CN113988331A - Elevator maintenance cycle determination method based on score card model - Google Patents

Elevator maintenance cycle determination method based on score card model Download PDF

Info

Publication number
CN113988331A
CN113988331A CN202111267543.6A CN202111267543A CN113988331A CN 113988331 A CN113988331 A CN 113988331A CN 202111267543 A CN202111267543 A CN 202111267543A CN 113988331 A CN113988331 A CN 113988331A
Authority
CN
China
Prior art keywords
elevator
maintenance
data
model
scoring
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.)
Pending
Application number
CN202111267543.6A
Other languages
Chinese (zh)
Inventor
邱中凯
张雷
万敏
蔡巍伟
靳旭哲
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.)
Zhejiang Xinzailing Technology Co ltd
Original Assignee
Zhejiang Xinzailing Technology Co ltd
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 Zhejiang Xinzailing Technology Co ltd filed Critical Zhejiang Xinzailing Technology Co ltd
Priority to CN202111267543.6A priority Critical patent/CN113988331A/en
Publication of CN113988331A publication Critical patent/CN113988331A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/20—Administration of product repair or maintenance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Engineering & Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Educational Administration (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Development Economics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Game Theory and Decision Science (AREA)
  • Maintenance And Inspection Apparatuses For Elevators (AREA)

Abstract

The invention relates to an elevator maintenance cycle determining method based on a score card model, which comprises the following steps: s1, collecting data needed by establishing a maintenance scoring model; s2, performing target selection learning from the data of the step S1; s3, establishing a maintenance scoring model; and S4, converting the maintenance scoring model into a standard scoring card, and determining the maintenance period of the elevator according to the standard scoring card. The elevator maintenance cycle determining method based on the scoring card model can solve the problems that the maintenance cycle in the prior art is poor in pertinence and cannot meet the requirement for determining the maintenance cycle of an elevator under different use conditions.

Description

Elevator maintenance cycle determination method based on score card model
Technical Field
The invention belongs to the technical field of elevator safety monitoring, and relates to an elevator maintenance cycle determining method based on a score card model.
Background
With the increasing preservation quantity of elevators and the increasing old elevators, the maintenance problem of the elevators is also more and more concerned by people. Normally, an elevator should be cleaned, lubricated, conditioned and monitored at least once every fifteen days, and maintenance is generally performed in four time-fixed manners of half a month, quarter, half a year and year. The maintenance mode plays an important role in timely discovering potential safety hazards of the elevator and guaranteeing safe operation of the elevator. However, the method has certain defects, and can not meet the individual maintenance requirements of elevators in different use environments, different use frequencies and different service lives. For example, in the elevator in the public transport terminal station such as subway, the service of its high load, heavy traffic, long period requires shorter maintenance cycle, and some elevators with low passenger density can suitably prolong the maintenance cycle.
Disclosure of Invention
The invention aims to provide an elevator maintenance cycle determining method based on a score card model, and solves the problems that the maintenance cycle in the prior art is poor in pertinence and cannot meet the requirement for determining the maintenance cycle of an elevator under different use conditions.
In order to achieve the above object, the present invention provides an elevator maintenance cycle determining method based on a score card model, comprising:
s1, collecting data needed by establishing a maintenance scoring model;
s2, performing target selection learning from the data of the step S1;
s3, establishing a maintenance scoring model;
and S4, converting the maintenance scoring model into a standard scoring card, and determining the maintenance period of the elevator according to the standard scoring card.
According to an aspect of the present invention, the data collected in step S1 includes security class data, operation class data, environment class data and attribute class data;
the safety data comprises fault data of people trapping, elevator stopping, emergency stopping, door fault, resetting and the like in the life cycle of the elevator;
the operation type data comprises operation intensity data in the life cycle of the elevator, and the operation intensity data comprises operation times, operation mileage, operation duration, door opening and closing times, number of people carried and bending times;
the attribute class data comprises the service life of the elevator, the brand, the maintenance unit and the property unit.
According to an aspect of the present invention, the step S3 of building a dimension score model includes:
s31, carrying out variable discretization on the data;
s32, performing evidence weight transformation;
s33, screening variables;
and S34, fitting the data by using logistic regression.
According to an aspect of the present invention, in step S31, optimal binning is performed on continuous variables based on the CART algorithm, median values in two adjacent element values are sequentially calculated, a data set is divided into two, a degree of degradation of a kini value from a split value is calculated when the point is used as a cut point, a point with the largest degree of degradation of the kini value is selected as an optimal cut point each time the cut point is performed, and the cut data set is cut according to the same principle until a termination condition.
According to one aspect of the invention, the termination condition is: a. the sample size of each leaf node of the CART algorithm is more than or equal to 2% of the total sample size; b. the minimum number of samples required by the subdivision of the internal nodes is more than or equal to 5% of the total sample size.
According to one aspect of the invention, the formula of the evidence weight is:
Figure BDA0003327318840000021
wherein, BadiBad sample number, Bad for ith binTGood is the total number of bad samplesiGood sample number for ith binTIs the total number of good samples.
According to one aspect of the invention, in step S33, variables are screened using the IV information value, screening variables having IV values greater than 0.02. IV is calculated as follows
Figure BDA0003327318840000022
Figure BDA0003327318840000031
Wherein n is the number of bins.
According to one aspect of the invention, in step S34, the data is fitted using logistic regression whose formula is:
Figure BDA0003327318840000032
wherein wTTo learn the parameters, x is the respective variable (after WOE conversion), and p is the bad sample probability.
According to one aspect of the invention, in step S4, the dimension score model is converted into a standard score card:
the probability of a bad sample is represented as p, then the probability of a good sample is 1-p, which can be obtained by the basic principle of logistic regression:
Figure BDA0003327318840000033
Figure BDA0003327318840000034
the score of the score card is defined as a linear expression of the log of ratios:
Figure BDA0003327318840000035
wherein A and B are constants.
Compared with the traditional expert model, the elevator maintenance cycle determining method based on the scoring card model can learn the most main influence factors of people trapping and faults of the elevator, thereby giving higher weight and having better rationality.
Drawings
Fig. 1 schematically shows a flow chart of an elevator maintenance cycle determination method based on a scorecard model according to the invention.
Detailed Description
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.
The present invention is described in detail below with reference to the drawings and the specific embodiments, which are not repeated herein, but the embodiments of the present invention are not limited to the following embodiments.
As shown in fig. 1, the present invention provides a method for determining an elevator maintenance cycle based on a scorecard model, comprising: s1, collecting data needed by establishing a maintenance scoring model; s2, performing target selection learning from the data of the step S1; s3, establishing a maintenance scoring model; and S4, converting the maintenance scoring model into a standard scoring card, and determining the maintenance period of the elevator according to the standard scoring card.
Wherein the data collected in step S1 includes security class data, operation class data, environment class data and attribute class data. The safety data comprises fault data such as people trapping, elevator stopping, emergency stopping, door fault, resetting and the like in the life cycle of the elevator.
The operation data comprises operation intensity data in the life cycle of the elevator, including the operation frequency, the operation mileage, the operation time length, the door opening and closing frequency, the number of people carried and the bending frequency.
The attribute class data comprises basic attribute data such as the service life of the elevator, the brand, the maintenance unit, the property unit and the like.
