Аннотация:

The paper is concerned with the formation methods on algorithmic support for navigation systems with variable structure based on the concept of system synthesis. During the functioning of aircrafts, it is highly needed to change the instrumental structures of navigation systems, and apply mathematical models constructed in flight to formulate correction algorithms for getting the most effective state variables on each interval of system operation. The identification of state variables or parameters in the models is undertook by utilizing the ensemble of selection criteria during the flight. Thus, the dynamic models constructed with the concept of system synthesis is dependent on the aircraft flight modes. Therefore, the mathematical models used in the information processing algorithms is with the highest possible degree of observability and identifiability of state variables.

Ключевые слова:

high-precision navigation complex, information processing algorithms, dynamic system synthesis, mathematical models, ensemble of selection criteria, observability degree criterion, controllability degree criterion, identifiability degree criterion.

Основной текст труда

Aircraft (AC) control is carried out on the basis of information from various navigation systems and orientation systems. Modern navigation systems are subject to high requirements for accuracy. To fulfill the established accuracy characteristics, the measuring systems are combined into navigation systems (NS). In turn, the increase in the accuracy of NS is carried out algorithmically [1, 2]. Information processing algorithms include algorithms for complexing, estimating, controlling, and building models [3]. The basic algorithm that combines measuring systems is the integration algorithm [4]. The main function of the integration algorithm is the joint processing of signals from navigation systems and the selection of navigation information. The aggregation method involves the use of an ensemble of information selection criteria. Algorithmic support of high-precision navidation complexes (NC) includes various intelligent components using elements of theories of intelligent systems [5,6,7]. The use of intelligent components [8, 9] in NC assumes the presence of a high performance computer on the aircraft.

The use of the concept of dynamic system synthesis in the operation of algorithmic support for NC makes it possible to form the best structure of algorithmic support at each flight interval, adequate models with improved characteristics, and determine the best architecture of NC during the flight of an aircraft.

The development of highly efficient software and algorithmic support for NC requires the use of new information technologies and approaches, for example, the concept of system synthesis [10,11].

The environment in which the NC aircraft operates is theoretically described by a large number of parameters. Some of these parameters are defining (key, dominant) and it is these parameters that are used in the NC algorithmic support. There are projections onto a subspace of a smaller number of variables that reflect the situation in the original space of variables with a sufficient degree of adequacy. Channels are formed from these most informative state variables. With respect to these channels, algorithms for NC aircraft are being developed. With the help of an ensemble of selection criteria, the dimension of the channel is determined, i.e. state variables are identified that fairly well reflect the process under study. In practical applications, the dimension of the channel, as a rule, is small. At the NC design stage, using an ensemble of selection criteria and a priori information about the processes under study, key parameters are selected and the NC architecture is determined, as well as algorithmic support models. However, during the operation of the NC aircraft, external perturbations and the eigenstate of the ТС can change significantly. Therefore, the key parameters no longer adequately reflect the real processes. With intensive maneuvering of the aircraft, parameters appear that were not decisive before, and now they characterize the state of the ТС, while other key parameters become insignificant and fall out of the channel. Evolutionary algorithms [12,13,14] are used to construct algorithmic support models for the NC aircraft. When using an inertial navigation system (INS) as a basic system in NC, as a rule, it is the models of its errors that are built first of all. The self-organization algorithm is based on the model selection hypothesis using an ensemble of criteria.: The use of criteria for the degree of observability, controllability and parametric identifiability in the algorithm for selection of models makes it possible to obtain models with improved qualitative characteristics. The self-organization algorithm allows you to automatically highlight the most significant state variables that are used in the generated model. Since the ensemble of NC selection criteria includes criteria for the degree of observability, controllability, and parametric identifiability, the algorithmic support for NC uses only well-observed and controllable state variables, as well as well-identifiable parameters of INS error models [14, 15, 16, 17, 18]. When changing the operating mode of the aircraft NC, the degrees of observability of the state variables are analyzed and the best NC structure is automatically selected. The mathematical model is used in the estimation algorithm to determine the state of the system under study, as well as in the ensemble of selection criteria. If at the first stage of the NC operation some components of the state vector were poorly observable and were not evaluated [19], then over time it becomes possible to use a more detailed model of the process under study and the degree of observability of these components may increase. In this case, components that were poorly observable in the past (become well observable) are included in the channel composition, i.e. go into the category of estimated components of the state vector. As useful information is accumulated, a more detailed model of the process under study is built using the self-organization algorithm.

If the use of a more detailed model leads to the fact that the degree of observability of a particular parameter increases, then the estimated state vector expands and, finally (in the case when all parameters of the complex become «well» observable), the transition from the reduced to the usual full state vector is carried out. The algorithmic support uses scalar estimation algorithms, therefore, in the case of a change in the dimension of the state vector, it is not necessary to change the matrices of the model of the process being evaluated and the formulas for calculating the gain matrices and covariance matrices of estimation errors. Models of the estimated process — INS errors are built using evolutionary algorithms for each variable. Within the framework of the concept of dynamic system synthesis, in the process of functioning of the NC aircraft, the choice of dominant state variables is carried out, which best determine the process under study. In flight, models with improved properties are built, and these models are used in the NC algorithmic support. Used qualitative characteristics — the degree of observability, controllability and parametric identifiability affect the accuracy of NC. When correcting in the NC structure with the help of a controller, models with improved properties of not only controllability, but also observability and parametric identifiability are used. An increase in the degree of observability of the model leads to an improvement in the accuracy of estimation, and an increase in the degree of parametric identifiability makes it possible to obtain a higher accuracy in the construction of the model. Together, the degree of observability and parametric identifiability allow building more accurate models, which are then used in the state controller. The degree of controllability is a property that determines the effectiveness of the controller and an increase in the degree of controllability leads to an increase in the efficiency of managing the state variables that are included in the model. Thus, the concept of dynamic system synthesis has been implemented in the algorithmic support of NC aircraft.. Thus, the concept of dynamic system synthesis was used to develop algorithmic support for high-precision aircraft navigation. In the NC complexing block, the ensemble of selection criteria includes the criteria for the degree of observability and controllability of the state variables and the identifiability of the parameters of the models used. Models are built in the process of NC operation using evolutionary algorithms. The use of the concept of dynamic system synthesis makes it possible to build models with improved characteristics and use only the dominant state variables during the flight of the aircraft for estimation and control, which increases the accuracy of the NC and greatly simplifies the implementation of the algorithmic support of the NC in the on-board computer of the aircraft.

Литература

- Neusypin K.A. Sovremennye sistemy i metody navedeniya, navigatsii i upravleniya letatelnymi apparatami. M., Izd. MGOU, 2009. – 500s.
- Proletarskiy A.V., Neusypin K.A., Shen Kay. Algoritmicheskie sposoby korrektsii avtonomnykh inertsialnykh navigatsionnykh sistem. V sb. Upravlenie v morskikh i aerokosmicheskikh sistemakh (UMAS – 2014). Sedmaya Rossiyskaya multikonferentsiya po problemam upravleniya: materialy konferentsii GNTs RF OAO «Kontsern «TsNII «Elektropribor». 2014. S. 637 – 641.
- Buy Van Kyong, Neusypin K.A. Algoritmicheskiy sposob povysheniya tochnosti navigatsionnykh sistem. Avtomatizatsiya. Sovremennye tekhnologii. 2005. № 7. S. 11-15.
- Ke Fang, Proletarsky A. V., Neusipin K. A. Selection of measured signals in the navigation measuring complex // Journal of Measurement Science and Instrumentation, 2011, Vol. 02,. No. 12. pp. 346–348.
- Vasilyev S. N., Zherlov A. K., Fedosov E. A., Fedunov B.E. Intellektualnoe upravlenie dinamicheskimi sistemami. M.: Fizmatlit, 2000.
- Ke Fang, Neusipin K. A. Algorithms in intelligent control systems of aerocrafts. China, Chendu: Sichuan University Press, 2011.
- Neusypin K.A. Kontseptualnyy sintez intellektualnykh sistem. Avtomatizatsiya. Sovremennye tekhnologii. 2000. № 6. - S. 23-25.
- Neusypin K.A. Napravleniya razvitiya intellektualnykh sistem. Avtomatizatsiya i sovremennye tekhnologii. 2002. № 12. - S. 12-16.
- Neusypin K.A., Loginova I.V. Voprosy teorii i realizatsii intellektualnykh sistem. – M.: Signal, MPU, 1999.
- Proletarskiy A. V. Kontseptsiya sistemnogo sinteza dinamicheskikh obektov // Avtomatizatsiya i sovremennye tekhnologii, 2007, №8. s.28–33.
- Neusypin K.A. Sistemnyy sintez sistem upravleniya s intellektualnoy komponentoy // Avtomatizatsiya i sovremennye tekhnologii. 2007. № 3. S. 35-39.
- Neusypin K.A. Razrabotka modifitsirovannykh algoritmov samoorganizatsii dlya korrektsii navigatsionnoy informatsii. Avtomatizatsiya. Sovremennye tekhnologii. 2009. № 1. - S. 37-39.
- Chan N.Kh., Selezneva M.S. Razrabotka algoritmov korrektsii navigatsionnykh sistem. Aktualnye voprosy fundamentalnykh nauk. Trudy mezhdunarodnoy nauchnoy konferentsii. M. 2014. S. 155-158
- Ivakhnenko A.G., Myuller Y.Ya. Samoorganizatsiya prognoziruyushchikh modeley. Kiev: Tekhnika, 1985.
- Shen Kay, Neusypin K.A.. Issledovanie kriteriev stepeni nablyudaemosti, upravlyaemosti i identifitsiruemosti lineynykh dinamicheskikh sistem // Mekhatronika, avtomatizatsiya, upravlenie. 2016. № 11. C. 723-731.
- Shen Kay, Neusypin K.A. Kriteriy stepeni nablyudaemosti peremennykh sostoyaniya nestatsionarnykh sistem // Avtomatizatsiya. Sovremennye tekhnologii, 2016, № 6. s. 10–16.
- Neusypin K.A., Proletarskiy A.V., Kuznetsov I.A. Sintez chislennogo kriteriya mery identifitsiruemosti parametrov modeley dinamicheskikh sistem. Avtomatizatsiya. Sovremennye tekhnologii. 2015. № 3. S. 9 – 13.
- Neusypin K. A., Tsibizova T. Y., Kuznetsov I. A. Development of criterion of degree of identifiability of parameters of dynamic systems. 2015 International Workshop on Computer Science and Engineering: Information Processing and Control Engineering, April 15-17, 2015. pp. 66–71.