Design of neuro fuzzy network for temperature control of furnance

Abstract

The design of an advanced control methodology for professional nuclear reactor has been done. This design will increase complexity and reduces the risk factor as nuclear reactor uses highly explosive and radioactive elements as a fuel and required a very highly precise and accurate control at real time. To achieve the control in this process combination of neural and fuzzy logic is used which provide autonomous control to heating and cooling of the reactor. The system accepts the weights from user which are to be loaded into artificial neural network nodes. When reactor is turned ON. The hardware that is installed in reactor senses the temperature with the help of temperature sensor, and process the measured data to a high speed microcontroller unit. The instant temperature readings are then passed to the computer system located in monitoring office far away from reactor. Reactors coolant and heating controls are also connected to microcontroller unit within the reactor. This system developed can support n number of reactors at a time i.e it can provide temperature measurement for multiple reactors simultaneously and can effectively control them. Application software is programmed on computer system which receives updated data from reactors microcontroller unit and display current temperature in it. It gets set limit values initially from user and whenever incoming temperature matches or crosses the set limits then it fires the various outputs according to values set to provide amount of coolant and designs to turn the control of reactor automatically. Soul of this application software is artificial intelligence which is provided by concepts of fuzzy sets or rules based expert systems of artificial neural networks. Intelligence of this application software is to drive nuclear reactor in a smart and control way by itself. It totally eliminates human interaction with controls of reactor. It itself decides that what amount of coolant is needed at which temperature range. This application software is developed of fuzzy logic and artificial neural network nodes technology base and the control system results in security and precise control system.