Step S2 of the present invention includes two parts, i.e. target selection and feature selection, specifically, assuming that no people trapping or failure will occur in the elevator in the future month, then no maintenance will be needed recently, but if it is determined that people trapping or failure will occur in the future month, then maintenance will be scheduled recently in order to avoid the occurrence of these failures as much as possible, based on this idea, whether maintenance is related to whether people trapping or failure occurs in the future month, so an optional learning goal is whether faults such as people trapping or elevator stopping occur in the future month.
The feature selection is to select a feature more relevant to the target for learning, and the selectable features are as follows:
for the safety data, the number of people trapped in the last 3 months/history, the number of stairs stopped in the last 3 months/history, the number of emergency stops in the last 3 months/history, the number of door faults in the last 3 months/history, the number of reset times in the last 3 months/history, etc. can be selected.
For the operation data, the average operation times of the last 3 months/history, the average operation mileage of the last 3 months/history, the average operation time of the last 3 months/history, the average door opening and closing times of the last 3 months/history, the average number of people carrying the last 3 months/history, the average bending times of the last 3 months/history, and the like can be selected.
For environmental data, the number of battery cars entering stairs in last 3 months/history, the number of doors repeatedly opened and closed in last 3 months/history, the number of shielding doors in last 3 months/history, the number of large luggage entering stairs in last 3 months/history and the like can be selected.
For the attribute data, the service life of the elevator, the brand of the elevator, the maintenance unit of the elevator, the property unit of the elevator and the like can be selected.
In the method of the present invention, the step S3 of establishing the maintenance scoring model includes: s31, carrying out variable discretization on the data; s32, performing evidence weight transformation; s33, screening variables; and S34, fitting the data by using logistic regression.
The variable discretization is variable binning treatment, and common methods include equidistant binning, equal-depth binning and optimal binning; the optional method in the scheme is optimal binning.
Specifically, the continuous variables are optimally binned based on a CART (classification and regression tree) algorithm, the CART is a binary tree, only binary classification is carried out each time, for the continuous variables, the method comprises the steps of sequentially calculating median of two adjacent element values, dividing a data set into two, calculating the descending degree of the kini value when the point is taken as a cutting point compared with the kini value before segmentation, selecting the point with the maximum kini descending degree as the optimal segmentation point each time, and segmenting the segmented data set according to the same principle until a termination condition. The invention relates to the termination conditions of CART binning: a. the sample size of each leaf node is 2% of the total sample size; b. minimum number of samples required for inner node subdivision ═ 5% of total sample size
In step S32, WOE transformation is performed, and WOE (weight of evidence) is called evidence weight. During the modeling analysis of the scoring card, the WOE conversion is needed to be carried out on each original characteristic, and the WOE value of the corresponding grading is calculated. Where the formula of WOE is:
Figure BDA0003327318840000051
wherein BadiBad sample number, Bad for ith binTGood is the total number of bad samplesiGood sample number for ith binTIs the total number of good samples. The bad sample means that there is trouble or failure in the future 1 month, and the good sample means that there is no trouble or failure in the future 1 month
And then variable screening is carried out. The initially defined variables may have low correlation with the target and do not need to be put into a model, so that variable screening is needed to remove the variables with low correlation. IV is calculated as follows
Figure BDA0003327318840000061
Figure BDA0003327318840000062
Wherein n is the number of bins.
And finally, fitting the data by using logistic regression, wherein the formula of the logistic regression is
Figure BDA0003327318840000063
Wherein wTTo learn the parameters, x is the variable (after WOE conversion), and p is the bad sample probability, i.e., there is a predicament or failure in the next 1 month. And then judging the fitting effect of the maintenance scoring model by utilizing ROC and AUC.
After the maintenance scoring model is established, in step S4, the maintenance scoring model is converted into a standard scoring card:
the probability of a bad sample is represented as p, then the probability of a good sample is 1-p, which can be obtained by the basic principle of logistic regression:
Figure BDA0003327318840000064
Figure BDA0003327318840000065
the score of the score card is defined as a linear expression of the log of ratios:
Figure BDA0003327318840000066
wherein A and B are constants. The negative sign of B front can make the lower the risk of stranded people or failure in the future 1 month, the higher the score, the high score of the invention represents low risk, the low score represents high risk
In the above formula, only a and B are unknown, and given a and B, a standard score card can be formulated as shown in the following table:
Figure BDA0003327318840000071
the values of A and B can be obtained by assuming that
(1) The expected score at a certain probability of default, i.e.
Figure BDA0003327318840000072
Is w0Is a score of P0
(2) Scoring of doubling of default Probability (PDO)
Based on the above assumptions, it can be obtained
P0＝A-B×log(w0)
P0-PDO＝A-B×log(2w0)
Solving the equation to obtain
Figure BDA0003327318840000073
A＝P0+B*log(w0)
The parameters of the scoring card selected by the invention are as follows: p0(basal score) 700 and PDO (double rate score) 50 (double rate per 50 higher). The final dimension score is the addition of the base and the score of each part.
After the score card is established, the higher the elevator score is, the lower the risk is, the maintenance cycle can be properly prolonged, and the lower the elevator score is, the higher the risk is, the maintenance cycle needs to be properly shortened; the specific segmentation mode can be comprehensively determined through the relative relation between the maintenance resources and the risk values, and the selectable partitioned interval can be divided into four intervals, namely a key maintenance interval, a high-frequency maintenance interval, a medium-frequency maintenance interval and a low-frequency maintenance interval.
Compared with the traditional expert model, the elevator maintenance cycle determining method based on the scoring card model can learn the most main influence factors of people trapping and faults of the elevator, thereby giving higher weight and having better rationality.
The above description is only one embodiment of the present invention, and is not intended to limit the present invention, and it is apparent to those skilled in the art that various modifications and variations can be made in the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (9)

