CN107991870A - A kind of fault pre-alarming and life-span prediction method of Escalator equipment - Google Patents

A kind of fault pre-alarming and life-span prediction method of Escalator equipment Download PDF

Info

Publication number
CN107991870A
CN107991870A CN201711263968.3A CN201711263968A CN107991870A CN 107991870 A CN107991870 A CN 107991870A CN 201711263968 A CN201711263968 A CN 201711263968A CN 107991870 A CN107991870 A CN 107991870A
Authority
CN
China
Prior art keywords
mrow
msub
data
equipment
tables
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201711263968.3A
Other languages
Chinese (zh)
Other versions
CN107991870B (en
Inventor
张新征
郭乾
刘新东
周曙
周政昊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jinan University
University of Jinan
Original Assignee
Jinan University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jinan University filed Critical Jinan University
Priority to CN201711263968.3A priority Critical patent/CN107991870B/en
Publication of CN107991870A publication Critical patent/CN107991870A/en
Application granted granted Critical
Publication of CN107991870B publication Critical patent/CN107991870B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05BCONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance

Abstract

The invention discloses a kind of fault pre-alarming and life-span prediction method of Escalator equipment, the health status that this method passes through Escalator modules state computation Escalator system, by Escalator health status compared with abnormal elevator degree time match function, obtain Escalator run time, Escalator run time is corresponded to according to Escalator fault threshold, you can estimation apparatus remaining life.Various algorithms employed in the inventive method, it is to be analyzed and designed based on equipment index, Escalator index in algorithm is changed to other equipment index, gather corresponding detection device real-time indicators data, it can obtain the health evaluating situation and remaining life of institute's detection device, therefore, this algorithmic system is suitable for the prediction of each key equipment, has versatility.

Description

A kind of fault pre-alarming and life-span prediction method of Escalator equipment
Technical field
The present invention relates to the acquisition of fault characteristic information and the processing of information uncertainty and mass data analyzing and processing skill Art field, and in particular to a kind of fault pre-alarming and life-span prediction method of Escalator equipment.
Background technology
The possibility omen that equipment fault early-warning and life prediction are obtained according to equipment moving law or observation, can set For before really breaking down, promptly and accurately the unusual condition of HERALD equipment, takes appropriate measures, and farthest reduces setting Loss caused by standby failure, ensures the safety and steady of the operational process of equipment.Therefore reliable condition monitoring technology has in time The monitoring of effect and diagnosis process exception just seem particularly urgent.
Existing equipment fault early-warning technology is broadly divided into three major types:Method, Knowledge based engineering side based on mechanism model Method and the method based on data-driven.
Method based on mechanism model mainly includes two stages:Produce residual error to assess with residual error, the former with equipment by transporting The difference that the mathematical model estimating system that row mechanism is established is exported between actual measured value show that the latter analyzes according to the former and sets It is standby whether to break down.Such method is combined closely with control theory, and most mechanism models are all linear system, therefore is worked as and faced Non-linear, the free degree is higher and during the complication system of Multivariable Coupling, can not detect the failure of equipment well, also Huge cost is paid to establish model, the limitation of various environment is also all so that this method monitoring effect is bad, it is impossible to extensively Using.
Knowledge based engineering method, it is desirable to which there is a large amount of knowledge and experience, complete database, the inspiration according to associated specialist Sex experience, connection relation, fault propagation pattern in automatic describing monitoring process between each unit etc..For single system, property Can preferably, for complicated system, then can be due to not complete database, enough experience inference deduction failure processes and band Carry out various problems, so versatility is poor.
Method based on data-driven relies on smart instrumentation and computer memory technical, and mass data is excavated, and finds Internal information resume number in process data, is monitored data so as to establish model, judges the malfunction of equipment.Rely on Intelligent equipment well cannot lay down a definition the situation inside equipment the analysis minings of data, and machine It is relatively narrow to practise the algorithm application surface of data-driven, technology is weak, can not be widely used in the fault pre-alarming of equipment.
In recent years, do not turn off and send new life-span prediction method, having considerable method to have been used to reality also has Still in experiment forecast period.At present, life prediction mainly has two methods, indirectly and directly measures two kinds.
Indirect life-span prediction method is based on the supplemental characteristic of component, the degree of injury of calculating unit, dependent on component Complete during operation, real data, have ignored the factor of material aging.
Direct life-span prediction method has nondestructive test method and destructive testing method, and destructive testing method needs to obtain identical Or similar sample, the data needed by destructive testing, analyze data, and projected life degree of injury, does Go out life appraisal.Nondestructive test method, in the shorter time, can diagnose more position, and energy regular monitoring, uses Narrow.
The research of these types of method mechanism is ripe, and the development and application of device are feasible, but the essences of life prediction Degree needs to further improve, and application range is also up for improving.
The content of the invention
The purpose of the invention is to can pre- measurement equipment whether the remaining life of exception and equipment, prevent equipment Failure, assures the safety for life and property of the people, there is provided a kind of equipment fault early-warning and life prediction system approach, for each The different equipment of kind of building block, in order to which whether pre- measurement equipment breaks down and remaining life, by calculating each portion of equipment Whether the probability of malfunction of part breaks down judging equipment and the remaining life of pre- measurement equipment.
The purpose of the present invention can be reached by adopting the following technical scheme that:
A kind of fault pre-alarming and life-span prediction method of Escalator equipment, the fault pre-alarming and life-span prediction method bag Include the following steps:
S1, using SQL statement, establish database relational table, be stored in oracle database, Matlab passes through ODBC bridges Connect oracle database and access data, complete database sharing;
An important factor for S2, analysis elevator basic structure, determine the critical component for causing elevator faults, and screening influences elevator Analogue simulation is carried out as the index for judging elevator faults, and to achievement data;
S3, collection elevator real time data, structure dynamic bayesian network (DBN), and by Monte carlo algorithm to dynamic The each node of Bayesian network, which solves, estimates probability of malfunction, forms CPT probabilities of malfunction, early warning is made to the failure of elevator;
Elevator achievement data in S4, calling oracle database, generates the sequence of abnormal elevator degree and run time, obtains To the abnormality degree function of time, by the abnormality degree function of time determine elevator run time compared with abnormality degree time threshold, So as to fulfill Escalator life prediction algorithm.
Further, the step S1 includes:
S101, using SQL statement, with reference to pivot, decode in Oracle, programming apparatus health evaluating table ELEIDX, Fault diagnosis table FAULTDB, data point table ELE forms, a series of relation tables are built up by original point of interface table select, storage In oracle database;
S102, base data and fault type according to equipment and part of appliance, using Devicetype tables, Index tables, Indexdb tables, Basicdb tables, Basicdb tables structure data system, using hierarchical structure database design health evaluating with Fault Diagnosis Database part, wherein, Devicetype tables storage device type and part of appliance and its corresponding numbering, Index tables storage facilities refines index species, and the record of Indexdb tokens is in real time and recent index, Basicdb tables are used to record reality When and the base data values collected in the recent period, Basic tables be used to express each base data classification and between each component and component index Contact;
S103, using equipment fault evaluation index table Elec_Index and base data table Elec_Basic structure health comment Estimate database, there are overlying relation between two tables, data are according to basic number in equipment fault evaluation index table Elec_Index Calculated according to table Elec_Basic, the renewal of base data table causes the renewal of index table data, and two tables its capacity be can Add;
S104, utilize equipment fault diagnosis table Faultdb and base data table Elec_Basic structure fault diagnosis datas Storehouse, there are overlying relation between two tables, data are according to base data table Elec_Basic in equipment fault diagnosis table Faultdb Calculate, the renewal of base data table causes the renewal of fault diagnosis table data, and two table capacity can add;
The client software of Oracle is installed, correctly under the client directory of configuration Oracle on S105, client machine Tnsname.ora files, there is provided early warning system access address and access port, start the oracle listener of server-side;
S106, database and Matlab connect the data inquired about using exec in database, fetch functions by ODBC bridgings Upper strata variable is returned to, update updates the data health evaluating table and fault diagnosis table in storehouse.
Further, the step S2 includes:
S201, analysis elevator configuration tractive driving, suspension arrangement, car frame and carriage, door system, weight balancing system, Electric drive system, electrical system and safety system, design objective;
S202, call database in Devicetype tables, Index tables and Basicdb tables, by the elevator data collected with Offline mode carries out big data analysis, forms Basic tables and Indexdb tables deposit database;
S203, with reference to the gloomy TE-evolution1 elevator structures of the base of a fruit, research object is divided into traction machine, steel wire rope and door system System, and propose corresponding index for each research object, and determine that fault diagnosis index is the change gear degree of wear in door system And wirerope diameter;
