RU2156496C1 - Method for on-line dynamic analysis of fuzzy state of object or process described by multiple parameters - Google Patents

Abstract

FIELD: computer engineering. SUBSTANCE: method may be used, in particular, for computer-aided and expert systems for recognition and diagnostics of functional condition of aircraft and spacecraft equipment, expert systems and situation centers for dynamic analysis of social power production or financial objects and processes, which are described by multiple parameters. Said method involves representation of identified current states of chosen characteristic as respective information color-code signals of visible spectrum in sequence in time with generalization over complete set of parameters in given time interval. In addition, converting is achieved by means of generation of information color-code signal, depends on results of evaluation of current fuzzy object state. Then, method involves displaying color-code signals using bar chart matrix of object states. Columns of matrix correspond to identified value of considered characteristic. Its lines correspond to given time intervals. Then, method involves detection of sequence and type of alteration of object states. EFFECT: increased speed of representation and analysis of current values of fuzzy states of objects. 2 dwg

Description

 The invention relates to the field of computer technology and can be used in various automated recognition systems (diagnostics), cluster analysis, identification of the technical and functional state of products of the aviation and space industry, as well as in expert systems and situational centers for dynamic analysis of social, energy and financial economic multi-parameter objects or processes.

 Known devices and methods for monitoring and diagnosing the state of a technical object (USSR, AS N-01504653, A1, G 06 F 15/46, 1989), during the implementation of which during monitoring and diagnostics slow changes in parameters are recorded for each cycle , and the obtained data is compared with the reference values, and based on the comparison, a conclusion is made about the state of the object, as well as a method for entering readable automatic digital data into grayscale images (ЕВП / EP /, N 0493053, A2, G 06 K 1/12, 19 / 16, 15/00, 1992) and a data processing method (EPP / EP /, N-0493105, A1, G 06 F 15/17, 1992 )..

 The proposed method and device does not allow to quickly diagnose the state of a multi-parameter dynamic object or process (MPO) for a large number of measuring parameters.

 A known method for the on-line diagnosis of MPO conditions according to measurement information (RF patent for the invention N 2125294 // Bulletin of the invention N 2, 1999). The method allows diagnosing MPO states by a large number of parameters, however, it does not allow assessing the magnitude and nature of the change in the integral state (class of states) of the controlled object over the entire set of observed measurement parameters, the current value of which can be fuzzy (blurry, vague, uncertain).

 The closest in technical essence is the method of monitoring and evaluating the technical condition of a multi-parameter object according to telemetric information (RF patent N 2099792 // Bulletin of inventions N 35, 1997, p. 595). The proposed method allows you to quickly detect the sources of disturbances and their occurrence in the controlled (telemetry) objects. However, the method does not allow to evaluate the magnitude and nature of the change in the integral state of the controlled object over the entire set of observed measurement parameters, the current value of which may be fuzzy.

 For complex MPOs, the current state of which can be ambiguous (uncertain), the unambiguous tolerance assessment of the fuzzy state (value) of each of the dynamic parameters with a large set of them causes certain difficulties, which is associated with the ambiguity of assigning the current fuzzy state of the dynamic parameter and, in general, MPO to one or another class of clear states (in the admission - not in the admission, serviceable - faulty). This is especially evident when there is a shortage of time, for example, when solving the problem of dynamic analysis in real time, the so-called at the pace of change in the current state of MPO.

 The purpose of the invention is the operational presentation and dynamic analysis of the fuzzy state of a multi-parameter object or process with the determination of the relative magnitude and nature of the change in its state, controlled over the entire set of observed measurement parameters, the current value of which may be fuzzy.

 The goal is achieved by the implementation of the proposed method for operational dynamic analysis of the fuzzy state of MPO, which allows to implement the principle of taking into account the history of the MPO based on the target decomposition of the initial fuzzy set of the multi-parameter space of the state of the object into classes of clear and fuzzy states (values) of heterogeneous dynamic parameters.

