CN-117131345-B - Multi-source data parameter evaluation method based on data deep learning calculation

CN117131345BCN 117131345 BCN117131345 BCN 117131345BCN-117131345-B

Abstract

The invention discloses a multisource data parameter evaluation method based on data deep learning calculation, and belongs to the technical field of data parameter evaluation. The invention discloses a multi-source data parameter evaluation method based on data deep learning calculation, which comprises the following steps: the method comprises the steps of collecting multi-source data in real time, processing collected multi-source data information, and calculating accuracy of multi-source data parameters based on a data deep learning calculation mode. The method solves the problems that the conventional multi-source data transmission cannot be carried out after the multi-source data parameter evaluation, so that the multi-source data transmission risk is high, the multi-source data transmission is unstable and the data transmission accuracy is low.

Inventors

LI CHANGFU
DUAN HONGYAN

Assignees

山西玖幺两航空运动有限公司

Dates

Publication Date: 20260512
Application Date: 20230830

Claims (10)

1. A multi-source data parameter evaluation method based on data deep learning calculation is characterized by comprising the following steps: S1, acquiring multi-source data in real time, acquiring multi-source data information, processing the acquired multi-source data information, calculating the accuracy of the multi-source data information based on a data deep learning calculation mode, determining a multi-source data parameter calculation result, and executing different multi-source data storage and transmission strategies according to the multi-source data parameter calculation result; S2, carrying out data transmission and storage on the multi-source data parameters according to the condition that the calculation result of the multi-source data parameters is within the accuracy qualified range, and independently storing and not transmitting the multi-source data parameters according to the condition that the calculation result of the multi-source data parameters is not within the accuracy qualified range; and S3, acquiring multi-source data parameters with unqualified accuracy, analyzing the multi-source data parameters according to data nodes of the multi-source data parameters, evaluating the data parameters of the multi-source data parameters through data node analysis, and presenting an evaluation result in a form of a table.
2. The method for evaluating the multi-source data parameters based on the data deep learning calculation of claim 1, wherein the collected multi-source data information is processed by the following steps: Extracting the acquired multi-source data information completely; Carrying out data retrieval on the multi-source data information, filtering out multi-source data information which is useless for multi-source data parameter evaluation according to multi-source data parameter evaluation requirements, and retaining the multi-source data information which is useful for multi-source data parameter evaluation; classifying the reserved multi-source data information, dividing the multi-source data information into a plurality of categories according to different keywords, wherein different multi-source data information is stored in each category; and calculating the classified multi-source data information, calculating the accuracy rate of the multi-source data information based on a data deep learning calculation mode, and determining a multi-source data parameter calculation result based on the accuracy rate calculation of the multi-source data information.
3. The method for evaluating the multi-source data parameters based on the data deep learning calculation of claim 2, wherein different multi-source data storage and transmission strategies are executed according to the multi-source data parameter calculation result, and the following operations are executed: acquiring a multi-source data parameter calculation result; comparing and analyzing the multi-source data parameter calculation result by referring to the accuracy qualification range of the stored multi-source data information; If the calculation result of the multi-source data parameters is within the qualified range of the accuracy rate of the multi-source data information, the multi-source data parameters are transmitted and stored; if the calculation result of the multi-source data parameters is not in the qualified range of the accuracy of the multi-source data information, the multi-source data parameters are stored independently and are not transmitted.
4. The method for evaluating the multi-source data parameters based on the data deep learning computation of claim 3, wherein the method is characterized by acquiring the multi-source data parameters with unqualified accuracy and searching for the data nodes of the multi-source data parameters, wherein the data nodes of the multi-source data parameters comprise a data node name, a data node creator, a data node creation type, a data node modification time, a data node modifier, an associated point number, a data source, a participation check field, a sorting field, a hash bit number and a description.
5. The method for evaluating the multi-source data parameters based on the data deep learning computation of claim 4, wherein the multi-source data parameters are analyzed according to the data nodes of the multi-source data parameters, and the following operations are executed: acquiring a data node of a multi-source data parameter; extracting data node multi-source information of multi-source data parameters, and analyzing the data node multi-source information of the multi-source data parameters and stored data node standard information; aiming at the condition that single information in the data node multi-source information of the multi-source data parameters is consistent with the data node standard information, the single information accuracy rate of the data node is qualified; aiming at the situation that single information in the multi-source information of the data node of the multi-source data parameter is inconsistent with standard information of the data node, the accuracy rate of the single information of the data node is not qualified; And carrying out data parameter evaluation on the multi-source data parameters according to the analysis result of the data nodes.
6. The method for evaluating the multi-source data parameters based on the data deep learning calculation of claim 5, wherein the data node multi-source information of the multi-source data parameters and the stored data node standard information are analyzed, and the following operations are performed: extracting single data node information in the multi-source information of the data nodes of the multi-source data parameters one by one; comparing and analyzing single information of a plurality of data nodes extracted one by one with standard information of the data nodes; And extracting the single information of the unqualified data nodes according to the condition that the single information of the data nodes is unqualified, counting, and uniformly storing the single information in an unqualified accuracy list.
7. The method for evaluating the multi-source data parameters based on the data deep learning calculation of claim 6, wherein the data parameter evaluation is performed on the multi-source data parameters according to the data node analysis, and the following operations are performed: Acquiring single information of the data nodes in the unqualified list of the accuracy rate, calculating the single information of the data nodes, and determining the number S of data parameter items with unqualified accuracy rate in the data nodes; Combining the number S of the unqualified data parameters of the accuracy rate, and evaluating the multi-source data parameters by integrating single information of a plurality of unqualified data nodes of the accuracy rate; Determining different data parameter evaluation results according to the data node conditions of different data parameters; and acquiring a plurality of groups of data parameter evaluation results, and presenting the evaluation results in a form of a table.
8. The method for evaluating the multi-source data parameters based on the data deep learning calculation of claim 7, wherein the multi-source data parameters are evaluated by combining the number S of the data parameters with unqualified accuracy, and the following operations are executed: If the number S of the data parameter items is more than or equal to 0 and less than 3, determining that the risk of the current multi-source data parameter is smaller as a result of evaluating the data parameter; If the number S of the data parameter items is less than or equal to 3 and is less than 5, the determined data parameter evaluation result is that the current multi-source data parameter risk is general; if the number S of the data parameter items is not more than 5, determining that the risk of the current multi-source data parameter is larger as a result of evaluating the data parameter; Based on the determined data parameter evaluation results, the evaluation results are presented in the form of a table, and corresponding data parameter evaluation risk conditions are presented.
9. The method for evaluating the multi-source data parameters based on the data deep learning calculation of claim 1, further comprising the steps of: The method for monitoring the multi-source data information, which is stored independently, in real time, and the multi-source data information, the calculated result of which is not in the qualified range of the accuracy rate of the multi-source data information, and selectively deleting the multi-source data information according to the storage space requirement comprises the following steps: detecting and monitoring a storage unit for independently storing multi-source data information, wherein the multi-source data information is not in the accuracy qualified range of the multi-source data information, and acquiring the residual storage capacity of the storage unit; extracting the data quantity of the multi-source data information, of which the multi-source data parameter calculation result is not in the qualified range of the accuracy rate of the multi-source data information, which is determined in each unit time, and taking the data quantity as a reference factor; and acquiring a residual space resource evaluation parameter of the storage unit by using the residual storage capacity of the storage unit and a reference factor, wherein the residual space resource evaluation parameter is acquired by the following formula: ; wherein R represents a residual space resource evaluation parameter; R 0 represents the remaining storage capacity of the current storage unit; the Δr represents a parameter compensation amount, n represents the total number of times the data amount of the multi-source data information within the accuracy qualification range of the multi-source data information generated in the next unit time exceeds the data amount of the multi-source data information within the accuracy qualification range of the multi-source data information generated in the last unit time, Δc i represents the number of times the data amount of the multi-source data information within the accuracy qualification range of the multi-source data information generated in the next unit time exceeds the data amount of the multi-source data information within the accuracy qualification range of the multi-source data information generated in the last unit time, S maxi represents the floating difference value of the corresponding data amount when the data amount of the multi-source data information within the accuracy qualification range of the multi-source data information generated in the i-th unit time exceeds the data amount of the multi-source data information generated in the last unit time, S maxi represents the corresponding data amount of the multi-source data information within the accuracy qualification range of the multi-source data information generated in the i-th unit time exceeds the data information generated in the last unit time, S t represents the risk grade type corresponding to the maximum data rate type, and the risk grade value of the multi-source grade data is more than the risk grade data of the risk grade of 3726 when the risk grade is greater than the risk grade of the risk grade data is equal to 35.25 f=0, f t =0.43, m represents the total number of unit time which has been currently experienced, C j represents the data amount of the multi-source data information within the qualified range of the accuracy rate of the multi-source data information generated in the jth unit time, m 2 represents the number of data items with general risk level in the existing data amount, m 3 represents the number of data items with larger risk level in the existing data amount, S 2i represents the number of data parameter items corresponding to the general data with the ith risk level, and S 3i represents the number of data parameter items corresponding to the data with larger risk level; and when the residual space resource evaluation parameter is lower than a preset evaluation parameter threshold, selectively deleting the multi-source data information.
10. The method for evaluating the multi-source data parameters based on the data deep learning calculation of claim 9, wherein the step of selectively deleting the multi-source data information comprises the steps of: Extracting each multi-source data information of the storage unit; Obtaining a storage value parameter of each multi-source data information by using a comprehensive value evaluation model, wherein the comprehensive value evaluation model is as follows: ; Wherein R f represents a storage value parameter, C 1 represents a data amount corresponding to multi-source data information within an accuracy qualification range of each multi-source data information, C represents a total data amount stored in the storage unit, P represents a total data calling number of the storage unit, and P 0i represents a calling number of multi-source data information within an accuracy qualification range of an ith multi-source data information; and deleting the multi-source data information with the stored value parameter lower than the preset value parameter threshold when the stored value parameter is lower than the preset value parameter threshold.

Description

Multi-source data parameter evaluation method based on data deep learning calculation Technical Field The invention relates to the technical field of data parameter evaluation, in particular to a multi-source data parameter evaluation method based on data deep learning calculation. Background Deep learning is a new research direction in the field of machine learning, and is introduced into machine learning to enable the machine learning to be closer to an original target, namely artificial intelligence, and is the internal law and representation level of learning sample data, and information obtained in the learning process is greatly helpful to interpretation of data such as characters, images and sounds. The final goal is to enable a machine to analyze learning ability like a person, recognize data such as text, image and sound, deep learning is a complex machine learning algorithm, and the effects achieved in terms of speech and image recognition far exceed those achieved in the prior art, and deep learning has achieved many results in search technology, data mining, machine learning, machine translation, natural language processing, multimedia learning, speech, recommendation and personalization technologies, and other related fields. The deep learning makes the machine imitate the activities of human beings such as audio-visual and thinking, solves a plurality of complex pattern recognition problems, and makes the related technology of artificial intelligence greatly advanced. The Chinese patent with publication number CN111858346A discloses a data quality multidimensional assessment technology based on a deep learning test dataset, based on the measurement assessment of the test dataset of a deep neural network, the measurement of two layers of input and output of the test dataset is given, and finally a quality report of the test dataset is generated. However, the above patent suffers from the following drawbacks: For transmission of multi-source data, the multi-source data cannot be transmitted after being evaluated based on parameters of the multi-source data, so that the risk of multi-source data transmission is high, the multi-source data transmission is unstable, and the accuracy of data transmission is low. Disclosure of Invention The invention aims to provide a multi-source data parameter evaluation method based on data deep learning calculation, which can reduce the risk of multi-source data transmission, stabilize multi-source data transmission and improve data transmission accuracy based on multi-source data parameter evaluation for multi-source data transmission and solve the problems in the background technology. In order to achieve the above purpose, the present invention provides the following technical solutions: A multi-source data parameter evaluation method based on data deep learning calculation comprises the following steps: S1, acquiring multi-source data in real time, acquiring multi-source data information, processing the acquired multi-source data information, calculating accuracy of multi-source data parameters based on a data deep learning calculation mode, determining a multi-source data parameter calculation result, and executing different multi-source data storage and transmission strategies according to the multi-source data parameter calculation result; S2, according to the situation that the calculation result of the multi-source data parameter is within the accuracy qualified range, the multi-source data parameter is subjected to data transmission and storage, and according to the situation that the calculation result of the multi-source data parameter is not within the accuracy qualified range, the multi-source data parameter is subjected to independent storage and is not transmitted; and S3, acquiring multi-source data parameters with unqualified accuracy, analyzing the multi-source data parameters according to data nodes of the multi-source data parameters, evaluating the data parameters of the multi-source data parameters through data node analysis, and presenting an evaluation result in a form of a table. Preferably, the collected multi-source data information is processed, and the following operations are performed: Extracting the acquired multi-source data information completely; Carrying out data retrieval on the multi-source data information, filtering out multi-source data information which is useless for multi-source data parameter evaluation according to multi-source data parameter evaluation requirements, and retaining the multi-source data information which is useful for multi-source data parameter evaluation; classifying the reserved multi-source data information, dividing the multi-source data information into a plurality of categories according to different keywords, wherein different multi-source data information is stored in each category; and calculating the classified multi-source data information, calculating the accuracy rate of the multi-source