CN107991870B - Fault early warning and service life prediction method for escalator equipment - Google Patents
Fault early warning and service life prediction method for escalator equipment Download PDFInfo
- Publication number
- CN107991870B CN107991870B CN201711263968.3A CN201711263968A CN107991870B CN 107991870 B CN107991870 B CN 107991870B CN 201711263968 A CN201711263968 A CN 201711263968A CN 107991870 B CN107991870 B CN 107991870B
- Authority
- CN
- China
- Prior art keywords
- data
- equipment
- database
- wear
- fault
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive 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
- G05B13/042—Adaptive 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
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Maintenance And Inspection Apparatuses For Elevators (AREA)
- Indicating And Signalling Devices For Elevators (AREA)
Abstract
The invention discloses a fault early warning and service life prediction method for escalator equipment. Various algorithms adopted in the method are designed based on equipment index analysis, escalator indexes in the algorithms are replaced by other equipment indexes, and real-time index data of corresponding detection equipment is collected, so that the health assessment condition and the residual life of the detection equipment can be obtained.
Description
Technical Field
The invention relates to the technical field of acquisition of fault characteristic information, processing of information uncertainty and analysis and processing of mass data, in particular to a fault early warning and service life prediction method for escalator equipment.
Background
According to the equipment fault early warning and service life prediction, the abnormal condition of the equipment can be accurately forecasted in time before the equipment really breaks down according to the equipment operation rule or the possibility precursor obtained by observation, corresponding measures are taken, loss caused by equipment faults is reduced to the maximum extent, and the safety and stability of the operation process of the equipment are ensured. Therefore, a reliable state monitoring technology is particularly urgent to timely and effectively monitor and diagnose process abnormalities.
The existing equipment fault early warning technology is mainly divided into three categories: a mechanism model based approach, a knowledge based approach and a data driven approach.
The method based on the mechanism model mainly comprises two stages: a residual error, which is derived from the difference between the system output and the actual measured value estimated by a mathematical model built on the plant operating mechanism, and a residual error estimate, according to which the plant is analyzed for the presence of a fault, are generated. The method is tightly combined with a control theory, most mechanism models are linear systems, so that when a complex system with nonlinearity, high degree of freedom and multivariable coupling is faced, the fault of equipment cannot be well detected, a huge cost is needed to establish the models, and the method is poor in monitoring effect due to various environmental restrictions and cannot be widely applied.
The knowledge-based method requires a complete database with a large amount of knowledge and experience, and automatically describes the connection relationship, the fault propagation mode and the like among all units in the monitoring process according to the heuristic experience of relevant experts. For simple systems, the performance is better, and for complex systems, the universality is poor due to various problems caused by insufficient complete database and enough empirical reasoning deduction fault processes.
The data driving-based method is characterized in that mass data are mined by means of an intelligent instrument and a computer storage technology, and the number of the resume of the internal information in the process data is found, so that a model is built to monitor the data and judge the fault state of the equipment. The intelligent equipment is relied on to analyze and mine the data, so that the internal condition of the equipment cannot be well explained, and the learning data-driven algorithm of the machine has narrow application range and weak technology, and cannot be widely applied to fault early warning of the equipment.
In recent years, new life prediction methods have been developed, and there are a lot of methods that have been used in practice and some that are still in the experimental prediction stage. Currently, there are two main methods for life prediction, indirect and direct measurement.
The indirect life prediction method is based on parameter data of the component, calculates the damage degree of the component, depends on complete and real data of the component during operation, and ignores the aging factor of the material.
The direct life prediction method comprises a non-destructive test method and a destructive test method, wherein the destructive test method needs to obtain the same or similar samples, obtain required data through a destructive test, analyze the data, calculate the damage degree of the life and evaluate the life. The nondestructive test method can diagnose more parts in a short time, can monitor periodically and has narrow application range.
The research on the mechanisms of the methods is mature, the development and application of the device are feasible, but the accuracy of the life prediction is still to be further improved, and the application range is also to be improved.
Disclosure of Invention
The invention aims to predict whether equipment is abnormal or not and the residual life of the equipment, prevent the equipment from being out of order and guarantee the safety of lives and properties of people.
The purpose of the invention can be achieved by adopting the following technical scheme:
a fault early warning and service life prediction method for escalator equipment comprises the following steps:
s1, establishing a database relation table by adopting an SQL statement, storing the database relation table in an Oracle database, and connecting Matlab with the Oracle database through an ODBC bridge to access data to complete the database construction;
s2, analyzing the basic structure of the elevator, determining key components causing elevator faults, screening important factors influencing the elevator as indexes for judging the elevator faults, and performing simulation on index data;
s3, collecting elevator real-time data, constructing a Dynamic Bayesian Network (DBN), solving the pre-estimated fault probability of each node of the dynamic Bayesian network through a Monte Carlo algorithm, forming a CPT fault probability, and giving an early warning to the elevator fault;
and S4, calling elevator index data in an Oracle database, generating a sequence of the abnormal degree and the running time of the elevator, obtaining an abnormal degree time function, and determining the running time of the elevator to be compared with an abnormal degree time threshold value through the abnormal degree time function, thereby realizing the escalator service life prediction algorithm.
