CN108158028A - Cigarette cut tobacco process multistage distributed monitoring and diagnostic method based on tile and hierarchy thought - Google Patents
Cigarette cut tobacco process multistage distributed monitoring and diagnostic method based on tile and hierarchy thought Download PDFInfo
- Publication number
- CN108158028A CN108158028A CN201711472774.4A CN201711472774A CN108158028A CN 108158028 A CN108158028 A CN 108158028A CN 201711472774 A CN201711472774 A CN 201711472774A CN 108158028 A CN108158028 A CN 108158028A
- Authority
- CN
- China
- Prior art keywords
- monitoring
- data
- sirox
- kld
- stage
- 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
- 238000012544 monitoring process Methods 0.000 title claims abstract description 118
- 241000208125 Nicotiana Species 0.000 title claims abstract description 87
- 235000002637 Nicotiana tabacum Nutrition 0.000 title claims abstract description 87
- 238000000034 method Methods 0.000 title claims abstract description 64
- 230000008569 process Effects 0.000 title claims abstract description 40
- 235000019504 cigarettes Nutrition 0.000 title claims abstract description 29
- 238000002405 diagnostic procedure Methods 0.000 title abstract 2
- 238000004519 manufacturing process Methods 0.000 claims abstract description 70
- 230000002159 abnormal effect Effects 0.000 claims abstract description 35
- 238000003745 diagnosis Methods 0.000 claims abstract description 15
- 239000011159 matrix material Substances 0.000 claims description 30
- 238000010438 heat treatment Methods 0.000 claims description 25
- 238000001035 drying Methods 0.000 claims description 19
- 238000005070 sampling Methods 0.000 claims description 12
- 238000009826 distribution Methods 0.000 claims description 10
- 230000000903 blocking effect Effects 0.000 claims description 7
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 239000000463 material Substances 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 3
- 230000000087 stabilizing effect Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 7
- 238000010792 warming Methods 0.000 abstract 1
- 230000005856 abnormality Effects 0.000 description 16
- 238000000513 principal component analysis Methods 0.000 description 8
- 238000012360 testing method Methods 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 238000010923 batch production Methods 0.000 description 3
- 238000001816 cooling Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000007704 transition Effects 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000007599 discharging Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 229920006395 saturated elastomer Polymers 0.000 description 2
- 238000003070 Statistical process control Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012847 principal component analysis method Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000007921 spray Substances 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
- 238000009736 wetting Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A24—TOBACCO; CIGARS; CIGARETTES; SIMULATED SMOKING DEVICES; SMOKERS' REQUISITES
- A24B—MANUFACTURE OR PREPARATION OF TOBACCO FOR SMOKING OR CHEWING; TOBACCO; SNUFF
- A24B7/00—Cutting tobacco
-
- A—HUMAN NECESSITIES
- A24—TOBACCO; CIGARS; CIGARETTES; SIMULATED SMOKING DEVICES; SMOKERS' REQUISITES
- A24B—MANUFACTURE OR PREPARATION OF TOBACCO FOR SMOKING OR CHEWING; TOBACCO; SNUFF
- A24B3/00—Preparing tobacco in the factory
- A24B3/10—Roasting or cooling tobacco
-
- A—HUMAN NECESSITIES
- A24—TOBACCO; CIGARS; CIGARETTES; SIMULATED SMOKING DEVICES; SMOKERS' REQUISITES
- A24B—MANUFACTURE OR PREPARATION OF TOBACCO FOR SMOKING OR CHEWING; TOBACCO; SNUFF
- A24B9/00—Control of the moisture content of tobacco products, e.g. cigars, cigarettes, pipe tobacco
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
- G07C3/00—Registering or indicating the condition or the working of machines or other apparatus, other than vehicles
- G07C3/005—Registering or indicating the condition or the working of machines or other apparatus, other than vehicles during manufacturing process
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- Manufacturing & Machinery (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Evolutionary Computation (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Manufacture Of Tobacco Products (AREA)
Abstract
The present invention discloses a kind of cigarette cut tobacco process multistage distributed monitoring and diagnostic method based on tile and hierarchy thought.Present invention identification cigarette cut tobacco process standby operation phase and the different monitoring requirements in steady production stage, using lateral equipment piecemeal, the thought of longitudinal time layering, build the multistage distributed monitoring model of cigarette cut tobacco process Sirox warming and humidifyings machine and KLD thin-plate cut-tobacco drier key equipments, realize that cigarette cut tobacco process is single and concurrent abnormal effective detection, and using drawing method is contributed to accurately identify abnormal cause.This invention removes the monitoring blind areas of cigarette cut tobacco process standby operation phase, improve the capturing ability of the concurrent exception information of process, are capable of the single and concurrent exception of accurate and effective monitoring different operation phase with diagnosis cigarette cut tobacco process key equipment.
Description
Technical Field
The invention relates to the field of monitoring of quality of products in process and equipment integrity in a cigarette tobacco shred making process, in particular to a multi-stage distributed monitoring and diagnosis method for the cigarette tobacco shred making process based on a blocking and layering thought.
Background
With the continuous improvement of the whole strength of the tobacco industry in China, the intelligent level and the efficient operation capability of effective improvement equipment become the focus of attention of cigarette factories, and state monitoring and abnormal diagnosis are important means for improving the intelligent and efficient operation of the equipment. At present, a single-variable statistical process control method is mainly used for monitoring the state and diagnosing the abnormality of equipment in the cigarette tobacco making process, and a rainbow diagram and a process capability index are used for analyzing steady-state data in a production stage. Such methods lack effective analysis of the dynamic information on the batch axis and also fail to reflect changes in the correlation between process variables. Wangwei and the like propose a multimode cigarette cut tobacco section fault monitoring method based on relative change Analysis, which integrates a cut tobacco section Sirox heating and humidifying machine and a KLD thin plate cut tobacco drying machine, and establishes Principal Component Analysis (PCA) models for three-dimensional data of different production brands for equipment state monitoring. The method aims at the steady-state data analysis in the production stage, the equipment state monitoring in the standby operation stage before feeding is not considered, and a monitoring blind zone exists in the standby operation stage; meanwhile, as the Sirox and the KLD are modeled and monitored as a whole, after the model alarms due to the abnormality of one device, the state monitoring of the other device is interfered, and the capturing capability of the method for the concurrent abnormal information in the cut tobacco section batch making process is limited. Therefore, if the information of two aspects of different equipment units and different production stages can be considered at the same time, and a multi-stage distributed monitoring model is established, the monitoring blind area of the standby operation stage can be eliminated, and the capturing capability of the concurrent abnormal information is improved.
Disclosure of Invention
Aiming at the monitoring blind area of the standby operation stage in the cigarette tobacco making section batch process and the mutual interference of the concurrency abnormality of different equipment to the overall monitoring model, the multi-stage distributed monitoring and diagnosis method based on the blocking and layering thought is provided, the capture capability of the monitoring model to the concurrency abnormality information of the tobacco making process is improved, the single or concurrency abnormality of the equipment in different stages is effectively monitored, and the distributed monitoring and abnormality diagnosis of the standby operation stage and the stable production stage of the tobacco making process is realized on the basis of the existing monitoring variables.
The invention adopts the following specific technical scheme:
a multi-stage distributed monitoring and diagnosis method for a cigarette tobacco shred making process based on a blocking and layering thought comprises the following steps:
the first step is as follows: according to existing measuring points and equipment monitoring requirements of a production site, respectively determining monitoring variables of a Sirox heating and humidifying machine and a KLD thin plate cut-tobacco drying machine in a standby operation stage and a stable production stage, wherein the monitoring variables of the Sirox heating and humidifying machine and the KLD thin plate cut-tobacco drying machine in the standby operation stage are respectively serial numbers 1-5 and 1-14, and the monitoring variables of the Sirox heating and humidifying machine and the KLD thin plate cut-tobacco drying machine in the stable production stage are respectively serial numbers 1-8 and 1-17;
TABLE 1 monitoring variables of Sirox heating humidifier, KLD thin plate cut tobacco dryer in standby operation stage and stable production stage
The second step is that: respectively determining validity judgment rules of offline modeling data of a Sirox heating and humidifying machine and a KLD sheet cut-tobacco drying machine in a standby operation stage and a stable production stage according to production process requirements; rejecting the broken material batch and abnormal batch data of the cut tobacco grade in the cut tobacco making centralized control system, and obtaining the original three-dimensional modeling data of the standby operation stage and the stable production stage of the Sirox heating humidifier and the KLD thin plate cut tobacco drying machine according to the modeling data validity judgment rule of table 2 Wherein I represents a batch, J represents a monitoring variable, K represents a sampling point, a subscript SS represents a stand-by operation stage of a Sirox temperature-increasing humidifier, KS represents a stand-by operation stage of a KLD sheet tobacco dryer, SP represents a stable operation stage of the Sirox temperature-increasing humidifier, KP represents a stable operation stage of the KLD sheet tobacco dryer, and the sampling frequency of the monitoring variable is 10 s/time;
TABLE 2 validity judgment rules for offline modeling data
The third step: three-dimensional data by adopting variable expansion modeExpansion into two-dimensional dataWherein, SS or KS or SP or KP is taken as the index, and is used for distinguishing the standby operation stage and the stable production stage of a Sirox heating humidifier and a KLD thin plate cut tobacco drying machine; for two-dimensional matrixEach variable inPerforming an average value reductionExcept for standard deviationTo obtain high-quality two-dimensional modeling data X by data standardization preprocessing*(I*K*×J*)；
