PNRPU Bulletin. Electrotechnics, Informational Technologies, Control Systems

The functioning of shipboard electrical equipment is closely related with the quality of electrical energy. Changing the frequency of the alternating current leads to a change in the operating modes of electrical systems in the ship electric power system. In particular, to a decrease in the performance of pumps and fans, to a interruption of the stability of the generators and turbines of the electricity power station and to a decrease in the overload capacity of induction motors. The influence of the frequency change introduces in the regulation of the moments commutation to the semiconductor valves. Pulses of control formed by phase-shifting devices of the rectifier control system. As a result, there is a deterioration in the quality of the regulatory process and a change in static and dynamic characteristics, as well as the possibility of dangerous situations for process equipment. The digital phase-shifting device with a memory unit was developed for stabilize the output parameters of the semiconductor converter. This device allows to correct the commutating moments of the semiconductor gate depending on the frequency of the alternating current. The purpose of this work is to analyze the functioning of various types of phase-shifting devices based on the static and dynamic characteristics of the relative error of average rectified voltage depending on the voltage frequency and the angle of control. To study the proposed device, a series of experiments was carried out on the basis of a simulation model in the MatLab environment. The processing of the results obtained during the study of the simulation model proposed by the digital phase-shifting device as part of a three-phase rectifier with asymmetric control together with the prototype device is presented. The results of the analysis showed the possibility of using the device with resistance to instability of changes in the frequency of the supply network as part of a three-phase rectifier in order to reduce the deviation of the rectified voltage and smoothly change the angle of control for powering DC consumers in ship power systems. The technical result of the developed device is to increase the time resolution in contrast to the device prototype.

Programmable logic integrated circuits are widely used in digital equipment, including equipment for critical applications, for example, in medicine, avionics, in control systems for spacecraft, atomic power stations, in military and special equipment. Currently, the number of logical elements in the most advanced FPGA samples is approaching tens of millions. Therefore, the diagnosis of logic FPGA responsible use is a very urgent task. This task is complicated by the fact that the number of variables increases and now there are individual elements that implement the functions of six, seven and even eight variables. We should expect a further increase in the dimension of the problem solved in a single logical element. Existing methods for diagnosing FPGA logic are based on boundary scanning using external hardware according to the IEEE 1149.1 standard, and embedded test tools are actively being implemented according to the IEEE P1500 standard. As a rule, a comparison of the results of calculations in several logical elements is used, and if the results of calculations in one of them differ from the results, for example, the group of three elements, then it is concluded that there is a fault in this separate element. All this takes a relatively long time. For some applications, this approach is unacceptable. At the same time, methods and means of self-control, as well as accelerated testing, are not fully developed. The same applies to the features of diagnosing new logical elements of the so-called “Green Logic”. Well-proven genetic algorithms require refinement to reflect these new factors in order to improve the built-in testing of FPGAs. Article investigates methods and tools for diagnosing the FPGA logic for highly reliable applications. Authors shows that advanced logical elements can improves logic reliability via diagnosis. The task is to perfect the methods for diagnosing the logic of the FPGA using new logic elements, based on genetic algorithms.

In the article on the example of synchronous generator the description of dynamic mathematical model of electric machine is considered. This mathematical description is presented in the natural three-phase coordinate system a,b,c. The advantages and disadvantages of this description are shown. On the example of a synchronous generator, a General form of recording for an arbitrary element of the electrical system is obtained. This form of recording is obtained in vector-matrix form. In such a General form of recording it is advisable to present all the main elements of the electrical system (generators, electric motors, electric load elements, power lines, etc.) in order to be able to model them together within a single electrical system. It is shown that in this mathematical description of the inductance of the electric machine depends on the position of the rotor. Two variants of the account of the specified dependence are considered: by means of analytical expressions and by means of matrix transformations. Modern computer technology does not encounter any difficulties in mathematical modeling of electric machines directly in the natural three-phase coordinate system. The widespread use of the transformed two-phase coordinate system can be largely explained by tradition. Description of electrical machines in the axes a,b,c seems particularly appropriate in the study of asymmetric modes of operation in electrical systems and the asymmetric design of the electric machine. This article is the first article. Following it in the next article will be considered modeling of interaction of electric machines and other elements among themselves within the framework of a single electrical system. Mathematical models considered in the article can be used to study dynamic processes in electrical systems. Including asymmetric dynamic processes. On the basis of the mathematical description obtained in the article, it is planned to create an original software package for modeling various dynamic modes of three-phase electrical systems.