S204, for change gear degree of wear index, quote in IBM algorithms and survey wear model, exceed original for predicting The abrasion of beginning surface roughness depth;
S205, for wirerope diameter index, using finite element theory, the mutation analysis of calculated diameter.
Further, the step S3 includes:
S301, Data Collection, from database call include device history achievement data Indexdb tables, Devicetype tables and Index tables;
S302, determine network node variable and causality, determines that network node variable and causality become including node Measure the selection of the measurement, types of variables (discrete and continuous) of name, the description of each part of system and amount of degradation;
S303, structure elevator DBN model, analyze each component coupled relation of Escalator, with reference to electric staircase failure diagnosis index Extraction module, builds Escalator Bayesian network fault diagnosis model figure, uses matrix-style to represent Bayesian network in BNT, Even node i has an arc to j, then (i, j) value is 1 in homography, is otherwise 0, establishes Bayesian network BNT, and by sample Learn_params () function is substituted into as training set to be learnt, study obtains conditional probability table CPT;
S304, the probability for estimating according to trained Bayesian network unit failure, result of calculation deposit database is good for In health condition evaluation table and fault diagnosis table.
Further, the step S4 includes:
S401, the CPT tables by Bayes's fault diagnosis model and network node, from database facility health Evaluation Table extraction equipment Real-time Monitoring Data, assesses equipment health status, output equipment abnormality degree, i.e. probability of malfunction;
Historical data in S402, calling health evaluating table, generates unit exception degree time series, is intended with Matlab curves Close tool box CFTOOL and determine fitting function exponent number, obtain unit exception degree and the fitting function of run time;
S403, call unit exception degree and the fitting function of run time, sets the corresponding abnormality degree threshold value of equipment scrapping A, corresponding service life are x, and the abnormality degree b of equipment at this time is obtained to system real time fail probability analysis using expert system, According to the fitting function of unit exception degree and run time, the remaining life w of equipment at this time usage time y, then equipment are calculated =x-y, realizes and system lifetim is predicted.
Further, the step S204 includes:
S2041, the basic assumption that wear model can be surveyed,
It is ENERGY E that abrasion that each number of pass times produces is consumed and number of pass times N the two variables according to wear extent Function this it is assumed that the relation between wear extent Q and number of pass times N and the ENERGY E consumed is the differential equation
Wherein, wear extent and the relation of number of pass times are
In formula, m is the given constant of system, when unlubricated,
S2042, according to the energy of consumption and the relation of wear process, the abrasion of A types can be divided into and worn with Type B, wherein, A types Abrasion energy of rupture in wear process remains unchanged, and belongs to heavy wear, and applied to dry friction and case of heavy load, Type B abrasion exists Destroy as number of pass times changes and change in wear process, be moderate transfer as a result, for the abrasion of door change gear, abrasion Represented with the following differential equation
In above formula, C is the constant coefficient of system, and the wear extent being determined by experiment under a certain number of pass times tries to achieve or by zero Abrasion gets up to try to achieve the analytic expression of C with the models coupling that can survey abrasion, and N is number of pass times, and S is length of the contact surface along glide direction Degree, Q is polishing scratch cross-sectional area,
It is assumed that destroy work(and τmxS is directly proportional, and the contact area of two articles changes with abrasion, therefore τmxS is not normal for one Amount,
Further, the step S205 is specific as follows:
Vernier caliper measurement of the diameter of steel wire rope with jaw, the width minimum of jaw will be enough adjacent across two Stock, measure and carried out on the straight position outside steel wire rope end head 15m, at a distance of at least in the two sections of 1m, and at same section Face measures two values, the measured diameter of the average value of four measurement results, as steel wire rope orthogonally.
Further, it is specific as follows that elevator DBN model is built in the step S303:
S3031, by fixed node and priori data build prior Bayesian network;
S3032, call data to carry out parameter learning to DBN networks, builds posterior Bayesian network;
S3033, calculate its probability of malfunction to each node in DBN model using Monte carlo algorithm, determines all-network The CPT probability distribution tables of node.
Further, the parameter learning is to be carried out using the EM algorithms learn_params_em of successive ignition optimizing Study, wherein, the process of EM Algorithm for Solving optimum structures is to converge to local optimum parameterProcess, occurred by random number Device produces sample training collection, fills up loss data, and data are carried out with the parameter that maximal possibility estimation learning simulation best suits structure, Comprise the following steps that:
E is walked, that is, it is expected:
Wherein E is desired value;D is training sample set;Represent the optimum structure parameter found, XiCodomain be:qiIt is configuration πiPut in order;NijkIt is to meet variate-value in data set DAnd πiThe bar of=j Part frequency:ylIt is the data amount check lost in D;S hIt is that bayesian network structure selection is assumed;
M is walked, i.e. maximum estimated, maximal possibility estimation function:
MAP estimation function:
N′ijkIt is the abundant statistical factors of priori, NijkIt is the abundant statistical factors of sample data,i,j,k,h,q∈N。
Further, the parameter learning is to be learnt using the SEM algorithms of successive ignition optimizing, wherein, SEM The specific implementation procedure of algorithm is as follows:The initial value of variable is set;The iteration since some initial Bayesian network;Call connection Reasoning computing of the tree reasoning algorithm completion to Bayesian network is closed, draws current optimum network;Utilized based on current optimum network EM algorithms carry out completion to data set, obtain complete data set to realize the maximization of parameter;All nodes in calculating network The summation of value number;The network structure different from initial network is created, using these structures as candidate network structure;Use BIC Score function gives a mark the network structure of above-mentioned candidate, and the ginseng so that score maximum is found from selected network structure Number;Go out the best network structure with data set fitting.
The present invention is had the following advantages relative to the prior art and effect:
1st, the present invention according to data acquisition real-time update data and can carry out data processing, draw the probability of malfunction of elevator with Trouble unit during elevator faults, and life prediction is carried out according to real time data, can be in first time by the health status of elevator Reality gives staff.
2nd, the present invention combines IBM algorithms and finite element theory algorithm during fault signature index is established, effectively The efficiency of data mining and the precision of data mining are improved, so that index is more reasonable.
3rd, in electric staircase failure diagnosis calculating, by Bayesian network intelligent algorithm and electric staircase failure diagnosis index extraction Technology is combined, and builds the Bayesian network fault diagnosis model figure of elevator, and combines EM algorithms and SEM algorithms etc., by data Bring into program and be trained, obtain conditional probability table, current elevator health status can be assessed.To improve Bayesian network The arithmetic speed of network, introduces the approximate resoning algorithm and importance sampling method of random sampling.
Brief description of the drawings
Fig. 1 is the equipment fault early-warning disclosed in the present invention and the basic framework frame of life prediction system approach principle Structure;
Fig. 2 is the process step figure of elevator faults early warning and life-span prediction method;
Fig. 3 is base data table form schematic diagram;
Fig. 4 is fault diagnosis sheet format schematic diagram;
Fig. 5 is health evaluating sheet format schematic diagram;
Fig. 6 is device databases graph of a relation;
Fig. 7 is assessment and diagnostic data base schematic diagram;
Fig. 8 is TE-evolution1 elevator primary structure block diagrams;
Fig. 9 is wirerope diameter instrumentation plan;
Figure 10 is elevator Bayesian network fault diagnosis model figure.
Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention In attached drawing, the technical solution in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is Part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art All other embodiments obtained without making creative work, belong to the scope of protection of the invention.
Embodiment
Present embodiment discloses a kind of equipment fault early-warning and life prediction system approach, its structure as shown in Figure 1, Figure 2, All steps of the present embodiment are all based on what Matlab development environments and Orcle databases were completed, and this method specifically includes The following steps:
S1, with SQL statement, establish database relational table, be stored in oracle database, Matlab passes through ODBC bridgings Connect oracle database and access data, complete database sharing;
The step is specific as follows:
S101, such as Fig. 3, Fig. 4, Fig. 5, require according to scheme, using SQL statement programming apparatus health evaluating table ELEIDX, Fault diagnosis table FAULTDB, data point table ELE forms, a series of relation tables are built up by original point of interface table select, storage It is used for the equipment real time data for storing collection in oracle database;
S102, such as Fig. 6, data system is built according to the base data and fault type of equipment and part of appliance using 5 tables, Devicetype tables storage device type and part of appliance and its corresponding numbering;Index tables storage facilities refines index species; Indexdb tables then record real-time and recent index;Basicdb tables are used to record base data values that are real-time and collecting in the recent period; Basic tables are used to express each base data classification and contacting between each component and component index;
For health evaluating and Fault Diagnosis Database part, using the design of database (such as Fig. 7) of hierarchical structure.
S103, such as table 1, table 2, health evaluating data base manipulation equipment fault evaluation index table (Elec_Index), basis Tables of data (Elec_Basic) is built, the overlying relation between two tables, and the data in assessment of failure index table are according to basis Tables of data calculates, and the renewal of underlying table data can cause the renewal of index table data, its capacity of two tables can add;
1. health evaluating table ELEIDX of table
FN Brake power exception percentage Brake input voltage exception percentage Traction machine working temperature abnormity Wirerope diameter abnormality degree Door lock relay status intensity of anomaly
FVA LU E66 0.1 0.1 0 0 0
FVA LU E67 0 0 0.5 0 0
FVA LU E68 0 0 0.3 1 0
FVA LU E69 0.2 0 0 0.1 0
FVA LU E70 0.1 0 0.2 0 0
FVA LU E71 0.1 0 0 0 0
FVA LU E72 0 0 0.5 0 0
FVA LU E73 0 0 0 0 1
FVA LU E74 0 0 0 0 0
FVA LU E75 0.3 0 0 0 1
FVA LU E76 0 0.3 0.2 0 0
FVA LU E77 0.1 0 0 0.4 0
FVA LU E78 0.2 0 0.1 0.1 0
FVA LU E79 0.1 0 0.5 0 0
FVA LU E80
2. basic data point table ELE of table
Data point describes EQ UTPM EN TLO CATD N equipment place PO IN TT YPE FVALUE VALUEDESCRIPTION Numbers illustrated
Elevator is normally applauded DI 1 0-32767 0-32767
The voltage of insulation resistance AI 220 0-32767V 0-32767V
The electric current of insulation resistance AI 30 0-32767mA 0-32767mA
Hot voltage AI 218 0-32767V 0-32767V
Firewire electric current AI 90 0-32767mA 0-32767mA
Brake input voltage AI 235 0-32767V 0-32767V
Brake both end voltage AI 110 0-32767V 0-32767V
Brake electric current AI 500 0-32767mA 0-32767mA
Brake force AI 200 0-32767n 0-32767n
Speed encoder power supply AI 8 0-32767V 0-32767V
Speed encoder operating temperature AI 78 0-32767Degree 0-32767Degree
Door lock relay status DI 1 0=RELEASE/1=PULL-IN 0=releases/1=is attracted
The door lock relay response time AI 600 0-32767ns 0-32767ns
Frequency converter input voltage AI 200 0-32767V 0-32767V
Inverter current AI 1000 0-32767mA 0-32767mA
Frequency converter rotating speed AI 1560 0-32767RPM 0-32767RPM
Frequency converter work throat sound DI 1 0=A BN O RM AL/1=N O RM AL 0=exception/1=are normal
Photoelectric induction device state DI 1 0=A BN O RM AL/1=N O RM AL 0=often/1=is normal
Voltage during door electromechanical source AI 250 0-32767V 0-32767V
Door electromechanical source electric voltage frequency AI 50 0-32767Hz 0-32767Hz
Door machine out-put supply AI 2000 0-32767MA 0-32767MA
Door machine output power DI 400 0-32767W 0-32767W
Door machine connecting line state AI 1 0=A BN O RM AL/1=N O RM AL 0=exception/1=are normal
Encoder power supply voltage AI 15 0-32767V 0-32767V
Shutdown is whether there is to hit DI 0 0=N O/1=HAVE 0=has without/1=
Lubricating oil temperature AI 180 0-32767Degree 0-32767 degree
Elevator door usage time AI 18 0-32767Months 0-32767Months
S104, such as table 3, Fault Diagnosis Database utilize equipment fault diagnosis table (FAULTDB), base data table (Elec_ Basic) build, the overlying relation between two tables, the data in fault diagnosis table (FAULTDB) are according to basic data meter Calculate, the renewal of base data table causes the renewal of fault diagnosis table, and two table capacity can add;
3. fault diagnosis table FAULTDB of table
Abnormality degree Monitoring part State Connection status Sub- malfunction RCTED RCTED RESERVED2 RESERVED3
0.19 Brake 0 1 NULL - 17 09.04.53.531000 afternoons ... of the 12-4 months ...
0.029 Steel wire rope 2 1 NULL - 17 11.51.17.213000 afternoons ... of the 12-8 months ...
0.035 Frequency converter 0 1 NULL - 17 08.48.01.917000 afternoons ... of the 12-4 months ...
0.197 Encoder 0 1 NULL - 17 09.07.55.421000 afternoons ... of the 12-4 months ..
0.83 Traction machine 2 1 NULL - 17 09.08.07.183000 afternoons ... of the 12-4 months ...
0.79 Door electromechanical source 2 1 NULL - 17 09.08.17.683000 afternoons ... of the 12-4 months ...
0.438 Door lock relay 1 1 NULL - 17 09.08.28.834000 afternoons ... of the 12-4 months ..
The client software of oracle database is installed on S105, client machine.Correct configuration oracle database Tnsname.ora files under client directory, there is provided early warning system access address and access port, start the monitoring of server-side Program, is connected with Matlab and data in oracle database is calculated by life prediction algorithm more than warning algorithm system, renewal Data in oracle database, recording equipment last state.
An important factor for S2, understand elevator basic structure, determines the critical component for causing elevator faults, and screening influences elevator Analogue simulation is carried out as the index for judging elevator faults, and to achievement data;
S201, understanding elevator configuration tractive driving, suspension arrangement, car frame and carriage, door system, weight balancing system, Electric drive system, electrical system and safety system and relevant parameter simultaneously consult amount of literature data, design corresponding Index is as shown in table 4, judges the health status of equipment in being calculated in the later stage by the change of achievement data;
The gloomy machine-roomless lift TE-Evolution1 index tables of 4. base of a fruit of table
S202, call oracle database design module Devicetype tables, Index tables and Basicdb tables.According to collection To elevator data carry out big data analysis off-line manner, form Basic tables and Indexdb tables deposit database, collection Data as latest data, compared with other data in table, the operating status of analytical equipment;
S203, such as Fig. 8, with reference to the gloomy TE-evolution1 elevator structures of the base of a fruit, to basis of the elevator faults in relation to factor On, main study subject is divided into three big module of traction machine, steel wire rope and door system, and propose accordingly for each section module Index, the comprehensive influence to elevator, the principal element for determining to cause elevator faults simultaneously determines that fault diagnosis index is door system Middle door hanging foot wheel abrasion degree and wirerope diameter, and determine to hang in door system foot wheel abrasion degree and wirerope diameter threshold value, When system equipment corresponding index as obtained by calculating the real time data of collection exceeds threshold range, you can judge that event occurs in equipment Barrier;
S204, for change gear degree of wear index, quote in IBM algorithms and survey wear model, exceed original for predicting The abrasion of beginning surface roughness depth.
1) basic assumption of wear model can be surveyed
It is ENERGY E that abrasion that each number of pass times produces is consumed and number of pass times N the two variables according to wear extent Function this hypothesis.Relation between wear extent Q and number of pass times N and the ENERGY E consumed is the differential equation
Show according to the experiment, the relation of wear extent and number of pass times is
In formula, m is the given constant of system, and related document is pointed out, when unlubricated,
2) according to the energy of consumption and the relation of wear process, the abrasion of A types can be divided into and worn with Type B.
Wherein, A types abrasion refers to energy of rupture in wear process and remains constant, belongs to serious abrasion.It is mainly used in Dry friction and case of heavy load.Do not introduce herein specifically.
Wherein, Type B abrasion refers to destroy in wear process and changes with the change of number of pass times, this is moderate turns Move result.The door change gear abrasion of this research just belongs to such case.This abrasion can be represented with the following differential equation
In formula, C is the constant coefficient of system, and the wear extent that must be generally determined by experiment under a certain number of pass times is tried to achieve. Zero abrasion and the models coupling that can survey abrasion can be got up try to achieve the analytic expression of C, N is number of pass times, and S is contact surface edge slip side To length, Q is polishing scratch cross-sectional area.
In this wear model, it has been assumed that destroying work(and τmxS is directly proportional, because the contact area of two articles is with abrasion And change, so τmxS is not a constant.
S205, for wirerope diameter index, using finite element theory, the mutation analysis of calculated diameter.Steel wire rope it is straight Vernier caliper measurement of the footpath with jaw, the width minimum of jaw as shown in Figure 9 will be enough to cross over two i.e. AB of adjacent stock Between at least to include two strands, measure and i.e. measurement point A is carried out on the straight position outside steel wire rope end head 15m surpass apart from rope end C Cross 15 meters, B points apart from rope end D more than 15 meters, at a distance of at least in the two sections of 1m, and same section it is orthogonal measure Two values rotating vernier slide calliper rule, according to two vertical direction measurements of left-right and front-back, the average value of four measurement results, is steel The measured diameter of cord.
S3, collection elevator real time data, structure dynamic bayesian network (DBN), and by Monte carlo algorithm to DBN nets The each node of network, which solves, estimates probability of malfunction, forms CPT (probability of malfunction), early warning is made to the failure of elevator;
Secondary development is carried out using Matlab technological frames, Bayesian Networks Toolbox (FULLBNT) is used, using pattra leaves This network algorithm, builds dynamic bayesian network, calculates the probability distribution of DBN (dynamic bayesian network), infers Escalator shape State simultaneously predicts remaining life.
S301, Data Collection:Established according to data system in step S2, being called from database includes device history index Indexdb tables, Devicetype tables and the Index table training dynamic bayesian networks of data simultaneously update the data equipment health in storehouse Assess table ELEIDX, fault diagnosis table FAULTDB;
S302, determine node variable and causality:Network node represents stochastic variable, and the node of Bayesian network allows It is any variable, node variable can be divided into chance variable, decision variable and auxiliary variable, can be continuous variable, also may be used It is discrete variable.Determine node variable and causality be named including node variable, the description of each part of system and amount of degradation Measurement, types of variables (discrete and continuous) selection;
S303, structure DBN:By analyzing each component coupled relation of Escalator, with reference to electric staircase failure diagnosis index extraction Module, builds Escalator Bayesian network fault diagnosis model figure, and Figure 10 gives the Bayesian network fault diagnosis mould of elevator Type figure, being lost in node medium between node insulation ag(e)ing has an arc, i.e. right when matrix represents Bayesian network in BNT It is 1 to answer matrix (dielectric loss, insulation ag(e)ing) value, is otherwise 0.The situation of all nodes is all inputted into BNT matrixes, establishes pattra leaves This net BNT, substitutes into learn_params () function as training set using obtained sample matrix and is learnt, study obtains bar Part probability tables CPT.With the increase of number of training, the conditional probability table learnt increasingly approaches real conditional probability Table;
The specific steps of elevator DBN model structure:
1) prior Bayesian network is built by fixed node and priori data.
2) call data to carry out parameter learning to DBN networks, build posterior Bayesian network.
Parameter Learning Algorithm function:Abundant parameter learning function is provided in BNT, in shortage of data, it is known that network is opened up In the case of flutterring structure, we carry out parameter learning using the EM algorithms learn_params_em of successive ignition optimizing
1. the process of EM Algorithm for Solving optimum structures really converges to local optimum parameterProcess.First by with Machine number generator (being applied in Matlab instruments) produces sample training collection, fills up loss data.Maximum likelihood is carried out to data to estimate Meter learning simulation best suits the parameter of structure, comprises the following steps that:
E walks (expectation):
Wherein E is desired value;D is training sample set;Represent the optimum structure parameter found, XiCodomain be:qiIt is configuration πiPut in order;NijkIt is to meet variate-value in data set DAnd πiThe bar of=j Part frequency:ylIt is the data amount check lost in D;S hIt is that bayesian network structure selection is assumed.
M walks (maximum estimated):
Maximal possibility estimation function:
MAP estimation function:
N′ijkIt is the abundant statistical factors of priori;Correspondingly, NijkIt is the abundant statistical factors of sample data, i, j, k, h, q ∈ N.
2. SEM algorithms, the optimization algorithm of EM algorithms, structuring expectation maximization SEM algorithms.The application process of SEM algorithms It is roughly divided into two step of search structure and parameter learning.The search inch of network structure is carried out, SEM algorithms are sufficiently counted using expectation The factor replaces the sufficient statistical factors being not present so that the form of scoring functions has decomposability, then carries out local Search, finds the structure of network scoring higher.Finally, the parameter of score maximum is screened in selected structure.
Its specific implementation procedure is as follows:The initial value of variable is set;The iteration since some initial Bayesian network; Call joint tree reasoning algorithm to complete the reasoning computing to Bayesian network, draw current optimum network;Based on current optimal net Network carries out completion using EM algorithms to data set, obtains complete data set to realize the maximization of parameter;Institute in calculating network There is the summation of node value number;All network structures different from initial network are created, using these structures as candidate network Structure;Given a mark to the network structure of above-mentioned candidate with BIC score functions, found from selected network structure so that score Maximum parameter;Go out the best network structure with data set fitting.
3) its probability of malfunction is calculated using Monte carlo algorithm to each node in DBN model, so that it is determined that all nets CPT (probability distribution) table of network node.
S304, the probability for estimating according to trained Bayesian network unit failure, database is deposited into by result of calculation In health Evaluation table and fault diagnosis table, storehouse is updated the data, realizes detection in real time, timely discovering device failure.
Elevator achievement data in S4, calling database, generates the sequence of abnormal elevator degree and run time, obtains exception The function of time is spent, determining elevator by the abnormality degree function of time, run time is compared with abnormality degree time threshold, so that real Existing Escalator life prediction algorithm.
Expertise is collected, writes expert system, according to the real-time analysis to system mode, by expert system to elevator State is analyzed, and the service life is predicted.
S401, the CPT tables by step S3 Bayes fault diagnosis model and network node, from database facility health shape Condition assesses table extraction equipment Real-time Monitoring Data, equipment health status is assessed, output equipment abnormality degree (probability of malfunction);
Historical data in S402, calling health evaluating table, generates unit exception degree time series, is intended with Matlab curves Close tool box CFTOOL and determine fitting function exponent number, obtain unit exception degree and the fitting function of run time, determine that equipment is transported Capable general status, for the remaining life of computing device;
S403, call unit exception degree and the fitting function of run time, sets the corresponding abnormality degree threshold value of equipment scrapping A, corresponding service life are x, obtain the abnormality degree b (b of equipment at this time to system real time fail probability analysis using expert system ≤ a), as b < a, equipment run time y is calculated when abnormality degree is b by abnormality degree and run time fitting function, then is set Standby remaining life w=x-y;As b=a, remaining life 0, realizes and system lifetim is predicted.
Above-described embodiment is the preferable embodiment of the present invention, but embodiments of the present invention and from above-described embodiment Limitation, other any Spirit Essences without departing from the present invention with made under principle change, modification, replacement, combine, simplification, Equivalent substitute mode is should be, is included within protection scope of the present invention.

