Intelligent analysis method for automobile supply chain bill data driven by deep learning

Abstract

The invention discloses a deep learning-driven intelligent analysis method for automobile supply chain bill data, and relates to the technical field of data management.The deep learning-driven intelligent analysis method comprises the steps that an automobile supply chain bill is scanned to obtain scanned images of the automobile supply chain bill, and then the multiple scanned images obtained through scanning are preprocessed periodically to obtain a ciphertext conversion sequence; converting the ciphertext conversion sequence, and determining an array ciphertext interval of each pre-processing period according to the number of pre-processing times and the number characteristic of the segmentation array of the currently pre-processed scanning image in the process of pre-processing the scanning image, according to an array ciphertext interval, 16 kinds of originally existing four-bit binary numbers are compressed into 8 kinds, and ciphertext characters are used for increasing fault tolerance of a ciphertext conversion sequence after forward transmission, so that the ciphertext conversion sequence obtained through final conversion can have 16 kinds of four-bit binary numbers on the whole or in groups, complexity and difficulty are increased for three-party cracking, and the conversion efficiency of the ciphertext conversion sequence is improved. And the security of the scanned image in transmission is further ensured.

Description

Technical Field

[0001] The present invention relates to the field of data management technology, and in particular to a deep learning-driven intelligent analysis method for automobile supply chain bill data. Background Art

[0002] With the continuous increase in business volume, the traditional manual processing and analysis of automobile supply chain bill data can no longer meet the needs of efficiency and accuracy. On the one hand, manual processing is prone to errors, especially when faced with massive bill data, the accuracy of data entry and analysis is difficult to guarantee; on the other hand, manual processing is slow, which will cause delays in business processes and affect the operating efficiency of the entire supply chain. In order to solve the problems of traditional processing methods, deep learning technology has gradually been introduced into the analysis of automobile supply chain bill data. Deep learning has powerful feature learning and data processing capabilities, and can automatically extract valuable information from a large amount of bill data. When using deep learning to analyze automotive supply chain bill data, one of the important links is the processing of bill scan images. Since bills may come from different suppliers and transaction links, their formats, clarity, paper quality, etc. may vary. Therefore, before these scanned images are used as input for deep learning models, they need to be processed in a series of ways. In order to meet the high requirements of deep learning models for computing resources, these bill scan images are usually processed with the help of powerful cloud computing capabilities. Cloud computing platforms have a large amount of computing resources and can quickly perform complex processing and analysis on images. However, in the process of transmitting the scanned images of automobile supply chain bills to the cloud, data security becomes a crucial issue. In order to protect the confidentiality of the image data and prevent it from being stolen or tampered with during transmission, key encryption is usually used to encrypt the image for transmission. However, this method has certain risks. Once the key is lost, the encrypted scanned image will become unsafe, which may cause the leakage of sensitive information contained therein (such as transaction amount, supplier information, etc.), causing huge losses to automobile supply chain companies. In addition, key management is a complex and resource-intensive process. Enterprises need to invest a lot of financial resources to establish a secure key storage system to prevent keys from being illegally obtained. At the same time, they also need to spend time and manpower to regularly update keys and monitor key usage to ensure the security and effectiveness of keys. The investment of these resources not only increases the operating costs of enterprises, but also there may still be loopholes in the key management process, and security risks cannot be completely eliminated. In order to solve the above problems, the present invention proposes a solution. Summary of the invention

[0003] The purpose of the present invention is to provide a deep learning driven intelligent analysis method for automobile supply chain bill data, in order to solve the problems raised in the above background technology; The purpose of the present invention can be achieved through the following technical solutions: The deep learning-driven intelligent analysis method for automobile supply chain bill data includes the following steps: Step 1: The image preprocessing module preprocesses the scanned images of all automobile supply chain documents stored in the current preprocessing cycle according to the preset preprocessing rules to obtain the ciphertext conversion sequence of the current preprocessing cycle, and transmits the ciphertext conversion sequence of the current preprocessing cycle to the cloud processing terminal; Step 2: The cloud processing terminal restores and stores the received ciphertext conversion sequence of the current preprocessing cycle.

[0004] Furthermore, the image preprocessing module updates and stores the processing count.

[0005] Furthermore, in step 1, the preprocessing rules for the ciphertext conversion sequence of the current preprocessing cycle are obtained as follows: S11: Acquire all scanned images stored in the current preprocessing cycle, perform binary conversion on all scanned images in turn to obtain the base conversion data of the corresponding scanned images, and splice the base conversion data of all scanned images stored in the current preprocessing cycle in the order of storage to obtain the data to be processed in the current preprocessing cycle; S12: specifying a cutting step length of 4, cutting the data to be processed from left to right to obtain a plurality of groups of segmentation arrays, and marking all the groups of segmentation arrays as A1, A2, ..., Aa, according to the order of the segmentation arrays in the data to be processed before cutting, where a≥1; S13: Obtain the processing count number B1 currently stored in the image preprocessing module. If either B1 or a is an odd number in value, the array ciphertext interval of the current preprocessing cycle is determined to be [0,7]; otherwise, the array ciphertext interval of the current preprocessing cycle is determined to be [8,15]; S14: according to the array ciphertext interval of the current preprocessing cycle, the ciphertext array and ciphertext characters of the segmented arrays A1, A2, ..., Aa are obtained according to the preset first or second acquisition rule; S15: in the order of ciphertext characters and ciphertext array, the ciphertext characters and ciphertext array of the segmentation array A1 are concatenated to obtain the ciphertext sequence of the segmentation array A1. Similarly, the ciphertext sequences of the segmentation arrays A2, A3, ..., Aa can be obtained in sequence. S16: In the order of the segmented arrays A1, A2, ..., Aa, the ciphertext sequences of the segmented arrays A1, A2, ..., