RU2649075C1 - Method for monitoring technical condition of building objects with processing of results determining state of monitoring object with use of soft computing - Google Patents

Abstract

FIELD: monitoring systems.

SUBSTANCE: invention relates to the field of automated monitoring systems for the technical condition of buildings and structures and can be used in the design and operation of buildings and structures. Proposed method involves the selection of controlled elements (constructions) of a building object, the status of which is used to determine the condition of the building object, recording of the values of the measured parameters, characterizing the state of the selected monitored elements, calculating the current values of the monitored parameters from the measurement results and their processing using the soft computing method. Then, the states of the monitored parameters are determined using the results of comparing the calculated values with the threshold values of the monitored parameters, which is followed by the identification of the state of the monitored elements and / or the building object as a whole based on the choice of the worst state of the relevant monitored parameters, and displaying the monitoring information and evaluation results of individual controlled elements and / or the building object as a whole in a visual form.

EFFECT: technical result consists in increased reliability of detecting structures that are in an emergency or pre-emergency condition, increased accuracy of determining the state of controlled structures and the object as a whole by eliminating the systematic error of measuring instruments and the possibility of interrelated analysis of measurements from various instruments, increased performance due to the reduction of the array of processed information.

1 cl

Description

FIELD OF THE INVENTION

The invention relates to the field of automated systems for monitoring the technical condition of buildings and structures and can be used in the design and operation of buildings and structures.

State of the art

The prior art devices of the same purpose as the claimed invention.

A known system for determining the stability of buildings and structures used to determine the stability of objects (buildings and structures), while the system for determining the stability of buildings and structures contains a shock device block, a block for generating an electric clock, a block for converting oscillations into an electric signal, an analog-to-digital conversion block an electric signal, a digital storage unit and a digital storage control unit, an experimental and / or input input unit even values of surface strength and / or bulk strength, and / or reinforcement parameters of structural elements of the object, and / or precipitation, and / or shear, and / or roll of the object, and / or the depth of the foundation, and / or its surface strength, and / or its bulk strength, and / or period of natural vibrations of the soil under the object, and / or around it, measured at least by the first tone of vibrations and / or the level of groundwater, a unit for comparing experimental data with normalized data calculated for data structures and m the material of the test object and the composition of the soil beneath and / or around it and the unit for reproducing the obtained data connected to the control and data buses with each other and with the other functional units of the system (RF patent for the invention No. 2245531, IPC G01M 7/00, publ. dated January 27, 2005).

Also known is a system for monitoring the technical condition of buildings and structures, comprising a shock device unit, a vibration sensor unit, a processing and output information unit, an object vibration acceleration measurement unit and / or an object vibration velocity measurement unit and / or an object vibration amplitude measurement unit and / or a measurement unit slopes and / or a unit for measuring deflections and / or a unit for measuring stresses and / or a unit for measuring loads and / or a unit for measuring absolute and uneven settlement and / or a unit for monitoring cracks, joints and seams and / or bl ok measuring the geodetic parameters, the gradation block of the output information, the output of the vibration sensor block and / or the output of the object vibration acceleration measuring unit and / or the output of the object vibration velocity measuring unit and / or the output of the object vibration amplitude measuring unit and / or the output of the slope measuring unit and / or the output of the unit for measuring deflections and / or the output of the unit for measuring stresses and / or the output of the unit for measuring loads and / or the output of the unit for measuring absolute and uneven settlement and / or the output of the unit for monitoring cracks, joints and seams and / or stroke parameters geodetic measuring unit connected to the input of the processing and output information, the output of which is connected to the input of the output gradation information (RF patent for utility model №66525, IPC G01M 7/00, publ. 09/10/2007).

The known method and the system for monitoring and predicting the technical condition of buildings and structures intended for its implementation (RF Patent for the invention No. 2381470, IPC G01M 7/00, publ. 10.02.2009). The method according to patent No. 2381470 includes the excitation of object vibrations at natural frequencies, registration of vibrations, and / or accelerations of vibrations, and / or vibration frequencies, and / or vibration amplitudes, and / or inclinations, and / or deflections, and / or stresses, and / or loads, and / or measurements of absolute and uneven settlement, and / or geodetic parameters, and / or control of cracks, joints, seams, filtering of the technical condition of buildings and structures, divided into two groups: a group of parameters of the technical condition of the lower part of the object and group couples technical parameters of the upper part of the object, determined using the parameters of the technical condition of the lower part of the object by mathematical (computer) modeling of the object, the calculated parameters of the building structures of the upper part of the object, compare the calculated parameters of the building structures of the upper part of the object with the same parameters of the building structures of the upper part of the object, determined by the results of field measurements from sensors to monitor the technical condition of the upper part of project, adjust the parameters of the mathematical model of the object, provided that the calculated parameters of building structures of the upper part of the object, determined by the results of mathematical modeling, differ from similar parameters of building structures of the upper part of the object, determined by the results of field measurements by an amount greater than a given threshold, determined by the measured parameters technical condition of the lower part of the object trends in the parameters of the technical state of the lower part of the object, extrapolate t The end values of the parameters of the technical condition of the lower part of the object for a given time interval are determined on the basis of the extrapolation of the parameters of the technical condition of the lower part of the object; the estimated design parameters of the technical condition of the building structures of the upper part of the object; parameters of the technical condition of building structures of the upper part of the object with Only valid values.

