The article presents the results of developing a model for diagnosing a ship complex technical system with incomplete data and its implementation in an intelligent system for assessing the risk of failures of subsystems, components, intercomponent links, which allows obtaining a priori information about the technical condition of a complex system. Types of technical condition of subsystems, components, intercomponent connections are determined on the basis of diagnostic features of a complex system using the example of a ship power plant to assess the risk of their failures. Predicting the type of technical state of a complex technical system was carried out using a posteriori inference in Bayesian belief networks. The studies presented in the article assessed the risk of failures as a result of the use of an intelligent system for diagnosing and predicting the risk of failures of a ship complex technical system. The model for diagnosing and predicting the risk of failures of subsystems, components, interconnections can be considered as a conceptual model of an intelligent system for diagnosing and predicting the risk of failures of complex technical systems on network infrastructures, which has a relative insensitivity to incomplete technological data.
Complex technical systems (CTS), having structural and functional diversity, differ in the
principles of operation and consist of numerous interconnected and interdependent subsystems,
components with complex intersystem, intercomponent links [
Diagnostics and prediction of TС helps to reduce the CTS risk of failures of subsystems,
components, intersystem and intercomponent connections at their operation stage [
2023 Copyright for this paper by its authors.
The CTS reliability can be assessed by the results of diagnosing the TС, and the prediction of changes in the TС makes it possible to operate the systems until signs of a dangerous decrease in reliability appear, while excluding premature dismantling of components and assemblies, as well as performing other labor-intensive work that is often of dubious usefulness for the CTS reliable operation.
To successfully solve the problem of ensuring the ship CTS reliability it is necessary to remove a
number of uncertainties. Such uncertainties include incomplete data on external, internal impacts on
systems and on the state of such systems. The removal of uncertainties can be based on solving
problems of assessing the risk of CTS failures and its prediction with relative insensitivity to
incomplete technological data FS, FС, FIC and FI [
Traditional automation of ship equipment includes, in addition to monitoring parameters and
controlling installations and mechanisms, also equipping them with complexes that allow us to create
hierarchical distributed integrated systems for diagnosing and predicting complex systems TС
[
Diagnostic and forecasting systems should: constantly carry out diagnostics of the ship CTS TС of functional FS, FC, FIC and FI; analyze trends in changes in the technical equipment of systems; perform failover and provide TS prediction. To implement such a technology, appropriate algorithmic and software tools are required. The diagnostic algorithms used, as a rule, are based on the tolerance control of individual diagnostic parameters. However, the analysis and integral assessment of FS, FC, FIC and FI TC, the development of control actions in most cases is carried out by ship operators based on heuristic rules. At the same time, the volume of measuring and diagnostic information, the number of connections, dependencies of systems diagnostic TC features and types can be significant.
In theory, engineering practice, various methods are used to assess the risk of failures FS, FC, FIC
and FI CTC. An example of the application of risk theory is the logical development of probabilistic
analysis to assess CTS failures risk [
Intellectualization of automated diagnostic systems involves solving a number of interrelated tasks
of a structural, functional, informational and organizational nature, which should be provided at the
stage of designing diagnostic systems for TC CTS [
In artificial intelligence, knowledge representation models are actively developing - Bayesian
Belief Networks (BBN), used to diagnose the TC of complex systems [
Thus, the problems associated with ensuring the reliable operation of ship CTS require further improvement and the search for new methods, models and algorithms aimed at promptly detecting emergency conditions of equipment, at solving the problems of diagnosing and predicting system failures risk under conditions of relative insensitivity to incomplete data on FS, FC, FIC and FI. Since all modern ships must be equipped with automation systems for technical means using artificial intelligence technologies, the introduction of approaches based on such methods, models and algorithms should help ensure ship’s CTS reliable operation.
That is, taking into account the existing problems in ensuring reliability during the operation of CTS, the intellectualization of diagnostics and predicting ship’s CTS failures risk by diagnostic features is a significant direction that allows us to influence the safety and reliability of systems and is an urgent task.
Statement of the problem: to substantiate the forecast failures risk of FS, FC, FIC and FI by intellectualizing the assessment ship complex systems TC by diagnostic features.
