CN111444477A - Glass insulator formula automatic generation method based on gradient lifting regression model - Google Patents
Glass insulator formula automatic generation method based on gradient lifting regression model Download PDFInfo
- Publication number
- CN111444477A CN111444477A CN202010246612.4A CN202010246612A CN111444477A CN 111444477 A CN111444477 A CN 111444477A CN 202010246612 A CN202010246612 A CN 202010246612A CN 111444477 A CN111444477 A CN 111444477A
- Authority
- CN
- China
- Prior art keywords
- data
- glass insulator
- regression model
- formula
- raw material
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- 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)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Algebra (AREA)
- Operations Research (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Insulators (AREA)
Abstract
The invention provides a glass insulator formula automatic generation method based on a gradient lifting regression model, which adopts a big data analysis method combined with a machine-learned gradient lifting regression model to process historical data of a glass insulator, trains the gradient lifting regression model through the processed historical data, and then generates a required glass insulator formula only by inputting the purity of each pure object in each raw material in the required glass insulator formula into the trained gradient lifting regression model, thereby realizing the automatic generation of the glass insulator formula. Meanwhile, the yield of the glass insulator can be gradually improved in continuous iteration of experimental data and actual production.
Description
Technical Field
The invention relates to the technical field of glass insulators, in particular to a glass insulator formula automatic generation method based on a gradient lifting regression model.
Background
The insulator has excellent electrical insulation performance and stable mechanical performance, and is widely applied to power transmission lines. In practical application, most glass insulators are used, and the glass insulators have the characteristics of high mechanical strength, excellent insulating performance and thermal stability, and the chemical components of the glass insulators are key for determining the performance of the glass insulators.
At present, the formula generation of the glass insulator is mainly realized manually, the method is time-consuming and labor-consuming, is easy to make mistakes, and consumes a lot of unnecessary human resources. Deviation exists in each formula calculation, the yield of products is greatly fluctuated, and the automatic generation technology of the glass insulator formula is in a blank state.
Disclosure of Invention
In view of the above situation, the present invention provides an automatic generation method for a glass insulator formula based on a gradient lifting regression model, so as to solve the problems of time consuming, labor consuming and error prone in the manual generation of the glass insulator formula.
The technical scheme of the invention is as follows:
a glass insulator formula automatic generation method based on a gradient lifting regression model comprises the following steps:
extracting historical formula data of the glass insulator, wherein the historical formula data of the glass insulator comprises raw material purity data and batching quality data put into actual production, the raw material purity data is the purity percentage of each batching where each purified object is in the batching, and the batching quality data put into actual production corresponds to the raw material purity data one by one;
preprocessing the historical formula data of the glass insulator, merging the raw material purity data in the preprocessed historical formula data of the glass insulator and the batching quality data put into practical production according to the corresponding relation of time to obtain a data set, splitting the data set into a training set and a testing set, and training a gradient lifting regression model through the training set and the testing set;
and in the trained gradient lifting regression model, taking the purity of each pure object in each raw material in the required glass insulator formula as an input to obtain the quality of each raw material in the required glass insulator formula, and then normalizing the quality of each raw material to obtain the required glass insulator formula.
According to the automatic generation method of the glass insulator formula based on the gradient lifting regression model, firstly, historical data of a glass insulator is processed by a big data analysis method in combination with a machine-learned gradient lifting regression model, the gradient lifting regression model is trained through the processed historical data, and then the required glass insulator formula can be generated only by inputting the purity of each pure object in each raw material in the required glass insulator formula into the trained gradient lifting regression model, so that the automatic generation of the glass insulator formula is realized. Meanwhile, the yield of the glass insulator can be gradually improved in continuous iteration of experimental data and actual production.
In addition, the automatic generation method of the glass insulator formula based on the gradient lifting regression model, provided by the invention, also has the following technical characteristics:
further, the evaluation parameters of the gradient boost regression model include: decision coefficient, correction decision coefficient, mean absolute error, mean square error, root mean square error.