1. An elevator maintenance cycle determining method based on a score card model comprises the following steps:
s1, collecting data needed by establishing a maintenance scoring model;
s2, performing target selection learning from the data of the step S1;
s3, establishing a maintenance scoring model;
and S4, converting the maintenance scoring model into a standard scoring card, and determining the maintenance period of the elevator according to the standard scoring card.
2. The elevator maintenance cycle determination method based on the score card model according to claim 1, wherein the data collected in the step S1 includes security class data, operation class data, environment class data and attribute class data;
the safety data comprises data of people trapping, elevator stopping, emergency stopping, door fault and reset fault in the life cycle of the elevator;
the operation data comprises operation intensity data in the life cycle of the elevator, including the operation times, operation mileage, operation time length, door opening and closing times, number of people carrying the elevator and bending times;
the attribute class data comprises the service life of the elevator, the brand, the maintenance unit and the property unit.
3. The method for determining the maintenance cycle of an elevator based on the scoring card model of claim 1, wherein the step S3 of establishing the maintenance scoring model comprises:
s31, carrying out variable discretization on the data;
s32, performing evidence weight transformation;
s33, screening variables;
and S34, fitting the data by using logistic regression.
4. The method for determining the maintenance cycle of the elevator based on the scorecard model according to claim 3, characterized in that in the step S31, continuous variables are optimally binned based on the CART algorithm, median values in two adjacent element values are sequentially calculated, the data set is halved, the degree of decline of the kini value when the point is taken as a cut point compared with the kini value before segmentation is calculated, the point with the maximum degree of decline of the kini is selected as the optimal cut point each time, and the data set after segmentation is segmented according to the same principle until the termination condition.
5. The elevator maintenance cycle determination method based on the scorecard model according to claim 4, characterized in that the termination condition is: a. the sample size of each leaf node of the CART algorithm is more than or equal to 2% of the total sample size; b. the minimum number of samples required by the subdivision of the internal nodes is more than or equal to 5% of the total sample size.
6. The elevator maintenance cycle determination method based on the scorecard model according to claim 5, characterized in that the formula of the evidence weight is:
Figure FDA0003327318830000021
wherein, BadiBad sample number, Bad for ith binTGood is the total number of bad samplesiGood sample number for ith binTIs the total number of good samples.
7. The method of claim 6, wherein the variables are screened using the IV information value, and the variables having an IV value greater than 0.02 are screened in step S33. IV is calculated as follows
Figure FDA0003327318830000022
Figure FDA0003327318830000023
Wherein n is the number of bins.
8. The method for determining an elevator maintenance cycle based on a scorecard model according to claim 7, characterized in that in step S34, the data is fitted using a logistic regression whose formula is:
Figure FDA0003327318830000024
wherein wTTo learn the parameters, x is the respective variable (after WOE conversion), and p is the bad sample probability.
9. The method for determining the maintenance cycle of an elevator based on a rating card model according to claim 8, wherein the maintenance rating model is converted into a standard rating card in step S4:
the probability of a bad sample is represented as p, then the probability of a good sample is 1-p, which can be obtained by the basic principle of logistic regression:
Figure FDA0003327318830000025
Figure FDA0003327318830000026
the score of the score card is defined as a linear expression of the log of ratios:
Figure FDA0003327318830000031
wherein A and B are constants.
CN202111267543.6A 2021-10-29 2021-10-29 Elevator maintenance cycle determination method based on score card model Pending CN113988331A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111267543.6A CN113988331A (en) 2021-10-29 2021-10-29 Elevator maintenance cycle determination method based on score card model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111267543.6A CN113988331A (en) 2021-10-29 2021-10-29 Elevator maintenance cycle determination method based on score card model

