CN110633516A - Method for predicting performance degradation trend of electronic device - Google Patents

Method for predicting performance degradation trend of electronic device Download PDF

Info

Publication number
CN110633516A
CN110633516A CN201910812431.0A CN201910812431A CN110633516A CN 110633516 A CN110633516 A CN 110633516A CN 201910812431 A CN201910812431 A CN 201910812431A CN 110633516 A CN110633516 A CN 110633516A
Authority
CN
China
Prior art keywords
data
codebook
output
electronic device
prediction
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201910812431.0A
Other languages
Chinese (zh)
Other versions
CN110633516B (en
Inventor
刘震
梅文娟
杜立
程玉华
黄建国
白利兵
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN201910812431.0A priority Critical patent/CN110633516B/en
Publication of CN110633516A publication Critical patent/CN110633516A/en
Application granted granted Critical
Publication of CN110633516B publication Critical patent/CN110633516B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a method for predicting performance degradation trend of an electronic device, which is characterized in that a prediction model of an initial correlation entropy extreme learning machine is established in an off-line manner, a dynamic correlation entropy extreme learning machine is updated on line, a code book mechanism combined with singular values is constructed, and then the singular values in historical data are identified, so that the influence of noise and the singular values in the data on the prediction model is overcome, and the final prediction effect of the prediction model is improved.

Description

Method for predicting performance degradation trend of electronic device
Technical Field
The invention belongs to the technical field of reliability of sub-components, and particularly relates to a method for predicting performance degradation trend of an electronic device.
Background
With the rapid development of big data, cloud computing and industrial internet, intelligent troubleshooting and health management of electronic devices have been receiving attention in recent years due to the advantages of sharing resources and related services. Meanwhile, along with the development of intelligent health management, the information interaction speed between the electronic equipment and the health management platform is increased, and the information scale is enlarged. In this respect, the acquisition of large data of the device helps to improve the accuracy of the prediction. On the other hand, the performance degradation trend data of the electronic equipment belongs to real-time flow, and has the characteristics of data output one by one, lower throughput compared with batch data in other prediction scenes, strict time delay requirement on a prediction model and the like, so that a new challenge is provided for the design of a performance degradation trend model. How to obtain accurate prediction results at high speed becomes one of the technical problems in the background of 4.0 of industry.
Most of the existing real-time flow prediction methods adopt a neural network to carry out off-line training on historical information of electronic equipment to obtain a prediction model, and then use the prediction model to carry out on-line prediction. However, for most electronic devices, the performance degradation trend is not obvious at an early stage, and the early history information cannot completely reflect the degradation rule thereof, so that a large error exists in the prediction of the performance degradation trend. On the other hand, the prediction model adopted by the existing method needs to adjust network parameters for many times in the training process, and the training time is long, so that the defect of long time delay exists in the application of real-time stream fault trend prediction. Different from the traditional neural network, the online extreme learning machine has the characteristics of rapidness and accuracy due to the characteristics of simple model, no need of adjusting hidden layer parameters and the like. Therefore, in recent years, the model and related improved methods have been widely used in real-time diagnostics. However, the method considers the statistical rule of all historical information of the device when establishing the model. When the device is degraded, the prediction result is affected by the health state information of the device and the early degradation information, so that the current performance degradation trend cannot be well reflected by the prediction result. Meanwhile, the online learning machine adopts the minimum mean square criterion as the training basis of the output layer, so that the prediction model is easily influenced by non-Gaussian noise and singular values, and the prediction effect is reduced.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a method for predicting the performance degradation trend of an electronic device.
In order to achieve the above object, the present invention provides a method for predicting a performance degradation trend of an electronic device, comprising the steps of:
(1) establishing a prediction model of an initial correlation entropy extreme learning machine in an off-line manner
(1.1) acquiring historical data { x) of the electronic device to be testedi,yiWherein, i-1, 2, …, k-1, k-1 represents the historical data number before real-time prediction, xiAs real-time data at the i-th moment of the electronic device under test, yiIs xiA corresponding expected degradation trend characteristic;
(1.2) randomly generating hidden layer weight { w) of related entropy extreme learning machine1,w2,…,wMAnd bias b1,b2,…,bmM represents the dimension of the hidden layer weight, and M represents the number of offsets;
(1.3) calculating hidden layer output H under historical data(k-1)；
Figure BDA0002185427720000021
Wherein T represents transpose, g (w)1,b1,x1) Representing the output of the 1 st historical data under the 1 st hidden layer node;
(1.4) calculating output layer output under historical data;
Figure BDA0002185427720000022
wherein,
Figure BDA0002185427720000023
eithe error between the predicted value and the true value of the model at the ith moment is represented, and sigma is the related entropy variance; t is(k-1)For the history truth of the electronic device at the k-1 timeOutputting the value;
(1.5) initializing a code book of the relevant entropy limit learning machine;
setting the length L of the codebookw：
Lw＝min{Lmax,k-1}
Wherein L ismaxThe maximum length used for a codebook;
record information in the codebook:
Figure BDA0002185427720000031
Figure BDA0002185427720000032
Figure BDA0002185427720000033
Figure BDA0002185427720000034
wherein, WcbFor prediction error of data in the codebook, HcbFor hidden layer output of corresponding data, acbFor the associated entropy of the data in the codebook, TcbFor the actual output of the data in the codebook,
Figure BDA0002185427720000035
is the L < th > of code bookwThe hidden layer corresponding to each data is output,
Figure BDA0002185427720000036
is the L < th > of code bookwActual output corresponding to the data;
(1.6) calculating information of singular value estimation;
Figure BDA0002185427720000038
wherein,
Figure BDA0002185427720000039
for the variance of the singular value estimate, alpha is the singular value estimate threshold, lambda1、λ2、λ3Is a constant;
(2) acquiring real-time data x of the tested electronic device at the kth momentkAs input to a predictive model;
(3) calculating real-time data x input at the kth moment based on the prediction modelkCorresponding hidden layer output H(k)；
H(k)＝{g(w1,b1,xk),g(w2,b2,xk),…,g(wM,bm,xk)}
(4) And calculating a prediction result y at the k-th time(k)；
y(k)＝β(k-1)H(k-1)
(5) Acquiring real-time data x at the kth momentkDegradation information t ofk；
(6) Using the real-time data x at the k-th timekUpdating K(k)；
Figure BDA00021854277200000310
(7) And calculating the real-time data x at the kth momentkLower output layer output beta(k)；
Figure BDA00021854277200000311
(8) Updating information in the code book;
(8.1) if the number of samples in the codebook is less than LmaxThen the information update codebook is as follows:
Figure BDA0002185427720000041
Figure BDA0002185427720000042
Figure BDA0002185427720000043
Figure BDA0002185427720000044
(8.2) if the number of samples in the codebook has reached the maximum value LmaxThen the codebook is updated as follows:
Figure BDA0002185427720000047
(9) updating information of singular value estimation;
Figure BDA0002185427720000049
Figure BDA00021854277200000410
wherein,
Figure BDA00021854277200000411
for the variance of the updated singular value estimate,
Figure BDA00021854277200000412
estimating a threshold value for the updated singular value;
(10) extracting singular value information in the historical data;
H*＝{hi|hi∈Hcb,wi≥αM 2}
Λ*＝diag{ai|ai∈acb,wi≥αM 2}
T*＝{ti|ti∈Tcb,wi≥αM 2}
(11) eliminating the influence of singular value information in a prediction model to obtain the output layer output beta at the kth moment(k)Thereby predicting the performance degradation trend of the tested electronic device at the kth moment;
β(k)＝(K(k)-H*TΛ*H*)-1(K(k)β(k)-H*TΛ*T*)。
the invention aims to realize the following steps:
the invention relates to a method for predicting performance degradation trend of an electronic device, which is characterized in that a prediction model of an initial correlation entropy extreme learning machine is established in an off-line manner, a code book mechanism combined with a singular value is constructed by updating a dynamic correlation entropy extreme learning machine on line, and then the singular value in historical data is identified, so that the influence of noise and the singular value in the data on the prediction model is overcome, and the final prediction effect of the prediction model is improved.
Meanwhile, the method for predicting the performance degradation trend of the electronic device further has the following beneficial effects:
(1) the dynamic correlation entropy extreme learning machine is constructed, and can be used for overcoming the influence of noise and singular values in data on a prediction model;
(2) the influence of the singular value in the prediction model is eliminated, so that the performance degradation trend of the electronic device can be accurately and quickly predicted;
(3) the invention has the characteristics of fast dynamic update, good prediction effect and high robustness.
Drawings
FIG. 1 is a flow chart of a method for predicting degradation trend of electronic device performance according to the present invention;
FIG. 2 is a graph of the predicted effect of the photocoupler;
fig. 3 is a comparison graph of prediction accuracy of the photocoupler.
Detailed Description
The following description of the embodiments of the present invention is provided in order to better understand the present invention for those skilled in the art with reference to the accompanying drawings. It is to be expressly noted that in the following description, a detailed description of known functions and designs will be omitted when it may obscure the subject matter of the present invention.
Examples
FIG. 1 is a flow chart of a method for predicting degradation trend of electronic device performance according to the present invention.
In this embodiment, as shown in fig. 1, a method for predicting performance degradation trend of an electronic device according to the present invention includes the following steps:
s1, establishing a prediction model of the initial correlation entropy extreme learning machine in an off-line manner
S1.1, acquiring historical data { x) of the electronic device to be testedi,yiWherein i-1, 2, …, k-1, k-1 represents the number of historical data before real-time prediction, and xiIs the i-th real-time data, y, of the electronic device to be testediIs xiA corresponding expected degradation trend characteristic;
s1.2, randomly generating hidden layer weight { w) of related entropy extreme learning machine1,w2,…,wMAnd bias b1,b2,…,bmM represents the dimension of the hidden layer weight, and M represents the number of offsets;
s1.3, calculating hidden layer output H under historical data(k-1)；
Figure BDA0002185427720000051
Wherein T represents transpose, g (w)1,b1,x1) Representing the output of the 1 st historical data under the 1 st hidden layer node;
s1.4, calculating output layer output under historical data;
Figure BDA0002185427720000052
wherein, eithe error between the predicted value and the true value of the model at the ith moment is represented, and sigma is the related entropy variance; t is(k-1)The historical true output value of the electronic device at the k-1 moment is obtained;
s1.5, initializing a code book of the relevant entropy extreme learning machine;
setting the length L of the codebookw：
Lw＝min{Lmax,k-1}
Wherein L ismaxThe maximum length used for a codebook;
record information in the codebook:
Figure BDA0002185427720000062
Figure BDA0002185427720000063
Figure BDA0002185427720000064
Figure BDA0002185427720000065
wherein, WcbFor prediction error of data in the codebook, HcbFor hiding of corresponding dataLayer output, acbFor the associated entropy of the data in the codebook, TcbFor the actual output of the data in the codebook,
Figure BDA0002185427720000066
is the L < th > of code bookwThe hidden layer corresponding to each data is output,
Figure BDA0002185427720000067
is the L < th > of code bookwActual outputs corresponding to the individual data, i.e., degradation information;
s1.6, calculating information of singular value estimation;
Figure BDA0002185427720000068
Figure BDA0002185427720000069
wherein,
Figure BDA00021854277200000610
is the variance of the singular value estimation, alpha is the singular value estimation threshold, and med (-) represents the median calculation;
s2, collecting real-time data x of the tested electronic device at the kth momentkAs input to a predictive model;
s3, calculating the real-time data x input at the k-th time based on the prediction modelkCorresponding hidden layer output H(k)；
H(k)＝{g(w1,b1,xk),g(w2,b2,xk),…,g(wM,bm,xk)}
S4, calculating the prediction result y at the k-th moment(k)；
y(k)＝β(k-1)H(k-1)
S5, acquiring the real-time data x at the kth momentkDegradation information t ofk；
S6, LiUsing real-time data x at the kth momentkUpdating K(k)；
Figure BDA0002185427720000071
S7, calculating the real-time data x at the k momentkLower output layer output beta(k)；
Figure RE-GDA0002277374300000072
S8, updating information in the code book;
s8.1, if the number of samples in the code book is less than LmaxThen the information update codebook is as follows:
Figure BDA0002185427720000073
Figure BDA0002185427720000074
Figure BDA0002185427720000075
Figure BDA0002185427720000076
s8.2, if the number of samples in the code book reaches the maximum value LmaxThen the codebook is updated as follows:
Figure BDA0002185427720000077
Figure BDA0002185427720000078
Figure BDA0002185427720000079
Figure BDA00021854277200000710
s9, updating information of singular value estimation;
Figure BDA00021854277200000711
s10, extracting singular value information in the historical data;
H*＝{hi|hi∈Hcb,wi≥αM 2}
Λ*＝diag{ai|ai∈acb,wi≥αM 2}
T*＝{ti|ti∈Tcb,wi≥αM 2}
s11, eliminating the influence of singular value information in the prediction model to obtain the output layer output beta at the kth moment(k)Thereby predicting the performance degradation trend of the tested electronic device at the kth moment;
β(k)＝(K(k)-H*TΛ*H*)-1(K(k)β(k)-H*TΛ*T*)。
examples of the invention
In order to illustrate the technical effects of the invention, the invention is verified by taking the direct current transmission ratio real-time current prediction of the photoelectric coupler as an example.
The photoelectric coupler is an electronic component for converting electric energy and light energy, which transmits electric signals by taking light as a medium, and is used for isolating input and output electric signals. The direct current transmission ratio of the photoelectric coupler can effectively reflect the health state of the device. In order to verify the effectiveness of the method, a prediction model is established by the method, and the trend of real-time stream data in the degradation state of the photoelectric coupler is predicted.
Meanwhile, the method is compared with a relevant entropy limit learning machine (RCC-ELM), an online limit learning machine (MOS-ELM), an M estimation online limit learning machine (MOS-ELM) and an extreme survival error learning machine (ESEP-ELM). The off-line training precision and the on-line training precision are shown in table 1.
Algorithm RCC-ELM ESEP-ELM OSELM MOSELM The invention
Off-line training accuracy 4.68E-05 1.88E-04 4.61E-05 4.68E-05 4.61E-05
Real-time stream prediction accuracy 6.76E-04 2.62E-04 1.91E-04 1.88E-04 9.65E-05
TABLE 1
As can be seen from Table 1, the prediction model generated by the invention can achieve the optimal off-line training precision and the prediction precision under real-time streaming. As can be seen from comparison of prediction curves generated by the models in fig. 2, the off-line prediction method RCC-ELM is only trained on off-line data, so that the data rule of the photocoupler during degradation cannot be accurately captured, and therefore reliable prediction is hardly performed. ESEP-ELM, OSELM and MOS-ELM can effectively and reasonably predict the model, but are influenced by noise in data, the prediction result has large fluctuation, and large errors exist on partial nodes. Compared with the prior art, the method overcomes the influence of noise, the prediction curve can accurately reflect the performance degradation trend of the fault, the prediction curve has no large fluctuation, and the prediction effect is stable.
Fig. 3 compares the prediction errors of the respective methods for the respective samples, and it can be seen that the prediction errors of the first 10 sample points of the five methods are all low, while the information of the following sample points can show that RCC-ELM, ESEP-ELM, OSELM and MOS-ELM all have a phenomenon of reduced prediction accuracy, but the prediction accuracy of the present invention is relatively stable and reduced, which indicates that the algorithm has the capability of stably reflecting the data change trend and provides a reliable prediction result.
Table 2 shows the time comparison between the training and prediction of the five algorithms, and it can be seen that the prediction model of the present invention has a low training time, the prediction time is controlled within 0.5 second, and the present invention has advantages compared with other real-time algorithms, and can well adapt to the time requirement of performance degradation trend prediction under the real-time streaming condition.
Algorithm RCCELM ESEP-ELM OSELM MOS-ELM The invention
Test time(s) 0.2972 0.05113 0.00815 0.03764 0.00964
Predicting time(s) 0.03667 0.03462 0.5085 0.55803 0.45394
TABLE 2
Although illustrative embodiments of the present invention have been described above to facilitate the understanding of the present invention by those skilled in the art, it should be understood that the present invention is not limited to the scope of the embodiments, and various changes may be made apparent to those skilled in the art as long as they are within the spirit and scope of the present invention as defined and defined by the appended claims, and all inventions utilizing the inventive concept are protected.

Claims (1)

1. A method for predicting performance degradation trend of an electronic device is characterized by comprising the following steps:
(1) establishing a prediction model of an initial correlation entropy extreme learning machine in an off-line manner
(1.1) acquiring historical data { x) of the electronic device to be testedi,yiWherein i-1, 2, …, k-1, k-1 represents the number of historical data before real-time prediction, and xiAs real-time data at the i-th moment of the electronic device under test, yiIs xiA corresponding expected degradation trend characteristic;
(1.2) randomly generating hidden layer weight { w) of related entropy extreme learning machine1,w2,…,wMAnd bias b1,b2,…,bm}；
(1.3) calculating hidden layer output H under historical data(k-1)；
Figure FDA0002185427710000011
Wherein T represents transpose, g (w)1,b1,x1) Representing the output of the 1 st historical data under the 1 st hidden layer node;
(1.4) calculating output layer output under historical data;
Figure FDA0002185427710000012
wherein,eithe error between the predicted value and the true value of the model at the ith moment is represented, and sigma is the related entropy variance; t is(k-1)The historical true output value of the electronic device at the k-1 moment is obtained;
(1.5) initializing a code book of the relevant entropy limit learning machine;
setting the length L of the codebookw：
Lw＝min{Lmax,k-1}
Wherein L ismaxThe maximum length used for a codebook;
record information in the codebook:
Figure FDA0002185427710000021
Figure FDA0002185427710000022
Figure FDA0002185427710000023
wherein, WcbFor prediction error of data in the codebook, HcbFor hidden layer output of corresponding data, acbFor the associated entropy of the data in the codebook, TcbFor the actual output of the data in the codebook,is the L < th > of code bookwThe hidden layer corresponding to each data is output,
Figure FDA0002185427710000026
is the L < th > of code bookwActual output corresponding to the data;
(1.6) calculating information of singular value estimation;
Figure FDA0002185427710000027
Figure FDA0002185427710000028
wherein,
Figure FDA0002185427710000029
is a strangeVariance of value estimate, α being singular value estimate threshold, λ1、λ2、λ3Constant, med (-) represents the median calculation;
(2) acquiring real-time data x of the tested electronic device at the kth momentkAs input to a predictive model;
(3) calculating real-time data x input at the kth moment based on the prediction modelkCorresponding hidden layer output H(k)；
H(k)＝{g(w1,b1,xk),g(w2,b2,xk),…,g(wM,bm,xk)}
(4) And calculating a prediction result y at the k-th time(k)；
y(k)＝β(k-1)H(k-1)
(5) Acquiring real-time data x at the kth momentkDegradation information t ofk；
(6) Using the real-time data x at the k-th timekUpdating K(k)；
Figure FDA00021854277100000210
(7) And calculating the real-time data x at the kth momentkLower output layer output beta(k)；
Figure FDA00021854277100000211
(8) Updating information in the code book;
(8.1) if the number of samples in the codebook is less than LmaxThen the information update codebook is as follows:
Figure FDA0002185427710000031
Figure FDA0002185427710000034
(8.2) if the number of samples in the codebook has reached the maximum value LmaxThen the codebook is updated as follows:
Figure FDA0002185427710000035
Figure FDA0002185427710000037
Figure FDA0002185427710000038
(9) updating information of singular value estimation;
Figure FDA0002185427710000039
Figure FDA00021854277100000310
wherein,
Figure FDA00021854277100000311
for the variance of the updated singular value estimate,
Figure FDA00021854277100000312
to be moreA new singular value estimation threshold;
(10) extracting singular value information in the historical data;
H*＝{hi|hi∈Hcb,wi≥αM 2}
Λ*＝diag{ai|ai∈acb,wi≥αM 2}
T*＝{ti|ti∈Tcb,wi≥αM 2}
(11) eliminating the influence of singular value information in a prediction model to obtain the output layer output beta at the kth moment(k)Thereby predicting the performance degradation trend of the tested electronic device at the kth moment;
β(k)＝(K(k)-H*TΛ*H*)-1(K(k)β(k)-H*TΛ*T*)。
CN201910812431.0A 2019-08-30 2019-08-30 Method for predicting performance degradation trend of electronic device Active CN110633516B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910812431.0A CN110633516B (en) 2019-08-30 2019-08-30 Method for predicting performance degradation trend of electronic device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910812431.0A CN110633516B (en) 2019-08-30 2019-08-30 Method for predicting performance degradation trend of electronic device

Publications (2)

Publication Number Publication Date
CN110633516A true CN110633516A (en) 2019-12-31
CN110633516B CN110633516B (en) 2022-06-14

Family

ID=68969553

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910812431.0A Active CN110633516B (en) 2019-08-30 2019-08-30 Method for predicting performance degradation trend of electronic device

Country Status (1)

Country Link
CN (1) CN110633516B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111143973A (en) * 2019-12-05 2020-05-12 云南电网有限责任公司玉溪供电局 Valve-regulated lead-acid storage battery degradation trend prediction method based on Gauss process regression
CN113361189A (en) * 2021-05-12 2021-09-07 电子科技大学 Chip performance degradation trend prediction method based on multi-step robust prediction learning machine

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140222349A1 (en) * 2013-01-16 2014-08-07 Assurerx Health, Inc. System and Methods for Pharmacogenomic Classification
CN107085704A (en) * 2017-03-27 2017-08-22 杭州电子科技大学 Fast face expression recognition method based on ELM own coding algorithms
CN107451278A (en) * 2017-08-07 2017-12-08 北京工业大学 Chinese Text Categorization based on more hidden layer extreme learning machines
US20180136617A1 (en) * 2016-11-11 2018-05-17 General Electric Company Systems and methods for continuously modeling industrial asset performance
CN109447272A (en) * 2018-09-29 2019-03-08 西安交通大学 A kind of extreme learning machine method based on center of maximum cross-correlation entropy criterion
CN109934295A (en) * 2019-03-18 2019-06-25 重庆邮电大学 A kind of image classification and method for reconstructing based on the hidden feature learning model that transfinites
US20190230119A1 (en) * 2018-01-19 2019-07-25 General Electric Company Dynamic concurrent learning method to neutralize cyber attacks and faults for industrial asset monitoring nodes

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140222349A1 (en) * 2013-01-16 2014-08-07 Assurerx Health, Inc. System and Methods for Pharmacogenomic Classification
US20180136617A1 (en) * 2016-11-11 2018-05-17 General Electric Company Systems and methods for continuously modeling industrial asset performance
CN107085704A (en) * 2017-03-27 2017-08-22 杭州电子科技大学 Fast face expression recognition method based on ELM own coding algorithms
CN107451278A (en) * 2017-08-07 2017-12-08 北京工业大学 Chinese Text Categorization based on more hidden layer extreme learning machines
US20190230119A1 (en) * 2018-01-19 2019-07-25 General Electric Company Dynamic concurrent learning method to neutralize cyber attacks and faults for industrial asset monitoring nodes
CN109447272A (en) * 2018-09-29 2019-03-08 西安交通大学 A kind of extreme learning machine method based on center of maximum cross-correlation entropy criterion
CN109934295A (en) * 2019-03-18 2019-06-25 重庆邮电大学 A kind of image classification and method for reconstructing based on the hidden feature learning model that transfinites

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
XING HJ 等: "Training extreme learning machine via regularized correntropy criterion", 《NEURAL COMPUT & APPLIC》 *
翟俊海 等: "基于MapReduce的大数据主动学习", 《计算机应用》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111143973A (en) * 2019-12-05 2020-05-12 云南电网有限责任公司玉溪供电局 Valve-regulated lead-acid storage battery degradation trend prediction method based on Gauss process regression
CN111143973B (en) * 2019-12-05 2021-01-26 云南电网有限责任公司玉溪供电局 Valve-regulated lead-acid storage battery degradation trend prediction method based on Gauss process regression
CN113361189A (en) * 2021-05-12 2021-09-07 电子科技大学 Chip performance degradation trend prediction method based on multi-step robust prediction learning machine
CN113361189B (en) * 2021-05-12 2022-04-19 电子科技大学 Chip performance degradation trend prediction method based on multi-step robust prediction learning machine

Also Published As

Publication number Publication date
CN110633516B (en) 2022-06-14

Similar Documents

Publication Publication Date Title
CN101477172B (en) Analogue circuit fault diagnosis method based on neural network
CN106847302B (en) Single-channel mixed voice time domain separation method based on convolutional neural network
CN110632572B (en) Radar radiation source individual identification method and device based on unintentional phase modulation characteristics
CN109376892B (en) Equipment state prediction method based on life cycle stage of equipment
CN110633516B (en) Method for predicting performance degradation trend of electronic device
CN111428201B (en) Prediction method for time series data based on empirical mode decomposition and feedforward neural network
Yin Quantized identification with dependent noise and Fisher information ratio of communication channels
CN108875126A (en) Electrolytic capacitor method for predicting residual useful life
CN109088770B (en) Electromechanical system interactive network modeling method based on self-adaptive symbol transfer entropy
CN110879927A (en) Sea clutter amplitude statistical distribution field modeling method for sea target detection
CN116992986A (en) Network traffic prediction system based on time sequence decomposition
CN115495924A (en) MOSFET service life prediction method based on ARIMA model
CN108984851B (en) Weighted Gaussian model soft measurement modeling method with time delay estimation
CN110187206A (en) Fault detection method for suspension system in non-Gaussian process under complex working condition
CN111901773A (en) Industrial park air quality detection method
CN111462479A (en) Traffic flow prediction method based on Fourier-recurrent neural network
CN110598334B (en) Performance degradation trend prediction method based on collaborative derivation related entropy extreme learning machine
CN112633550A (en) RNN-based catenary fault trend prediction method, equipment and storage medium
CN113297540A (en) APP resource demand prediction method, device and system under edge Internet of things agent service
CN113821419A (en) Cloud server aging prediction method based on SVR and Gaussian function
CN113627621A (en) Active learning method for regression estimation of transmission quality of optical network
CN116223044B (en) Bearing residual life prediction system and method based on multi-sensor fusion and Bi-TACN
CN117610397B (en) Intelligent analysis method and device for using effect based on sewage treatment equipment
CN117668623B (en) Multi-sensor cross-domain fault diagnosis method for leakage of ship pipeline valve
CN109521176B (en) Virtual water quality monitoring method based on improved deep extreme learning machine

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant