CN116085245A - Online compressor performance prediction method and system based on OS-ELM - Google Patents
Online compressor performance prediction method and system based on OS-ELM Download PDFInfo
- Publication number
- CN116085245A CN116085245A CN202310084524.2A CN202310084524A CN116085245A CN 116085245 A CN116085245 A CN 116085245A CN 202310084524 A CN202310084524 A CN 202310084524A CN 116085245 A CN116085245 A CN 116085245A
- Authority
- CN
- China
- Prior art keywords
- sample set
- elm
- compressor
- online
- data
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
- F04B—POSITIVE-DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS
- F04B51/00—Testing machines, pumps, or pumping installations
-
- 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
Abstract
The invention discloses an on-line prediction method and system for compressor performance based on OS-ELM, comprising the following steps: s1, collecting compressor operation performance data as a training sample set; s2, preprocessing a training sample set; s3, constructing a compressor performance prediction model based on the OS-ELM by using the preprocessed training sample set; s4, dividing data in the online sliding window into a new training sample set and a test sample set by setting the sliding window online, verifying the matching of the compressor performance prediction model based on the OS-ELM and the working condition by the test sample set, and if the prediction error of the compressor performance prediction model based on the OS-ELM is smaller than a set value, putting the model into use, otherwise, entering the step S5; and S5, inputting the new training sample set in the online sliding window into the training sample set in the step S2, and executing the steps S3-S4. The invention effectively solves the problem of model-working condition mismatch caused by variable operation working conditions of the compressor, and realizes the online real-time updating application of the model.
Description
Technical Field
The invention belongs to the technical field of compressor performance prediction model modeling, and particularly relates to an on-line compressor performance prediction method and system based on an OS-ELM.
Background
Compressors have been widely used in industrial fields because of their high operating efficiency, large delivery flow rate, and the like. The main way to obtain the performance curve of the compressor is to test the measurement or perform performance conversion based on known data. The former directly measures the performance of the compressor, so that the obtained performance curve is more real and reliable, but the test cost is more expensive and the time is longer. The latter is based on historical operating data of the compressor, whose performance is scaled under the assumption of certain similarity, so-called data-driven modeling. Although this approach can reduce test time and cost, it is highly dependent on the accuracy of the similarity assumptions. In general, as long as there is enough reliable operation data information, the performance prediction accuracy of the model will be higher, but if the compressor needs to operate in a stable working condition range, once the operation environment or working condition changes, the performance prediction accuracy of the model will also be drastically reduced, and the problem of model-working condition mismatch caused by frequent working condition changes cannot be effectively solved. Therefore, how to quickly establish and update the performance prediction model of the compressor on line, and effectively solve the problem of on-line model update become a hot spot of current research.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide an on-line prediction method and system for the performance of a compressor based on an OS-ELM, which solve the problems that the prediction precision of a performance prediction model of the compressor in the prior art is reduced and the model-working condition mismatch caused by frequent working condition change cannot be effectively solved.
In order to achieve the above purpose, the invention is realized by adopting the following technical scheme: an on-line prediction method for compressor performance based on OS-ELM comprises the following steps:
s1, collecting compressor operation performance data as a training sample set;
s2, preprocessing a training sample set;
s3, constructing a compressor performance prediction model based on the OS-ELM by using the preprocessed training sample set;
s4, dividing data in the online sliding window into a new training sample set and a test sample set by setting the sliding window online, verifying the matching of the compressor performance prediction model based on the OS-ELM and the working condition by the test sample set, and if the prediction error of the compressor performance prediction model based on the OS-ELM is smaller than a set value, putting the model into use, otherwise, entering the step S5;
and S5, inputting the new training sample set in the online sliding window into the training sample set in the step S2, and executing the steps S3-S4.
Preferably, the medium inlet pressure, flow, temperature and rotation speed of the compressor in the data samples are used as input values, and the output pressure ratio and the temperature ratio of the compressor are used as output values.
Preferably, the step S2 performs normalization processing on the training sample set, and maps the training sample set to the [ -1,1] interval uniformly.
Preferably, the mapping is performed using equation (1),
wherein Y is normalized data, and X is dataSample, X min Minimum value, X, in data samples max The maximum value in the data samples.
Preferably, the data ratio between the new training sample set and the test sample set in step S4 is 3:1.
Preferably, the step S3 includes the steps of:
step S31, concentrating the training samples
Inputting an OS-ELM network for parameter training, and randomly selecting an input connection weight value a i Threshold b with hidden layer node i I=1, …, L, where N 0 Is the number of data samples and N 0 ≥L；
Step S32, calculating an initial hidden layer output matrix H 0 ，
Wherein g (a) i x i +b i ) To activate the function, L is an hidden layer node, x i Meaning the medium inlet pressure, flow, temperature and rotational speed, y of the compressor i Meaning the output pressure ratio and the temperature ratio of the compressor;
step S33, calculating an initial output connection weight vector
Wherein->
Juxtaposition k=0, K being the number of iterations;
step S34, new training sample set (x K+1 ,y K+1 ) Adding training sample set Ω 0 Calculating the K+1st hidden layer output matrix H K+1 And updates P according to K+1 And beta K+1 ；
And step S35, K=K+1, returning to step S34 for updating iteration until training of all sample data is finished, and obtaining the compressor performance prediction model based on the OS-ELM.
Preferably, the step S4 includes online setting a sliding window length M, an error threshold E and a data updating frequency F, dividing the data in the online sliding window into a new training sample set and a test sample set, verifying the matching between the compressor performance prediction model based on OS-ELM and the working condition by the test sample set, performing effect evaluation on the prediction output data, wherein the model prediction error frequency is greater than F and the prediction output error is greater than E, inputting the new training sample set in the online sliding window into the training sample set of the step S2, and executing the steps S3-S4 until the prediction error frequency is less than F and the prediction output error is less than E, otherwise, online applying the model.
Preferably, the prediction error in step S4 is evaluated by using a criterion of root mean square error RMSE and maximum error absolute value MAE, and the N test samples are:
wherein y is i For the actual output value, Y i The output values are predicted for the model.
The invention also discloses an OS-ELM-based compressor performance online prediction system which is characterized by comprising a data acquisition module, a data processing module, a model building module and a model updating module;
the data acquisition module is used for acquiring the operation performance data of the compressor as a data sample;
the data processing module is used for preprocessing the data samples;
the model building module is used for building a compressor performance prediction model based on the OS-ELM by adopting the preprocessed training sample set;
the updating model module is used for guiding online updating of the compressor performance prediction model based on the OS-ELM by utilizing the online sliding window when the working condition is changed and predicting an output result online.
Compared with the prior art, the invention has the following beneficial effects: in addition, the invention can track the change of the operation condition in real time through an online sliding window technology, realize the rapid and accurate establishment of the compressor performance prediction model under the whole working condition, effectively solve the problem of model-working condition mismatch caused by the variable operation condition of the compressor and realize the online real-time updating application of the model.
Further, the data ratio of the training sample set to the test sample set is 3:1, and the compressor performance prediction model based on the OS-ELM is better in effectiveness.
Further, the online real-time update of the model is realized by adopting the online learning mode of the OS-ELM algorithm to perform real-time data learning.
Furthermore, the method and the device avoid the problem that the algorithm instantaneity is poor due to frequent model updating caused by noise or other factors by an online sliding window technology.
Furthermore, the prediction precision of the model is evaluated by utilizing the criterion of Root Mean Square Error (RMSE) and maximum error absolute value (MAE), the actual situation of the prediction error of the model is better reflected, and the prediction precision of the prediction model of the compressor performance based on the OS-ELM is higher.
Drawings
FIG. 1 is a flow chart of an OS-ELM based compressor performance online prediction method of the present invention;
FIG. 2 is an output pressure ratio of the development of a compressor performance prediction model of the present invention;
FIG. 3 is an output temperature ratio of the present invention for developing a predictive model of compressor performance.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The invention is described in further detail below with reference to the attached drawing figures:
an on-line prediction method for compressor performance based on OS-ELM, see FIG. 1, comprises the following steps:
s1, collecting compressor operation performance data as a training sample set;
s2, preprocessing a training sample set;
s3, constructing a compressor performance prediction model based on the OS-ELM by using the preprocessed training sample set;
s4, dividing data in the online sliding window into a new training sample set and a test sample set by setting the sliding window online, verifying the matching of the compressor performance prediction model based on the OS-ELM and the working condition by the test sample set, and if the prediction error of the compressor performance prediction model based on the OS-ELM is smaller than a set value, putting the model into use, otherwise, entering the step S5;
and S5, inputting the new training sample set in the online sliding window into the training sample set in the step S2, and executing the steps S3-S4.
