2687-0517 14 3 2021 1-75

RAR RUS 7-19 Peter the Great St. Petersburg Polytechnic University Konstantinov Andrei andrue.konst@gmail.com

St. Petersburg, Russian Federation

Deep gradient boosting for regression problems Deep Forest is a new machine-learning algorithm that combines the advantages of Deep Neural Networks and Decision Trees. It uses representation learning and allows building accurate compositions with a small amount of training data. A significant disadvantage of this approach is the inability to apply it directly to regression problems. First, feature generation method should be determined. Secondly, when replacing classification models with regression models, the set of distinct values of the Deep Forest model becomes limited by the set of values of its last layer. To eliminate the shortcomings, a new model, the Deep gradient boosting is proposed. The main idea is to iteratively improve the prediction using a new feature space. Features are generated based on the predictions of previously constructed cascade layers, by transforming predictions to a probability distribution. To reduce the time of model construction and overfitting, a mechanism for points screening is proposed. Experiments show the effectiveness of the proposed algorithm, in comparison with many existing methods for solving regression problems. 10.18721/JCSTCS.14301 004 regression Deep Forest gradient boosting ensembles decision trees https://infocom.spbstu.ru/article/2021.70.1/

RAR RUS 20-32 Peter the Great St. Petersburg Polytechnic University Kudriashov Nikita niki94@yandex.ru

St. Petersburg, Russian Federation

Dynamic energy consumption rationing based on machine learning algorithms for oil refining tasks Energy consumption rationing is necessary for high-quality production planning, and allows optimizing their use. This paper provides an analysis of various approaches to building a model of energy consumption, describes their limitations and new approaches to dynamic rationing. As the object of modeling the ELOU-AVT-6 (CDU/VDU-6) unit has been taken. Such units are intended for desalination and primary fractionation of oil. Functional requirements for the algorithms have been formed, based on real production needs. As the solution, models based on machine learning algorithms have been analyzed. These algorithms include CatBoost Regressor, Gradient tree boosting, Random Forest, ElasticNet and artificial neural networks. The analysis of the modeling results and comparison of the accuracy of the models is carried out. The paper also demonstrates a scenario of using a dynamic rationing model to analyze the causes of deviations of the actual consumption values from the planned ones. 10.18721/JCSTCS.14302 004 energy consumption rationing machine learning digital twin oil refining factor analysis https://infocom.spbstu.ru/article/2021.70.2/

RAR RUS 33-42 St. Petersburg Federal Research Center of the Russian Academy of Sciences Svistunova Aliaksandra svistunova_alexandra@bk.ru

St. Petersburg, Russian Federation

St. Petersburg Federal Research Center of the Russian Academy of Sciences Khasanov Dmitry

St. Petersburg, Russian Federation

Improving the efficiency of traffic management in a metropolis based on computer simulation This article considers an algorithm for traffic management. The process of simulation modeling of traffic flow at a road intersection in a metropolis is described. The optimization of the obtained simulation model is performed in the software system of simulation maintenance Anylogic. The application of the developed model is shown on the real example of a megalopolis road intersection. We considered the classification of highways, and analyzed the built model on the basis of the existing megalopolis crossroads, which allowed us to obtain the data comparable with the existing system. We considered and adopted methods to solve the high traffic problem by re-organizing the intersection using simulation modeling. The final system showed a 27 % throughput growth in the whole system and a 40 % growth in the main direction without traffic jams. The analysis shows the prospects of using the proposed simulation model to study real-world traffic flow management processes in order to study their behavior. 10.18721/JCSTCS.14303 656.1, 004.94 megalopolis control simulation modeling algorithm model transport stream car roads https://infocom.spbstu.ru/article/2021.70.3/

RAR RUS 43-55 Peter the Great St. Petersburg Polytechnic University Rachitskaya Antonina antonina_92@list.ru

St. Petersburg, Russian Federation

Semi-natural modeling for GNSS integrity monitoring algorithm The paper considers the suboptimal version of GNSS integrity monitoring algorithm involving multichannel signal processing. This algorithm was examined in terms of probability-based cha-racteristics obtained during semi-natural modeling. Such modeling assumes that multichannel snapshots are getting from real channels of multichannel GNSS receiver including antenna array when all subsequent procedures are implemented in Matlab later. Probability-based characteristics obtained in such way consequently checked with similar characteristics obtained by Matlab simulation ideal model, which ignored probable effects of signal transmission and reception in real environment. It was shown the level of similarity between characteristics of both types, and also clarified the conditions when the characteristics are close to each other, and the conditions when the difference between them is significant. The main reason of such difference was found out empirically. 10.18721/JCSTCS.14304 621.37 generalized maximum likelihood ratio test probability-based characteristics confi-dence interval non-identity of channels multichannel receiver https://infocom.spbstu.ru/article/2021.70.4/

RAR RUS 56-63 Aslanov Gadarbek uits@dstu.ru Daghestan State Technical University Aslanov Tagirbek tabasik@gmail.com

Makhachkala, Republic of Daghestan, Russian Federation

Kazibekov Rustam kazibrus11@mail.ru Musaibov Rashid rashid_musaibov@mail.ru Influence of transients in the information processing channel of the airport automatic radio direction finder on the direction finding accuracy The article is devoted to the study of the transients influence in the information processing channel of an aerodrome automatic direction finder (ADF) on the accuracy of direction finding. This task is of great importance, since improving the accuracy of navigation equipment allows setting the separation standards higher and improving the aircraft flights safety. When the phase difference between the signals on the neighboring vibrators of the antenna system (AS) is equal to 180°, a signal loss is observed in the low-frequency filter of the ADF due to the amplitudes equality of the component signals from the kth and k + 1th vibrators, that leads to failures (the appearance of abnormal errors) in the ADF operation. When using the radio direction finders operation to find speech-modulated signals (an amplitude-modulated signal), certain gaps emerge in the direction-finding signal due to the operation of automatic gain control in the ADF receiver, which leads to the accuracy deterioration of the direction-finding. We propose methods of reducing the transients influence on the accuracy of radio direction finding. To eliminate the influence of transients caused by the operation of the Automatic Gain Control, taking into account the fact that the processing of direction finding information in the ADF is carried out on a channel microprocessor, it is necessary to assign weight coefficients to the directions calculated for eight switching cycles of the AS elements and calculate the value of the weighted average finding. 10.18721/JCSTCS.14305 621.396 radio direction finder antenna system transients finding accuracy radio direction finder failures https://infocom.spbstu.ru/article/2021.70.5/

RAR RUS 64-71 Salahaddin University Assim Ara Abdulsatar araabdulsattar@gmail.com

Erbil, Iraq

Balashov Evgeny balashov_ev@mail.ru Zero-drift operational amplifiers This article considers the design and implementation of four different zero-drift operational amplifiers with 50 nm technology CMOS and compares their characteristics. The aim is minimizing input offset voltage and flicker noise. Offset voltage is unavoidable in operational amplifiers, because no two transistors can be identical. A small difference in their dimensions (length or width) gives rise to this undesirable effect, the value of offset voltage in common operational amplifiers is less than 10 mV. In this article, two major techniques of dynamic offset cancellation, chopping and auto-zeroing, are considered. The operational amplifier with chopping shows the best result among the four amplifiers. 10.18721/JCSTCS.14306 621.375 zero-drift operational amplifiers auto-zeroing chopper amplifier offset voltage reduction CMOS https://infocom.spbstu.ru/article/2021.70.6/