Academic literature on the topic 'Linearization of the mathematical model'

The purpose of this article is to create a mathematical model of a vehicle using dynamic equations of motion and simulation of perturbations acting on a vehicle. It is assumed that the tire in the car model behaves linearly. Because the vehicle model is nonlinear, the model will need to be linearized in order to find the transfer function between the angle of rotation of the front wheel and the lateral position of the vehicle. For this purpose, simple dynamic models of the car were created, which reflect its lateral and longitudinal dynamics. These types of models are usually used with a linearized form of mechanical and mathematical equations that are required when designing controllers, active suspension and other driver assistance systems.

A novel approach to build a Takagi-Sugeno (T-S) fuzzy model of an unknown nonlinear system from experimental data is presented in the paper. The neuro-fuzzy models or, more specifically, fuzzy basis function networks (FBFNs) are trained from input–output data to approximate the nonlinear systems for which analytical mathematical models are not available. Then, the T-S fuzzy models are derived from the direct linearization of the neuro-fuzzy models. The operating points for linearization are chosen using the evolutionary strategy to minimize the global approximation error so that the T-S fuzzy models can closely approximate the original unknown nonlinear system with a reduced number of linearizations. Based on T-S fuzzy models, optimal controllers are designed and implemented for a nonlinear two-link flexible joint robot, which demonstrates the possibility of implementing the well-established model-based optimal control method onto unknown nonlinear dynamic systems.

This work explores a model-free adaptive PID (MFA-PID) control for nonlinear discrete-time systems with rigorous mathematical analysis under a data-driven framework. An improved compact form dynamic linearization (iCFDL) is proposed to transfer the original nonlinear system into an affined linear data model including a nonlinear residual term. Both a time-difference estimator and a gradient parameter estimator are designed to estimate the nonlinear residual uncertainties and the unknown parameters in the iCFDL model. Subsequently, a novel improved CFDL based MFA-PID (iCFDL-MFA-PID) control is proposed by incorporating these two estimators. The results are extended by the use of improved partial format dynamic linearization (iPFDL) and full format dynamic linearization (iFFDL). The theoretical results are shown using contraction mapping principle-based mathematical analysis, as well as simulations.

Manufacturing process optimization is an ever-actual goal. Within this goal, machining parameters optimization is a very important task. Machining parameters strongly influence the manufacturing costs, process productivity and piece quality. Literature presents a series of optimization methods. The results supplied by these methods are comparable and it is difficult to establish which method is the best. For machining parameters optimization, this paper proposes a novel, simple and efficient method, without additional costs related to new software packages. This approach is based on linear mathematical programming. The optimization mathematical models are, however, nonlinear. Therefore, mathematical model linearization is required. The major and difficult problem is the linearization of the objective function. This represents the key element of the proposed optimization method. In this respect, the paper proposes an original mathematical procedure for calculating the part of the objective function that refers to the analytical integration of the tool life into the model. This calculus procedure was transposed into an original software tool. For demonstrating the validity of the method, a comparison is presented among the results obtained by certain optimization techniques. It results that the proposed method is simple and as good as those presented by the literature.

This paper proposes a control methodology based on feedback linearization for a doubly fed induction generator (DFIG) incorporating the magnetic saturation. The feedback linearization algebraically converts a nonlinear system model into a linear model, allowing the use of linear control techniques. Feedback linearization control depends on the model of the system and is therefore sensitive to parameter variations. The doubly fed induction generator (DFIG) operating under the magnetic saturation conditions results in the nonlinear variation of magnetizing inductance, which affects the performance of the control algorithm. From this stand point, on the basis of the dynamic model of the doubly fed induction generator considering magnetic saturation, the feedback linearizing control technique has been formulated. The mathematical model of the doubly fed induction generator, integrating the magnetic saturation has been formulated in the stator flux-oriented reference frame with rotor current and stator magnetizing current as state variables. Simulation studies demonstrate that the inclusion of magnetic saturation in the feedback linearization control of the doubly fed induction generator model increases its accuracy and results in a more efficient and reliable synthesis of the control algorithm.