Device for implementing the above method

The system for monitoring the technical condition of buildings and structures includes a shock device, a processing unit and output information, a gradation block for output information, and / or object vibration measurement sensors, and / or object vibration acceleration sensors, and / or object vibration velocity sensors, and / or sensors for measuring the amplitudes of oscillations of the object, and / or sensors for measuring slopes, and / or sensors for measuring deflections, and / or sensors for measuring stresses, and / or sensors for measuring loads, and / or sensors for measuring a absolute and uneven settlement, and / or sensors for monitoring cracks, joints and seams, and / or sensors for measuring geodetic parameters, pressure sensors (including for monitoring the pressure of the object on the ground and / or the pressure of the soil on the object), and / or measurement sensors deformations, and / or temperature measurement sensors, and / or humidity measurement sensors (in this case, all of the above sensors are combined in one block, a block of sensors and equipment of an automated monitoring system), a block for calculating the parameters of the technical condition of the object, a filter block parameters of the technical state of the object, a unit for determining trends and extrapolation of the parameters of the technical state of the lower part of the object, a comparison unit, a threshold device, a block of mathematical modeling and calculation of parameters of the technical condition of the upper part of the object, a block for adjusting the parameters of the mathematical model of the object, an electronic key, a display unit for forecasting and monitoring information, and the output of the sensor unit and the equipment of the automated monitoring system is connected to the input of the calculation unit parameters of the technical state of the object, the first output of which is connected to the input of the filtering unit of the parameters of the technical state of the object, and the second output is connected to the input of the processing unit and output information, the output of which is connected to the input of the gradation block of output information, the first output of the filtering block of the parameters of the technical state of the object is connected to the first input of the block of mathematical modeling and calculation of the parameters of the technical condition of the upper part of the object, the second output of the block of filtering the parameters of technical the state of the object is connected to the input of the comparison unit, the output of the comparison unit is connected to the input of the threshold device, the first output of which is connected to the input of the parameter adjustment block of the mathematical model of the object and the first control input of the electronic key, and the second output is connected to the second control input of the electronic key, the output of the block adjusting the parameters of the mathematical model of the object is connected to the second input of the block of mathematical modeling of the object and calculating the parameters of the technical condition of the upper hour the object, the first output of which is connected to the input of the comparison unit, and the second output is connected to the first input of the predictive and monitoring information display unit, the third input of the object mathematical modeling unit and the calculation of the technical parameters of the upper part of the object is connected to the output of the trend determination and extrapolation unit of the technical state of the lower part of the object, the input of which is connected to the output of the electronic key.

There is also a method of monitoring the technical condition of construction objects (RF Patent for the invention No. 2460980 IPC G01M 7/00, publ. 09/10/2012), which includes the determination of the controlled elements of a construction object based on an analysis of threats and / or design features and / or location , and / or external influences, and / or analysis of the stress-strain state of the building object, the formation of symmetrical pairs of controlled elements of the building object and / or their parts, the definition of controlled parameters, the state of the formed set of controlled elements and / or their parts, determining a set of measured parameters, on the basis of which it is possible to determine controlled parameters, determining acceptable values or intervals of acceptable values of controlled parameters, according to which the technical condition of the object is determined, measurement for symmetrical controlled elements of the building of the object and / or their parts of the parameters, on the basis of which the absolute and relative values of parameters being monitored, comparing the absolute values of the monitored parameters with their permissible values or ranges of permissible values, and the relative monitored parameters with the permissible measurement error, judging by the comparison results obtained at the initial time interval, the adequacy of the mathematical model of the object, when judging by the results of the comparison of inadequacy correction of the mathematical model of the object, the formation of conclusions about the current technical condition of the object on the basis of Ia absolute and / or relative values of monitored parameters with their permissible values or ranges of permissible values specified in the form of specific values or ranges. The described method increases the speed of the method and system for monitoring and forecasting the technical condition of buildings and structures protected by patent No. 2381470.

A common drawback of the listed technical solutions is the insufficient accuracy of diagnosing the current technical condition of the construction site.

These technical solutions determine the technical condition of the building and structure only at the time the sensors take various characteristics (periods of natural vibrations, inclines, etc.) or carry out a predictive assessment of the future technical condition of the building and structure for a long time interval. The experience of using automated monitoring systems has shown that the measurement results are a large array of data and for its correct processing and obtaining reliable results for assessing the technical condition of the object, it is necessary to perform analytical processing of the measured data.

In addition to the aforementioned disadvantages of the analogues known from the prior art [1] - [23], the components characterizing the state of the controlled elements (structures) are not allocated or evaluated in the sensor readings either. So, the readings of the sensor measuring the angle of inclination indicate a change in the angle at a particular point, which can be caused by a change in the state of several structural elements. For example, a sensor whose readings are used to monitor the condition of a foundation slab can be installed (for reasons of ease of installation) on a column mounted on this slab. If the sensor readings indicate an emergency, then it is necessary, referring to the readings of other sensors, the readings of which determine the state of the foundation slab and those structural elements on which each of the sensors is installed, determine the reason for the change in the readings of the sensors (for example, the angle has changed due to that the foundation plate or column on the foundation plate on which the sensor is mounted) has changed its position in space). With an interconnected analysis of the readings of various sensors installed on various controlled elements, it is possible to identify the causes of changes in the monitoring results and increase the accuracy of the monitoring system.

Another disadvantage of the known analogues is that the assessment of the technical condition of the object based on the results of instantaneous measurements can contain a large percentage of false estimates. For example, instantaneous measurement and evaluation by the results of measurements of tilt sensors can lead to a false assessment of the state of structures if a machine drives near the sensor (for example, in a parking lot in the underground part of the building) or a person stomps near the measuring station or equipment that creates equipment near the measuring station vibrations. With short-term impacts on the object near the measuring devices, the controlled parameters can go beyond the controlled values, which in reality does not indicate a deterioration in the technical condition of building structures.

In addition to these shortcomings, these monitoring methods and systems involve processing significant amounts of information due to the fact that it is not intended to select controlled elements depending on the degree of their influence on the state of the object, their significance in emergency situations for the whole or its parts, in significantly affecting the condition of the facility being monitored.

The well-known [1] - [23] solutions do not raise the question of taking into account the systematic errors of measuring instruments, which also negatively affects the assessment of the current or forecasted state of the monitoring object.

The closest technical solution supporting the development of a design method for MIS based on the concept of soft computing is the intelligent Grid system [24], which provides system integration of computing and information components, the formalized control logic of which is associated with solving resource-intensive problems in the study of complex phenomena and laws of dynamic systems using the block of intellectual support for the functioning of the Grid system, interconnected by a block of software controls with human-computer interaction blocks and a block of applied grid services, the intellectual support block containing an expert system that ensures the functioning of the grid system in a given computing environment and decision-making on managing computing processes, an adaptation block that implements adaptive learning procedures due to the possibility of control a computational process with dynamically changing information, the choice of the preferred computational data processing technology, tuning logical models for perceiving new information and extracting “hidden” knowledge, a composite application generator that implements the functions of developing alternative solutions, a human-computer interaction unit containing an intelligent interface that supports user interaction with the computing environment under conditions of heterogeneity of computing resources, uncertainty in the characteristics of the task and incompleteness source information.

The main disadvantage of information processing technology in the Grid system as applied to the monitoring object is the neglect of the uncertainty factor that determines the design process of monitoring systems based on the concept of soft computing [25] - [26] using intelligent technologies and high-performance information processing tools.

The properties of the monitoring object are characterized by significant a priori uncertainty, as well as environmental factors of its functioning, while most of the properties themselves and the factors that determine them cannot be directly observed, and the experimental information about them is incomplete and inaccurate. Of the many existing methods of soft computing, as shown by the results of semi-natural experiments using a mathematical model of the monitoring object and a number of parameter sensors characterizing the state of the monitoring object, the Bayesian intelligent measurements based on the Bayesian regulatory approach showed the best results.

Application of a traditional measurement methodology in such an information situation,

The technical result, to which the method of the claimed invention is directed, is to increase the reliability of determining structures that are in emergency or pre-emergency state, increase the accuracy of determining the state of controlled structures and the object as a whole by eliminating the systematic error of measuring devices and the possibility of an interrelated analysis of measurements from various devices, increased performance by reducing the array of processed information.

The essence of the claimed invention.

The claimed method for monitoring the technical condition of construction objects allows us to eliminate the above disadvantages by analytically processing the measurement results using soft calculation methods.

The properties of the monitoring object are characterized by significant a priori uncertainty, as well as environmental factors of its functioning; however, most of the properties themselves and the factors determining them cannot be directly observed, and the experimental information about them is incomplete and inaccurate. The application of the traditional measurement methodology in such an information situation, on the basis of which the measurement result can only be presented in the form of a numerical value and obtained only on the basis of experimental numerical information in compliance with the principles of measurement uniformity, gives poor quality results with an uncontrolled level of residual uncertainty.

To enable analytical processing of measurement results, a formalization of the subject area is necessary.

The monitoring object in this case is represented as a set of separate building elements (load-bearing structures) forming a set K.

Due to the fact that, as a rule, any load-bearing structure A _i is responsible, either a manufacturing defect during its manufacture or defects during construction work can be allowed in any design, and it is not economically feasible to control absolutely all structures of the object, then the determination of the optimal composition of structural elements and control parameters is the main task in designing a monitoring system, which is solved individually for each object based on the consideration of such various fact Orov as the location of the object, its responsibility, reliability of design decisions, financial constraints.

At the same time, factors such as location (climatic and engineering-geological conditions for finding the object) and the reliability of design decisions (the use of complex atypical design units, long-spans and consoles, unapproved design decisions and materials, etc.) determine potential threats, the implementation of which can entail the deterioration of the state of structural elements or their destruction.

Of the many building structures, a subset stands out, the elements of which are subject to control by a monitoring system (many controlled elements). For each of these elements, one or several parameters are determined, the changes in the values of which will make it possible to judge the change in the deformation state of the building element. This array of parameters forms a set, each of whose elements can be obtained on the basis of measurements taken by sensors of various physical nature. The resulting array of parameters is processed using the following method.

The continuous development of the methodological information-technological and technical bases of measuring systems has led to such a significant expansion of the scope of their application that, based on these tools, the tasks of assessing and controlling the properties of complex objects (SO), as well as managing them, have been successfully solved. Technological objects (technological processes, production systems and complexes, transmission networks of information, energy, material resources), as well as all natural phenomena, processes, and ecosystems can be classified as SOs. Information processes in such systems are implemented on the basis of a measuring approach, which assumes compliance with the principle of measurement uniformity at each measurement stage and continuous metrological support of intermediate and final results. The measuring approach is used for parametric and structural identification, production process management, image classification and processing, product quality assessment, as well as in monitoring ecosystems, their components (including socio-economic components) and environmental management.

As a rule, in practice, such tasks are accompanied by a complex information situation characterized by significant a priori uncertainty of knowledge about the properties of the controlled object and the influencing factors of its functioning environment, the impossibility of directly observing many of them, and the inaccuracy and incompleteness of experimental information about them, which distinguishes the cognitive function of the methodology for solving them as fundamental.

The application of the classical measurement methodology in such an information situation, on the basis of which the measurement result can only be presented in the form of a numerical value and obtained only on the basis of experimental numerical information, as well as the application of measurement information processing methods without observing the principles of measurement uniformity do not only solve these problems ineffective, but often and generally practically unsuitable due to their poor quality and uncontrolled level of residual indefinitely ty.

Therefore, the conclusion made in the works on the need to improve the methodological base of measuring systems in the direction of enhancing the role of the cognitive measurement function seems quite fair and relevant, which makes it necessary to obtain the results of such "generalized" measurements in the form of knowledge (analytical expressions for models, as well as conclusions and recommendations with their full metrological substantiation in the form of sets of quality indicators of these decisions) based on the entire volume of a priori and incoming in the process of measuring experiment data, including non-numeric.

The fulfillment of this requirement contributed to the involvement of apparatuses of theories of optimal solutions, artificial intelligence, fuzzy systems in the measuring medium. The desire to measure the non-quantitative properties of objects has led to the creation of a general (representative) measurement theory. Currently, work is underway to study and use the semantics of various types of measuring scales to increase the efficiency of measuring processes. On the other hand, measure theory and scaling theory are widely used in the modern theory of fuzzy sets. The concept of "measurement" is used in determining the membership function, the degree of fuzziness. The types of measuring scales that are most effective for implementing logical inference in decision systems are also determined. The list of such examples of the interpenetration of these methodologies can, of course, be continued.

As a result of such integration in the eighties of our century, the concept of intelligent measurements was formulated.

The terms computational intelligence and soft computing were introduced by L. 3ade in 1994. Then he formulated the main principle of soft computing - tolerance for inaccuracy and partial truth to achieve interpretability, flexibility and low cost of the solution.

Hard computing is based on accurate models, which include reasoning based on symbolic logic and classical methods of computing and searching for information. Soft calculations are based on approximate models, including approximate reasoning methods and computational methods based on functional approximation, random search and optimization.

Approximate reasoning methods that are part of soft computing are based on two main inference mechanisms - conditional inference and inference rule modus ponens.

The first mechanism (conditional conclusion) includes:

1. Probabilistic models - Nielson's probabilistic logic; Nguyen's fuzzy probabilistic logic, Pearl's probabilistic reasoning on Bayesian networks, subjective Bayesian methods.

2. Methods based on confidence functions — Demster-Scheifer theory, Smetz confidence functions, Fagin-Halpern upper and lower probabilities.