According to the invention, 255 groups of data samples of an earlier period of operation time and 300 groups of data samples of a latest period of operation time are selected from historical operation data of different periods respectively and used for OS-ELM network training, and in addition, 100 groups of data samples are selected from the operation data of the latest period of time and used as test sample sets for verifying the matching of the model. The test sample set described above was used to verify the predicted performance of the compressor network model for each period of time, with the predicted pressure ratio and temperature such as shown in fig. 2 and 3. For better illustration, the predicted results of the compressor mechanism model and the actual operating data of the compressor are all in the same graph. As can be seen from fig. 2 and fig. 3, compared with the mechanism model, the OS-ELM-based compressor performance prediction model has a better prediction effect, is more consistent with the actual output value of the compressor, and can solve the problem of a larger prediction error of the mechanism model to a certain extent. In addition, the RMSE and MAE predicted by each model are listed in table 1, respectively. It can be seen that the OS-ELM based compressor performance prediction model has higher prediction accuracy compared to the mechanism model. Meanwhile, a network model obtained by training operation data in an earlier period is utilized, the RMSE and the MAE of the output pressure ratio prediction error of the compressor are respectively 0.023 and 0.085, and the RMSE and the MAE of the output temperature ratio prediction error of the compressor are respectively 0.005 and 0.015; by using the online sliding window technique, the OS-ELM is updated with the latest period of operating data, the RMSE and MAE of the output pressure ratio prediction error of the compressor are 0.012 and 0.046, respectively, and the RMSE and MAE of the output temperature ratio prediction error of the compressor are 0.002 and 0.011, respectively. Therefore, the prediction error of the network model after online updating is obviously reduced, and the rapid online updating of the model is realized.
Table 1 RMSE versus MAE for each model
The above is only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited by this, and any modification made on the basis of the technical scheme according to the technical idea of the present invention falls within the protection scope of the claims of the present invention.
Claims (9)
1. An on-line prediction method for compressor performance based on OS-ELM is characterized by comprising the following steps:
s1, collecting compressor operation performance data as a training sample set;
s2, preprocessing a training sample set;
s3, constructing a compressor performance prediction model based on the OS-ELM by using the preprocessed training sample set;
s4, dividing data in the online sliding window into a new training sample set and a test sample set by setting the sliding window online, verifying the matching of the compressor performance prediction model based on the OS-ELM and the working condition by the test sample set, and if the prediction error of the compressor performance prediction model based on the OS-ELM is smaller than a set value, putting the model into use, otherwise, entering the step S5;
and S5, inputting the new training sample set in the online sliding window into the training sample set in the step S2, and executing the steps S3-S4.
2. The method of claim 1, wherein the medium inlet pressure, flow rate, temperature and rotation speed of the compressor in the data samples are used as input values, and the output pressure ratio and the temperature ratio of the compressor are used as output values.
3. The online prediction method of compressor performance based on OS-ELM according to claim 1, wherein the step S2 performs normalization processing on the training sample set, and maps the normalization processing to [ -1,1] intervals.
4. An OS-ELM-based compressor performance online prediction method according to claim 3, wherein the mapping is performed using equation (1),
wherein Y is normalized data, X is a data sample, X min Minimum value, X, in data samples max The maximum value in the data samples.
5. The online prediction method of compressor performance based on OS-ELM according to claim 1, wherein the data ratio between the new training sample set and the test sample set in step S4 is 3:1.
6. The online prediction method of compressor performance based on OS-ELM according to claim 3, wherein the step S3 comprises the steps of:
step S31, concentrating the training samples
Inputting an OS-ELM network for parameter training, and randomly selecting an input connection weight value a i Threshold b with hidden layer node i I=1, …, L, where N 0 Is the number of data samples and N 0 ≥L；
Step S32, calculating an initial hidden layer output matrix H 0 ，
Wherein g (a) i x i +b i ) To activate the function, L is an hidden layer node, x i Meaning the medium inlet pressure, flow, temperature and rotational speed, y of the compressor i Meaning the output pressure ratio and the temperature ratio of the compressor;
step S33, calculating initial inputOutput connection weight vector
Wherein->
Juxtaposition k=0, K being the number of iterations;
step S34, new training sample set (x K+1 ,y K+1 ) Adding training sample set Ω 0 Calculating the K+1st hidden layer output matrix H K+1 And updates P according to K+1 And beta K+1 ；
And step S35, K=K+1, returning to step S34 for updating iteration until training of all sample data is finished, and obtaining the compressor performance prediction model based on the OS-ELM.
7. The online prediction method of the performance of the compressor based on the OS-ELM according to claim 6, wherein the step S4 comprises online setting of a sliding window length M, an error threshold E and a data updating frequency F, dividing data in the online sliding window into a new training sample set and a test sample set, verifying the matching of a prediction model of the performance of the compressor based on the OS-ELM and a working condition by the test sample set, performing effect evaluation on the prediction output data, inputting the new training sample set in the online sliding window into the training sample set of the step S2, and executing the steps S3-S4 until the prediction error frequency is smaller than F and the prediction output error is smaller than E, otherwise, performing online application of the model.
