Home – Home – Jornals – Avtometriya 2021 number 4

In this paper we address the problem of velocity vector control of aircraft in the flight path axis system. The development of an automatic flight path control system requires reaching a desired flight speed with coordinated control of rudders and engine thrust. For that purpose the nonlinear controller based on method of structural synthesis is built. Simulation results of the complete closed-loop system demonstrate the effectiveness of the proposed approach.

In this paper the developed and implemented software and hardware complex for flight experiments is presented. The complex ensures operation in the hardware-in-the-loop simulation mode, aircraft crew training mode, and ground control station mode. The purpose and functional features of each of the operating modes of the complex are highlighted. The results of research performed by using the developed complex by an example of a remotely controlled aircraft - a flying laboratory - are presented.

A constructive method for synthesizing the optimal energy-efficient control for distributed parameter systems with a given-precision uniform approximation of the space distribution of the controlled variable with respect to the desired state is proposed. This approach is based on the earlier developed alternance method for constructing optimal open-loop control algorithms. It is shown that the optimal controller equations are reduced to linear output feedback equations with time-dependent coefficients that are calculated based on incomplete measurements of the state of the system. The transmission factors here are known functions of time and final values of the controlled parameter determined by preliminary calculations of the programmed control.

In this paper the procedure of the controller synthesis for a linearized system “double inverted pendulum on a cart” based on the sliding modes technique is described. The main peculiarity of this method is the two-stage controller calculation. At the first stage, the canonical basis transformation of the model object is performed, and the control algorithm is organized in the form of a function of canonical variables. At the second stage, the controller definition related to the initial basis is obtained via the inverse transformation of variables. As the direct measurement of several internal variables for the type of the object under research is possible, it is needed to estimate the remaining object variables for the controller implementation. The use of several state estimation filters of the same kind is proposed for this purpose, which allows definition of the entire state vector, although only some of state variables can be measured directly. The calculated controller ensures pendulum stabilization in the upper position with the low value of angle deviation from the vertical. The results of the numerical simulation of the system with the help of the MatLab Simulink integrated environment illustrate the main properties of the system.

A modified structure of a proportional-integral resonant (PIR) controller and a method of adjusting its parameters are considered. It is shown that application of the time-scale separation method allows selection of the proportional-integral component of the controller independently from selection of the parameters of the resonant components of the controller. The proposed approach to calculation of the controller parameters allows providing robust perfect tracking and disturbance rejection in the presence of plant model uncertainties and harmonic disturbances. The results of numerical modeling are also reported.

In the well-known publications devoted to the method of potential fields in mobile robotics, the key importance is attached to the mechanism of force influence on the movement of the robot, which determines the law of change in the speed of its movement. In this article, a new methodology for using virtual fields is proposed: the force aspect is excluded, and the lines of intensity of the generated field define only the desired trajectories of the robot movement, and following a given trajectory is ensured by a two-contour control system for the direction and speed of the robot movement. The question of constructing attracting and repulsive potentials is discussed. The results of the simulation of the trajectory control of a two-wheeled robot are presented.

The paper proposes a way to apply the shuffled frog leaping algorithm as a tool to expand a rule base of the fuzzy classifier. Its application is relevant in the case where the existing rules are not sufficient for the qualitative recognition of all classes, for example, in the presence of data imbalance. Additional rules generated by metaheuristics are able not only to improve the classification quality, but also to provide a more complete description of the domain under study. To create a compact initial structure of the classifier, the algorithm based on extreme values of features in classes is used. The combination under consideration is tested on 36 imbalanced datasets from the Knowledge Extraction based on Evolutionary Learning repository and showed an increase in the mean geometric accuracy on 34 datasets, as well as satisfactory results compared to analogs. The advantages of the proposed algorithm of structure formation are the absence of the necessity of data augmenting with synthetic samples, low scatter of results in individual runs, and the ability to improve the classification quality by adding a few rules.

In order to reduce the complexity of combinational schemes, a new methodology for synthesizing control automata is proposed, based on entering a limited number of empty operators into the control graph scheme and three ways to organize the address subsystem: by entering a state counter and unitary encoding, entering a multiplexer into the control automaton structure to select the address of one logical condition from the entire set {α}, and entering an address block consisting of q two-input elements "And," one "OR," and an RS trigger to select one α_j∈{α}. In all versions of the control automata implementation, the number of inputs to the programmable logic matrix decreases by 1.5-3 times, depending on the control algorithm complexity.

Using the hyperstability criterion, fast-acting filters-correctors, and implicit reference models under the conditions of control signal constraints and a priory uncertainty, the synthesis of combined nonlinear control algorithms for a system of inverted pendulums connected by a spring is discussed. With the help of computational experiments, the quality of the obtained control system is studied.

An algorithm for controlling the flight of an unmanned aerial vehicle (UAV) over a railway in the automatic mode without participation of an operator and the use of satellite navigation systems is proposed. The algorithm is based on a computer analysis of the video image from the UAV camera, identifying the rails, determining their direction relative to the UAV movement, and controlling the flight characteristics (pitch, yaw, and roll) so that the flight is carried out above the railway at a given height. An advantageous feature of the algorithm is its simplicity, which does not require significant computing resources. The implementation of the algorithm affects neither the weight and size parameters and power consumption of the UAV nor the range and duration of its flight. Flight tests have confirmed the algorithm reliability.

The increased usage of computer vision for biometric identification systems and as a security measure leads to an increase in attempts of falsification. In this regard, the number of ways to automatically detect such situations is also increasing. However, the accuracy of the detection of spoofing attacks can be reduced in some cases like in other systems that use computer vision. In the present study, the existing approaches to the detection of spoofing attacks are considered, and an assessment of their resistance to changes in the conditions of data recording is given.

The use of convolutional neural networks for analyzing graphic images for the presence of data introduced by steganography methods is investigated. It is shown that a deep convolutional neural network is trained to classify the presence of hidden data in graphic images, achieving accuracy according to the weighted AUC metric of 0.928. The hypothesis of the effectiveness of applying the concept of "transfer learning" for the sphere of steganography is tested. The effectiveness of the proposed technology is confirmed by a large number of experiments.

Dynamic test benches with starry sky simulators are designed for laboratory tests of the star tracker working cycle. Raster screens (computer or television) are often used now to obtain the image of the starry sky. However, star images on such screens have significant differences from the real sky, which must be taken into account when carrying out dynamic tests of star trackers and processing their results.

The synthesis of a digital low-frequency quasi-Gaussian filter with a finite pulse-transient response, a unity gain at zero frequencies, and non-negative weighting factors is considered. The filter is designed to reduce noise in Poisson observations. A Gaussian function is used for the prototype filter, and an approximation functional is formed. For the synthesis, the stochastic global minimization of this functional with a modified algorithm for simulating "annealing" is used. The quasi-Gaussian filter is investigated using model and experimental Poisson observations, and its efficiency is confirmed.