The second mechanism (modus ponens) includes multi-valued logics (algebras), fuzzy logic and theory of possibilities.

Computational methods based on functional approximation, random search and optimization, included in soft computing, are mainly divided into mechanisms of local search (neural networks) and global search (evolutionary computing).

Many approaches that are part of soft computing are universal, however, they complement each other well and are used in various combinations to create hybrid intelligent systems. Therefore, when creating systems that work with uncertainty, it is necessary to understand which of the constituent parts of soft computing or which combination of them are best suited to solve the problem.

Hybrid intelligent systems can conditionally be divided into the following classes:

1. Hybrid systems with functional substitution. They use one model, one of the elements of which is replaced by another model, for example:

a) recalculation of weights in the backpropagation procedure using the genetic algorithm. The number of iterations to obtain solutions is reduced;

b) selection of membership functions in a fuzzy controller using a genetic algorithm. Functions are much better than with manual selection.

2. Hybrid systems with interaction. Independent modules are used that exchange information and perform various functions in order to obtain a common solution. If the task is divided into pattern recognition, output and optimization, then these functions are taken over by neural networks, expert systems and genetic algorithms.

3. Polymorphic hybrid systems. One model is used to simulate the functioning of other models. So reasoning using a chain of rules can be modeled using a neural network (also genetic search).

In the direction of the development of the concept of smart measurements in the early 90s, a Bayesian smart metering methodology (BII) was developed based on the regularizing Bayesian approach (RBP). The BPO is a modification of the Bayesian approach to obtaining optimal solutions to these problems under conditions of significant a priori uncertainty, observing the principles of the uniformity of measurements in the process of forming a solution.

The concept of BII, based on the RBP, is a methodology for the synthesis of a new type of scales for the implementation of "generalized" measurements in order to achieve a high-quality solution to the applied problem on the basis of a comprehensive knowledge of the properties of CO and its functioning environment. Such scales are realized in the metric spaces of the dynamic compacts of their carriers and are called scales with dynamic constraints (SDO), which is due to their ability to adaptively change their structure in the process of accumulating information about the development of the properties of CO or its functioning environment. The process of solving the applied problem on the basis of the BII methodology is implemented as a process of purposeful transformation of the hierarchical structure of SDO, capable of adequately reflecting the properties of an evolving SD. The hierarchical SDO can reflect the properties of CO and the environment, geometrically representing a kind of hypercube of conditionally connected one-dimensional SDOs.

SDO results can be:

- numerical value of the parameter;

- analytical form of functional dependence;

- systems of analytical dependencies that determine the state of CO;

- linguistic meanings and expressions defining conclusions and decisions regarding the properties and their states for OM;

- to ensure the sustainable functioning of OM.

SDO BII for determining the state of CO or the degree of intensity of manifestation of its controlled property can be represented in the form of a three-link structure of SDO BII, which is a screen form of a management decision support system (SPPUER) that implements the methodology of BII. This SDO consists of a priori, current, and posterior SDOs and is used to convolve information flows that are diverse in the form of presentation of information, its content and degree of uncertainty.

SDO BII has in its structure a conjugated linguistic scale, the three-link carrier space of which is a hierarchical development of the known structure of the space of a fuzzy thermal set of a linguistic variable, for which the base set X is the carrier of the current BII scale of a controlled property of CO. At the same time, for the first time, it becomes possible to take into account linguistic information based on the formalistic principles of BPM in the process of obtaining a measurement result, which allows using additional linguistic information to reduce the degree of uncertainty and improve the quality of the measurement result. In addition, such a three-link structure of linguistic SDO allows you to integrate a priori and current linguistic information on the basis of the BII methodology, obtaining an a posteriori linguistic solution with a metrological justification of its quality, as well as identifying or adapting the form of membership functions, objectifying it by attracting additional information (and numerical) during measurement or processing.

A fuzzy dynamic model of the same parameter is obtained on a dynamic SDI BII. With its help, a number of analytical dependencies were obtained for possible alternative models of population dynamics, each of which is accompanied by its own set of metrological indicators, which include the accuracy of the model, its reliability and reliability (a posteriori Bayesian probability). The risk of the decision and the amount obtained during the measurement are also monitored. The complex of metrological characteristics, in addition to standardized indicators of the quality of measurements (such as accuracy of the result), includes the most important characteristics of the quality of solutions recommended in a number of works by well-known domestic and foreign scientists.

Thus, the basic principles of BII are:

- Integration of information of a variety of presentation forms in order to improve or achieve the required quality of the result;

- metrological substantiation of the obtained solutions in the form of quantitative indicators of the measure of posterior (residual) uncertainty, for example, indicators of accuracy, reliability, reliability;

- implementation of the principle of self-development of models of measurement objects and the environment of their functioning on the basis of adaptation of SDO structures to the properties of CO, known in the process of BII.

The formal record of the BII equation in the optimization form has the following form:

{h _{k, t} ⎟MX _{k, t} } = {argminC ^(B) [φ _{j, t} (x _{i, t} ⎟ {y _{i, t} })]}

h _k ∈H _{K, T} ; k = 1, K,

x _{i, t} ∈X _{I, T} ; t = 1, T

φ∈F _{J, T}

where {h _k } is the list of BII results, the reliability of each of which is defined as the posterior Bayesian probability; NK, T is the set of BII results, with variable boundaries and volume, which in the work will be called a dynamic set. C ^(B) is the Bayesian decision rule (BRI) for deciding on an algorithm from a dynamic set of algorithms Ф _{J, T} under a set of experimental data and measurement conditions y _{i, t} ∈ Y _{i, T} where

Y _{i, T} = {Ai, t} * {Mi, t} * {Oi, t}

Ai, t, Mi, t, Oi, t are dynamic sets of flows of a priori information, metrological requirements and restrictions, respectively * * is a convolution sign.

The choice of the Bayesian approach ideology as the conceptual basis of the BII is determined by the need to attract additional knowledge about the uncontrolled properties of CO to ensure the required quality of solutions with inaccurate, incomplete, and fuzzy a priori information, which is typical for CO-oriented tasks.

The results of such measurements should be accompanied at each stage of their determination by complexes of metrological characteristics, having the following structure:

MX _{k, t} {ζ _{k, t} ; V _{k, t} ; {P _{k, t} }}

where ζ _{k, t} is the accuracy of the obtained solution on the SDE, V _{k, t} is the reliability of the obtained solution on the SDE, determined by the level of errors of the 1st and 2nd kind {P _{k, t} } is a list of the uncertainty indicators, which include measure of reliability, measure of opportunity, measure of trust, etc.

Complexes of metrological characteristics are calculated on the basis of current indicators of the quality of the solution and a priori characteristics of the quality of the solution.

MX _{k, j} = MX ^(T) _{k, j} * MX _{k, t-1} ⎟y _{i, t}

where MX ^(T) _{k, j} is the complex of metrological characteristics of the solution obtained on the current SDS scale; MX _{k, t-1} - a complex of metrological characteristics of the a priori solution.

Currently, on the basis of the RBM methodology, a wide range of methods and tools have been developed for monitoring JI and supporting decision-making.

It has been analytically proven that BII are a generalization of known types of measurements, such as direct, indirect, aggregate, joint, adaptive, algorithmic, statistical and others, and add new advantages to the advantages of these types of measurements.

However, these advantages of BII are incomplete without such a flexible mechanism for managing the quality of solutions as customizable logics, which are widely used in various fuzzy inference systems.

The ability to quickly change the method of processing fuzzy knowledge (up to changing the method in the process of inference) gives the use of fuzzy logics. In fuzzy logic, reliability is represented as a fuzzy truth value (some arbitrary subjective value that does not have any statistical meaning, in contrast to probability). Fuzzy logic based on the triangular Zadeh norm (minimax) is most common in IP, and other T norms can also be used: probabilistic, Lukashevich, etc. A variety of T - norms gives a variety of fuzzy logics. If for evaluation we use a fuzzy linguistic scale, then in a dialogue with IP we can operate with linguistic concepts, not numbers. This linguistic fuzzy scale represents a set (of 5-9 objects), ordered by some selected criterion. When outputting on fuzzy scales, the result is approximated. It turns out the discrete logic defined on this set. The use of the transition table when outputting fully defines fuzzy logic and has the following advantages:

1) the process of fuzzy inference is faster;

2) the meaning of the transition tables is understandable to a person without a mathematical education;

3) the change in the method of recalculating uncertainty consists in changing one table of transition to another, therefore, such a change can occur at any time, including during the derivation process;

4) new transition tables can be created by a person using methods of cognitive graphics, the computer will provide verification of the fulfillment of properties - norms.

In further discussions, by fuzzy logic we mean logic, which uses the norm and the conorm satisfying a system of axioms as generalized conjunction and disjunction operators. In the output module of the expert system, such a logic (or a family of such logics) is used as a mechanism for managing uncertainty.

- the norm is a conjunction operator specified on the degrees of uncertainty of two or more conditions in the same product, which satisfies the following properties:

T (0,0) = 0

T (p, I = T (I, p) = p

T (p, q) = T (q, p)

T (p, q) <T (r, s), if p <r, q <s

T (p, (T (q, r) = T (T (p, q), r)

- the conorm S (p, q) calculates the degree of uncertainty of the conclusion drawn from two or more rules. This is a disjunction operator satisfying the following properties:

S (I, I) = I

S (p, 0) = T (0, p) = p

S (p, q) = S (q, p)

S (p, q) <S (r, s) if p <r, q <sS (p, (S (q, r)) = S (S (p, q), r)

- norm and - conorm are interconnected as follows:

( _P , q) = 1-S (1-p, 1-q)

With the help of - norms and - conorms, when choosing the appropriate negation, one can determine the implication and construct the derivation rule modus ponens, which is quite enough to calculate the degree of uncertainty of the conclusion depending on the degree of uncertainty of the premises and the rule itself.

The degree of uncertainty can be either a number from the interval [0,1], or a linguistic variable - a function ƒ (x): [0,1] → [0,1]. In this case, the number of values of fuzzy logic is limited by the number of terms of the linguistic scale. As the function ƒ (x), standard functions are widely used - fuzzy marks, trapezoidal functions, S-functions. Typically, the number of terms ranges from 5 to 9. According to experimental studies, significantly different results when deriving from a term-set of no more than 9 elements give pairs (T _i , S _i ) - Lukashevich’s logic, - probabilistic logic, (T _З , S _H ) - Zadeh's logic.

After performing any operation, linguistic approximation of the result by the elements of the scale is carried out. Since the number of inputs of any binary operation is finite, the result (after a linguistic operation) can be represented by a 9 × 9 matrix. Thus, each of the three logics will correspond to a set of matrices and the role of an expert will be to choose the number of this set.

With the help of - norms and - conorms, when choosing the appropriate negation, one can determine the implication and construct the derivation rule modus ponens, which is quite enough to calculate the degree of uncertainty of the conclusion depending on the degree of uncertainty of the premises and the rule itself.

Fuzzy linguistic meanings are represented in the system in the form of membership functions defined on the interval [0,1], where 0 is treated as a lie and 1 is treated as true. In order to reduce the number of calculations, the so-called trapezoidal functions are used as membership functions.

The conjunction and disjunction operators applied to these functions will produce as a result a new membership function for which the closest membership function from the existing set must be found. This operation is called approximation and is necessary to ensure that the term set is closed over all logical operations. The closest membership function is sought by the Euclidean metric in the space of mathematical expectations (1st dimension) and areas (2nd dimension).

After obtaining the result of a logical operation - applying - a norm or - a conorm or an implication to all pairwise values from a term-set according to the above rules and after approximation, we obtain matrices that store the results of operations and which can later be used instead of recalculating according to formulas.

The second way to manage uncertainty is to use fuzzy labels that change the degree of confidence in a fact.

There are two types of tags:

I. The “blurring” statement (increasing the degree of uncertainty) is (0, a), and ∈ [0,1].

II. The “concretizing” statement (reducing the degree of uncertainty) is (a, 1), and ∈ [0,1].

All operations can be reduced to arithmetic over the parameters a and b of a fuzzy label.

The system uses a set of fuzzy labels as a term set. For an already known thermal set, you can pre-calculate and store the results of these operations in matrices. By combining these operations, the confidence in the fact on the right side of the product is calculated.

The number of logics received by the system may depend on the following factors:

- the size of the matrix, and therefore the number of possible values of its elements;

- additional axioms that significantly narrow the set of generated matrices.

