Please use this identifier to cite or link to this item:
Title: Розвиток методів та моделей енергоефективних компонентів комп’ютерно-інтегрованих систем ресурсозберігаючого технологічного обладнання спеціального призначення
Other Titles: Development of methods and models of energy-efficient components of computer-integrated systems of resource-saving technological equipment for special purposes
Authors: Лукашенко, Андрій Германович
Keywords: методи;модель;компоненти;перетворювачі;обчислювачі
Issue Date: 29-Dec-2020
Abstract: Робота присвячена підвищенню ефективності компонентів комп’ютерно-інтегрованих систем ресурсозберігаючого технологічного обладнання спеціального призначення, основу яких становлять обчислювальні перетворювачі, формувачі лінійно-ламаних функцій та обчислювачі передавальних функцій коригувальних ланок субсистеми. В дисертації на єдиному методологічному й інформаційному базисі вирішено науково-технічну проблему створення цифрових, комбінованих, гібридних обчислювальних перетворювачів і субсистеми, що забезпечують одночасно високу надійність, швидкодію, точність при малих габаритах і вазі, малому енергоспоживанні та низькій вартості за рахунок розробки нових і вдосконалення наявних методів та моделей компонентів і субсистеми, ефективність якихобґрунтована теоретично, підтверджена розрахунками та експериментами. Основні теоретичні та практичні результати дослідження впроваджені в ряді підприємств та державних університетах України.
The work is devoted to increasing the efficiency of computer-integrated systems of resource-saving technological equipment for special purposes through improving the theory aimed at overcoming contradictions and difficulties in the engineering of precision, highly reliable, high-speed with small dimensions, weight, cost and energy consumption components. In the thesis, on a single methodological and informational basis, the scientific and technical problem of creating highly efficient digital, combined, hybrid computing converters, generators of linear-broken functions and calculators of transfer functions of the correcting links of the subsystem is solved.The theoretical basis for creating a compositional logical-mathematical model (CLMM) on the basis of formal logic has been improved. A feature of CLMM is the formation of unipolar positionally ordered binary-code (OPVDK) operands as a composition of "0" and "1" regardless of the number systems and, in the absence of an analytical relationship between the digits. This expands the class of code converters and the scope of their application, increases the volume of production and, as a result, decreases the cost of the product. On the basis of KLMM, a bitwise tuple-table logical-revolving (RKTLO) method has been developed, the main procedures of which are: formation of correspondence tables based on KLMM; determination of the values of correcting constants using the properties of the XOR operation of the Zhegalkin algebra; elimination of information redundancy due to the introduction of tuple decomposition; the formation of the result of computational precision information is carried out on the basis of the rank distribution of the results of transformation of low-order tuples according to the rules of the concatenation operation. A distinctive feature of the RKTLO method consists in converting an array of input operands of various types to the corresponding array of output values of the code combination with correcting constants, using only logical operations; direct and inverse transformation of the values of the input OPVDK operands into the corresponding output for many functions is carried out using the same data of the correcting constants (pre-calculated), which reduces the number of tables by 2 times. At the same time, the following characteristics are provided: high reliability (through the hardware implementation in a single crystal of multifunctional precision models, the introduction of corrective and informational feedbacks inside the crystal, the number of active and passive elements: triggers, reducing the number of numeric memory blocks, generators of linear-broken functions and external outputs); high speed (through parallel processing of input digital information and only by logical operations and high-speed analog processes when generating information results in analog form); high accuracy (provided by multi-digit capacity, increasing the noise immunity zone and eliminating the influence of threshold voltages in analog transistor switches when transmitting input information); small dimensions and weight are provided by a decrease in structural complexity when using micro-, nanotechnology; low power consumption of precision models is ensured by using one register for input and output, decreasing by k times the numerical memory blocks and  times the generators of linear-broken functions (k and  the number of functions in multifunctional models); low cost is provided by a decrease in complexity, an increase in the same type of topology of model elements, an increase in the production of LSIs through the expansion of the application scope due to multifunctionality and a high percentage of the yield of suitable crystals from the plate. The combination of the KLMM and the RKTLO method provides the creation of a universal (with the possibility of increasing) homogeneous microarchitecture of a numerical memory block, in which a table of low-bit correction constants is stored.The versatility of a small-size numerical memory block reduces the cost, simplifies and speeds up the process of preparing constants due to bitwise processing of information, in which there is no influence of previous results on obtaining subsequent ones. At the same time, for multi-bit operands, it is not required to use high-performance precision computing tools when calculating constants and to create a special program for calculating the values of correcting constants, which significantly reduces energy, time and material costs. In the thesis, a multifunctional model of a combined calculator was developed, which converts analog, digital information and pulse duration into an output analog, using a unified model of a linear-broken function generator. This increases reliability and reduces energy and time costs; a hybrid multifunctional model of the calculator (GMMV) performs six functions: digital functional converter; digital-to-analog multiplier; digital-to-analog converter; analog-to-digital divider; analog inverted adder; operational amplifier. A special feature of the GMMV is the control cascade of two-position transistor switches, which provides high performance, the power consumption per cascade is not more than 0.1mW at T = 373K, and at T = 77K - 0.348 mW; input minimum current is 〖110〗^(-11 ) A, leakage current - at T = 77 K no more than 〖110〗^(-12 )A, and at T = 308K - no more than 〖110〗^(-9 )A. The paper proposes a generalized indicator for evaluating the effectiveness of the component models that are analyzed. A distinctive feature is the lack of subjectivity in the assessment procedure. This facilitates and accelerates the process of comparative analysis when determining from a variety of existing component models a model close to the optimal in many parameters for computer-integrated systems, which significantly reduces the cost of the product. In the work, based on conditional modeling and formed (based on the theory of dimensions, the method of zero power complexes and heuristics) multiparametric quality criteria corresponding to the physical interpretation, a method for determining the best of the set, serially producing integral components simultaneously in many parameters, and with full lack of a mathematical description of the connections between them, took further development. In this case, the key quality criterion is the criterion characterizing the energy reserve. On the basis of quality criteria, a criterion equation is created, sign models are built for the considered set of components in one, two, or four quadrants with the coordinates of multiparametric quality criteria. A feature is the visibility of establishing the relationship between many parameters at the same time when analyzing the constructed sign models. In addition, the paper proposes a highly economical method based on LSI self-modeling with a high energy reserve of the crystal, which provides a reduction in energy, time and material costs by almost 50%. A feature of the new LSI model is the change in the corresponding parameters (for example: either U, or f, or T) while maintaining the LSI topology. The high efficiency of the developed new and improved existing methods and models is substantiated theoretically, confirmed by calculations and experiments. The main theoretical and practical results of the research have been implemented in a number of enterprises and state universities of Ukraine (confirmed by acts of implementation).
Number of Pages: 383
Specialization: 05.13.05 – комп'ютерні системи та компоненти
Appears in Collections:05.13.05 – комп'ютерні системи та компоненти

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.