CN116226754A - Equipment health state assessment method and system based on equipment modeling - Google Patents
Equipment health state assessment method and system based on equipment modeling Download PDFInfo
- Publication number
- CN116226754A CN116226754A CN202310484475.1A CN202310484475A CN116226754A CN 116226754 A CN116226754 A CN 116226754A CN 202310484475 A CN202310484475 A CN 202310484475A CN 116226754 A CN116226754 A CN 116226754A
- Authority
- CN
- China
- Prior art keywords
- slice
- time sequence
- time
- historical
- category
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 230000036541 health Effects 0.000 title claims abstract description 11
- 238000006731 degradation reaction Methods 0.000 claims description 16
- 230000015556 catabolic process Effects 0.000 claims description 9
- 230000003862 health status Effects 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 4
- 238000012545 processing Methods 0.000 abstract description 3
- 230000005856 abnormality Effects 0.000 abstract description 2
- 238000004364 calculation method Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 206010033799 Paralysis Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000009194 climbing Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000005507 spraying Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
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
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention relates to the technical field of electric signal processing, and provides a device health state assessment method and system based on device modeling, wherein the method comprises the following steps: acquiring the history and current vibration signals of passenger conveying equipment, acquiring corresponding bearing information, and carrying out time sequence slicing on the history and current vibration signals; acquiring experience fault probability of each historical time sequence slice; acquiring first characteristics according to bearing information corresponding to the time sequence slice, acquiring second characteristics according to signal amplitude expression, and classifying according to the first characteristics and the second characteristics to obtain a plurality of first categories; acquiring a matched first category of each time sequence slice on the same day according to the first characteristic and the second characteristic of each time sequence slice on the same day; and according to the experience fault probability of the current time sequence slice matched with each historical time sequence slice in the first category, acquiring the fault prediction probability of the current equipment, evaluating the equipment state and taking corresponding measures. The invention aims to solve the problem that the abnormality detection of the traditional equipment is limited to the performance of a vibration signal.
Description
Technical Field
The invention relates to the field of electric signal processing, in particular to a device health state assessment method and system based on device modeling.
Background
Many cities in China are opened or track traffic is under construction, and subway electromechanical equipment in the track traffic comprises passenger conveying equipment, an AFC (automatic control) system, a central air conditioner, a fire-fighting spraying system, a power supply control system and the like; the device is similar to a step escalator, a spiral escalator, a belt type moving pavement, a double-line moving pavement and the like, belongs to passenger conveying equipment in rail traffic, is used as special equipment necessary for carrying passengers at a station, and has huge daily carrying capacity in subway urban rails; the equipment moving parts are exposed outside, so that the subway operation is briefly paralyzed once the failure is light, and the personal safety of passengers is threatened when the failure is heavy; therefore, the safety operation of the passenger conveying equipment is ensured, and the improvement of maintenance, repair and management levels is a problem that the operation units and the production enterprises need to continuously study, excavate and improve.
The passenger conveyor is composed of two conveyor belts, one is a chain conveyor belt, the other is a friction handrail conveyor belt, both conveyor belts are dragged by the same driving main shaft, and the linear speeds of the conveyor belts are kept consistent. The prior published patent CN215931068U proposes to convert vibration in the running process of the conveying equipment into an electric signal by using a sensor and collect the electric signal, and to help maintenance personnel to determine the cause of the fault by analyzing the abnormality of the vibration electric signal and combining with the historical fault, or to predict the fault of the passenger conveying equipment running currently; however, the traditional vibration electric signal anomaly analysis method is almost realized through anomaly threshold values and data fluctuation, is too dependent on an algorithm, is limited to waveform anomalies shown by the vibration signals, and has no high practicability in actual fault prediction.
Disclosure of Invention
The invention provides a device health state assessment method and a device health state assessment system based on device modeling, which aim to solve the problem that the existing traditional device anomaly detection is limited to the representation of vibration signals, and the adopted technical scheme is as follows:
in a first aspect, an embodiment of the present invention provides a device health status assessment method based on device modeling, the method including the steps of:
acquiring a history and current vibration signal of equipment, dividing the history and current vibration signal according to time to acquire a plurality of history time sequence slices and current time sequence slices, and recording the number of passengers and the number of passengers in a motion state in a period corresponding to each time sequence slice;
acquiring a plurality of F nodes in the historical vibration signals, and acquiring the experience fault probability of each historical time sequence slice according to the time difference between the historical time sequence slice and the F nodes;
decomposing the vibration signal of each time sequence slice to obtain the number of high-frequency signals and the number of low-frequency signals in each time sequence slice, and obtaining the first characteristic of each time sequence slice according to the ratio of the number of the high-frequency signals to the number of the low-frequency signals in each time sequence slice and the number of passengers in the corresponding time period and the number of passengers in the motion state;
trend fitting is carried out on the vibration signals of each time sequence slice to obtain a plurality of fitting differences of each time sequence slice, and second characteristics of each time sequence slice are obtained according to the fitting differences; obtaining the similarity of any two historical time sequence slices according to the first features and the second features, and obtaining a plurality of first categories according to the similarity;
according to the first characteristic and the second characteristic of each time sequence slice on the same day and the difference between the first characteristic and the second characteristic of each history time sequence slice in each first category, the difference degree of each time sequence slice on the same day and each first category is obtained, and the first category with the smallest difference degree in a plurality of difference degrees of each time sequence slice on the same day is used as the matched first category of each time sequence slice on the same day;
and acquiring the fault prediction probability of the current day equipment according to the experience fault probability of all the current day time sequence slices matched with each historical time sequence slice in the first category.