We introduce the definition of a triangular norm for the case of a discrete set of values.

Let A be an ordered set of such linguistic values = {A ₁ , A ₂ , ..., A _N }, where A ₁ ≤A _j if i <j, which means that the confidence expressed by A _{i is} less than the confidence expressed by A _j , which is in full agreement with the order relation on fuzzy numbers introduced above.

The following operations of negation, disjunction, and conjunction necessary for constructing fuzzy logic can be determined.

The operator N is a negation if it decreases monotonically, that is, N (A _i ) ≥N (A _j ), if I≤j for i, j = 1, ..., n н N [N (A _i )] = A _i for i = 1, n. The only function that has this property is N (Ai) = A _{n + 1-i} for i = 1, ..., n.

Conjunction and disjunction are related by the expression

T (A _i , A _j ) = N [S (NA _i ), N (A _j )}] (*)

- the norm is a conjunction operator specified on the degrees of uncertainty of two or more conditions in the same product, which satisfies the following properties:

ABSORPTION T (A _i , A _j ) = A ₁ for i = 1, ..., n

DOMINATION T (A _n , A _i ) = A _i , for i = 1, ..., n

SYMMETRICITY T (A _i , A _j ) = T (A _j , A _i ) for i = 1, ..., n, j = 1 ..., n (**),

MONOTONITY T (A _i , A _j ) ≥ (A _i-1 , A _j ) for i = 2, ..., n, j = 1, ..., n

ASSOCIATIVITY T [T (A _i , A _j ), A _k ] = T [A _i , T (A _j , A _k )] for I, j, k = 1 ..., n is the conorm S (A _i , A _j ) calculates the degree of uncertainty of the conclusion drawn from two or more rules. Its properties follow from (*) and (**).

But for large n this number is - norms are quite large. To further reduce their number, two additional axioms are introduced

STRICT T (A _j , A _i ) ≠ A ₁ for i, j ≠ 1

α-Uniformity If T (N _r, A -) = Ak, m (_ _b A _d Af = Ap (***) and T (Ai, Aj_i) = Aq , then k - a <p, q <k for and E [1, years].

This means that there cannot be a jump of more than a on the scale divisions between adjacent positions.

и n=9 при α=1 будет 34 матрицы и 3 класса, при α=2 будет 62 матрицы и 7 классов при, α=3 будет 65 матриц и 6 классов.In addition, since most tables differ only in a few cells, and this difference is not significant, one can introduce a partition into equivalence classes, for example, using the Hamming metric. Then, when doing

and n = 9 for α = 1 there will be 34 matrices and 3 classes, for α = 2 there will be 62 matrices and 7 classes for, α = 3 there will be 65 matrices and 6 classes.

с последующей классификацией.Such an approach allows the expert to present the entire possible set of fuzzy logics, and he will only choose the appropriate one. The second possible way is to fill an expert with only a few fixed positions with filling the remaining, taking into account axioms

with subsequent classification.

In cases where there are too many generated logics, you can specify the number of clusters into which the set of all received logics will be divided.

Custom logic can be selected in accordance with the quality of the resulting solution.

{l _t , | MX _{l, t} ) = argextrC ^(B) _{t, l} [φ ^(l) _{j, t} (MX _{l, t-1} ) (h _{k, t-1} ; x _{i, t-1} ; y _{i, t-1} ; l _t-1 )

where C ^(B) _{t, l} is the criterion for the choice of logic, φ ^(l) _{j, t} ∈ Ф ^(L) _{J, T} is the algorithm for processing metrological characteristics in order to select the optimal logic, l _t-1 is the logic for which the solution h _{k, t-1 is} obtained.

To use this mechanism, you need to move on to the implementation of BII on the spectrum of customizable logics.

Since customizable logics are the main element of the theory of fuzzy systems in the direction called soft computing, it is obviously right to call this new type of measurement soft measurements SM (soft measurements SM), moreover, if the criterion for decision making is a network, evolutionary calculations, including evolutionary strategies, evolutionary programs , genetic algorithms and genetic programming are special cases of the MI equation. Provided that they are implemented on the principles of measurement uniformity with a complete metrological justification of the obtained solutions, obviously, new types of soft measurements, such as fuzzy, neuro, evolutionary (genetic) measurements, can be obtained, provided that they are implemented on the principles of the uniformity of measurements with a complete metrological justification of the obtained solutions .

The direction of soft measurements is a branch of the direction of soft calculations, in which the information technologies of the components of soft calculations are implemented on the principles of the uniformity of measurements with full coverage of the metrological support of solutions by their chain. Then it is obvious that it is possible to create separate branches of soft measurements in the form of: fuzzy measuring systems

neural measuring networks (neuromeasurements)

evolutionary dimensions

where h _{k, t} ∈ H _{K, T} is a solution from a dynamic set of solutions to SDO, φ _jt ∈ Ф _{J, T} is a processing algorithm from a set of processing algorithms, x _{i, t} ∈X _{I, T} is an information stream from a dynamic set of information flows, y _{i, t} ∈ Y _{I, T} - a set of measurement conditions from a dynamic set of sets of measurement conditions.

The classification of structures that implement soft computing illustrates various combinations of components of this direction. Moreover, the resulting combinations of system components are called hybrid. The obvious drawback of such systems is the impossibility of integrating numerical information in the calculation process, the lack of metrological justification and the possibility of planning a strategy for implementing decision-making processes in order to ensure their required quality, and in a number of options (fuzzy systems) the lack of self-development based on self-learning. Structures implementing the equations of fuzzy measuring systems, neural measuring networks, and evolutionary measurements will obviously be free from these drawbacks.

The composition of several types of such measurements will produce hybrid types of soft measurements. Information systems built on such a methodological basis can be called fuzzy measuring systems, neural measuring networks, evolutionary measuring systems, and hybrid soft measurement systems (when they are composed).

In our opinion, the integration of soft computing systems with soft measurement systems can be carried out at several levels of integration, which will provide different depths of mutual penetration of methodologies.

At the macro level, hybrid structures realizing the composition of soft measurement tools and known soft computing components can be proposed. So, for example, the well-known structure of a fuzzy controller can be modified using the soft measurement methodology in various ways: applying the soft measurement technology to form the input matrix, replacing the neural network weight matrix with a hypercube, and extracting many control decisions. This creates additional opportunities for attracting numerical information, quality control of the management process and its optimal planning.

