CC BY
0Вывод. Из расчета траектории различных масс (размеров) окатышей видно, что более крупные окатыши двигаются относительно быстрее мелких, создавая некоторые предпосылки для циркуляции первых вокруг вторых, что должно благоприятствовать окомкованию. Кроме того, расчеты показывают, что окатыши с одним и тем же углом отрыва уо имеют траектории непараллельные и непересекающиеся, т.е. поток сыпучего материала в фазе ссы-пания является расходящим, что подтверждается и кинограммой процесса.
Список литературы
1. Охотин В.В. Физические и механические свойства грунтов в зависимости от минералогического состава и степени дисперсности. - М.: Изд.ГУШДора, 1937. - 16 с.
2. Крылов А. Н. О некоторых дифференциальных уравнениях математической физики, имеющих приложение в технических вопросах [Текст] / А. Н. Крылов. -Л.: Изд-воАНСССР, 1933. - 368 с.
3. Исаев Е.А., Чернецкая И.Е., Крахт Л.Н.. Современная теория окомкования сыпучих материалов. - Старый Оскол: Тонкие наукоемкие технологии, 2001. - 244 с.
Nguyen Minh Tien
Candidate of technical sciences, Trainee of Department 'Operation of road vehicles and car service' of Moscow Automobile and Road Institute, lecturer of 'Engineering dynamics' subject (Vietnam State Technical
University named after Le Quy Don)
DIAGNOSTICS OF HYDRAULIC POWER TRANSMISSION OF ARMORED VEHICLE M113 ON THE
BASE OF NEURO-FUZZY SYSTEM
ABSTRACT
Definition method of technical state of hydraulic power transmission was reviewed, which uses the artificial intellect technologies and is based on neuro-fuzzy systems application. It was noted that this method must have four main stages. Diagnostic system structure, realizing this method, is shown here. Fault identification in control system of hydraulic power transmission is provided here.
Keywords: neuro-fuzzy system, hydraulic power transmission, diagnostics.
Introduction
Hydraulic power transmission (HPT) is an essential part of armored vehicles, which determines its technical and operational characteristics in many respects. At the same time, data about faults and malfunctions of armored vehicles indicate the low level of operation reliability of HPT as a whole, and its friction coupling particular. Technical influence, timely and reasonable by depth and volume directed to support of HPT friction couplings in technically sound condition are possible only if objective diagnostic information is available. However, existing diagnostic methods are focused on usage of external stationary diagnostics means, and usage of complicated, expensive and scarce drum benches of towing performance. Foreign investigations in this sphere are directed to creation of built-in diagnostics means aimed mostly on diagnostics of HPT control system and its electronic units, and cannot define technical conditions of mechanical part of transfer gearbox. In this situation, the most reasonable is development of new diagnostics method for HPT.
Background
During operation of armored vehicle M-113, the following troubles can take place in hydraulic power transmission:
• low pressure in the main line in neutral position of gear stick because of insufficient pump performance; sharp pressure change in the main line during gear actuation due to wear and tear of control shaft seals, main shaft distributor and friction coupling piston;
• low oil pressure in hydraulic transformer due to wear and tear of impeller hub and insufficient pump performance of transformer power section (hydraulic
transformer power section);
• increased oil temperature in hydraulic transformer during vehicle movement on level road due to malfunction of hydraulic transformer parts;
• absence of vehicle movement after switching of one of transmissions at normal oil pressure because of damage friction coupling of switched transmission; vehicle movement in neutral position of gear stick because of sintering of driven and driver disks of any friction couplings;
Many scientists investigated issue of mechanism diagnostics and development of theoretical issues of technical diagnostics. However, the methods, which are used by majority of authors, have a range of limits, because they do not allow taking into account variety of parameters, conditions and situations, which can take place during operation of vehicle. Uncertain and unclear conditions of automation objects functioning include many difficulties to procedure and processing of huge information and do not allow to make control of these objects and their mechanisms, protection and other functions on one-valued features obtained by famous traditional methods. First of all, diagnostics of the object is very difficult in real time. Secondly, and the most important, is the difficulty of technical diagnostics issue, because nowadays it is impossible to create some effective universal method, which takes into account indicated variety of any and all circumstances.
Published works on vehicle diagnostics are very local ones. They are devoted to usage of assessment methods of technical state of separate mechanisms, mostly stationary or mobile technical means. At that, there is no systematic theoretical approach to the
problem of vehicle diagnostics, methodological question of technical diagnostics based on modern means and new technologies are not developed. Moreover, majority of works are devoted to indirect investigation methods. Application of direct methods in engineering practice during solution of diagnostic tasks is limited by complication of mathematic description and analysis of internal dynamic processes in the object, which states the problem of further development of technical diagnostic methods improvement. It is obvious, that new approaches, ways and methods that differ from the classic ones, are required to solve the mentioned problems. First of all, these ways and methods must be concept-based; secondly they should be based on fundamental theoretic developments and relevant mathematical tools.