Further, in the step of preprocessing the historical formula data of the glass insulator, the following preprocessing is respectively performed on the raw material purity data and the batching quality data put into actual production:
and (4) filling the average value of the missing data, and deleting the repeated redundant data.
Further, the data sets are differentiated according to product types.
Further, the ratio of the training set to the test set is 8: 2.
Further, when the data set is divided into a training set and a test set, the data set is divided in a cross validation mode, and the number of cross times is set to 10.
Further, in the trained gradient boost regression model, the purity of each pure substance in each raw material in the required glass insulator formula is used as an input, and the quality of each raw material in the required glass insulator formula is obtained.
Drawings
FIG. 1 is a schematic diagram of the classification of four raw materials and the purity of each of the different raw materials.
Detailed Description
To facilitate an understanding of the invention, the invention will now be described more fully with reference to the accompanying drawings. Several embodiments of the invention are presented in the drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
The embodiment of the invention provides a glass insulator formula automatic generation method based on a gradient lifting regression model, which comprises the steps of S1-S3.
And S1, extracting historical formula data of the glass insulator, wherein the historical formula data of the glass insulator comprises raw material purity data and batching quality data put into actual production, the raw material purity data are the purity percentages of the respective batching of each pure object in the batching, and the batching quality data put into actual production correspond to the raw material purity data one by one.
In the present embodiment, after analyzing the ingredient formula of the glass insulator, the purity percentages of ingredients where fourteen pure substances in four raw materials are respectively located are extracted, and the extracted purity percentages are used as raw material purity data, as shown in fig. 1, the raw materials are divided into 4 types of mixtures, which are quartz sand (1), potassium feldspar (2), dolomite (3) and limestone (4), and the pure substances in the four raw materials are respectively silica _1, iron trioxide _1 and aluminum oxide _1 in the quartz sand (1); silicon dioxide _2, ferric oxide _2, aluminum oxide _2, potassium oxide _2 and sodium oxide _2 in the potassium feldspar (2); calcium oxide _3, ferric oxide _3, magnesium oxide _3 and aluminum oxide _3 in the dolomite (3); calcium oxide _4 and ferric oxide _4 in the limestone (4). It should be noted that the raw materials should also be soda ash, potassium carbonate, barium carbonate and mirabilite, and since these four raw materials are all industrial grade purities, the purity is 100% by default, and the purity analysis is not needed.
In addition, according to the time sequence, the quality data of the ingredients put into practical production are extracted and are in one-to-one correspondence with the purity data of the raw materials, and the data are historical formula data of the glass insulator.
S2, preprocessing the historical formula data of the glass insulator, merging the raw material purity data in the preprocessed historical formula data of the glass insulator and the batching quality data put into practical production according to the corresponding relation of time to obtain a data set, dividing the data set into a training set and a testing set, and training a gradient lifting regression model through the training set and the testing set.
In the step of preprocessing the historical formula data of the glass insulator, the following preprocessing is respectively carried out on the raw material purity data and the batching quality data put into practical production:
and (4) filling the average value of the missing data, and deleting the repeated redundant data.
The data set is used for training and testing a formula prediction regression model, and is obtained by analyzing and sorting historical production data of the glass insulator. Including historical purity data for each item of ingredient, as well as historical recipe data for that purity data. The data sets are differentiated according to product type. In this embodiment, the data sets are divided into 9 types (named as T8F, T8, T12F, T12, T16G, T16H, T30F, T30P, and T42, respectively) according to the existing product models, and the data set corresponding to each product model contains one thousand to several thousand pieces of data.
When the data set is split into a training set and a test set, the ratio of the training set to the test set is 8: 2. Specifically, the data set is divided by adopting a cross validation mode, and the number of cross is set to be 10.
The data information of the training set comprises the mass of 8 raw materials (adding up to 500kg of the mass of a pair of formula raw materials), the purity of 14 purified materials in four mixture raw materials of quartz sand, potassium feldspar, dolomite and limestone, and the time for the pair of formula raw materials to enter a factory for production.