Optionally, the obtaining the experience fault probability of each historical time sequence slice according to the time difference between the historical time sequence slice and the F node includes the following specific methods:
wherein ,represents the empirical failure probability of the ith historical timing slice during degradation of the v-th F node,/>Represents the time of the ith history timing slice during degradation of the v-th F node, +.>Indicating the time at which the v F node appears,indicate->The time of occurrence of the F nodes; the time sequence slices correspond to a period of time, and the time center value of each time sequence slice is taken as the time of each time sequence slice.
Optionally, the acquiring the first feature of each time sequence slice includes the following specific methods:
wherein ,first feature representing the jth sequential slice, < >>Indicating the number of passengers in motion state in the j-th time slice,/->Indicating the number of passengers in the j-th time slice,/->Representing the ratio of the number of high frequency signals to the number of low frequency signals in the j-th timing slice.
Optionally, the trend fitting is performed on the vibration signal of each time sequence slice to obtain a plurality of fitting differences of each time sequence slice, and the second feature of each time sequence slice is obtained according to the fitting differences, including the following specific methods:
trend fitting is carried out on the vibration signals of each time sequence slice to obtain a fitting function, the positions of wave crests and wave troughs in the vibration signals of each time sequence slice are taken as amplitude points, and the difference between the signal amplitude of each amplitude point and the fitting function value is taken as the fitting difference of each amplitude point;
and taking the product of the difference between the maximum value and the minimum value of the fitting difference in each time sequence slice and the variance of the fitting difference as a second characteristic of each time sequence slice.
Optionally, the obtaining the similarity of any two historical time sequence slices according to the first feature and the second feature, and obtaining a plurality of first categories according to the similarity includes the following specific methods:
wherein ,representing the similarity of the jth time slice and the kth time slice,/for each time slice>And->First characteristics of the jth time series slice and the kth time series slice are respectively represented, and +.>And->Representing second features of the jth time series slice and the kth time series slice, respectively;
classifying two historical time sequence slices with similarity larger than a first preset threshold value into one category, classifying all the historical time sequence slices, and marking the obtained classification result as a plurality of first categories.
Optionally, the method for obtaining the difference degree between each time sequence slice on the same day and each first category includes the following specific steps:
wherein ,representing the number of history time slices in the q-th first category,/for each of the first and second categories>First feature representing the p-th day time slice,>representing the first characteristic of the nth history time slice in the qth first category, +.>Second feature representing the p-th day time slice,>representing a second characteristic of an nth historical timing slice in the qth first category.
Optionally, the method for obtaining the failure prediction probability of the current day equipment includes the following specific steps:
wherein ,representing the probability of failure prediction of the current day device, F representing the number of current day time series slices already recorded on the current day, W representing the number of historical time series slices in the historical vibration signal, +.>Representing the number of history time slices in the matching first category of the p-th current day time slice,/>Representing the empirical fault probability mean of the p-th current day time slice matching all of the historical time slices in the first category.
In a second aspect, another embodiment of the present invention provides a device health status assessment system based on device modeling, the system comprising:
the data acquisition module acquires historical and current vibration signals of the equipment, divides the historical and current vibration signals according to time to acquire a plurality of historical time sequence slices and current time sequence slices, and records the number of passengers and the number of passengers in a motion state in a period corresponding to each time sequence slice;
and a signal matching module: acquiring a plurality of F nodes in the historical vibration signals, and acquiring the experience fault probability of each historical time sequence slice according to the time difference between the historical time sequence slice and the F nodes;
decomposing the vibration signal of each time sequence slice to obtain the number of high-frequency signals and the number of low-frequency signals in each time sequence slice, and obtaining the first characteristic of each time sequence slice according to the ratio of the number of the high-frequency signals to the number of the low-frequency signals in each time sequence slice and the number of passengers in the corresponding time period and the number of passengers in the motion state;
trend fitting is carried out on the vibration signals of each time sequence slice to obtain a plurality of fitting differences of each time sequence slice, and second characteristics of each time sequence slice are obtained according to the fitting differences; obtaining the similarity of any two historical time sequence slices according to the first features and the second features, and obtaining a plurality of first categories according to the similarity;
according to the first characteristic and the second characteristic of each time sequence slice on the same day and the difference between the first characteristic and the second characteristic of each history time sequence slice in each first category, the difference degree of each time sequence slice on the same day and each first category is obtained, and the first category with the smallest difference degree in a plurality of difference degrees of each time sequence slice on the same day is used as the matched first category of each time sequence slice on the same day;
a state evaluation module: and acquiring the fault prediction probability of the current day equipment according to the experience fault probability of all the current day time sequence slices matched with each historical time sequence slice in the first category.
