Dr. Dipankar Dasgupta
Dr. Dipankar Dasgupta, IEEE Fellow
William Hill Professor in Cyber Security
Director, Center for Information Assurance
Director, Intelligent Security Systems Research Laboratory
Dr. Dasgupta will organize symposium on Computational Intelligence in Cyber Security (CICS) at the IEEE Symposium Series on Computational Intelligence (SSCI) in Xiamen, China from Dec 6 - 9, 2019.Click here for more details    Dr. Dasgupta has organized symposium on Computational Intelligence in Cyber Security (CICS) at the IEEE Symposium Series on Computational Intelligence (SSCI) in Bengaluru, India from November 18-21, 2018.Click here for more details    Dr. Dasgupta has organized IEEE Symposium on Computational Intelligence in Cyber Security (CICS 2017) at Hawaii, USA from November 27-December 1, 2017.     Program Committee Member of the 1st IEEE International Workshop on Cyber Resiliency Economics (CRE 2016) , Vienna, Austria, August 1-3, 2016.   

Fault Detection in Manufacturing using an Immunologically Inspired Technique

Project Objective:

The objective of this research is to develop an efficient fault detection algorithm based on immunological principles. From an information-processing perspective, the immune system is a remarkable, parallel and distributed adaptive system. It uses learning, memory, and associative retrieval to solve recognition and classification tasks (of its defense mechanism). The proposed immunity-based fault detection algorithm is a probabilistic method that uses a negative selection mechanism to detect any changes in the normal behavior of a monitored system.

Proposed Research:

Manufacturers are always looking for ways to improve productivity without compromising on quality of manufacturing processes. To this end, much attention has been directed towards automated manufacturing. Detecting fault or anomaly in sensory measurements is a problem of great practical interest in many manufacturing and signal processing applications, where it is necessary to detect anomalies or imperfections in system or process behavior. For example, in drilling or high-speed milling industries, on-line monitoring of the tool breakage is a key component in unmanned machining operations. In safety-critical applications, it is essential to detect the occurrence of unnatural events as quickly as possible before any significant performance degradation results. This can be done by continuous monitoring of the system for changes from the normal behavior patterns.

We have experimented with several data sets including some real sensory data and the initial results are very encouraging. In one such experiments, we applied the algorithm for Tool Breakage Detection in a milling operation. In this implementation, the tool breakage detection problem is formulated as the problem of detecting temporal changes in the cutting force pattern that results from a broken cutter. That is, the new data patterns are monitored to check for whether or not the current pattern is different from the established normal pattern, where a difference implies a shift in the cutting force dynamics. The detection algorithm was successful in detecting the existence of broken teeth from simulated cutting force signals in a milling process.

The goal of this work is to develop an efficient detection algorithm that can be used to alert an operator to any changes in steady-state characteristics of a monitored system. This approach collects knowledge about the normal behavior of a system from an historical data set, and generates a set of detectors that probabilistically notice any deviation from the normal behavior of the system. The detection system can be updated by generating a new set of detectors as the normal system behavior shifts due to aging, system modifications, change in operating environments, etc.