In this embodiment, a Gradient Boost Regression (GBR) model is selected as a calculation model, and the accuracy of the trained GBR model is evaluated, wherein the evaluated parameters are R ^2 (decision coefficient), R ^2_ adjusted (correction decision coefficient), MSE (mean square error), RMSE (root mean square error), and MAE (mean absolute error). As shown in formulas (1), (2), (3), (4) and (5).
Wherein, the formula (1) represents a determination coefficient, the higher the goodness of fit, the higher the interpretation degree of the independent variable on the dependent variable, and the higher the percentage of the variation caused by the independent variable to the total variation. The denser the observation points are near the regression line. Equation (2) is a correction decision coefficient, and is a complement to equation (1), where n is the number of samples and p is the number of features. The decision coefficient R can be used for evaluating the quality of the regression equation, but the R is increased along with the increase of the number of independent variables. The correction decision coefficient of equation (2) is introduced to solve the problem that the decision coefficient R becomes larger when the independent variable is large. The MSE in the formula (3) is Mean Squared Error, which is an expected value of the square of the difference between a parameter estimation value and a parameter true value; the MSE can evaluate the change degree of the data, and the smaller the value of the MSE is, the better the accuracy of the prediction model for describing the experimental data is. The RMSE root mean square error of equation (4) is the arithmetic square root of the mean square error. The MAE in the formula (5) is the Mean Absolute Error, and the Mean Absolute Error is an average value of Absolute errors, and can better reflect the actual situation of predicted value errors. The experimental results in table 1 were obtained based on historical data for a glass insulator model number T12F. The experimental result shows that all evaluation parameters of the gradient lifting regression model are excellent in performance.
TABLE 1
Gradient lifting regression model | Quartz sand | Potassium feldspar | Dolomite | Limestone | Soda ash | Potassium carbonate | Barium carbonate | Natrii sulfas |
Cross validation results | 0.9976 | 0.9973 | 0.9984 | 0.9966 | 0.9987 | 0.9929 | 0.9992 | 0.9848 |
Determining coefficients | 0.9989 | 0.9991 | 0.9992 | 0.9987 | 0.9758 | 0.9925 | 0.9923 | 0.9931 |
Correction decision coefficient | 0.9977 | 0.9983 | 0.9984 | 0.9974 | 0.9511 | 0.9848 | 0.9874 | 0.986 |
Mean absolute error of MAE | 0.0199 | 0.0209 | 0.00111 | 0.0134 | 0.0021 | 0.0124 | 0.0137 | 0.0005 |
Mean square error of MSE | 0.0196 | 0.0184 | 0.0063 | 0.0091 | 0.0013 | 0.0047 | 0.0007 | 0.00002 |
RMSE root mean square error | 0.1399 | 0.1355 | 0.0796 | 0.0952 | 0.0356 | 0.0685 | 0.0264 | 0.004 |
And S3, in the trained gradient lifting regression model, taking the purity of each pure object in each raw material in the required glass insulator formula as input to obtain the quality of each raw material in the required glass insulator formula, and then normalizing the quality of each raw material to obtain the required glass insulator formula.
The purity of each kind of purified substances in the raw materials is used as input, wherein the purity of the purified substances in fourteen raw materials of silica _1, ferric oxide _1, aluminum oxide _1, silica _2, ferric oxide _2, aluminum oxide _2, potassium oxide _2, sodium oxide _2, calcium oxide _3, ferric oxide _3, magnesium oxide _3, aluminum oxide _3, calcium oxide _4 and ferric oxide _4 in the four raw materials of quartz sand (1), potash feldspar (2), dolomite (3) and limestone (4) is included. The other four raw material data (soda ash, potassium carbonate, barium carbonate and mirabilite) are pure substances by default. Since the regression model can only yield one value at a time, the model needs to be run eight times, each time to predict the quality of one material. And finally, the mass of each auxiliary material is fixed, and after the mass of the eight raw materials is predicted, the eight raw materials are normalized to obtain the required glass insulator formula.