8. An OS-ELM-based compressor performance online prediction method of claim 3The method is characterized in that the prediction error in the step S4 is evaluated by adopting a criterion of a Root Mean Square Error (RMSE) and a maximum error absolute value (MAE), and the N test samples are as follows:
wherein y is i For the actual output value, Y i The output values are predicted for the model.
9. The compressor performance online prediction system based on the OS-ELM is characterized by comprising a data acquisition module, a data processing module, a model building module and a model updating module;
the data acquisition module is used for acquiring the operation performance data of the compressor as a data sample;
the data processing module is used for preprocessing the data samples;
the model building module is used for building a compressor performance prediction model based on the OS-ELM by adopting the preprocessed training sample set;
the updating model module is used for guiding online updating of the compressor performance prediction model based on the OS-ELM by utilizing the online sliding window when the working condition is changed and predicting an output result online.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310084524.2A CN116085245A (en) | 2023-01-31 | 2023-01-31 | Online compressor performance prediction method and system based on OS-ELM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310084524.2A CN116085245A (en) | 2023-01-31 | 2023-01-31 | Online compressor performance prediction method and system based on OS-ELM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116085245A true CN116085245A (en) | 2023-05-09 |
Family
ID=86186689
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310084524.2A Pending CN116085245A (en) | 2023-01-31 | 2023-01-31 | Online compressor performance prediction method and system based on OS-ELM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116085245A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116735146A (en) * | 2023-08-11 | 2023-09-12 | 中国空气动力研究与发展中心低速空气动力研究所 | Wind tunnel experiment method and system for establishing aerodynamic model |
-
2023
- 2023-01-31 CN CN202310084524.2A patent/CN116085245A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116735146A (en) * | 2023-08-11 | 2023-09-12 | 中国空气动力研究与发展中心低速空气动力研究所 | Wind tunnel experiment method and system for establishing aerodynamic model |
| CN116735146B (en) * | 2023-08-11 | 2023-10-13 | 中国空气动力研究与发展中心低速空气动力研究所 | Wind tunnel experiment method and system for establishing aerodynamic model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108647428B (en) | Turbofan engine adaptive component level simulation model construction method | |
| CN107392368B (en) | Meteorological forecast-based office building dynamic heat load combined prediction method | |
| CN108920863B (en) | Method for establishing energy consumption estimation model of robot servo system | |
| CN108876021B (en) | Medium-and-long-term runoff forecasting method and system | |
| CN111967688B (en) | Power load prediction method based on Kalman filter and convolutional neural network | |
| US10331810B2 (en) | Method for determining a model of an output quantity of a technical system | |
| CN110083910B (en) | NSGA-II based chaotic time sequence prediction sample acquisition method | |
| CN113657660A (en) | Heat source load prediction method based on substation load and heat supply network hysteresis model | |
| CN116085245A (en) | Online compressor performance prediction method and system based on OS-ELM | |
| CN112113146B (en) | Synchronous self-adaptive check method for roughness coefficient and node water demand of water supply pipe network pipeline | |
| CN102708294A (en) | Self-adaptive parameter soft measuring method on basis of semi-supervised local linear regression | |
| CN112418495A (en) | Building energy consumption prediction method based on longicorn stigma optimization algorithm and neural network | |
| Dong et al. | Short-term building cooling load prediction model based on DwdAdam-ILSTM algorithm: A case study of a commercial building | |
| CN111859249A (en) | Ocean numerical forecasting method based on analytical four-dimensional set variation | |
| CN109753634B (en) | Historical data steady-state value-based dynamic system gain estimation method | |
| CN116707331B (en) | Inverter output voltage high-precision adjusting method and system based on model prediction | |
| CN114583767B (en) | Data-driven wind power plant frequency modulation response characteristic modeling method and system | |
| CN114707753B (en) | Regional LSTM flood forecasting method | |
| CN116663727A (en) | Photovoltaic power prediction method and system | |
| CN110909492A (en) | Sewage treatment process soft measurement method based on extreme gradient lifting algorithm | |
| CN115936236A (en) | Method, system, equipment and medium for predicting energy consumption of cigarette factory | |
| CN112348700B (en) | Line capacity prediction method combining SOM clustering and IFOU equation | |
| CN114970698A (en) | Metering equipment operation performance prediction method based on improved LWPLS | |
| CN113159395A (en) | Deep learning-based sewage treatment plant water inflow prediction method and system | |
| CN113408076A (en) | Small sample mechanical residual life prediction method based on support vector machine model |
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 |