At an average level, when integrating methodologies, one of them is governing in relation to the other.

But the most effective is the integration of methodologies at the micro level, when the methodologies are combined at the level of implementation of elementary actions of the stages of the decision-making process. An example of such integration is the use of customizable logics at the level of SDO BII.

Obviously, such SDO can be called SDI BMI.

According to one of the leading experts in the field of artificial intelligence, D.A. Pospelova, the main directions of development of intelligent systems are determined by the following five global ideas, to which it is obviously advisable to add two more: the sixth - intellectual measurements and the seventh - soft measurements.

1. The rejection of rigid reasoning schemes based on a deductive procedure. Instead of closed formal systems that model subject areas, the laws of which are a priori known, the attention of the researcher will be directed to the study of quasi-axiomatic systems in which some of the axioms are replaceable, which makes it possible to carry out nonmonotonic reasoning and allows the use of plausible inference and plausible argumentation.

2. The widespread use of the idea of applied semiotics to build semantic knowledge bases (SBZ). Information units in SBZ, unlike ordinary knowledge bases, are signs, that is, structures that allow the three sides of any entity (name, concept, representation) to be displayed in a single way. This makes it possible to bridge the gap between left-sided representations and reasoning mechanisms based on symbols and gestalt patterns. Thus, the mechanisms of applied semiotics make it possible to take a fundamental step along the path of approximating cognitive structures in the knowledge base to those structures used by human thinking.

3. The development of means of cognitive graphics of a new approach not only to solving problems, but also to the search for the formulation of these tasks. Given the absence of any models in this area, it can be assumed that the theory of cognitive computing will become the central point in the development of work in the field of creative processes.

4. The ongoing development of robots poses a number of problems for researchers regarding the logic of action. The experience gained in this area shows that the logic of actions goes beyond formal systems to the field of semiotic modeling and multi-agent systems. The development of the logic of action should ultimately lead to the development of the mathematical theory of dynamic open systems.

5. The development of soft computing, that is, the complex development and use of computational methods based on fuzzy logic, neural computing, genetic computing and probabilistic computing, implemented in various combinations in hybrid intelligent systems.

6. Creation and development of the direction of intelligent measurements, as measurements based on the receipt and use of metrologically sound knowledge in the process of obtaining the result. In a regularizing Bayesian approach, such measurements are a branch of intelligent measurements called Bayesian measurements (BII). BII results are a type of fuzzy measurement results obtained on conjugate numerical and linguistic scales of a special structure, called scales with dynamic constraints (SDO). These types of measurements are focused on complex informational situations with significant a priori uncertainty of knowledge about the measurement object, its functioning environment and measurement tool, the use of different types of information, complete and continuous metrological substantiation of the results, compliance with the principles of measurement uniformity, self-development of measuring scales and technologies.

7. Creation and development of the direction of soft measurements - that is, fuzzy measurements, neuromeasurements, evolutionary (genetic) measurements and probabilistic measurements, implemented in various combinations in intelligent measuring systems based on soft measurements, or, in short, in soft measurement systems, the direction of soft measurements, generally coincides with the structure of the problem area for soft computing.

The choice of the most suitable method of approximate reasoning for the problem under consideration, not to mention the optimal structure of the intellectual system, remains a rather difficult problem. You can compare the methods of reasoning, using any pragmatic considerations, for example, efficiency, or simply experimentally, but until now there was practically no single formalism for presenting information within which these methods could be objectively compared. The absence of such a formalism for representing approximate reasoning methods and systems based on it, within the framework of soft computing, makes it especially relevant to create a single integrated approach to information processing to represent fuzzy, probabilistic methods and methods based on a confidence function. Such an approach naturally fits into the concept of “soft” measurements. An example of such a single integrated approach is the BII methodology and, in particular, the soft measurement methodology. The creation of decision support systems based on this formalism will allow the user to most adequately present their knowledge of the problem, first breaking it into relatively small fragments, and choose their own formalism for each of them, based on the restriction on the use of certain methods within the given fragment soft computing. One of the main problems in creating soft measurement technology is the development of a theory that unifies probabilistic, probabilistic, and fuzzy approaches to solving the problems of classifying complex statistical objects. The developed methods for the analysis of invariant fuzzy measures make it possible to effectively restore fuzzy measures of events in sigma-algebras with a small number of statistical observations. The proposed variational approach to constructing optimal measures of proximity of metrized relations allows us to find proximity measures with a priori given properties, in particular, smoothing errors of expert data, which is important when creating intelligent systems. In soft measurement systems, fuzzy-valued probabilistic logic can also be used.

Nedosekin D.D., Prokopchina S.V., Chernyavsky E.A. Information technology intellectualization of measuring processes. St. Petersburg: Energoatomizdat, 1995, 185 p.

Prokopchina S.V., Koinash B.V. Regularizing Bayesian approach to the classification of objects by image. - Preprint of the Academy of Sciences of the USSR, ILA, L., 1991, 61 p.

Prokopchina S.V. Development of methods and tools for Bayesian intellectualization of measurements in monitoring tasks of complex objects. - Abstract, St. Petersburg: TETU, 1995.33 s.

The implementation of the invention.

Based on the above analysis, a subset of E is allocated from the set of building structures K, the elements of which are subject to control by a monitoring system (many controlled elements).

For each element e _i from the set E, one or several parameters p _{i are} determined; changes in the values of which will allow us to judge the change in the deformation state of the building element. This array of parameters forms a set of P (a set of controlled parameters).

A plurality of monitored parameters P can be obtained by performing various measurements, while the same parameter p _i can be determined in various ways based on data from various measuring equipment. For example, a controlled parameter characterizing the slope of the foundation slab can be obtained using tilometers installed on the foundation slab and measuring the tilt angles at the installation point, and based on the readings of the geodetic equipment (total stations, levels) that measures the movements of the given points along which You can calculate the roll of the foundation plate.

The composition of the measuring equipment to obtain a set of P is determined individually for each object, depending on such factors as the required accuracy of measuring the parameter p _i , the possibility and convenience of the location of the corresponding equipment on the object, the cost of equipment, etc.

