Understanding Smart Sensors - Nomads.usp

Control Techniques 163ObserverHB−+X e (n+1)+z −1 z −1AX e (n)DD´ø e+−velEstimatedvelocityuControlsignalB+X (n+1)+X (n)DøOutput positionA(Motor and drive)Figure 7.10 The block diagram of an observer for closed-loop velocity control.delay operator, and H is the observer vector. The lower portion of Figure 7.10shows the block diagram of the actual motor, and the upper portion containsthe observer algorithm that is in the software [20].The observer estimates both velocity and position. A comparison of theestimated position with the measured position is used to correct for the nextsampling period. Estimation error is used to achieve a better estimate for thenext sampling interval. The observer gain vector H is used to modify theobserver dynamics.Feedforward techniques use sensors to modify the estimated values forsystem inputs that are in the observer model [21]. The growing complexity ofautomotive engine controls to meet increasingly stringent emissions targetsmay require systems that anticipate, act, and then adjust algorithms based onmeasurements on a cylinder-by-cylinder basis. Model-based methods can alsoreduce the calibration effort that takes considerable time and effort foreach internal combustion engine that a manufacturer produces. Sensor-basedintelligent decision making has also been investigated as a possible solutionin computer-based untended machining [22]. Those and other techniquesthat expand the capability of neural networks rely heavily on sensor-systemsynergy.