CN113837591A - Equipment health assessment method oriented to multi-working-condition operation conditions - Google Patents
Equipment health assessment method oriented to multi-working-condition operation conditions Download PDFInfo
- Publication number
- CN113837591A CN113837591A CN202111097810.XA CN202111097810A CN113837591A CN 113837591 A CN113837591 A CN 113837591A CN 202111097810 A CN202111097810 A CN 202111097810A CN 113837591 A CN113837591 A CN 113837591A
- Authority
- CN
- China
- Prior art keywords
- health
- working conditions
- parameters
- equipment
- working
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Abstract
The invention discloses a device health assessment method for multi-working-condition operation conditions, which comprises the following specific processes: screening working condition parameters and performance parameters sensitive to recession, and performing smooth noise reduction processing on data; dividing the operation working conditions of the equipment by a clustering method to obtain corresponding circulating working condition labels; calculating the MMD distance between the sample set to be tested and the health set under different working conditions by sliding the time window, and constructing an evaluation index representing the performance degradation state of the equipment by using the distance value; and the health indexes under different working conditions are combined to describe the degradation process of the equipment, so that the idea of health assessment under multiple working conditions is embodied.
Description
Technical Field
The invention relates to the technical field of health assessment, in particular to a health assessment method for equipment under multiple working conditions.
Background
Modern industry is continuously developing, the requirements of complex equipment systems such as aerospace vehicles, large-scale electrical equipment and the like on performance are increasingly improved, and the reliability and the safety of the complex equipment systems are also increasingly important. As one of the main activities of health management, health assessment assesses the health status of a system by analyzing sensor monitoring data, thereby providing support for maintenance decisions and ensuring smooth operation of the system. The data-driven method is a health assessment method which is widely applied at present, and because the method does not need to construct an accurate physical model and obtain a large amount of prior information and the mode of obtaining the equipment operation data is more and more perfect in recent years, the data-driven health assessment method has a wide application prospect.
In the existing data-driven health assessment method, for example, a fault degree evaluation matrix is established by using an analytic hierarchy process in a distribution equipment health state assessment method and a system CN201811510608.3, and equipment health state scores are obtained by using a fuzzy evaluation method and a weighting membership principle, and a subjective weighting method used by the method cannot fully reflect data information of evaluation indexes. An improved method for evaluating the health state of a spacecraft structure vibration test is CN201910596085.7, a vibration curve is collected according to a vibration monitoring sensor, a vibration transfer function curve is calculated, and whether health hidden dangers exist in a spacecraft is judged by comparing the vibration transfer function curve with the vibration transfer function curve. Meanwhile, the prior art carries out health assessment on the system under single-working-condition operation, and ignores the problem of frequent change of the operation working condition of large-scale complex equipment. For large equipment under complex operating conditions, monitoring data of the large equipment is multidimensional time sequence data, and the defects of the existing method have influence on the accuracy of health assessment.
Disclosure of Invention
The technical problem solved by the scheme provided by the embodiment of the invention is that the health assessment of various working conditions often existing in the operation process of large-scale complex equipment cannot be realized.
The equipment health assessment method facing the multi-working-condition operation condition is characterized by comprising the following steps:
dividing the obtained original equipment parameters into N working condition parameters and M individual performance parameters, and performing decay sensitive screening processing on the M individual performance parameters under different working conditions to obtain decay sensitive performance parameters;
obtaining a working condition label corresponding to each working cycle by dividing the N working condition parameters into N working conditions;
acquiring a training set and a test set of equipment to be tested, taking the training set and the test set of the equipment to be tested as a sample set to be tested, simultaneously determining a health set from the training set, and dividing the health set into n sub-health sets under different working conditions according to the working condition labels;
calculating MMD distance values of the sample set to be tested and the sub-health set under n different working conditions through a sliding window, and constructing a health evaluation index for representing a performance degradation state of the equipment by utilizing the MMD distance values under the n different working conditions;
merging the health evaluation indexes used for representing the equipment performance decline state to obtain a degradation curve graph after merging working conditions;
wherein, M, N and N are positive integers.
Preferably, the obtaining of the degradation-sensitive performance parameters by performing degradation-sensitive screening processing on the M performance parameters under different working conditions includes:
according to the change trend of the performance parameters, selecting a primary function and a secondary function to perform function fitting on the performance parameters to obtain the fitted performance parameters;
and establishing evaluation indexes according to the fitted performance parameters, and screening out the performance parameters with obvious decline trends and consistent change trends under different working conditions by comparing the evaluation indexes of the performance parameters.
Preferably, the evaluation index includes:
coefficient of the highest order term: linear slope k (y ═ kx + b) and conic a value (y ═ ax)2+ bx + c), selecting the parameter with larger variation trend as the evaluation parameter;
variance of absolute values of parameters and fit line: representing the dispersion degree of a parameter distribution distance fitting line, and setting a parameter with the variance less than 0.045 as an evaluation parameter;
percentage of points within upper and lower thresholds of the fit line: setting 0.2 above and below the fitting straight line and 0.15 above and below the fitting curve as threshold values, calculating the parameter percentage in the threshold value range, and taking the parameter with the point number of more than 75% in the setting range as an evaluation parameter.
Preferably, after obtaining the fading sensitive performance parameters, the method further comprises:
and carrying out normalization and smooth noise reduction processing on the decay sensitivity performance parameters, and simultaneously carrying out normalization processing on the working condition parameters so as to improve the overall quality of the parameters.
Preferably, the dividing the N operating condition parameters into N operating conditions includes:
and dividing the working condition parameters after the normalization processing into n working conditions by adopting a K-Means clustering method according to the type number n of the operating working conditions of the equipment.
Preferably, the training set refers to a time sequence from P health to failure containing decay sensitivity performance parameters; the test set of the equipment to be tested refers to a time sequence comprising decay sensitivity performance parameters, which is formed by Q devices and ends before failure.
Preferably, the determining the health set from the training set comprises: and selecting the initial healthy state of the training set as a healthy set.
Preferably, after calculating the MMD distance values of the sample set to be tested and the sub-healthy set under n different working conditions through a sliding window, the method further includes:
and carrying out sub-working condition normalization processing on the MMD distance value.
Preferably, the obtaining of the degradation curve graph after the combined working conditions by combining the health assessment indicators for characterizing the equipment performance degradation state includes:
corresponding the health evaluation indexes under different working conditions to the working cycles corresponding to the sliding time windows, and combining the health evaluation indexes according to the time sequence to obtain a health index sequence combining a plurality of working conditions;
and performing visualization processing on the health index sequence combined with the plurality of working conditions to obtain a degradation curve graph after the working conditions are combined.
Preferably, the method further comprises the following steps:
and smoothing the degradation curve graph after the working conditions are combined by using a local weighted regression method so as to more intuitively observe the health state change trend of the equipment.
According to the scheme provided by the embodiment of the invention, the method for evaluating the health of the equipment under multiple working conditions based on MMD (Maximum Mean difference) distance measurement relates to the method processes of parameter screening and data preprocessing of the equipment, working condition division of the equipment, construction of health indexes under multiple working conditions and the like. The specific flow of the method comprises the steps of screening working condition parameters and performance parameters sensitive to recession, and carrying out smooth noise reduction treatment on data; dividing the operation working conditions of the equipment by a clustering method to obtain corresponding circulating working condition labels; calculating the MMD distance between the sample set to be tested and the health set under different working conditions by sliding the time window, and constructing an evaluation index representing the performance decline state of the equipment by using the distance value; and the health indexes under different working conditions are combined to describe the degradation process of the equipment, so that the idea of health assessment under multiple working conditions is embodied. The method makes full use of the characteristics of complex equipment operation conditions and strong operation data time sequence, systematically establishes the health assessment model suitable for various operation conditions, and further improves the accuracy and the applicability of the health assessment.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:
FIG. 1 is a flow chart of an equipment health assessment method for multi-condition operating conditions according to an embodiment of the present invention;
FIG. 2 is a flowchart of an equipment health assessment method for multi-condition operating conditions according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a health index under multiple operating conditions constructed by the MMD method provided by the embodiment of the invention;
FIG. 4 is a chart of a parameter variation trend under the working conditions provided by the embodiment of the present invention;
FIG. 5 is a parameter graph provided by an embodiment of the present invention without function fitting;
FIG. 6 is a graph of a linear fit parameter screen provided by an embodiment of the present invention;
FIG. 7 is a graph of a quadratic curve fit parameter screening provided by an embodiment of the present invention;
FIG. 8 is a schematic illustration of data preprocessing for Engine number 2 parameter 4 provided by an embodiment of the present invention;
FIG. 9 is a schematic diagram of CV value sets under different operating conditions of a number 32 test engine according to an embodiment of the present invention;
fig. 10 is a visual graph of a training engine No. 220 and a testing engine No. 32 provided by an embodiment of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, and it should be understood that the preferred embodiments described below are only for the purpose of illustrating and explaining the present invention, and are not to be construed as limiting the present invention.