Claims (10)

1. a kind of fault pre-alarming and life-span prediction method of Escalator equipment, it is characterised in that the fault pre-alarming and service life Forecasting Methodology comprises the following steps:
S1, using SQL statement, establish database relational table, be stored in oracle database, Matlab is connect by ODBC bridgings Oracle database accesses data, completes database sharing;
An important factor for S2, analysis elevator basic structure, determine the critical component for causing elevator faults, and screening influences elevator conduct Judge the index of elevator faults, and analogue simulation is carried out to achievement data;
S3, collection elevator real time data, build dynamic bayesian network, and by Monte carlo algorithm to dynamic bayesian network Each node, which solves, estimates probability of malfunction, forms CPT probabilities of malfunction, early warning is made to the failure of elevator;
Elevator achievement data in S4, calling oracle database, generates the sequence of abnormal elevator degree and run time, obtains different The normal manner function of time, by the abnormality degree function of time determine elevator run time compared with abnormality degree time threshold so that Realize Escalator life prediction algorithm.
2. the fault pre-alarming and life-span prediction method of a kind of Escalator equipment according to claim 1, it is characterised in that institute The step S1 stated includes:
S101, using SQL statement, with reference to pivot, decode in Oracle, programming apparatus health evaluating table ELEIDX, failure Diagnostics table FAULTDB, data point table ELE forms, build up a series of relation tables by original point of interface table select, are stored in In oracle database;
S102, base data and fault type according to equipment and part of appliance, using Devicetype tables, Index tables, Indexdb tables, Basicdb tables, Basicdb tables structure data system, using hierarchical structure database design health evaluating with Fault Diagnosis Database part, wherein, Devicetype tables storage device type and part of appliance and its corresponding numbering, Index tables storage facilities refines index species, and the record of Indexdb tokens is in real time and recent index, Basicdb tables are used to record reality When and the base data values collected in the recent period, Basic tables be used to express each base data classification and between each component and component index Contact;
S103, utilize equipment fault evaluation index table Elec_Index and base data table Elec_Basic structure health evaluating number According to storehouse, there are overlying relation between two tables, data are according to base data table in equipment fault evaluation index table Elec_Index Elec_Basic is calculated, and the renewal of base data table causes the renewal of index table data, and two tables its capacity can add;
S104, build Fault Diagnosis Database using equipment fault diagnosis table Faultdb and base data table Elec_Basic, and two There are overlying relation between table, data are calculated according to base data table Elec_Basic in equipment fault diagnosis table Faultdb, The renewal of base data table causes the renewal of fault diagnosis table data, and two table capacity can add;
The client software of Oracle is installed, correctly under the client directory of configuration Oracle on S105, client machine Tnsname.ora files, there is provided early warning system access address and access port, start the oracle listener of server-side;
S106, database and Matlab connect the data inquired about using exec in database by ODBC bridgings, and fetch functions return Upper strata variable, update update the data health evaluating table and fault diagnosis table in storehouse.
3. the fault pre-alarming and life-span prediction method of a kind of Escalator equipment according to claim 1, it is characterised in that institute The step S2 stated includes:
S201, analysis elevator configuration tractive driving, suspension arrangement, car frame and carriage, door system, weight balancing system, electric power Dragging system, electrical system and safety system, design objective;
S202, call Devicetype tables, Index tables and Basicdb tables in database, by the elevator data collected with offline Mode carry out big data analysis, form Basic tables and Indexdb tables deposit database;
S203, with reference to the gloomy TE-evolution1 elevator structures of the base of a fruit, research object is divided into traction machine, steel wire rope and door system, and Corresponding index is proposed for each research object, and determines that fault diagnosis index is the change gear degree of wear and steel wire in door system Rope diameter;
S204, for change gear degree of wear index, quote in IBM algorithms and survey wear model, exceed original table for predicting The abrasion of surface roughness depth;
S205, for wirerope diameter index, using finite element theory, the mutation analysis of calculated diameter.
4. the fault pre-alarming and life-span prediction method of a kind of Escalator equipment according to claim 1, it is characterised in that institute The step S3 stated includes:
S301, Data Collection, call Indexdb tables, the Devicetype tables for including device history achievement data from database With Index tables;
S302, determine network node variable and causality, determines that network node variable and causality take including node variable The selection of the measurement, types of variables (discrete and continuous) of name, the description of each part of system and amount of degradation;
S303, structure elevator DBN model, analyze each component coupled relation of Escalator, with reference to electric staircase failure diagnosis index extraction Module, builds Escalator Bayesian network fault diagnosis model figure, uses matrix-style to represent Bayesian network in BNT, even Node i has an arc to j, then (i, j) value is 1 in homography, is otherwise 0, establishes Bayesian network BNT, and using sample as Training set substitutes into learn_params () function and is learnt, and study obtains conditional probability table CPT;
S304, the probability for estimating according to trained Bayesian network unit failure, database health shape is stored in by result of calculation Condition is assessed in table and fault diagnosis table.
5. the fault pre-alarming and life-span prediction method of a kind of Escalator equipment according to claim 1, it is characterised in that institute The step S4 stated includes:
S401, the CPT tables by Bayes's fault diagnosis model and network node, carry from database facility health Evaluation table Taking equipment Real-time Monitoring Data, assesses equipment health status, output equipment abnormality degree, i.e. probability of malfunction;
Historical data in S402, calling health evaluating table, generates unit exception degree time series, with Matlab curve matching works Tool case CFTOOL determines fitting function exponent number, obtains unit exception degree and the fitting function of run time;
S403, the fitting function for calling unit exception degree and run time, set equipment scrapping corresponding abnormality degree threshold value a, right The service life answered is x, and the abnormality degree b of equipment at this time, foundation are obtained to system real time fail probability analysis using expert system The fitting function of unit exception degree and run time, calculates the remaining life w=x- of equipment usage time y, then equipment at this time Y, realizes and system lifetim is predicted.
6. the fault pre-alarming and life-span prediction method of a kind of Escalator equipment according to claim 3, it is characterised in that institute The step S204 stated includes:
S2041, the basic assumption that wear model can be surveyed,
It is the letter for ENERGY E and number of pass times N the two variables that abrasion that each number of pass times produces is consumed according to wear extent This is counted it is assumed that the relation between wear extent Q and number of pass times N and the ENERGY E consumed is the differential equation
<mrow> <mi>d</mi> <mi>Q</mi> <mo>=</mo> <msub> <mrow> <mo>(</mo> <mfrac> <mrow> <mo>&amp;part;</mo> <mi>Q</mi> </mrow> <mrow> <mo>&amp;part;</mo> <mi>N</mi> </mrow> </mfrac> <mo>)</mo> </mrow> <mi>E</mi> </msub> <mi>d</mi> <mi>N</mi> <mo>+</mo> <msub> <mrow> <mo>(</mo> <mfrac> <mrow> <mo>&amp;part;</mo> <mi>Q</mi> </mrow> <mrow> <mo>&amp;part;</mo> <mi>E</mi> </mrow> </mfrac> <mo>)</mo> </mrow> <mi>N</mi> </msub> <mi>d</mi> <mi>E</mi> </mrow>
Wherein, wear extent and the relation of number of pass times are
<mrow> <mi>Q</mi> <mo>=</mo> <msup> <mi>CN</mi> <mfrac> <mn>1</mn> <mi>m</mi> </mfrac> </msup> <mo>,</mo> <mi>w</mi> <mo>&gt;</mo> <mn>0</mn> </mrow>
In formula, m is the given constant of system, when unlubricated,
<mrow> <mi>Q</mi> <mo>=</mo> <msub> <mi>C</mi> <mn>1</mn> </msub> <msup> <mi>N</mi> <mfrac> <mn>1</mn> <mrow> <mn>3</mn> <mi>x</mi> <mn>3</mn> </mrow> </mfrac> </msup> <mo>,</mo> <mn>2</mn> <mo>&amp;le;</mo> <mi>N</mi> <mo>&amp;le;</mo> <mn>2000</mn> <mo>;</mo> </mrow>
S2042, according to the energy of consumption and the relation of wear process, the abrasion of A types can be divided into and worn with Type B, wherein, the abrasion of A types Energy of rupture remains unchanged in wear process, belongs to heavy wear, and applied to dry friction and case of heavy load, Type B abrasion is being worn During destroy as number of pass times changes and change, be moderate transfer as a result, being worn for door change gear, under abrasion use The row differential equation represents
<mrow> <mi>d</mi> <mo>&amp;lsqb;</mo> <mfrac> <mi>Q</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>&amp;tau;</mi> <mrow> <mi>m</mi> <mi>x</mi> </mrow> </msub> <mi>S</mi> <mo>)</mo> </mrow> <mfrac> <mn>9</mn> <mn>2</mn> </mfrac> </msup> </mfrac> <mo>&amp;rsqb;</mo> <mo>=</mo> <mi>C</mi> <mi>d</mi> <mi>N</mi> </mrow>
In above formula, C is the constant coefficient of system, and the wear extent being determined by experiment under a certain number of pass times tries to achieve or by zero abrasion Get up to try to achieve the analytic expression of C with the models coupling that can survey abrasion, N is number of pass times, and S is length of the contact surface along glide direction, Q For polishing scratch cross-sectional area,
It is assumed that destroy work(and τmxS is directly proportional, and the contact area of two articles changes with abrasion, therefore τmxS is not a constant,
<mrow> <mi>d</mi> <mi>Q</mi> <mo>=</mo> <mo>&amp;lsqb;</mo> <mi>C</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>&amp;tau;</mi> <mrow> <mi>m</mi> <mi>x</mi> </mrow> </msub> <mi>S</mi> <mo>)</mo> </mrow> <mfrac> <mn>9</mn> <mn>2</mn> </mfrac> </msup> <mo>&amp;rsqb;</mo> <mi>d</mi> <mi>N</mi> <mo>+</mo> <mfrac> <mi>W</mi> <mrow> <msub> <mi>&amp;tau;</mi> <mrow> <mi>m</mi> <mi>x</mi> </mrow> </msub> <mi>S</mi> </mrow> </mfrac> <mi>d</mi> <mrow> <mo>(</mo> <msub> <mi>&amp;tau;</mi> <mrow> <mi>m</mi> <mi>x</mi> </mrow> </msub> <mi>S</mi> <mo>)</mo> </mrow> <mo>.</mo> </mrow>
7. the fault pre-alarming and life-span prediction method of a kind of Escalator equipment according to claim 3, it is characterised in that institute The step S205 stated is specific as follows:
Vernier caliper measurement of the diameter of steel wire rope with jaw, the width minimum of jaw will be enough adjacent across two Stock, measures and is carried out on the straight position outside steel wire rope end head 15m, apart at least in the two sections of 1m, and in same section Two values, the measured diameter of the average value of four measurement results, as steel wire rope are measured orthogonally.
8. the fault pre-alarming and life-span prediction method of a kind of Escalator equipment according to claim 4, it is characterised in that institute It is specific as follows that elevator DBN model is built in the step S303 stated:
S3031, by fixed node and priori data build prior Bayesian network;
S3032, call data to carry out parameter learning to DBN networks, builds posterior Bayesian network;
S3033, calculate its probability of malfunction to each node in DBN model using Monte carlo algorithm, determines all-network node CPT probability distribution tables.
9. the fault pre-alarming and life-span prediction method of a kind of Escalator equipment according to claim 8, it is characterised in that institute The parameter learning stated is to be learnt using the EM algorithms learn_params_em of successive ignition optimizing, wherein, EM Algorithm for Solving The process of optimum structure is to converge to local optimum parameterProcess, by randomizer produce sample training collection, fill out Mend and lose data, data are carried out with the parameter that maximal possibility estimation learning simulation best suits structure, is comprised the following steps that:
E is walked, that is, it is expected:
<mrow> <msub> <mi>E</mi> <mrow> <mi>p</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>|</mo> <mi>D</mi> <mo>,</mo> <mi>&amp;theta;</mi> <mi>s</mi> <mo>,</mo> <msup> <mi>s</mi> <mi>h</mi> </msup> <mo>)</mo> </mrow> </mrow> </msub> <mrow> <mo>(</mo> <msub> <mi>N</mi> <mrow> <mi>i</mi> <mi>j</mi> <mi>k</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <mi>p</mi> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>i</mi> <mi>k</mi> </msubsup> <mo>,</mo> <msubsup> <mi>&amp;pi;</mi> <mi>i</mi> <mi>j</mi> </msubsup> <mo>|</mo> <msub> <mi>y</mi> <mn>1</mn> </msub> <mo>,</mo> <msub> <mover> <mi>&amp;theta;</mi> <mo>&amp;OverBar;</mo> </mover> <mi>s</mi> </msub> <mo>,</mo> <msup> <mi>s</mi> <mi>h</mi> </msup> <mo>)</mo> </mrow> </mrow>
Wherein E is desired value;D is training sample set;Represent the optimum structure parameter found, XiCodomain be:qiIt is configuration πiPut in order;NijkIt is to meet variate-value in data set DAnd πiThe bar of=j Part frequency:ylIt is the data amount check lost in D;ShIt is that bayesian network structure selection is assumed;
M is walked, i.e. maximum estimated, maximal possibility estimation function:
<mrow> <msub> <mi>&amp;theta;</mi> <mrow> <mi>i</mi> <mi>j</mi> <mi>k</mi> </mrow> </msub> <mo>=</mo> <mfrac> <mrow> <msub> <mi>E</mi> <mrow> <mi>p</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>|</mo> <mi>D</mi> <mo>,</mo> <msub> <mi>v</mi> <mi>s</mi> </msub> <mo>,</mo> <msup> <mi>s</mi> <mi>h</mi> </msup> <mo>)</mo> </mrow> </mrow> </msub> <mrow> <mo>(</mo> <msub> <mi>N</mi> <mrow> <mi>i</mi> <mi>j</mi> <mi>k</mi> </mrow> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>q</mi> <mi>i</mi> </msub> </munderover> <msub> <mi>E</mi> <mrow> <mi>p</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>|</mo> <mi>D</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>s</mi> </msub> <mo>,</mo> <msup> <mi>s</mi> <mi>h</mi> </msup> <mo>)</mo> </mrow> </mrow> </msub> <mrow> <mo>(</mo> <msub> <mi>N</mi> <mrow> <mi>i</mi> <mi>j</mi> <mi>k</mi> </mrow> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> </mrow>
MAP estimation function:
<mrow> <msub> <mi>&amp;theta;</mi> <mrow> <mi>i</mi> <mi>j</mi> <mi>k</mi> </mrow> </msub> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>N</mi> <mrow> <mi>i</mi> <mi>j</mi> <mi>k</mi> </mrow> <mo>,</mo> </msubsup> <mo>+</mo> <msub> <mi>E</mi> <mrow> <mi>p</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>|</mo> <mi>D</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>s</mi> </msub> <mo>,</mo> <msup> <mi>s</mi> <mi>h</mi> </msup> <mo>)</mo> </mrow> </mrow> </msub> <mrow> <mo>(</mo> <msub> <mi>N</mi> <mrow> <mi>i</mi> <mi>j</mi> <mi>k</mi> </mrow> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <msup> <mi>q</mi> <mi>i</mi> </msup> </munderover> <mrow> <mo>(</mo> <msubsup> <mi>N</mi> <mrow> <mi>i</mi> <mi>j</mi> <mi>k</mi> </mrow> <mo>,</mo> </msubsup> <mo>+</mo> <msub> <mi>E</mi> <mrow> <mi>p</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>|</mo> <mi>D</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>s</mi> </msub> <mo>,</mo> <msup> <mi>s</mi> <mi>h</mi> </msup> <mo>)</mo> </mrow> </mrow> </msub> <mo>(</mo> <msub> <mi>N</mi> <mrow> <mi>i</mi> <mi>j</mi> <mi>k</mi> </mrow> </msub> <mo>)</mo> <mo>)</mo> </mrow> </mrow> </mfrac> </mrow>
N′ijkIt is the abundant statistical factors of priori, NijkIt is the abundant statistical factors of sample data, i, j, k, h, q ∈ N.
10. the fault pre-alarming and life-span prediction method of a kind of Escalator equipment according to claim 8, it is characterised in that The parameter learning is to be learnt using the SEM algorithms of successive ignition optimizing, wherein, the specific implementation procedure of SEM algorithms It is as follows:The initial value of variable is set;The iteration since some initial Bayesian network;Joint tree reasoning algorithm is called to complete Reasoning computing to Bayesian network, draws current optimum network;Based on current optimum network using EM algorithms to data set into Row completion, obtains complete data set to realize the maximization of parameter;The summation of all node value numbers in calculating network;Wound The network structure different from initial network is built, using these structures as candidate network structure;With BIC score functions to above-mentioned candidate Network structure give a mark, from selected network structure find so that score maximum parameter;Go out and data set fitting Best network structure.
CN201711263968.3A 2017-12-05 2017-12-05 Fault early warning and service life prediction method for escalator equipment Active CN107991870B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711263968.3A CN107991870B (en) 2017-12-05 2017-12-05 Fault early warning and service life prediction method for escalator equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711263968.3A CN107991870B (en) 2017-12-05 2017-12-05 Fault early warning and service life prediction method for escalator equipment

