An immune network approach to sensor-network with self-organization for sensor and process faults

Abstract

The self-organizing diagnosis has been studied by applying the idea of autonomous and decentralized systems extracted from the concept of immune network. The model implements network-level recognition by connecting information from local recognition units by dynamical evaluation chain. The model has been further elaborated for engineering concerns of identifying not only sensor faults but process faults. The sensor faults will be identied by evaluating reliability of data from sensor, while the process faults will be identied by evaluating that of constraints that must be satised among these data. We have demonstrated that the extended sensor network will work against both sensor faults and process faults by an illustrative example.