Further, the step S1 includes:
s101, compiling equipment health evaluation table ELEIDX, fault diagnosis table FAULTDB and data point table ELE formats by adopting SQL statements and combining with pivot and decode in Oracle, and establishing an original interface point table select into a series of relation tables which are stored in an Oracle database;
s102, constructing a data system by utilizing a DeviceType table, an Index table, an Indexdb table, a Basicdb table and a Basic table according to Basic data and fault types of equipment and equipment components, and designing a health assessment and fault diagnosis database part by adopting a database with a hierarchical structure, wherein the DeviceType table stores the equipment types and the equipment components and corresponding numbers thereof, the Index table stores equipment refinement Index types, the Indexdb table records real-time and recent indexes, the Basicdb table is used for recording real-time and recently collected Basic data values, and the Basic table is used for expressing the various Basic data types and the relation between the various parts and component indexes;
s103, a health assessment database is constructed by utilizing the equipment fault assessment Index table Elec _ Index and the Basic data table Elec _ Basic, an upper-layer relation and a lower-layer relation exist between the two tables, data in the equipment fault assessment Index table Elec _ Index are calculated according to the Basic data table Elec _ Basic, updating of the Basic data table causes updating of data of the Index table, and the capacities of the two tables can be added;
s104, constructing a fault diagnosis database by using the equipment fault diagnosis table Faultdb and the Basic data table Elec _ Basic, wherein the upper-layer and lower-layer relations exist between the two tables, data in the equipment fault diagnosis table Faultdb are calculated according to the Basic data table Elec _ Basic, the updating of the Basic data table causes the updating of data of the fault diagnosis table, and the capacities of the two tables can be added;
s105, installing Oracle client software on a client machine, correctly configuring a tnsname.
And S106, the database and the Matlab are connected through the ODBC bridge, data in the database are inquired through exec, a fetch function returns upper-layer variables, and the update updates a health evaluation table and a fault diagnosis table in the database.
Further, the step S2 includes:
s201, analyzing structural traction driving, a suspension device, a car frame, a car, a door system, a weight balance system, an electric traction system, an electric system and a safety protection system of the elevator, and designing indexes;
s202, calling a Devicetype table, an Index table and a Basic table in a database, and performing big data analysis on the collected elevator data in an offline mode to form a Basic table and an Indexdb table which are stored in the database;
s203, referring to a Tinsen TE-evolution1 elevator structure, dividing research objects into a tractor, a steel wire rope and a door system, proposing corresponding indexes for each research object, and determining fault diagnosis indexes as the wear degree of a change gear in the door system and the diameter of the steel wire rope;
s204, referring to a measurable wear model in an IBM algorithm aiming at the wear degree index of the change gear, and predicting the wear exceeding the original surface roughness depth;
s205, calculating the change analysis of the diameter by adopting a finite element theory according to the diameter index of the steel wire rope.
Further, the step S3 includes:
s301, data collection, namely calling an Indexdb table, a DeviceType table and an Index table which comprise equipment historical Index data from a database;
s302, determining network node variables and causal relationships, wherein the network node variables and causal relationships comprise node variable naming, description of each part of the system, measurement of degradation quantity and selection of variable types (discrete and continuous);
s303, constructing an elevator DBN model, analyzing the coupling relation of all parts of the escalator, combining an escalator fault diagnosis index extraction module, constructing an escalator Bayesian network fault diagnosis model graph, wherein a Bayesian network is represented in a BNT in a matrix mode, namely if nodes i to j have an arc, the value (i, j) in the corresponding matrix is 1, otherwise, the value is 0, establishing a Bayesian network BNT, substituting a sample as a training set into a learn _ params () function to learn, and learning to obtain a conditional probability CPT;
s304, according to the probability of the failure of the components estimated by the trained Bayesian network, the calculation result is stored in a database health condition evaluation table and a failure diagnosis table.
Further, the step S4 includes:
s401, extracting real-time equipment monitoring data from a database equipment health condition evaluation table through a Bayesian fault diagnosis model and a CPT table of a network node, evaluating the health condition of equipment, and outputting equipment abnormality degree, namely fault probability;
s402, calling historical data in the health evaluation table to generate an equipment abnormality degree time sequence, and determining the order of a fitting function by using a Matlab curve fitting tool box CFTOOL to obtain a fitting function of the equipment abnormality degree and the running time;
s403, calling a fitting function of the abnormality degree and the running time of the equipment, setting an abnormality degree threshold value a corresponding to equipment scrapping, wherein the corresponding service life is x, applying an expert system to carry out real-time fault probability analysis on the system to obtain the abnormality degree b of the equipment at the moment, calculating the used time y of the equipment at the moment according to the fitting function of the abnormality degree and the running time of the equipment, and realizing the service life prediction of the system, wherein the residual service life w of the equipment is x-y.
Further, the step S204 includes:
s2041, basic assumption of a measurable abrasion model,
the relationship between the amount of wear Q and the number of passes N and the amount of energy consumed E is a differential equation based on the assumption that the amount of wear is a function of two variables, the amount of energy consumed E and the number of passes N, for each pass
Wherein the relationship between the wear amount and the number of passes is
Wherein m is a constant given by the system, and when no lubrication is available,
s2042, according to the relation between consumed energy and a wear process, the wear can be divided into A-type wear and B-type wear, wherein damage energy of the A-type wear in the wear process can be kept unchanged, the wear belongs to severe wear, the wear is applied to the conditions of dry friction and heavy load, damage of the B-type wear in the wear process changes along with the change of passing times, the damage is a transfer result of a medium degree, and for the wear of a door hanging wheel, the wear is expressed by the following differential equation
In the above formula, C is a constant coefficient of the system, and is obtained by measuring the abrasion loss under a certain passing frequency through experiments or by combining a model with zero abrasion and measurable abrasion, N is the passing frequency, S is the length of the contact surface along the sliding direction, Q is the cross-sectional area of the grinding mark,
assuming destructive power and taumxS is proportional, the contact area of the two objects changes with the abrasion, so taumxS is not a constant value and is,
further, step S205 is specifically as follows:
the diameter of the wire rope is measured with a vernier caliper having a jaw whose width is at least sufficient to span two adjacent strands, the measurement is performed on a straight portion of the wire rope outside the 15m end, two values are measured on two cross sections at a distance of at least 1m and perpendicularly to each other on the same cross section, and the average of the four measurements is the measured diameter of the wire rope.
Further, the building of the elevator DBN model in step S303 is specifically as follows:
s3031, constructing a prior Bayesian network by the determined nodes and the prior data;
s3032, calling data to perform parameter learning on the DBN, and constructing a posterior Bayesian network;
s3033, calculating the fault probability of each node in the DBN model by using a Monte Carlo algorithm, and determining the CPT probability distribution table of all network nodes.
Further, the parameter learning is learning by adopting an EM algorithm lear _ params _ EM of multiple iterative optimization, wherein the process of solving the optimal structure by the EM algorithm is to converge to local optimal parameters
The process of (1) generating a sample training set through a random number generator, filling up lost data, and carrying out maximum likelihood estimation learning simulation on the data to simulate parameters which most accord with a structure, specifically comprises the following steps:
step E, namely expecting:
wherein E is an expected value; d is a training sample set;
representing the sought optimal structural parameter, XiThe value range of (1) is:
qiis to configure piiThe arrangement order of (a); n is a radical ofijkIs to satisfy the variable value in the data set D
And piiNumber of occurrences of condition j:
ylis the number of data lost in D; shIs a bayesian network structure selection hypothesis;
m steps, i.e. maximum estimation, maximum likelihood estimation function:
maximum a posteriori estimation function:
N′ijkis a priori sufficient statistical factor, NijkIs the sufficient statistical factor of sample data, i, j, k, h, q belongs to N.
Further, the parameter learning is performed by using an EM algorithm that is iteratively optimized for multiple times, wherein the EM algorithm specifically performs the following process: setting an initial value of a variable; starting iteration from a certain initial Bayesian network; calling a joint tree reasoning algorithm to complete the reasoning operation on the Bayesian network to obtain the current optimal network; completing the data set by utilizing an EM algorithm based on the current optimal network to obtain a complete data set so as to realize maximization of parameters; calculating the sum of the value numbers of all nodes in the network; creating network structures different from the initial network, and taking the structures as candidate network structures; scoring the candidate network structures by using a BIC scoring function, and searching a parameter which enables the score to be maximum from the selected network structures; the best network structure to fit to the data set is found.
Compared with the prior art, the invention has the following advantages and effects:
1. the elevator fault prediction system can update data in real time and process the data according to data acquisition to obtain the fault probability of the elevator and fault components when the elevator is in fault, can predict the service life according to the real-time data, and can realize the health condition of the elevator to workers at the first time.
2. The method combines the IBM algorithm and the finite element theory algorithm in the process of establishing the fault characteristic index, effectively improves the efficiency and the precision of data mining, and thus ensures that the index is more reasonable.
3. In the escalator fault diagnosis calculation, a Bayesian network intelligent algorithm is combined with an escalator fault diagnosis index extraction technology, a Bayesian network fault diagnosis model graph of an elevator is built, data are brought into a program for training by combining an EM algorithm, an SEM algorithm and the like, a condition probability table is obtained, and the current elevator health condition can be evaluated. In order to improve the operation rate of the Bayesian network, an approximate inference algorithm of random sampling and an importance sampling method are introduced.
Drawings
FIG. 1 is a basic framework architecture of the device failure early warning and life prediction system research methodology disclosed in the present invention;
FIG. 2 is a flow chart of a method of elevator fault early warning and life prediction;
FIG. 3 is a schematic diagram of an underlying data table format;
FIG. 4 is a schematic diagram of a fault diagnosis table format;
FIG. 5 is a schematic diagram of a health assessment table format;
FIG. 6 is a device database relationship diagram;
FIG. 7 is a schematic view of an assessment and diagnosis database;
fig. 8 is a block diagram of the main structure of a TE-evolution1 elevator;
FIG. 9 is a schematic representation of a wire rope diameter measurement;
fig. 10 is a diagram of a bayesian network fault diagnosis model of an elevator.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Examples
The embodiment discloses a method for researching a system for early warning of equipment failure and predicting service life, which has a structure shown in fig. 1 and fig. 2, wherein all steps of the embodiment are completed based on a Matlab development environment and an oval database, and the method specifically comprises the following steps:
s1, using SQL sentences to establish a database relation table, storing the database relation table in an Oracle database, connecting Matlab with the Oracle database through an ODBC bridge to access data, and completing database construction;
the method comprises the following steps:
s101, as shown in the figures 3, 4 and 5, writing equipment health evaluation table ELEIDX, fault diagnosis table FAULTDB and data point table ELE formats by utilizing SQL statements according to scheme requirements, building an original interface point table select into a series of relation tables, and storing the relation tables in an Oracle database for storing collected equipment real-time data;
s102, as shown in FIG. 6, a data system is constructed by utilizing 5 tables according to the basic data and the fault types of the equipment and the equipment components, and the DeviceType table stores the equipment types and the equipment components and the corresponding numbers thereof; the Index table stores the detailed Index types of the equipment; the Indexdb table records real-time and recent indexes; the Basicdb table is used for recording base data values collected in real time and in the near future; the Basic table is used for expressing various base data categories and the relation between various components and component indexes;
for the health assessment and troubleshooting database portion, a hierarchical database design scheme (e.g., FIG. 7) is employed.
S103, as shown in tables 1 and 2, the health assessment database is constructed by using an equipment fault assessment Index table (Flec _ Index) and a Basic data table (Flec _ Basic), the upper layer and the lower layer of relation between the two tables is formed, data in the fault assessment Index table are calculated according to the Basic data table, the data in the Basic table can be updated to cause the data in the Index table to be updated, and the capacities of the two tables can be increased;
TABLE 1 health evaluation Table ELEIDX
| FN | Percentage of brake power anomaly | Brake input voltage anomaly percentage | Working temperature difference of traction machine | Wire rope diameter anomaly | Abnormal degree of door lock relay state |
| FVALUE66 | 01 | 0.1 | 0 | 0 | 0 |
| FVALUE67 | 0 | 0 | 0.5 | 0 | 0 |
| FVALUE68 | 0 | 0 | 0.3 | 1 | 0 |
| FVALUE69 | 02 | 0 | 0 | 01 | 0 |
| FVALUE70 | 01 | 0 | 0.2 | 0 | 0 |
| FVALUE71 | 01 | 0 | 0 | 0 | 0 |
| FVALUE72 | 0 | 0 | 0.5 | 0 | 0 |
| FVALUE73 | 0 | 0 | 0 | 0 | 1 |
| FVALUE74 | 0 | 0 | 0 | 0 | 0 |
| FVALUE75 | 03 | 0 | 0 | 0 | 1 |
| FVALUE76 | 0 | 0.3 | 0.2 | 0 | 0 |
| FVALUE77 | 01 | 0 | 0 | 0.4 | 0 |
| FVALUE78 | 02 | 0 | 0.1 | 01 | 0 |
| FVALUE79 | 01 | 0 | 0.5 | 0 | 0 |
| FVALUE80 |
TABLE 2 basic data Point Table FIJE
S104, as shown in the table 3, the fault diagnosis database is constructed by using a device fault diagnosis table (FAULTDB) and a Basic data table (Elec _ Basic), the upper and lower layer relation between the two tables is realized, data in the fault diagnosis table (FAULTDB) is calculated according to the Basic data table, the updating of the Basic data table causes the updating of the fault diagnosis table, and the capacities of the two tables can be increased;
TABLE 3 Fault diagnosis Table FAULTDB
And S105, installing client software of the Oracle database on the client machine. Correctly configuring tnsname.
S2, knowing the basic structure of the elevator, determining key components causing elevator faults, screening important factors influencing the elevator as indexes for judging the elevator faults, and performing simulation on index data;
s201, knowing the structure traction drive, the suspension device, the car frame and the car, a door system, a weight balance system, an electric traction system, an electric system, a safety protection system and related parameters of the elevator, looking up a large amount of literature data, designing corresponding indexes as shown in a table 4, and judging the health condition of equipment through the change of index data in later-stage calculation;
TABLE 4 Ttson elevator without machine room TE-Evolution1 index table
S202, calling a Devicetype table, an Index table and a Basicdb table of an Oracle database design module. Performing big data analysis in an off-line mode according to the collected elevator data to form a Basic table and an Indexdb table, storing the Basic table and the Indexdb table into a database, comparing the collected data serving as the latest data with other data in the table, and analyzing the running state of equipment;
s203, referring to a Tinsen TE-evolution1 elevator structure, dividing main research objects into three modules of a tractor, a steel wire rope and a door system on the basis of relevant factors of elevator faults, providing corresponding indexes for each module, synthesizing the influence on the elevator, determining main factors causing the elevator faults, determining fault diagnosis indexes as the abrasion degree of a door hanging wheel in a door system and the diameter of the steel wire rope, determining the abrasion degree of the wheel hanging wheel in the door system and the diameter threshold of the steel wire rope, and determining that the equipment faults can be determined when the corresponding indexes of the equipment obtained by the system through calculation of collected real-time data exceed the threshold range;
and S204, referring to a measurable wear model in an IBM algorithm aiming at the wear degree index of the change gear, and predicting the wear exceeding the original surface roughness depth.
1) Basic assumptions for measurable wear models
According to the assumption that the abrasion amount is a function of two variables of the energy consumed by abrasion generated by each pass number E and the pass number N. The relation between the wear amount Q and the number of passes N and the consumed energy E is a differential equation
Experiments show that the relation between the abrasion loss and the passing times is
Where m is a constant given by the system, the literature indicates that, in the absence of lubrication,
2) based on the relationship between the consumed energy and the abrasion process, the abrasion process can be classified into type A abrasion and type B abrasion.
Wherein, the A-type abrasion means that the damage can be kept unchanged all the time in the abrasion process, and belongs to serious abrasion. The method is mainly applied to dry friction and heavy load conditions. Details are not described here.
Among them, type B abrasion means that the damage during abrasion varies with the number of passes, which is a moderate transfer result. This is the case with the wear of the door hanger wheels of this study. This wear can be expressed by the following differential equation
In the formula, C is a constant coefficient of the system, and generally must be obtained by measuring the wear amount at a certain number of passes through experiments. The model of zero wear and measurable wear can also be combined to obtain the analytic expression of C, N is the passing frequency, S is the length of the contact surface along the sliding direction, and Q is the cross-sectional area of the grinding mark.
In this wear model, it has been assumed that the failure work is associated with τmxS is proportional, τ because the contact area of two objects changes with wearmxS is not a constant.
S205, calculating the change analysis of the diameter by adopting a finite element theory according to the diameter index of the steel wire rope. The diameter of the steel wire rope is measured by a vernier caliper with a jaw, as shown in fig. 9, the jaw has the minimum width enough to span two adjacent strands, namely AB, and at least comprises two strands, the measurement is carried out on a straight part outside 15m of the end of the steel wire rope, namely a measurement point A is more than 15 meters away from a rope end C, a measurement point B is more than 15 meters away from a rope end D, two numerical value rotating vernier calipers are vertically measured on the same section at least 1m away from the rope end C, the measurement is carried out according to the left and right vertical directions, and the average value of four measurement results is the measured diameter of the steel wire rope.
S3, collecting elevator real-time data, constructing a Dynamic Bayesian Network (DBN), solving the pre-estimated fault probability of each node of the DBN through a Monte Carlo algorithm, forming a CPT (fault probability), and giving early warning to the elevator fault;
and carrying out secondary development by utilizing a Matlab technical framework, using a Bayesian network tool kit (FULLBNT), constructing a dynamic Bayesian network by adopting a Bayesian network algorithm, calculating probability distribution of a DBN (dynamic Bayesian network), deducing the state of the escalator and predicting the residual life.
S301, data collection: establishing a data system according to the step S2, calling an Indexdb table, a Devicetype table and an Index table which comprise equipment historical Index data from a database to train a dynamic Bayesian network, and updating an equipment health evaluation table ELEIDX and a fault diagnosis table FAULTDB in the database;
s302, determining node variables and causal relations: the network nodes represent random variables, the nodes of the Bayesian network are allowed to be any variables, and the node variables can be divided into opportunity variables, decision variables and auxiliary variables, and can be continuous variables or discrete variables. Determining the node variable and the causal relationship comprises the steps of naming the node variable, describing each part of the system and measuring the degradation quantity, and selecting the type (discrete and continuous) of the variable;
s303, constructing a DBN: an escalator Bayesian network fault diagnosis model graph is constructed by analyzing the coupling relation of all parts of the escalator and combining an escalator fault diagnosis index extraction module, the Bayesian network fault diagnosis model graph of the elevator is shown in the graph 10, an arc exists between the medium loss of a node and the insulation aging of the node, namely when the matrix in the BNT represents the Bayesian network, the value of a corresponding matrix (medium loss and insulation aging) is 1, and otherwise, the value is 0. And inputting the conditions of all the nodes into a BNT matrix, establishing a Bayesian network BNT, substituting the obtained sample matrix as a training set into a least _ params () function for learning, and learning to obtain a conditional probability table CPT. With the increase of the number of training samples, the learned conditional probability table is more and more approximate to the real conditional probability table;
the method comprises the following specific steps of building an elevator DBN model:
1) and constructing a prior Bayesian network by the determined nodes and the prior data.
2) And calling data to carry out parameter learning on the DBN, and constructing a posterior Bayesian network.
Parameter learning algorithm function: the BNT provides a rich parameter learning function, and in the case of data loss and known network topology, the parameter learning is performed by adopting an EM algorithm, namely, lean _ params _ EM, which is optimized by multiple iterations.
The process of solving the optimal structure by the EM algorithm is actually converged to the local optimal parameters
The process of (1). First, a sample training set is generated by a random number generator (applied in Matlab tool) to fill in the missing data. Carrying out maximum likelihood estimation learning on data to simulate parameters which most accord with a structure, and specifically comprising the following steps:
step E (expected):
wherein E is an expected value; d is a training sample set;
representing the sought optimal structural parameter, XiThe value range of (1) is:
qiis to configure piiThe arrangement order of (a); n is a radical ofijkIs to satisfy the variable value in the data set D
And piiNumber of occurrences of condition j:
ylis the number of data lost in D; shIs a bayesian network structure selection hypothesis.
M steps (maximum estimate):
maximum likelihood estimation function:
maximum a posteriori estimation function:
N'ijkis a priori sufficient statistical factor; corresponding to, NijkIs the sufficient statistical factor of sample data, i, j, k, h, q belongs to N.
And the SEM algorithm, the optimization algorithm of the EM algorithm and the structured expectation maximization SEM algorithm. The application process of the SEM algorithm is roughly divided into two steps of structure searching and parameter learning. And (4) searching the network structure, replacing the nonexistent sufficient statistical factors by the SEM algorithm with the expected sufficient statistical factors to enable the form of the scoring function to have resolvability, and then, carrying out local search to find the structure with higher network score. Finally, the parameter with the largest score is screened on the selected structure.
The specific implementation process is as follows: setting an initial value of a variable; starting iteration from a certain initial Bayesian network; calling a joint tree reasoning algorithm to complete the reasoning operation on the Bayesian network to obtain the current optimal network; completing the data set by utilizing an EM algorithm based on the current optimal network to obtain a complete data set so as to realize maximization of parameters; calculating the sum of the value numbers of all nodes in the network; creating all network structures different from the initial network, and taking the structures as candidate network structures; scoring the candidate network structures by using a BIC scoring function, and searching a parameter which enables the score to be maximum from the selected network structures; the best network structure to fit to the data set is derived.
3) And calculating the fault probability of each node in the DBN model by using a Monte Carlo algorithm, thereby determining a CPT (probability distribution) table of all network nodes.
S304, according to the probability of the faults of the components estimated by the trained Bayesian network, the calculation results are stored in a database health condition evaluation table and a fault diagnosis table, the database is updated, real-time detection is realized, and equipment faults are found in time.
S4, calling elevator index data in the database, generating a sequence of the degree of abnormality and the operation time of the elevator, obtaining a time function of the degree of abnormality, and determining the comparison between the operation time of the elevator and a time threshold value of the degree of abnormality according to the time function of the degree of abnormality, thereby realizing the life prediction algorithm of the escalator.
Collecting expert knowledge, compiling an expert system, analyzing the state of the elevator through the expert system according to the real-time analysis of the state of the system, and predicting the service life.