The object of study in the article is the vortex effect of temperature separation in a rotating gas flow, which is realized in small-sized vortex energy separators. The subject matter is the models that describe the physical processes of energy conversion in small-sized vortex energy separators as objects of automatic control. The goal is to obtain models of a vortex energy separator reflecting its static and dynamic properties as an automatic control object. The tasks to be solved are: to develop a three-dimensional computer model of a small-sized vortex energy separator which will allow analyzing the parameters of the gas flow and physical processes of energy conversion directly inside the object and obtaining its static characteristics. A linearization method of static characteristics on the interval of input and output values is proposed which will expand the operating range without loss of linearization accuracy. A method of structural-parametric identification based on experimental logarithmic magnitude-frequency characteristics is proposed which will allow for the same set of experimental points to select the structure of the mathematical model of varying complexity depending on the specified accuracy. As a result of the work, the scheme for modeling the automatic control object was formed, consisting of the drive unit, sensor unit, and vortex energy separator, with the reflection of all the obtained operating modes. The methods used are the method of graphic linearization, Laplace transform, structural-parametric identification. The following results were obtained: a computer and linearized mathematical model of the small-sized vortex energy separator as an automatic control object reflecting its properties in the time and frequency domains was obtained. A comparative analysis of the reactions of the model and the real object to the same input action was carried out. Conclusions. The scientific novelty of the results obtained is as follows: 1) multiple graphic linearizations of one static characteristic to use the full range of the operation mode of vortex energy separator, which distinguishes it from the known;2) mathematical model structural-parametric identification for vortex energy separator with the help of known points of the Bode magnitude plots by using the interpolation polynomial and its derivatives graphs.

In this paper, according to the AC permanent magnet synchronous servo motor of the laboratory, using appropriate method to deal with servo motor makes its physical model be established. The nonlinear dynamic mathematical model of permanent magnet synchronous servo motor is established on the basis of the physical model. Based on nonlinear dynamic mathematical model of the permanent magnet synchronous servo motor, and through the coordinate transformation and state feedback, the input-output linearization is realized and the system decoupling is achieved. According to the system's linear model, a speed tracking controller is designed. The Simulink model of Svpwm is established. The control algorithm and the model of Svpwm are verified based on theMatlab7.6/Simulink & SimPowerSystems toolbox. The simulation results show that the controller designed has a very good control effect while the feedback linearization design is simple.

To control wheeled inverted pendulum is a good way to test all kinds of theories of control, On the basis of Newtonian mechanics and dynamic characteristics, the dynamical equation of DC motor, cart and pendulum is studied. Then the approximate linear model of wheeled inverted pendulum is concluded, which is based on the way of linearization of small Angle.

A linearization control research based on α-th order inverse system method has been developed for an axial hybrid magnetic bearing, which is a nonlinear system. The configuration of the axial hybrid magnetic bearing is briefly introduced, the working principle of the hybrid magnetic bearing is analyzed, and then the suction equations are set up. Based on expounding of α-th order inverse system method, and aiming at dynamics model of the axial hybrid magnetic bearing, the feasibility of linearization control is discussed in detail, the linearization control arithmetic based on α-th order inverse system method is deduced, and then close system controller is designed. Finally, the simulation system is set up with MATLAB software. The step response of system, the start up displacement curve of rotor and the performance of anti-disturbance of system are simulated. The simulation results have shown that the α-th order inverse system control strategy can realize accurate linearization for nonlinear mathematical model of the axial hybrid magnetic bearing, and the designed close control system has good dynamic and static performance.

Permanent magnet synchronous motors (PMSMs) have been broadly applied in servo-drive applications. It is necessary to improve the performance of PMSM. An explicit controller designed for PMSM based on multi-point linearization is proposed to reduce the linearized model error caused by different running status of PMSM. The mathematical model of PMSM system in the synchronous rotating frame and the problem formulation are introduced at first. Then, the preliminaries about explicit model predictive control (MPC) algorithm are presented in this article. Based on this, the multi-point linearization model is created for explicit MPC controller design. Moreover, the block diagram of the proposed method for PMSM system is presented. Finally, the simulation results are provided to demonstrate that the proposed explicit MPC controller based on multi-point linearization achieves better performance than that based on traditional single-point linearization, but requires the same online computation time because of the offline optimization of explicit MPC.

Robots have been a revolutionizing force in manufacturing in the 20th and 21st century but have proven too dangerous around humans to be used in many other fields including medicine. We describe a new control algorithm for robots developed by the Brigham Young University Robotics and Dynamics and Robotics Laboratory that has shown potential to make robots less dangerous to humans and suitable to work in more applications. We analyze the computational complexity of this algorithm and find that it could be a feasible control for even the most complicated robots. We also show conditions for a system which guarantee local stability for this control algorithm.