Publications (1)

Publication Number Publication Date
CN113988331A true CN113988331A (en) 2022-01-28

Family

ID=79743952

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111267543.6A Pending CN113988331A (en) 2021-10-29 2021-10-29 Elevator maintenance cycle determination method based on score card model

Country Status (1)

Country Link
CN (1) CN113988331A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115564069A (en) * 2022-09-28 2023-01-03 北京百度网讯科技有限公司 Method for determining server maintenance strategy, method for generating model and device thereof

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115564069A (en) * 2022-09-28 2023-01-03 北京百度网讯科技有限公司 Method for determining server maintenance strategy, method for generating model and device thereof

Similar Documents

Publication Publication Date Title
CN111047082B (en) Early warning method and device of equipment, storage medium and electronic device
CN110441629B (en) Fault diagnosis method and device based on turnout switch machine operating current curve
CN106909933B (en) A kind of stealing classification Forecasting Methodology of three stages various visual angles Fusion Features
Insua et al. A framework for risk management decisions in aviation safety at state level
CN112365066A (en) Elevator fault prediction method, system, device, computer equipment and storage medium
WO2019196869A1 (en) Method for determining list of patrolling base stations, and patrolling apparatus
CN111178598A (en) Passenger flow prediction method and system for railway passenger station, electronic device and storage medium
CN112884325A (en) Method and system for application analysis and health condition evaluation of customer station equipment
CN111931978A (en) Urban rail transit passenger flow state prediction method based on space-time characteristics
CN110647446A (en) Log fault association and prediction method, device, equipment and storage medium
CN109472075B (en) Base station performance analysis method and system
CN114267178B (en) Intelligent operation maintenance method and device for station
CN110690701A (en) Analysis method for influence factors of abnormal line loss
CN113988331A (en) Elevator maintenance cycle determination method based on score card model
Lin et al. Short-term metro passenger flow prediction based on random forest and LSTM
CN110751338A (en) Construction and early warning method for heavy overload characteristic model of distribution transformer area
CN111310275A (en) Bridge disease big data analysis method
US20170236071A1 (en) Alarm management system
Wen et al. Cause-specific investigation of primary delays of Wuhan–Guangzhou HSR
CN113837473A (en) Charging equipment fault rate analysis system and method based on BP neural network
Xu et al. Comparisons of logistic regression and artificial neural network on power distribution systems fault cause identification
CN117034001A (en) Wind turbine generator system fault prediction method and device and electronic equipment
Attanayake et al. On the necessity of using supervised machine learning for risk-based screening of distribution transformers: An industrial case study
CN112434955B (en) Distribution network line operation risk sensing method based on multi-data fusion
Bautista et al. A condition-based maintenance for complex systems consisting of two different types of components

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