Compared with the prior art, the invention has the beneficial effects that: carrying out slicing processing on vibration signals of passenger conveying equipment in historical data, synchronously recording bearing information in each time sequence slice, taking the relevance of the bearing information and a signal structure and the noise content in the signal as the characteristics of each time sequence slice, and constructing a similarity model for classification; matching the current real-time vibration signal with the historical data category to match the average probability of the time sequence slices in the P-F interval in the completed historical data category, and taking the average probability as the fault prediction probability of the current passenger conveying equipment; compared with the traditional method, the method has the advantages that the real-time fault prediction result is more accurate by extracting and matching the characteristics of the historical data, and the real-time fault prediction result can be mutually verified with the historical data, so that the reliability and the robustness of the prediction result are greatly improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, it being obvious that the drawings in the description below are only some embodiments of the invention, and that other drawings can be obtained according to these drawings without inventive faculty for a person skilled in the art.
FIG. 1 is a schematic flow chart of a device health status assessment method based on device modeling according to an embodiment of the present invention;
FIG. 2 is a block diagram of a device health status assessment system based on device modeling according to another embodiment of the present invention;
fig. 3 is a schematic diagram of degradation curves.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, a flowchart of a device health status assessment method based on device modeling according to an embodiment of the present invention is shown, and the method includes the following steps:
step S001, acquiring the history and current day vibration signals of the passenger conveying equipment, simultaneously acquiring corresponding bearing information, and carrying out time sequence slicing on the history and current day vibration signals.
The objective of the present embodiment is to predict the failure probability of the passenger conveyor on the day according to the matching relationship between the vibration signal on the day and the historical vibration signal, so that the vibration signal on the day and the vibration signal collected in the history need to be obtained first, and meanwhile, in order to make the matching relationship more accurate, the bearing information corresponding to the time relationship between the vibration signal on the day and the historical vibration signal needs to be collected.
The method for specifically acquiring the vibration signal and the bearing information comprises the following steps: the mechanical vibration signal sensor is arranged on a main machine, a reduction gearbox, step vibration, a step chain wheel, a chain wheel bearing and other parts of the passenger conveying equipment, the vibration signal of the passenger conveying equipment is continuously recorded from the installation date of the equipment, and the time span of the collected historical vibration signal is limited to be more than one year and can be applied to the fault detection and prediction of the current date of the equipment; shooting passengers on the passenger conveying equipment by using a monitoring system above the station, inputting the passengers into a person recognition module and an action recognition module, and acquiring the number of the passengers and the movement state of the passengers on the passenger conveying equipment at each moment and taking the number of the passengers and the movement state of the passengers as corresponding bearing information; the motion state is that when the speed of the passenger in the motion direction is greater than the linear speed of the passenger conveying equipment, the passenger is considered to be in the motion state, and the passenger speed is equal to the linear speed of the conveying equipment, namely, the passenger is in a static state; it should be noted that, the person recognition module and the action recognition module both adopt the prior art, and this embodiment is not repeated.
Further, since the vibration signal is continuous in time sequence, the vibration signal can be split into a plurality of time sequence slices, the vibration signal and corresponding bearing information in the same time sequence slice are synchronously stored so as to facilitate subsequent calculation, and the concrete method for splitting the vibration signal into a plurality of time sequence slices is as follows: the ratio of the length of a conveyor belt in the passenger conveying equipment to the running speed is used as the time length of a time sequence slice, namely the time taken by the equipment to run for one circle is used as the time length of one time sequence slice, the vibration signals, the number of passengers and the number of passengers in the same time sequence slice are recorded in the same time sequence slice, and the historical vibration signals, the current-day vibration signals and the corresponding bearing information are respectively recorded in a plurality of historical time sequence slices and the current-day time sequence slice.
Step S002, obtaining the experience fault probability of each historical time sequence slice according to the time performance of the degradation curve and the historical time sequence slice.
Referring to fig. 3, a schematic diagram of a degradation curve is shown, where an abscissa indicates time, an ordinate indicates device performance, a point a indicates that the device performance starts to degrade, a point P indicates that a latent fault occurs in the device, a point F indicates that a functional fault occurs in the device, a point P' indicates a time corresponding to the latent fault, a point F indicates a time corresponding to the functional fault, and an interval from the latent fault to the functional fault is referred to as a state detection interval and is denoted as a P-F interval.
It should be noted that the potential failure refers to a degradation phenomenon that is shown before the occurrence of the functional failure, and is obtained from external observation, and is not that the failure has occurred; the P-F interval is quite common in the field of equipment management and is the optimal equipment maintenance period, namely, the period of time when the equipment shows obvious degradation phenomenon but has not failed; for historical vibration signals of passenger conveying equipment, wherein F nodes with a plurality of faults found exist, all vibration signals between every two adjacent F nodes are degradation processes of a later F node in the two F nodes, and the degradation process of a first F node is all vibration signals when equipment is installed between the first F node; for each historical timing slice in each degradation process, the closer it is to the next F node in time, the greater the empirical probability of failure for that historical timing slice.