 The method allows visual dynamic analysis of both clear (uniquely identified) and fuzzy (ambiguously identified) current states of the object from the screen of one multicolor video monitor and quickly (in real time) to determine the relative magnitude and nature of the change in the diagnosed generalized state of MPO with sequence estimation ( prehistory) its changes. All this in combination provides a reduction in the time of analyzing the state of MPO and the used technical properties of the display of dynamic parameters for information support of decision-making on diagnostics of the state of MPO and which is an element of an automated diagnostic system.

где A₁ ⁿ(t_i) - штатное четкое состояние n-го параметра, т.е. параметр однозначно находится в t_i-й момент времени в допуске; A₃ ⁿ(t_i) - нештатное четкое состояние n-го параметра, т.е. параметр однозначно находится в t_i-й момент времени не в допуске;

- нечеткое состояние n-го параметра, т. е. система обработки в t_i-й момент времени не может в силу тех или иных причин однозначно определить в допуске или не в допуске динамический параметр. Нечеткость состояния параметра определяется тем, что его текущее значение в виде размытости, расплывчивости может быть отнесено как штатному, так и нештатному состоянию.When analyzing the state of MPO observed by the nth dynamic parameter (indicator of the state of MPO), its current state at the t _i -th time moment can be represented as an expression

where A ₁ ⁿ (t _i ) is the regular clear state of the nth parameter, i.e. the parameter is uniquely located at the t _i -th time moment in the tolerance; A ₃ ⁿ (t _i ) is an abnormal clear state of the nth parameter, i.e. the parameter is unambiguously at the t _i- th moment of time out of tolerance;

- the fuzzy state of the nth parameter, that is, the processing system at the t _i- th moment of time cannot unambiguously determine the dynamic parameter in the tolerance or not in the tolerance. The fuzzy state of the parameter is determined by the fact that its current value in the form of blurriness, vagueness can be attributed to both regular and abnormal conditions.

 Generalizing expression (1) over the entire set of dynamic parameters n∈N or MPO as a whole, we obtain possible generalized estimates of the state classes.

Состояние динамических параметров, оцененных в соответствии с выражением (2) по этапам функционирования (движения) объекта, определяет соответственно его обобщенное (интегральное) состояние и переходы объекта из одного состояния в другое (динамику состояний). Например, в простейшем случае множества

определяют соответствующие классы штатных K^ш и нештантых K^н состояний МПО. Очевидно, чем более неопределенней (неоднозначней) ситуация на выходе МПО и чем большее влияние оказывают различные дестабилизирующие факторы на измерительный сигнал (динамический параметр), тем труднее провести оценку текущего нечеткого состояния МПО с идентификацией исходного многопараметрического пространства на классы { A₁(t_i), A₃(_i)} четких состояний динамических параметров и тем большее количество динамических параметров попадает в класс

нечетких состояний.

The state of dynamic parameters evaluated in accordance with expression (2) from the stages of the functioning (movement) of an object determines its generalized (integral) state and transitions of an object from one state to another (state dynamics), respectively. For example, in the simplest case, the sets

determine appropriate classes of regular K ^w and K ⁿ neshtantyh MPO states. Obviously, the more uncertain (ambiguous) the situation at the MPO output and the greater the influence of various destabilizing factors on the measuring signal (dynamic parameter), the more difficult it is to evaluate the current fuzzy state of the MPO with identification of the initial multi-parameter space to the classes {A ₁ (t _i ) , A ₃ ( _i )} of clear states of dynamic parameters, and the greater the number of dynamic parameters falls into the class

fuzzy conditions.

можно представить в виде соотношения

где ΔA₁ и ΔA₃ - значение погрешностей определения A₁(t_i) и A₃(t_i) при оценке параметров из исходного множества A(t_i), которые не позволяют однозначно идентифицировать (классифицировать, распознавать) анализируемый параметр и отнести его к одному из четких подмножеств A₁(t_i) и A₃(t_i).Fuzzy subset

can be represented as a ratio

where ΔA ₁ and ΔA ₃ is the value of the errors of determination of A ₁ (t _i ) and A ₃ (t _i ) when evaluating the parameters from the initial set A (t _i ) that do not allow to uniquely identify (classify, recognize) the analyzed parameter and attribute it to one of the clear subsets A ₁ (t _i ) and A ₃ (t _i ).

где N - общее количество динамических параметров, N(t_i) - количество динамических параметров, текущее значение которых в t_i-й момент времени отнесено к одному из подмножеств (классов)

Используя введенные характеристики (2) и (4), можно получать различные цветокодовые матрицы-гистограммы. Так, кодируя определенным цветовым кодом видимого спектра каждый из выделенных классов состояний (2) динамических параметров и представляя относительную величину A_j ^*(t_i) в виде информационного поля соответствующего подмножества динамических параметров получаем 3-х уровневую цветокодовую матрицу-гистограмму состояния МПО.The exception of the complex operation of identification (classification, recognition) of fuzzy states of dynamic parameters only into classes of clear states (equivalence classes), as proposed in the prototype, and the transition to a flexible classification based on the introduction of a fuzzy state class (tolerance class) of a dynamic parameter can significantly simplify the operational process processing and analysis of fuzzy states of dynamic parameters and the fuzzy state of MPO as a whole. We introduce the generalized characteristic

where N is the total number of dynamic parameters, N (t _i ) is the number of dynamic parameters, the current value of which at the t _i th moment of time is assigned to one of the subsets (classes)

Using the introduced characteristics (2) and (4), it is possible to obtain various color-code matrix histograms. So, encoding with a certain color code of the visible spectrum each of the distinguished classes of states (2) of dynamic parameters and representing the relative value A _j ^* (t _i ) in the form of an information field of the corresponding subset of dynamic parameters, we obtain a 3-level color-code matrix-histogram of the MPO state.

 As an estimated MPO indicator observed using a dynamic parameter, there may be: pressure, temperature, vibration, etc. Accordingly, as the estimated characteristics of the process (state) of the dynamic parameter can be: amplitude, frequency, dispersion, gradient, etc. dynamic parameters, for example, rapidly changing (vibration parameters), slowly changing parameters, path parameters, signal parameters, etc., are used as an estimated characteristic.

 The essence of the method, therefore, is that in order to provide an operational dynamic analysis of the fuzzy state of a multi-parameter object or process, the results of the tolerance assessment of the clear states of homogeneous dynamic parameters are converted into the corresponding information color-code signals with a generalization over the entire set of dynamic parameters at a given time interval, while the estimated characteristics of the process can be the amplitude, frequency, etc., as The two parameters of the evaluated characteristic use heterogeneous dynamic parameters, the results of the tolerance assessment of the fuzzy states of the dynamic parameters are used as the assessment results, the conversion operation is carried out by generating the corresponding information color code signal of the visible spectrum depending on the results of the tolerance assessment of the current states of the dynamic parameter with its subsequent assignment to one of classes of clear or fuzzy states (in tolerance, not in tolerance, not known stn) with a generalization over the entire set of parameters at a given time interval, information color-code signals are displayed by means of a three-level matrix of state histograms whose columns correspond to the relative values of the estimated classes of clear and fuzzy states of the object's dynamic parameters, the rows correspond to given time intervals, and the given color ( color code) - the corresponding class of state of the dynamic parameter, determine the relative magnitude and nature of the change in the integral state could parametric object by changing directions and relative values of the change in time information tsvetokodovyh signals.

Thus, the fuzzy multiparameter state space of the entire set (flow) of dynamic parameters, regardless of its power, can be represented on the screen of one multicolor video monitor (display) in the form of a 3-level color-code matrix-histogram of MPO states [A _j ^* (t _i ) • t _i • z (A _j ⁿ (t _i ))], where A _j ^* (t _i ) is the relative number of dynamic parameters belonging to the jth class of states, j = 1,2,3; t _i is the registration time; z (A _j ⁿ (t _i )) is the color (color code) corresponding to the jth class of the parameter state.

