CN113642251B - Data analysis and prediction method and system for building ceramic spray powder preparation quality - Google Patents
Data analysis and prediction method and system for building ceramic spray powder preparation quality Download PDFInfo
- Publication number
- CN113642251B CN113642251B CN202111016978.3A CN202111016978A CN113642251B CN 113642251 B CN113642251 B CN 113642251B CN 202111016978 A CN202111016978 A CN 202111016978A CN 113642251 B CN113642251 B CN 113642251B
- Authority
- CN
- China
- Prior art keywords
- powder
- data
- variable
- slurry
- spray
- 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
- 239000000843 powder Substances 0.000 title claims abstract description 164
- 239000007921 spray Substances 0.000 title claims abstract description 120
- 238000000034 method Methods 0.000 title claims abstract description 84
- 239000000919 ceramic Substances 0.000 title claims abstract description 49
- 238000002360 preparation method Methods 0.000 title claims abstract description 29
- 238000007405 data analysis Methods 0.000 title claims abstract description 26
- 238000012545 processing Methods 0.000 claims abstract description 13
- 208000025174 PANDAS Diseases 0.000 claims abstract description 12
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 claims abstract description 12
- 235000016496 Panda oleosa Nutrition 0.000 claims abstract description 12
- 230000002159 abnormal effect Effects 0.000 claims abstract description 11
- 238000010276 construction Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000011156 evaluation Methods 0.000 claims abstract description 5
- 239000002002 slurry Substances 0.000 claims description 85
- 230000006870 function Effects 0.000 claims description 20
- 238000012549 training Methods 0.000 claims description 18
- 238000004364 calculation method Methods 0.000 claims description 16
- 240000004718 Panda Species 0.000 claims description 11
- 238000005507 spraying Methods 0.000 claims description 10
- 239000002245 particle Substances 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 8
- 238000002485 combustion reaction Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 6
- 238000001514 detection method Methods 0.000 claims description 4
- 230000005484 gravity Effects 0.000 claims description 4
- 238000003860 storage Methods 0.000 claims description 4
- 238000012546 transfer Methods 0.000 claims description 4
- 238000013145 classification model Methods 0.000 claims description 3
- 238000005086 pumping Methods 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 2
- 238000004519 manufacturing process Methods 0.000 abstract description 25
- 238000004458 analytical method Methods 0.000 abstract description 5
- 238000013178 mathematical model Methods 0.000 abstract description 3
- 240000000220 Panda oleosa Species 0.000 abstract 1
- 230000000875 corresponding effect Effects 0.000 description 15
- 238000011049 filling Methods 0.000 description 8
- 238000010298 pulverizing process Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 3
- 239000003595 mist Substances 0.000 description 3
- 238000013523 data management Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000001035 drying Methods 0.000 description 2
- 238000003801 milling Methods 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 238000001694 spray drying Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000001704 evaporation Methods 0.000 description 1
- 230000008020 evaporation Effects 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000004886 process control Methods 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Feedback Control In General (AREA)
Abstract
The invention relates to the field of building ceramic spray powder preparation, in particular to a data analysis and prediction method and a system for the quality of building ceramic spray powder preparation; the data analysis and prediction method comprises the following steps: the method comprises the steps of obtaining flow data, constructing a traceable data link, constructing a preprocessing data set, calculating the total sample number of the preprocessing data set and the statistical condition of the number of variables by utilizing pandas, judging and processing missing values and abnormal values, and establishing model evaluation by using Xgboost function packages; after the data analysis and prediction system applies the data analysis and prediction method, the situation of the powder production quality of the spray tower can be monitored through construction and analysis of a mathematical model, and factor suggestions which can influence the product quality are rapidly given, so that technicians are assisted in rapid adjustment, the stability of the product quality is ensured, and the influence on the production of the subsequent process is reduced.
Description
Technical Field
The invention relates to the field of building ceramic spray powder preparation, in particular to a data analysis and prediction method and a system for the quality of building ceramic spray powder preparation.
Background
The spray drying tower is equipment required for preparing ceramic powder particles in the production process of building ceramics, the slurry is dehydrated and dried under the action of hot air through being sprayed out by a spraying device to form ceramic powder particles with certain moisture and granularity, namely, slurry fog drops and hot air form a mixed flow mode, the slurry fog drops form spheres under the action of surface tension, and the moisture is quickly evaporated and dried at high temperature to finally obtain the dried spherical ceramic powder particles.
In the ceramic building industry, the preparation of ceramic powder by using a drying tower is called a spray pulverizing process. The production, management and quality control in the spray pulverizing process still keep the characteristics of the traditional manufacturing industry, the production adjustment is totally dependent on the experience of technicians, no more scientific and standard indexes are used for guiding how to adjust production parameters in the production process, and the production efficiency and the powder quality are difficult to improve.
In order to solve the problem, the industry generally carries out adjustment experiments on spray powder preparation process parameters and spray powder preparation procedures of a production line by spray powder preparation workers. However, there are problems in that the adjustment time is long, trial and error costs are high, and only a worker with sufficient experience can perform an efficient adjustment experiment on the production line.
Disclosure of Invention
Aiming at the defects, the invention aims to provide a data analysis and prediction method and a data analysis and prediction system for the powder preparation quality of building ceramic spray, which solve the technical problems that in the prior art, the adjustment time is long, the trial-and-error cost is high, and only workers with enough experience can carry out effective adjustment experiments on a production line.
To achieve the purpose, the invention adopts the following technical scheme:
A data analysis and prediction method for the spray powder preparation quality of building ceramics comprises the following contents:
A. Acquiring flow data, spray tower parameters and slurry performance data, and preparing powder performance data of ceramic powder;
B. Setting up a data link of three processes of feeding the slurry into a spray tower, spraying powder preparation and detecting ceramic powder by using the process data as connection information to obtain a traceable data link;
C. Importing the traceable data link into a python dataset to obtain a preprocessed dataset, and calculating the total sample number and the statistics of the variable number of the preprocessed dataset by using pandas;
D. According to the total sample size and the statistics quantity of each variable, the judgment and the treatment of the missing value and the abnormal value are completed;
E. Taking powder performance data as an output variable X, taking spray tower parameters and slurry performance data as an input variable Y, and calculating characteristic correlation information of each input variable and each output variable by using a corr function of python; modeling and training single appointed parameters in powder performance data by using Xgboost function packages, and adjusting three super parameters including maximum depth (max_depth), learning rate (learning_rate) and evaluator number (n_ estimators) by using automatic parameter adjustment during training; the model effect is evaluated through the R2_SCORE index, and the model formed by different combinations of three super parameters of maximum depth (max_depth), learning rate (learning_rate) and estimator number (n_ estimators) is evaluated, wherein the R2_SCORE calculation formula is as follows:
The model with the training set and the testing set with the R2_SCORE value approaching to 1 is selected, the variable characteristics with the top ranking are output, and the variable characteristics with the top ranking are input variables with larger influence on the corresponding single specified parameter in the powder performance data.
Specifically, the step a further includes the following: the flow data comprises date and time information, a pulp pool number, a transfer cylinder number and a spray tower number; the spray tower parameters include: hearth temperature, tower top temperature, tower middle temperature, outlet temperature, combustion air frequency, grate frequency, screw pump frequency, spray gun count, spray sheet aperture, slurry feeding pressure, negative pressure, fan current and/or large fan frequency; the slurry performance data includes; slurry moisture, slurry flow rate, and/or slurry property-specific gravity; the powder performance data includes: powder name, powder moisture, powder bulk weight and/or particle size; and classifying the parameters.
Specifically, the step B further includes the following: and setting up a data link of the whole process from the process that slurry enters the spray tower to the process that powder preparation of the spray tower is finished and the powder quality is detected by taking the process data recorded by the slurry pumping time, the slurry pool number, the middle-rotating cylinder number, the spray tower, the date and time recorded by the spray tower, the spray tower number, the powder name and the powder detection time as connection information, so as to obtain a traceable data link.
Specifically, the step C further includes the following:
And importing the constructed traceable data link into a python data set to serve as a preprocessing data set, and obtaining statistical indexes of maximum value, maximum value bit number, minimum value position, 25% quantile, median, 75% quantile, mean value, average absolute deviation, variance, standard deviation, skewness and kurtosis of each variable by utilizing pandas, wherein the statistical indexes are used for knowing the total sample number and the statistical condition of the number of each variable.
More preferably, the step D further includes the following:
judging and processing the missing value: comparing the difference between the total sample size and the statistical quantity of each variable, and judging the condition of the missing value of each variable; determining a method for deleting or filling the missing value according to the actual requirements of the process experience and the model and by combining the actual conditions of the data; when the sample size of the data set is large or the field with more missing values is not an important variable, namely after deleting the missing values or missing value data, the number of samples/field entering the model can still ensure the validity of the model, and then the missing values or missing value samples can be directly removed; this can be achieved by Pandas or Numpy; filling the missing value can be performed by adopting a plurality of methods such as random forests, and when the data management stage cannot judge which method should be used for filling the missing value, the method can enter the model establishment stage for processing;
Judging and processing abnormal values: judging the distribution and deviation of numerical values in each variable according to the statistical indexes such as the maximum and minimum values, quartiles and the like, and judging whether abnormal values exist in the data; the outlier is automatically corrected.
More preferably, a data description statistics step is further arranged between the step D and the step E, and the step comprises the following steps: based on the process standard, comparing the slurry performance data, the powder performance data and the spray tower parameters produced in the stage to reach the standard; setting the standard range of the process parameters of target variables including slurry feeding pressure, large fan frequency, hearth temperature, tower top temperature, tower middle temperature and outlet temperature, powder moisture, powder volume weight and granularity as a number; setting a number which does not reach the standard and a null array which does not reach the standard proportion; establishing a circulation statistics substandard quantity and substandard proportion; and in the data set, if the variable in the corresponding stat_cols is not in the corresponding array standard range, calculating the number lower than the minimum standard value and the number higher than the maximum standard value, and outputting a nogood array of numbers which record the substandard and a nogood _rate array of substandard.
Specifically, the step of calculating the characteristic correlation information of each input variable and each output variable by using the corr function of python by using the powder performance data as the output variable and using the spray tower parameter and the slurry parameter as the input variables includes the following steps: the input variable X is hearth temperature, tower top temperature, tower middle temperature, outlet temperature, combustion air, spray gun count, slurry feeding pressure, slurry moisture, slurry flow rate, negative pressure, fan current or large fan frequency; the output variable Y is powder moisture, powder volume weight or powder granularity; and calculating the characteristic correlation of each input variable X and each output variable Y by using the corr function of python, and analyzing the correlation strength of the output variable Y and the input variable X by combining the calculation result.
Specifically, the correlation strong and weak correlation evaluation index of the output variable Y and the input variable X is as follows: calculating the characteristic correlation of each input variable X and each output variable Y by using the corr function of python, and judging that the output variable Y has weak correlation with the corresponding input variable X if the absolute value of the correlation coefficient of the calculated result exceeds 0.3; if the absolute value of the correlation coefficient of the calculated result exceeds 0.6, judging that the output variable Y and the corresponding input variable X have strong correlation; if the absolute value of the correlation coefficient of the calculation result is positive, the absolute value is negative, and if the absolute value is positive, the absolute value is negative.
The data analysis and prediction system for the construction ceramic spray powder quality is applied to the data analysis and prediction method for the construction ceramic spray powder quality; the data analysis and prediction system comprises:
the data acquisition module is used for acquiring flow data, spray tower parameters and slurry performance data and preparing powder performance data of the ceramic powder;
the data link construction module is used for constructing a data link of three processes of feeding the slurry into a spray tower, spraying powder preparation and detecting ceramic powder by taking the process data as connection information to obtain a traceable data link;
The data preprocessing module is used for importing a traceable data link into the python data set to obtain a preprocessed data set, and calculating the total sample number and the statistics of the variable number of the preprocessed data set by utilizing pandas;
the data judging and processing module is used for judging and processing the missing value and the abnormal value according to the total sample size and the statistical quantity of each variable;
The system also comprises a data description statistical module, a control module and a control module, wherein the data description statistical module is used for comparing the slurry performance data, the powder performance data and the spray tower parameters produced in the stage to reach the standard on the basis of the process standard;
The characteristic correlation calculation module is used for calculating characteristic correlation information of each input variable and each output variable by using a corr function of python by taking powder performance data as an output variable X and taking spray tower parameters and slurry parameters as input variables Y;
Xgboost a classification model building module, which builds a model by using Xgboost function packages, respectively carries out modeling training on single specified parameters in powder performance data, and three super parameters, namely maximum depth (max_depth), learning rate (learning_rate) and evaluator number (n_ estimators), are adjusted by using automatic parameter adjustment during training; the model effect is evaluated through the R2_SCORE index, and the model formed by different combinations of three super parameters of maximum depth (max_depth), learning rate (learning_rate) and estimator number (n_ estimators) is evaluated, wherein the R2_SCORE calculation formula is as follows:
The model with the training set and the testing set with the R2_SCORE value approaching to 1 is selected, the variable characteristics with the top ranking are output, and the variable characteristics with the top ranking are input variables with larger influence on the corresponding single specified parameter in the powder performance data.
A computer storage medium storing computer instructions which when invoked are used to perform a method of data analysis prediction of building ceramic spray pulverizing quality as described above.
The embodiment of the invention has the beneficial effects that:
According to the data analysis and prediction method and system, the influence correlation of the spray tower parameters and the slurry performance data on the powder performance characteristics in the building ceramic spray pulverizing process is combined, the condition of the spray tower pulverizing quality can be monitored through construction and analysis of a mathematical model, factor suggestions which possibly influence the production quality are rapidly given, technicians are assisted in rapid adjustment, stability of the product quality is guaranteed, and influence on production in the subsequent process is reduced.
Drawings
FIG. 1 is a schematic flow chart of a method for predicting the quality of powder produced by spraying architectural ceramic according to an embodiment of the invention;
FIG. 2 is a schematic flow chart of a method for predicting the quality of powder produced by spraying architectural ceramic according to another embodiment of the present invention;
FIG. 3 is a schematic diagram of a modeling flow in calculating feature correlation information for each input variable and each output variable in one embodiment of the invention.
Detailed Description
The technical scheme of the invention is further described below by the specific embodiments with reference to the accompanying drawings.
In one embodiment of the present application, as shown in fig. 1 and 2, a method for predicting the quality of spray powder of architectural ceramic by data analysis includes the following steps:
A. Acquiring flow data, spray tower parameters and slurry performance data, and preparing powder performance data of ceramic powder; specifically, the method comprises two stages of data dimension definition and data preparation.
The data dimension is clear: the spray drying is to spray ceramic slurry into mist droplets in a tower by a spray gun, and the mist droplets are contacted with hot air generated by a hot blast stove, and the drying is completed almost instantaneously due to the fact that the mist droplets are thin and have huge evaporation area, so that the desirable granular powder is obtained, and the powder is accumulated at the tower bottom and discharged by a discharge valve. In the actual production process of building ceramics, slurry enters a spray tower, process operators set parameters of the spray tower according to the quality requirements of powder in a production formula so as to obtain powder meeting the quality requirements, and the related data dimensions are three: slurry performance data (slurry test prior to entering the spray tower milling process), spray tower parameters (parameters during the spray tower equipment milling process), powder performance data (test data for spray tower process products), detailed variables including but not limited to those shown in table 1 below.
TABLE 1 variable detail table
As shown in the table above, the flow data includes date and time information, a pulp pool number, a transfer cylinder number and a spray tower number; the spray tower parameters include: hearth temperature, tower top temperature, tower middle temperature, outlet temperature, combustion air frequency, grate frequency, screw pump frequency, spray gun count, spray sheet aperture, slurry feeding pressure, negative pressure, fan current and/or large fan frequency; the slurry performance data includes; slurry moisture, slurry flow rate, and/or slurry property-specific gravity; the powder performance data includes: powder name, powder moisture, powder bulk weight and/or particle size.
Data preparation: the acquisition stage is divided into two cases according to different bases of factory digitization degree. One is that the digitization degree is higher, and the factory conditionally automatically collects the electronic data, and then directly calls the data from the database; in the other case, if the factory lacks automatic data acquisition conditions, according to the definite data dimension, the tables in the record variable detail list are combed in the daily production flow record of the factory, and data collection and arrangement are carried out; and the data which are not recorded in the daily production process are newly added, so that the complete recording and effective storage of the data recorded in the daily production process are ensured. The collection of powder performance data includes that part of factories can detect the physical and chemical properties of powder in multiple links, for example, some factories detect the powder performance after finishing powder preparation by a spray tower and record relevant powder quality data, and in the top link of a press, the powder entering a powder bin can be detected for the second time and record relevant powder quality data, so that two data for evaluating the powder quality are formed. In this case, considering that the present invention is directed to the study of the salient factors affecting the quality of the powder in the spray tower, it is suggested to use the powder quality data detected and recorded after the powder is prepared by the spray tower.
