CN108205432B - Real-time elimination method for observation experiment data abnormal value - Google Patents
Real-time elimination method for observation experiment data abnormal value Download PDFInfo
- Publication number
- CN108205432B CN108205432B CN201611165505.9A CN201611165505A CN108205432B CN 108205432 B CN108205432 B CN 108205432B CN 201611165505 A CN201611165505 A CN 201611165505A CN 108205432 B CN108205432 B CN 108205432B
- Authority
- CN
- China
- Prior art keywords
- experimental data
- abnormal value
- abnormal
- value
- observed
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
- G06F7/22—Arrangements for sorting or merging computer data on continuous record carriers, e.g. tape, drum, disc
- G06F7/24—Sorting, i.e. extracting data from one or more carriers, rearranging the data in numerical or other ordered sequence, and rerecording the sorted data on the original carrier or on a different carrier or set of carriers sorting methods in general
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
Abstract
The invention provides a real-time elimination method for observation experiment data abnormal values, which comprises the following steps: counting the existing experimental data of a certain variable set X, and determining an abnormal value threshold; and detecting and removing the abnormal value of the experimental data in real time according to the abnormal value threshold and a given mathematical model for engineering. The invention determines the threshold of the abnormal value by counting the historical experimental data of the continuously changed variable to be observed, and then dynamically judges the current experimental data observed by the experimental system in real time by the dynamic processing mathematical model, so that the abnormal value can be judged and eliminated, and the invention is beneficial to more truly evaluating, predicting or restoring the real variable to be observed. The method provided by the invention is not limited by using conditions, has a wider application range, and provides an effective dynamic processing method for observing abnormal values in experimental data.
Description
Technical Field
The invention belongs to the technical field of experimental data analysis, relates to a real-time elimination method for observation experimental data abnormal values, and particularly relates to a real-time elimination method for abnormal values in observation experimental data with continuous change characteristics.
Background
Most of the existing experimental data abnormal value processing methods are static processing methods, for example, in statistical analysis of experimental data (military training teaching materials editorial working committee of general equipment of the national liberation force of China 2001, 1 month national defense industry publishing society ISBN 7-118-02370-1P 67-P77), abnormal values (or named field values and field points) are processed by using methods such as an extreme value deviation method and a range ratio method, and whether larger or smaller data in a group of experimental data obviously deviates from other experimental data is judged by sequencing the group of experimental data belonging to the same variable set. When the static methods are used for processing abnormal values, firstly, a variable set is required to be a certain fixed invariant, a group of experimental data is obtained through multiple observation of the invariant, and then abnormal value judgment and processing are carried out; if the variable set is a time sequence of changes, the static processing method cannot effectively judge abnormal values in the variable set; in addition, the experimental data judged when the abnormal value processing is performed by the static processing method is not the current observed value of the experimental system, and the real-time performance is poor.
A dynamic abnormal value processing method is proposed in individual publications, for example, in optimal estimation and application (Jade, Juzheng Takou, publication of Jumbo, 12 month Union letters, 15031.609P 280-P312, 1984), an abnormal value processing method for real-time discrimination of observed quantity by using forecast residual error statistic in Kalman filtering is a dynamic processing method, which can remove observed experimental data in real time, but can only be used in a Kalman filtering mathematical model, and the application range is strictly limited.
The invention mainly solves the problems of detection and elimination of abnormal values in observation experiment data, can detect and eliminate any data to be observed with continuous change characteristics in real time on site after the method is adopted, effectively avoids the interference of the experiment abnormal values on the observation results, and ensures that the results of parameter estimation by referring to the observation data are more real and accurate.
It should be noted that the method is also applicable to the real-time elimination of the abnormal values of the observation experiment data of the invariant, but the method is not applicable to the real-time elimination of the abnormal values of the observation experiment data of the discontinuous change (such as step change).
Disclosure of Invention
The invention aims to provide a real-time elimination method for observing abnormal values in experimental data, which solves the problems that a static processing method cannot judge the experimental data in real time and the use limitation of the currently known real-time elimination method.