 The novelty of the proposed method compared to the prototype and known methods for presenting and analyzing the state of a multi-parameter object or process consists in the fact that a logical sequence of actions has been developed for presenting and analyzing the fuzzy state of MPO, which leads to the achievement of the goal of the invention. The basis of this approach is proposed to use a new approach to the decomposition of the multi-parameter space of fuzzy states of dynamic parameters into classes of clear and fuzzy states, which greatly facilitates the implementation of classification algorithms (clustering), recognition (diagnostics), identification and information support for decision-making on MPO management.

 Thus, the set of essential features leading to the desired result in the patent and scientific literature is not found, which indicates the "inventive step" of the proposed technical solution.

The essence of the proposed method is well illustrated in the analysis of various vibrational parameters. Let the amplitude be chosen as the estimated characteristic of the process, the vibration parameters are used as the parameters of the estimated characteristic, and the results of the tolerance evaluation of the fuzzy states of the vibration parameters are used as the evaluation results. Then, an operational dynamic analysis of the fuzzy state of an object or process will be considered using the example of estimating the power and dynamics of shock distribution (local explosion) of MPO (Fig. 1 and Fig. 2). In FIG. Figure 1 shows the traditional presentation of graphs of changes in some dynamic MPO parameters with measuring sensors installed on it n = 1,2, ..., N, where N = 7, which form the corresponding dynamic parameters, the amplitude of which varies from A _min to A _max . In FIG. Figure 2 shows the color-code matrix diagram of the power distribution and the nature of the impact (local explosion) of the MPO with estimates of the amplitude of its current particular (parametric) clear and fuzzy states. In the general case, the representation of the process of changing the fuzzy state of MPO can be represented as a multilevel matrix-histogram of the MPO state [A _j ^* (t _i ) • t _i • z (A _j ⁿ (t _i ))], where t _i-1 is the beginning and t _{i + 3} is the end of the shock process at MPO.

где K^ш - класс штатного четкого состояния МПО;

- класс нештатного нечеткого состояния, где нечеткость определяется в том числе и нечеткостью множества

- класс штатного нечеткого состояния МПО, нечеткость которого определяется возможными последствиями возникновения нештатной ситуации на интервале t_i-1 - t_i+3. Так несмотря на штатное состояние всех динамических параметров (вибропараметров) после t_i+3-го момента времени, в объекте могли произойти какие-нибудь необратимые изменения, ненаблюдаемые с помощью рассматриваемой группы вибропараметров. Для идентификации наличия этих возможных изменений, т. е. устранения нечеткости необходима дополнительная обработка другой группы параметров (как правило, невибрационных).The fuzzy state of MPO is determined in the general case by the set of both clear and fuzzy states of dynamic parameters. We define the set (class) of K states of MPO in the form of three subsets

where K ^W - class regular clear state of MPO;

- class of contingency fuzzy state, where fuzziness is determined including fuzziness of the set

- the class of the regular fuzzy state of the MPO, the fuzziness of which is determined by the possible consequences of an emergency in the interval t _i-1 - t _{i + 3} . So, despite the regular state of all dynamic parameters (vibration parameters) after t _{i + the 3rd} moment of time, some irreversible changes could have occurred in the object that were not observed using the considered group of vibration parameters. To identify the presence of these possible changes, i.e., to eliminate fuzziness, additional processing of another group of parameters (usually non-vibrational) is necessary.

вышедших за поле допуска;
оценить интегральное распределения мощности удара (локального взрыва) на объекте во времени по общему количеству вибрационных параметров, распределенных по множествам

Таким образом, способ позволяет обеспечить динамический анализ изменения интегрального нечеткого состояния многопараметрического объекта с экрана многоцветного видеомонитора, оперативно (в реальном масштабе времени) обнаруживать изменение класса состояния МПО и оценивать динамику состояния по всему множеству динамических параметров. Такое представление позволяет комплексно оценить динамику состояния МПО по большому множеству динамических параметров, текущее состояние которых может быть нечетким, а сами динамические параметры могут быть разнотипными (аналоговыми или цифровыми, быстро меняющимися и медленно меняющимися и т.п.). Все это в комплексе позволяет оперативный динамический анализ изменения состояния МПО, обеспечивает сокращение сроков обработки и анализа информации и используемых технических средств ее отображения для информационной поддержки принятия решений обработчиком-аналитиком.The analysis of the present representation (Fig. 2), revealing the essence of the proposed method, allows for an operational dynamic analysis of the state of MPO, including:
evaluate the nature of the impact (local explosion) of an object, by the type (character) of a change in its integral state. So from FIG. 2 clearly implies that the impact was pulsed with a maximum load at time t _i-1 , a uniform decrease (attenuation) in the interval t _i - t _{i + 3} and the complete termination of the impact at time t _{i + 3} ;
to evaluate in relative terms the maximum impact power (local explosion) in the interval t _i-1 - t _i according to the total number of vibration parameters, unambiguously (A ₃ ) and ambiguous

out of the field of admission;
to estimate the integral distribution of the impact power (local explosion) at the object in time by the total number of vibration parameters distributed over the sets

Thus, the method allows for dynamic analysis of changes in the integral fuzzy state of a multi-parameter object from the screen of a multi-color video monitor, to quickly (in real time) detect a change in the MPO state class and evaluate the dynamics of the state over the entire set of dynamic parameters. This representation allows us to comprehensively assess the dynamics of the MPO state by a large number of dynamic parameters, the current state of which can be fuzzy, and the dynamic parameters themselves can be of different types (analog or digital, rapidly changing and slowly changing, etc.). All this in combination allows an operational dynamic analysis of changes in the state of MPO, provides a reduction in the processing time and analysis of information and the technical means used to display it for information support decision-making processor-analyst.

 The undoubted advantage of the proposed method is the pronounced cognitiveness of the presentation of fuzzy situations, so the resulting flower-mode descriptions of the distribution of subsets of fuzzy-estimated parameters in time carry generalized information about the nature of the observed process (impact), as well as the color-code descriptions of the distribution of sets of clearly-estimated parameters. To solve the problems of operational diagnostics of the dynamics of shock processes, the proposed approach avoids the use of complex classification algorithms for fuzzy multiparameter space with the MPO parameter, and, therefore, significantly improves the efficiency and complexity of dynamic analysis of rapidly changing processes.

 It is advisable to use for dynamic analysis of complex parameter objects and processes in various practical applications.

Claims (1)

 A method for the on-line dynamic analysis of the fuzzy state of a multi-parameter object or process, which consists in quickly converting the results of the tolerance assessment of the states of dynamic parameters into corresponding information color-code signals with a generalization over the entire set of dynamic parameters in a given time interval, characterized in that either amplitude, or frequency, or dispersion, or gradient, as parameters of the character The critics use heterogeneous dynamic parameters, as the results of the tolerance assessment, use the tolerance estimates of the fuzzy states of the dynamic parameters, with the mentioned transformation, the formation of the corresponding information color-code signal is carried out in the visible spectrum depending on the results of the tolerance assessment of the current state of the dynamic parameter with its subsequent assignment to one of the classes of clear or fuzzy with respect to tolerance states with a generalization over the entire set of parameter in a given time interval, information color-code signals are displayed by means of a three-level matrix of histograms of states, the columns of which correspond to the relative values of the estimated classes of clear and fuzzy states of the dynamic parameters of the object, and the rows correspond to the given time intervals, determine the relative value and nature of the change in the integral state of the multi-parameter object from the directions of change and the relative magnitudes of this change in time of information color signals.

Images