CN115959549A - Escalator fault diagnosis method based on digital twinning - Google Patents
Escalator fault diagnosis method based on digital twinning Download PDFInfo
- Publication number
- CN115959549A CN115959549A CN202211739600.0A CN202211739600A CN115959549A CN 115959549 A CN115959549 A CN 115959549A CN 202211739600 A CN202211739600 A CN 202211739600A CN 115959549 A CN115959549 A CN 115959549A
- Authority
- CN
- China
- Prior art keywords
- escalator
- fault
- data
- characteristic value
- value
- 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
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02B—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO BUILDINGS, e.g. HOUSING, HOUSE APPLIANCES OR RELATED END-USER APPLICATIONS
- Y02B50/00—Energy efficient technologies in elevators, escalators and moving walkways, e.g. energy saving or recuperation technologies
Landscapes
- Escalators And Moving Walkways (AREA)
Abstract
The invention discloses a digital twin-based escalator fault diagnosis method, which comprises the following steps of: acquiring and loading an escalator online fault monitoring model; acquiring escalator measuring point data, wherein the measuring point data comprises vibration waveform data, process quantity data and electromechanical system information; calculating a time domain characteristic value and a frequency domain characteristic value of a vibration signal waveform according to the vibration waveform data, and normalizing the time domain characteristic value and the frequency domain characteristic value; inputting the normalized time domain characteristic value and frequency domain characteristic value into an escalator fault monitoring model to obtain an independent vibration monitoring result of an escalator measuring point; and inputting the independent vibration monitoring result, the process quantity data and the electromechanical system information of the escalator measuring point into the escalator fault classification tree model, carrying out fault positioning, and obtaining an analyzed escalator fault comprehensive diagnosis result through linkage fault analysis. The diagnosis system based on the digital prototype has the advantages of high automation degree, advanced pre-judgment, accurate fault positioning, low false alarm rate, zero missing report rate and the like.
Description
Technical Field
The invention relates to the technical field of mechanical fault diagnosis, in particular to a digital twin-based escalator fault diagnosis method.
Background
An escalator, also known as an escalator for moving pedestrians, is a kind of escalator with circularly running steps, is a fixed electric driving device used for transporting passengers upwards or downwards, and is commonly used in public facilities such as shopping malls, overpasses, subway stations and the like. Due to the fact that the number of users is large, safety and maintenance problems of the escalator are very important, and an electromechanical system or a sensor is used for recording escalator operation data in the operation process and detecting according to a diagnosis model, so that the escalator operation data is an important means for detecting faults. The escalator operation data comprises process quantity signals representing operation states, such as motor rotating speed, key phase values, motor power, motor current and the like, vibration signals of measuring points, such as the motor, the reduction gearbox, the main driving wheel, the step chain tension wheel and the like, and electromechanical system information, such as escalator operation states, motor operation states, escalator uplink and downlink parameters, fault codes and the like.
The major failure of escalator installations is mostly a developmental, long-term and early-characterized failure. Early failure or failure precursors of escalator equipment can be often detected through modern detection means based on data drive and mechanism knowledge, and the escalator equipment early failure or failure precursors are favorable for timely reminding relevant personnel to carry out night shutdown detection and maintenance or temporary failure maintenance, so that the occurrence of serious accidents is avoided. The escalator digital prototype built by the digital twin technology is a mathematical model which has high value and high potential, can reflect the authenticity of a physical real machine, is worthy of data mining and deep analysis, and can play important roles in monitoring the normal operation of the escalator, providing escalator maintenance suggestions and diagnosing escalator fault types by forming a superior fault diagnosis model through secondary development.
The existing physical real machine alarm system mainly judges faults according to fault codes of an electromechanical system or evaluates and maintains the escalator condition by depending on inspection personnel according to self experiences, and has the defects of high manual dependence, high hysteresis, unclear afterward maintenance target due to unclear fault identification and the like.
Disclosure of Invention
In order to solve the problems in the prior art, the embodiment of the invention provides a digital twin-based escalator fault diagnosis method. The technical scheme is as follows:
in a first aspect, a digital twin-based escalator fault diagnosis method is provided, and comprises the following steps:
acquiring and loading an escalator online fault monitoring model;
acquiring escalator measuring point data, wherein the measuring point data comprises vibration waveform data, process quantity data and electromechanical system information;
calculating a time domain characteristic value and a frequency domain characteristic value of a vibration signal waveform according to the vibration waveform data, and normalizing the time domain characteristic value and the frequency domain characteristic value;
inputting the normalized time domain characteristic value and frequency domain characteristic value into an online escalator fault monitoring model to obtain an independent vibration monitoring result of an escalator measuring point;
and inputting the independent vibration monitoring result, the process quantity data and the electromechanical system information of the escalator measuring point into the escalator fault classification tree model, carrying out fault positioning, and obtaining an analyzed escalator fault comprehensive diagnosis result through linkage fault analysis.