Publications (2)

Publication Number Publication Date
CN107991870A true CN107991870A (en) 2018-05-04
CN107991870B CN107991870B (en) 2020-10-27

Family

ID=62035539

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711263968.3A Active CN107991870B (en) 2017-12-05 2017-12-05 Fault early warning and service life prediction method for escalator equipment

Country Status (1)

Country Link
CN (1) CN107991870B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108469783A (en) * 2018-05-14 2018-08-31 西北工业大学 Deep hole deviation from circular from prediction technique based on Bayesian network
CN109034010A (en) * 2018-07-06 2018-12-18 北京天泽智云科技有限公司 A kind of lubrication failure on-line prediction method of automatic door unit
CN109534140A (en) * 2018-12-27 2019-03-29 北京交通大学 The modeling of Escalator step chains and fault simulation method based on SIMPACK
CN110133508A (en) * 2019-04-24 2019-08-16 上海博强微电子有限公司 The safe early warning method of electric automobile power battery

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102651054A (en) * 2012-04-10 2012-08-29 北京航空航天大学 Probability method of electronic product service life model based on Bayesian theory
CN102932194A (en) * 2011-08-09 2013-02-13 中国银行股份有限公司 Bayes method-based Internet application service monitoring system and method
CN103245861A (en) * 2013-05-03 2013-08-14 云南电力试验研究院(集团)有限公司电力研究院 Transformer fault diagnosis method based on Bayesian network
CN103245911A (en) * 2013-05-03 2013-08-14 云南电力试验研究院(集团)有限公司电力研究院 Breaker fault diagnosis method based on Bayesian network
CN103678952A (en) * 2013-11-14 2014-03-26 昆明理工大学 Elevator risk evaluation method
CN104392122A (en) * 2014-11-17 2015-03-04 北京航空航天大学 Probabilistic life evaluation method based on crack detection probability model
CN104944240A (en) * 2015-05-19 2015-09-30 重庆大学 Elevator equipment state monitoring system based on large data technology
CN107013473A (en) * 2017-04-19 2017-08-04 武汉惜源科技有限公司 A kind of pumping plant real time on-line monitoring and energy efficiency managing method and system
CN107473036A (en) * 2017-07-31 2017-12-15 浙江省特种设备检验研究院 Elevator traction machine remote online detects and diagnostic system and its checkout and diagnosis method

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102932194A (en) * 2011-08-09 2013-02-13 中国银行股份有限公司 Bayes method-based Internet application service monitoring system and method
CN102651054A (en) * 2012-04-10 2012-08-29 北京航空航天大学 Probability method of electronic product service life model based on Bayesian theory
CN103245861A (en) * 2013-05-03 2013-08-14 云南电力试验研究院(集团)有限公司电力研究院 Transformer fault diagnosis method based on Bayesian network
CN103245911A (en) * 2013-05-03 2013-08-14 云南电力试验研究院(集团)有限公司电力研究院 Breaker fault diagnosis method based on Bayesian network
CN103678952A (en) * 2013-11-14 2014-03-26 昆明理工大学 Elevator risk evaluation method
CN104392122A (en) * 2014-11-17 2015-03-04 北京航空航天大学 Probabilistic life evaluation method based on crack detection probability model
CN104944240A (en) * 2015-05-19 2015-09-30 重庆大学 Elevator equipment state monitoring system based on large data technology
CN107013473A (en) * 2017-04-19 2017-08-04 武汉惜源科技有限公司 A kind of pumping plant real time on-line monitoring and energy efficiency managing method and system
CN107473036A (en) * 2017-07-31 2017-12-15 浙江省特种设备检验研究院 Elevator traction machine remote online detects and diagnostic system and its checkout and diagnosis method

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
冯鑫等: "基于贝叶斯模型的电梯安全评估方法研究", 《起重运输机械》 *
张星辉等: "基于动态贝叶斯网络的设备故障预测方法研究", 《科技广场》 *
李志强等: "视情维修条件下的多状态控制单元可用性建模与分析", 《兵工学报》 *
杨志波等: "动态贝叶斯网络在设备剩余寿命预测中的应用研究", 《计算机集成制造系统》 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108469783A (en) * 2018-05-14 2018-08-31 西北工业大学 Deep hole deviation from circular from prediction technique based on Bayesian network
CN108469783B (en) * 2018-05-14 2021-02-02 西北工业大学 Deep hole roundness error prediction method based on Bayesian network
CN109034010A (en) * 2018-07-06 2018-12-18 北京天泽智云科技有限公司 A kind of lubrication failure on-line prediction method of automatic door unit
CN109534140A (en) * 2018-12-27 2019-03-29 北京交通大学 The modeling of Escalator step chains and fault simulation method based on SIMPACK
CN109534140B (en) * 2018-12-27 2019-12-06 北京交通大学 Escalator step chain modeling and fault simulation method based on SIMPACK
CN110133508A (en) * 2019-04-24 2019-08-16 上海博强微电子有限公司 The safe early warning method of electric automobile power battery

Also Published As

Publication number Publication date
CN107991870B (en) 2020-10-27

Similar Documents

Publication Publication Date Title
Zhang et al. Towards a fuzzy Bayesian network based approach for safety risk analysis of tunnel‐induced pipeline damage
Wang et al. A two-stage data-driven-based prognostic approach for bearing degradation problem
CN104820716B (en) Equipment Reliability appraisal procedure based on data mining
CN105809255B (en) A kind of thermal power plant&#39;s rotating machinery health control method and system based on Internet of Things
Okoh et al. Overview of remaining useful life prediction techniques in through-life engineering services
Asadzadeh et al. An integrated systemic model for optimization of condition-based maintenance with human error
CN102765643B (en) Elevator fault diagnosis and early-warning method based on data drive
Heng et al. Rotating machinery prognostics: State of the art, challenges and opportunities
Gorjian et al. A review on degradation models in reliability analysis
Boutros et al. Detection and diagnosis of bearing and cutting tool faults using hidden Markov models
Javed et al. State of the art and taxonomy of prognostics approaches, trends of prognostics applications and open issues towards maturity at different technology readiness levels
Wu et al. A neural network integrated decision support system for condition-based optimal predictive maintenance policy
CN106200616B (en) Periodic decomposition analysis for remote machine monitoring
JP5530020B1 (en) Abnormality diagnosis system and abnormality diagnosis method
US7743005B2 (en) Diagnosis of abnormal operation modes of a machine utilizing self organizing map
CA2545695C (en) Method and system for predicting remaining life for motors featuring on-line insulation condition monitor
Sheu Dynamic relief-demand management for emergency logistics operations under large-scale disasters
JP4601492B2 (en) Quality prediction system and method for production process
CN103592141B (en) For the method for the reliability for testing complication system
Di Maio et al. Combining relevance vector machines and exponential regression for bearing residual life estimation
MAzurkiewicz Computer-aided maintenance and reliability management systems for conveyor belts
Tobon-Mejia et al. A data-driven failure prognostics method based on mixture of Gaussians hidden Markov models
Li et al. Remaining useful life prediction based on a general expression of stochastic process models
AU2008311268B2 (en) Systems and methods for improving haul road conditions
US8566070B2 (en) Apparatus abnormality monitoring method and system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant