Oilfield business main data aggregation and classification method

Abstract

The invention discloses an oil field business main data aggregation and classification method, which belongs to the field of enterprise main data processing, and comprises the following steps of: firstly, dividing data into two types of data, namely digital data and text data, and respectively classifying the two types of data; and secondly, the text type data is processed into multi-dimensional vector data, that is, data capable of relating to context data and digital type data are subjected to data cleaning, and the integrity, accuracy and normalization of digital type main data are ensured through data cleaning. The method comprises the following steps of: extracting multi-dimensional vector data and standard digital type data, performing aggregation classification on the multi-dimensional vector data and the standard digital type data through an unsupervised machine learning model to obtain an aggregation result, and finally, extracting common features in a certain type of data according to the aggregation classification result, and screening the type of data according to the common features to realize accurate classification of the oil field business main data.

Description

Technical Field

[0001] The invention belongs to the field of enterprise master data processing, and in particular relates to a method for aggregating and classifying oilfield business master data. Background Art

[0002] As the level of informatization of enterprises improves and the number of system platforms used increases, the demand for shared use of enterprise master data by different business units within the enterprise has gradually become prominent. Due to the lack of correct understanding of master data by enterprises during the information system construction period, the master data standards are huge and complicated.

[0003] Enterprise master data is unified and sharable business data within an enterprise and is the most important data asset of an enterprise, such as the designed well depth, actual well depth, crude oil and natural gas production, ground engineering technical standards, master data of gathering and transportation pipeline coordinates, ground information, well numbers in the petroleum and petrochemical industry, site information, enterprise employee information, customer information and supplier information, etc. The classification and sorting of existing master data mainly relies on manual identification and retrieval, which consumes a lot of manpower, economic and time costs.

[0004] Existing methods for classifying enterprise master data are time-consuming and costly. Summary of the invention

[0005] The purpose of the present invention is to provide a method for aggregating and classifying oilfield business master data, so as to solve the technical problems of time-consuming and high-cost methods for classifying enterprise master data in the prior art.

[0006] To achieve the above purpose, the technical solution of a method for aggregating and classifying oilfield business master data provided by the present invention is:

[0007] A method for aggregating and classifying oilfield business master data comprises the following steps: obtaining oilfield business master data to be aggregated and classified, dividing the obtained master data into two data sets of text type and digital type; converting data in the text type data set into multidimensional vector data, performing data cleaning on the digital type data set and converting it into data of labeled digital type; building an aggregation classification model using an unsupervised machine learning model, taking the multidimensional vector data as input of the model, outputting the aggregation classification result of the text data, taking the labeled digital type data as input of the model, outputting the aggregation classification result of the digital data; extracting common features in a certain type of data according to the aggregation classification result, and screening the data of this type according to the common features.

[0008] The beneficial effects are as follows: first, the data is divided into two types, digital and text, and classified separately; second, the text type data is processed into multidimensional vector data, that is, data that can be linked to context data, and the digital type data is cleaned. Through data cleaning, the integrity, accuracy and standardization of the digital type master data are guaranteed. Then, the multidimensional vector data and the standard digital type data are aggregated and classified through the unsupervised machine learning model to obtain the aggregation results. Finally, the common features in a certain type of data are extracted according to the aggregation classification results, and the data of this type is screened according to the common features, so as to improve the accuracy of the oilfield business master data to be aggregated and classified.

[0009] As a further improvement, when converting the data in the text type data set into multidimensional vector data, the word2vec text feature extraction vector model is used for conversion. As a further improvement, the process of using the word2vec text feature extraction vector model to obtain multidimensional vector data is: using CBOW to process the text type data set, the processed data is used as sample data, and the multidimensional vector data is expressed by the formula: L = ∑ wec logp(ω|context(ω)), where ω is the original value of the sample data and context(ω) is the context data.

[0010] The beneficial effect is: processing data by the CBOW method improves the data foundation for the connection context of a certain master data.

[0011] As a further improvement, the unsupervised machine learning model uses K-means to aggregate and classify the input data according to the set K value.

[0012] As a further improvement, when obtaining the oilfield business master data to be aggregated and classified, the data extraction and integration framework ETL is used to obtain data from the oilfield data source.

[0013] The beneficial effect is that obtaining data through the data extraction integration framework ETL can improve the efficiency of obtaining data.

[0014] As a further improvement, the method also includes displaying the aggregate classification results using a scatter plot or a table.

[0015] The beneficial effects are: the clustered data is displayed through a scatter plot, making the data more intuitive, and the clustered data is displayed through a table, which is convenient for quickly locating data of different categories.

[0016] As a further improvement, data cleaning refers to: deleting special characters and abnormal data in digital data sets.

[0017] The beneficial effect is that the clarified data can improve the efficiency of aggregation and the accuracy of classification. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic flow chart of a method for aggregating and classifying oilfield business master data according to the present invention;

[0019] Figure 2 A scatter plot of a result set of text type for the method for aggregating and classifying oilfield business master data of the present invention;

[0020] Figure 3 A scatter plot of a result set of digital types for the method of aggregating and classifying oilfield business master data of the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention, that is, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. The components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0022] An embodiment of a method for aggregating and classifying oilfield business master data:

[0023] This application takes the relevant data of an oilfield enterprise as an example. The master data of the enterprise includes the designed well depth, actual well depth, crude oil and natural gas production, ground engineering technical standards, gathering and transportation pipeline coordinate master data, ground information, well numbers of the petroleum and petrochemical industry, site information, enterprise employee information, customer information, supplier information, etc. This application takes well number data and completion data as an example.

[0024] like Figure 1 As shown, the method for aggregating and classifying oilfield business master data includes the following steps:

[0025] 1) The extraction and integration framework ETL is used to obtain the oilfield business master data to be aggregated and classified from the database, and the master data is divided into two data sets: text type and digital type.

[0026] Among them, the well number data is text type data, and the data set of the well number data is recorded as S1; the number of completed wells is digital type data, and the data set of the number of completed wells is recorded as S2. The specific data of the data sets of S1 and S2 are shown in the following Table 1:

[0027] Table 1

[0028] Hashtag S1 Number of completed wells S2 Maoba 138-1 3525 Celebration 23 -4588 Article 77-1 3883？ Pulu 3 2360 Wei 77-10 4600

[0029] 2) Use the word2vec text feature extraction vector model to convert the data in the text type data set into multi-dimensional vector data, and clean the digital type data set and convert it into standard digital type data. The data cleaning standard is to remove unnecessary special characters, format errors, data anomalies, data missing, and repeated impurities to ensure the integrity, accuracy, and standardization of the digital type master data. Convert the digital data type to the standard digital data type.

[0030] Among them, for the text type oilfield business master data S1, the word2vec text feature extraction vector model is used to convert S1 into multidimensional vector data to obtain sample set S1. The converted multidimensional vector data is shown in Table 2 below. For the digital type oilfield business master data S2, the data is cleaned according to the standard, and S2 is converted into standard data digital type data to obtain sample set S2. The standard digital type data is shown in Table 3 below.

[0031] Table 2

[0032]

[0033]

[0034] Table 3

[0035] Number of completed wells2 Number of completed wells2 Number of completed wells2 Number of completed wells2 Number of completed wells2 3525 4588 3883 2360 4600

[0036] 3) Use the K-means unsupervised machine learning model to build an aggregation classification model, take the multidimensional vector data as the input of the aggregation classification model, learn according to the K value, and output the aggregation classification results of the data. Take the data of the vertebral number type as the input of the aggregation classification model, and output the aggregation classification results of the data through unsupervised autonomous learning.

[0037] Specifically, the sample set S1 is used as the training set S1, and the training set S1 is used as the input of the aggregation classification model. After the K value clusters are divided, unsupervised automatic learning is performed to output the results of the aggregation classification of the main data as the result set S1. The classification results are shown in Table 4 below:

[0038] Table 4

[0039] Maoba 138-1 Wei 77-10 Pulu 3 / Article 77-1 Celebration 23

[0040] The sample set S2 is used as the training set S2, and the training set S2 is used as the input of the aggregate classification model. Through autonomous learning, the results of the data aggregation classification are output. The classification results are shown in Table 5 below:

[0041] Table 5

[0042] 2360 3883 4588 / 3525 4700

[0043] 4) If Figure 2 , Figure 3 As shown in the figure, according to the aggregation classification results, the common features in a certain type of data are extracted, and the data of this type is screened according to the common features. For example, Wen 103, 2 Wen 103, and Wen 103-2 are the same type of data. The common feature of the data is that they are all in the format of a text plus a number. Through analysis and screening, it can be seen that 2 Wen 103 and Wen 103-2 may be wrong well numbers, and the staff needs to pay special attention to them. This method can provide oilfield enterprise users with a more accurate well number data set, improve the standardization and unique identification of well numbers, reduce the workload of data retrieval and query, and have a great role in promoting enterprise data governance. Enterprise employee information Zhang San, Zhang San 6, and Zhang San are the same type of data. Through analysis and screening, it can be seen that the common feature of the data is the format of text. Zhang San 6 and Zhang San may be wrong employee names, and the staff needs to pay special attention to them. This method can provide oilfield enterprise users with a more accurate employee information set, improve the standardization and unique identification of names, reduce the workload of data retrieval and query, and have a great role in promoting enterprise data governance.

[0044] In addition, the aggregate classification results are displayed using scatter plots or tables. The data sets, sample sets, training sets, and result sets of S1 and S2 are comprehensively displayed, sorted, and statistically analyzed to form a scatter plot for representing the results of the aggregate classification.

[0045] Specifically, Figure 2 The scatter plot of the text class clustering results is shown. Figure 2 The results show that the closer the texts are, the denser the data points are, and the farther the data points are, the more different the texts they correspond to. For example, the points formed by Wei 77-10 and Wen 77-1 are close in distance, while the points formed by Pu Lu 3 and Qing 23 are close in distance, and Mao Ba 138-1 is relatively independent. At the same time, if different cluster values ​​are selected, the clustering results will also be different.

[0046] like Figure 3 The scatter plot of the clustering results of the digital class is shown. Figure 3 From the results shown, we can see that data values ​​with similar numbers are located in similar areas. The scattered points formed by 3525 and 3883 are distributed close to each other, while the distribution distances of 4588 and 4700 are close to each other, and 2360 is relatively independent.

[0047] In oil and gas exploration and development, the production system captures data according to data needs, classifies the data, and screens the data according to the classification results. The master data such as the well numbers of each system in the screened data can remain highly consistent, ensuring that the automatically collected real-time data can be pushed normally, reducing the workload of grassroots personnel; taking data as the key element, it can accelerate the advancement of oil and gas production Internet of Things, scientific research collaboration, business process informatization, and intelligent decision-making and command, and realize the exploration and development, scientific research collaboration, production experience and other systems to effectively support high-quality development.

[0048] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention is described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions recorded in the aforementioned embodiments without creative work, or replace some of the technical features therein with equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for aggregating and classifying oilfield business master data, characterized in that: The method comprises the following steps: Obtain the oilfield business master data to be aggregated and classified, and divide the acquired master data into two data sets: text type and numeric type; Convert the data in the text type data set into multidimensional vector data, clean the data of the digital type data set and convert it into the standard digital type data; Use unsupervised machine learning models to build aggregate classification models, take multidimensional vector data as input to the model, output aggregate classification results of text data, take data with numeric type as input to the model, output aggregate classification results of numeric data; Based on the aggregate classification results, common features in a certain type of data are extracted, and the data of this type is screened based on the common features.

2. The method for aggregating and classifying oilfield business master data according to claim 1, characterized in that: When converting data in a text-type dataset into multi-dimensional vector data, the word2vec text feature extraction vector model is used for conversion.

3. The method for aggregating and classifying oilfield business master data according to claim 2, characterized in that: The process of using the word2vec text feature extraction vector model to obtain multidimensional vector data is as follows: CBOW is used to process the text type data set, and the processed data is used as sample data. The multidimensional vector data is expressed by the formula: L = Σ ω∈C logP(ω|context(ω)), where ω is the sample data and context(ω) is the context data.

4. The method for aggregating and classifying oilfield business master data according to claim 1, characterized in that: The unsupervised machine learning model uses K-means to aggregate and classify the input data according to the set K value.

5. The method for aggregating and classifying oilfield business master data according to claim 1, characterized in that: When obtaining the oilfield business master data to be aggregated and classified, the data extraction and integration framework ETL is used to obtain data from the oilfield data source.

6. The method for aggregating and classifying oilfield business master data according to any one of claims 1 to 5, characterized in that: The method also includes displaying the aggregate classification results using a scatter plot or table.

7. The method for aggregating and classifying oilfield business master data according to any one of claims 1 to 5, characterized in that: Data cleaning refers to deleting special characters and abnormal data in digital data sets.

Images