Specifically, taking the ith historical time sequence slice in the degradation process of the ith F node as an example, the empirical fault probability of the historical time sequence slice is obtainedThe calculation method of (1) is as follows:
wherein ,representing the time of the ith historical time sequence slice in the degradation process of the v-th F node, wherein the time sequence slice corresponds to a period of time, and taking the time center value of each time sequence slice as the time of each time sequence slice; />Representing the time of occurrence of the v-th F node, < >>Indicate->The time of occurrence of the F nodes; specially, when->When (I)>I.e. < ->Namely the installation time of the equipment; />Representing the total time taken for the degradation process of the v-th F node,/for>The time distance between the ith historical time sequence slice and the v F node in the degradation process of the v F node is shown; at this time, the smaller the time distance is at the duty ratio of the total time, the closer the time distance of the time-series slice F node is, the larger the empirical failure probability is.
And acquiring the experience fault probabilities of all the historical time sequence slices according to the method, and predicting the fault probability of the current time sequence slice according to the probability expression of the historical time sequence slices.
Step S003, a first characteristic of each time sequence slice is obtained according to the associated expression between the bearing information corresponding to the time sequence slice and the signal structure, a second characteristic of each time sequence slice is obtained according to the signal amplitude expression of the time sequence slice, a similarity model is built according to the first characteristic and the second characteristic, and the historical time sequence slices are classified to obtain a plurality of first categories.
It should be noted that, since the running power of the conveying device varies with the load capacity, when the number of passengers is different, there may be a large difference in the vibration signals collected by the sensors; for example, an increase in operating power may cause the signal amplitude to become large and create a high frequency noise signal in the signal; when the borne passengers move, such as climbing a ladder, the vibration generated by the movement of the passengers belongs to low-frequency vibration compared with mechanical vibration, so that a large amount of low-frequency noise signals can be generated due to the influence of the movement of the person in the collected vibration signals; the bearing information of each time sequence slice is different, namely the number of passengers and the number of passengers in a motion state are different, the bearing information of each time sequence slice is quantized through the relation between a high-frequency component and a low-frequency component in a vibration signal in each time sequence slice, and the quantized association relation is used as a first characteristic of each time sequence slice.
Specifically, the vibration signal is decomposed through wavelet transformation, the vibration signal of each time sequence slice is decomposed by using a Mallat algorithm, the vibration signal is decomposed into a high-frequency signal component and a low-frequency signal component, the number of amplitude points in the signal component is recorded as the number of signals, the number of the amplitude points is the position of a wave crest and a wave trough, and the jth time sequence slice is taken as an example, so that the bearing information of the jth time sequence slice has the following relation with the number of the high-frequency and the low-frequency signals:
wherein ,representing the ratio of the number of high frequency signals to the number of low frequency signals in the j-th time slice, +.>Indicating the number of passengers in the j-th time slice,/->Indicating the number of passengers in motion state in the j-th time slice,/->Representing the correlation coefficient of the bearing information and the quantity of the high-frequency and low-frequency signals in the j-th time sequence slice; the number of passengers affects the high-frequency components of the vibration signals, the more the number of passengers is, the larger the running power of equipment is, the more the high-frequency components in the vibration signals are, and the more the number of the high-frequency signals is; the number of passengers in a motion state affects the low-frequency noise component of the vibration signal, the more the number of passengers in a motion state is, the more the generated low-frequency noise is, the more the number of low-frequency signals is, and the load information is the number of passengers in a time sequence slice and the number of passengers in the motion state, so that the relation between the load information and the numbers of high-frequency and low-frequency signals is determined.
According to the above relation, there is a unique correlation coefficient for the j-th time sequence sliceThe calculation method comprises the following steps:
wherein ,indicating the number of passengers in motion state in the j-th time slice,/->Indicating the number of passengers in the j-th time slice,/->Representing the ratio of the number of high frequency signals to the number of low frequency signals in the jth time sequence slice; association coefficient +.>As the quantized association relation between the bearing information of the jth time sequence slice and the signal structure, and as the first characteristic of the jth time sequence slice; and acquiring first characteristics of all the historical time sequence slices and the current time sequence slices according to the method, and using the first characteristics for subsequent classification of the historical time sequence slices and matching of the current time sequence slices.
It should be further noted that, due to the difference of the operating power of the devices in each time-series slice, the amplitude of the vibration signal is different, and the high-frequency noise represented by the different differences is accompanied, so that the second characteristic of each time-series slice is obtained according to the amplitude of the signal and the high-frequency noise.
Specifically, trend item fitting is firstly carried out on the vibration signals in each time sequence slice, namely least square fitting is carried out on all amplitude points in the vibration signals of each time sequence slice, and a fitting function is obtainedThe amplitude point is the position of the wave crest and the wave trough; it should be noted that, the signal noise has the characteristics of high frequency and discrete, so when the trend item is fitted, according to the principle that the sum of the distances between the amplitude point and the trend fitted curve is minimum, the difference between the discrete signal noise and the trend item is necessarily larger; if a discrete noise signal is present in the vibration signal, the difference between the vibration signal and the fitting term increases, and the larger the difference value, the more discrete noise in the vibration signal.
