Applications

Real-Time Analysis of Dynamic System Behavior

For many projects, the objective is to develop a dynamic simulation to study the response to certain stimuli that cannot be analyzed in any other way. For example, an aircraft is made up of many subsystems that all have to interact with each other and the most effective way to study the impact of one system on others is to test it in a virtual prototype of the aircraft. Also, by putting an operator in the loop, such as a test pilot, many design flaws can be addressed long before a physical prototype is built.

Physical Component Testing with Virtual System

There is a growing trend towards testing new engineering components in a dynamic test cell that simulates the behavior of the system in which the component will be installed. For example, a vehicle transmission can be installed in a virtual vehicle test cell that behaves like a real test vehicle. In this way the transmission manufacturer saves on the cost of a test vehicle, does not need to rely on the stamina of the driver and the tests are entirely repeatable.

Controller Development

The V cycle is a representation of the typical process used for the development of control systems. Regardless of the application, automotive, aerospace, robotics or power-electronics, the use of real-time and non-real-time simulation usually stems from the need to develop a robust control strategy for a complex system. RT-LAB-based systems can help every step of the way.

The first step of the cycle is the requirements analysis. At this stage, it is necessary to define the extent of the system to control (an engine, motor, robot arm, aircraft), referred to using the generic term plant, as well as the inputs that the controller can influence (fuel flow, input voltage or frequency, actuator torque) and outputs to monitor (speed, position, altitude). It is also very important to define the performance (stabilization time, overshoot, phase margin, gain margin) that the controlled system needs to achieve. This will determine if the controller is acceptable or not.

The next stage is the high-level design, also called software-in-the-loop. Here, the plant model is used in conjunction with a controller model to test the validity of the planned control strategies. This step does require simulation, but usually not in real time since both system models (plant and controller) are part of a common mathematical model. However, it might be useful to distribute the processing of this model on multiple processors in order to accelerate the simulation and move to the next step faster. This is something that RT-LAB handles transparently.

Once the control strategy has been selected, we move on to the function prototyping. This is the implementation stage of the control algorithm into a system capable of running this algorithm in real time and interfacing with the real plant. Throughout the development cycle, it is desirable to limit the use of a real plant, as this is time-consuming and expensive, or even impossible if the system for which it is being destined is being developed in parallel. Here, however, it is necessary in order to validate the control strategy and to adjust the control parameters. The more realistic the plant model is, the less time is necessary here because the high-level design stage will have yielded a better controller already. A compact RT-LAB target system can be used as a prototyping platform, complete with I/O interfaces. For research projects and low-volume productions, the RT-LAB system can be used as the final controller platform so the prototyping can actually be done directly on the same hardware.

The control algorithm should now work as expected. It is thus time translate the algorithm to production code and implant it into the prototype platform. The product of this is a real controller, perhaps with some added diagnostics functions, in a production-ready format.

With the prototype controller, the first step is to perform unit testing. This low-level testing usually involves the validation of the input and output interfaces and the basic operation of the controller. Because the tests performed here involve only basic functionalities, a simple open-loop simulator can be used. This means that the test system simply excites the inputs of the controller and monitors its outputs. There is no instance of the plant model needed to "close" the functional loop. The system used here can be a dedicated open-loop test system, such as Opal-RT's TestDrive system, which can also be used at the production stage for the functional testing of the production controllers. Alternatively, using an RT-LAB Engineering Simulator it is possible to use the full HIL system used for integration testing with a simplified model.

Once the basic functionalities have been validated, integration testing or hardware-in-the-loop tests will be done to validate the extent of the control algorithm as implemented in the prototype hardware. This includes all standard operating conditions and as many corner cases as possible. In order to do this, a real-time simulator running the plant model is the best test system there is. This is both more cost-effective and more flexible in generating and duplicating normal operating and failure scenarios than using a real plant. Depending on the model complexity, a single-target or distributed RT-LAB simulator can be used to run the plant model in real-time.

Once all of this is done, the prototype controller should have been thoroughly tested and can be connected to a real plant. Operational testing will allow the final calibration and adjustments of the control parameters to take place.

Testing of Production Controllers

Once the controller has gone into production, instead of inspecting the controllers with a real plant, they can be readily tested using the same plant simulator that was used in the prototyping stage, allowing rigorous, repeatable, automated tests to be carried out. As a prime example, the TestDrive system was designed with production testing in mind. From integrated power-moding to software-configurable signal conditioning, this system implements everything for the quick changeover required for production testing.

High-Speed Data Logging and Event Capture

RT-LAB's unique high-speed data logging and event capture features allow data to be acquired and streamed to disk on the target system, making an ideal tool for data logging applications. Added to this, its embedded support allows you to create standalone systems for unattended or remote operation.

Operator Training

Since the mathematical model of a system has been developed for the design stage, RT-LAB provides a very economical platform for training operators of the system. RT-LAB's API allows the model to be connected to the real operator interface in order to provide a close-to-reality environment in which the operator can learn procedures and deal with faults in the system without causing damage or injury.