Further, before calculating a time domain characteristic value and a frequency domain characteristic value of a vibration signal waveform according to the vibration waveform data, the method further comprises preprocessing the measuring point data, wherein the preprocessing comprises denoising, filtering, screening invalid points, screening non-key phase points and screening signal abnormal points.
Further, the time domain feature values include: mean, standard deviation, effective value, peak-to-peak value, kurtosis, skewness, crest factor and kurtosis coefficient; the frequency domain feature values include: gIE eigenvalue, pass frequency characteristic, double frequency characteristic, teager energy factor.
Further, the escalator online fault monitoring model is established in an offline manner by analyzing a local digital prototype integrated data packet, and the establishing method comprises the following steps:
acquiring and analyzing an integrated data packet of an escalator data prototype model, wherein the integrated data packet comprises a measured point time domain characteristic value and a measured point frequency domain characteristic value which are calculated;
normalizing the measuring point time domain characteristic value and the frequency domain characteristic value;
forming a feature vector matrix by the normalized measuring point time domain feature value and the normalized measuring point frequency domain feature value, and removing the extreme points;
inputting the characteristic vector matrix into an LOF algorithm to obtain an outlier detection model;
and sequencing according to the size of the LOF dimensionless value to obtain an LOF sequence of a feature vector matrix from small to large, and dividing the data set into a safety set, a middle set and an edge set according to a preset proportion.
Further, the escalator online fault monitoring model is established online by collecting real-time data packets of an escalator digital prototype, and the establishing method comprises the following steps:
acquiring real-time operation data of a digital prototype of the escalator, and preprocessing the real-time operation data, wherein the preprocessing comprises denoising, filtering, screening invalid points, screening keyless phase points and screening signal abnormal points;
calculating a time domain characteristic value and a frequency domain characteristic value according to the preprocessed real-time operation data, and accumulating an escalator operation vibration data set;
after the vibration data set reaches a preset data volume, normalizing the time domain characteristic value and the frequency domain characteristic value;
forming a characteristic vector matrix by the vibration data set after the normalization processing, and removing extreme points in the characteristic vector matrix;
inputting the feature vector matrix into an LOF algorithm to obtain an outlier detection model;
and sequencing according to the size of the LOF dimensionless value to obtain an LOF sequence with a characteristic vector matrix from small to large, and dividing the data set into a safety set, a middle set and an edge set according to a preset proportion.
Further, the step of inputting the independent vibration monitoring result, the process quantity data and the electromechanical system information of the escalator measuring point into the escalator fault classification tree model for fault positioning, and obtaining the analyzed escalator fault comprehensive diagnosis result through linkage fault analysis comprises the following steps:
normalizing the vibration state monitoring result and the process quantity data of each part of the escalator and the information structure of an electromechanical system, and dividing the result into a discrete characteristic and a continuous characteristic;
establishing an escalator fault classification tree model by using a CART algorithm, and judging the overall state of a part to which a measuring point belongs through an escalator measuring point vibration signal, a process quantity signal and electromechanical information to perform fault positioning;
after the CART tree obtains the component operation state classification, the escalator fault source judgment is carried out based on the linkage fault analysis and induction relation table and the escalator multi-component fault diagnosis result, and various escalator faults are judged to be self faults or linkage faults;
for self faults, directly outputting fault diagnosis results;
for linkage faults, single state feedback information of the components is input, upper and lower level escalator components corresponding to the faults are found through linkage fault analysis and induction relation table fault tracing retrieval, and comprehensive inference results of fault diagnosis are output.
Further, the building of the escalator fault classification tree model by using the CART algorithm comprises the following steps:
(1) For the data set of the current node, if the number of samples is less than a threshold value or no characteristic exists, returning to a decision sub-tree, and stopping recursion of the current node;
(2) Calculating the kini coefficient of the data set, if the kini coefficient is smaller than a threshold value, returning to a decision tree subtree, and stopping recursion of the current node;
(3) Calculating the kini indexes of all values of all existing characteristics of the current node;
(4) Selecting the characteristic with the minimum damping coefficient and the corresponding value thereof as the optimal characteristic and the optimal dividing point from the damping coefficients of all the values of all the characteristics, and dividing the data set of the node into two parts D according to the optimal characteristic and the optimal dividing point 1 And D 2 Simultaneously generating two child nodes of the current node, wherein the data set of the left node is D 1 The data set of the right node is D 2 。
(5) And (4) recursively repeating the steps (1) to (4) for the left and right child nodes to generate an escalator fault classification tree model.
Further, after the escalator fault classification tree model is generated, the method further comprises pruning by using a cost complexity method:
defining each non-leaf node in the classification tree to calculate its surface error rate gain value
Wherein T is t Is a sub-tree with T as the root node, | T t L is the number of leaf nodes contained in the subtree, C (t) is the error cost of the single-node tree by t, the node is pruned, C (t) = r (t) × p (t), r (t) is the error rate of the node t, and p (t) is the proportion of data on the node t to all the users; c (T) t ) Is T t If the node is not pruned, it equals the subtree T t The sum of the error costs of all the leaf nodes is obtained, and when the alpha values of a plurality of non-leaf nodes reach the minimum simultaneously, the absolute value T is taken t Pruning is performed with the maximum.
The technical scheme provided by the embodiment of the invention has the following beneficial effects: in the embodiment of the invention, an escalator online fault monitoring model is obtained and loaded; acquiring escalator measuring point data, wherein the measuring point data comprises vibration waveform data, process quantity data and electromechanical system information; calculating a time domain characteristic value and a frequency domain characteristic value of a vibration signal waveform according to the vibration waveform data, and normalizing the time domain characteristic value and the frequency domain characteristic value; inputting the normalized time domain characteristic value and frequency domain characteristic value into an online escalator fault monitoring model to obtain an independent vibration monitoring result of an escalator measuring point; and inputting the independent vibration monitoring result, the process quantity data and the electromechanical system information of the escalator measuring point into the escalator fault classification tree model, carrying out fault positioning, and obtaining an analyzed escalator fault comprehensive diagnosis result through linkage fault analysis. The diagnosis system based on the digital prototype has the advantages of high automation degree, advanced pre-judgment, accurate fault positioning, low false alarm rate, zero missing report rate and the like, solves the core pain point of the existing system, and provides safe and reliable simulation data support on equipment alarm and predictive maintenance analysis. The escalator operation state is detected and diagnosed in a digital prototype modeling mode, the hidden danger of an electromechanical system is eliminated in time, the safety of the escalator serving as a key node in public facilities is ensured, and the digital prototype detection and diagnosis method has important significance for landing application of a digital twin system.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a flow chart of a digital twin-based escalator fault diagnosis method provided by an embodiment of the invention;
fig. 2 is a flowchart of an escalator fault diagnosis method based on a classification tree according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
The digital twin-based escalator fault diagnosis process shown in fig. 1 will be described in detail below with reference to specific embodiments, and the contents may be as follows:
and S101, acquiring and loading an escalator online fault monitoring model.
The escalator operation side edge computing node can acquire the established escalator online fault monitoring model from the cloud server, can also perform offline establishment of an escalator online fault monitoring model by analyzing the local digital prototype integrated data packet, or can acquire the escalator digital prototype data packet in real time through the digital prototype data acquisition card, and further establish the escalator online fault monitoring model online. The escalator online fault monitoring model comprises normalized parameters, namely zscore parameters (mean value and standard deviation) of characteristic vectors.
Optionally, the method for establishing the escalator online fault monitoring model in an offline manner is as follows:
(1) And analyzing an integrated data packet (Json format) of the escalator data prototype model to obtain the calculated measuring point time domain characteristic value and the calculated measuring point frequency domain characteristic value.
The time domain characteristic values comprise a mean value, a standard deviation, an effective value, a peak-to-peak value, kurtosis, skewness, a crest factor and a kurtosis coefficient; the frequency domain eigenvalues include gIE eigenvalues, pass frequency eigenvalues (center of gravity frequency, variance frequency, mean square frequency), double frequency eigenvalues (center of gravity frequency, variance frequency, mean square frequency), teager energy factor.
(2) And performing zscore normalization on the time domain characteristic value and the frequency domain characteristic value obtained by analysis.
zscore is normalized in a manner of
Wherein
σ x Mean and standard deviation of x, respectively.
(3) And forming a vector matrix by the characteristic parameters after the normalization processing according to rules, removing points with obvious outliers according to a 3 sigma rule, and primarily removing extreme points in the points.
(4) And inputting the characteristic vector matrix into an LOF algorithm to obtain an outlier detection model (an escalator online fault monitoring model).
The LOF algorithm comprises the following specific processes:
(1) defining a K distance
The k-distance of the data object q is defined as the distance from the nearest k-th point in the data set to the data object q, denoted as k-distance (q), and is referred to herein as the Euclidean distance.
(2) K distance neighborhood
The set of data points in the data set having a distance to the data object q of no more than k, i.e. N k -distance (q) (p) = { p ∈ D { q } | D (p, q) ≦ k-distance (q) }. d (p, q) denotes the European expression between p and qDistance.
(3) Reachable distance
p and q are any two points in the data set, and the reachable distance from p to q is defined as: reach-distk (p, q) = max { d (p, q), k-distance (q) }.
(4) Local achievable density
The local reachable density of q refers to the inverse of the average reachable distance of q to all points in the neighborhood, and is calculated as follows:
wherein, | N k (q) | is the number of points in the k neighborhood of q. If lrd k The larger (q) indicates the higher density of q and the more normal the q point.
(5) Local outlier factor
If LOF >1, it indicates that the q-point density is far from the overall data density, i.e., the outliers. LOF is close to 1, then point q is more normal.
(5) And sequencing the size of the LOF dimensionless values to obtain a LOF sequence of the feature matrix from small to large. The data sets are divided into a security set, a middle set and an edge set in a proportion of 75%, 85% and 95%.
Optionally, the method for online establishing the escalator online fault monitoring model is as follows:
(1) The method comprises the steps of collecting operation data of a digital prototype of the escalator in real time, and preprocessing the operation data, wherein the preprocessing comprises denoising, filtering, and screening out invalid points, keyless phase points and signal abnormal points.
(2) And calculating the time domain characteristic value and the frequency domain characteristic value of the escalator, and accumulating the escalator operation vibration data set.
The time domain characteristic values comprise a mean value, a standard deviation, an effective value, a peak-to-peak value, kurtosis, skewness, a crest factor and a kurtosis coefficient; the frequency domain eigenvalues include gIE eigenvalues, pass frequency characteristics (center of gravity frequency, variance frequency, mean square frequency), octager frequency characteristics (center of gravity frequency, variance frequency, mean square frequency), teager energy factor. The escalator operation vibration data set can manually or automatically set the data volume according to the computer resource condition, and the data volume can be accumulated when reaching the preset data volume, so that the subsequent processing is further continued.
(3) After the vibration data set meets the quantity condition, zscore normalization is carried out on the vibration data set.
Wherein
σ x Mean and standard deviation of x, respectively.
(4) And filtering extreme values of the normalized eigenvectors, and forming a vector matrix according to rules.
(5) And inputting the characteristic vector matrix into an LOF algorithm to obtain an outlier detection model.
The LOF algorithm comprises the following specific processes:
(1) defining a K distance
The k-distance of the data object q is defined as the distance from the nearest kth point to q in the data set to the data object q, and is denoted as k-distance (q), and is referred to as Euclidean distance.
(2) K distance neighborhood
The set of data points in the data set having a distance to the data object q of no more than k, i.e. N k -distance (q) (p) = { p ∈ D { q } | D (p, q) ≦ k-distance (q) }. d (p, q) refers to the Euclidean distance between p and q.
(3) Reachable distance
p and q are any two points in the data set, and the reachable distance from p to q is defined as: reach-distk (p, q) = max { d (p, q), k-distance (q) }.
(4) Local achievable density
The local reachable density of q refers to the inverse of the average reachable distance of q to all points in the neighborhood, and is calculated as follows:
wherein, | N k (q) | is the number of points in k neighborhood of q. If lrd k The larger (q) indicates the higher density of q and the more normal the q point.
(5) Local outlier factor
If LOF >1, it indicates that the q-point density is far from the overall data density, i.e., the outliers. LOF is close to 1, then point q is more normal.
(6) And sequencing the size of the dimensionless LOF values to obtain a LOF sequence of the feature matrix from small to large. The data sets are divided into a security set, a middle set and an edge set in a proportion of 75%, 85% and 95%.
Step S102, collecting escalator measuring point data, wherein the measuring point data comprises vibration waveform data, process quantity data and electromechanical system information.
And (3) acquiring measuring point data of each measuring point of the escalator on line in real time by the escalator operation side edge computing node, wherein the measuring point data comprises vibration waveform data, process quantity data and electromechanical system information of the measuring point. And then, carrying out preprocessing such as measuring point data denoising, filtering, invalid point screening, keyless phase point screening, signal abnormal point screening and the like by the escalator operation side edge computing node.
Step S103, calculating a time domain characteristic value and a frequency domain characteristic value of the vibration signal waveform according to the vibration waveform data, and normalizing the time domain characteristic value and the frequency domain characteristic value.
The escalator operation side edge calculation node calculates a time-frequency domain characteristic value of the waveform of the vibration signal according to the collected vibration waveform data, wherein the time-frequency domain characteristic value comprises a mean value, a standard deviation, an effective value, a peak-peak value, kurtosis, skewness, a crest factor and a kurtosis coefficient; the frequency domain eigenvalues include gIE eigenvalues, pass frequency characteristics (center of gravity frequency, variance frequency, mean square frequency), octager frequency characteristics (center of gravity frequency, variance frequency, mean square frequency), teager energy factor. Then, the time domain feature value and the frequency domain feature value obtained by the calculation are normalized by using the normalization parameter pair obtained in step 101.
And step S104, inputting the normalized time domain characteristic value and frequency domain characteristic value into an escalator online fault monitoring model to obtain an independent vibration monitoring result of an escalator measuring point.
And taking the characteristic value of a frame of vibration wave mode signal as a characteristic vector, searching K nearest neighbor points in the LOF model and solving the LOF value, wherein the operation process adopts a pruning mode to reduce the number of global calculation points.
And judging the high-dimensional space position of the Hibert to which the current point belongs according to an outlier identification strategy, namely which set (a security set, a middle set and an edge set) the K nearest neighbors of the current point belong to and the LOF value.
The specific details of the outlier identification strategy are as follows:
(1) If LOF >10, determining the current point as a height outlier, and defining an outlier index P = LOF multiplied by 1;
(2) If 10>
(3) If LOF is less than 1, the current point is determined as a safety point, and the indicator is outlier
Wherein, | N-saf k (q) | is the number of k neighborhood interior points of q belonging to the security set, | N-Mid k (q) | is the number of the k neighborhood inner points of q belonging to the middle set, and | N-Mark (q) | is the number of the k neighborhood inner points of q belonging to the edge set.
When height outliers or moderate outliers (considered as height outliers) with an outlier index P >5 occur continuously (generally considered as more than 5 continuous), or the mean value of the outlier index P in a unit time window exceeds a fixed limit, the escalator is considered to be in an abnormal condition currently. The fixed limit for the outlier P may be determined from historical statistics or from administrator experience.
And according to the actual application requirement, the LOF model is intermittently and normally updated.
The steps for updating the LOF model are as follows:
(1) And collecting new feature vectors with normal judgment results.
(2) The LOF model is inverse normalized using the normalization parameters.
(3) The LOF model eliminates the point with the earliest time, and if the earliest time point is multiple, the point with the largest LOF is eliminated.
(4) The LOF model adds the newly listed feature vectors and calculates the model LOF value.
(5) The edge lines of the security set, the middle set and the edge set are updated.
And S105, inputting the independent vibration monitoring result, the process quantity data and the electromechanical system information of the escalator measuring point into an escalator fault classification tree model, carrying out fault positioning, and analyzing the escalator fault comprehensive diagnosis result through linkage fault analysis.
The method comprises the steps of firstly discretizing vibration state monitoring results and process quantity information of all parts of the escalator, and establishing an escalator fault classification tree model by adopting a CART classification tree.
The escalator fault classification comprises main driving wheel abnormity, reduction box fault, reduction box foot fixing abnormity, handrail belt mounting abnormity, handrail belt fault, step chain tension wheel abnormity and motor abnormity, and is accompanied with a linkage fault analysis and summary relation table.
The escalator fault diagnosis method comprises the following detailed steps:
step S201, normalizing the vibration state monitoring result and the process quantity data of each part of the escalator and the information structure of the electromechanical system, and dividing the vibration state monitoring result and the process quantity data into a discrete characteristic part and a continuous characteristic part.
Step S202, a CART algorithm is used for building an escalator fault classification tree model, and the overall state of a part of a measuring point is judged through an escalator measuring point vibration signal, a process quantity signal and electromechanical information, so that fault positioning is carried out.
And S203, after the operation states of the parts are classified by the CART tree, carrying out escalator fault source judgment based on a linkage fault analysis and induction relation table and escalator multi-part (motor, step chain tension wheel, handrail belt, main driving wheel and reduction gearbox) fault diagnosis results, and judging various escalator faults as self faults and linkage faults.
And judging that the component has a fault for the fault source, and directly outputting a fault diagnosis result.
And if the fault source is determined to be a component linkage fault, according to the component vibration state monitoring result, the process quantity data and the electromechanical system information which are processed and normalized in the step S201, searching the fault tracing source through a linkage fault analysis and induction relation table preset by a worker to find the upper and lower escalator components corresponding to the fault, and outputting a comprehensive inference result of fault diagnosis.