In modern power systems, block thermal power plants are involved in the regulation of frequency and power in normal and emergency modes. The implementation of power control tasks for power units in the mode of regulating the parameters of the power system requires the creation of special frequency and power control systems. Since the frequency and power of the power unit are interrelated, the task of maintaining the frequency at a given value is implemented by the power control system of the power unit. The article analyzes and evaluates the influence of external disturbing influences on the power control system of a once-through boiler of a block-type TPP. To date, the automatic power control system has been considered without taking into account the real nature of the disturbing influences on the load and the frequency control poses the problem of taking into account the stochastic properties of power system perturbations. In the normal operation mode, the stochastic properties of the equivalent disturbance can be determined from the results of a computational experiment using the power unit's automated process control system database. To obtain an equivalent perturbation, models were constructed for the steam pressure channel for the boiler unit and the power channel for the turbine unit, and the values of the expectation and standard deviation of the autocorrelation functions were calculated. A comparative analysis of the region of the most intense perturbing frequencies and resonant frequencies of the regulator BPR and TPC is performed. The proposed method allows to obtain a probabilistic model of the equivalent perturbation of the boiler unit and the turbine unit in normal operation. Equivalent perturbation models can be used to assess the technological effectiveness of the proposed solutions to improve the performance of the power unit control system.

The problems of state estimation of thermal power system operation and identification of mathematical model parameters have not been acceptable solved due to the complexity of objects of studies and their mathematical models, and the lack of effective methods, algorithms and computer programs to solve the required mathematical tasks. The results of solving the indicated problems are of importance in themselves and play a large part in the qualitative solution to thermal power units’ control problems, e.g., the tasks of optimal load distribution among the thermal power plants units and optimal control of thermal power units operating conditions. The current thermal power units, such as boiler units and steam turbines, are technical systems with rather complex engineering flow diagrams, diverse elemental composition and operation conditions. Hence, the main instruments to study thermal power equipment are the methods of mathematical modeling and optimization of its schemes and parameters. The paper describes a description an improved technique for identification (adjustment) of mathematical models for current main thermal power equipment. The technique allows us to reveal more effectively the gross errors in measurements of control parameters used to identify the mathematical model of the equipment under study, evaluate the correctness and correct the errors in constructing the mathematical model itself and improve the accuracy of identification. Among other things, the article discusses the issue of assessing the accuracy of identification the parameters of mathematical models of heat and power equipment, depending on the accuracy of measurements of control parameters used to set up the model, as well as on the correctness of the construction of the mathematical model itself and the calculation method used.

The article provides a survey about the implementation of neural networks on Programmable Logic Device (PLDs) such as FPGA (Field Programmable Gate Array) and Application-Specific Integrated Circuit (ASIC) from 2009 to 2019. Advantages and restrictions of this approach are given using on various hardware platforms. The reasons for the appropriateness to use FPGAs and ASICs based on them on such or another platform are indicated. In particular, FPGAs work best in low-power mobile systems, while ASICs, being a specialized solution, demonstrate the highest possible performance with, however, the development cost too high. We compare the performance of neural networks of various architectures (perceptrons, convolutional, binarized, recurrent, and also their modifications) based on FPGA with respect to other hardware solutions according to the criteria of processing speed and power consumption in relation to the price and ease of deployment. The high interest in FPGAs is investigated and confirmed as a result of their high energy efficiency and performance in solving a number of tasks. It is shown that GPUs are best suited for the implementation of convolutional neural networks, while FPGAs are suitable for recurrent ones. We note that with binarization of neural networks, the performance of gate arrays increases significantly, approaching the performance of specialized microcircuits. A promising direction for subsequent research is to further increase the performance of binarized neural networks implemented on the basis of FPGA, by improving both the mathematical apparatus underlying the network and the internal architecture of the gate array and its logic elements.