As a measuring equipment, any combination of various types of sensors that make up the set S can be used, such as vibration sensors, object vibration acceleration measurement sensors, object vibration velocity measurement sensors, object vibration amplitude measurement sensors, inclination measurement sensors, deflection measurement sensors, voltage measurement sensors , load measurement sensors, pressure measurement sensors, strain measurement sensors, absolute and / or uneven draft measurement sensors, control sensors cracks, joints and seams, sensors for measuring geodetic parameters, sensors for measuring climatic parameters. The specific composition of the sensors used is selected based on considerations of an economic nature as well as on considerations of ease of installation of the sensor and its maintenance during operation.

Based on the readings of the measuring equipment that makes up the set S, a set of V is formed from the physical parameters characterizing the state of the controlled objects.

Having many controlled elements E, the state of which is characterized by many parameters P, which in turn are determined on the basis of the measured values forming the set V, it is necessary to obtain an assessment of the technical condition of building structures and / or the whole object. Parameters P are estimated using the Bayesian soft measurement method, which allows one to solve problems associated with the uncertainty of both the monitoring object and the external environment, which are largely determining these parameters.

Information may be displayed on a computer monitor. In this case, the controlled elements (structural elements) can be displayed in any form convenient for the operator’s perception, in particular, in the form of elements of a three-dimensional model of a building or structure.

From the above list of sources used in compiling the description of the claimed invention, the popularity of the means on the basis of which the claimed device is built and the declared is carried out follows, which indicates the compliance of the claimed invention with the patentability condition “industrial applicability”.

List of sources used

1. RF patent №66525 for utility model "System for monitoring the technical condition of buildings and structures", IPC G01M 7/00, publ. 09/10/2007.

2. RF patent No. 82048 for utility model “Structured monitoring and control system for engineering systems of buildings and structures - SMIS”, IPC G05 17/02, publ. 04/10/2009.

3. RF patent No. 77429 for utility model “Device for dynamic studies of earthquake resistance of buildings and structures”, IPC G01M 7/00, publ. 10/20/2008.

4. RF patent No. 108602 for utility model “System for monitoring the technical condition of building structures”, IPC G01B 21/22, publ. 09/20/2011.

5. RF patent No. 123949 for utility model "System for monitoring changes in the state of load-bearing structures of buildings and structures", IPC G01M 7/00, publ. 01/10/2013.

6. RF patent №148119 for utility model “Device for monitoring cracks and joints of a building”, IPC E04G 23/00, publ. 07/15/2014.

7. RF patent No. 149873 for utility model “Device for determining the static and dynamic parameters of load-bearing structures of buildings and structures”, IPC G01M 7/00, publ. 01/20/2015.

8. RF patent №2245531 for the invention "A method for determining the stability of buildings and structures and a system for determining the stability of buildings and structures", IPC G01M 7/00, publ. January 27, 2005

9. RF patent No. 2284518 for the invention "A method for diagnosing the beginning of the destruction process in the structural elements of an object", IPC G01N 29/04, published on September 25, 2006.

10. RF patent №2327105 for the invention "Method for monitoring the state of a building or engineering construction and device for its implementation", IPC G01B 7/16, G01M 7/00, publ. June 20, 2008

11. RF patent No. 2357205, for the invention "System for determining the deformations of building structures", IPC G01B 11/16, publ. May 27, 2009

12. RF patent No. 2378457 for the invention "Method for monitoring a building under the influence of disturbances from the displacement of its foundation", IPC E02D 33/00, published on January 10, 2010

13. RF patent No. 2381470 for the invention "Method for monitoring and forecasting the technical condition of buildings and structures and a system for monitoring and forecasting the technical condition of buildings and structures", IPC G01M 7/00, publ. 02/10/2009.

14. RF patent No. 2395786 for the invention "Method for the diagnosis of design", IPC G01B 7/16, publ. 07/27/2010.

15. RF patent No. 2413193 for the invention “Method for monitoring the safety of load-bearing structures, structural elements of buildings and structures and a system for its implementation”, IPC G01M 7/00, publ. 02/27/2011.

16. RF patent No. 2448225 for the invention "System for monitoring the state of cracks and joints of buildings and structures", IPC E04G 23/00, publ. 04/20/2012.

17. RF patent No. 2460980 for the invention "Method for monitoring the technical condition of objects", IPC G01M 7/00, publ. 09/10/2012.

18. RF patent No. 2460981 for the invention "Method for monitoring and predicting the technical condition of objects", IPC G01M 7/00, publ. 09/10/2012.

19. RF patent No. 2461847 for the invention "Method for continuous monitoring of the physical condition of buildings or structures and a device for its implementation", IPC G01V 1/28, G01V 7/02, publ. 09/20/2012.

20. RF patent No. 2472129 for the invention "Monitoring system for the safe operation of buildings and civil engineering structures", IPC G01M 7/00, publ. 01/10/2013.

21. RF patent No. 2482445 for the invention "Device for monitoring the state of a building structure or civil engineering construction", IPC G01B 7/16, G01M 5/00, publ. 05/20/2013.

22. RF patent No. 2557343 for the invention, IPC G01M 7/00, publ. July 20, 2015

23. Patent of the Russian Federation No. 2557343 for the invention "Method for determining signs and localization of the place of change of the stress-strain state of buildings, structures", IPC G01M 7/00, publ. July 20, 2015

24. RF patent No. 2411574 for the invention "Intelligent system for high-performance data processing", IPC G06F 15/16 publ. from 02/10/2011.

25. Prokopchina S.V. Development of methods and tools for Bayesian intellectualization of measurements in monitoring tasks of complex objects. Abstract of St. Petersburg: TETU. 1995.

26. Nedosekin D.D., Prokopchina S.V., Chernyavsky E.A. Information technology intellectualization of measuring processes. SPb. Energoatomizdat, 1995.

27. Prokopchina S.V., Koinash B.V. Regularizing Bayesian approach to the classification of objects by image. - Preprint of the Academy of Sciences of the USSR, ILA, L., 1991.

Claims (1)

A method for monitoring the technical condition of building objects, including the selection of controlled elements (structures) of a building object, according to which they judge the state of the building object, recording the values of the measured parameters characterizing the state of the selected controlled elements, calculating the current values of the controlled parameters according to the measurement results and processing them using the soft method calculations, determining the states of controlled parameters by comparing the calculated values with horny values of the monitored parameters, identification of the state of the monitored elements and / or the construction site as a whole based on the selection of the worst state of the corresponding monitored parameters, visualization of monitoring information and evaluation results of the individual monitored elements and / or the building as a whole.