QMC DATA MINING PROGRAM for Windows
SENSORFAULT METER PROGRAM
The SensorFault Meter uses cause and effect modeling technologies to establish the validity of sensor measurements. This approach allows verification by the VB Model, Neural Networks, e Model, and Multivariable Regression to determine statistically when measurements are in error.
As an example, predictions of a measured polymer property determines the product quality. The measured value is determined by lab analysis and can be subject to error due to faulty equipment, calibration changes, or analysis errors. Poor product quality will cause a downgraded selling price of the polymer. The SensorFault Meter uses modeling techniques to validate actual values with predicted values. The measured polymer property is indicated by a collaboration of four (4) model predictions. This corroborates the lab measurement of the polymer property since the predicted property is at the desired product specification. The SensorFault Flag uses a voting protocol statistically derived to ensure that neither the lab readings nor the sensors are in error. Process sensitivities can be adjusted to meet plant requirements. The benefit of the SensorFault Meter is detection of faulty sensors or inaccurate readings that leads to degradation in product quality and profit.
System Requirements
Available for: Win95/98/NT
Minimum CPU Requirements: 486
Minimum Ram: 8 MB
Version: 6
Published: 2000