S401, extracting real-time equipment monitoring data from a database equipment health condition evaluation table through a Bayesian fault diagnosis model and a CPT table of a network node in the step S3, evaluating the health condition of the equipment, and outputting the abnormal degree (fault probability) of the equipment;
s402, calling historical data in the health evaluation table to generate an equipment abnormality degree time sequence, using a Matlab curve fitting tool box CFTOOL to determine the order of a fitting function, obtaining the fitting function of the equipment abnormality degree and the running time, determining the general running state of the equipment, and calculating the residual life of the equipment;
s403, calling a fitting function of the abnormality degree and the running time of the equipment, setting an abnormality degree threshold value a corresponding to equipment scrapping, wherein the corresponding service life is x, applying an expert system to carry out real-time fault probability analysis on the system to obtain the abnormality degree b (b is less than or equal to a) of the equipment at the moment, when b is less than a, calculating the running time y of the equipment when the abnormality degree is b through the abnormality degree and running time fitting function, and then setting the residual service life w of the equipment to be x-y; when b is equal to a, the residual life is 0, and life prediction of the system is realized.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.
Claims (9)
1. A fault early warning and service life prediction method for escalator equipment is characterized by comprising the following steps:
s1, establishing a database relation table by adopting an SQL statement, storing the database relation table in an Oracle database, and connecting Matlab with the Oracle database through an ODBC bridge to access data to complete the database construction;
s2, analyzing the basic structure of the elevator, determining key components causing elevator faults, screening important factors influencing the elevator as indexes for judging the elevator faults, and performing simulation on index data;
s3, collecting elevator real-time data, constructing a dynamic Bayesian network, solving the pre-estimated fault probability of each node of the dynamic Bayesian network through a Monte Carlo algorithm, forming a CPT fault probability, and giving an early warning to the elevator fault;
s4, calling elevator index data in an Oracle database, generating a sequence of the degree of abnormality and the running time of the elevator, obtaining an abnormality time function, and determining the running time of the elevator to be compared with an abnormality time threshold value through the abnormality time function, so as to realize the escalator service life prediction algorithm, wherein the step comprises the following steps:
s401, extracting real-time equipment monitoring data from a database equipment health condition evaluation table through a Bayesian fault diagnosis model and a CPT table of a network node, evaluating the health condition of equipment, and outputting equipment abnormality degree, namely fault probability;
s402, calling historical data in the health evaluation table to generate an equipment abnormality degree time sequence, and determining the order of a fitting function by using a Matlab curve fitting tool box CFTOOL to obtain a fitting function of the equipment abnormality degree and the running time;
s403, calling a fitting function of the abnormality degree and the running time of the equipment, setting an abnormality degree threshold value a corresponding to equipment scrapping, wherein the corresponding service life is x, applying an expert system to carry out real-time fault probability analysis on the system to obtain the abnormality degree b of the equipment at the moment, calculating the used time y of the equipment at the moment according to the fitting function of the abnormality degree and the running time of the equipment, and realizing the service life prediction of the system, wherein the residual service life w of the equipment is x-y.
2. The method for predicting the malfunction advance and the life span of an escalator installation according to claim 1, wherein said step S1 comprises:
s101, compiling equipment health evaluation table ELEIDX, fault diagnosis table FAULTDB and data point table ELE formats by adopting SQL statements and combining with pivot and decode in Oracle, and establishing an original interface point table select into a series of relation tables which are stored in an Oracle database;
s102, constructing a data system by utilizing a DeviceType table, an Index table, an Indexdb table, a Basicdb table and a Basic table according to Basic data and fault types of equipment and equipment components, and designing a health assessment and fault diagnosis database part by adopting a database with a hierarchical structure, wherein the DeviceType table stores the equipment types and the equipment components and corresponding numbers thereof, the Index table stores equipment refinement Index types, the Indexdb table records real-time and recent indexes, the Basicdb table is used for recording real-time and recently collected Basic data values, and the Basic table is used for expressing the various Basic data types and the relation between the various parts and component indexes;
s103, a health assessment database is constructed by utilizing the equipment fault assessment Index table Elec _ Index and the Basic data table Elec _ Basic, an upper-layer relation and a lower-layer relation exist between the two tables, data in the equipment fault assessment Index table Elec _ Index are calculated according to the Basic data table Elec _ Basic, updating of the Basic data table causes updating of data of the Index table, and the capacities of the two tables can be added;
s104, constructing a fault diagnosis database by using the equipment fault diagnosis table Faultdb and the Basic data table Elec _ Basic, wherein the upper-layer and lower-layer relations exist between the two tables, data in the equipment fault diagnosis table Faultdb are calculated according to the Basic data table Elec _ Basic, the updating of the Basic data table causes the updating of data of the fault diagnosis table, and the capacities of the two tables can be added;
s105, installing Oracle client software on a client machine, correctly configuring a tnsname.
And S106, the database and the Matlab are connected through the ODBC bridge, data in the database are inquired through exec, a fetch function returns upper-layer variables, and the update updates a health evaluation table and a fault diagnosis table in the database.
3. The method for predicting the malfunction advance and the life span of an escalator installation according to claim 2, wherein said step S2 comprises:
s201, analyzing structural traction driving, a suspension device, a car frame, a car, a door system, a weight balance system, an electric traction system, an electric system and a safety protection system of the elevator, and designing indexes;
s202, calling a Devicetype table, an Index table and a Basic table in a database, and performing big data analysis on the collected elevator data in an offline mode to form a Basic table and an Indexdb table which are stored in the database;
s203, dividing the research objects into a tractor, a steel wire rope and a door system, proposing corresponding indexes for each research object, and determining the fault diagnosis indexes as the wear degree of a change gear in the door system and the diameter of the steel wire rope;
s204, referring to a measurable wear model in an IBM algorithm aiming at the wear degree index of the change gear, and predicting the wear exceeding the original surface roughness depth;
s205, calculating the change analysis of the diameter by adopting a finite element theory according to the diameter index of the steel wire rope.
