Detection and correction method for related data stream exception

A dual data stream, anomaly detection technology, applied in complex mathematical operations and other directions, can solve problems such as poor applicability, inability to reflect individual changes in data, lowering the accuracy of data streams, etc., to achieve the effect of improving quality and output

Active Publication Date: 2018-07-31
WUXI INSPECTION TESTING & CERTIFICATION INST
View PDF6 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] At present, for the single data stream anomaly detection method, the first type is based on the statistical analysis characteristics of data: the method is relatively simple to implement and easy to implement, but because the window statistical characteristics are used to replace the data itself, the changes between data individuals are encapsulated in Expressed in the statistical feature mode, it cannot reflect the individual changes of the data, and it is difficult to detect the abnormal points of the data flow in the window
The second type is the anomaly detection method based on the distribution characteristics of the data stream: the distribution model is established according to the data distribution characteristics, and when the real-time arriving data does not conform to the distribution model, it is considered as abnormal data. In the application, it is difficult to determine the distribution form of the data flow, so the applicability is not strong
The above-mentioned common types of methods can realize anomaly detection of a single data stream, but none of them consider the correlation between data streams and the impact on data uncertainty, and this influencing factor will reduce the data flow in the data stream. Accuracy when performing uncertainty detection

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Detection and correction method for related data stream exception
  • Detection and correction method for related data stream exception
  • Detection and correction method for related data stream exception

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] Below in conjunction with shown in accompanying drawing, the present invention is described in further detail:

[0041] In the liquor brewing process, especially in the solid-state brewing process, the steam pressure and steam flow rate affect the quality and yield of the distilled wine (see literature: [1] Pan Aizhen, Zhao Xuemin, Zhang Guangteng, etc. Solid-liquid two-state mixed distillation equipment Preliminary research on fuel ethanol distillation process and its equipment[J].Agricultural Machinery,2012(16):158-161.[2]Zhang Guangteng,Pan Aizhen,Si Zhenjun.Research on fuel ethanol new distillation process and its equipment[J].Value Engineering,2012 ,31(28):32-33.[3] Rao Jiaquan, Feng Bo, Du Liquan, et al. Research on the relationship between the speed of the retort and the yield and quality of koji wine[J]. Wine Science and Technology, 2012(1):27-29.[4 ]Yang Yaru, Liu Dengfeng, Xu Guoqiang, et al. LabVIEW-based automatic liquor retort distillation monitoring system...

Embodiment 2

[0069] Embodiment 2: the inventive method application instance data

[0070] Table 1 and Table 2 are the data detection and correction tables of a certain steam pressure and steam flow data flow respectively.

[0071] In this example, Table 1 contains 4 types of data: time index number, original data, comparative data and corrected data. The physical meanings of the 4 types of data are as follows: time index serial number I=1,2,...,108, that is, each round of production process is divided into n=108 time periods; the original data is the online collected steam pressure data X* ; The comparison data is the cluster center data CCS calculated by the offline process x , used to proofread the original data; the corrected data is the data corrected by using the comparative data after the original data is detected with uncertain data.

[0072] Table 1 shows an application example of the invention with a time series length of 108: at each time index number, the vapor pressure data v...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a detection and correction method for related data stream exception, and belongs to the field of industrial process modeling and data processing. In the method, clustering analysis and window matching are used for realizing the exceptional point detection and correction of the single input data stream. A GPR (Gaussian Process Regression) model is introduced to establish theprediction model of an input-output data stream, the output of the data stream and the prediction model are observed in real time to carry out window comparison, and the detection and the correctionof the exceptional data in the output data stream are further solved. The method not only considers the general features of the data stream but also utilizes correlation among the data streams to solve the problem that input/ output multi-data-stream exceptional points are difficult to be correctly detected and corrected in practice.

Description

technical field [0001] The invention relates to a method for abnormal detection and correction of correlated double data streams, and belongs to the field of industrial process modeling and data processing. Background technique [0002] After continuous collection of discrete production data, a data flow is formed macroscopically. These real-time collected data streams are affected by many uncertain factors such as measurement accuracy, measurement error, noise, and surrounding environment interference, which makes the data stream obtained by the actual system not only have the inherent characteristics of real-time, continuous, orderly, and fast arrival, but also There is also uncertainty. For example, in the production monitoring process of liquor brewing, two types of data streams, steam pressure and steam flow, need to be measured. Similar to the general measurement process, the measured data stream must contain many uncertainties, forming an uncertain data stream. But ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06F17/18
CPCG06F17/18
Inventor 吴宏杰张聪章晓明秦宁宁朱树才
Owner WUXI INSPECTION TESTING & CERTIFICATION INST
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap