CN115203905A - Equipment health assessment method integrating expert experience and intelligent algorithm - Google Patents
Equipment health assessment method integrating expert experience and intelligent algorithm Download PDFInfo
- Publication number
- CN115203905A CN115203905A CN202210724695.2A CN202210724695A CN115203905A CN 115203905 A CN115203905 A CN 115203905A CN 202210724695 A CN202210724695 A CN 202210724695A CN 115203905 A CN115203905 A CN 115203905A
- Authority
- CN
- China
- Prior art keywords
- state
- equipment
- health
- current
- evaluation model
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention relates to the technical field of equipment monitoring and maintenance, in particular to an equipment health assessment method integrating expert experience and an intelligent algorithm, which comprises the following steps: s1, collecting signal data of equipment, synchronously evaluating the health state of the equipment by an expert, and labeling corresponding label data; s2, preprocessing the acquired signal data and then performing calculation analysis to obtain state parameters of the equipment; s3, processing and clustering the state parameters of the equipment to obtain clustering centers of all states of the equipment, and establishing a state matrix of the equipment according to the clustering centers of all states of the equipment; and S4, after processing the clustering center, the state parameters and the label data corresponding to a certain state of the equipment, training a preset basic model to obtain an initial evaluation model corresponding to the state of the equipment. The method can reduce the data volume collected and can accurately evaluate the health state of single equipment.
Description
Technical Field
The invention relates to the technical field of equipment monitoring and maintenance, in particular to an equipment health assessment method integrating expert experience and an intelligent algorithm.
Background
The health status of industrial equipment, the quality level of industrial products produced in the event and the stability and fluency of the production process. In order to ensure the quality of industrial production, the state of the equipment needs to be monitored and evaluated.
At present, most of equipment health assessment methods are obtained based on data-driven models, namely, after mass data are collected, the data are organized to form information, relevant information is integrated and refined, and an automated industrial equipment health state assessment model is formed through training and fitting on the basis of the data. Belongs to a model establishing method for decision and action based on data as center. And after the evaluation model is obtained, analyzing and evaluating the state of the equipment according to the evaluation model and the acquired equipment state data. Although the method becomes the creation idea of the machine learning algorithm model which is mainstream at present and has been developed more maturely, the health status assessment of the industrial equipment by using the method has the following problems:
1. the method needs to acquire massive data when the evaluation model is established, but on the site of industrial equipment, the problems of high cost, unstable communication, limited storage space and the like exist in massive data acquisition, and the marking after data acquisition also costs very high labor cost;
2. the industrial field environment is complex, the same equipment has different characteristics under different working conditions, and the influence of industrial and mining cannot be eliminated by data preprocessing and characteristic engineering, so that the evaluation model obtained by the method has poor generalization capability and cannot be suitable for different individuals, and the health state of the single equipment cannot be accurately evaluated.
Therefore, how to reduce the amount of data collected and enhance the adaptability of the single device becomes a problem to be solved.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides the equipment health assessment method integrating the expert experience and the intelligent algorithm, which can reduce the acquired data volume and can accurately assess the health state of single equipment.
In order to solve the technical problems, the invention adopts the following technical scheme:
an equipment health assessment method integrating expert experience and an intelligent algorithm comprises the following steps:
s1, collecting signal data of equipment, synchronously evaluating the health state of the equipment by an expert, and labeling corresponding label data;
s2, preprocessing the acquired signal data and then performing calculation analysis to obtain state parameters of the equipment;
s3, processing and clustering the state parameters of the equipment to obtain clustering centers of all states of the equipment, and establishing a state matrix of the equipment according to the clustering centers of all states of the equipment;
s4, after processing a clustering center, state parameters and label data corresponding to a certain state of the equipment, training a preset basic model to obtain an initial evaluation model corresponding to the state of the equipment; repeating the process until initial evaluation models of all states of the equipment are obtained respectively;
s5, acquiring and processing signal data of the equipment in actual operation to obtain current state parameters of the equipment; and after the current state of the equipment is obtained according to the current state parameters and the state matrix, calling an evaluation model corresponding to the current state, and evaluating the current health state of the equipment by combining the current state parameters of the equipment.
Preferably, in S2, the preprocessing includes outlier elimination, missing value filling, and time series signal filtering.
Preferably, in S2, the state parameters include a time domain state parameter, a frequency domain state parameter, and a time-frequency domain state parameter.
