each sensor, static testing was performed on a 3 degree of freedom (DOF) precision mount capable of rotating each axis by a defined amount [Figure 1]. The static testing was done to evaluate the natural noise and drift of the sensors, while dynamic testing was performed to assess sensor accuracy while undergoing a range of motion that may be expected from a helicopter. Dynamic tests were performed by securing the 3 DOF mount to a small electromagnetic shaker platform (Wilcoxon F4/F7 electromagnetic/piezoelectric shaker system) [Figure 2]. The dynamic tests were performed as they better represent the vibration environment that would be experienced by the IMU when placed inside an operational helicopter. While the performances of the two IMUs in consideration were not as good as the reference IMU, they still demonstrated reasonably accurate results, with the Ultimate IMU showing a tendency for its readings to drift over time.

The results of the benchmarking process were used to verify that the IMU selected for our application would provide sufficient accuracy and reliability. While our requirements for an IMU were unique to the design goals defined by NRC, the selection and evaluation process will be of interest to others looking to use COTS IMU sensors for other applications.

Figure 1: Tri-axial tripod mount used to evaluate IMU boards


Figure 2: F4/F7 electromagnetic shaker platform used for testing sensor accuracy in a vibration heavy environment