Optionally, the CART classification tree algorithm comprises the following steps:
(1) Discretizing the continuous type features.
For the samples M, the number is | M |, the continuous characteristics A have | M | values, the samples are arranged from small to large, the CART takes the average of two adjacent sample values as division points, the total number is M-1, wherein the ith division point T is i Expressed as:
T i ＝(α i +α i+1 )/2
the kini coefficients when the m-1 points are regarded as binary classification points are calculated, respectively. And selecting the point with the minimum Keyney coefficient as the binary discrete classification point of the continuous characteristic. For example, the point where the minimum Keyney coefficient is taken is α t Is smaller than α t Is of class 1, greater than α t Is category 2, thus achieving discretization of the continuous features.
(2) The kini coefficient of each feature is calculated. The CART classification tree algorithm establishes a classification tree in a binary tree form, the characteristics are selected by using the kini coefficient, the kini coefficient represents the impure degree of the model, and the lower the kini coefficient is, the lower the impure degree is, and the better the characteristics are. In the continuous case, the data set D,
under the discrete condition, for the sample D, the number is | D |, and according to the possible value a of the characteristic A, the D is divided into two parts D1 and D2.
D 1 ＝(x，y)∈D|A(x)＝a，D 2 ＝D-D 1
For feature a, the kini coefficient for sample D is defined as:
(3) Establishing CART classification tree (namely staircase fault classification tree model)
The CART classification tree establishment steps are as follows:
inputting: training set D, threshold value of the kini coefficient and threshold value of the minimum number of samples to be segmented.
And (3) outputting: a CART tree.
Starting from the root node, the CART classification tree is recursively built using a training set.
(1) If the data set of the current node is D, if the number of samples is less than the threshold value or no characteristic exists, a decision sub-tree is returned, and the current node stops recursion.
(2) And calculating the kini coefficient of the data set D, if the kini coefficient is smaller than a threshold value, returning to the subtree of the decision tree, and stopping recursion of the current node.
(3) And calculating the Gini indexes of all values of all existing characteristics of the current node.
(4) And selecting the feature A with the minimum Keyny coefficient and the corresponding value a thereof as the optimal feature and the optimal segmentation point from the calculated Keyny coefficients of the values of the features. Then, according to the optimal characteristics and the optimal segmentation points, the data set of the node is divided into two parts D 1 And D 2 Simultaneously generating two child nodes of the current node, wherein the data set of the left node is D 1 Data set of right node is D 2 。
(5) And recursively calling the left and right child nodes for 1-4 steps to generate the CART classification tree.
(4) Pruning using cost complexity method
The cost complexity pruning method is as follows:
defining each non-leaf node in a classification tree to calculate its surface error rate gain value
Wherein T is t Is a sub-tree with T as the root node, | T t L is the number of leaf nodes contained in the subtree, C (t) is the error cost of a single-node tree by taking t as the error cost, the node is pruned, C (t) = r (t) × p (t), r (t) is the error rate of the node t, and p (t) is the proportion of data on the node t to all users; c (T) t ) Is T t If the node is not pruned, it equals the subtree T t Sum of error costs of all leaf nodes.
The pruning method aims to find the non-leaf node with the minimum alpha value, make the left and right nodes NULL, and take the absolute value of T when the alpha values of a plurality of non-leaf nodes reach the minimum simultaneously t Pruning is performed with the maximum.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment may be implemented by software plus a necessary general hardware platform, and may also be implemented by hardware. With this in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute the methods of the various embodiments or some parts of the embodiments.
The present invention is not limited to the above preferred embodiments, and any modifications, equivalent replacements, improvements, etc. within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (8)
1. The escalator fault diagnosis method based on the digital twin is characterized by comprising the following steps of:
acquiring and loading an escalator online fault monitoring model;
acquiring escalator measuring point data, wherein the measuring point data comprises vibration waveform data, process quantity data and electromechanical system information;
calculating a time domain characteristic value and a frequency domain characteristic value of a vibration signal waveform according to the vibration waveform data, and normalizing the time domain characteristic value and the frequency domain characteristic value;
inputting the normalized time domain characteristic value and frequency domain characteristic value into an online escalator fault monitoring model to obtain an independent vibration monitoring result of an escalator measuring point;
and inputting the independent vibration monitoring result, the process quantity data and the electromechanical system information of the escalator measuring point into the escalator fault classification tree model, carrying out fault positioning, and obtaining an analyzed escalator fault comprehensive diagnosis result through linkage fault analysis.
2. The method of claim 1, wherein prior to calculating the time domain feature value and the frequency domain feature value of the vibration signal waveform from the vibration waveform data, the method further comprises preprocessing the survey point data, wherein the preprocessing comprises denoising, filtering, screening invalid points, screening non-key phase points, and screening signal outliers.
3. The method of claim 1, wherein the time-domain eigenvalues comprise: mean, standard deviation, effective value, peak-to-peak, kurtosis, skewness, crest factor, kurtosis coefficient; the frequency domain feature values include: gIE eigenvalue, pass frequency characteristic, double frequency characteristic, teager energy factor.
4. The method of claim 1, wherein the escalator online fault monitoring model is established offline by parsing a local digital prototype integrated data packet, the establishing method comprising:
acquiring and analyzing an integrated data packet of an escalator data prototype model, wherein the integrated data packet comprises a measured point time domain characteristic value and a frequency domain characteristic value which are calculated;
normalizing the measuring point time domain characteristic value and the frequency domain characteristic value;
forming a feature vector matrix by the normalized measuring point time domain feature value and the normalized measuring point frequency domain feature value, and removing the extreme points;
inputting the characteristic vector matrix into an LOF algorithm to obtain an outlier detection model;
and sequencing according to the size of the LOF dimensionless value to obtain an LOF sequence of a feature vector matrix from small to large, and dividing the data set into a safety set, a middle set and an edge set according to a preset proportion.
5. The method of claim 1, wherein the escalator online fault monitoring model is established online by collecting escalator digital prototype real-time data packets, and the establishing method comprises the following steps:
acquiring real-time operation data of a digital prototype of the escalator, and preprocessing the real-time operation data, wherein the preprocessing comprises denoising, filtering, screening invalid points, screening keyless phase points and screening signal abnormal points;
calculating a time domain characteristic value and a frequency domain characteristic value according to the preprocessed real-time operation data, and accumulating an escalator operation vibration data set;
after the vibration data set reaches a preset data volume, normalizing the time domain characteristic value and the frequency domain characteristic value;
forming a characteristic vector matrix by the vibration data set after the normalization processing, and removing extreme points in the characteristic vector matrix;
inputting the feature vector matrix into an LOF algorithm to obtain an outlier detection model;
and sequencing according to the size of the LOF dimensionless value to obtain an LOF sequence of a feature vector matrix from small to large, and dividing the data set into a safety set, a middle set and an edge set according to a preset proportion.
6. The method as claimed in claim 1, wherein the step of inputting the independent vibration monitoring result, the process quantity data and the electromechanical system information of the escalator measuring point into the escalator fault classification tree model for fault location, and obtaining the analyzed escalator fault comprehensive diagnosis result through linkage fault analysis comprises the steps of:
normalizing the vibration state monitoring result and the process quantity data of each part of the escalator and the information structure of the electromechanical system into a discrete characteristic part and a continuous characteristic part;
establishing a staircase fault classification tree model by using a CART algorithm, judging the overall state of a part to which a measuring point belongs through a staircase measuring point vibration signal, a process quantity signal and electromechanical information, and carrying out fault positioning;
after the CART tree obtains the component operation state classification, the escalator fault source judgment is carried out based on the linkage fault analysis and induction relation table and the escalator multi-component fault diagnosis result, and various escalator faults are judged to be self faults or linkage faults;
for self faults, directly outputting fault diagnosis results;
and for linkage faults, inputting single state feedback information of the component, searching the source of the faults through a linkage fault analysis and induction relation table to find the upper and lower escalator components corresponding to the faults, and outputting a comprehensive inference result of fault diagnosis.
7. The method of claim 6, wherein the building an escalator fault classification tree model using the CART algorithm comprises:
(1) For the data set of the current node, if the number of samples is less than a threshold value or no characteristic exists, returning to a decision sub-tree, and stopping recursion of the current node;
(2) Calculating the kini coefficient of the data set, if the kini coefficient is smaller than a threshold value, returning to a decision tree subtree, and stopping recursion of the current node;
(3) Calculating the kini indexes of all values of all existing characteristics of the current node;
(4) Selecting the characteristic with the minimum damping coefficient and the corresponding value thereof as the optimal characteristic and the optimal dividing point from the damping coefficients of all the values of all the characteristics, and dividing the data set of the node into two parts D according to the optimal characteristic and the optimal dividing point 1 And D 2 Simultaneously generating two child nodes of the current node, wherein the data set of the left node is D 1 Data set of right node is D 2 。
(5) And (4) recursively repeating the steps (1) to (4) for the left and right child nodes to generate an escalator fault classification tree model.
8. The method of claim 7, wherein after generating the staircase fault classification tree model, the method further comprises pruning using a cost complexity method:
defining each non-leaf node in a classification tree to calculate its surface error rate gain value
Wherein T is t Is a sub-tree with T as the root node, | T t L is the number of leaf nodes contained in the subtree, C (t) is the error cost of the single-node tree by t, the node is pruned, C (t) = r (t) × p (t), r (t) is the error rate of the node t, and p (t) is the proportion of data on the node t to all the users; c (T) t ) Is T t If the node is not pruned, it equals the subtree T t The sum of the error costs of all the leaf nodes is obtained, and when the alpha values of a plurality of non-leaf nodes reach the minimum simultaneously, the absolute value T is taken t Pruning is performed with the maximum.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211739600.0A CN115959549A (en) | 2022-12-30 | 2022-12-30 | Escalator fault diagnosis method based on digital twinning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211739600.0A CN115959549A (en) | 2022-12-30 | 2022-12-30 | Escalator fault diagnosis method based on digital twinning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115959549A true CN115959549A (en) | 2023-04-14 |
Family
ID=87357675
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211739600.0A Pending CN115959549A (en) | 2022-12-30 | 2022-12-30 | Escalator fault diagnosis method based on digital twinning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115959549A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117436024A (en) * | 2023-12-19 | 2024-01-23 | 湖南翰文云机电设备有限公司 | Fault diagnosis method and system based on drilling machine operation data analysis |
| CN118130070A (en) * | 2024-04-03 | 2024-06-04 | 上海辉度智能系统有限公司 | Escalator fault prediction diagnosis method, device and system |
-
2022
- 2022-12-30 CN CN202211739600.0A patent/CN115959549A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117436024A (en) * | 2023-12-19 | 2024-01-23 | 湖南翰文云机电设备有限公司 | Fault diagnosis method and system based on drilling machine operation data analysis |
| CN117436024B (en) * | 2023-12-19 | 2024-03-08 | 湖南翰文云机电设备有限公司 | Fault diagnosis method and system based on drilling machine operation data analysis |
| CN118130070A (en) * | 2024-04-03 | 2024-06-04 | 上海辉度智能系统有限公司 | Escalator fault prediction diagnosis method, device and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107941537B (en) | A kind of mechanical equipment health state evaluation method | |
| CN115959549A (en) | Escalator fault diagnosis method based on digital twinning | |
| CN113344134B (en) | Low-voltage distribution monitoring terminal data acquisition abnormality detection method and system | |
| CN108069308A (en) | A kind of electric staircase failure diagnosis method based on sequential probability | |
| CN115760484B (en) | Method, device and system for improving hidden danger identification capability of power distribution area and storage medium | |
| CN113562562A (en) | Elevator safety early warning monitoring system and judgment method thereof | |
| CN113610188B (en) | Bow net contact force non-section abnormality identification method and device | |
| CN116380445A (en) | Equipment state diagnosis method and related device based on vibration waveform | |
| CN114723285A (en) | Power grid equipment safety evaluation prediction method | |
| CN109034076A (en) | A kind of automatic clustering method and automatic cluster system of mechanical fault signals | |
| CN113551927A (en) | Mechanical equipment fault early warning method and system based on vibration signals | |
| CN113742993A (en) | Method, device, equipment and storage medium for predicting life loss of dry-type transformer | |
| CN113987905A (en) | Escalator braking force intelligent diagnosis system based on deep belief network | |
| CN116166992A (en) | Non-invasive load decomposition method and system based on multi-feature event classification | |
| CN117437933A (en) | Feature cluster combination generation type learning-based unsupervised detection method for fault of voiceprint signal of transformer | |
| CN118051601A (en) | Industrial Internet of things data fusion and analysis method | |
| CN117591857A (en) | Motor detection method based on deep learning | |
| CN111428772A (en) | Photovoltaic system depth anomaly detection method based on k-nearest neighbor adaptive voting | |
| CN118536082A (en) | Inverter fault prediction method and device based on artificial intelligence, terminal equipment and storage medium | |
| CN118050591A (en) | Distribution network 10KV overhead line fault early warning and positioning system based on AI | |
| CN117523836A (en) | System and method for predicting traffic flow by adopting graph neural network | |
| CN116126807A (en) | Log analysis method and related device | |
| CN112816788B (en) | Capacity value estimation method of locomotive traction converter direct current support capacitor and application system thereof | |
| CN114295981A (en) | Motor fault diagnosis method based on RN network current multi-feature form recognition | |
| CN112850408A (en) | Elevator emergency stop trapped person fault detection method based on multi-model fusion |
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 |