The flow data, the spray tower parameters, the slurry performance data and the powder performance data are classified according to the respective parameters, and are specifically shown in table 2.
TABLE 2 variable type partitioning
| Variable dimension | Variable name | Unit (B) | Variable type |
| Flow data | Date and time | / | |
| Flow data | Tower number | / | |
| Powder performance data | Powder name | Uncontrollable variable | |
| Spray tower parameters | Hearth temperature | ℃ | Controllable variable |
| Spray tower parameters | Overhead temperature | ℃ | Controllable variable |
| Spray tower parameters | Temperature in the column | ℃ | Controllable variable |
| Spray tower parameters | Outlet temperature | ℃ | Controllable variable |
| Spray tower parameters | Frequency of combustion-supporting air | Hz | Controllable variable |
| Spray tower parameters | Grate/screw pump frequency | Hz | Controllable variable |
| Spray tower parameters | Spray gun count | Controllable variable | |
| Spray tower parameters | Hole diameter of spray sheet | mm | Controllable variable |
| Spray tower parameters | Slurry feed pressure | Mpa | Controllable variable |
| Slurry performance data | Slurry moisture | ％ | Uncontrollable variable |
| Slurry performance data | Slurry flow rate | Second of | Uncontrollable variable |
| Spray tower parameters | Negative pressure | pa | Controllable variable |
| Spray tower parameters | Blower current | A | Controllable variable |
| Spray tower parameters | Big fan frequency | Hz | Controllable variable |
| Powder performance data | Moisture of powder | ％ | Target variable |
| Powder performance data | Powder volume weight | Target variable | |
| Powder performance data | Particle size | ％ | Target variable |
| Slurry performance data | Slurry properties-specific gravity | Uncontrollable variable |
B. Setting up a data link of three processes of feeding the slurry into a spray tower, spraying powder preparation and detecting ceramic powder by using the process data as connection information to obtain a traceable data link;
Specifically, the process data of the slurry pumping time, the slurry pool number, the middle transfer cylinder number, the date and time recorded by the spray tower, the spray tower number, the powder name and the powder detection time are taken as connection information, and a data link of the whole process from the slurry entering the spray tower to the powder preparation completion of the spray tower and the powder quality detection is constructed, so that a traceable data link is obtained.
C. Importing the traceable data link into a python dataset to obtain a preprocessed dataset, and calculating the total sample number and the statistics of the variable number of the preprocessed dataset by using pandas;
specifically, the constructed traceable data link is imported into a python data set to be used as a preprocessing data set, and pandas is utilized to obtain statistical indexes of maximum value, maximum value bit number, minimum value position, 25% bit number, median, 75% bit number, mean value, average absolute deviation, variance, standard deviation, skewness and kurtosis of each variable, wherein the statistical indexes are used for knowing the total sample number and the statistical condition of the number of each variable.
D. According to the total sample size and the statistics quantity of each variable, the judgment and the treatment of the missing value and the abnormal value are completed; the method specifically comprises the following steps:
Judging and processing the missing value:
Comparing the difference between the total sample size and the statistical quantity of each variable, and judging the condition of the missing value of each variable; determining a method for deleting or filling the missing value according to the actual requirements of the process experience and the model and by combining the actual conditions of the data;
when the sample size of the data set is large or the field with more missing values is not an important variable, namely after deleting the missing values or missing value data, the number of samples/field entering the model can still ensure the validity of the model, and then the missing values or missing value samples can be directly removed; this can be achieved by Pandas or Numpy;
Filling the missing value can be performed by adopting a plurality of methods such as random forests, and when the data management stage cannot judge which method should be used for filling the missing value, the method can enter the model establishment stage for processing;
Judging and processing abnormal values:
Judging the distribution and deviation of numerical values in each variable according to the statistical indexes such as the maximum and minimum values, quartiles and the like, and judging whether abnormal values exist in the data; the outlier is automatically corrected.
Specifically, there are mainly two cases (1) filling in errors for outliers: judging the normal value range of a single variable according to experience, for example, the slurry feeding pressure of spray tower equipment parameters in a general spray tower procedure, wherein the normal value range is 0.9MPa-1.4MPa, but a certain amount of 12MPa exists in data, so that 12MPa can be judged to be a filling error of missing decimal points, and can be directly modified into 1.2MPa; (2) The outliers are defined, and data points exceeding the mean value by + -3 times of standard deviation are generally judged as outliers from statistical operation (the judging standard needs to be determined according to different conditions, the invention judges the data points exceeding the mean value by + -3 times of standard deviation as outliers), and the outliers exceeding the range are removed.
In some embodiments, the method further comprises the steps of:
D1. The data description and statistics step comprises the following steps:
Based on the process standard, comparing the slurry performance data, the powder performance data and the spray tower parameters produced in the stage to reach the standard;
Setting the controllable parameters including slurry feeding pressure, large fan frequency, hearth temperature, tower top temperature, tower middle temperature and outlet temperature, and the technological parameter standard range of target variables including powder moisture, powder volume weight and granularity as an array;
For example:
setting up array＝[(0.9,1.4),(39,49),(980,1040),(600,650),(85,1000),(50,1000),(6.5,7.3),(0,10),(-1,0.8),(55,65),(85,95),(97,120)];
Setting various variable names in the array corresponding to stat_cols= [ ' slurry feeding pressure ', ' big fan frequency ', ' hearth temperature ', ' tower top temperature ', ' tower middle temperature ', ' outlet temperature ', ' powder water ', ' powder volume weight ', ' granularity ];
Setting a number which does not reach the standard and a null array which does not reach the standard proportion;
such as: setting an up-to-standard quantity array nogood = [ ];
Setting a nonstandard proportion array nogood _rate= [ ];
Establishing a circulation statistics substandard quantity and substandard proportion;
and in the data set, if the variable in the corresponding stat_cols is not in the corresponding array standard range, calculating the number lower than the minimum standard value and the number higher than the maximum standard value, and outputting a nogood array of numbers which record the substandard and a nogood _rate array of substandard.
According to the calculation results of the substandard quantity and the substandard proportion, a reference value is obtained for the situation of controlling technological process parameters and judging the approximate level of the powder quality in the process of preparing the powder by the staged spray tower, and the larger the substandard quantity or the substandard proportion is, the more certain problems or the larger fluctuation exists in the technological process control and the powder quality level.
Based on sample data, counting the quantity of various slurries and various powders, and comparing the quality and the fluctuation condition of operation of various parameters in the stage production process; mining production adjustment nodes presented by the data about optimizable production;
E. And calculating characteristic correlation information of each input variable and each output variable by using the corr function of python by taking the powder performance data as an output variable X and taking the spray tower parameter and the slurry performance data as an input variable Y.
Specifically, as shown in fig. 3, the modeling flow is shown in the drawing, and the input variable X is the hearth temperature, the tower top temperature, the tower middle temperature, the outlet temperature, the combustion air, the spray gun count, the slurry feeding pressure, the slurry moisture, the slurry flow rate, the negative pressure, the fan current or the large fan frequency; the output variable Y is powder moisture, powder volume weight or powder granularity; and calculating the characteristic correlation of each input variable X and each output variable Y by using the corr function of python, and analyzing the correlation strength of the output variable Y and the input variable X by combining the calculation result.
The correlation strength and weakness correlation judgment indexes of the output variable Y and the input variable X are as follows: calculating the characteristic correlation of each input variable X and each output variable Y by using the corr function of python, and judging that the output variable Y has weak correlation with the corresponding input variable X if the absolute value of the correlation coefficient of the calculated result exceeds 0.3; if the absolute value of the correlation coefficient of the calculated result exceeds 0.6, judging that the output variable Y and the corresponding input variable X have strong correlation; if the absolute value of the correlation coefficient of the calculation result is positive, the absolute value is negative, and if the absolute value is positive, the absolute value is negative.
For example: the correlation results of the output variable Y and the corresponding input variable X are output as shown in table 3.
TABLE 3 correlation result output
| Correlation coefficient | Weight per unit | Particle size |
| Hearth temperature | 0.075 | -0.306 |
| Overhead temperature | 0.082 | 0.286 |
| Temperature in the column | 0.051 | 0.207 |
| Outlet temperature | 0.019 | 0.192 |
| Frequency of combustion-supporting air | 0.287 | -0.374 |
| Spray gun count | -0.055 | 0.280 |
| Slurry feed pressure | 0.135 | -0.237 |
| Slurry moisture | -0.074 | 0.367 |
| Slurry flow rate | -0.094 | -0.168 |
| Negative pressure | -0.191 | 0.237 |
| Blower current | -0.407 | 0.157 |
| Big fan frequency | -0.170 | 0.457 |
In the table above, the volume weight of the output variable powder has weak correlation with the current of the input variable fan, and is in negative correlation, namely, when the volume weight of the spray tower powder is not ideal in the production process, the reason of the fan current can be preferentially considered; the granularity has weak correlation with the frequency of the large fan and is positively correlated. Other variable correlations with absolute values less than 0.3 are not apparent. The larger the absolute value of the correlation coefficient, the more the output variable is affected by the corresponding input variable.
E1. Modeling and training single appointed parameters in powder performance data by using Xgboost function packages, and adjusting three super parameters including maximum depth (max_depth), learning rate (learning_rate) and evaluator number (n_ estimators) by using automatic parameter adjustment during training; the model effect is evaluated through the R2_SCORE index, and the model formed by different combinations of three super parameters of maximum depth (max_depth), learning rate (learning_rate) and estimator number (n_ estimators) is evaluated, wherein the R2_SCORE calculation formula is as follows:
The model with the training set and the testing set with the R2_SCORE value approaching to 1 is selected, the variable characteristics with the top ranking are output, and the variable characteristics with the top ranking are input variables with larger influence on the corresponding single specified parameter in the powder performance data.
Through the construction and analysis of the mathematical model, the condition of the powder preparation quality of the spray tower can be monitored, the suggestion of factors which can influence the product quality is rapidly given, the rapid adjustment of technicians is assisted, the stability of the product quality is ensured, and the influence on the production of the subsequent process is reduced.
Example two
The data analysis and prediction system for the construction ceramic spray powder quality is applied to the data analysis and prediction method for the construction ceramic spray powder quality; the data analysis and prediction system comprises:
the data acquisition module is used for acquiring flow data, spray tower parameters and slurry performance data and preparing powder performance data of the ceramic powder;
the data link construction module is used for constructing a data link of three processes of feeding the slurry into a spray tower, spraying powder preparation and detecting ceramic powder by taking the process data as connection information to obtain a traceable data link;
The data preprocessing module is used for importing a traceable data link into the python data set to obtain a preprocessed data set, and calculating the total sample number and the statistics of the variable number of the preprocessed data set by utilizing pandas;
the data judging and processing module is used for judging and processing the missing value and the abnormal value according to the total sample size and the statistical quantity of each variable;
The system also comprises a data description statistical module, a control module and a control module, wherein the data description statistical module is used for comparing the slurry performance data, the powder performance data and the spray tower parameters produced in the stage to reach the standard on the basis of the process standard;
The characteristic correlation calculation module is used for calculating characteristic correlation information of each input variable and each output variable by using a corr function of python by taking powder performance data as an output variable X and taking spray tower parameters and slurry parameters as input variables Y;
The Xgboost classification model building module uses Xgboost function packages to build models, respectively carries out modeling training on single specified parameters in powder performance data, and uses automatic parameter adjustment to adjust three super parameters of maximum depth (max_depth), learning rate (learning_rate) and evaluator number (n_ estimators) during training; the model effect is evaluated through the R2_SCORE index, and the model formed by different combinations of three super parameters of maximum depth (max_depth), learning rate (learning_rate) and estimator number (n_ estimators) is evaluated, wherein the R2_SCORE calculation formula is as follows:
The model with the training set and the testing set with the R2_SCORE value approaching to 1 is selected, the variable characteristics with the top ranking are output, and the variable characteristics with the top ranking are input variables with larger influence on the corresponding single specified parameter in the powder performance data.
Example III
A computer storage medium storing computer instructions which when invoked are used to perform a method of data analysis prediction of the quality of a ceramic spray powder of a building as described above.
The advantage of the data chain construction in this embodiment is: at present, each process of most ceramic manufacturing enterprises has a large degree of data island, including a powder preparation process. In the embodiment of the invention, the data link is required to be built, so that the problem of data tracing is solved to a certain extent. The analysis prediction method can adaptively adjust the super parameters of the model, and select the optimal super parameters through the model evaluation index.
It should be noted that, the application of the data in this embodiment is not limited to the listed 3 data dimensions, but also to the listed data variables, and the parameters that the process experience considers to have an influence on the powder quality can be increased, and the parameters are put into the method as input X, so that the method and the system can be used for evaluation, analysis and prediction.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the present application. As used herein, the singular is also intended to include the plural unless the context clearly indicates otherwise, and furthermore, it is to be understood that the terms "comprises" and/or "comprising" when used in this specification are taken to specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof.
The relative arrangement of the components and steps, numerical expressions and numerical values set forth in these embodiments do not limit the scope of the present invention unless it is specifically stated otherwise. Meanwhile, it should be understood that the sizes of the respective parts shown in the drawings are not drawn in actual scale for convenience of description. Techniques, methods, and apparatus known to one of ordinary skill in the relevant art may not be discussed in detail, but should be considered part of the specification where appropriate. In all examples shown and discussed herein, any specific values should be construed as merely illustrative, and not a limitation. Thus, other examples of the exemplary embodiments may have different values. It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further discussion thereof is necessary in subsequent figures.
The technical principle of the present invention is described above in connection with the specific embodiments. The description is made for the purpose of illustrating the general principles of the invention and should not be taken in any way as limiting the scope of the invention. Other embodiments of the invention will be apparent to those skilled in the art from consideration of this specification without undue burden.
Claims (8)
1. The data analysis and prediction method for the powder preparation quality of the building ceramic spray is characterized by comprising the following steps of:
A. Acquiring flow data, spray tower parameters and slurry performance data, and preparing powder performance data of ceramic powder;
B. Setting up a data link of three processes of feeding the slurry into a spray tower, spraying powder preparation and detecting ceramic powder by using the process data as connection information to obtain a traceable data link;
C. Importing the traceable data link into a python dataset to obtain a preprocessed dataset, and calculating the total sample number and the statistics of the variable number of the preprocessed dataset by using pandas;
D. According to the total sample size and the statistics quantity of each variable, the judgment and the treatment of the missing value and the abnormal value are completed;
E. Taking powder performance data as an output variable X, taking spray tower parameters and slurry performance data as an input variable Y, and calculating characteristic correlation information of each input variable and each output variable by using a corr function of python; modeling training is carried out on single specified parameters in powder performance data respectively by using Xgboost function packages, and three super parameters including maximum depth max_depth, learning rate learning_rate and evaluator number n_ estimators are adjusted by using automatic parameter adjustment during training; and evaluating the model effect through R2_SCORE indexes, and evaluating models formed by different combinations of three super parameters of maximum depth max_depth, learning rate learning_rate and estimator number n_ estimators, wherein the R2_SCORE calculation formula is as follows:
Selecting models with the values of R2_SCORE of a training set and a testing set approaching 1, and outputting variable characteristics with the top ranking, wherein the variable characteristics with the top ranking are input variables with larger influence on corresponding single specified parameters in powder performance data;
And a data description statistical step is further arranged between the steps D and E, and comprises the following steps:
Based on the process standard, comparing the slurry performance data, the powder performance data and the spray tower parameters produced in the stage to reach the standard;
Setting the controllable parameters including slurry feeding pressure, large fan frequency, hearth temperature, tower top temperature, tower middle temperature and outlet temperature, and the technological parameter standard range of target variables including powder moisture, powder volume weight and granularity as an array;
Setting various variable names in the array corresponding to stat_cols= [ ' slurry feeding pressure ', ' big fan frequency ', ' hearth temperature ', ' tower top temperature ', ' tower middle temperature ', ' outlet temperature ', ' powder water ', ' powder volume weight ', ' granularity ];
Setting a number which does not reach the standard and a null array which does not reach the standard proportion;
Establishing a circulation statistics substandard quantity and substandard proportion;
and in the data set, if the variable in the corresponding stat_cols is not in the corresponding array standard range, calculating the number lower than the minimum standard value and the number higher than the maximum standard value, and outputting a nogood array of numbers which record the substandard and a nogood _rate array of substandard.
2. The method for predicting the quality of spray powder of architectural ceramic according to claim 1, wherein the step a further comprises the following steps: the flow data comprises date and time information, a pulp pool number, a transfer cylinder number and a spray tower number; the spray tower parameters include: hearth temperature, tower top temperature, tower middle temperature, outlet temperature, combustion air frequency, grate frequency, screw pump frequency, spray gun count, spray sheet aperture, slurry feeding pressure, negative pressure, fan current and/or large fan frequency; the slurry performance data includes; slurry moisture, slurry flow rate, and/or slurry property-specific gravity; the powder performance data includes: powder name, powder moisture, powder bulk weight and/or particle size; and classifying the parameters.
3. The method for predicting the quality of spray powder of architectural ceramic according to claim 2, wherein the step B further comprises the following steps:
and setting up a data link of the whole process from the process that slurry enters the spray tower to the process that powder preparation of the spray tower is finished and the powder quality is detected by taking the process data recorded by the slurry pumping time, the slurry pool number, the middle-rotating cylinder number, the spray tower, the date and time recorded by the spray tower, the spray tower number, the powder name and the powder detection time as connection information, so as to obtain a traceable data link.
4. The method for predicting the quality of spray powder of architectural ceramic according to claim 2, wherein said step C further comprises the following steps:
And importing the constructed traceable data link into a python data set to serve as a preprocessing data set, and obtaining statistical indexes of maximum value, maximum value bit number, minimum value position, 25% quantile, median, 75% quantile, mean value, average absolute deviation, variance, standard deviation, skewness and kurtosis of each variable by utilizing pandas, wherein the statistical indexes are used for knowing the total sample number and the statistical condition of the number of each variable.
5. The method for predicting the quality of powder by spraying and analyzing the data of the ceramic powder by building according to claim 1, wherein the step of calculating the characteristic correlation information of each input variable and each output variable by using the corr function of python by taking the powder performance data as the output variables and taking the spray tower parameter and the slurry parameter as the input variables comprises the following steps:
The input variable X is hearth temperature, tower top temperature, tower middle temperature, outlet temperature, combustion air, spray gun count, slurry feeding pressure, slurry moisture, slurry flow rate, negative pressure, fan current or large fan frequency; the output variable Y is powder moisture, powder volume weight or powder granularity; and calculating the characteristic correlation of each input variable X and each output variable Y by using the corr function of python, and analyzing the correlation strength of the output variable Y and the input variable X by combining the calculation result.
6. The method for predicting the data analysis of the powder preparation quality of the architectural ceramic spray according to claim 5, wherein the correlation strength and weakness correlation evaluation index of the output variable Y and the input variable X is as follows: calculating the characteristic correlation of each input variable X and each output variable Y by using the corr function of python, and judging that the output variable Y has weak correlation with the corresponding input variable X if the absolute value of the correlation coefficient of the calculated result exceeds 0.3; if the absolute value of the correlation coefficient of the calculated result exceeds 0.6, judging that the output variable Y and the corresponding input variable X have strong correlation; if the absolute value of the correlation coefficient of the calculation result is positive, the absolute value is negative, and if the absolute value is positive, the absolute value is negative.
7. A data analysis and prediction system for the spray powder quality of building ceramics is characterized in that the data analysis and prediction method for the spray powder quality of building ceramics is applied according to any one of claims 1-6,
The data analysis and prediction system for the spray powder preparation quality of the building ceramic comprises the following components:
the data acquisition module is used for acquiring flow data, spray tower parameters and slurry performance data and preparing powder performance data of the ceramic powder;
the data link construction module is used for constructing a data link of three processes of feeding the slurry into a spray tower, spraying powder preparation and detecting ceramic powder by taking the process data as connection information to obtain a traceable data link;
The data preprocessing module is used for importing a traceable data link into the python data set to obtain a preprocessed data set, and calculating the total sample number and the statistics of the variable number of the preprocessed data set by utilizing pandas;
the data judging and processing module is used for judging and processing the missing value and the abnormal value according to the total sample size and the statistical quantity of each variable;
The system also comprises a data description statistical module, a control module and a control module, wherein the data description statistical module is used for comparing the slurry performance data, the powder performance data and the spray tower parameters produced in the stage to reach the standard on the basis of the process standard;
The characteristic correlation calculation module is used for calculating characteristic correlation information of each input variable and each output variable by using a corr function of python by taking powder performance data as an output variable X and taking spray tower parameters and slurry parameters as input variables Y;
Xgboost a classification model building module, which builds a model by using Xgboost function packages, respectively carries out modeling training on single specified parameters in powder performance data, and adjusts three super parameters including maximum depth max_depth, learning rate learning_rate and evaluator number n_ estimators by using automatic parameter adjustment during training; and evaluating the model effect through R2_SCORE indexes, and evaluating models formed by different combinations of three super parameters of maximum depth max_depth, learning rate learning_rate and estimator number n_ estimators, wherein the R2_SCORE calculation formula is as follows:
The model with the training set and the testing set with the R2_SCORE value approaching to 1 is selected, the variable characteristics with the top ranking are output, and the variable characteristics with the top ranking are input variables with larger influence on the corresponding single specified parameter in the powder performance data.