4. The method for predicting the malfunction advance and the life span of an escalator installation according to claim 2, wherein said step S3 comprises:
s301, data collection, namely calling an Indexdb table, a DeviceType table and an Index table which comprise equipment historical Index data from a database;
s302, determining network node variables and causal relationships, wherein the network node variables and causal relationships comprise node variable naming, description of each part of the system and measurement of degradation quantity, and selection of discrete or continuous variable types;
s303, constructing an elevator DBN model, analyzing the coupling relation of all parts of the escalator, combining an escalator fault diagnosis index extraction module, constructing an escalator Bayesian network fault diagnosis model graph, wherein a Bayesian network is represented in a BNT in a matrix mode, namely if nodes i to j have an arc, the value (i, j) in the corresponding matrix is 1, otherwise, the value is 0, establishing a Bayesian network BNT, substituting a sample as a training set into a learn _ params () function to learn, and learning to obtain a conditional probability CPT;
s304, according to the probability of the failure of the components estimated by the trained Bayesian network, the calculation result is stored in a database health condition evaluation table and a failure diagnosis table.
5. The method as claimed in claim 3, wherein the step S204 includes:
s2041, basic assumption of a measurable abrasion model,
the relationship between the amount of wear Q and the number of passes N and the amount of energy consumed E is a differential equation based on the assumption that the amount of wear is a function of two variables, the amount of energy consumed E and the number of passes N, for each pass
Wherein the relationship between the wear amount and the number of passes is
Wherein m is a constant given by the system, and when no lubrication is available,
s2042, according to the relation between consumed energy and a wear process, the wear can be divided into A-type wear and B-type wear, wherein damage energy of the A-type wear in the wear process can be kept unchanged, the wear belongs to severe wear, the wear is applied to the conditions of dry friction and heavy load, damage of the B-type wear in the wear process changes along with the change of passing times, the damage is a transfer result of a medium degree, and for the wear of a door hanging wheel, the wear is expressed by the following differential equation
In the above formula, C is a constant coefficient of the system, and is obtained by measuring the abrasion loss under a certain passing frequency through experiments or by combining a model with zero abrasion and measurable abrasion, N is the passing frequency, S is the length of the contact surface along the sliding direction, Q is the cross-sectional area of the grinding mark,
assuming destructive power and taumxS is proportional, the contact area of the two objects changes with the abrasion, so taumxS is not a constant value and is,
6. the method for fault pre-warning and life prediction of escalator equipment according to claim 3, wherein said step S205 is as follows:
the diameter of the wire rope is measured with a vernier caliper having a jaw whose width is at least sufficient to span two adjacent strands, the measurement is performed on a straight portion of the wire rope outside the 15m end, two values are measured on two cross sections at a distance of at least 1m and perpendicularly to each other on the same cross section, and the average of the four measurements is the measured diameter of the wire rope.
7. The method for fault early warning and life prediction of escalator equipment as claimed in claim 4, wherein said step S303 of constructing an elevator DBN model is as follows:
s3031, constructing a prior Bayesian network by the determined nodes and the prior data;
s3032, calling data to perform parameter learning on the DBN, and constructing a posterior Bayesian network;
s3033, calculating the fault probability of each node in the DBN model by using a Monte Carlo algorithm, and determining the CPT probability distribution table of all network nodes.
8. The method of claim 7, wherein the parameter learning is learning by using a multi-iteration optimized EM algorithm least _ params _ EM, and wherein the EM algorithm for solving the optimal structure converges to a local optimal parameter
The process of (1) generating a sample training set through a random number generator, filling up lost data, and carrying out maximum likelihood estimation learning simulation on the data to simulate parameters which most accord with a structure, specifically comprises the following steps:
step E, namely expecting:
wherein E is an expected value; d is a training sample set;
representing the sought optimal structural parameter, XiThe value range of (1) is:
qiis to configure piiThe arrangement order of (a); n is a radical ofijkIs to satisfy the variable value in the data set D
And piiNumber of occurrences of condition j:
ylis the number of data lost in D; shIs a bayesian network structure selection hypothesis;
m steps, i.e. maximum estimation, maximum likelihood estimation function:
maximum a posteriori estimation function:
N′ijkis a priori sufficient statistical factor, NijkIs the sufficient statistical factor of sample data, i, j, k, h, q belongs to N.
9. The method for fault pre-warning and life prediction of escalator equipment as claimed in claim 7, wherein said parameter learning is performed by using an EM algorithm with multiple iterative optimization, wherein the EM algorithm is specifically performed as follows: setting an initial value of a variable; starting iteration from a certain initial Bayesian network; calling a joint tree reasoning algorithm to complete the reasoning operation on the Bayesian network to obtain the current optimal network; completing the data set by utilizing an EM algorithm based on the current optimal network to obtain a complete data set so as to realize maximization of parameters; calculating the sum of the value numbers of all nodes in the network; creating network structures different from the initial network, and taking the structures as candidate network structures; scoring the candidate network structures by using a BIC scoring function, and searching a parameter which enables the score to be maximum from the selected network structures; the best network structure to fit to the data set is found.
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 CN107991870A (en) | 2018-05-04 |
| CN107991870B true 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) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108469783B (en) * | 2018-05-14 | 2021-02-02 | 西北工业大学 | Deep hole roundness error prediction method based on Bayesian network |
| CN109034010B (en) * | 2018-07-06 | 2021-05-14 | 北京天泽智云科技有限公司 | On-line prediction method for lubrication failure of automatic door system |
| CN109271705B (en) * | 2018-09-14 | 2023-03-24 | 湘潭大学 | Machine predictive maintenance method based on deep learning |
| 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 |
| CN109872003B (en) * | 2019-03-06 | 2021-08-13 | 中国科学院软件研究所 | Object state prediction method, object state prediction system, computer device, and storage medium |
| CN110133508B (en) * | 2019-04-24 | 2022-04-01 | 上海博强微电子有限公司 | Safety early warning method for power battery of electric automobile |
| CN112668729A (en) * | 2019-10-15 | 2021-04-16 | 深圳怡化电脑股份有限公司 | Equipment data analysis method and device, computer equipment and readable storage medium |
| CN111275306B (en) * | 2020-01-16 | 2023-12-05 | 上海三菱电梯有限公司 | Elevator reconstruction evaluation method and elevator reconstruction evaluation system |
| CN112390127B (en) * | 2020-12-12 | 2022-12-23 | 中铁第四勘察设计院集团有限公司 | Health degree model-based preventive maintenance strategy generation method for escalator |
| CN113377595B (en) * | 2021-06-11 | 2023-06-23 | 上海壁仞智能科技有限公司 | Fault diagnosis method, device, electronic equipment and storage medium |
| CN113720716B (en) * | 2021-09-01 | 2022-08-19 | 桂林电子科技大学 | Quantitative analysis and service life prediction method for wear degree of elevator traction sheave |
| CN114297255B (en) * | 2021-12-17 | 2024-04-19 | 中电信数智科技有限公司 | Network quality work order fault early warning method based on log analysis |
| CN114418101B (en) * | 2022-01-11 | 2024-05-31 | 中国人民解放军国防科技大学 | Bayesian network reasoning method and system |
| CN114715752B (en) * | 2022-06-08 | 2022-08-23 | 凯尔菱电（山东）电梯有限公司 | Abnormity detection method and system for elevator |
| CN114781762B (en) * | 2022-06-21 | 2022-09-23 | 四川观想科技股份有限公司 | Equipment fault prediction method based on life consumption |
| CN115659812B (en) * | 2022-10-29 | 2023-06-09 | 思维实创(哈尔滨)科技有限公司 | Escalator life prediction method, system, equipment and medium based on urban rail ISCS |
| CN115557349B (en) * | 2022-12-05 | 2023-03-14 | 苏州大名府电梯有限公司 | Intelligent home elevator based on Internet of things and detection method |
| CN116308300B (en) * | 2023-05-11 | 2023-07-21 | 四川新迎顺信息技术股份有限公司 | Power equipment state monitoring evaluation and command method and system |
Citations (9)
| 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 |
| 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 |
| 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 |
| 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 |
-
2017
- 2017-12-05 CN CN201711263968.3A patent/CN107991870B/en active Active
Patent Citations (9)
| 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 |
| 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 |
| 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 |
| 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)
| Title |
|---|
| 动态贝叶斯网络在设备剩余寿命预测中的应用研究;杨志波等;《计算机集成制造系统》;20070930;全文 * |
| 基于动态贝叶斯网络的设备故障预测方法研究;张星辉等;《科技广场》;20100531;参见第31页 * |
| 基于贝叶斯模型的电梯安全评估方法研究;冯鑫等;《起重运输机械》;20160520;全文 * |
| 视情维修条件下的多状态控制单元可用性建模与分析;李志强等;《兵工学报》;20171130;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107991870A (en) | 2018-05-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107991870B (en) | Fault early warning and service life prediction method for escalator equipment | |
| CN109886430B (en) | Equipment health state assessment and prediction method based on industrial big data | |
| CN110764493B (en) | PHM application system, method and storage medium suitable for high-speed railway | |
| Wang et al. | Real-time defect detection of high-speed train wheels by using Bayesian forecasting and dynamic model | |
| Di Maio et al. | Combining relevance vector machines and exponential regression for bearing residual life estimation | |
| Tobon-Mejia et al. | A data-driven failure prognostics method based on mixture of Gaussians hidden Markov models | |
| CN103617110B (en) | Server device condition maintenance system | |
| US20220057367A1 (en) | Method for evaluating pipe condition | |
| Dong et al. | Equipment PHM using non-stationary segmental hidden semi-Markov model | |
| CN108873859B (en) | Bridge type grab ship unloader fault prediction model method based on improved association rule | |
| CN105095990A (en) | Method and device for predicting maintenance | |
| CN114519923A (en) | Intelligent diagnosis and early warning method and system for power plant | |
| Wang et al. | Reliability analysis for automobile engines: Conditional inference trees | |
| Edward et al. | Survival function in the analysis of the factors influencing the reliability of water wells operation | |
| KR102656115B1 (en) | Remaining useful life prediction method of fuel cell system and digital twin device performing the same | |
| KR101242060B1 (en) | The Reliability and Availability Management System of the Urban Transit EMU' Maintenance | |
| CN113094826A (en) | Task reliability-based remaining life prediction method for multi-state manufacturing system | |
| Wagner et al. | An overview of useful data and analyzing techniques for improved multivariate diagnostics and prognostics in condition-based maintenance | |
| Li et al. | Health state prediction and performance evaluation of belt conveyor based on dynamic Bayesian network in underground mining | |
| Zhou et al. | Quantification of value of information associated with optimal observation actions within partially observable Markov decision processes | |
| Chen et al. | BIM-and IoT-Based Data-Driven Decision Support System for Predictive Maintenance of Building Facilities | |
| CN110083933A (en) | A kind of corrosion pipeline Bayes's degradation analysis method considering stochastic effects | |
| Cheng et al. | Optimal opportunistic maintenance model of multi-unit systems | |
| CN115600695A (en) | Fault diagnosis method of metering equipment | |
| CN115496233A (en) | Markov model-based data middlebox operation fault prediction method |
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 |