The fourth step: modeling data X for two dimensions*PCA decomposition is carried out, and monitoring dies of a Sirox heating humidifier, a KLD thin plate cut tobacco drying machine standby operation stage and a stable production stage are respectively establishedType, obtaining the number of principal elements A of each model*A load matrix P*And diagonal matrix S*(ii) a Based on modeling data for each modelAnd SPE*Statistics, using F distributions and χ2Distributing to obtain the statistic control limit of the model;
the formula for the PCA decomposition is:
wherein, T*In a principal component subspace (I)*K*×A*) A scoring matrix of dimensions; p*In a principal component subspace (J)*×A*) A load matrix of dimensions; e*In the residual subspace (I)*K*×J*) A residual matrix of dimensions; a. the*Representing the number of the principal elements, and determining by an accumulative contribution rate method; diagonal matrix S*＝diag(λ*1,λ*2,…,λ**A*) From modeling data X*Of the covariance matrix sigma*＝X* TX*/(I*K*-1) front A*Forming a characteristic value;
calculating from the F distributionMonitoring control limits of statisticsComprises the following steps:
where α is the confidence, Fα(A*,I*K*-A*) Is towith confidence of α and degree of freedom of A*、I*K*-A*F distribution threshold value of (a);
according to chi2Distributed computing SPE*Control limit ctr for monitoring statisticsSPE*Comprises the following steps:
wherein, g*＝v*/(2n*)，n*、v*Are respectively SPE*Monitoring the mean and variance of the statistics;
the fifth step: according to the state monitoring requirement, determining the validity judgment rule of the online monitoring data of the Sirox heating and humidifying machine and the KLD thin plate cut-tobacco drier shown in the table 3 to obtain the current valid data in the standby operation stage Or stabilizing currently valid data of production stageThe sampling frequency of the monitoring variable is 10 s/time;
TABLE 3 validity judgment rules for on-line monitoring data
And a sixth step: according to the current leaf shred grade and the equipment operation stage to which the current effective data belongs, the same grade corresponding stage Sirox temperature increasing and humidifying machine and the KLD thin plate tobacco dryer monitoring model are utilized to monitorValue ofExcept for standard deviationCarrying out standardized preprocessing on the data to obtain the current data x to be analyzedSSnew(1×JSS)、xKSnew(1×JKS) Or xSPnew(1×JSP)、xKPnew(1×JKP)；
The seventh step: projecting the current data to be analyzed to a Sirox heating and humidifying machine and a KLD thin plate cut-tobacco drier monitoring model at the corresponding stage of the same mark, and utilizing a load matrix P of the monitoring model*And diagonal matrix S*Calculating current data to be analysedAndstatistics; the calculation formula is as follows:
wherein,represents (1 XJ) obtained by reconstruction*) Estimating a vector in dimension;
eighth step: the introduction of an abnormal alarm, which is questioned by communication with process personnel, is defined as: at the end of the standby operation phase, advance T of 12 sets of raw data within 2 minutes2And SPE statistics exceed the control limit; t of 6 consecutive sets of raw data within 1 minute during the stationary production phase2And SPE statistics exceed the control limit; and at the moment of an abnormal alarm point, calculating the contribution rate of each monitoring variable to the overrun statistic by adopting a contribution graph method, and separating the cause variable causing the overrun of the statistic.
The invention has the beneficial effects that:
the multi-stage distributed monitoring model is constructed in a cigarette tobacco shred making process by transverse equipment blocking and longitudinal time layering, and the monitoring blind area of the existing monitoring method in the standby operation stage and the mutual interference of the concurrent abnormality of different equipment on the whole monitoring model are overcome. The invention fully utilizes the expert experience and production data of the cigarette tobacco shred making process, realizes the distributed monitoring and diagnosis of different production stages of key equipment units, greatly improves the monitoring performance of the cigarette tobacco shred making process, and is very favorable for the development and implementation of an intelligent analysis and abnormity diagnosis system of the cigarette tobacco shred making process.
Drawings
FIG. 1 is a diagram of a multi-stage distributed monitoring model structure for a cigarette cut tobacco making process based on a block layering idea.
FIG. 2 is a flow chart of on-line monitoring and abnormality diagnosis in the tobacco shred making process.
FIG. 3 is a monitoring diagram of normal batch test data in the cigarette cut tobacco making process.
FIG. 4 is a monitoring diagram of test data of an abnormal batch 1 in the tobacco shred manufacturing process of cigarettes.
FIG. 5 is a monitoring diagram of abnormal batch 2 test data in the tobacco shred manufacturing process.
FIG. 6 is a graph of the contribution of the Sirox monitor variable at the 143 th data point in exception batch 1.
FIG. 7 is a graph of the contribution of the Sirox monitor variable at the time of the 121 th data point in exception batch 2.
Fig. 8 is a graph of KLD monitor variable contribution at the 127 th data point time in exception batch 2.
Fig. 9 is a graph of KLD monitor variable contribution at the 209 th data point time in exception batch 2.
Detailed Description
In order to better understand the technical scheme of the invention, the following description is further provided for the embodiment of the invention with the accompanying drawings.
The invention relates to a key equipment unit for a cigarette leaf shred making process, which comprises the following steps: a Sirox heating and humidifying machine and a KLD thin plate cut tobacco drying machine. The Sirox temperature-increasing and humidity-increasing machine has the main functions of expanding and wetting cut tobacco before drying cut tobacco, improving the temperature and the water content of the cut tobacco and improving the filling value of the dried cut tobacco. The cut tobacco reaches the airflow cyclone separator through the charging chute and the impeller gate, the shaft of the airflow cyclone separator is hollow, and a threaded hole is formed in the shaft. Saturated steam flows out of holes in the hollow shaft and falls on the cut tobacco to expand and wet the cut tobacco, and the cut tobacco is conveyed to a discharging vibration groove conveyor with a cover plate by the airflow cyclone separator. The shaft of the cyclone separator is driven by a frequency converter to control the rotation speed of the shaft. The opening and closing of the steam spray pipeline are realized by adjusting valves, and the steam quantity is adjusted according to the operation requirement. The KLD sheet tobacco dryer mainly has the advantages that the saturated steam is used for heating the cylinder body to dry the cut tobacco, the sensory quality of the cut tobacco is improved, and the processing requirements of the subsequent procedures are met. The cut tobacco enters the cut tobacco dryer through an opening in the inlet cover plate by the vibrating trough conveyor, and the cut tobacco in the cut tobacco dryer is continuously conveyed by a rotating roller. The wall of the roller is provided with blades which can be heated by steam, and the blades drive the cut tobacco to rotate together when the roller rotates and convey the cut tobacco forwards through the slope of the roller. The rotation of the drum and its inclination cause the strands to be in constant contact with the blades inside the drum, during which the strands are stirred and heated uniformly. The temperature of the wall of the drum can be adjusted by steam pressure or steam flow, and the temperature of the hot air in the drum can be adjusted by the steam pressure of the heat exchanger.
The invention discloses a multi-stage distributed monitoring and diagnosis method for a cigarette cut tobacco making process based on a blocking and layering thought, which mainly comprises the following steps:
(1) multi-stage distributed monitoring modeling
The structure of a multi-stage distributed monitoring and diagnosis model for the tobacco shred making process of cigarettes based on the idea of blocking and layering is shown in figure 1. Taking a certain cut tobacco grade multi-stage distributed monitoring modeling as an example, aiming at the characteristics of multi-stage data, firstly respectively determining monitoring variables and a modeling data validity judgment rule of a standby operation stage and a stable production stage, obtaining modeling data of different stages of Sirox and KLD, and further respectively establishing PCA monitoring models of different stages of Sirox and KLD by utilizing a PCA method and obtaining a control limit.
The first step is as follows: according to existing measuring points and equipment monitoring requirements of a production field, monitoring variables of a Sirox heating and humidifying machine and a KLD sheet cut-tobacco drying machine in a standby operation stage and a stable production stage are respectively determined, as shown in table 1, the monitoring variable numbers of the Sirox and the KLD in the standby operation stage are respectively 5 and 14, and the monitoring variable numbers of the Sirox and the KLD in the stable production stage are respectively 8 and 17.
TABLE 1 monitoring variables of Sirox, KLD Standby run phase and Stable production phase
The second step is that: and respectively determining validity judgment rules of the offline modeling data of the Sriox and KLD standby operation stage and the stable production stage according to the production process requirements. Rejecting the broken material batch and abnormal batch data of the cut tobacco grade in the cut tobacco making centralized control system, and obtaining the original three-dimensional modeling data of the Sirox and KLD standby operation stage and the stable production stage according to the modeling data validity judgment rule of table 2Wherein I represents batch, J represents monitoring variable, K represents sampling point, subscript SS represents Sirox standby operation stage, KS represents KLD standby operation stage, SP represents Sirox stable operation stage, KP represents KLD stable operation stage, sampling frequency of monitoring variableThe rate was 10 s/time.
TABLE 2 validity judgment rules for offline modeling data
The third step: three-dimensional data by adopting variable expansion modeExpansion into two-dimensional dataWherein, SS or KS or SP or KP can be taken as the raw material for distinguishing the Sirox and KLD standby operation stage and the stable production stage. For two-dimensional matrixEach variable inPerforming an average value reductionExcept for standard deviationTo obtain high-quality two-dimensional modeling data X by data standardization preprocessing*(I*K*×J*)。
The fourth step: modeling data X for two dimensions*PCA decomposition [10 ]]Respectively establishing Sirox and KLD standby operation stagesMonitoring models in section and stable production stage to obtain the number A of principal elements of each model*A load matrix P*And diagonal matrix S*. Based on modeling data for each modelAnd SPE*Statistics, using F distributions and χ2The distribution obtains the statistical control limit for the model.
The formula for the PCA decomposition is:
wherein, T*In a principal component subspace (I)*K*×A*) A scoring matrix of dimensions; p*In a principal component subspace (J)*×A*) A load matrix of dimensions; e*In the residual subspace (I)*K*×J*) A residual matrix of dimensions; a. the*Representing the number of the principal elements, and determining by an accumulative contribution rate method; diagonal matrix S*＝diag(λ*1,λ*2,…,λ**A*) From modeling data X*Covariance matrix ofFront A of*A characteristic value constitution
Calculating from the F distributionMonitoring control limits of statisticsComprises the following steps:
wherein, α is the confidence，Fα(A*,I*K*-A*) is corresponding to a confidence of α and a degree of freedom of A*、I*K*-A*F distribution threshold of (a).
According to chi2Distributed computing SPE*Monitoring control limits of statisticsComprises the following steps:
wherein, g*＝v*/(2n*)，n*、v*Are respectively SPE*The mean and variance of the statistics are monitored.
(2) Online monitoring and anomaly diagnosis
And (3) determining the leaf shred grades to be produced according to the shred manufacturing and production scheduling plan, and calling the leaf shred grades to be produced as indication variables to call monitoring models in corresponding stages of the same grades to respectively calculate monitoring statistics of Sirox and KLD in standby operation and stable production stages. A standby operation stage, monitoring whether a stable production environment meeting the process requirements is established at the later stage (namely before feeding) of the stage, and if the stable production environment is not established before the stage is finished, performing exception alarm and identifying the reason of exception; and in the stable production stage, calculating statistic by using the stable production data with the transient transition period removed, judging whether the control limit is continuously exceeded or not, discovering process abnormity in time and identifying abnormal reasons. The algorithm flow of the on-line monitoring and abnormality diagnosis in the cigarette tobacco shred making process is shown in figure 2.
The fifth step: according to the state monitoring requirement, determining the validity judgment rule of the online monitoring data of the Sriox heating humidifier and the KLD sheet tobacco dryer shown in the table 3 to obtain the current valid data in the standby operation stageOr stabilizing currently valid data of production stageThe sampling frequency of the monitored variable was 10 s/time.
TABLE 3 validity judgment rules for on-line monitoring data
And a sixth step: according to the current leaf shred grade and the equipment operation stage to which the current effective data belongs, the mean value of the Sirox and KLD monitoring model at the corresponding stage of the same grade is utilizedExcept for standard deviationCarrying out standardized preprocessing on the data to obtain the current data x to be analyzedSSnew(1×JSS)、xKSnew(1×JKS) Or xSPnew(1×JSP)、xKPnew(1×JKP)。
The seventh step: projecting the current data to be analyzed to Sirox and KLD monitoring models at corresponding stages with the same grade, and utilizing a load matrix P of the monitoring models*And diagonal matrix S*Calculating current data to be analysedAndstatistics are obtained. The calculation formula is as follows:
wherein,represents (1 XJ) obtained by reconstruction*) The vector is estimated dimensionally.
Eighth step: the introduction of an abnormal alarm, which is questioned by communication with process personnel, is defined as: at the end of the standby operation phase, advance T of 12 sets of raw data within 2 minutes2And SPE statistics exceed the control limit; t of 6 consecutive sets of raw data within 1 minute during the stationary production phase2And SPE statistics both exceed control limits. At the moment of an abnormal alarm point, a contribution graph method is adopted [11 ]]And calculating the contribution rate of each monitoring variable to the overrun statistic, and separating the cause variable causing the statistic overrun.
The effectiveness of the method of the present invention is illustrated below with reference to the example of the process of producing shredded tobacco from line F of Hangzhou cigarette factory.
Selecting historical operating data of an F line 'Liqun A shredded tobacco grade' of a shredded tobacco making section in 2016, 7-12 months of Hangzhou cigarette factories for multi-stage distributed monitoring modeling and model verification. Screening historical operation data by a production process manager, selecting 13 normal batches in 7-9 months as modeling data, and performing variable expansion and standardization pretreatment on original data to obtain high-quality two-dimensional modeling data XSS(612×5)、XKS(612×14)、XSP(6321X 8) and XKP(6321 × 17). Monitoring models of Sirox and KLD in a standby operation stage and monitoring models of Sirox and KLD in a stable production stage are respectively established by adopting a principal component analysis method, the confidence coefficient of a control limit is 0.99, and the number of principal components and the control limit of each monitoring model are shown in a table 4.
TABLE 4 "Liqun A leaf shred number" number of principal elements and control limits of each monitoring model
And (3) re-collecting data of a certain normal batch of the cut tobacco number of 10 months for model verification, wherein the batch contains 1002 groups of original sampling data, and obtaining high-quality test data of a standby operation stage and a stable production stage respectively through validity judgment of on-line monitoring data, wherein the on-line monitoring result of the test data is shown in fig. 3. T of Sirox and KLD at the end of the stand-by operation phase2And SPE statistic is within control limit, and in stable production stage, there is T of individual original data2And SPE statistics exceed control limits, but no abnormal alarm is caused. Therefore, the method can accurately indicate the normal state of the standby operation stage and the stable production stage, and has the capability of accurately and effectively monitoring the normal state of the cut tobacco processing section batch process.
And (3) acquiring 2 abnormal batch data of the cut tobacco brand numbers of 10 months and 12 months again for model verification, wherein the abnormal batch 1 comprises 1140 groups of original sampling data, and the abnormal batch 2 comprises 1050 groups of original sampling data. Through the validity judgment of the online monitoring data, high-quality test data 1 and high-quality test data 2 of the abnormal batch 1 and the abnormal batch 2 in the standby operation stage and the stable production stage are respectively obtained, and the online monitoring results of the two batches of test data are shown in fig. 4 and 5.
As can be seen from fig. 4, the batch has a condition of stopping feeding, the stable production environment is established by the Sirox at the 227 th data point of the standby operation stage, and the stable production environment is established by the KLD at the 93 th and 217 th data points of the standby operation stage. According to the definition of an abnormal alarm, the Sirox alarm appears at the 143 th data point moment, which shows that the feeding is started when a stable production environment is not established in the Sirox standby operation stage, the contribution rate of each monitoring variable to the statistic overrun is shown in figure 6, the contribution rate of 2 variables of the steam mass flow and the steam film valve opening is large, and the feeding should be stopped immediately to avoid producing unqualified cut tobacco. Further analyzing the abnormality to find that in the standby operation stage, the set value of the Sirox steam mass flow is 400kg/h, the detection value is about 1033kg/h, the opening of the steam film valve is 0%, and the temperature of Sirox outlet cut tobacco is close to the ambient temperature after feeding, so that the Sirox steam mass flow detection value can be judged to be abnormal. In actual production, due to lack of effective monitoring in a standby operation stage, the abnormality cannot be discovered until the moisture of the cut tobacco after the KLD is dried cannot be controlled to meet the requirements of process indexes, and feeding is stopped manually by operators, so that certain unqualified cut tobacco is generated in the period.
As can be seen from FIG. 5, when the batch has a condition of material stop, the statistics of Sirox in the standby operation stage and the stable production stage exceed the corresponding control limits, and the statistics in different stages are relatively stable, which indicates that there is a long-term constant deviation of the monitored variable in the production process of the batch. According to the definition of the abnormal alarm, the abnormal alarm of the Sirox which appears earliest is at the 121 th data point moment, the contribution rate of each monitoring variable to the statistic overrun is shown in figure 7, and the contribution rates of 4 variables of the steam pressure, the steam temperature, the steam volume flow and the steam film valve opening before the Sirox valve are all large. Further analysis of the anomaly reveals that the pressure reducing valve of the steam main pipe is replaced in the period, and the steam pressure before the Sirox valve is replaced is not adjusted to the central value required by the process. Due to PID closed-loop control of the steam mass flow, the steam mass flow detection value can track a corresponding set value, and an operator does not find the abnormality in time.
The KLD does not generate abnormal alarm in the first standby operation stage, abnormal alarm occurs in SPE statistic in a residual subspace at the 127 th data point moment of the first stable production stage, the contribution rate of each monitoring variable to statistic overrun at the moment is shown in figure 8, and the contribution rate of the temperature of the dried cut tobacco is large. Further analysis of the abnormality finds that the temperature detection value of the cut tobacco after drying is lower than that of the normal batch by about 1 ℃ in the early stage of the first stable production stage of the batch, and the cut tobacco cannot enter the stable state of the normal batch within the transition time of 5 minutes.
KLD does not generate abnormal alarm in the middle and later stages of the first stable production stage, but the feeding is stopped in the production process. One possibility is that the model fails to report the anomaly, and the other possibility is that the anomaly is not reflected in the current monitored variables. Further analyzing the abnormality, the feeding stop is caused by the air separation blockage caused by the abnormal position of the baffle of the cooling air separator, a monitoring variable capable of reflecting the abnormality of the cooling air separator is not available at present, and if the abnormality is found, the existing measuring point of the cooling air separator needs to be optimally configured to obtain more monitoring information.
The statistics of the KLD in the second standby operation stage all exceed the corresponding control limits, an abnormal alarm occurs at the 209 th data point time, the contribution rate of each monitored variable to the statistic overrun at the time is as shown in fig. 9, and the contribution degree of the moisture-removing negative pressure and the hot air temperature is large. Further analysis on the abnormity shows that when the standby operation stage is finished, the moisture-discharging negative pressure detection value is 3 times that of the normal batch, the hot air temperature detection value is about 1 degree lower than that of the normal batch, the stable production environment meeting the process requirement cannot be established, and the early stage of the stable production stage after the transition time of 5 minutes, namely T2And SPE statistics fluctuate significantly and a small number of alarms occur.
According to the analysis of the abnormal batches, the method can correctly indicate the abnormal states of the standby operation stage and the stable production stage, and has the capability of accurately and effectively monitoring the abnormal states of the cut tobacco processing section batch process.
Claims (1)
1. The multi-stage distributed monitoring and diagnosis method for the cigarette cut tobacco making process based on the blocking and layering thought is characterized by comprising the following steps:
the first step is as follows: according to existing measuring points and equipment monitoring requirements of a production site, respectively determining monitoring variables of a Sirox heating and humidifying machine and a KLD thin plate cut-tobacco drying machine in a standby operation stage and a stable production stage, wherein the monitoring variables of the Sirox heating and humidifying machine and the KLD thin plate cut-tobacco drying machine in the standby operation stage are respectively serial numbers 1-5 and 1-14, and the monitoring variables of the Sirox heating and humidifying machine and the KLD thin plate cut-tobacco drying machine in the stable production stage are respectively serial numbers 1-8 and 1-17;
TABLE 1 monitoring variables of Sirox heating humidifier, KLD thin plate cut tobacco dryer in standby operation stage and stable production stage
The second step is that: respectively determining validity judgment rules of offline modeling data of a Sirox heating and humidifying machine and a KLD sheet cut-tobacco drying machine in a standby operation stage and a stable production stage according to production process requirements; rejecting the broken material batch and abnormal batch data of the cut tobacco grade in the cut tobacco making centralized control system, and obtaining the original three-dimensional modeling data of the standby operation stage and the stable production stage of the Sirox heating humidifier and the KLD thin plate cut tobacco drying machine according to the modeling data validity judgment rule of table 2 Wherein I represents a batch, J represents a monitoring variable, K represents a sampling point, a subscript SS represents a stand-by operation stage of a Sirox temperature-increasing humidifier, KS represents a stand-by operation stage of a KLD sheet tobacco dryer, SP represents a stable operation stage of the Sirox temperature-increasing humidifier, KP represents a stable operation stage of the KLD sheet tobacco dryer, and the sampling frequency of the monitoring variable is 10 s/time;
TABLE 2 validity judgment rules for offline modeling data
The third step: three-dimensional data by adopting variable expansion modeExpansion into two-dimensional dataWherein, SS or KS or SP or KP is taken as the index, and is used for distinguishing the standby operation stage and the stable production stage of a Sirox heating humidifier and a KLD thin plate cut tobacco drying machine; for two-dimensional matrixEach variable inPerforming an average value reductionExcept for standard deviationTo obtain high-quality two-dimensional modeling data X by data standardization preprocessing*(I*K*×J*)；
The fourth step: modeling data X for two dimensions*PCA decomposition is carried out, monitoring models of a Sirox heating humidifier, a KLD thin plate cut tobacco drying machine standby operation stage and a stable production stage are respectively established, and the number A of principal components of each model is obtained*A load matrix P*And diagonal matrix S*(ii) a T based on modeling data of each model* 2And SPE*Statistics, using F distributions and χ2Distributing to obtain the statistic control limit of the model;
the formula for the PCA decomposition is:
wherein, T*In a principal component subspace (I)*K*×A*) A scoring matrix of dimensions; p*In a principal component subspace (J)*×A*) A load matrix of dimensions; e*In the residual subspace (I)*K*×J*) A residual matrix of dimensions; a. the*Representing the number of the principal elements, and determining by an accumulative contribution rate method; diagonal matrixFrom modeling data X*Of the covariance matrix sigma*＝X* TX*/(I*K*-1) front A*Forming a characteristic value;
calculating T from F distribution* 2Monitoring control limits of statisticsComprises the following steps:
where α is the confidence, Fα(A, I K-A) is the corresponding confidence coefficient alpha and the degree of freedom A*、I*K*-A*F distribution threshold value of (a);
according to chi2Distributed computing SPE*Monitoring control limits of statisticsComprises the following steps:
wherein, g*＝v*/(2n*)，n*、v*Are respectively SPE*Monitoring the mean and variance of the statistics;
the fifth step: according to the state monitoring requirement, determining the validity judgment rule of the online monitoring data of the Sirox heating and humidifying machine and the KLD thin plate cut-tobacco drier shown in the table 3 to obtain the current valid data in the standby operation stage Or stabilizing currently valid data of production stageThe sampling frequency of the monitoring variable is 10 s/time;
TABLE 3 validity judgment rules for on-line monitoring data
And a sixth step: according to the current leaf shred grade and the equipment operation stage to which the current effective data belongs, the same grade corresponding stage Sirox temperature increasing and humidifying machine and KLD thin plate tobacco dryer are utilized to monitor the mean value of the modelExcept for standard deviationCarrying out standardized preprocessing on the data to obtain the current data x to be analyzedSSnew(1×JSS)、xKSnew(1×JKS) Or xSPnew(1×JSP)、xKPnew(1×JKP) (ii) a The seventh step: projecting the current data to be analyzed to a Sirox heating and humidifying machine and a KLD thin plate cut-tobacco drier monitoring model at the corresponding stage of the same mark, and utilizing a load matrix P of the monitoring model*And diagonal matrix S*Calculating current data to be analysedAnd SPE*newStatistics; the calculation formula is as follows:
wherein,represents (1 XJ) obtained by reconstruction*) Estimating a vector in dimension;
eighth step: the introduction of an abnormal alarm, which is questioned by communication with process personnel, is defined as: at the end of the standby operation phase, advance T of 12 sets of raw data within 2 minutes2And SPE statistics exceed the control limit; t of 6 consecutive sets of raw data within 1 minute during the stationary production phase2And SPE statistics exceed the control limit; and at the moment of an abnormal alarm point, calculating the contribution rate of each monitoring variable to the overrun statistic by adopting a contribution graph method, and separating the cause variable causing the overrun of the statistic.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711472774.4A CN108158028A (en) | 2017-12-29 | 2017-12-29 | Cigarette cut tobacco process multistage distributed monitoring and diagnostic method based on tile and hierarchy thought |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711472774.4A CN108158028A (en) | 2017-12-29 | 2017-12-29 | Cigarette cut tobacco process multistage distributed monitoring and diagnostic method based on tile and hierarchy thought |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108158028A true CN108158028A (en) | 2018-06-15 |