A feedback design procedure known as extended linearization consists in replacing a mathematical model of a nonlinear dynamical system with its family of linearizations, parametrized by the operating point, and then combining feedback gains designed for representatives of the family into a single nonlinear feedback law. The principles of the procedure, applicable both to lumped-parameter and distributed-parameter systems, are discussed at the outset. The development shows limits on feedback laws that can be designed, as well as nonuniqueness of solutions, inherent in the method.
Ph. D.

The paper is focused on the Predator-Prey model modified in the case of harvesting one or both populations. Firstly there is given a short description of the basic model and the sensitivity analysis. The first essential modification is percentage harvesting. This model could be easily converted to the basic one using a substitution. The next modification is constant harvesting. Solving this system requires linearization, which was properly done and brought valuable results applicable even for the basic or the percentage harvesting model. The next chapter describes regulation models, which could be used especially in applying environmental policies. All reasonable regulation models are shown after distinguishing between discrete and continuous harvesting. The last chapter contains an algorithm for maximizing the profit of a harvester using econometrical modelling tools.

The thesis describes various modifications of the predator-prey model. The modifications are considering several harvesting methods. At the beginning a solution and a sensitivity analysis of the basic model are provided. The first modification is the percentage harvesting model, which could be easily converted to the basic model. Secondly a constant harvesting including a linearization is derived. A significant part is devoted to regulation models with special a focus on environmental applications and the stability of the system. Optimization algorithms for one and both species harvesting are derived and back-tested. One species harvesting is based on econometrical tools; the core of two species harvesting is the modified Newton's method. The economic applications of the model in macroeconomics and oligopoly theory are expanded using the methods derived in the thesis.

Рассматривается задача линеаризации математических моделей, описывающих технологические процессы, с целью получения удобного инструмента для управления ими. Задача линеаризации решается с помощью геометрической теории управления (ГТУ). Привлекательность ГТУ связана с получением эквивалентных нелинейным моделям линейных моделей, которые удобно использовать для решения задач управления, получая структуры регуляторов или законы управления. После чего осуществляется обратный переход из пространства линейных систем в пространство исходной нелинейной системы. При этом основные аналитические преобразования автоматизированы с помощью специализированного программного обеспечения. Поиск функций преобразования, связывающих переменные линейной и нелинейной моделей, осуществляется с помощью нового конструктивного метода решения системы дифференциальных уравнений в частных производных.
The problem of linearization of mathematical models describing technological processes with the purpose of obtaining a convenient tool for managing them is considered. The problem of linearization is solved by means of a geometric control theory (GCT). The attractiveness of GCT is connected, first of all, with obtaining equivalent nonlinear linear models, which are convenient for solving management problems and receiving regulatory structures or control laws. After that performed the reverse transition from the space of linear systems to the space of the original nonlinear system. A wider application of the geometric control theory is hindered by cumbersome analytical transformations connected with the calculation of the derivatives and the Lie brackets, the definition of the involutivity of distributions, and so on, and also the problem of determining the transformation functions connecting the variables of linear models in the form of Brunovsky and initial non-linear models of control objects. The authors developed specialized software that automates the main analytical transformations of GCT. The search for the transformation functions connecting the variables of the linear and nonlinear models is carried out using a new constructive method for solving the system of partial differential equations.

Feedforward is known to be one of the best methods for power amplifier linearization due to its superior linearization performance and broadband stable operation. However feedforward systems have relatively poor power efficiency and are complicated due to the presence of two nonlinear amplifiers and the requirements of amplitude, phase and delay matching within two different loops. In this thesis stochastic characterization of a simple feedforward system with autocorrelation analysis has been presented for Code Division Multiple Access (CDMA) applications taking the amplitude and delay mismatches into consideration. It has been assumed that, the input signal can be represented as Gaussian noise, main and error amplifiers can be modeled with third order AM/AM nonlinearities and there exists no phase mismatch within the loops. Hence closed form expressions, which relate the main channel and distorted adjacent channel power at any point in the feedforward circuitry to the system parameters, have been obtained. Consequently, a mathematical handy tool is achieved towards specifying the circuit parameters rapidly for optimum linearity performance and efficiency. The developed analytical model has been verified by Radio Frequency (RF) and system simulations. An alternative approach towards modeling feedforward systems for arbitrary signals has also been brought into consideration and has been verified with system simulations.

Дисертація на здобуття наукового ступеня доктора філософії (PhD) за спеціальністю 123 – Комп’ютерна інженерія – Національний технічний університет “Харківський політехнічний інститут”, Харків, 2020. Об’єктом дослідження є процеси управління тяговим рухомим складом за допомогою бортової комп’ютерної системи, що використовується у дизель-потягах серії ДЕЛ-02. Предметом дослідження є моделі, методи та відповідні програмні компоненти, які використовуються в комп’ютерній системі тягового рухомого складу, та розширюють область застосування геометричної теорії управління при синтезі оптимальних керувань рухомим складом, а також методи і засоби розробки сучасних програмних комплексів в рамках розробки комп’ютерної системи підтримки прийняття рішень машиніста дизель-потяга серії ДЕЛ-02. У вступі акцентовано увагу та обгрунтовано актуальність теми, що досліджується, показано зв’язок роботи з науковими програмами, планами та темами, наведено наукову новизну, а також, сформульовано практичне значення отриманих результатів. В першому розділі здійснено аналітичний огляд моделей, методів та програмних компонентах, що використовуються в комп’ютерних системах управління тяговим рухомим складом. Розглянуто особливості структури та роботи подібних систем на залізничному транспорті в Україні та світі (Китай, Індія, Німеччина, країни СНД). На прикладі роботи таких систем було розглянуто їх структуру, технічні характеристики, області застосування та особливості використання. В рамках першого розділу, також, було розглянуто математичну модель об’єкта управління, приклад методу лінеаризації даної математичної моделі, метод пошуку функцій перетворення, що пов’язують змінні лінійної та нелінійної математичної моделей. Також, було розглянуто можливості використання нейромережевої асоціативної пам’яті в системах управління та проаналізовано методи синтезу оптимальних систем управління. В результаті, було обрано основні напрямки досліджень та поставлено основні задачі дисертаційної роботи. В другому розділі було розглянуто питання перетворення нелінійних математичних моделей в еквівалентні лінійні математичні моделі в формі Бруновського. Також, було розглянуто методи спрощення аналітичних перетворень під час виконання процесу лінеаризації за рахунок перетворення до лінійного виду нелінійних систем з різною кількістю одночленів в правих частинах диференційних рівнянь початкового об’єкту, а також, відокремлення лінійного рівняння від системи в цілому. Дані методи було перевірено шляхом моделювання руху по відрізку шляху початкового об’єкту у вигляді нелінійної системи диференційних рівняннь та об’єкту перетвореного у лінійну форму Бруновського, з подальшим порівнянням отриманих результатів, які показали співпадіння, що свідчить про те, що у разі використання даного методу лінеаризації отримується лінійна математична модель, що є повністю еквівалентною початковій недінійній моделі. Додатково, було виконано лінеаризацію більш складної нелінійної математичної моделі, що описує роботу потяга з двома окремими двигунами, перевірка результатів моделювання лінійної моделі показала повну еквівалентність її початковій формі. Результати досліджень дозволили отримати ряд наукових результатів: − визначено залежність кількості та складності розрахунків під час проведення лінеаризації та пошуку функцій перетворень від кількості одночленів в правій частині рівнянь нелінійної математичної моделі; − запропоновано два нових методи пошуку функцій перетворення, що пов’язують змінні лінійної та нелінійної моделей, що дозволяють розширити область застосування ГТУ на об’єкти, праві частини диференційних рівнянь яких містять більше двох одночленів; − запропоновано метод зниження кількості обчислень при виконанні лінеаризації за рахунок відокремлення лінійного рівняння від системи; − виконано перевірку даного методу, який показа свою роботоспроможність на більш складних математичних моделях, зокрема на моделі, що описує роботу потяга з використанням двох еквівалентних двигунів. В третьому розділі розглянуто питання створення нового методу для пошуку функцій перетворення з використанням нейронних мереж. В рамках даного розділу запропоновано нову нейронну мережу, яка може бути використана для пошуку функцій перетворення. Наряду з цим в даному розділі було запропоновано новий табличний метод пошуку функцій перетворення, який є простим та наочним, там може використовуватися для швидкого отримання результатів при виконанні розрахунків. Дослідження, проведені в даному розділі дозволили отримати наступні наукові результати: − створено та запропоновано нову нейронну мережу для пошуку функцій перетворення, що пов’язують змінні нелінійної та лінійної моделей об’єкта управління, а це, в свою чергу, розширює область застосування геометричної теорії управління; − запропоновано новий табличний метод для пошуку функцій перетворення, який є досить простим для сприйняття та достатньо наочним. В рамках цього, запропоновано систему рівнянь в частинних похідних з обмеженнями у вигляді диференціальних нерівностей представляти у вигляді відповідної таблиці, яка дає змогу в наочному вигляді отримувати залежність функцій перетворень від аргументів, також формувати системи лінійних однорідних рівнянь, за допомогою яких можна буде звужувати область пошуку функцій перетворення. В четвертому розділі присвячено увагу програмним компонентом бортової комп’ютерної системи, а також розробленому програмному забезпеченню, що дозволяє розширити область застосування геометричної теорії управління. А саме, було розглянуто нові функціональні можливості розробленого програмного забезпечення, та описано його основні характеристики та структуру. В рамках опису розробленого програмного забезпечення особливу увагу приділено структурі та опису роботи окремих функціональних блоків програми, розробці структури інтерфейсу, надійності програмного забезпечення, компонентів для вирішення завдань управління за допомогою геометричної теорії управління, оцінці якості програмного забезпечення. Також, в даному розділі приведено приклад роботи розробленого програмного забезпечення. Крім того, в даному розділі приведено результати рішення завдання оптимального руху дизель-потягу по маршруту його прямування, в рамках чого було виконано моделювання руху потяга по маршруту та та порівняння отриманих даних з даними руху реального потяга, а також виконано спробу підвищити ефективність руху потяга за рахунок оптимізації окремих множин перегонів з урахуванням особливостей маршруту прямування. В рамках даного розділу були отримані наступні наукові результати: − розроблено нове програмне забезпечення, яке отримало подальший розвиток завдяки використанню можливостей сучасних мов програмування. Розроблене програмне забезпечення є більш стабільним завдяки блоку тестування, більш зручним завдяки створеному графічному інтерфейсу користувача, більш функціональним, адже воно може виконувати процес лінеаризації та пошуку функцій перетворення, але при цьому багато функціональних можливостей є автоматизованими, в вихідних даних наявні коментарі та пояснення, що збільшує рівень зручності користування даним програмним забезпеченням, крім того, характеристики програми відповідають вимогам стандарту з якості програмного забезпечення; − було виконано дослідження залежності кількості спожитого палива під час руху потяга від особливостей рельєфу місцевості, стилю ведення потяга та розкладу його руху. − було запропоновано та протестовано метод зниження кількості спожитого палива, використовуючи особливості рельєфу місцевості, допустимі відставання чи випередження графіку руху потяга, а також визначення оптимального стилю руху як для маршруту в цілому, так і для його окремих частин; − було виконано моделювання руху потяга по реальному маршруту, а результати порівняні з реальним потягом, що курсує цим маршрутом, результати показали правильність моделювання. Отже, дисертаційна робота присвячена розвязанню науково-прикладної задачі, а саме, розробки моделей, методів та програмних компонентів компютерної системи тягового рухомого складу, яка створена на основі узагальнених математичних моделей, розробленого програмного забезпечення, а також засобів оптимізації управління рухомими об’єктами з використанням нових методів, а також використання нової стуктури нейронних мереж для пошуку функцій перетворення, що дозволило розширити область застосування геометричної теорії управління, що створює передумови для розробки автоматичних систем управління потягом та дозволяє поліпшити характеристики, повязані з об’ємами споживання енергоресурсів. Вдосконалена модель дизель-потяга враховує основні види взаємодії потяга та профілю шляху, а саме, повороти, нахили, а також роботу двигунів потяга, що адекватно відображає протікаючі в реальному дизель-потязі процеси. Було створено спеціалізоване програмне забезпечення, що має графічний інтерфейс користувача, а також відповідає вимогам оцінки якості програмного забезпечення. Дане програмне забезпечення реалізує вдосконалену структуру людино-машинної системи, дає можливість виконати автоматизацію аналітичних перетворень геометричної теорії управління у формі Бруновського. Нова структура нейронних мереж, базується на нейронних мережах типу АРТ, що дозволяє вирішувати завдання, що мають декілька рішень. Це дозволило виконати розробку нового методу пошуку функцій перетворення, які зв’язують змінні нелінійних та лінійних моделей у формі Бруновського. Для збільшення ефективності процесу лінеаризації було запропоновано декілька методів спрощення процесу розрахунків за рахунок зменшення кількості елементів в правій частині початкової системи диференційних рівнянь, та за рахунок відокремлення першого рівняння, яке саме по собі вже є лінійним, від загальної системи в цілому. Виконані дослідження та розробки дозволили вдосконалити структуру бортової компютерної системи підтримки прийняття рішень машиніста дизель потяга, що дозволило, в реальних умовах руху динамічного об’єкту, під час змін дорожніх умов, виконувати перерахунки та видавати машиністу нові закони керування, які дозволять продовжити рух по маршруту з дотриманням графіку та мінімальними витратами паливо-енергетичних ресурсів. Проведені дослідження на реальному обєкті та математичних моделях. Результати досліджень підтвердили правильність використовуваних інструментів, методів та алгоритмів, на основі яких були запропоновані відповідні рішення, які лягли в основу розробленого програмного забезпечення.
The thesis is submitted to obtain a scientific degree of Doctor of Philosophy, specialty 123 – Computer Engineering – National Technical University “Kharkiv Polytechnic Institute” , Kharkiv, 2020. The object of the research is the processes of managing the traction rolling stock with the help of an on-board computer system used in the DEL-02 series diesel trains. The subject of research are models, methods and corresponding software components used in the computer system of traction rolling stock, which extend the using scope of geometric control theory for the synthesis of optimal controls of rolling stock, as well as methods and tools for the development of modern software complexes in the development of computer decision support systems of the diesel train driver of the DEL-02 series trains. The introduction focused and explained on the relevance of the topic being researched, shows the relationship with scientific programs, plans and topics, presents the scientific novelty, as well as formulates the practical significance of the results. The first section provides an analytical overview of models, methods and software components used in computerized decision support systems of the diesel train driver and train control systems. The peculiarities of the structure and peculiarities of using such systems on rail transport in Ukraine and in the world (China, India, Germany, CIS countries) are considered. On the example of the operation of such systems considered their structure, specifications, applications and features of use. The first section also deals with the mathematical model of a control object, an example of a method of linearization of a given mathematical model, a method of finding transform functions that relate variables of linear and nonlinear mathematical models. Also, the possibility of using neural network associative memory in control systems was considered and methods of synthesis of optimal control systems were analyzed. As a result, the main directions of research were selected and the main tasks of the dissertation were set. In the second section, the question of converting nonlinear mathematical models into equivalent linear mathematical models in the form of Brunovsky was considered. Also, methods of simplifying analytical transformations during the linearization process by converting to a linear kind of nonlinear systems with different numbers of monomials in the right-hand sides of the differential equations of the initial object, as well as separating the linear equation from the other part of the system of equations, were considered. These methods were verified by modeling the motion along the path of the initial object in the form of a nonlinear system of differential equations and the object transformed into a linear Brunovsky form, with further comparison of the results obtained, which showed coincidence, which indicates that in the case of using this the linearization method allows to obtain a linear mathematical model that is completely equivalent to the original non-linear model. Additionally, linearization of a more complex nonlinear mathematical model describing the operation of a train with two separate engines was performed, and the verification of the results of the linear model simulation showed complete equivalence to its original form. Research results have yielded a number of scientific results: − dependence of quantity and complexity of calculations during linearization and search of transformation functions on the number of monomials in the right part of equations of nonlinear mathematical model is determined; − two new methods of finding transform functions are proposed that relate variables of linear and nonlinear models that extend the scope of geometric control theory to objects whose right-hand sides of differential equations contain more than two monomials; − was proposed a method of reducing the number of calculations when performing linearization by separating a linear equation from the system; − this method was tested, which showed its workability on more complex mathematical models, in particular, on a model that describing the operation of a train using two equivalent motors. In the third section of the paper, the question of creating a new method for finding functions of transformation using neural networks was considered. In this section proposes a new neural network that can be used to search for conversion functions. In addition, this section proposes a new tabular method of finding conversion functions, which is simple and clear and can be used to get results when performing the calculation process. The studies conducted in this section have yielded the following scientific results: − a new neural network has been created and proposed for searching the conversion functions that relate variables to nonlinear and linear models of a control object, which in turn widens the scope of geometric control theory; − a new tabular method for finding conversion functions is proposed, which is simple enough to understand and sufficiently visual. In this context, it is proposed to present a system of partial differential equations with constraints in the form of differential inequalities in the form of a corresponding table, which allows to visualize the dependence of transformation functions on arguments, as well as to form systems of linear homogeneous equations by which it is possible to narrow the search area of conversion functions. The fourth section focuses on the software components of the on-board computer system, as well as the developed software that extends the scope of geometric control theory. Specifically, shows with new functionality of designed software and describes its main characteristics and structure. In the framework of the description of the developed software, special attention is paid to the structure and description of the operation of individual functional blocks of the program, the development of the interface structure, the reliability of the software, components for solving control problems using geometric control theory, evaluation of the quality of the software. Also, this section gives an example of how the developed software works. In addition, this section presents the results of solving the problem of optimal motion of the diesel train along the route of its direction, in which the simulation of the train movement along the route was performed and the comparison of the obtained data with the data of the movement of the real train, as well as an attempt to improve the efficiency of train movement due to the optimization of individual sets of routes, taking into account the features of the route. The following scientific results have been obtained within this section: − new software has been developed that has been further developed through the use of modern programming languages. The developed software is more stable due to the testing unit, more convenient due to the created graphical user interface, more functional, because it can perform the process of linearization and search of conversion functions, many of the functionality are automated, there are comments and an explanation that increases the ease of use of this software, in addition, the characteristics of the program meet the requirements of the standard of program quality; − the study of the dependence of the amount of fuel consumed during train movement on the features of terrain, the style of running the train and its schedule; − a method of reducing the amount of fuel consumed was proposed and tested, using terrain features, permissible lag or advance of the train timetable, as well as determining the optimal driving style for the route as a whole and for its individual parts; − the train simulation was performed on a real route, and the results showed that the simulation was correct, because it was compared to the real train running on this route. Therefore, the dissertation is devoted to the solution of the scientific-applied problem, namely, the development of models, methods and software components of the computer system of traction rolling stock, which is created on the basis of generalized mathematical models, developed software, as well as the means of optimizing the control of moving objects new methods, as well as the use of a new neural network structure to search for transformation functions, which made it possible to extend the scope of geometric control theory it breeds the preconditions for developing automatic train control systems and improves performance related to energy consumption. The advanced diesel train model takes into account the main types of interaction between the train and the track profile, namely, turns, slopes, as well as the performance of the train engines, which adequately reflects the processes in real diesel train. Specialized software has been created that has a graphical user interface and complies with software quality assessment requirements. This software implements an advanced structure of the human-machine system, makes it possible to perform automation of analytical transformations of geometric control theory to the form of Brunovsky. The new neural network structure is based on ART-type neural networks to solve multiple-choice tasks. This made it possible to develop a new method of finding transform functions that relate variables of nonlinear and linear models in the form of Brunovsky. To increase the efficiency of the linearization process, several methods have been proposed to simplify the calculation process by reducing the number of elements in the right-hand side of the initial differential equation system, and by separating the first equation, which itself is linear, from the general system of equations. The performed research and development allowed to improve the structure of the on-board computer system of decision support of the driver of the diesel train, which allowed, under real conditions of movement of the dynamic object, during changes of road conditions, to perform recalculations and to give the driver new control laws which will allow to continue the movement on the route adhering to the timetable and minimum cost of fuel and energy resources. Appropriate researches were conducted on real object and mathematical models. The results of the researches confirmed the correctness of the used tools, methods and algorithms, on the basis of which the appropriate solutions that formed the basis of the developed software were proposed.

A recent development in the design of control system for a jet engine is to use a suitable, fast and accurate model running on board. Development of linear models is particularly important as most engine control designs are based on linear control theory. Engine control performance can be significantly improved by increasing the accuracy of the developed model. Current state-of-the-art is to use piecewise linear models at selected equilibrium conditions for the development of set point controllers, followed by scheduling of resulting controller gains as a function of one or more of the system states. However, arriving at an effective gain scheduler that can accommodate fast transients covering a wide range of operating points can become quite complex and involved, thus resulting in a sacrifice on controller performance for its simplicity. This thesis presents a methodology for developing a control oriented analytical linear model of a jet engine at both equilibrium and off-equilibrium conditions. This scheme requires a nonlinear engine model to run onboard in real time. The off-equilibrium analytical linear model provides improved accuracy and flexibility over the commonly used piecewise linear models developed using numerical perturbations. Linear coefficients are obtained by evaluating, at current conditions, analytical expressions which result from differentiation of simplified nonlinear expressions. Residualization of the fast dynamics states are utilized since the fast dynamics are typically outside of the primary control bandwidth. Analytical expressions based on the physics of the aerothermodynamic processes of a gas turbine engine facilitate a systematic approach to the analysis and synthesis of model based controllers. In addition, the use of analytical expressions reduces the computational effort, enabling linearization in real time at both equilibrium and off-equilibrium conditions for a more accurate capture of system dynamics during aggressive transient maneuvers. The methodology is formulated and applied to a separate flow twin-spool turbofan engine model in the Numerical Propulsion System Simulation (NPSS) platform. The fidelity of linear model is examined by validating against a detailed nonlinear engine model using time domain response, the normalized additive uncertainty and the nu-gap metric. The effects of each simplifying assumptions, which are crucial to the linear model development, on the fidelity of the linear model are analyzed in detail. A case study is performed to investigate the case when the current state (including both slow and fast states) of the system is not readily available from the nonlinear simulation model. Also, a simple model based control is used to illustrate benefits of using the proposed modeling approach.