For example, the purity of each pure type of the input raw material is divided into:
the purity of the silica _1 is 98.47, the purity of the iron trioxide _1 is 0.15, the purity of the aluminum oxide _1 is 0.77, the purity of the silica _2 is 66.94, the purity of the iron trioxide _2 is 0.73, the purity of the aluminum oxide _2 is 17.12, the purity of the potassium oxide _2 is 9.86, the purity of the sodium oxide _2 is 1.36, the purity of the calcium oxide _3 is 31.99, the purity of the iron oxide _3 is 0.2, the purity of the magnesium oxide _3 is 19.95, the purity of the aluminum oxide _3 is 0.01, the purity of the calcium oxide _4 is 55.57, and the purity of the iron trioxide _4 is 0.02.
And inputting the data into a trained gradient lifting regression model to finally obtain the formula of the glass insulator. The error of the obtained recipe data compared to the actual recipe data is shown in table 1.
In summary, according to the method for automatically generating a glass insulator formula based on the gradient lifting regression model provided by this embodiment, firstly, the historical data of the glass insulator is processed by using a big data analysis method in combination with the GBR gradient lifting regression model learned by the machine, the gradient lifting regression model is trained by using the processed historical data, and then, the required glass insulator formula can be generated only by inputting the purity of each pure object in each raw material in the required glass insulator formula into the trained gradient lifting regression model, so that the automatic generation of the glass insulator formula is realized. Meanwhile, the yield of the glass insulator can be gradually improved in continuous iteration of experimental data and actual production.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (7)
1. A glass insulator formula automatic generation method based on a gradient lifting regression model is characterized by comprising the following steps:
extracting historical formula data of the glass insulator, wherein the historical formula data of the glass insulator comprises raw material purity data and batching quality data put into actual production, the raw material purity data is the purity percentage of each batching where each purified object is in the batching, and the batching quality data put into actual production corresponds to the raw material purity data one by one;
preprocessing the historical formula data of the glass insulator, merging the raw material purity data in the preprocessed historical formula data of the glass insulator and the batching quality data put into practical production according to the corresponding relation of time to obtain a data set, splitting the data set into a training set and a testing set, and training a gradient lifting regression model through the training set and the testing set;
and in the trained gradient lifting regression model, taking the purity of each pure object in each raw material in the required glass insulator formula as an input to obtain the quality of each raw material in the required glass insulator formula, and then normalizing the quality of each raw material to obtain the required glass insulator formula.
2. The method for automatically generating a glass insulator formula based on a gradient lifting regression model according to claim 1, wherein the evaluation parameters of the gradient lifting regression model comprise: decision coefficient, correction decision coefficient, mean absolute error, mean square error, root mean square error.
3. The method for automatically generating a glass insulator formula based on a gradient lifting regression model according to claim 1, wherein in the step of preprocessing the historical formula data of the glass insulator, the raw material purity data and the batching quality data which are put into practical production are preprocessed as follows:
and (4) filling the average value of the missing data, and deleting the repeated redundant data.
4. The method of claim 1, wherein the data sets are differentiated according to product type.
5. The method for automatically generating a glass insulator formula based on a gradient lifting regression model according to claim 4, wherein the ratio of the training set to the testing set is 8: 2.
6. The method for automatically generating a glass insulator formula based on a gradient lifting regression model according to claim 5, wherein when the data set is divided into a training set and a testing set, the data set is divided in a cross validation mode, and the number of crossing times is set to 10.
7. The method according to claim 1, wherein in the step of obtaining the quality of each raw material in the glass insulator formulation, the quality of each raw material is obtained by running the gradient lifting regression model once, in the trained gradient lifting regression model, using the purity of each pure substance in each raw material in the glass insulator formulation as an input.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010246612.4A CN111444477B (en) | 2020-03-31 | 2020-03-31 | Glass insulator formula automatic generation method based on gradient lifting regression model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010246612.4A CN111444477B (en) | 2020-03-31 | 2020-03-31 | Glass insulator formula automatic generation method based on gradient lifting regression model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111444477A true CN111444477A (en) | 2020-07-24 |
CN111444477B CN111444477B (en) | 2023-03-21 |
Family
ID=71654032
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010246612.4A Active CN111444477B (en) | 2020-03-31 | 2020-03-31 | Glass insulator formula automatic generation method based on gradient lifting regression model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111444477B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100002021A1 (en) * | 2008-06-30 | 2010-01-07 | Nec Electronics Corporation | Display panel driving method and display apparatus |
CN108873829A (en) * | 2018-05-28 | 2018-11-23 | 上海新增鼎数据科技有限公司 | A kind of phosphoric acid production parameter control method promoting decision tree based on gradient |
CN110276241A (en) * | 2019-03-28 | 2019-09-24 | 广东工业大学 | A kind of stockbridge damper recognition methods based on template matching |
-
2020
- 2020-03-31 CN CN202010246612.4A patent/CN111444477B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100002021A1 (en) * | 2008-06-30 | 2010-01-07 | Nec Electronics Corporation | Display panel driving method and display apparatus |
CN108873829A (en) * | 2018-05-28 | 2018-11-23 | 上海新增鼎数据科技有限公司 | A kind of phosphoric acid production parameter control method promoting decision tree based on gradient |
CN110276241A (en) * | 2019-03-28 | 2019-09-24 | 广东工业大学 | A kind of stockbridge damper recognition methods based on template matching |
Non-Patent Citations (1)
Title |
---|
邱志贤: "关于污秽条件下交流绝缘子选取中的修正因数", 《电瓷避雷器》 * |
Also Published As
Publication number | Publication date |
---|---|
CN111444477B (en) | 2023-03-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109297689B (en) | Large-scale hydraulic machinery intelligent diagnosis method introducing weight factors | |
CN109255159B (en) | Circuit path delay fluctuation prediction method based on machine learning | |
Lauzon-Gauthier et al. | The Sequential Multi-block PLS algorithm (SMB-PLS): Comparison of performance and interpretability | |
CN107203690A (en) | A kind of GM(1,1)Model Transformer failure prediction method | |
CN103852513A (en) | Method and system based on HCD mass spectrogram and ETD mass spectrogram for peptide fragment de novo sequencing | |
CN112149342B (en) | Model method for optimizing quality of cement raw material | |
CN111444477B (en) | Glass insulator formula automatic generation method based on gradient lifting regression model | |
CN110705397A (en) | YOLOv3 pruning identification insulator defect method suitable for small field sample amount | |
CN114239921A (en) | Chip performance grade prediction and packaging method and device | |
CN110009062A (en) | Disaggregated model training method and device | |
CN111061246B (en) | Method for analyzing failure mode, influence and hazard in mechanical product assembly process | |
CN116861224A (en) | Intermittent process soft measurement modeling system based on intermittent process soft measurement modeling method | |
CN112509642B (en) | Online prediction method for blast furnace slag viscosity | |
CN115436342A (en) | Method and device for reducing LIBS detection uncertainty among multiple batches of samples | |
CN112949169B (en) | Coal sample test value prediction method based on spectral analysis | |
CN115187102A (en) | Model evaluation improvement method based on sinter quality prediction | |
CN112992284A (en) | Sintered mineral quality forecasting method based on PSO-ELM algorithm | |
CN112199854A (en) | Method for constructing efficiency analysis model of power industry | |
Etipola et al. | High fidelity clinker quality forecasting and controlling system using machine learning | |
EP3616012B1 (en) | Method and apparatus for estimating throughput of production line | |
Chang et al. | Composition analysis and identification of ancient glass products based on gray correlation | |
CN117388374B (en) | Method and system for detecting strength of concrete for building | |
CN113533133B (en) | Asphalt dynamic viscoelastic characteristic prediction method based on machine learning | |
Lan | Research on component content model of ancient glass products based on statistical analysis | |
Fan et al. | Automatic generation of glass insulator formulations based on time-scale uniformity |
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 |