Fig. 1 is a flowchart of an equipment health assessment method for a multi-condition operating condition according to an embodiment of the present invention, as shown in fig. 1, including:
step S101: dividing the obtained original equipment parameters into N working condition parameters and M individual performance parameters, and performing decay sensitivity screening processing on the M individual performance parameters under different working conditions to obtain decay sensitivity performance parameters;
step S102: obtaining a working condition label corresponding to each working cycle by dividing the N working condition parameters into N working conditions;
step S103: acquiring a training set and a test set of equipment to be tested, taking the training set and the test set of the equipment to be tested as a sample set to be tested, simultaneously determining a health set from the training set, and dividing the health set into n sub-health sets under different working conditions according to the working condition labels;
step S104: calculating MMD distance values of the sample set to be tested and the sub-healthy set under n different working conditions through a sliding window, and constructing a health evaluation index for representing the equipment performance decline state by utilizing the MMD distance values under the n different working conditions;
step S105: merging the health evaluation indexes used for representing the equipment performance decline state to obtain a degradation curve graph after the working conditions are merged;
wherein, M, N and N are positive integers.
The screening processing of recession sensitivity is carried out on the M performance parameters under different working conditions, and the obtaining of the recession sensitivity performance parameters comprises the following steps: according to the change trend of the performance parameters, selecting a primary function and a secondary function to perform function fitting on the performance parameters to obtain the fitted performance parameters; and establishing evaluation indexes according to the fitted performance parameters, and screening out the performance parameters with obvious decline trends and consistent change trends under different working conditions by comparing the evaluation indexes of the performance parameters.
Specifically, the evaluation index includes:
coefficient of the highest order term: linear slope k (y ═ kx + b) and conic a value (y ═ ax)2+ bx + c), selecting the parameter with larger variation trend as the evaluation parameter;
variance of absolute values of parameters and fit line: representing the dispersion degree of a parameter distribution distance fitting line, and setting a parameter with the variance less than 0.045 as an evaluation parameter;
percentage of points within upper and lower thresholds of the fit line: setting 0.2 above and below the fitting straight line and 0.15 above and below the fitting curve as threshold values, calculating the parameter percentage in the threshold value range, and taking the parameter with the point number of more than 75% in the setting range as an evaluation parameter.
After obtaining the parameter of the decline sensitivity performance, the invention also comprises: and carrying out normalization and smooth noise reduction treatment on the decay sensitivity performance parameters, and simultaneously carrying out normalization treatment on the working condition parameters so as to improve the overall quality of the parameters.
Wherein the dividing the N operating condition parameters into N operating conditions comprises: and dividing the normalized working condition parameters into n working conditions by adopting a K-Means clustering method according to the type number n of the operating working conditions of the equipment.
Specifically, the training set refers to a time sequence from P from health to failure, wherein the time sequence comprises decay sensitivity performance parameters; the test set of the equipment to be tested refers to a time sequence comprising decay sensitivity performance parameters, which is formed by Q devices and ends before failure. Wherein the determining a health set from the training set comprises: and selecting the initial health state of the training set as a health set.
After the MMD distance values of the sample set to be tested and the sub-health set under n different working conditions are calculated through the sliding window, the method also comprises the following steps: and carrying out sub-working condition normalization processing on the MMD distance value.
The obtaining of the degradation curve graph after the working conditions are combined by combining the health assessment indexes for representing the equipment performance degradation state comprises the following steps:
corresponding the health evaluation indexes under different working conditions to the working cycles corresponding to the sliding time windows, and combining the health evaluation indexes according to the time sequence to obtain a health index sequence combining a plurality of working conditions;
and performing visualization processing on the health index sequence combined with the plurality of working conditions to obtain a degradation curve graph after the working conditions are combined.
The invention also includes: and smoothing the degradation curve graph after the working conditions are combined by using a local weighted regression method so as to more intuitively observe the health state change trend of the equipment.
The invention carries out parameter screening and data preprocessing on the monitoring data of the sensor and divides the working condition of the equipment by using a clustering algorithm. And then, determining a health set according to the health state of the training set, taking all data of the training set and the test set as a sample set to be tested, calculating MMD distance value sequences of the sample set to be tested and the health set under different working conditions through a sliding window, constructing a health index by using the distance values, and finally merging the sub-working condition health indexes according to flight circulation to obtain a degradation curve after merging the working conditions. The flow frame is shown in fig. 2, and the specific method flow and steps are as follows:
step 1: parameter screening and data preprocessing
Step 1.1 parameter screening
The invention requires enough equipment performance monitoring data samples, and the data needs to have the following characteristics:
obtaining equipment parameters under various operating conditions;
continuously acquiring equipment parameters in each working cycle, wherein the parameters are uninterrupted time sequence data;
the collected parameters can represent the degradation trend of the equipment, namely the performance monitoring parameters of the equipment are non-constant value parameters.
Complicated equipment systems such as aerospace vehicles, large-scale electrical equipment and the like have complicated internal structures and numerous monitoring parameters, and in order to research the degradation condition of the equipment under multiple working conditions, the parameters need to be divided into working condition parameters and performance parameters. Because different performance monitoring parameters are different in the capability of representing the health state of the engine, parameter data which do not change along with time, do not change obviously along with time or have inconsistent fading trends under different working conditions can appear under different working conditions. Therefore, to ensure the accuracy of the health assessment, a decline-sensitive performance parameter screening is required. In the process of screening performance parameters, according to the parameter variation trend, selecting curves such as a primary function, a secondary function and the like to perform function fitting on the curves, and constructing three evaluation indexes for quantitatively evaluating the fitting condition:
(1) coefficient of the highest order term: and judging the degree of the parameter variation trend.
(2) Variance of absolute values of parameters and fit line: the discrete degree of the parameter distribution distance fitting line is represented, and the smaller the variance is, the more stable the parameter change is.
(3) Percentage of points within upper and lower thresholds of the fit line: the fluctuation of the parameter change is shown, and the higher the percentage is, the smaller the fluctuation of the parameter change is, and the folding property is better.
Setting specific index value requirements according to the data condition of the equipment, and screening out performance parameters with obvious decline trend and consistent change trend by comparing the evaluation indexes of the parameters.
Step 1.2 data normalization
The working condition parameters and the decay-sensitive performance parameters are standardized to be in the range of [0,1] by adopting a maximum-minimum normalization method, and dimensional influence is removed, so that the subsequent health assessment work is more reasonable.
Step 1.3 data denoising
Due to the fact that complex equipment systems such as aerospace vehicles and large electrical equipment are complex in operation environment, random noise interference exists in collected data, smooth noise reduction processing is conducted on the data through a local weighted regression method, and the overall quality of the data is improved.
Step 2: division of operating modes
And (3) assuming that the equipment has Q operating conditions, dividing the processed operating condition parameters into Q operating conditions by adopting a K-Means clustering method, and outputting operating condition labels corresponding to each working cycle.
And step 3: construction of health set and sample set to be tested
Suppose that the training set (historical data) is a P health-to-fault time series containing k decay-sensitive performance parameters, and each sample is recorded as
The test set (real-time data) is a time series of Q fail-before-fail containing k decay-sensitive performance parameters, each sample noted
Selecting the initial health state of the training set as the health set and recording the initial health state as the health set
Because the influence of the working conditions needs to be considered, the working condition labels obtained according to the clustering correspond to the working cycles, and the health set of the training sample is divided into the health sets under a plurality of working conditions. The sample set to be tested is sample data of all the training sets and the test set.
And 4, step 4: health index metric
Step 4.1, calculating the MMD distance between the sample set to be detected and the health set by sliding the window
As shown in the figure, a sliding window with a window length of L and a step length of 1 is used to segment a multi-dimensional sample set to be tested, and a time window of each sample set to be tested is marked as { Ω [ ]1,Ω2,...,Ωl(Ti)-L+1The data of the sample set to be tested corresponding to each time window is
Wherein l (Ti) is the time sequence length of the sample set to be tested. If the last duty cycle t of the time windowi+L-1And if the corresponding working condition is k, measuring the MMD distance between the data of all the multi-dimensional sample sets to be tested belonging to the working condition k in the time window and the health set under the working condition, wherein the distance value represents the health state of the last working cycle in the time window, and the smaller the distance between the data of all the multi-dimensional sample sets to be tested and the health set under the working condition k, the smaller the difference between the state of the equipment under the working cycle and the health state is. And starting from the first time window, sliding the time window to generate a one-dimensional distance sequence of corresponding cycles under different working conditions of the training set sample and the test set sample.
Step 4.2 normalizing MMD distance
As the health indexes under different working conditions need to be combined subsequently, the distance sequence needs to be normalized by working conditions, and the normalization standard under each working condition is the same. Under the k-th working condition, the maximum distance value in all the distance sequences is
The minimum distance value is
The standard distance value is calculated by the formula
And obtaining the standardized distance value sequences of the training set sample and the test set sample under different working conditions.
Step 4.3 construction of health indicators
The health index is measured by a CV value, the CV value of the equipment is gradually reduced along with the degradation process, and the later CV value is smaller, which represents that the degradation is more obvious. Since the distance value of the MMD algorithm metric represents the difference metric between the state of the device at the moment and the full health state, the CV value is expressed as
Wherein the content of the first and second substances,
representing normalized distance value, CV, according to the respective operating conditionsiAnd the health index of the corresponding moment under the working condition is shown. Finally, the CV value sets of all the training set samples and the test set samples under different working conditions can be obtained, as shown in fig. 3.
And 5: device health status characterization under multiple operating conditions
Step 5.1 merging conditions
Enabling CV indexes under different working conditions to correspond to working cycles corresponding to the sliding time windows, combining the CV indexes according to a time sequence to obtain a health index sequence combining a plurality of working conditions, and recording as CV ═ CV1,CV2,...,CVl(Ti)-L+1The sequence represents the health state change condition of the equipment under various working conditions.
Step 5.2 generating a degradation curve
And visualizing the health index sequence, drawing a CV curve graph by taking the flight cycle as an abscissa and the health index as an ordinate, smoothing the curve by using a local weighted regression method to ensure degradation monotonicity, finally obtaining a degradation curve graph of the equipment, and more intuitively observing the health state change trend of the equipment.
Example analysis
An engine Data set from Data Challenge in the international conference on PHM 2008 was selected for the example analysis, and this Data included test Data for 150 training engines and 260 test engines under 6 operating conditions.
Step 1: the original flight data is divided into 3 operating condition parameters and 21 gas path parameters.
As shown in fig. 4, 7 constant value parameters that do not change with time are excluded, and then decay sensitive gas circuit parameter screening based on function fitting is performed on the remaining 14 parameters, wherein the specific fitting scheme is as follows:
(1) the parameters with different change trends under different working conditions are not fitted, as shown in fig. 5, the parameters 9 of the No. 13 training engine are change trend graphs under 6 different working conditions, and the graphs show that the parameters have opposite change trends under different working conditions, and have no research significance for multiple working conditions facing equipment, so that the parameters are excluded.
(2) The parameters are in a linear function with time under different working conditions, and the linear function is used for fitting. As shown in fig. 6, the variation trend of parameter 2 of the No. 11 training engine under 6 different working conditions and the fitted straight line graph show that the variation trend of the parameter basically conforms to the situation of a linear function, so that the parameter is screened according to the situation of the linear function fitting.
(3) The parameters are in quadratic function with time under different working conditions, and quadratic curve fitting is adopted. Fig. 7 is a variation trend and fitting curve diagram of parameter 11 of No. 11 training engine under 6 different working conditions, and it can be known from fig. 7 that the variation trend of the parameter basically conforms to the situation of quadratic function, so that the parameter is screened according to the situation of quadratic function fitting.
Three evaluation indexes are constructed, and the specific index requirements are as follows:
(1) coefficient of the highest order term: linear slope k (y ═ kx + b) and conic a value (y ═ ax)2+ bx + c), selecting the parameter with larger variation trend as the evaluation parameter.
(2) Variance of absolute values of parameters and fit line: and characterizing the dispersion degree of the parameter distribution distance fitting line, and setting the parameter with the variance less than 0.045 as an evaluation parameter.
(3) Percentage of points within upper and lower thresholds of the fit line: setting 0.2 above and below the fitting straight line and 0.15 above and below the fitting curve as threshold values, calculating the parameter percentage in the threshold value range, and setting the parameter with the point number more than 75% in the range as an evaluation parameter.
By calculating evaluation indexes, 4 parameters which do not fit due to different change trends under different working conditions and 3 gas path parameters which do not meet the requirement of the indexes are eliminated, and finally 7 gas path parameters which are sensitive to fading are screened out. Further normalizing the working condition parameters and the screened gas path parameters, and performing smooth noise reduction processing on the gas path parameters, wherein fig. 8 is a data preprocessing process of the No. 2 engine parameters 4.
Step 2: and dividing the working condition parameters into 6 categories by adopting a K-Means clustering method to obtain a working condition label corresponding to each flight cycle of the engine.
And step 3: and selecting 130 circular data samples after the training of the engine is removed as a health set according to a piecewise linear RUL target function and by combining the degradation condition of the engine, and dividing the health set into 6 health sets under different working conditions according to working condition labels obtained by clustering. All training engine and test engine operating data are taken as a sample set to be tested.
And 4, step 4: and setting the length L of the sliding window as 20, wherein the step length is 1, measuring the MMD distance between the sample set to be measured belonging to the cycle working condition at the end of the time window and the health set under the working condition, carrying out sub-working condition normalization on the distance result, and calculating the health index corresponding to the distance value according to a formula. Taking test engine number 117 as an example, CV values of corresponding cycles under various working conditions are generated, and fig. 9 is a CV value set of test engine number 32 under different working conditions.
And 5: the health indexes under the 6 working conditions are combined according to flight cycles to obtain a CV value sequence representing the health state of the engine, and the visualization is shown in FIG. 10. As can be seen from fig. 10, since the training engine is a life-cycle example from health to failure, the health status of the No. 220 training engine is continuously degraded with the increase of the running time, and the health index is obviously reduced; the decline trend of test engine No. 32 was not obvious and the health index generally declined due to the incomplete case where the test engine ended before failure. The analysis result shows that the health state of the engine under various working conditions can be well evaluated.
According to the scheme provided by the embodiment of the invention, the health state of the engine under various working conditions can be well evaluated.
Although the present invention has been described in detail hereinabove, the present invention is not limited thereto, and various modifications can be made by those skilled in the art in light of the principle of the present invention. Thus, modifications made in accordance with the principles of the present invention should be understood to fall within the scope of the present invention.
Claims (10)
1. A health assessment method for equipment facing multi-working-condition operation conditions is characterized by comprising the following steps:
dividing the obtained original equipment parameters into N working condition parameters and M individual performance parameters, and performing decay sensitivity screening processing on the M individual performance parameters under different working conditions to obtain decay sensitivity performance parameters;
obtaining a working condition label corresponding to each working cycle by dividing the N working condition parameters into N working conditions;
acquiring a training set and a test set of equipment to be tested, taking the training set and the test set of the equipment to be tested as a sample set to be tested, simultaneously determining a health set from the training set, and dividing the health set into n sub-health sets under different working conditions according to the working condition labels;
calculating MMD distance values of the sample set to be tested and the sub-health set under n different working conditions through a sliding window, and constructing a health evaluation index for representing the equipment performance degradation state by utilizing the MMD distance values under the n different working conditions;
merging the health assessment indexes used for representing the equipment performance decline state to obtain a degradation curve graph after the working conditions are merged;
wherein, M, N and N are positive integers.
2. The method according to claim 1, wherein the obtaining of degradation-sensitive performance parameters by performing degradation-sensitive screening processing on the M performance parameters under different working conditions comprises:
according to the change trend of the performance parameters, selecting a primary function and a secondary function to perform function fitting on the performance parameters to obtain the fitted performance parameters;
and establishing evaluation indexes according to the fitted performance parameters, and screening out the performance parameters with obvious decline trends and consistent change trends under different working conditions by comparing the evaluation indexes of the performance parameters.
3. The method according to claim 2, wherein the evaluation index includes:
coefficient of the highest order term: straight barThe line slope k (y ═ kx + b) and the value of the quadratic curve a (y ═ ax)2+ bx + c), selecting the parameter with larger variation trend as the evaluation parameter;
variance of absolute values of parameters and fit line: representing the dispersion degree of a parameter distribution distance fitting line, and setting a parameter with the variance less than 0.045 as an evaluation parameter;
percentage of points within upper and lower thresholds of the fit line: setting 0.2 above and below the fitting straight line and 0.15 above and below the fitting curve as threshold values, calculating the parameter percentage in the threshold value range, and setting the parameter with the point number accounting for more than 75% in the range as an evaluation parameter.
