Actuator performance envelope through nonlinear test bed
With the increased importance of actuator performance in robotic applications, many robot manufacturers have focused on assuring quality, reliability, conformance to design, and application suitability in their actuators. Performance metrics for actuator evaluation have been proposed to improve understanding of an actuator’s nonlinear characteristics. Thorough actuator testing can provide necessary feedback information on the effectiveness of the robot actuator for a specific task. This report describes the development and use of performance maps and performance envelopes for electromechanical actuators using an advanced dynamic dynamometer test bed. Initial test data in the form of performance maps is created for a range of rated output power and dynamic loading for brushless DC motors. The core contributions of this research are the creation and manipulation of actuator performance maps and envelopes, and the design and development of a nonlinear test bed for electromechanical actuators. To create a performance map, specific criteria must be developed to describe the performance characteristics of an actuator. The performance criteria are formulated to define, anticipate, and evaluate the performance responses of the system and can provide metrics for measuring and controlling actuator performance and help identify nonlinear components in actuator modeling. These criteria form the basis for performance maps, which is basically a data representation of the criteria over a range of input and output conditions. A representative set of ten actuator performance maps are suggested based on several independent control parameters such as current and voltage, and output conditions such as torque and speed. After completing the development and analysis of these ten performance maps, the process for building unique performance envelopes is described. Performance envelopes are created from performance maps and are defined as the combined optimum surface in the full set of maps, which satisfy a combination of criteria in each operational region. Performance envelopes can safely increase the number of control surfaces without violating actuator constraints. This provides more choices in an operational range as compared to statically developed torque-speed curves. For the creation of these performance maps, five test regimes are suggested such as dynamic loading, static loading, transient response, torque ripple and statically changing temperature tests. Test regimes are evaluated under two types of loads: a static loading and a sinusoidal loading. For these test protocols, a Nonlinear Test Bed for Actuators (NTBA) has been designed and built to measure and record an array of physical performance properties. The test bed is specifically designed with the capability to emulate high bandwidth complex duty cycle loads. It is comprised of a 7.2 kW servo motor (which serves as a dynamic load emulator), a 26 Nm hysteresis brake, an electromechanical clutch, full sensor array (includes current, temperature, torque, magnetic flux density sensors) for comprehensive monitoring of test variables, and supporting motion control hardware to operate the test bed. Ten performance maps (i.e., operational margin, efficiency, power losses, rise time, torque-current, torque ripple, acceleration, max magnetic flux energy, magnetic flux density, and temperature) for a commercial brushless DC motor are developed in 3D surface displays using the test bed. The performance maps collected from repeated tests (more than 20 times for each test protocol) of the test motor provide an insight into the complexity of the electromechanical response and also provide metrics for measuring motor current states during operation. Standard deviation and average values are also calculated and used to develop the upper and lower error bounds of the magnetic flux energy performance map. In addition, the Box-Jenkins ARIMA method is used to analyze all of the test data and build a mathematical representation of sensor derived data with respect to time. This stochastic ARIMA equation is used to create each data point in a performance map. The performance map plots are normalized by using the maximum values over the range of motor operation. One performance envelope plot with respect to torque and speed is illustrated in this research. Finally, the performance data plots used in the trade literature and research papers are compared with the test result plots presented as actuator performance maps. The performance plots in the literature have several drawbacks for users, which are the lack of operational values in abnormal operating conditions and the lack of reference plots to fully describe and enable the management of the performance of the motor considering all of the operational conditions. On the other hand, the maps obtained in the experiments using the NTBA provide enough metric values to manage the actuator to achieve optimum performance for a multitude of desired operating conditions (maximum efficiency, soft start, low noise, maximum life, etc.).