Preferably, S3 comprises:
s31, forming the state parameters at each moment into a vector form according to a time sequence, and recording the vector form as the state vector at each moment;
s32, carrying out unsupervised learning on the state vector at each moment, and clustering by adopting a K-Means algorithm to obtain each running state when the equipment runs;
and S33, acquiring the clustering centers of all the running states in the clustering result to form a state matrix of the equipment.
Preferably, S4 comprises:
s41, calculating residual error values of the state vector of the signal data at each moment in a certain state of the equipment and the clustering center of the state to obtain a residual error sequence of the state;
s42, calculating the standard deviation of the residual sequence of the state according to a preset sliding window;
s43, training a preset basic model through the residual value and the standard deviation of the state and corresponding label data to obtain an initial evaluation model of the state;
and S44, repeating S1-S43 until initial evaluation models of all states of the equipment are obtained respectively.
Preferably, S5 comprises:
s51, processing actually acquired signal data to obtain a current state vector, and determining the current state of the equipment according to the current state vector and the state matrix;
s51, calculating a residual error value of the current state vector according to the current state vector and the corresponding clustering center, and obtaining a current residual error sequence;
s52, calculating the standard deviation of the current residual sequence according to a preset sliding window;
and S53, calling an evaluation model corresponding to the current state, and evaluating the current health state of the equipment according to the current residual value and standard deviation.
Preferably, after S5, S6 is further included, the evaluation result and the actual state of the apparatus are analyzed, and if the evaluation result of the evaluation model in a certain state is that there is an abnormality or the apparatus actually has an abnormality in the state, the expert gives corresponding label data and then uses the label data as new training data, and the evaluation model corresponding to the state is retrained again to obtain a training updated evaluation model of the state.
Preferably, S6 comprises:
s61, analyzing the evaluation result and the actual state of the device, and if the evaluation result of the evaluation model in a certain state is abnormal or the equipment is actually abnormal in the state, carrying out real-time evaluation by an expert and providing corresponding label data;
s62, processing signal data corresponding to real-time evaluation of the expert in the S61 to obtain corresponding residual errors and standard deviations;
and S63, adding the label data, the corresponding residual error and the standard deviation which are given by the expert in real time into the historical training data of the evaluation model in the corresponding state, and training and updating the evaluation model to obtain the updated evaluation model after training in the state.
Preferably, in S1, the signal data acquisition device includes one or more of a vibration sensor, a temperature sensor and a noise sensor.
Compared with the prior art, the invention has the following beneficial effects:
1. by using the method, each running state of a single specific device under specific working conditions can be modeled respectively, and the modeling mode is similar to a customized modeling mode, so that the accuracy is ensured, the data quantity required to participate in training of a single evaluation model is small, the data acquisition and processing are convenient, and the model training efficiency is high. Besides the signal data of the equipment, the label data evaluated by experts is used as the training data of the evaluation model, so that the effectiveness of the evaluation model is ensured. Compared with the prior art that the use environment of the device is not considered and a unified evaluation model is established for all the devices, the method can reduce the collected data volume and can accurately evaluate the health state of single equipment.
2. In the method, after the state matrix is obtained through the analysis and the processing of the state parameters, the evaluation models in different states are constructed, so that the state parameters can be fully utilized, and the accuracy and the comprehensiveness of the evaluation on the health degree of the equipment can be ensured. When the health state of the equipment is subsequently evaluated, the corresponding evaluation model is called to evaluate according to the current state of the equipment, so that the pertinence is provided for the equipment, the pertinence is provided for the state of the equipment, and the problem of inaccurate evaluation caused by different equipment states (such as a full-load state and an idle state) can be avoided.
3. In the method, the training data of the evaluation model of each state are from the equipment to be detected, so that the influence caused by equipment industrial and mining difference and data distribution difference can be effectively avoided.
4. In the using process of each evaluation model, if an evaluation result corresponding to a certain state is abnormal or the device actually has abnormal in the state, an expert evaluates and approves the state to obtain corresponding label data, adds the label data, corresponding residual errors and standard differences into historical training data of the evaluation model corresponding to the state, and trains the evaluation model again, so that updating iteration of the evaluation model is realized, and the latest data can be effectively utilized.
Drawings
For purposes of promoting a better understanding of the objects, aspects and advantages of the invention, reference will now be made in detail to the present invention as illustrated in the accompanying drawings, in which:
fig. 1 is a flowchart in the embodiment.
Detailed Description
The following is further detailed by the specific embodiments:
example (b):
the embodiment of the invention discloses
As shown in fig. 1, a method for evaluating the health of a device by fusing expert experience and an intelligent algorithm includes the following steps:
s1, collecting signal data of equipment, synchronously evaluating the health state of the equipment by an expert, and labeling corresponding label data. In particular implementations, the signal data acquisition device includes one or more of a vibration sensor, a temperature sensor, and a noise sensor. Of course, those skilled in the art may also set up a specific collection device according to the type of the monitored equipment and the operating condition.
And S2, preprocessing the acquired signal data and then performing calculation analysis to obtain the state parameters of the equipment. Wherein the preprocessing comprises the elimination of abnormal values, the filling of missing values and the filtering processing of time sequence signals. The state parameters comprise time domain state parameters, frequency domain state parameters and time-frequency domain state parameters.
And S3, processing and clustering the state parameters of the equipment to obtain the clustering centers of all the states of the equipment, and establishing a state matrix of the equipment according to the clustering centers of all the states of the equipment. Specifically, S3 includes:
s31, forming the state parameters at each moment into a vector form according to a time sequence, and recording the vector form as the state vector at each moment;
s32, carrying out unsupervised learning on the state vector at each moment, and clustering by adopting a K-Means algorithm to obtain each running state when the equipment runs;
and S33, acquiring the clustering centers of all the running states in the clustering result to form a state matrix of the equipment.
S4, after processing a clustering center, state parameters and label data corresponding to a certain state of the equipment, training a preset basic model to obtain an initial evaluation model corresponding to the state of the equipment; and the process is repeated until an initial evaluation model of all states of the device is obtained, respectively. Specifically, S4 includes:
s41, calculating residual error values of the state vector of the signal data at each moment in a certain state of the equipment and the clustering center of the state to obtain a residual error sequence of the state;
s42, calculating the standard deviation of the residual error sequence of the state according to a preset sliding window;
s43, training a preset basic model through the residual value and the standard deviation of the state and corresponding label data to obtain an initial evaluation model of the state; when the basic model is selected, the existing conventional machine learning model is directly used, which is not the innovation point of the invention of the application, and a person skilled in the art can select the machine learning model familiar with himself, and details are not described herein.
And S44, repeating S1-S43 until initial evaluation models of all states of the equipment are obtained respectively.
S5, acquiring and processing signal data of the equipment in actual operation to obtain current state parameters of the equipment; and after the current state of the equipment is obtained according to the current state parameters and the state matrix, calling an evaluation model corresponding to the current state, and evaluating the current health state of the equipment by combining the current state parameters of the equipment. Specifically, S5 includes:
s51, processing actually acquired signal data to obtain a current state vector, and determining the current state of the equipment according to the current state vector and the state matrix;
s51, calculating a residual error value of the current state vector according to the current state vector and the corresponding clustering center, and obtaining a current residual error sequence;
s52, calculating the standard deviation of the current residual sequence according to a preset sliding window;
and S53, calling an evaluation model corresponding to the current state, and evaluating the current health state of the equipment according to the current residual value and standard deviation.
And S6, analyzing the evaluation result and the actual state of the device, if the evaluation result of the evaluation model in a certain state is abnormal or the equipment is actually abnormal in the state, giving out corresponding label data by an expert to serve as new training data, and performing retraining on the evaluation model corresponding to the state to obtain the evaluation model updated by the training of the state. Specifically, S6 includes:
s61, analyzing the evaluation result and the actual state of the device, and if the evaluation result of the evaluation model in a certain state is abnormal or the equipment is actually abnormal in the state, carrying out real-time evaluation by an expert and providing corresponding label data;
s62, processing signal data corresponding to real-time evaluation of the expert in the S61 to obtain corresponding residual errors and standard deviations;
and S63, adding the label data, the corresponding residual error and the standard deviation which are given by the expert in real time into the historical training data of the evaluation model in the corresponding state, and training and updating the evaluation model to obtain the updated evaluation model after training in the state.
The method can respectively model each running state of a single concrete device under a concrete working condition, and the modeling mode is similar to a customized modeling mode, so that the accuracy is ensured, the data quantity of a single evaluation model needing to participate in training is small, the data acquisition and processing are convenient, and the model training efficiency is high. Besides the signal data of the equipment, the label data evaluated by experts is used as the training data of the evaluation model, so that the effectiveness of the evaluation model is ensured. Compared with the prior art that the use environment of the device is not considered and a uniform evaluation model is established for all the devices, the method can reduce the acquired data volume and can accurately evaluate the health state of single equipment. In addition, the method obtains the state matrix through the analysis and the processing of the state parameters, and then constructs the evaluation models in different states, so that the state parameters can be fully utilized, and the accuracy and the comprehensiveness of the evaluation on the health degree of the equipment can be ensured. When the health state of the equipment is evaluated subsequently, the corresponding evaluation model is called according to the current state of the equipment for evaluation, so that the pertinence to the equipment is achieved, the state of the equipment is pointed, and the problem of inaccurate evaluation caused by different equipment states (such as a full load state and an empty load state) can be avoided.
In addition, in the using process of each evaluation model, if the evaluation result corresponding to a certain state is abnormal or the device actually has abnormality in the state, the expert evaluates and approves the state to obtain corresponding label data, adds the label data, the corresponding residual error and the standard deviation into historical training data of the evaluation model in the corresponding state, and trains the evaluation model again, so that the updating iteration of the evaluation model is realized, and the latest data can be effectively utilized. In addition, the training data of the evaluation model of each state are from the equipment to be detected, so that the influence caused by equipment industrial and mining difference and data distribution difference can be effectively avoided.
Compared with the prior art, the method can effectively reduce the acquired data volume and can accurately evaluate the health state of single equipment.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present invention and not for limiting the technical solutions, and those skilled in the art should understand that modifications or equivalent substitutions can be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions, and all that should be covered by the claims of the present invention.
Claims (9)
1. An equipment health assessment method integrating expert experience and an intelligent algorithm is characterized by comprising the following steps:
s1, collecting signal data of equipment, synchronously evaluating the health state of the equipment by an expert, and labeling corresponding label data;
s2, preprocessing the acquired signal data and then performing calculation analysis to obtain state parameters of the equipment;
s3, processing and clustering the state parameters of the equipment to obtain clustering centers of all states of the equipment, and establishing a state matrix of the equipment according to the clustering centers of all states of the equipment;
s4, after processing a clustering center, state parameters and label data corresponding to a certain state of the equipment, training a preset basic model to obtain an initial evaluation model corresponding to the state of the equipment; repeating the process until initial evaluation models of all states of the equipment are obtained respectively;
s5, acquiring and processing signal data of the equipment in actual operation to obtain current state parameters of the equipment; and after the current state of the equipment is obtained according to the current state parameters and the state matrix, calling an evaluation model corresponding to the current state, and evaluating the current health state of the equipment by combining the current state parameters of the equipment.
2. The method for evaluating the health of a device incorporating expert experience and an intelligent algorithm as set forth in claim 1, wherein: in S2, the preprocessing comprises elimination of abnormal values, filling of missing values and filtering of time sequence signals.
3. The method for evaluating the health of a device incorporating expert experience and an intelligent algorithm as set forth in claim 2, wherein: in S2, the state parameters include a time domain state parameter, a frequency domain state parameter, and a time-frequency domain state parameter.
4. The method for evaluating the health of equipment by combining expert experience and an intelligent algorithm according to claim 3, wherein the method comprises the following steps: s3 comprises the following steps:
s31, forming the state parameters at each moment into a vector form according to a time sequence, and recording the vector form as the state vector at each moment;
s32, carrying out unsupervised learning on the state vector at each moment, and clustering by adopting a K-Means algorithm to obtain each running state when the equipment runs;
and S33, acquiring the clustering centers of the running states in the clustering result to form a state matrix of the equipment.
5. The method for evaluating the health of a device incorporating expert experience and intelligent algorithms as claimed in claim 4, wherein: s4 comprises the following steps:
s41, calculating residual error values of a state vector of signal data at each moment in a certain state of the equipment and a clustering center of the state to obtain a residual error sequence of the state;
s42, calculating the standard deviation of the residual sequence of the state according to a preset sliding window;
s43, training a preset basic model through the residual value and the standard deviation of the state and corresponding label data to obtain an initial evaluation model of the state;
and S44, repeating S1-S43 until initial evaluation models of all states of the equipment are obtained respectively.
6. The method for evaluating the health of a device incorporating expert experience and intelligent algorithms as claimed in claim 5, wherein: s5, the method comprises the following steps:
s51, processing actually acquired signal data to obtain a current state vector, and determining the current state of the equipment according to the current state vector and the state matrix;
s51, calculating a residual error value of the current state vector according to the current state vector and the corresponding clustering center, and obtaining a current residual error sequence;
s52, calculating the standard deviation of the current residual sequence according to a preset sliding window;
and S53, calling an evaluation model corresponding to the current state, and evaluating the current health state of the equipment according to the current residual value and standard deviation.
7. The method for evaluating the health of equipment by combining expert experience and an intelligent algorithm according to claim 6, wherein the method comprises the following steps: and S6, analyzing the evaluation result and the actual state of the device, giving out corresponding label data by an expert as new training data if the evaluation result of the evaluation model in a certain state is abnormal or the equipment is actually abnormal in the state, and re-training the evaluation model corresponding to the state to obtain a training updated evaluation model of the state.
8. The method for evaluating the health of equipment by combining expert experience and an intelligent algorithm according to claim 7, wherein the method comprises the following steps: s6 comprises the following steps:
s61, analyzing the evaluation result and the actual state of the device, and if the evaluation result of the evaluation model in a certain state is abnormal or the equipment is actually abnormal in the state, carrying out real-time evaluation by an expert and providing corresponding label data;
s62, processing signal data corresponding to real-time evaluation of the expert in the S61 to obtain corresponding residual errors and standard deviations;
and S63, adding the label data, the corresponding residual error and the standard deviation which are given by the expert in real time into the historical training data of the evaluation model in the corresponding state, and training and updating the evaluation model to obtain the updated evaluation model after training in the state.
9. The method for evaluating the health of a device incorporating expert experience and an intelligent algorithm as set forth in claim 1, wherein: in S1, the signal data acquisition device comprises one or more of a vibration sensor, a temperature sensor and a noise sensor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210724695.2A CN115203905A (en) | 2022-06-24 | 2022-06-24 | Equipment health assessment method integrating expert experience and intelligent algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210724695.2A CN115203905A (en) | 2022-06-24 | 2022-06-24 | Equipment health assessment method integrating expert experience and intelligent algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115203905A true CN115203905A (en) | 2022-10-18 |
Family
ID=83578484
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210724695.2A Pending CN115203905A (en) | 2022-06-24 | 2022-06-24 | Equipment health assessment method integrating expert experience and intelligent algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115203905A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115758277A (en) * | 2022-11-30 | 2023-03-07 | 重庆忽米网络科技有限公司 | Online health state evaluation method for rotary equipment |
-
2022
- 2022-06-24 CN CN202210724695.2A patent/CN115203905A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115758277A (en) * | 2022-11-30 | 2023-03-07 | 重庆忽米网络科技有限公司 | Online health state evaluation method for rotary equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108320043B (en) | Power distribution network equipment state diagnosis and prediction method based on electric power big data | |
| CN111368403B (en) | Self-adaptive non-linear degradation residual life prediction method | |
| JP2019527413A (en) | Computer system and method for performing root cause analysis to build a predictive model of rare event occurrences in plant-wide operations | |
| CN111310981B (en) | Reservoir water level trend prediction method based on time series | |
| CN116129366A (en) | Digital twinning-based park monitoring method and related device | |
| CN107423414A (en) | A kind of process industry complex electromechanical systems fault source tracing method based on information transmission model | |
| CN101436057A (en) | Numerical control machining tool heat error Bayes network compensation method | |
| CN113610381B (en) | Water quality remote real-time monitoring system based on 5G network | |
| CN110727665A (en) | Internet of things equipment reported data quality analysis method and system | |
| CN114444231B (en) | Online self-adaptive prediction method, device, equipment and medium for residual life of mold | |
| CN115438726A (en) | Device life and fault type prediction method and system based on digital twin technology | |
| CN117390591B (en) | Operation monitoring method and system for coal conveying belt sampling machine based on electric parameter analysis | |
| CN115203905A (en) | Equipment health assessment method integrating expert experience and intelligent algorithm | |
| CN115617606A (en) | Equipment monitoring method and system, electronic equipment and storage medium | |
| CN117473253A (en) | Power acquisition data quality assessment method | |
| CN117038079A (en) | Child heart disease risk assessment and early warning system based on neural network | |
| CN116861503A (en) | Method for constructing digital twin model of power transformer based on big data | |
| CN116842684A (en) | Electric energy meter, evaluation method and system of operation reliability of electric energy meter and electric energy meter processor | |
| CN116205044A (en) | Real-time health analysis method and system for complex aerospace equipment | |
| CN113868119A (en) | Embedded software testing system and method based on Internet of things | |
| Prabowo et al. | Improving Internet of Things Platform with Anomaly Detection for Environmental Sensor Data | |
| CN117560300B (en) | Intelligent internet of things flow prediction and optimization system | |
| CN117870034B (en) | Control method, device and system for environmental parameters of clean room | |
| CN116957367B (en) | Parameter multi-time scale prediction method and system for comprehensive energy system operation strategy | |
| CN115859201B (en) | Chemical process fault diagnosis method and system |
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 |