A.F. Vasbieva - Student, Department of Automation and Telemechanics, Perm National Research Polytechnic University. E-mail: email@example.com
L.M. Oniskiva - Ph.D. (Eng.), Associate Professor, Department of Applied Mathematics, Perm National Research Polytechnic University. E-mail: firstname.lastname@example.org
A.I. Posyagin - Assistant, Department of Automation and Telemechanics, Perm National Research Polytechnic University. E-mail: email@example.com
A.A. Yuzhakov - Dr.Sc. (Eng.), Professor, Head of a Department of Automation and Telemechanics, Perm National Research Polytechnic University
One of the important elements of automatic-control systems (ACS) is the analog-to-digital converter (ADC) as the matching device between detectors\' analog signals and the microcontroller. The contemporary researches in the field of new structures and algorithms of construction of ADC are aimed, first of all, at the creation of the devices with streaming dynamic architecture and the increase of fault-tolerance. In order to solve these problems the usage of the neural network (NN) was suggested since using NN there is the possibility to designate universal elements in the structure and their combination can help to change ADC\'s capacity, as well as their substitution provides the high level of fault-tolerance. In order to execute the analog-to-digital conversion the method of discharge weighting when receipt of the bit value requires two cycles: at the first cycle the bit is set to one and at the second cycle is either left at one, if the ba-lancing voltage doesn\'t exceed the input voltage, or is set to zero, if the input voltage exceeds the balancing one. The balancing voltage arrives from the usage of the R-2R matrix.
The article concerns the basic measuring neuron (BMN) in the self-routing analog-to-digital converter which is divided into 3 blocks: the measuring block, the directing block and the controlling block and also 2 auxiliary blocks. The structural scheme of BMN is presented where interaction principles and connection with other neurons mechanisms are described. The auxiliary blocks constitute input and output keys which are essential for connecting the measuring parts of several neurons into an individual analog-to-digital converter (IADC). The measuring part of BMN constitutes the element of the R-2R matrix, the elements of receipt and holding a discharge and also shift register trigger for the transmission of the counting value to the output neuron. The directing block is designed for all other blocks coordination as well as the conversion process defining the sequence of neurons combined into IADC operation. The controlling block is intended to check the working capacity of a particular BMN and by the use of the availability operating flag to transmit this information to the other points of the neural network. On the basis of the obtained BMN structure the functional scheme of all blocks is developed and also the model is created in the MultiSim program in order to examine all the algorithms put into this structure. On the basis of the BMN model the model of the one-input ADC is created which is capable of measuring one constant analog signal with the varying capacity from 2 to 8, receiving the stable value of the output signal. A part of NN in the created model of ADC is hand-operated since the full value operating, apart from BMN, requires modeling of the commutators, input and output neurons.