The technical scheme of the invention comprises the following steps:
a real-time elimination method for observation experiment data abnormal values comprises the following steps:
the method comprises the following steps: and (3) counting the existing experimental data (namely historical experimental data) of a certain variable set X, and determining an abnormal value threshold. The method comprises the following steps:
1) with { xii 1, 2.. k represents experimental data of a set of variables X obtained in a previous experiment, and may be regarded as a time series, and Y representsi：
Yi＝|xi+1-xi| (1)
Representing a sequence xiThe absolute values of the differences between the front and rear points are arranged in the order from small to large, i.e.
Y(1)≤Y(2)≤...≤Y(m)≤...≤Y(k-1) (2)
2) And determining the abnormal value detection level sigma, wherein the sigma can take a value within a range of 1-10 per mill according to the requirement of an experimental system, and the smaller the sigma is, the lower the detection level of the abnormal value in the X experimental data of the corresponding variable set is, otherwise, the higher the detection level is. For a general system, the detection level of the abnormal value can be 3 per thousand.
3) Calculating the ordering of the abnormal value threshold, wherein the formula is as follows:
b＝σ×(k-1) (3)
the parameter b in the above equation represents the number of abnormal values in the k pieces of historical experimental data participating in statistics of the set of variables X, calculated according to the detection level σ. Since the result of the above formula cannot be guaranteed to be an integer exactly, if m is a minimum positive integer greater than or equal to "2 × b", m is the arrangement order of the outlier threshold in the formula (2). To ensure that m is greater than 0, k should be large enough, i.e., enough historical experimental data to participate in the statistics.
4) And (3) determining an abnormal value threshold, obtaining an abnormal value threshold value Y (m) corresponding to the historical experimental data of the variable set X according to the arrangement sequence of the abnormal value thresholds by the formula (2), and regarding the Y (m) as the abnormal value threshold of the variable set X.
Step two: and (3) detecting and removing the abnormal value of the experimental data in real time according to the abnormal value threshold and a given mathematical model for engineering, wherein the method comprises the following steps:
1) setting the time sequence of real-time experimental data obtained by the same variable set in the corresponding step one as { x }ii＝1，2，......n}，xnFor the experimental data observed at time n,
is to xnAnd (3) recording and processing the condition that the abnormal value continuously appears in the experimental data according to a formula (4) according to the result obtained by the abnormal value detection and elimination processing, wherein the formula is as follows:
where a (n) is the raw experimental data { X) for the set of variables XiThe number of outliers that appear in succession. The initial value of a (n) is 0, i.e. representing the first observed dataConsidered to be normal; for subsequent observed data, if the absolute value of the difference between the subsequent observed data and the previous abnormal value detection and removal processing result is smaller than an abnormal value threshold Y (m), the observed experimental data of the current observed point is judged to be a normal value, the number a (n) of the corresponding abnormal values is 0, otherwise, the a (n) is accumulated to be 1;
2) calculating an abnormal value detection processing result of experimental data observed in real time according to the following formula:
that is, x is determined according to whether a (n) is 0 or notnWhether or not it is an abnormal value, if xnIf the value is abnormal, x is addednRemoving abnormal value at n-1 time, and processing the result
As observed values of experimental data at time n; if xnIf the value is normal, the processing such as elimination is not performed.
And through the first step and the second step, the abnormal values in the observed experimental data can be eliminated in real time.
Compared with the prior art, the invention has the beneficial effects that:
the invention determines the threshold of the abnormal value by counting the historical experimental data of the continuously changed variable to be observed, and then dynamically judges the current experimental data observed by the experimental system in real time by the dynamic processing mathematical model, so that the abnormal value can be judged and eliminated, and the invention is beneficial to more truly evaluating, predicting or restoring the real variable to be observed. The method is not limited by using conditions, has wider application range, and provides an effective dynamic processing method for observing abnormal values in experimental data.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention. It is obvious that the drawings in the following description are only some embodiments of the invention, and that for a person skilled in the art, other drawings can be derived from them without inventive effort.
FIG. 1: is a general flow chart of the invention;
FIG. 2: a scatter plot of a set of experimental data according to the present invention.
Detailed Description
The preferred embodiment of the present invention will be further described with reference to fig. 1.
As shown in fig. 1: the technical scheme of the invention comprises the following steps:
a real-time elimination method for observation experiment data abnormal values comprises the following steps:
the method comprises the following steps: and (3) counting the existing experimental data (namely historical experimental data) of a certain variable set X, and determining an abnormal value threshold. The method comprises the following steps:
1) with { xii 1, 2.. k represents experimental data of a set of variables X obtained in a previous experiment, and may be regarded as a time series, and Y representsi：
Yi＝|xi+1-xi| (1)
Representing a sequence xiThe absolute values of the differences between the front and rear points are arranged in the order from small to large, i.e.
Y(1)≤Y(2)≤...≤Y(m)≤...≤Y(k-1) (2)
2) And determining the abnormal value detection level sigma, wherein the sigma can take a value within a range of 1-10 per mill according to the requirement of an experimental system, and the smaller the sigma is, the lower the detection level of the abnormal value in the X experimental data of the corresponding variable set is, otherwise, the higher the detection level is. For a general system, the detection level of the abnormal value can be 3 per thousand.
3) Calculating the ordering of the abnormal value threshold, wherein the formula is as follows:
b＝σ×(k-1) (3)
the parameter b in the above equation represents the number of abnormal values in the k pieces of historical experimental data participating in statistics of the set of variables X, calculated according to the detection level σ. Since the result of the above formula cannot be guaranteed to be an integer exactly, if m is a minimum positive integer greater than or equal to "2 × b", m is the arrangement order of the outlier threshold in the formula (2). To ensure that m is greater than 0, k should be large enough, i.e., enough historical experimental data to participate in the statistics.
4) And (3) determining an abnormal value threshold, obtaining an abnormal value threshold value Y (m) corresponding to the historical experimental data of the variable set X according to the arrangement sequence of the abnormal value thresholds by the formula (2), and regarding the Y (m) as the abnormal value threshold of the variable set X.
Step two: and (3) detecting and removing the abnormal value of the experimental data in real time according to the abnormal value threshold and a given mathematical model for engineering, wherein the method comprises the following steps:
1) setting the time sequence of real-time experimental data obtained by the same variable set in the corresponding step one as { x }ii＝1，2，......n}，xnFor the experimental data observed at time n,
is to xnAnd (3) recording and processing the condition that the abnormal value continuously appears in the experimental data according to a formula (4) according to the result obtained by the abnormal value detection and elimination processing, wherein the formula is as follows:
where a (n) is the raw experimental data { X) for the set of variables XiThe number of outliers that appear in succession. The initial value of a (n) is 0, namely the first observed data is considered as a normal value; for subsequent observed data, if the absolute value of the difference between the subsequent observed data and the previous abnormal value detection and removal processing result is smaller than an abnormal value threshold Y (m), the observed experimental data of the current observed point is judged to be a normal value, the number a (n) of the corresponding abnormal values is 0, otherwise, the a (n) is accumulated to be 1; in addition, considering the possibility of occurrence of slice (continuous) abnormal values in the engineering, when a (n) is more than or equal to a certain upper limit, the abnormal value detection and elimination are carried out on the observation experiment data again, namely when a (n) is more than or equal to f, the range of f is 2-5, and the abnormal value detection and elimination are carried out on the observation experiment data again. The upper limit of the number of continuously occurring outliers can be generally selected from a smaller integer (e.g., 2-5), which is the case in this embodimentThe upper limit of (a), (n) in (1) is 3.
2) Calculating an abnormal value detection processing result of experimental data observed in real time according to the following formula:
that is, x is determined according to whether a (n) is 0 or notnWhether or not it is an abnormal value, if xnIf the value is abnormal, x is addednRemoving abnormal value at n-1 time, and processing the result
As observed values of experimental data at time n; if xnIf the value is normal, the processing such as elimination is not performed.
And through the first step and the second step, the abnormal values in the observed experimental data can be eliminated in real time.
To further explain the invention, a specific example is now given:
firstly, counting historical experimental data of a variable set X, taking an absolute value of a difference between the previous experimental data and the next experimental data, and arranging results in a sequence from small to large; then determining an abnormal value detection level, such as 3 ‰; then, calculating the arrangement sequence of the abnormal value threshold value in the sequence statistic according to the abnormal value detection level; the outlier threshold value y (m) for the set of variables X can thus be determined.
After the outlier threshold is determined, any set of experimental data { X ] of the variable set X can be subjected toiCarry out dynamic abnormal value processing, firstly, if the experimental data xiFor the first observation point of this experiment, i is 1, x is satisfiediJudging as a normal value; if xiIf the current point is not the first observation point, judging whether the absolute value of the difference between the currently observed experimental data and the experimental data processed by the abnormal value at the previous point is smaller than the threshold value of the abnormal value, and if so, judging the current point as a normal value; otherwise, judging whether the number of the abnormal values continuously appearing at present exceeds 3, if so, judging the current point as a normal value, otherwise, judging the current point as an abnormal value; normal value is not processedAnd removing the abnormal values, replacing the abnormal values of the current point with the experimental data processed by the previous abnormal value, and taking the abnormal values as the observed values of the current point.
The detailed technical scheme is as follows:
firstly, counting historical experimental data to obtain an abnormal value threshold:
the number of data points of the variable set X in the historical experiment is as follows: 297, point 085;
determining the abnormal value detection level to be 3 per mill;
the number of theoretical outliers can be calculated: (297, 085-1). times.3 ‰ 892 point;
outlier threshold:
wherein
Is the expression "| Xi-Xi-1The order statistic of | ", the subscript Max indicates that the expression is in ascending order, and the expression form here is consistent with the expression of formula (2). It should be noted that in the method for dynamically processing an abnormal value according to the present invention, the abnormal value has bilateral characteristics, that is, the abnormal value point is an abnormal value with respect to both the previous point and the subsequent point thereof, and therefore, when the abnormal value detection level is set to 3%, the sequence { | X needs to be actually seti-Xi-1The large value of 6 ‰ in | }, i.e. corresponding to 892 × 2 point, in this embodiment, since there are many historical experimental data, the statistical work of the abnormal value threshold can be realized by simple computer software, and finally the abnormal value threshold is obtained to be 2.1962.
Then, the dynamic abnormal value processing is carried out on the experimental data which belongs to the variable set X and is observed in real time, and the specific steps are as follows:
as shown in fig. 2, the existing set of real-time experimental data includes 10 observation data {12.4309, 11.5780, 12.4522, 11.5780, 12.1537, 11.4501, 10.9170, 16.3969, 11.5354, 11.2369 }.
According to the dynamic processing mathematical model:
for x112.4309, has a (1) ═ a0；
For x2～x7And (3) calculating according to the formula (3) and the formula (4):
a(k)＝0，
k＝2，......，7；
for x816.3969 because
Thus determining x8The point is an abnormal value point, and accordingly:
a(8)＝1；
for x911.5354, having a (9) 0;
for x1011.2369, having a (10) 0;
in the present embodiment, only x8The points are abnormal value points, and the rest are normal observation points.
The invention has not been described in detail and is in part known to those of skill in the art.
Claims (1)
1. A real-time elimination method for observation experiment data abnormal values is characterized by comprising the following steps:
the method comprises the following steps: counting the existing experimental data of a certain variable set X, and determining an abnormal value threshold, wherein the method comprises the following specific steps:
step 101, with { xii 1, 2.. k } is indicated in the specificationThe experimental data obtained in the experiment for the set of variables X can be regarded as a time series, and the sequence { X is expressed by formula (1)iThe absolute value of the difference between the front point and the rear point:
Yi＝|xi+1-xi|； (1)
then arranged in the order of small to large, i.e.
Y(1)≤Y(2)≤...≤Y(m)≤...≤Y(k-1)； (2)
Step 102, determining an abnormal value detection level sigma, wherein the value range of sigma is 1-10 per mill;
step 103, calculating the order of the outlier threshold using equation (3):
b＝σ×(k-1) (3)
the parameter b in the above formula represents the number of abnormal values in k pieces of historical experimental data participating in statistics of the variable set X calculated according to the detection level sigma, and if m is a minimum positive integer greater than or equal to "2 × b", m is the arrangement sequence of the abnormal value threshold in the formula (2);
step 104, determining an abnormal value threshold, obtaining an abnormal value threshold value Y (m) corresponding to the historical experimental data of the variable set X according to the arrangement sequence of the abnormal value threshold by the formula (2), and regarding the Y (m) as the abnormal value threshold of the variable set X;
step two: detecting and removing the abnormal value of the experimental data in real time according to a given mathematical model for engineering according to the abnormal value threshold, and the steps are as follows:
step 201, setting the time sequence of the real-time experimental data obtained by the same variable set in the step one as { x }ii＝1，2，......n}，xnFor the experimental data observed at time n,
is to xnAnd (3) recording and processing the condition that the abnormal value continuously appears in the experimental data according to a formula (4) if the result obtained by the abnormal value detection and elimination processing is obtained:
in the formulaa (n) is the raw experimental data { X) for the set of variables XiThe number of outliers appearing in succession, the initial value of a (n) is 0, i.e. it represents the first observation considered as a normal value; for subsequent observed data, if the absolute value of the difference between the subsequent observed data and the previous abnormal value detection and removal processing result is smaller than an abnormal value threshold Y (m), the observed experimental data of the current observed point is judged to be a normal value, the number a (n) of the corresponding abnormal values is 0, otherwise, the a (n) is accumulated to be 1;
step 202, determining an abnormal value detection processing result of the experimental data observed in real time according to the formula (5):
when a (n) is not 0, x is judgednIs an abnormal value, if xnIf the value is abnormal, x is addednRemoving abnormal value at n-1 time, and processing the result
As observed values of experimental data at time n; when a (n) is 0, it is determined that x is not zeronIf the value is a normal value, the elimination process is not performed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611165505.9A CN108205432B (en) | 2016-12-16 | 2016-12-16 | Real-time elimination method for observation experiment data abnormal value |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611165505.9A CN108205432B (en) | 2016-12-16 | 2016-12-16 | Real-time elimination method for observation experiment data abnormal value |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108205432A CN108205432A (en) | 2018-06-26 |
| CN108205432B true CN108205432B (en) | 2020-08-21 |
Family
ID=62601395
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611165505.9A Active CN108205432B (en) | 2016-12-16 | 2016-12-16 | Real-time elimination method for observation experiment data abnormal value |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108205432B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109884735B (en) * | 2019-03-27 | 2019-11-19 | 山东省气象局大气探测技术保障中心 | A method of precipitation is observed based on rainy quantity sensor |
| CN110906929B (en) * | 2019-12-12 | 2022-01-25 | 北京中科宇航探索技术有限公司 | Satellite abnormal data identification method and device |
| CN111579751A (en) * | 2020-05-08 | 2020-08-25 | 广东农工商职业技术学院(农业部华南农垦干部培训中心) | High-precision soil sensor |
| CN112965966B (en) * | 2021-02-08 | 2023-09-08 | 北京军懋国兴科技股份有限公司 | Rapid preprocessing method and system based on actually measured flight parameter data and computer related product |
| CN113157684B (en) * | 2021-05-13 | 2021-12-10 | 中山大学 | Water conservancy mass data error checking method |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101354316A (en) * | 2007-07-27 | 2009-01-28 | 通用电气公司 | Fleet anomaly detection method |
| US20120134576A1 (en) * | 2010-11-26 | 2012-05-31 | Sharma Avinash | Automatic recognition of images |
| CN102945320A (en) * | 2012-10-29 | 2013-02-27 | 河海大学 | Time series data abnormity detection method and device |
| CN104537034A (en) * | 2014-12-22 | 2015-04-22 | 国家电网公司 | Electric-transmission-and-transformation-equipment state monitoring data cleaning method based on time series analysis |
| CN104902509A (en) * | 2015-05-19 | 2015-09-09 | 浙江农林大学 | Abnormal data detection method based on top-k(sigma) algorithm |
| CN105259136A (en) * | 2015-11-24 | 2016-01-20 | 江南大学 | Measuring-point-free temperature correction method of near-infrared correction model |
| CN106096311A (en) * | 2016-06-24 | 2016-11-09 | 南京信息工程大学 | A kind of ionosphere VTEC value method for detecting abnormality |
-
2016
- 2016-12-16 CN CN201611165505.9A patent/CN108205432B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101354316A (en) * | 2007-07-27 | 2009-01-28 | 通用电气公司 | Fleet anomaly detection method |
| US20120134576A1 (en) * | 2010-11-26 | 2012-05-31 | Sharma Avinash | Automatic recognition of images |
| CN102945320A (en) * | 2012-10-29 | 2013-02-27 | 河海大学 | Time series data abnormity detection method and device |
| CN104537034A (en) * | 2014-12-22 | 2015-04-22 | 国家电网公司 | Electric-transmission-and-transformation-equipment state monitoring data cleaning method based on time series analysis |
| CN104902509A (en) * | 2015-05-19 | 2015-09-09 | 浙江农林大学 | Abnormal data detection method based on top-k(sigma) algorithm |
| CN105259136A (en) * | 2015-11-24 | 2016-01-20 | 江南大学 | Measuring-point-free temperature correction method of near-infrared correction model |
| CN106096311A (en) * | 2016-06-24 | 2016-11-09 | 南京信息工程大学 | A kind of ionosphere VTEC value method for detecting abnormality |
Non-Patent Citations (1)
| Title |
|---|
| "基于模型的过程工业时间序列异常值检测方法";苏卫星,朱云龙,胡琨元,刘芳;《仪器仪表学报》;20120915(第201209期);2080-2087 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108205432A (en) | 2018-06-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108205432B (en) | Real-time elimination method for observation experiment data abnormal value | |
| CN105279365B (en) | For the method for the sample for learning abnormality detection | |
| US11003738B2 (en) | Dynamically non-gaussian anomaly identification method for structural monitoring data | |
| Yang et al. | A novel CUSUM median control chart | |
| CN108830006B (en) | Linear-nonlinear industrial process fault detection method based on linear evaluation factor | |
| CN113449262A (en) | Data change trend judgment method and device | |
| CN107909216A (en) | A kind of part actual production period forecasting method | |
| CN112817955B (en) | Regression model-based data cleaning method | |
| Liu et al. | Gaussian noise level estimation in SVD domain for images | |
| CN108508860B (en) | Process industrial production system data monitoring method based on coupling relation | |
| CN103777519B (en) | A kind of production process method of quality control based on self starting technologies | |
| CN107154947B (en) | Network flow abnormity detection and classification method based on effective frequent flow characteristics | |
| Katemee et al. | Control charts for zero-inflated Poisson models | |
| CN107784015B (en) | Data reduction method based on online historical data of power system | |
| CN115099490A (en) | Yarn quality prediction method and related device | |
| CN1216338C (en) | Data merging method based linear constrainted cut minimum binary multiply | |
| US20210144171A1 (en) | A Method of Digital Signal Feature Extraction Comprising Multiscale Analysis | |
| CN105938623B (en) | A kind of Real-time Two-dimensional method for tracking target based on Two-way Feedback particle filter algorithm | |
| Buch | Decision based non-linear filtering using interquartile range estimator for Gaussian signals | |
| Chang | Comparison of two sampling intervals and three sampling intervals VSI charts for monitoring both means and variances | |
| Berset et al. | The effect of the initial network configuration on preferential attachment | |
| US20220027248A1 (en) | Automatic Window Generation for Process Trace | |
| JP4354496B2 (en) | Traffic leap detection device and traffic leap detection method | |
| CN107766290A (en) | Convergent multiple regression engineering statistics new method | |
| CN117708645A (en) | Industrial data stream concept drift detection method based on singular spectrum analysis |
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 |