The analysis of the problem area of application of medium power frequency-regulated electric drives in the conditions of influence of ionizing radiation is carried out in the work.
The radioactive waste storage bridge crane uses general industrial design of frequency-regulated electric drives, scilicet no activity and special technologies are used for radiation protection of equipment. The value of the absorbed dose of ionizing radiation was chosen to take into account the radiation effects. The structure and materials of each of the electric drive elements review was performed.
It was found that semiconductor devices have the lowest resistance to ionizing radiation, in particular, the degradation of silicon properties begins with an absorbed dose of 1 krad, in contrast to dielectrics (1 Mrad) and structural steel (100 Mrad). Further research focuses on the weakest link in the system that is semiconductor devices.
The structure of the power semiconductor channel of the frequency-regulated electric drive is considered. It is revealed that insulated gate bipolar transistors are used as inverter switches for medium power frequency converters. Since the radiation effects study of such transistors in the literature has not been identified, it was decided to use a substitution scheme for an insulated gate bipolar transistor. To study the radiation effects in insulated gate bipolar transistor, a substitution scheme based on elements whose radiation effects are known was chosen.
Two types of inverter power switch control drivers are considered – based on field effect (metal-oxide-semiconductor) and bipolar transistors. The electrical circuits of the frequency converter power semiconductor channel with these drivers are constructed.
Mathematical models of frequency converter semiconductor power channels for accounting of radiation effects are synthesized and investigated. To do this, the radiation effects in diodes, bipolar and field effect (metal-oxide-semiconductor) transistors and optocouplers were considered.
In the process of comparison the two options, it was found that the system with drivers based on bipolar transistors is more stable than the system with drivers based on field-effect transistors.
The synthesized model exploration allowed to determine the semiconductor device in the frequency converter power channel, which change of parameters most significantly affects the operation of the system. It was found that in a system with drivers based on bipolar transistors, the use of radiation-insensitive bipolar transistor in the IGBT provides constant frequency converter output voltage amplitude with increasing dose up to 800 krad, after which there is a functional failure. It was found that the radiation change of the optocoupler parameters has the most significant effect on the operation of the entire system.
Since the optical data transmission system is, in fact, an optocoupler, which emitter and photodetector are separated from each other in space by a considerable distance (up to 150 meters), it requires additional research.
The functional scheme and technical characteristics of the optical data transmission system are considered. Bit error rate is chosen as a criterion for data transfer quality. The analysis of factors that degrade the quality of information transmission is performed, the following factors include: weakening of the received signal power due to not infinitely narrow pattern of the transmitter; noise interference of the transmitter, channel and receiver; degradation caused by ionizing radiation; natural laser degradation; attenuation of laser radiation in the atmosphere.
To improve the quality of transmission in the optical data transmission system, the use of an artificial neural network as part of a demodulator has been proposed. Such demodulator operates with vectors instead of instantaneous signal values, which allows to reliably classify the signal at low values of the signal-to-noise ratio. A training sample for an artificial neural network was formed and processed. With considering the classification task, nine types of artificial neural network architectures have been selected, trained and tested. A positive result was achieved with only two types: feed-forward with two delay lines and probablistic radial basis function network. The feed-forward network with 24 hidden layer neurons and 2 input delay lines showed the lowest bit error rate at a signal-to-noise ratio of more than 0.7 dB. At lower signal-to-noise ratios, its curve coincides with the curve of an artificial neural network with 25 neurons and they are lower than the curve of a traditional demodulator.
Key words: electric drive, frequency converter, inverter, electrotechnical device, hoisting machine, bridge crane, radioactive waste, ionizing radiation, optical data transmission system, mathematical modelling.