Modern stage of techniques development is characterized by wide implementation into control systems of microelectronics, which allows providing complex automation of vehicles and technical areas. In this case, automatic devices are able to execute many functions simultaneously. Analysis of problem state shows that classic theory of automatic control does not allow accounting of whole variety of the object functioning conditions, and efficiency of control automation and diagnostics on its scientific principles appear to be much lower than expected one.
A distinctive feature of diagnosed objects is uncertainly of processes that take place during operation, and unpredictable behavior of diagnostics parameters, variety of situations and regimes, incomplete and limited information.
In connection behalf, the important task is appeared: to gather more diagnostic information and further to process it quickly and properly. With appearance of modern means of gathering and submitting of information, as well as due to colossal possibilities of board microelectronics, good backgrounds for diagnostics on brand new level are appeared.
Modern diagnostic methods closely work with information, such as scientific knowledge, because it gives real preconditions and is a base for correct and accurate technical diagnosis.
There are two main methods of acquiring scientific knowledge as follows: theoretical and experimental. They are widely used in diagnostics.
Theoretical method is based on carrying out of analytical procedures, theoretical experiments and simulation models, usage of mathematic tools, analysis of theoretical data, etc.
The test of object must be performed in experimental method of knowledge (iron bird, stand, operational and trial) to make technical diagnosis. Naturally, after that, these data obtained in form of oscillograms, graphics, tables, photos, etc., are processed on the base of mathematic tools or with the help of visual assessment and comparative analysis.
Theoretical and experimental methods are equally used in diagnostic methods. These methods are divided into two big classes, which are reviewed below.
In spite of big variety and wide possibilities of diagnostic methods used traditionally, they have a
range of disadvantages. The main disadvantages and limitations of these methods are as follows:
- complicity during assessment of technical state by measured parameters;
- sufficient labor intensity of works performed during diagnostics process;
- imperfection of methods and ways of diagnostics;
- limited functional opportunities of traditional diagnostic means;
- low efficiency of traditional diagnostics methods;
- low accuracy at technical diagnosing;
- low reliability of diagnostics.
Diagnostics method that is based on fuzzy logic.
Operation of the objects and mechanisms of vehicles takes place in condition of fuzziness, blurring and incompleteness of information about change of parameters. It means, that you need to use relevant mathe-matic tool, if you want to solve diagnostics tasks.
The author has developed a unique method of diagnostics that is based on fuzzy logics. It comes down to the following.
In order to receive information, necessary to form production rules of fuzzy logic, you have to build functional dependencies of one diagnostic parameters from the other,
dk = f(dl); k, l = 1, m, k # l (1)
where d k , dl - diagnostics parameters; m - their quantity.
These dependencies allow defining impact of similar diagnostic parameters.
In order to asses an impact of diagnostic parameters on characteristics of mechanisms and elements of diagnostics object, you can obtain dependencies of the following type:
y = f (dk); j = 1, n; k = 1, m (2)
where yj - object features; n - quantity of features.
Then you need to build an expert system, intended for definition of character and degree of malfunctions of hydraulic gear of HPT. The stages of technical identification of HPT, including complex of scientific-and-technical and research works on stands and laboratories of plants.
Diagnostic method is based on neuro-fuzzy identification. The essence of method is as follows. Analysis of occurrence causes of armored vehicles transitional operation modes is based on assessment of correlations between the changes of parameters of X input and Y output of certain system.
Identification stages of HPT technical state are as follows:
• arming of HPT with sensors and measuring tools;
• performance of experimental studies on factory stands and obtaining of features;
• connection of expert intellectual system to HPT and obtaining of technical diagnosis by expert on display.
Neuro-fuzzy network identifies certain malfunctions of the object, provided that it was previously trained how to define one or another fault or malfunction. The classic mathematic training principles of neuro or neuro-fuzzy networks are used. This method contains several stages.
Stage 1. Gathering of expert information by faults and malfunctions; formation of knowledge base. At this stage, knowledge base is formed with the help of experimental data, expert's opinion, and with the help of other reliable informational sources. It represents a combination of training selections, characterizing factors and signs of malfunctions; these selections further passed to neuro-fuzzy system input.
Thus, there is a task to define technical state of the elements of hydraulic power transmission of armored vehicle M-113, which is operated in Vietnam conditions. Source information about HPT functioning, behavior of its parameters, possible malfunctions and their reasons is formed as expert knowledge base on the basis of reliability of data obtained from different sources. The results of experimental tests in stand conditions and in scope of trial tests, as well as statistics data on faults can serve as additional sources of knowledge base formation.
Formed knowledge base is used during training of neuro-fuzzy networks for identification of malfunctions. The quantity of production rules of similar knowledge base contains several dozens for description of technical state of the separate mechanism or several hundred for the whole vehicle.
Stage 2. Creation of neuro-fuzzy model. Neuro-fuzzy model is formed on mathematic theory of neural networks and means of neuro-fuzzy logic (pic. No.1). Network inputs are represented by information variables X (diagnostic parameters), described by membership functions i = 1, NJ = 1, M (N - number of information variables, M - linguistic variables by each
parameter). Network output - vector Y - is represented by various categories, which define measure of efficiency, quality and safety of functioning of hydraulic power transmission of armored vehicle M-113 (coefficient of reduction ratio, vibration, compression oscillation, leakage, etc.)
Picture 1. Neuro-fuzzy network
Parameters are fuzzified to transform strong signals into fuzzy type. Each of diagnostics parameters a¡j is described by several (three-five) terms of linguistic variable. Generally, triangle and Gaussian membership functions are used for these purposes.
Specialized software (for example, Fuzzy Logic Toolbox of MATLAB 7.0 software) is used for neuro-fuzzy modelling. Approximating model includes the program with *.fis, extension, based on representation of input parameters of fuzzy rules with fuzzy variables, and the programs of test and training formation *.m on the base of real function processes of HPT. During the usage of model in real time mode, the sourced data about HPT parameter change come from register - a microprocessor or board computer.
The structure of adaptive fuzzy network -ANFIS, approximating output of diagnostics system, is shown at the Picture 2. At that, the number of network inputs is equal to the quantity of diagnostics parameters.
Picture. 2. Structure of ANFIS model.
Stage 3. Training of neuro-fuzzy system. During the training of network, a set of pairs of training samples is put on its input, characterizing a combination
of diagnostics parameters for different types of technical state of HPT mechanisms. Definition, received on network output, is interpreted in the following way:
Input layer Inner layer Inner layer Inner layer Output layer
Picture. 3. Adaptive neuro-fuzzy network - ANFIS
Training method for model, shown in the Picture 3, with use of computer with Windows XP Professional operation system and installed computation model MATLAB 7.0 takes only 10 s. The result of training was reached within 10 periods.
Stage 4. Neuro-fuzzy identification and conclusion. This stage is final in the process of technical state definition of HPT elements.
At that, to receive technical diagnosis, trained neuro-fuzzy system is used, and information about parameters is put on network input, characterizing real process in HPT. For convenience of diagnostic system, figure information, received on network output is submitted to additional interpretation and is given in verbal form (linguistic form).
At that, special interpretation windows are used, in which information of expert system about technical state of HPT mechanisms for chosen function mode is output in visual-verbal way. In review window of expert system, one can see graphic information about behavior of parameter in control and diagnostics modes. In special lines, different messages are output: types and names of diagnostics parameters, their current values, warning about emergency situations or reaching of critical values, etc.
Interface elements of expert system can be rather easy changeable according to user's wish. Expert system for diagnostics of HPT, realizing this created method, is accompanied by software, developed on modern software languages with attraction of special means of visual development.
Offered method provides obtaining of reliable knowledge base, promptness of information processing, accuracy of technical diagnosis and opportunity of fast definition of armored vehicle mechanisms in real time mode.
Benefits of the proposal method contain as follows:
- to use extended reliable knowledge base;
- high speed of information processing;
- high accuracy of technical diagnosis;
- ability of prompt diagnostics in real time mode;
- convenient regulatory diagnostics on special posts, maintenance services, on ore-dressing enterprises, etc.
Conclusion
Diagnostics of complex technical objects have the following: there are separate parameters among parameters of complex technical objects, which are signs of its technical state, they are compared with reference signs of source alphabet of grades. Perspective directions of diagnostics methods development and means are the methods, based on fuzzy logic and fuzzy sets, expert systems and neuron networks. Artificial neuron networks are used for identification of control objects, pattern recognition and forecast of technical system state. Application of INS allows obtaining increased speed of diagnostics means due to torrent paralleling of diagnostics information. Diagnostic method of complex technical objects is based on application of neuro-fuzzy networks, adapted to diagnostics tasks of hydraulic mechanical transmission of armored vehicle M-113.
References
1. Birger I.A. Technical diagnostics. - M.: Machine building, 1978. - page No. 240.
2. Viktorova Ye.V. Application of fuzzy neuron networks for technical diagnostics of road vehicles/ Ye.V. Viktorova // Vestnik of Kharkiv National Road-Transport University, - 2012, - issue No. 56. - pages 98-102.
3. Khakhanov V.I., Shcherba O.V. Application of artificial neuron networks for digital network diagnostics. / V.I. Khakhanov, O.V. Shcherba// Radioelec-tronic and computer systems. - 2010. - No. 5 (46), -pages 15-20.
4. Kruglova T.N. Fuzzy expert method of diagnostics of technical state of cutter-loader/ T.N. Kruglova // Scientific works DonNTU. - 2010. -No. 18(172). - Pages 179-185.