The purpose of the work: ensuring the reliability and safety of ship CTS by reducing the risk of failures FS, FC, FIC and FI. 3. Model diagnostics technical condition intellectualization and prediction ship complex systems failures risk
In a formalized form, BBN for diagnosing the TC and predicting the risk of ship CTS failures contains an acyclic directed graph G, a set of vertex variables and directed links between them. A formalized generalized model for the intellectualization of TC diagnostics and predicting failures risk of FS, FC, FIC and FI by diagnostic features can be described as follows:
G, S (C), I S (IC ), RS (C) , RI S(IC ) , L , where: S (C ) ,set FS (FС);
I S (IC ) , set FIС (FI); RS (C) , RI S (IC ) , set of failures risk diagnostic assessments FS (FС) и FIС (FI) СТS;
L - mapping relationships between sets S (C), I S (IC ) и RS (C) , RIS (IC ) , based on the fault tree of the CTS diagnostic mode.
ТС FS (FС) and FIС (FI) sets with S (C), I S (IC ) is determined based on the failure tree (Failure
Tree), presented as a set failures risk of FS, FC, FIC and FI. The tree consists of failure risk
sequences, which are a multi-level graphological structure diagnostic model causal relationships,
obtained by tracking failures in the reverse order of the structure, in order to find occurrence their
possible causes. The advantage of a fault tree over other failure scoring models is that failure analysis
is limited to identifying only those FS, FC, FIC, and FI of the system and the events that lead to a
particular system failure or crash. The fault tree allows us to identify all the paths leading to the
failure of the CTS, and provides the determination of the minimum number of combinations of events
that cause the system to fail [
The set FS (FC) of the CTS, taking into account their hierarchical levels, is determined
S(C) = { nsm(cs)(c) | ns(c) = 1, NS (C) ; ms(c) = 1, M S (C)}, where ms(c) , each FS (FС) condition;
ns(c) ns(c) , FS (FС) number; ms(c) , hierarchical level number FS (FС); N S (C) , FS (FС) value;
The state of each FS and FC CTC is expressed as: mS(C) = {W 0 nS(C) S(C)n(m) ,W f S(C)n(m) ,ainS(C) n(m) ,аonS(C)n(m) }, where W 0 S(C) n(m) , full working capacity FS (FС); (1) (2) (3) W f
S (C)n(m) a inS(C)n(m) , а onS (C)n(m) in, on, sequence number of incoming and outgoing FIС, FI.
Operability of FS and FC at different degrees of their losses: , performance FS (FC) at different degrees of their losses;
, incoming and outgoing FS (FС); In (4) f =0 is the correct state of the STS, f=1 is the failure of the CTS The set of FIC and FI CTS is determined by:
W f S(C)n(m)
= {W fns(c) ,ms(c) | f = 0,1;ns(c) = 1, NS(C); ms(c) = 1, M S(C)},
IS(C) = { ISa(,Сb),z,q | a = 1, A;b = 1, B; z = 1, Z;q = 1,Q}, where a,b, z,q , state of each FIC (FI);
I S(С) a, FIС number; z, FI number; b, hierarchical level number FIС; q, hierarchical level number FI; A, FIС value; Z, FI value; B, hierarchical level value FIС; Q, hierarchical level value FI.
State of each FIC and FI: where W 0
IS(C)a(z) W f IS(C)a(z)
, operability FIC (FI) at different degrees of their loss.
Operability of FIC and FI at different degrees of their loss: a,b,z,q = {W 0
I S (С) I S (C)a(z) ;W f I S (C)a(z) }, , full working capacity FIС (FI); n(m) FS (FС) failure risk: a(z) FIС (FI) failure risk:
W f IS(C)a(z)
= {W fa,z | f = 0,1;a = 1, A; z = 1, Z;} RS (C)n(m)
= DS (C)n(m) PS (C)n(m) (t) R
IS (C )a( z) = D
IS (C )a( z) PIS (C )a( z) (t) (4) (5) (6) (7) (8) (9)
Modeling assumptions and limitations are that FS and FC CTSs can have a failure risk level
distributed based on the Harrington desirability function [
(t) = exp(−T0 ) = exp(−T0 )
T0 = P(t)dt = 1/
0 PS (С)n(m) = vS (C) n(m) where – distribution parameter, taken according to the test results equal to time to failure estimate.
Mean time to failure:
The failure probability FS and FC CTS, taking into account the number of failures of a certain subsystem (component), can also be determined: where PS (C) n(m) , failure probability n(m) FS (FС); vS (C)n(m) , number of failures n(m) FS (FС); = 10 6 hours - the period of statistical testing.
The probability of failure FIC and FI CTS, taking into account the number of failures of a certain intersystem (intercomponent) connection, can be determined:
An available source for the reliability statistics FS, FC, FIC and FI of ship CTS when choosing the
values of the conditional probabilities of their failures to determine failures risk is the OREDA
offshore database [
PI S (C)a(z) v
IS (C)a(z) where , conditional failure probability a(z) FIС and FI; , number of failures a(z) FIС and FI.
Quantitative assessment of damage FS, FC from failure of a subsystem (component) to determine the risk of failure:
DS (C )n(m) =
Nun(m) (Cn(m) + Cin(m) + Cdn(m) ) ;
Nfn(m) where
С n(m) , FS (FС) price; Сi n(m) , FS (FС) installation cost; Сd n(m) , FS (FС) disposal cost; Nf n(m) , number of failures FS (FС); ds(c)n(m) , damage from abandonment FS (FС).
Quantification of FIC and FI damage from failure a(z) FIС and FI:
DIS (C)a(z) =
Nua(z) (Ca(z) + Cia(z) + Cda(z) ) ;
Nfa(z) (15) where
Nu a( z) , number of unrecoverable failures FIС (FI); С a( z) , price FIС (FI); Сi n(m) , installation cost FIС (FI); Сd a(z) , disposal cost FIС (FI); Nf a( z) , number of failures FIС (FI); d
, damage from abandonment FIС (FI). The verbal form is used to describe the category of
is(c)a( z)
damage from refusal FS (FC). To compare numerical estimates for different classes of damage, the
Harrington scale is used [
Based on the generated fault tree, the general structure of the SPP is formed, which makes it possible to formalize the model for diagnostics. Symbols of subsystems, components of the SPP in BBN: Input element - IE; Fire fighting system - FFS; Compressed air system - CAS; Manual control of the main engine - MCME; Control system - CS; Remote automated control system of the main engine - RACSME; Intermediate component - P1; Ship power plant - SPP; Main engine - ME; Ballast drainage system - BDS; Emergency drive propulsion and steering complex - ED PSC; Control system for propulsion and steering complex -CSPSC; Boiler plant - BP; Transfer of power from the main engine to the propeller - TPMEP; Intermediate component - P2; Propulsion and steering complex PSC; Output component - EXIT. Tab. 1 reflects the correspondence of symbols on the fault tree S and FS, FС BBN.
Event characteristics
Violation of the IE element Violation of the FFS, CAS, MCME elements Violation of the RACSME, P1, SPP elements
Violation of the CS, BDS, BP elements Violation of the ME, ED_PSC, CSPSC elements
Violation of the TPMEP, P2, PSC elements
The structure of the BBN SPP, shown in Fig. 2, is a multi-level subsystem location system, consisting of 13 subsystems, 7 levels with the addition of specialized intermediate nodes P1 and P2, providing the implementation of a multi-level network structure.
For the subsystems of the upper level SPP BBN structure, conditional failure probabilities are specified, taking into account the influence of subsystems of a lower hierarchical level on subsystems of a higher level.
As an BBN example for interconnected power plant units (Fig. 2) IE, CAS, SPP and interconnections IE-CAS, CAS - SPP, sets of failure risk at the initial time and taking into account the dynamics of technical conditions over time based on a priori data on intensities bounce
R(Work11,,32,,83 )t =0 = 0 ; R( Not _ work11,,32,,83 )t =0 = 1 ;
R(Work I2E,3−CAS ,CAS −SPP )t =0 = 0 ; R( Not _ work I2E,3−CAS,CAS−SPP )t =0 = 1 ;
R((Work11,,32,,83 )t /(Work11,,32,,83 )t −1) = 0,1 ;
R((WorkI2E,3_ CAS,CAS _ SPP )t /(WorkI2E,3_ CAS,CAS−SPP )t −1) = 0,1 • the risk failure level is assessed as acceptable, the consequences of the accident are insignificant at:
R((Not _ work11,,32,,83)t /(Work11,,32,,83)t −1) = 0,1− 0,2 ; R((Not _ workI2E,3_ CAS,CAS _ SPP )t /(Work1IE,3_ CAS,CAS −SPP )t −1) = 0,1 − 0,2 ;
R((Not _ work11,,32,,83)t /(Work11,,32,,83)t −1) = 0,2 − 0,37 ;
R((Not _ workI2E,3_ CAS,CAS _ SPP )t /(Work1IE,3_ CAS,CAS −SPP )t −1) = 0,2 − 0,37 ;
Sets of risk of failures at the current moment of time, taking into account the previous state of subsystems and intersystem communications, can be within: • the level of risk of failure is estimated as minimal, the consequences of an accident are minimal at:
The construction and study of BBN failure risk assessments FS, FC, FIC and FI CTS was carried
out using the software product GеNIe [
The use of the GеNIe environment makes it possible to diagnose the TC of each FS, FC, FIC and FI CTS. Perform a regression analysis of the influence of network each parent element on its corresponding child element. Implement a graphical display of the failures risk assessment predicting • the risk failure level is estimated as maximum, the consequences of the accident are significant at:
R((Not _ work11,,32,,83)t /(Work11,,32,,83)t −1) = 0,37 − 0,63 ;
R((Not _ workI2E,3_ CAS,CAS _ SPP )t /(WorkI2E,3_ CAS,CAS −SPP )t −1) = 0,37 − 0,63 ; • the failure risk level is assessed as critical at:
R((Not _ work11,,32,,83)t /(Work11,,32,,83)t −1) = 0,63 −1 ; R((Not _ workI2E,3_ CAS,CAS _ SPP )t /(WorkI2E,3_ CAS,CAS −SPP )t −1) = 0,63 − 1 (16) (17) (18) (19) (20) results FS, FC, FIC and FI CTC. Calculate the values of the working capacity loss probability, damage from failures and assessments of failures risk FS, FC, FIC and FI CTC.
When modeling the SPP’s BBN (Fig. 2), for various failure risk values the input component, the failures risk values of functionally interconnected and interacting FS for 20,000 hours SPP operation were determined (Fig. 3).
The operating state and failure, for example, of the CS subsystem for the risk of failure at the input element of the SPP 0.26 when simulating the BBN of the SPP is shown in Fig. 4.
The purpose of using BBN in assessing the risk of failures FS, FC CTC is a posteriori conclusion. The a priori data are dynamically recalculated and form a posterior failure risk estimate, which is a priori information, to process the new information.
Post hoc inference is based on procedures for analyzing data obtained from the use of BBN.
When implementing this approach in research, modeling on a priori and a posteriori data, the predicted TС FS, FC of the power plant are determined, which have the greatest impact on the performance of the main engine and the operation of the entire system for various time periods. It follows from the research results that the predicted maximum non-operating state during the operation of the SPP is 20,000 hours. Corresponds to the most vulnerable subsystems ME, VR (Fig. 2).
Because subsystems ME, VR are dependent at the level of the hierarchical structure of the SPP, therefore, in the future, it is necessary to regularly check the subsystems in order to find possible causes of their failure, thereby increasing operation reliability of ME, VR, and hence the whole SPP at all.
Thus, based on the intellectualization TC assessment FS, FC, FIC and FI of the CTS by diagnostic features, it is possible to substantiate the forecast failures risk FS, FC, FIC and FI of the SPP.
The considered principle intelligent system functioning, its structure, in terms of the technical and technological foundations of construction on the example of SPP, reflected in the method and model for assessing and predicting FS, FC, FIC and FI CTS failures risk can be considered as conceptual task.
The described method, the developed model of an intelligent system for assessing and predicting the risk of CTS failures on network infrastructures, as a research result, confirmed the relative insensitivity to incomplete technological data FS, FC, FIC and FI.
Application of research results allows providing: • formation of principles for the intelligent system for diagnosing and predicting the CTS failures risk construction and operation; • intellectualization model of TC estimation and forecasting ship CTS failures risk by diagnostic features, which has a relative insensitivity to incomplete technological data FS, FC, FIC and FI CTS creation; • intellectualization model for ES evaluation based on the use of a priori information about failures, linking the types of TC FS, FC, FIC and FI of complex systems and their diagnostic features in the failure risk form creation; • identifying the most vulnerable FS, FC, FIC and FI CTS and solving the problem of determining the failures causes depending on failures risk in the TC diagnostics.
The results of the development of a diagnostic model for a complex technical system with incomplete technological data and its implementation in an intelligent system for assessing and predicting FS, FC, FIC and FI ship CTS failures risk made it possible to obtain a priori information about the technical condition of a complex system.
The types of technical condition FS, FC, FIC and FI are determined on the basis of diagnostic features of a complex system using ship power plant example.
Predicting complex technical system technical state type was carried out using a posteriori inference in Bayesian belief networks.
The conducted studies presented in the article evaluated the results of functioning of an intelligent system for diagnosing and predicting complex technical system failures risk, which makes it possible to identify the most vulnerable FS, FC, FIC and FI CTS and predict their TC.
The model for diagnosing and predicting the risk of failures of subsystems, components, interconnections can be considered as an intelligent system conceptual model for diagnosing and predicting complex technical systems failures risk on network infrastructures, which has a relative insensitivity to incomplete technological data.
The use of the developed method and model, taking into account the hierarchical levels FS, FC, FIC and FI, when searching for the causes of failures in complex technical systems, allows us to control failures risk in systems when information about failures in their structures is received according to TC. The application of the method and model allows predicting trends in the risk of system failures, taking into account changes in individual FS failures risk of FC, FIC and FI in order to further choose a strategy for their restoration or replacement.