Power station austenitic stainless steel seamless steel pipe quality dynamic analysis method and system

Abstract

The invention discloses a power station austenitic stainless steel seamless steel tube quality dynamic analysis method and system. The method comprises the steps that the grade of a new power station austenitic stainless steel seamless steel tube is judged; speculating the source of a new product; the correlation and difference of the power station austenitic stainless steel seamless steel tubes of different manufacturers in the aspect of detection attribute indexes are analyzed, and the method and the system can be used for carrying out quality analysis on the power station austenitic stainless steel seamless steel tubes.

Description

Technical Field

[0001] The present invention belongs to the technical field of inspection and analysis of power station metal materials, and relates to a method and system for dynamic analysis of the quality of austenitic stainless steel seamless steel pipes for power stations. Background Art

[0002] Austenitic stainless steel seamless steel pipes are important components of the heating surface of power station boilers and are widely used in high-temperature superheaters and reheaters. Due to the harsh service conditions of high temperature and high pressure, the quality of its products directly affects the service safety of power station boilers. Therefore, higher and higher requirements are put forward for the quality of austenitic stainless steel seamless steel pipes for power stations. At present, the products of S30432 and HR3C seamless steel pipes in austenitic stainless steel seamless steel pipes have gradually achieved localization. At this stage, the contradiction between supply and demand of austenitic stainless steel seamless steel pipe products has gradually transformed from a shortage of high-end products to a stable supply of high-quality products. The urgent demand for high-quality stainless steel pipe products has promoted the iron and steel industry to urgently improve product quality. The key work to improve product quality is to carry out product quality analysis and quality comparison, focus on quality shortboards, and clarify improvement goals.

[0003] At present, the quality analysis of austenitic stainless steel seamless steel pipes for power stations in the iron and steel industry mainly records data in real time through a laboratory or factory information management system, and then conducts manual comparison and analysis of charts, standards, etc. The process of data processing and further difference analysis is cumbersome and complex, and the in-depth excavation of product quality analysis and evaluation is insufficient and the application is weak. Especially with the development of big data technology, it is becoming more and more necessary to strengthen the application of informatization and intelligent data analysis in industrial production.

[0004] Therefore, in view of the above problems, it is urgent to propose an effective product quality analysis method, aiming at the historical precipitation data in the production manufacturers, purchasers and market product quality inspection information record databases of austenitic stainless steel seamless steel pipes, as well as the current production line and laboratory quality data, using a data-driven method, combined with advanced big data processing, analysis and visualization display technologies, deeply excavate and analyze product quality data, conduct quality grading evaluation, construct a quality grading evaluation system, promote the improvement of product quality of austenitic stainless steel seamless steel pipe manufacturing enterprises and the iron and steel industry, and continuously meet the urgent needs of China's power industry for high-quality austenitic stainless steel pipes. Summary of the Invention

[0005] The purpose of the present invention is to overcome the above-mentioned disadvantages of the prior art, and provide a method and system for dynamic analysis of the quality of austenitic stainless steel seamless steel pipes for power stations, which can analyze the quality of austenitic stainless steel seamless steel pipes for power stations.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions:

[0007] On one hand, the present invention provides a method for dynamic quality analysis of seamless austenitic stainless steel pipes for power plants, including:

[0008] Judging the grade of newly incoming seamless austenitic stainless steel pipes for power plants;

[0009] Speculating the source of newly incoming products;

[0010] Analyzing the correlation and differences in detection attribute indexes of seamless austenitic stainless steel pipes for power plants from different manufacturers.

[0011] The process of judging the grade of newly incoming seamless austenitic stainless steel pipes for power plants is as follows:

[0012] Preliminarily classifying the sample data of seamless austenitic stainless steel pipes for power plants according to the product manufacturers in the database;

[0013] According to the characteristics and correlations of product attribute indexes, using a clustering model to divide the preliminarily classified index sample data into several categories;

[0014] Determining the weights and scores of each category of each index according to product characteristics, calculating the grade score of the product, and judging the grade of newly incoming seamless austenitic stainless steel pipes for power plants based on this.

[0015] The process of speculating the source of newly incoming products is as follows:

[0016] According to the integrity, index importance and manufacturer characteristics of product sample data, screening out the attribute indexes required to establish the matching models for each manufacturer and establishing a characteristic index system;

[0017] For the sample data of each product manufacturer for which a model needs to be established and the sample data of other manufacturers, using the Smote algorithm to perform sample balancing processing to obtain the balanced sample data of the product;

[0018] Based on the established characteristic index system and the balanced sample data of the product, using the Light GBM algorithm to establish the matching models for each manufacturer;

[0019] Based on the matching models for each manufacturer, analyzing the similarity and matching probability of the newly incoming product with the products of each manufacturer, and speculating the source of the newly incoming product.

[0020] On the other hand, the present invention provides a dynamic quality analysis system for seamless austenitic stainless steel pipes for power plants, including:

[0021] A clustering analysis subsystem for judging the grade of newly incoming seamless austenitic stainless steel pipes for power plants;

[0022] A production model analysis subsystem for speculating the source of newly incoming products;

[0023] A correlation analysis subsystem for analyzing the correlation and differences in the detection attribute indicators of austenitic stainless steel seamless steel pipes for power stations from different manufacturers.

[0024] The clustering analysis subsystem includes:

[0025] A source classification module for preliminarily classifying the sample data of austenitic stainless steel seamless steel pipes for power stations according to the product manufacturers in the database;

[0026] A clustering processing module for dividing the preliminarily classified index sample data into several categories by using a clustering model according to the characteristics and correlations of the product attribute indicators;

[0027] A product grade analysis module for determining the weights and scores of each category of each index according to the product characteristics, calculating the grade score of the product, and judging the grade of the newly introduced austenitic stainless steel seamless steel pipes for power stations based on this.

[0028] The production model analysis subsystem includes:

[0029] An index screening module for screening out the attribute indicators required to establish the matching models for each manufacturer according to the integrity of the product sample data, the importance of the indicators, and the characteristics of the manufacturers, and establishing a characteristic index system;

[0030] A sample balancing module for performing sample balancing processing on the sample data of each product manufacturer that needs to establish a model and the sample data of other manufacturers by using the Smote algorithm to obtain the balanced product sample data;

[0031] A model establishment module for establishing the matching models for each manufacturer based on the established characteristic index system and the balanced product sample data by using the Light GBM algorithm;

[0032] A matching analysis module for analyzing the similarity and matching probability of the newly introduced product with the products of each manufacturer based on the matching models for each manufacturer, and inferring the source of the newly introduced product.

[0033] The correlation analysis subsystem includes:

[0034] An index correlation module for analyzing the correlation between the detection attribute indicators of the same manufacturer according to the product sample data;

[0035] A manufacturer correlation module for analyzing the correlation and differences in the detection attribute indicators of austenitic stainless steel seamless steel pipes for power stations from different manufacturers.

[0036] The clustering processing module includes:

[0037] An attribute index selection unit, configured to determine product detection indexes for cluster analysis according to the integrity of product sample data, the importance of quality evaluation influencing factors, and product characteristics;

[0038] A data processing unit, configured to process the sample data of the product detection indexes based on the Minitab data analysis method;

[0039] A cluster analysis unit, configured to use a Q-type clustering model to subdivide the processed sample data of each index into several categories in combination with standard requirements.

[0040] The product grade analysis module includes:

[0041] A newly introduced product grade evaluation unit, configured to obtain the overall product score by calculating the weight of a single index and the scores of each category of the corresponding index according to the index category analysis, and judge the grade of the product based on this;

[0042] An early warning unit, configured to give a warning when the grade of the product falls into the category of deviation products.

[0043] The correlation analysis subsystem uses the KMO test algorithm and the Pearson correlation coefficient method for correlation analysis.

[0044] The present invention has the following beneficial effects:

[0045] When the quality dynamic analysis method and system of the power station austenitic stainless steel seamless steel pipe described in the present invention are specifically operated, by deeply mining the product database, analyzing the correlation between product attribute indexes, and adopting a quantitative evaluation method, a quality ability grading evaluation result of a specific product grade is formed, optimizing the threshold-type quality evaluation system, establishing the mutual relationship between products and manufacturers, promoting the quality improvement of enterprises and industries, and enhancing the influence of domestic brands. Description of the Drawings

[0046] The specification drawings forming a part of the present invention are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0047] Figure 1 is the system structure diagram of the present invention;

[0048] Figure 2 is the working principle diagram of the cluster analysis subsystem 1. Detailed Embodiments

[0049] To enable those skilled in the art to better understand the solution of the present invention, the following will clearly and completely describe the technical solution in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments, and are not intended to limit the scope of the present invention disclosure. In addition, in the following description, the description of known structures and technologies is omitted to avoid unnecessarily confusing the concepts disclosed in the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0050] The schematic structural diagrams according to the disclosed embodiments of the present invention are shown in the accompanying drawings. These figures are not drawn to scale, where for the purpose of clear expression, some details are enlarged and some details may be omitted. The shapes of various regions and layers shown in the figures, as well as their relative sizes and positional relationships, are only exemplary. In practice, there may be deviations due to manufacturing tolerances or technical limitations, and those skilled in the art can design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0051] Embodiment 1

[0052] The quality dynamic analysis system for austenitic stainless steel seamless steel pipes in power stations according to the present invention includes:

[0053] Judging the grade of newly incoming austenitic stainless steel seamless steel pipes in power stations;

[0054] Speculating the source of newly incoming products;

[0055] Analyzing the correlation and differences in the detection attribute indicators of austenitic stainless steel seamless steel pipes in power stations from different manufacturers.

[0056] The process of judging the grade of newly incoming austenitic stainless steel seamless steel pipes in power stations is as follows:

[0057] Pre - classifying the sample data of austenitic stainless steel seamless steel pipes in power stations according to the product manufacturers in the database;

[0058] According to the characteristics and correlations of product attribute indicators, using a clustering model to divide the pre - classified index sample data into several categories;

[0059] Determine the weights and scores of each category of each index according to product characteristics, calculate the grade score of the product, and judge the grade of the newly incoming austenitic stainless steel seamless steel pipes in power stations based on this.

[0060] The process of speculating the source of newly incoming products is as follows:

[0061] According to the integrity of product sample data, the importance of indicators, and the characteristics of manufacturers, select the attribute indicators required to establish the matching models for each manufacturer, and establish a characteristic index system;

[0062] For the sample data of each product manufacturer for which a model needs to be established and the sample data of other manufacturers, use the Smote algorithm to perform sample balancing processing to obtain the balanced product sample data;

[0063] Based on the established characteristic index system and the balanced product sample data, use the Light GBM algorithm to establish the matching models for each manufacturer;

[0064] Based on the matching models of each manufacturer, analyze the similarity and matching probability between the new product and the products of each manufacturer, and infer the source of the new product.

[0065] Embodiment 2

[0066] Reference Figure 1 , Figure 1 is the structural schematic diagram of the present invention. As Figure 1 shown, the system mainly includes: 1 clustering analysis subsystem 1, 2 production model analysis subsystem 2, and 3 correlation analysis subsystem 3. The following will separately elaborate on 1 clustering analysis subsystem 1, 2 production model analysis subsystem 2, and 3 correlation analysis subsystem 3. Figure 2 is the working principle diagram of the clustering analysis subsystem 1.

[0067] As Figure 2 can be seen, the clustering analysis subsystem 1 mainly, based on the classification of product manufacturers, optimizes important attribute indicators through the system, performs clustering using the Minitab processing and Q-type clustering analysis methods, generally divides each attribute indicator into different categories, based on the value range of the categories, uses scores and weights to divide the grades of new products, and at the same time gives early warnings for defective products, discovers problems in advance, and improves the risk control ability.

[0068] Based on the above principle, this embodiment provides a clustering analysis subsystem 1, and the clustering analysis subsystem 1 mainly includes:

[0069] A source division module 11, used to preliminarily classify the sample data of power station austenitic stainless steel seamless steel pipes according to the product manufacturers in the database;

[0070] A clustering processing module 12, used to further subdivide the sample data of each index after preliminary classification into several categories using a clustering model according to the characteristics and correlations of the product attribute indicators;

[0071] The product grade analysis module 13 is used to determine the weights and scores of each category of each index according to the product characteristics, calculate the product grade score, and judge the grade of the newly entered power station austenitic stainless steel seamless steel pipe;

[0072] Taking the 10Cr18Ni9NbCu3BN stainless steel pipe as an example, the detailed process of the above-mentioned clustering analysis subsystem 1 for identifying and analyzing the sample attribute indexes and product grades is described below, and the composition of each functional module is further explained.

[0073] (I) Regarding data acquisition

[0074] In this embodiment, the data acquisition module 01 runs through each subsystem, and is mainly used to obtain product sample data for clustering, modeling, and index analysis in the large database of the information management system. The information management system has integrated the historical data of inspection information records, the current production line, and the data of the laboratory quality inspection link, and has accumulated a large amount of product sample data, and the inspection standards, methods information, and manufacturer historical information can all be obtained in the system. In this embodiment, the data acquisition module 01 mainly includes the following units:

[0075] The data cleaning unit 011 is used to perform preliminary screening on the received product index detection data, delete outliers, values with magnitudes that do not conform to the actual values, and error values, to ensure the accuracy and reliability of subsequent results; for example, negative values appear, or three-digit values appear for temperature-related values.

[0076] The data parsing unit 012 is used to unify the dimension or perform structural transformation on the cleaned data, so that the sample data has a unified format and can be directly applied for processing subsequently. For example, hardness HRB / HBW / HV needs to be converted to a unified dimension.

[0077] (II) Regarding source division

[0078] In this embodiment, the source division module 11 is used to preliminarily divide the product sample data according to the production source of the stainless steel pipe; for example, for the 10Cr18Ni9NbCu3BN stainless steel pipe, the billet manufacturer is A, the steel pipe manufacturer is S, and the subsequent processor is B. The source division is preliminarily divided into the S factory according to the manufacturing producer.

[0079] (III) Regarding clustering processing

[0080] In this embodiment, the clustering processing module 12 mainly includes the following units:

[0081] The attribute index selection unit 121 is used to determine the product detection indexes for clustering analysis according to the integrity of the product sample data, the importance of quality evaluation influencing factors, and the product characteristics;

[0082] A data processing unit 122 for processing the sample data of the product detection indicators based on the Minitab data analysis method;

[0083] A clustering analysis unit 123 for further subdividing the processed sample data of each indicator into several categories by using a Q-type clustering model in combination with standard values.

[0084] The clustering processing module 12 explores the sample attribute indicators of the products of each manufacturer in the database without any prior knowledge, selects the attribute indicators with relatively complete data, great influence on quality, and representing product characteristics, and further clusters and subdivides them into different grades according to the distribution range, standard value, confidence interval, standard deviation, etc. of the indicator values.

[0085] In addition, since the Q-type clustering algorithm makes samples with similar characteristics gather together and separates those with large differences by the degree of difference between individuals (measured by distance), it is necessary to unify the dimension of the indicator values to reduce the influence of dimension difference on the clustering effect.

[0086] Combined with the aforementioned source division situation, the clustering analysis unit 123 is used for further clustering and subdivision of the corresponding attribute indicator values of each manufacturer, clearly presenting the distribution of the indicator values of different manufacturers, and quantifying the evaluation of the grades of newly introduced products.

[0087] (4) Regarding product grade analysis

[0088] The product grade analysis module 13 determines the product grade by analyzing the product detection indicator values, clarifying the categories after the indicator clustering division, and weighted calculating the total score of all indicators. For example, if one of the indicator value categories is a deviation product, the warning system alarms, the score is displayed as 0, and it is directly archived as a deviation product. In other cases, only the total score needs to be concerned, and the product grade will be directly displayed, and the details can be consulted.

[0089] The following is the product grade analysis result obtained in this embodiment.

[0090] 1. Clustering attribute indicators

[0091] According to the integrity of the product sample data, the importance of quality evaluation influencing factors, and product characteristics, the attribute indicators for the clustering analysis of 10Cr18Ni9NbCu3BN stainless steel pipes are: "cold rolling deformation amount", "rolling passes", "solution temperature", "chemical composition elements", "tensile strength", "yield strength", "hardness", "elongation after fracture", "grain size", "grain elongation".

[0092] 2. Indicator related values

[0093] Use Minitab to analyze and process the correlation values of the above clustering indicators for different manufacturers: maximum value, minimum value, mean value, standard deviation, obtain their respective overall distribution intervals, and calculate the 95% confidence interval.

[0094] 3. Clustering results

[0095] For the data of each indicator after unifying the dimension, use the Q-type clustering model to cluster the indicator values into 4 categories with relatively clear boundaries: deviation value, lower limit value, median value, and upper limit value.

[0096] 4. Grade analysis

[0097] Determine the category of each indicator of the newly introduced product, calculate the score of each indicator using the weighted method, calculate the comprehensive score S, and evaluate the grade. The calculation formula is:

[0098]

[0099] S = 0 is a deviation product, S = 60 is a lower limit product, S = 75 is a median product, S = 85 is an upper limit product

[0100] Next, introduce the production model analysis subsystem 2 proposed in this embodiment. The production model analysis subsystem 2 includes:

[0101] An index screening module 21, which is used to screen the attribute indexes required to establish the matching models of each manufacturer according to the product sample data and establish a characteristic index system;

[0102] A sample balancing module 22, which is used to perform sample balancing processing on the sample data of each product manufacturer that needs to establish a model and the sample data of other manufacturers by using the Smote algorithm;

[0103] A model establishment module 23, which is used to establish the matching models of each manufacturer by preferentially using the Light GBM algorithm based on the established characteristic index system and the balanced product sample data;

[0104] A matching analysis module 24, which is used to analyze the similarity and matching probability between the newly introduced product and the products of each manufacturer based on the formed matching model and infer its source.

[0105] Next, take the 10Cr18Ni9NbCu3BN stainless steel pipe as an example to describe the detailed process of analyzing the similarity and matching probability between the newly introduced product and the products of each manufacturer by the above system, and further explain the composition of each functional module of the system.

[0106] (1) Regarding data acquisition

[0107] In this embodiment, the production model analysis subsystem 2 further includes a data acquisition module 01, which is mainly used to acquire product sample data from different manufacturers. The data acquisition module 01 mainly includes the following units:

[0108] A data cleaning unit 011, which is used to perform a preliminary screening on the received product index detection data, delete outliers, values with magnitudes that do not conform to the actual values, and error values, to ensure the accuracy and reliability of subsequent results; for example, the appearance of negative values, three-digit values for temperature-related values, etc.

[0109] A data parsing unit 012, which is used to unify the dimension or perform a structural transformation on the cleaned data, so that the sample data has a unified format and can be directly applied for processing subsequently. For example, hardness HRB / HBW / HV needs to be converted to a unified dimension.

[0110] In this embodiment, large manufacturers are preferably selected, and their product sample data is relatively large, which is sufficient to characterize product features to establish a product manufacturer matching model with a relatively high accuracy. Therefore, in this embodiment, an oil product manufacturer matching model is only established for manufacturers with a relatively large amount of product sample data. When the product sample data of other manufacturers accumulates to a certain quantity, a model is established in a similar method.

[0111] After statistics, the product batch quantities of the top 6 manufacturers of 10Cr18Ni9NbCu3BN stainless steel pipes currently account for 91% of the total sample quantity. In this embodiment, only the historical product sample data of these top 6 manufacturers is collected.

[0112] (II) Regarding index screening

[0113] In this embodiment, the processed sample data is analyzed. The index screening module 21 preferably screens the attribute indexes required to establish the matching models for each manufacturer according to the integrity of the sample data, the importance of the indexes, and the characteristics of the manufacturers, and establishes a characteristic index system. Among them, in addition to the original indexes, two important indexes, namely 10 5 h creep strength and dimensional tolerance, are also selected. Specifically, in this embodiment, the indexes determined for establishing the manufacturer matching model are: "cold rolling deformation amount", "rolling passes", "solution temperature", "chemical composition elements", "tensile strength", "yield strength", "hardness", "elongation after fracture", "grain size", "grain elongation", "10 5 h creep strength", "dimensional tolerance". Among them: the calculation method of "10 5 h creep strength" is: extrapolate the 10 5 h creep strength using the L-M parameter method, and the calculation method of "dimensional tolerance" is: the difference between the actual size and the designed size, and calculate the positive and negative deviation value ranges.

[0114] (III) Regarding sample balance

[0115] In this embodiment, for the problem of unbalanced sample ratio in the modeling process, the sample balancing module 22 preferably uses the Smote algorithm to balance the positive and negative samples, so as to improve the model accuracy.

[0116] (IV) Regarding model establishment and matching analysis

[0117] The model establishment module 23 preferably uses the Light GBM algorithm.

[0118] It is applicable to large datasets and is still very robust in the face of problems such as data missing and a large number of input fields. In this embodiment, taking manufacturer S as an example, after balancing the samples, 70% is taken as the training set, a classification model is established on the training set, 30% is taken as the test set, and predictions are made on the test set to evaluate the effect of the established classification model.

[0119] First, taking manufacturer S as an example, a classification model is established for it. The main differences between a certain manufacturer and other manufacturers are manifested in characteristic indicators such as "cold rolling deformation amount", "solution temperature", "five harmful elements", "tensile strength", "yield strength", "hardness", "grain size", "grain elongation", "10 5 h creep strength", etc., and each index is presented in the form of a bar chart according to the importance degree. The effect of the model is evaluated by calculating the precision rate and recall rate of the model on the test set.

[0120] Through the manufacturer matching model, the matching probability between the newly introduced product and the manufacturer is determined, which is used to characterize the product similarity between the newly introduced product and the corresponding manufacturer.

[0121] The following introduces the correlation analysis subsystem 3 proposed in this embodiment. The correlation analysis subsystem 3 mainly includes:

[0122] The index correlation module 31 is used to analyze the correlation between the detection attribute indexes of the same manufacturer according to the product sample data;

[0123] The manufacturer correlation module 32 is used to analyze the correlation and differences in the detection attribute indexes of the austenitic stainless steel seamless steel pipes of different manufacturers in power stations.

[0124] (I) Regarding index correlation analysis

[0125] Index correlation analysis mainly explores the correlation relationship between detection indexes and discovers indexes with strong correlation. The relationship between quantitative variables is described by the correlation coefficient. The sign (±) of the correlation coefficient indicates the direction of the relationship (positive correlation or negative correlation), and the magnitude of its value indicates the strength of the relationship.

[0126] In this embodiment, the KMO test algorithm is adopted to quantify the dependence relationship between product detection indicators and find out the indicators with strong correlation. Combining the data volume of each detection indicator, the index correlation analysis is carried out on 13 indicators of stainless steel pipes in total. The specific indicators include: "cold rolling deformation amount", "rolling passes", "solution temperature", "main component elements", "control amount of five harmful elements", "tensile strength", "yield strength", "hardness", "elongation after fracture", "grain size", "grain elongation", "10 5 h creep strength" and "dimensional tolerance range".

[0127] The correlation analysis of manufacturers is completed based on the index correlation, and the basic method is the same as that of the index correlation analysis method.

[0128] To sum up the above analysis, the quality dynamic analysis system of power station austenitic stainless steel seamless pipes based on big data provided by this embodiment can simplify and intelligentize the data processing process, deeply mine the product database, analyze the correlation between product attribute indicators, adopt a quantitative evaluation method, form a quality ability grading evaluation result of a specific product grade, optimize the "threshold" quality evaluation system, establish the mutual relationship between products and manufacturers, promote the quality improvement of enterprises and industries, and improve the influence of domestic brands.

[0129] Embodiment 3

[0130] Taking the 07Cr25Ni21NbN stainless steel pipe as an example, this embodiment provides a clustering analysis subsystem 1. The methods for data acquisition, source division, clustering processing, and product grade analysis are all the same as above. The following are the product grade analysis results obtained in this embodiment.

[0131] 1. Clustering attribute indicators

[0132] According to the integrity of product sample data, the importance of quality evaluation influencing factors, and product characteristics, the attribute indicators for the clustering analysis of 07Cr25Ni21NbN stainless steel pipes are: "rolling ratio", "solution temperature", "chemical composition elements", "tensile strength", "yield strength", "hardness", "impact toughness", "elongation after fracture", and "grain size".

[0133] 2. Index correlation values

[0134] Use Minitab to analyze and process the correlation values of the above clustering indicators of different manufacturers: maximum value, minimum value, mean value, standard deviation, obtain their respective overall distribution intervals, and calculate the 95% confidence interval.

[0135] 3. Clustering results

[0136] For the indicator data after unification of dimensions, the Q-type clustering model is used to cluster the indicator values ​​into four categories with relatively clear boundaries, namely deviation value, lower limit value, median value, and upper limit value.

[0137] 4. Grade analysis

[0138] Determine the category of each indicator of the new product, use the weighted method to calculate the score of each indicator, calculate the comprehensive score S, and evaluate the grade. The calculation formula is:

[0139]

[0140] S=0 is a deviation product, S=60 is a lower limit product, S=75 is a median product, and S=85 is an upper limit product.

[0141] Taking 07Cr25Ni21NbN stainless steel pipe as an example, the detailed process of the similarity and matching probability analysis between the new product and the products of each manufacturer by the production model analysis subsystem 2 is described. The data acquisition, indicator screening, sample balance analysis method, model establishment and matching analysis method are the same as above.

[0142] According to statistics, the number of product batches from the top four manufacturers of 07Cr25Ni21NbN stainless steel pipes currently accounts for 90% of the total sample volume. In this embodiment, only the historical product sample data of the top four manufacturers are collected.

[0143] In this embodiment, the processed sample data is analyzed to establish the attribute indicators required by the matching model of each manufacturer and to establish a characteristic indicator system. In addition to the original indicators, 10 5 h endurance strength and dimensional tolerance are two important indicators. Specifically, in this embodiment, the indicators used to establish the manufacturer matching model are: "rolling ratio", "solution temperature", "chemical elements", "tensile strength", "yield strength", "hardness", "impact toughness", "elongation after fracture", "grain size", "10 5 h endurance strength", "dimensional tolerance". Among them: "10 5 The calculation method of "h endurance strength" is: use the LM parameter method to extrapolate 10 5 h endurance strength, the calculation method of "dimensional tolerance" is: the difference between the actual size and the design size, and the calculation range of positive and negative deviation values.

[0144] Take manufacturer W as an example, and establish a classification model for it. The main differences between a manufacturer and other manufacturers are reflected in "rolling ratio", "solid solution temperature", "five harmful elements", "tensile strength", "yield strength", "hardness", "grain size", "10 5Characteristic indicators such as "long-term strength", and each indicator is presented in the form of a bar chart according to the importance. The model effect is evaluated by calculating the precision and recall rate of the model on the test set.

[0145] Through the manufacturer matching model, determine the matching probability between the newly introduced product and the manufacturer, which is used to characterize the product similarity between the newly introduced product and the corresponding manufacturer's product.

[0146] Taking the 07Cr25Ni21NbN stainless steel pipe as an example, describe the detailed process of the production model analysis subsystem 2 for analyzing the similarity and matching probability between the newly introduced product and the products of each manufacturer.

[0147] Next, introduce the correlation analysis subsystem 3 proposed in this embodiment.

[0148] Taking the 07Cr25Ni21NbN stainless steel pipe as an example, describe the correlation analysis subsystem 3.

[0149] Adopt the KMO test algorithm to quantify the dependence relationship between product detection indicators and find the indicators with strong correlation. Combine the data volume of each detection indicator to conduct an index correlation analysis on 11 indicators of stainless steel pipes. The specific indicators include: "rolling ratio", "solution temperature", "main component elements", "control amount of five harmful elements", "tensile strength", "yield strength", "hardness", "elongation after fracture", "grain size", "10 at 650°C 5 h long-term strength", "dimension tolerance range". The manufacturer correlation analysis is completed based on the index correlation, and the basic method is the same as the index correlation analysis method.

[0150] To sum up the above analysis, the quality dynamic analysis system of power station austenitic stainless steel seamless pipes based on big data provided in this embodiment can simplify and intelligentize the data processing process, deeply mine the product database, analyze the correlation between product attribute indicators, adopt a quantitative evaluation method, form a quality ability grading evaluation result of a specific product grade, optimize the "threshold" quality evaluation system, establish the mutual relationship between products and manufacturers, promote the quality improvement of enterprises and the industry, and improve the influence of domestic brands.

[0151] The present invention has the following characteristics:

[0152] In the present invention, the clustering analysis subsystem 1 can identify and analyze the product attribute characteristics in the database. Through data processing and clustering analysis methods, combined with standard values, the attribute indicators are divided into various categories (upper limit value, median value, lower limit value, deviation value, etc.), and the characteristics of each manufacturer's attribute indicator categories are described in detail. The scores are calculated using scores and weights to obtain the product grade.

[0153] In the present invention, the clustering analysis subsystem 1 can characterize the product features of manufacturers and the product differences between different manufacturers from two aspects: the distribution of manufacturers in the database and the distribution of product indicators of manufacturers, and at the same time can give early warnings about the grades of newly detected product deviations, discover problems in advance, improve the risk control ability, and optimize the product quality.

[0154] In the present invention, the production model analysis subsystem 2 can screen out product attribute indicators according to product sample data, establish a characteristic index system, perform sample balancing processing for each product manufacturer that needs to establish a model, and then, based on the established characteristic index system and the balanced product sample data, preferably use the Light GBM algorithm to establish matching models for each manufacturer, analyze the similarity and matching probability between newly introduced products and the products of each manufacturer, and infer their sources.

[0155] In the present invention, the correlation analysis subsystem 3 can analyze the correlation between detection indicators and the correlation and differences in detection attribute indicators of austenitic stainless steel seamless steel pipes in power stations of different manufacturers, automatically judge the rationality of detection indicators according to the correlation, and perform index comparison analysis according to the strength of the correlation, which is beneficial to further characterize the product features and differences of each manufacturer.

[0156] The division of modules in the embodiments of the present application is illustrative, merely a logical function division. In actual implementation, there may be other division methods. In addition, in each embodiment of the present application, each functional module can be integrated in a processor, can also exist physically alone, or two or more modules can be integrated in one module. The above integrated modules can be implemented in the form of hardware or in the form of software function modules.

[0157] Embodiment Four

[0158] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method for dynamically analyzing the quality of austenitic stainless steel seamless steel pipes for power plants are implemented. For example, it includes: judging the grade of newly incoming austenitic stainless steel seamless steel pipes for power plants; inferring the source of the newly incoming products; analyzing the correlation and differences in the detected attribute indexes of austenitic stainless steel seamless steel pipes from different manufacturers. Among them, the memory may include a memory, such as a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk memory, etc.; the processor, network interface, and memory are interconnected through an internal bus, and this internal bus can be an Industry Standard Architecture bus, a Peripheral Component Interconnect standard bus, an Extended Industry Standard Architecture bus, etc., and the bus can be divided into an address bus, a data bus, a control bus, etc. The memory is used to store programs. Specifically, the program can include program code, and the program code includes computer operation instructions. The memory can include a memory and a non-volatile memory, and provides instructions and data to the processor.

[0159] Embodiment Five

[0160] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for dynamically analyzing the quality of austenitic stainless steel seamless steel pipes for power plants are implemented. For example, it includes: judging the grade of newly incoming austenitic stainless steel seamless steel pipes for power plants; inferring the source of the newly incoming products; analyzing the correlation and differences in the detected attribute indexes of austenitic stainless steel seamless steel pipes from different manufacturers. Specifically, the computer-readable storage medium includes, but is not limited to, for example, a volatile memory and / or a non-volatile memory. The volatile memory can include a random access memory (RAM) and / or a cache memory, etc. The non-volatile memory can include a read-only memory (ROM), a hard disk, a flash memory, an optical disc, a magnetic disk, etc.

[0161] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program code.

[0162] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices produce a means for implementing the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 a means for implementing the functions specified in one block or multiple blocks.

[0163] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction means, and the instruction means implements the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 a means for implementing the functions specified in one block or multiple blocks.

[0164] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 a means for implementing the functions specified in one block or multiple blocks.

[0165] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: it is still possible to modify the specific implementation manners of the present invention or make equivalent replacements, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the protection scope of the claims of the present invention.

Claims

1. A dynamic analysis method for the quality of austenitic stainless steel seamless pipes for power plants, characterized in that: include: Determine the grade of austenitic stainless steel seamless pipes newly delivered to power stations; Speculate on the origin of new products; Analyze the correlation and differences in the detection property indicators of power plant austenitic stainless steel seamless pipes from different manufacturers.

2. The dynamic analysis method for the quality of austenitic stainless steel seamless pipes for power plants according to claim 1 is characterized in that: The process of judging the grade of the newly-imported austenitic stainless steel seamless steel pipe is as follows: Preliminary classification of power station austenitic stainless steel seamless pipe sample data according to product manufacturers in the database; According to the characteristics and relevance of product attribute indicators, the clustering model is used to divide the indicator sample data that has been initially classified into several categories; The weight and score of each category of each indicator are determined according to the product characteristics, the grade score of the product is calculated, and the grade of the new austenitic stainless steel seamless steel pipe for the power station is judged based on this.

3. The dynamic analysis method for quality of austenitic stainless steel seamless pipes for power plants according to claim 1 is characterized in that: The process of inferring the source of new products is as follows: According to the completeness of product sample data, the importance of indicators and the characteristics of manufacturers, select the attribute indicators required to establish the matching model of each manufacturer, and establish a characteristic indicator system; For each product manufacturer sample data that needs to establish a model and other manufacturer sample data, the Smote algorithm is used to perform sample balancing to obtain the balanced product sample data; Based on the established feature index system and balanced product sample data, the Light GBM algorithm is used to establish matching models for each manufacturer; Based on the matching models of each manufacturer, the similarity and matching probability between the new product and the products of each manufacturer are analyzed to infer the source of the new product.

4. A dynamic analysis system for the quality of austenitic stainless steel seamless pipes in power plants, characterized in that: include: Cluster analysis subsystem (1), used to determine the grade of austenitic stainless steel seamless pipes newly delivered to the power station; Production model analysis subsystem (2), used to infer the source of new incoming products; The correlation analysis subsystem (3) is used to analyze the correlation and differences in the detection property indicators of power plant austenitic stainless steel seamless pipes from different manufacturers.

5. The power plant austenitic stainless steel seamless pipe quality dynamic analysis system according to claim 4 is characterized in that: The cluster analysis subsystem (1) comprises: A source classification module (11) is used to preliminarily classify the sample data of power station austenitic stainless steel seamless pipes according to the product manufacturers in the database; A clustering processing module (12) is used to classify the index sample data that has been preliminarily classified into several categories using a clustering model according to the characteristics and relevance of the product attribute indicators; The product grade analysis module (13) is used to determine the weight and score of each category of each indicator according to the product characteristics, calculate the product grade score, and use it to judge the grade of the newly-imported power station austenitic stainless steel seamless steel pipe.

6. The power plant austenitic stainless steel seamless pipe quality dynamic analysis system according to claim 4 is characterized in that: The production model analysis subsystem (2) comprises: An indicator screening module (21) is used to screen out the attribute indicators required for establishing the matching model of each manufacturer according to the completeness of the product sample data, the importance of the indicators and the characteristics of the manufacturer, and to establish a characteristic indicator system; A sample balancing module (22) is used to perform sample balancing processing using a Smote algorithm on sample data of each manufacturer of products for which a model needs to be established and sample data of other manufacturers, thereby obtaining balanced product sample data; A model building module (23) is used to build matching models for each manufacturer using the Light GBM algorithm based on the established feature index system and the balanced product sample data; The matching analysis module (24) is used to analyze the similarity and matching probability between the new product and the products of each manufacturer based on the matching models of each manufacturer, and to infer the source of the new product.

7. The power plant austenitic stainless steel seamless pipe quality dynamic analysis system according to claim 4 is characterized in that: The correlation analysis subsystem (3) comprises: An indicator correlation module (31) is used to analyze the correlation between the detection attribute indicators of the same manufacturer based on product sample data; The manufacturer correlation module (32) is used to analyze the correlation and difference in the detection attribute indicators of power plant austenitic stainless steel seamless pipes from different manufacturers.

8. The power station austenitic stainless steel seamless pipe quality dynamic analysis system according to claim 5 is characterized in that: The cluster processing module (12) comprises: An attribute index selection unit (121) is used to determine product testing indicators for cluster analysis based on the integrity of product sample data, the importance of quality evaluation influencing factors and product characteristics; A data processing unit (122), used for processing the sample data of the product detection index based on the Minitab data analysis method; The cluster analysis unit (123) is used to use the Q-type clustering model to subdivide the processed sample data of each indicator into several categories in combination with standard requirements.

9. The power station austenitic stainless steel seamless pipe quality dynamic analysis system according to claim 5, characterized in that: The product grade analysis module (13) comprises: A new product grade evaluation unit (131) is used to obtain a score for the entire product based on the indicator category analysis by calculating the weight of a single indicator and the score of each category of the corresponding indicator, and to judge the grade of the product based on this score; The early warning unit (132) is used to give an alert when the grade of the product falls into the category of deviant products.

10. The power station austenitic stainless steel seamless pipe quality dynamic analysis system according to claim 4, characterized in that: The correlation analysis subsystem (3) uses the KMO test algorithm and the Pearson correlation coefficient method to perform correlation analysis.

Images