8. A computer storage medium storing computer instructions which, when invoked, are adapted to perform a method of data analysis and prediction of the quality of a ceramic building spray powder process according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111016978.3A CN113642251B (en) | 2021-08-31 | 2021-08-31 | Data analysis and prediction method and system for building ceramic spray powder preparation quality |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111016978.3A CN113642251B (en) | 2021-08-31 | 2021-08-31 | Data analysis and prediction method and system for building ceramic spray powder preparation quality |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113642251A CN113642251A (en) | 2021-11-12 |
| CN113642251B true CN113642251B (en) | 2024-05-28 |
Family
ID=78424804
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111016978.3A Active CN113642251B (en) | 2021-08-31 | 2021-08-31 | Data analysis and prediction method and system for building ceramic spray powder preparation quality |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113642251B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116992390B (en) * | 2023-09-26 | 2023-12-05 | 北京联创高科信息技术有限公司 | Configuration and display method of abnormal data |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108375808A (en) * | 2018-03-12 | 2018-08-07 | 南京恩瑞特实业有限公司 | Dense fog forecasting procedures of the NRIET based on machine learning |
| CN108595803A (en) * | 2018-04-13 | 2018-09-28 | 重庆科技学院 | Shale gas well liquid loading pressure prediction method based on recurrent neural network |
| WO2020234899A2 (en) * | 2019-05-17 | 2020-11-26 | Tata Consultancy Services | Method and system for adaptive learning of models for manufacturing systems |
| CN112687349A (en) * | 2020-12-25 | 2021-04-20 | 广东海洋大学 | Construction method of model for reducing octane number loss |
| CN112785091A (en) * | 2021-03-04 | 2021-05-11 | 湖北工业大学 | Method for performing fault prediction and health management on oil field electric submersible pump |
| CN112818601A (en) * | 2021-02-05 | 2021-05-18 | 河海大学 | Hydroelectric generating set health assessment method based on GA-BP neural network and error statistical analysis |
| CN112990311A (en) * | 2021-03-15 | 2021-06-18 | 中国建设银行股份有限公司 | Method and device for identifying admitted client |
| CN113008805A (en) * | 2021-02-07 | 2021-06-22 | 浙江工业大学 | Radix angelicae decoction piece quality prediction method based on hyperspectral imaging depth analysis |
| WO2021143067A1 (en) * | 2020-05-28 | 2021-07-22 | 平安科技（深圳）有限公司 | Method and apparatus for predicting workpiece quality, and computer device |
-
2021
- 2021-08-31 CN CN202111016978.3A patent/CN113642251B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108375808A (en) * | 2018-03-12 | 2018-08-07 | 南京恩瑞特实业有限公司 | Dense fog forecasting procedures of the NRIET based on machine learning |
| CN108595803A (en) * | 2018-04-13 | 2018-09-28 | 重庆科技学院 | Shale gas well liquid loading pressure prediction method based on recurrent neural network |
| WO2020234899A2 (en) * | 2019-05-17 | 2020-11-26 | Tata Consultancy Services | Method and system for adaptive learning of models for manufacturing systems |
| WO2021143067A1 (en) * | 2020-05-28 | 2021-07-22 | 平安科技（深圳）有限公司 | Method and apparatus for predicting workpiece quality, and computer device |
| CN112687349A (en) * | 2020-12-25 | 2021-04-20 | 广东海洋大学 | Construction method of model for reducing octane number loss |
| CN112818601A (en) * | 2021-02-05 | 2021-05-18 | 河海大学 | Hydroelectric generating set health assessment method based on GA-BP neural network and error statistical analysis |
| CN113008805A (en) * | 2021-02-07 | 2021-06-22 | 浙江工业大学 | Radix angelicae decoction piece quality prediction method based on hyperspectral imaging depth analysis |
| CN112785091A (en) * | 2021-03-04 | 2021-05-11 | 湖北工业大学 | Method for performing fault prediction and health management on oil field electric submersible pump |
| CN112990311A (en) * | 2021-03-15 | 2021-06-18 | 中国建设银行股份有限公司 | Method and device for identifying admitted client |
Non-Patent Citations (3)
| Title |
|---|
| 基于机器学习模型的专利质量预测初探;刘夏;黄灿;余骁锋;;情报学报(第04期);全文 * |
| 基于机器学习的产油量主控因素分析;林霞;刘宗尚;高宇;武博宇;;信息***工程(第12期);全文 * |
| 基于质量预测与统计分析的烧结过程性能监控***研究;朱占军;王明侠;王春林;;烧结球团(第01期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113642251A (en) | 2021-11-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111461555B (en) | Production line quality monitoring method, device and system | |
| CN111103854A (en) | System and method for improving production stability of tobacco cut-tobacco drier | |
| CN111144667A (en) | Tobacco conditioner discharged material water content prediction method based on gradient lifting tree | |
| CN113642251B (en) | Data analysis and prediction method and system for building ceramic spray powder preparation quality | |
| CN104503402B (en) | Method for inspecting cigarette rolling quality stability in cigarette processing | |
| US20220142225A1 (en) | Intelligent control system and method of thin plate drier for cut tobacco | |
| CN107330555A (en) | Silk making process parameter weighting method based on random forest regression | |
| CN112327960B (en) | Intelligent control system for loosening and dampening equipment | |
| CN115936262B (en) | Yield prediction method, system and medium based on big data environment interference | |
| CN109342279B (en) | Mixed soft measurement method based on grinding mechanism and neural network | |
| CN111695730A (en) | ARIMA and RNN-based vertical mill vibration prediction method and device | |
| CN111667156B (en) | Method for evaluating physical quality consistency of cigarettes produced by multiple points | |
| CN112923727A (en) | Roasting furnace real-time furnace condition evaluation method based on temperature trend characteristic extraction | |
| CN115099457A (en) | On-line predicting and analyzing system for tobacco shred structure | |
| CN108519760A (en) | Filament making process steady-state identification method based on variable point detection theory | |
| CN108205432A (en) | A kind of real-time eliminating method of observation experiment data outliers | |
| CN111077876B (en) | Power station equipment state intelligent evaluation and early warning method, device and system | |
| CN117350028A (en) | Mushroom dryer intelligent management method and system based on data processing | |
| CN108052559A (en) | Distribution terminal defect mining analysis method based on big data processing | |
| CN105707974A (en) | Cigarette production process stability monitoring method | |
| CN116757354A (en) | Tobacco redrying section key parameter screening method based on multilayer perceptron | |
| CN116414095A (en) | Data-driven optimization method for technological parameters in traditional Chinese medicine manufacturing process | |
| CN106262996A (en) | A kind of Nicotiana tabacum L. processing technique controlled that is capable of homogenizing | |
| CN110910021A (en) | Method for monitoring online defects based on support vector machine | |
| CN106779322B (en) | Method for evaluating capacity index of part-counting value process obeying binomial distribution |
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 |