COGNITIVE PROCESSOR DESIGN BASED ON KNOWLEDGE GROWING SYSTEM (KGS)

• Main Author: Olivia Sereati , Catherine
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/26212
• Institution: Institut Teknologi Bandung

The Intelligence of Smart Instrumentation System starts to be developed on sensors and actuators known as Smart Sensors and Smart Actuators that are able to do their own calibration. Furthermore, a smart data bus was developed which has ability to detect the fault and present the Fault Tolerant Instrumentation System. Computer intervention began the era of a more intelligent and independent Instrumentation System based on the concept of Artificial Neural Networks, Fuzzy Logic, and Genetic Algorithm. But it is still uses computers outside the instrumentation system. This dissertation research discusses the development of intelligent processors that are able to grow their knowledge independently so that they can realize an Intelligent Instrumentation System that is self-regulating independently. Processor development that has artificial cognitive abilities as a Cognitive Processor can improve intelligence and independence of intelligent instrument systems. The Cognitive Processor developed in this dissertation research is capable of information fusion based on a new concept in the field of Artificial Intelligence (AI) called Knowledge Growing System (KGS) which was discovered in 2009 by Arwin Datumaya Wahyudi Sumari, Adang Suwandi Ahmad, Aciek Ida Wuryandari, and Jaka Sembiring. KGS is a system built to emulate the process of growing new knowledge that occurs in the human brain when getting new information from its sensing organs, over time. The Cognitive Processor Architecture is designed by implementing the KGS algorithm into computational KGS based on the ASSA2010 and OMASSA2010 formulas to obtain the data path of the processor. By using the VHDL programming language then it is implemented into the FPGA board, resulting in a Cognitive Processor design. The complexity of the circuit is determined by the number of sensor combinations and hypotheses that are input for this processor. From this research, there are two alternative designs of processor cognitive architecture, namely: Fully-combinational-based architecture that allows the KGS computation process to be run in real-time without delay in computing time. With this design the number of logic and element resources will increase as the combination of sensors and hypotheses is increased. Power usage on the processor will also increase, due to the large number of logic elements used. The second alternative architecture is a systolic-array based architecture that allows designs to be made with fewer elemental resources. This results in reduced design area, and savings in processor power. Compensation is that the computation time becomes slower, due to the time-delay due to <br /> <br /> scheduling the use of several processor elements each time the processor gets a hypothesis-sensor value that changes every time. This design has been able to represent the process of knowledge growing that occurs at each time of observation. The representation of knowledge growth is shown in the form of the value of the Degree of Certainty (DoC) obtained from KGS computing. The DoC value shows the best hypothesis value of all possible hypotheses that may arise from each time of observation. The results of this research dissertation show that the Cognitive Processor will realize a system that has Cognitive ability, knowledge growing. Furthermore, this Cognitive Processor will open wider opportunities for the development of a more autonomous Smart Instrumentation System including monitoring and early detection and control functions.