Industrial hydraulic systems are complex, and show nonlinear dynamic behavior because of their nature. When it is not easy to deal with the nonlinear models, hydraulic systems are usually described by linear or linearized models around operating points. In this work a nonlinear dynamic and mathematic model for the position control of a double rod hydraulic actuator was developed. Three control strategies were implemented: PID control, optimal control (LQR) and control by Feedback Linearization. For the PID control and optimal control (LQR) strategies a linearized model of the hydraulic actuator was developed around a specific operating point, contrary to the Feedback Linearization control that have a wide operation range and the nonlinear model was used. These mathematical models were represented on Simulink environment, in order to compare and analyze the response and dynamic behavior. The optimal control (LQR) shows better settling time than the PID control, both without overshoot; and the Feedback Linearization show the best dynamic performance in terms of settling time with a little overshoot and disturbance tolerance.

This paper aims to combine neural network modelling with model-based fault detection. An accurate and robust model is critical in model-based fault detection. However, the development of such a model is the most difficult task especially when a non-linear system is involved. The problem comes not only from the lack of concerned information about model parameters, but also from the inevitable linearization. In order to solve this problem, neural networks are introduced in this paper. Instead of using conventional neural network modelling, the neural network is only used to approximate the non-linear part of the system, leaving the linear part to be represented by a mathematical model. This new scheme of integration between neural network and mathematical model (NNMM) allows the compensation of the error from conventional modelling methods. Simultaneously, it keeps the residual signatures physically interpretable.