Taking the j-th time sequence slice as an example, the second characteristic of the time sequence slice is obtainedThe calculation method of (1) is as follows:
wherein ,representing the fitting difference of the jth amplitude point in the jth time series slice,/for>Representing a fitting function value representing the jth amplitude point in the jth time series slice, +.>Representing the signal amplitude of the ith amplitude point in the jth timing slice;
wherein ,and->Representing the maximum fitting difference and the minimum fitting difference of all amplitude points in the jth time series slice, respectively, +.>Representing the variance of the fitting differences for all amplitude points in the jth time series slice; quantifying the noise content representing the time series slice by the range of the fitting difference of all amplitude points in each time series slice, namely the difference value of the maximum value and the minimum value of the fitting difference and the variance of the fitting difference, wherein the range and the variance of the fitting difference are the unique characteristics of each time series slice, and the second characteristics of each time series slice are obtained according to the method and are used for classifying the historical time series sliceAnd matching of time-of-day slices.
It should be further noted that, at this time, the first feature and the second feature of each time sequence slice have been obtained, where the first feature reflects the association relationship between the vibration signal and the bearing information in the time sequence slice, and the second feature reflects the waveform expression of the vibration signal in the time sequence slice, and if the first feature and the second feature of any two time sequence slices are similar, it indicates that the vibration signals of the two time sequence slices are similar to the bearing information, and the corresponding operation states of the devices are similar.
Specifically, a similarity model is constructed according to the first feature and the second feature of each time sequence slice, and the calculation method is as follows:
wherein ,representing the similarity of the jth time slice and the kth time slice,/for each time slice>And->First characteristics of the jth time series slice and the kth time series slice are respectively represented, and +.>And->Representing second features of the jth time series slice and the kth time series slice, respectively; the similarity of the two time sequence slices is obtained by adopting a cosine similarity calculation method, and the closer the similarity is to 1, the closer the corresponding equipment running states of the two time sequence slices are, the more similar the equipment running states are
Further, the similarity of any two historical time sequence slices in all the historical time sequence slices is obtained according to the method, and a first preset threshold value is givenThe present embodiment employs +.>And (3) calculating, namely classifying two historical time sequence slices with similarity larger than a first preset threshold value into one category, classifying all the historical time sequence slices, ensuring that the similarity of any two historical time sequence slices in the same category is larger than the first preset threshold value, and marking the classification result obtained at the moment as a plurality of first categories.
So far, classifying all the historical time sequence slices to obtain a plurality of first categories, wherein the running states of the equipment corresponding to the historical time sequence slices in the same first category are similar or identical, and the running states of the equipment corresponding to the historical time sequence slices in different first categories are greatly different.
Step S004, according to the first characteristic and the second characteristic of each time sequence slice on the same day and the difference expression of the first characteristic and the second characteristic of each history time sequence slice in each first category, the matching first category of each time sequence slice on the same day is obtained.
It should be noted that, the similar or identical running states of the devices include similar or identical vibration signal structures and similar or identical bearing information between the time sequence slices, and then the historical time sequence slices corresponding to the current time sequence slice can be used as references to obtain the matching first category of each current time sequence slice.
Specifically, firstly, the matching degree of each current time sequence slice and each first category is obtained, taking the p-th current time sequence slice as an example, the difference degree of the p-th current time sequence slice and the q-th first categoryThe calculation method of (1) is as follows:
wherein ,represents the q-thThe number of historical timing slices in the first category, +.>First feature representing the p-th day time slice,>representing the first characteristic of the nth history time slice in the qth first category, +.>Second feature representing the p-th day time slice,>a second feature representing an nth historical timing slice in a qth first category; the matching relation between the current time sequence slice and the historical time sequence slice is represented by calculating the Euclidean norm of the difference value of the first and second characteristics, and the matching degree between the current time sequence slice and the first category is represented by the Euclidean norm mean value; the smaller the Euclidean norm of the difference value between the first and second features is, the closer the current time sequence slice is to the running state of the equipment corresponding to the historical time sequence slice, the smaller the Euclidean norm average value is, and the current time sequence slice is more similar to the running state of the equipment corresponding to all the historical time sequence slices in the first category; and obtaining the difference degree of the p-th time sequence slice and all the first categories according to the method, and taking the first category with the smallest difference degree as the matched first category of the time sequence slice.
According to the method, the first category matched with each time sequence slice on the same day is obtained, the historical time sequence slices in the first category matched with the time sequence slices on the same day are similar to the running state of equipment on the time sequence slices on the same day, and the fault probability of the time sequence slices on the same day can be predicted.
And S005, acquiring the failure prediction probability of the passenger conveying equipment on the same day according to the experience failure probability of all the time sequence slices on the same day, which are matched with each history time sequence slice in the first category, evaluating the equipment state according to the failure prediction probability, and taking corresponding measures.
It should be noted that, in step S002, the empirical fault probability of each historical time series slice has been obtained, and the fault probability of the current time series slice is predicted according to the empirical fault probability average value of all the historical time series slices in the first category, the larger the probability average value, the larger the empirical fault probability of the current time series slice, the more likely the fault will occur; meanwhile, as the historical time sequence slices with faults are fewer and cannot exist in the first category with larger volume, the larger the number of the historical time sequence slices in the first category is, the more normal the vibration signals of the historical time sequence slices are, and the lower the fault probability of the time sequence slices on the same day is matched with the vibration signals.
Specifically, the failure prediction probability of the passenger conveying device on the same day is obtainedThe calculation method of (1) is as follows:
wherein F represents the number of time-series slices of the day that have been recorded on the same day, W represents the number of time-series slices of the history in the history vibration signal,representing the number of history time slices in the matching first category of the p-th current day time slice,/>Representing an empirical fault probability mean value of all historical time sequence slices in the first category matched with the p-th current time sequence slice; at this time, the larger the number of the history time series slices in the first category is matched, the larger the ratio of the total number of the history time series slices is, and the smaller the fault probability of the time series slices on the same day matched with the total number of the history time series slices is; the larger the average value of the experience fault probabilities of all the historical time sequence slices in the first category is matched, the larger the probability of faults of the time sequence slices on the same day is matched with the experience fault average value; and (3) averaging the fault probabilities of all the time sequence slices of the current day, which are recorded on the current day, and obtaining the fault prediction probability of the passenger conveying equipment of the current day.
According to the failure prediction probability of the passenger conveying equipment on the same day, giveA second preset threshold valueFor fault determination, the present embodiment employs +.>And calculating, namely, when the fault prediction probability is larger than a second preset threshold value, considering that the equipment possibly fails, the state of the equipment is unhealthy, carrying out fault pre-alarming, and dispatching maintenance personnel to check and maintain the equipment.
Referring to fig. 2, a block diagram of a device health status assessment system based on device modeling according to another embodiment of the present invention is shown, where the system includes:
the data acquisition module S101 acquires the history and current day vibration signals of the passenger conveying equipment, acquires corresponding bearing information at the same time, and performs time sequence slicing on the history and current day vibration signals.
Signal matching module S102:
(1) Acquiring experience fault probability of each historical time sequence slice;
(2) Acquiring first characteristics of each time sequence slice according to the associated expression between the bearing information corresponding to the time sequence slice and the signal structure, acquiring second characteristics of each time sequence slice according to the signal amplitude expression of the time sequence slice, constructing a similarity model according to the first characteristics and the second characteristics, and classifying historical time sequence slices to obtain a plurality of first categories;
(3) And obtaining a matched first category of each current time sequence slice according to the difference expression of the first characteristic and the second characteristic of each current time sequence slice and the first characteristic and the second characteristic of each historical time sequence slice in each first category.
The state evaluation module S103 obtains the failure prediction probability of the passenger conveying device on the same day according to the experience failure probability of all the time sequence slices on the same day matching each history time sequence slice in the first category, evaluates the device state according to the failure prediction probability and takes corresponding measures.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (8)
1. A device health state assessment method based on device modeling, characterized in that the method comprises the following steps:
acquiring a history and current vibration signal of equipment, dividing the history and current vibration signal according to time to acquire a plurality of history time sequence slices and current time sequence slices, and recording the number of passengers and the number of passengers in a motion state in a period corresponding to each time sequence slice;
acquiring a plurality of F nodes in the historical vibration signals, and acquiring the experience fault probability of each historical time sequence slice according to the time difference between the historical time sequence slice and the F nodes;
decomposing the vibration signal of each time sequence slice to obtain the number of high-frequency signals and the number of low-frequency signals in each time sequence slice, and obtaining the first characteristic of each time sequence slice according to the ratio of the number of the high-frequency signals to the number of the low-frequency signals in each time sequence slice and the number of passengers in the corresponding time period and the number of passengers in the motion state;
trend fitting is carried out on the vibration signals of each time sequence slice to obtain a plurality of fitting differences of each time sequence slice, and second characteristics of each time sequence slice are obtained according to the fitting differences; obtaining the similarity of any two historical time sequence slices according to the first features and the second features, and obtaining a plurality of first categories according to the similarity;
according to the first characteristic and the second characteristic of each time sequence slice on the same day and the difference between the first characteristic and the second characteristic of each history time sequence slice in each first category, the difference degree of each time sequence slice on the same day and each first category is obtained, and the first category with the smallest difference degree in a plurality of difference degrees of each time sequence slice on the same day is used as the matched first category of each time sequence slice on the same day;
and acquiring the fault prediction probability of the current day equipment according to the experience fault probability of all the current day time sequence slices matched with each historical time sequence slice in the first category.
2. The method for evaluating the health state of equipment based on equipment modeling according to claim 1, wherein the obtaining the empirical fault probability of each historical time series slice according to the time difference between the historical time series slice and the F node comprises the following specific steps:
wherein ,represents the empirical failure probability of the ith historical timing slice during degradation of the v-th F node,/>Represents the time of the ith history timing slice during degradation of the v-th F node, +.>Representing the time of occurrence of the v-th F node, < >>Indicate->The time of occurrence of the F nodes; the time sequence slices correspond to a period of time, and the time center value of each time sequence slice is taken as the time of each time sequence slice.
3. The method for evaluating the health status of a device based on device modeling according to claim 1, wherein the step of obtaining the first characteristic of each time series slice comprises the following specific steps:
wherein ,first feature representing the jth sequential slice, < >>Indicating the number of passengers in motion state in the j-th time slice,/->Indicating the number of passengers in the j-th time slice,/->Representing the ratio of the number of high frequency signals to the number of low frequency signals in the j-th timing slice.
4. The method for evaluating the health state of equipment based on equipment modeling according to claim 1, wherein the trend fitting is performed on the vibration signal of each time series slice to obtain a plurality of fitting differences of each time series slice, and the second characteristic of each time series slice is obtained according to the fitting differences, comprising the following specific steps:
trend fitting is carried out on the vibration signals of each time sequence slice to obtain a fitting function, the positions of wave crests and wave troughs in the vibration signals of each time sequence slice are taken as amplitude points, and the difference between the signal amplitude of each amplitude point and the fitting function value is taken as the fitting difference of each amplitude point;
and taking the product of the difference between the maximum value and the minimum value of the fitting difference in each time sequence slice and the variance of the fitting difference as a second characteristic of each time sequence slice.
5. The method for evaluating the health state of equipment based on equipment modeling according to claim 1, wherein the method for obtaining the similarity of any two historical time sequence slices according to the first feature and the second feature and obtaining a plurality of first categories according to the similarity comprises the following specific steps:
wherein ,representing the similarity of the jth time slice and the kth time slice,/for each time slice>And->First characteristics of the jth time series slice and the kth time series slice are respectively represented, and +.>And->Representing second features of the jth time series slice and the kth time series slice, respectively;
classifying two historical time sequence slices with similarity larger than a first preset threshold value into one category, classifying all the historical time sequence slices, and marking the obtained classification result as a plurality of first categories.
6. The method for evaluating the health status of a device based on device modeling according to claim 1, wherein the step of obtaining the degree of difference between each time-of-day slice and each first category comprises the following specific steps:
wherein ,representing the number of history time slices in the q-th first category,/for each of the first and second categories>First feature representing the p-th day time slice,>representing the first characteristic of the nth history time slice in the qth first category, +.>Second feature representing the p-th day time slice,>representing a second characteristic of an nth historical timing slice in the qth first category.
7. The method for evaluating the health state of a device based on device modeling according to claim 1, wherein the obtaining the probability of failure prediction of the device on the same day comprises the following specific steps:
wherein ,representing the probability of failure prediction of the current day device, F representing the number of current day time series slices already recorded on the current day, W representing the number of historical time series slices in the historical vibration signal, +.>Representing the number of history time slices in the matching first category of the p-th current day time slice,/>Representing the empirical fault probability mean of the p-th current day time slice matching all of the historical time slices in the first category.
8. A device health status assessment system based on device modeling, the system comprising:
the data acquisition module acquires historical and current vibration signals of the equipment, divides the historical and current vibration signals according to time to acquire a plurality of historical time sequence slices and current time sequence slices, and records the number of passengers and the number of passengers in a motion state in a period corresponding to each time sequence slice;
and a signal matching module: acquiring a plurality of F nodes in the historical vibration signals, and acquiring the experience fault probability of each historical time sequence slice according to the time difference between the historical time sequence slice and the F nodes;
decomposing the vibration signal of each time sequence slice to obtain the number of high-frequency signals and the number of low-frequency signals in each time sequence slice, and obtaining the first characteristic of each time sequence slice according to the ratio of the number of the high-frequency signals to the number of the low-frequency signals in each time sequence slice and the number of passengers in the corresponding time period and the number of passengers in the motion state;
trend fitting is carried out on the vibration signals of each time sequence slice to obtain a plurality of fitting differences of each time sequence slice, and second characteristics of each time sequence slice are obtained according to the fitting differences; obtaining the similarity of any two historical time sequence slices according to the first features and the second features, and obtaining a plurality of first categories according to the similarity;
according to the first characteristic and the second characteristic of each time sequence slice on the same day and the difference between the first characteristic and the second characteristic of each history time sequence slice in each first category, the difference degree of each time sequence slice on the same day and each first category is obtained, and the first category with the smallest difference degree in a plurality of difference degrees of each time sequence slice on the same day is used as the matched first category of each time sequence slice on the same day;
a state evaluation module: and acquiring the fault prediction probability of the current day equipment according to the experience fault probability of all the current day time sequence slices matched with each historical time sequence slice in the first category.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310484475.1A CN116226754B (en) | 2023-05-04 | 2023-05-04 | Equipment health state assessment method and system based on equipment modeling |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310484475.1A CN116226754B (en) | 2023-05-04 | 2023-05-04 | Equipment health state assessment method and system based on equipment modeling |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116226754A true CN116226754A (en) | 2023-06-06 |
CN116226754B CN116226754B (en) | 2023-06-27 |
Family
ID=86577221
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310484475.1A Active CN116226754B (en) | 2023-05-04 | 2023-05-04 | Equipment health state assessment method and system based on equipment modeling |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116226754B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117235615A (en) * | 2023-11-14 | 2023-12-15 | 泰安维创游乐设备有限公司 | Potential safety hazard risk early warning system for large amusement park equipment |
CN117874544A (en) * | 2024-03-12 | 2024-04-12 | 徐州阿卡控制阀门有限公司 | Valve fault intelligent diagnosis method based on time sequence analysis |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101293529A (en) * | 2007-04-29 | 2008-10-29 | 余亚莉 | Intelligent monitoring and early warning system for passenger transportation ability and operation safety of vehicle mounted rail traffic |
US20160062950A1 (en) * | 2014-09-03 | 2016-03-03 | Google Inc. | Systems and methods for anomaly detection and guided analysis using structural time-series models |
CN111089726A (en) * | 2020-01-16 | 2020-05-01 | 东南大学 | Rolling bearing fault diagnosis method based on optimal dimension singular spectrum decomposition |
CN115941433A (en) * | 2021-08-16 | 2023-04-07 | 中国电信股份有限公司 | Network slice performance optimization guarantee method and system, storage medium and electronic equipment |
-
2023
- 2023-05-04 CN CN202310484475.1A patent/CN116226754B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101293529A (en) * | 2007-04-29 | 2008-10-29 | 余亚莉 | Intelligent monitoring and early warning system for passenger transportation ability and operation safety of vehicle mounted rail traffic |
US20160062950A1 (en) * | 2014-09-03 | 2016-03-03 | Google Inc. | Systems and methods for anomaly detection and guided analysis using structural time-series models |
CN111089726A (en) * | 2020-01-16 | 2020-05-01 | 东南大学 | Rolling bearing fault diagnosis method based on optimal dimension singular spectrum decomposition |
CN115941433A (en) * | 2021-08-16 | 2023-04-07 | 中国电信股份有限公司 | Network slice performance optimization guarantee method and system, storage medium and electronic equipment |
Non-Patent Citations (2)
Title |
---|
崔颖;赵军;赖欣欢;: "基于EEMD和变尺度随机共振的轴承故障诊断", 测控技术, no. 07 * |
赵阳: "城市轨道交通牵引电机轴承的故障识别与剩余寿命预测", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117235615A (en) * | 2023-11-14 | 2023-12-15 | 泰安维创游乐设备有限公司 | Potential safety hazard risk early warning system for large amusement park equipment |
CN117235615B (en) * | 2023-11-14 | 2024-01-30 | 泰安维创游乐设备有限公司 | Potential safety hazard risk early warning system for large amusement park equipment |
CN117874544A (en) * | 2024-03-12 | 2024-04-12 | 徐州阿卡控制阀门有限公司 | Valve fault intelligent diagnosis method based on time sequence analysis |
CN117874544B (en) * | 2024-03-12 | 2024-05-31 | 徐州阿卡控制阀门有限公司 | Valve fault intelligent diagnosis method based on time sequence analysis |
Also Published As
Publication number | Publication date |
---|---|
CN116226754B (en) | 2023-06-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN116226754B (en) | Equipment health state assessment method and system based on equipment modeling | |
CN109726524B (en) | CNN and LSTM-based rolling bearing residual service life prediction method | |
CN110647133B (en) | Rail transit equipment state detection maintenance method and system | |
US20190285517A1 (en) | Method for evaluating health status of mechanical equipment | |
Jardine et al. | A review on machinery diagnostics and prognostics implementing condition-based maintenance | |
CN110361180B (en) | Intelligent train pantograph service performance dynamic monitoring and evaluating method and system | |
CN108304941A (en) | A kind of failure prediction method based on machine learning | |
CN104091070A (en) | Rail transit fault diagnosis method and system based on time series analysis | |
Li et al. | Unsupervised machine anomaly detection using autoencoder and temporal convolutional network | |
CN107844067B (en) | A kind of gate of hydropower station on-line condition monitoring control method and monitoring system | |
CN109492790A (en) | Wind turbines health control method based on neural network and data mining | |
CN112766618B (en) | Abnormality prediction method and device | |
CN114417699A (en) | Pump valve fault detection method | |
CN111353640B (en) | Method for constructing wind speed prediction model by combination method | |
Martin-del-Campo et al. | Dictionary learning approach to monitoring of wind turbine drivetrain bearings | |
CN109583794A (en) | A kind of method of determining elevator failure time | |
CN114861349B (en) | Rolling bearing RUL prediction method based on model migration and wiener process | |
Atamuradov et al. | Degradation-level assessment and online prognostics for sliding chair failure on point machines | |
CN113435228A (en) | Motor bearing service life prediction and analysis method based on vibration signal modeling | |
Wang | Unsupervised anomaly detection in railway catenary condition monitoring using autoencoders | |
CN113987905A (en) | Escalator braking force intelligent diagnosis system based on deep belief network | |
CN117093911A (en) | Bridge sliding support abrasion identification method based on longitudinal temperature-induced displacement amplitude monitoring of main beam | |
Kim et al. | Anomaly Detection of an Air Compressor from Time-series Measurement Data | |
CN117992810B (en) | Road and bridge intensity detection device | |
Chen | Approaches for diagnosis and prognosis of asset condition: application to railway switch systems |
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 |