An Immune Network Approach to Sensor-based Diagnosis by Self-organization
Yoshiteru Ishida
Graduate School of Information Science,
Nara Institute of Science and Technology, Ikoma,
Nara, 630-01 Japan
Abstract
Self-organizing diagnosis is studied by applying the dynamic propagation of active states derived from the concept of an immune network. The proposed model is a mutual vote network which is a modification of the majority net. The model implements network-level recognition by connecting information from local recognition agents by dynamical evaluation chains. The model has been further elaborated to address the engineering concerns of identifying not only sensor faults but also process faults. The sensor faults are identified by evaluating the reliability of data from a sensor. Process faults are identified by evaluating constraints that must be satisfied among these data. We demonstrate that the sensor network works against both sensor faults and process faults by an illustrative example.
