The Reduction of the Dimensionality of Redundant Sensor Data Using Principal Component Analysis

The safety related sensors of nuclear power plant are redundant. Redundant sensor-systems achieve fault tolerance by duplication of components. It increases the ability of systems to interact with their environment by combining independent sensor readings into logical representations. Sensor integration of highly redundant systems offers these advantages: 1) Multiple inaccurate sensors can cost less than a few accurate sensors; 2) Sensor reliability may increase; 3) Sensor efficiency and performance can be enhanced; 4) Self-calibration can be attained. But the feasibility of the systems requires attention be paid to both reliability bounds and cost. Several on-line monitoring techniques have been developed that calculate the parameter estimate using only the measurements from a group of redundant instrument channels. These techniques are commonly referred to as redundant sensor calibration monitoring models. In this paper, we reduced the dimensionality of redundant sensor data using principal component analysis