Input Shaping Algorithm
Advanced Crane Control Laboratory
Input Shaping Algorithm.
(Click to Enlarge)
The PLC was programmed to run various input shaping algorithms on the crane. One such algorithm, a Zero Vibration Input Shaper, is shown in the above figure. The algorithm works by taking a desired command and convolving it with a series of impulses. The amplitudes and times of the impulses are tuned to the system parameters in such a way that vibration induced by the first impulse is canceled by the second impulse. The resulting command will not induce vibration when it is run on the system.

In our particular case the PLC is constantly monitoring the control pendant for button presses. As soon as a button press is detected it automatically calculates the velocity set-points according to the above figure. These set-points are fed to the motor drives and used to control the trolley’s velocity. The result is that the crane can be moved around without any swing. All of these calculations take place inside the PLC, and are totally transparent to the crane operator.

The figure below compares payload swing with and without input shaping. This figure shows real data measured from the crane. Notice that there is a significant reduction of swing when input shaping is used.
Payload Time Response With and Without Input Shaping.
(Click to Enlarge)

Obstacle Course With and Without Input Shaping
(Click to Enlarge)
Because input shaping reduces crane swing it makes operating the crane much safer and much more efficient. Below is a diagram of an obstacle course with a starting point and a goal. The crane was driven through the obstacle course with and without input shaping. Notice that without input shaping the payload collides with many of the obstacles, which could be very dangerous in a real-life working environment. Also notice that when the crane reaches the goal it has significant residual swing. In a manufacturing setting, the operator would have to wait for this swing to die out before continuing operation, which leads to inefficient throughput times. The input shaped does not suffer from either of these two problems.