.
A scientist from the National University of Technology Research (MISS) has proposed a new alternative neuronal network, an interconnected set of virtual neurons created by software programs similar to that of a biological neuron.
Dr. Anton Gloshenko of the Department of Management Systems and Informatics at the Stary Oscell Institute of Technology at the University of Mississauga said: "The nerve network modulator, designed at the University of Mississauga, aims to increase the energy efficiency of metal heating furnaces with energy consumption Up to 100 megawatts. "
As it is known, during the operation of the furnaces are exposed to a variety of disturbances – for example, opening the separation barrier for unloading and loading the metals leads to heat loss, and the pollution of the gas burners leads to lower efficiency of the combustion process, . Because these furnaces are usually controlled by a fixed-line linear controller, the characteristic change factor is not considered over time, which reduces the quality of the control and results in loss of power.
In an interview with the agency ".", Anton Goloshenko said:
"In order to solve traditional problems, we propose building a adaptive control system in the form of a neural network modulator, which simultaneously adjusts the linear control parameters so that the quality of the furnace control in all areas of the operating systems is equally high, thus reducing consumption energy".
The researcher pointed out that this modern approach is caused by the combination of intelligent techniques in the morph, and talk about the neural network and knowledge bases. The neural network calculates the values of the linear regulators used in the oven and learns during operation to monitor changes in the furnace.
The Russian scientist in this context, saying:
"The key questions are when and how fast the neural network can learn, and of course, these questions answer the knowledge base that reflects the experience of the automation engineer, describing the situation and when to adjust the regulator and the equations to calculate the speed of the neural network learning. The use of the neural network modulator does not require the formulation of a controller model nor a clear model, and it helps to track the task schedule when the oven indicators change and compensate for the disturbance of the furnace. "
The module is designed as a task unit that can be placed in the RAM of widely distributed logical controllers in the mining industry. The outputs and inputs of this unit are connected to a linear regulator connected to the console and signals obtained from the outside.
"The use of the microprocessor requires no more expense because, from the point of view of hardware and software, nothing will change in the current control system in the furnace, but the application of this approach will allow more efficient use of heating furnaces," said Anton Gluchenko. For metals by between 5-10%. "
It is worth mentioning that an experimental furnace was built in Russia to smelter iron from mineral waste. As reported by the University of Mississippi, it is also planned to expand the scope of the uses of the synthesizer after it has been processed and tested in various electric motors.
Read more scientific material from the University of "Mays"
0 Comments