Method and apparatus for anomaly detection on a doctor blade of a papermaking machine, and computing device

Patent metadata

Field	Value
Publication number	US-12371857-B2
Application number	US-202217986240-A
Country	US
Kind code	B2
Filing date	Nov 14, 2022
Priority date	Dec 23, 2021
Publication date	Jul 29, 2025
Grant date	Jul 29, 2025

Abstract

A method for anomaly detection on a doctor blade of a papermaking machine includes obtaining doctor blade-related data, wherein the doctor blade-related data includes working condition data, status monitoring data, and design parameter data of the doctor blade, performing a predetermined data pre-processing on the doctor blade-related data to obtain a pre-processed data set, wherein the pre-processed data set includes processed data respectively corresponding to the working condition data, the status monitoring data, and the design parameter data; performing a predetermined feature extraction on the pre-processed data set, and fusing extracted features to obtain fused feature data; and analyzing the fused feature data based on different anomaly detection processes respectively, to perform a comprehensive anomaly detection on the doctor blade of the papermaking machine.

First claim

What is claimed is:

1. A method for anomaly detection on a doctor blade currently in use by a papermaking machine, comprising:
obtaining doctor blade-related data, wherein the doctor blade-related data includes real-time working condition data, real-time status monitoring data, and design parameter data of the doctor blade currently in use by the papermaking machine;
performing data pre-processing on the doctor blade-related data in a predetermined pre-processing manner to obtain a pre-processed data set, wherein the pre-processed data set includes processed data respectively corresponding to the real-time working condition data, the status monitoring data, and the design parameter data;
performing feature extraction on the pre-processed data set in a predetermined feature extraction manner, and fusing extracted features to obtain fused feature data; and
analyzing the fused feature data based on different anomaly detection processes respectively, to perform a comprehensive anomaly detection on the doctor blade currently in use by the papermaking machine.

2. The method according to claim 1 , wherein the real-time working condition data includes Yankee cylinder rotational speed data, coating material type data, and pulp raw material type data; the real-time status monitoring data includes acceleration data of real-time vibration of bearings at a driving end and a non-driving end of a fixing bracket for the doctor blade and temperature data; and the design parameter data includes doctor blade material type data,
wherein performing data pre-processing on the doctor blade-related data in a predetermined pre-processing manner to obtain a pre-processed data set comprises:
performing at least one of the following pre-processing processes based on different types of data of the doctor blade-related data: a data deduplication process, a data denoising process, a data encoding process, and a data filtering process.

3. The method according to claim 1 , wherein performing feature extraction on the pre-processed data set in a predetermined feature extraction manner and fusing extracted features to obtain fused feature data comprises: for a current detection occasion,
extracting a data description feature corresponding to each feature parameter related to the real-time working condition data in the pre-processed data set within a preset time period related to the detection occasion, as a working condition feature;
extracting a data description feature corresponding to each feature parameter related to the status monitoring data in the pre-processed data set within the preset time period, as a status monitoring feature;
extracting a data description feature corresponding to each feature parameter related to the design parameter data in the pre-processed data set within the preset time period, as a design parameter feature; and
obtaining the fused feature data through a feature fusion processing and based on the data description features corresponding to each feature parameter related to the real-time working condition data, the status monitoring data, and the design parameter data.

4. The method according to claim 3 , further comprising:
assigning different weights to the working condition feature, the status monitoring feature and the design parameter feature according to corresponding sensitivity for characterization of a status of the doctor blade, and assigning, for at least one feature type of the working condition feature, the status monitoring feature and the design parameter feature, different weights to each data description feature under each feature type,
wherein obtaining the fused feature data through a feature fusion processing and based on the data description features corresponding to each feature parameter related to the real-time working condition data, the real-time status monitoring data, and the design parameter data comprises:
multiplying a feature value of the data description feature corresponding to each feature parameter with a corresponding weight to obtain a weighted feature value of the data description feature of the feature parameter; and
performing synchronous concatenation fusion for each weighted feature value, to obtain the fused feature data.

5. The method according to claim 1 , wherein analyzing the fused feature data based on different anomaly detection processes respectively to perform comprehensive anomaly detection on the doctor blade of the papermaking machine comprises:
analyzing the fused feature data based on different anomaly detection models respectively, to obtain detection results of all anomaly detection models; and
fusing the detection results of all anomaly detection models to obtain an anomaly detection indicator of the doctor blade of the papermaking machine.

6. The method according to claim 5 , wherein each anomaly detection model is pre-trained or modeled based on a training sample set, wherein the training sample set is obtained from historical working condition data, historical status monitoring data, and historical design parameter data, and
wherein the anomaly detection model includes a local outlier factor detection model, an MSET algorithm model, an isolation forest algorithm model or a support vector machine model, and wherein the detection results of all anomaly detection models are weighted to obtain the anomaly detection indicator of the doctor blade of the papermaking machine.

7. The method according to claim 6 , further comprising:
performing a remaining useful life prediction on the doctor blade based on the anomaly detection indicator of the doctor blade of the papermaking machine and by using different remaining useful life prediction models, and obtaining prediction results for each remaining useful life prediction model; and
fusing the prediction results for all remaining useful life prediction models to obtain a predicted remaining useful life of the doctor blade of the papermaking machine.

8. The method according to claim 7 , wherein the remaining useful life prediction models include a theoretical degradation model and an anomaly loss model,
wherein fusing the prediction results for all remaining useful life prediction models to obtain a predicted remaining useful life of the doctor blade of the papermaking machine comprises:
predicting by using the theoretical degradation model to obtain a theoretical remaining useful life at current detection occasion;
predicting by using the anomaly loss model based on the anomaly detection indicator at the current detection occasion to obtain a loss-based remaining useful life at the current detection occasion; and
obtaining the predicted remaining useful life based on the theoretical remaining useful life and the loss-based remaining useful life.

9. A computing device, comprising:
a computer processor; and
a memory having stored thereon a computer program which, when executed, causes the processor to implement respective steps of the method for anomaly detection on a doctor blade according to claim 1 .

10. A non-transient computer-readable storage medium having computer-readable instructions stored thereon, wherein when the instructions are executed by a computer, the method of claim 1 is performed.

11. An apparatus for anomaly detection on a doctor blade of an operating papermaking machine, comprising a computer processor configured to:
obtain doctor blade-related data of a doctor blade currently in use by the papermaking machine, wherein the doctor blade-related data includes real-time working condition data, real-time status monitoring data, and design parameter data of the doctor blade;
perform data pre-processing on the doctor blade-related data in a predetermined

Assignees

• Skf Ab

Inventors

• Zhang Kaihuan
• Ding Liang
• Cheng Gang

Classifications

• G05B23/0221

  Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods · CPC title

• G05B23/024 Primary

  Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks · CPC title

• G05B23/0283

  Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL] (preventive maintenance, i.e. planning maintenance according to the available resources without monitoring the system G06Q10/06) · CPC title

• D21G9/0054

  details of algorithms or programs · CPC title

• Y02P90/30

  Computing systems specially adapted for manufacturing · CPC title