4. The method of claim 1, further comprising, after obtaining the fade-sensitive performance parameter:
and carrying out normalization and smooth noise reduction processing on the decay sensitivity performance parameters, and simultaneously carrying out normalization processing on the working condition parameters so as to improve the overall quality of the parameters.
5. The method of claim 4, wherein the dividing the N operating condition parameters into N operating conditions comprises:
and dividing the working condition parameters after the normalization processing into n working conditions by adopting a K-Means clustering method according to the type number n of the operating working conditions of the equipment.
6. The method of claim 5, wherein the training set is a time sequence of P health-to-failure events including decay sensitivity performance parameters; the test set of the equipment to be tested refers to a time sequence comprising decay sensitivity performance parameters, which is formed by Q devices and ends before the failure.
7. The method of claim 6, wherein the determining the health set from the training set comprises: and selecting the initial health state of the training set as a health set.
8. The method according to claim 1, further comprising, after calculating the MMD distance values of the sample set to be tested and the sub-health set under n different operating conditions through a sliding window:
and carrying out sub-working condition normalization processing on the MMD distance value.
9. The method of claim 1, wherein the obtaining a degradation curve graph after combined conditions by combining the health assessment indicators for characterizing equipment performance degradation states comprises:
corresponding the health evaluation indexes under different working conditions to the working cycles corresponding to the sliding time windows, and combining the health evaluation indexes according to the time sequence to obtain a health index sequence combining a plurality of working conditions;
and performing visualization processing on the health index sequence combined with the plurality of working conditions to obtain a degradation curve graph after the working conditions are combined.
10. The method of claim 9, further comprising:
and smoothing the degradation curve graph after the working conditions are combined by using a local weighted regression method so as to more intuitively observe the health state change trend of the equipment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111097810.XA CN113837591A (en) | 2021-09-18 | 2021-09-18 | Equipment health assessment method oriented to multi-working-condition operation conditions |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111097810.XA CN113837591A (en) | 2021-09-18 | 2021-09-18 | Equipment health assessment method oriented to multi-working-condition operation conditions |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113837591A true CN113837591A (en) | 2021-12-24 |
Family
ID=78959866
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111097810.XA Pending CN113837591A (en) | 2021-09-18 | 2021-09-18 | Equipment health assessment method oriented to multi-working-condition operation conditions |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113837591A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114298616A (en) * | 2022-03-11 | 2022-04-08 | 西南石油大学 | Equipment health state evaluation method and device and computer equipment |
| CN114722641A (en) * | 2022-06-09 | 2022-07-08 | 卡松科技股份有限公司 | Lubricating oil state information integrated evaluation method and system for detection laboratory |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106094570A (en) * | 2016-07-13 | 2016-11-09 | 北京航空航天大学 | A kind of aero-engine complete machine health evaluating method under variable working condition based on operating mode's switch and this distance of paddy |
| CN106289777A (en) * | 2016-08-01 | 2017-01-04 | 北京航空航天大学 | A kind of multi-state rolling bearing performance appraisal procedure based on geometry tolerance |
| CN111985546A (en) * | 2020-08-10 | 2020-11-24 | 西北工业大学 | Aircraft engine multi-working-condition detection method based on single-classification extreme learning machine algorithm |
| CN112051468A (en) * | 2020-09-08 | 2020-12-08 | 南京航空航天大学 | Method for evaluating health state of aviation static converter under complex working conditions |
| CN112668105A (en) * | 2021-01-14 | 2021-04-16 | 北京航空航天大学 | Helicopter transmission shaft abnormity judgment method based on SAE and Mahalanobis distance |
-
2021
- 2021-09-18 CN CN202111097810.XA patent/CN113837591A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106094570A (en) * | 2016-07-13 | 2016-11-09 | 北京航空航天大学 | A kind of aero-engine complete machine health evaluating method under variable working condition based on operating mode's switch and this distance of paddy |
| CN106289777A (en) * | 2016-08-01 | 2017-01-04 | 北京航空航天大学 | A kind of multi-state rolling bearing performance appraisal procedure based on geometry tolerance |
| CN111985546A (en) * | 2020-08-10 | 2020-11-24 | 西北工业大学 | Aircraft engine multi-working-condition detection method based on single-classification extreme learning machine algorithm |
| CN112051468A (en) * | 2020-09-08 | 2020-12-08 | 南京航空航天大学 | Method for evaluating health state of aviation static converter under complex working conditions |
| CN112668105A (en) * | 2021-01-14 | 2021-04-16 | 北京航空航天大学 | Helicopter transmission shaft abnormity judgment method based on SAE and Mahalanobis distance |
Non-Patent Citations (4)
| Title |
|---|
| 宋登巍;吕琛;齐乐;王景霖;吴英建;: "基于健康基线和马氏距离的液压系统变工况健康评估", 系统仿真技术, no. 03, pages 20 - 27 * |
| 张妍;韩光威;陆宁云;姜斌;支有冉;: "基于JS散度的轨道车辆门系统健康状态评估方法", 机械设计与制造工程, no. 11, pages 126 - 131 * |
| 杨绪升: "燃气轮机机群气路故障诊断研究", 中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑, pages 44 - 47 * |
| 郑韩飞: "基于多工况性能监测数据的剩余使用寿命预测方法研究", 中国优秀硕士学位论文全文数据库 基础科学辑, pages 12 - 36 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114298616A (en) * | 2022-03-11 | 2022-04-08 | 西南石油大学 | Equipment health state evaluation method and device and computer equipment |
| CN114298616B (en) * | 2022-03-11 | 2022-05-06 | 西南石油大学 | Equipment health state evaluation method and device and computer equipment |
| CN114722641A (en) * | 2022-06-09 | 2022-07-08 | 卡松科技股份有限公司 | Lubricating oil state information integrated evaluation method and system for detection laboratory |
| CN114722641B (en) * | 2022-06-09 | 2022-09-30 | 卡松科技股份有限公司 | Lubricating oil state information integrated evaluation method and system for detection laboratory |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6192317B1 (en) | Statistical pattern analysis methods of partial discharge measurements in high voltage insulation | |
| CN109186813B (en) | Temperature sensor self-checking device and method | |
| CN109816031B (en) | Transformer state evaluation clustering analysis method based on data imbalance measurement | |
| CN111459700B (en) | Equipment fault diagnosis method, diagnosis device, diagnosis equipment and storage medium | |
| CN116879662B (en) | Transformer fault detection method based on data analysis | |
| CN113837591A (en) | Equipment health assessment method oriented to multi-working-condition operation conditions | |
| CN113344134B (en) | Low-voltage distribution monitoring terminal data acquisition abnormality detection method and system | |
| CN115982602B (en) | Photovoltaic transformer electrical fault detection method | |
| DE102012103652A1 (en) | Method, computer program and system for performing interpolation on sensor data for high system availability | |
| CN110333962B (en) | Electronic component fault diagnosis model based on data analysis and prediction | |
| CN111879349A (en) | Sensor data deviation self-adaptive correction method | |
| CN115828170B (en) | Fault detection method based on electronic control data of tunneling machine | |
| CN111678992A (en) | Nondestructive testing method for identifying damage type of concrete structure | |
| CN111967717A (en) | Data quality evaluation method based on information entropy | |
| CN116295948A (en) | Abnormality detection method, system and storage medium of industrial temperature sensor in large temperature difference environment | |
| CN111382029B (en) | Mainboard abnormity diagnosis method and device based on PCA and multidimensional monitoring data | |
| CN111507374A (en) | Power grid mass data anomaly detection method based on random matrix theory | |
| CN110967184B (en) | Gearbox fault detection method and system based on vibration signal distribution characteristic recognition | |
| CN113654651A (en) | Strong robust signal early degradation feature extraction and equipment running state monitoring method | |
| CN112445632A (en) | HPC reliability evaluation method based on fault data modeling | |
| CN112699609A (en) | Diesel engine reliability model construction method based on vibration data | |
| CN117353462B (en) | Power grid operation monitoring analysis method and platform based on artificial intelligence | |
| CN116379043B (en) | Fault detection method and system for spin-on oil filter | |
| CN114861536A (en) | Multi-head CNN-RNN-based track circuit fault detection method | |
| CN117251041A (en) | Method, system, device and medium for screening potential power failure of server |
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 |