Family
ID=62520044
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711472774.4A Pending CN108158028A (en) | 2017-12-29 | 2017-12-29 | Cigarette cut tobacco process multistage distributed monitoring and diagnostic method based on tile and hierarchy thought |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108158028A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109188905A (en) * | 2018-09-07 | 2019-01-11 | 浙江大学 | A kind of on-line monitoring method of the sound feature Cooperative Analysis towards gigawatt extra-supercritical unit |
| CN109283912A (en) * | 2018-08-07 | 2019-01-29 | 浙江大学 | A kind of distributed sound coordination with the synthesis monitoring method towards intelligent power plant's large size Thermal generation unit pulverized coal preparation system |
| CN110897192A (en) * | 2019-12-12 | 2020-03-24 | 河南中烟工业有限责任公司 | Tobacco stem quality control method and system based on tobacco stem temperature increase |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104699050A (en) * | 2015-02-13 | 2015-06-10 | 浙江中烟工业有限责任公司 | Leaf-shred preparation segment on-line monitoring and fault diagnosing method for cigarette filament treatment driven by data |
| CN105242660A (en) * | 2015-07-15 | 2016-01-13 | 浙江中烟工业有限责任公司 | Multi-modal cigarette primary processing process online monitoring and fault diagnosis method based on relative change analysis |
-
2017
- 2017-12-29 CN CN201711472774.4A patent/CN108158028A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104699050A (en) * | 2015-02-13 | 2015-06-10 | 浙江中烟工业有限责任公司 | Leaf-shred preparation segment on-line monitoring and fault diagnosing method for cigarette filament treatment driven by data |
| CN105242660A (en) * | 2015-07-15 | 2016-01-13 | 浙江中烟工业有限责任公司 | Multi-modal cigarette primary processing process online monitoring and fault diagnosis method based on relative change analysis |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109283912A (en) * | 2018-08-07 | 2019-01-29 | 浙江大学 | A kind of distributed sound coordination with the synthesis monitoring method towards intelligent power plant's large size Thermal generation unit pulverized coal preparation system |
| CN109283912B (en) * | 2018-08-07 | 2020-06-30 | 浙江大学 | Distributed dynamic and static cooperative comprehensive monitoring method for pulverizing system of large coal-fired generator set of intelligent power plant |
| CN109188905A (en) * | 2018-09-07 | 2019-01-11 | 浙江大学 | A kind of on-line monitoring method of the sound feature Cooperative Analysis towards gigawatt extra-supercritical unit |
| CN109188905B (en) * | 2018-09-07 | 2019-10-15 | 浙江大学 | A kind of on-line monitoring method of the sound feature Cooperative Analysis towards gigawatt extra-supercritical unit |
| CN110897192A (en) * | 2019-12-12 | 2020-03-24 | 河南中烟工业有限责任公司 | Tobacco stem quality control method and system based on tobacco stem temperature increase |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108158028A (en) | Cigarette cut tobacco process multistage distributed monitoring and diagnostic method based on tile and hierarchy thought | |
| CN105242660A (en) | Multi-modal cigarette primary processing process online monitoring and fault diagnosis method based on relative change analysis | |
| CN104699050A (en) | Leaf-shred preparation segment on-line monitoring and fault diagnosing method for cigarette filament treatment driven by data | |
| CN106649755B (en) | Threshold value self-adaptive setting abnormity detection method for multi-dimensional real-time power transformation equipment data | |
| CN102147613B (en) | Model predictive control method and system for moisture and temperature of tobacco redryer | |
| CN110150711B (en) | Moisture control method and system for moisture regain and humidification process based on multiple regression | |
| CN107676923A (en) | A kind of Air conditioning System for Hospitals cooling tower failure automatic judging method and device | |
| CN107239066B (en) | A kind of Vertical Mill operation closed-loop control device and method based on data mining | |
| CN117459073B (en) | Intelligent management method for heat pump system operation data | |
| CN112132316A (en) | System and method for monitoring abnormality of on-line equipment in silk making link | |
| CN106897542A (en) | Based on the cigarette cut tobacco segment fault diagnostic method that notable failure variable is extracted | |
| CN112859977A (en) | Dryer control method and system based on artificial intelligence and computer equipment | |
| CN114115393A (en) | Method for controlling moisture and temperature at outlet of cut tobacco dryer for sheet cut tobacco making line | |
| CN114326486B (en) | Process monitoring method based on probability slow feature analysis and elastic weight consolidation | |
| CN111275331A (en) | Monitoring method and device for production system | |
| CN111728253B (en) | Control method and system for tobacco airflow drying strength | |
| CN108664696A (en) | A kind of assessment method and device of handpiece Water Chilling Units operating status | |
| CN116757354A (en) | Tobacco redrying section key parameter screening method based on multilayer perceptron | |
| CN114747794B (en) | Cut stem dryer outlet water content control method based on accumulation and strategy | |
| Liu et al. | Multi-PCA based Fault Detection Model Combined with Prior knowledge of HVAC | |
| CN112800671B (en) | Data processing method and device and electronic equipment | |
| Zanoli et al. | Complexity-based methodology for Fault Diagnosis: application on a centrifugal machine | |
| Yuwono et al. | Automatic feature selection using multiobjective cluster optimization for fault detection in a heating ventilation and air conditioning system | |
| CN118052451A (en) | Method and device for predicting water content quality of cut tobacco at outlet of cut tobacco dryer | |
| Qu et al. | A wind turbine fault diagnosis method with self-updating model based on SCADA data mining |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180615 |