Abstract This paper investigates robust control problem of structural vibrations using shape memory alloy (SMA) wires as actuators. The mathematical model for these SMA actuators is derived with emphasis in model uncertainty. The linearization of the relation between stress and temperature dynamics of SMA actuators is analyzed for active control. To handle the uncertainties caused by the linearization and the neglected high frequency dynamics, optimal H∞ control was employed to design a controller. An example is used to demonstrate the design procedures and the control system is tested in a nonlinear environment.

Abstract In this paper we develop a mathematical model for a nonlinear analog of the Euler-Bernoulli beam which possesses linear damping and nonlinear stiffness properties. This nonlinear model is used as a basis for approximation and linearization. Feedback design for the linear problem is applied to compute gains which are then used in the nonlinear system.

The challenges of providing safe and high performance marine vehicles present strict and often conflicting constraints that require rational and holistic analysis methodologies to obtain efficient design solutions. This paper presents a mathematical framework for stability analysis, which is one of the key elements in the design and operation of ships and floating bodies that still require considerable improvement. The method is based on the application of the Lyapunov stability analysis concept, which has been highly successful in some other engineering and scientific disciplines. The paper presents the fundamental concepts on the applicability of the Lyapunov method to ship motions stability analysis. Governing mathematical models are derived from first principles and interpreted in the context of geometrical and physical interrelationships. The analytical models are primarily developed for the generalized case of non-linear forced non-conservative systems and simplified by linearization in the case of coupled motion for detailed analysis and characterization of stability conditions and domain. The concept of “motion boundedness” is introduced to satisfy requirements of the Lyapunov method to ship motions subjected to continuous excitations. The analysis leads to some valuable deductions and insight that would be useful in the formulation of stability criteria for ships and marine vehicles in general. The most significant contribution is the possibility of explicit determination of geometric and hydrostatics/hydrodynamics parameters that govern ship stability characteristics.

Abstract The mathematical modeling of dynamic systems is an important task in the design, analysis, and implementation of advanced automotive control systems. Although most vehicle control algorithms tend to use model-free calibration architectures, a need exists to migrate to model-based control algorithms which offer greater operating performance. However, in many instances, the analytical descriptions are too complex for real-time powertrain and chassis model-based control algorithms. Therefore, model reduction strategies may be applied to transform the original model into a simplified lower-order form while preserving the dynamic characteristics of the original high-order system. In this paper, an empirical gramian balanced nonlinear model reduction strategy is examined for the simplification process of dynamic system descriptions. The empirical gramians may be computed using either experimental or simulation data. These gramians are then balanced and unimportant system dynamics truncated. For comparison purposes, a Taylor Series linearization will also be introduced to linearize the original nonlinear system about an equilibrium operating point and then a balanced realization linear reduction strategy will be applied. To demonstrate the functionality of each model reduction strategy, two nonlinear dynamic system models are investigated and respective transient performances compared.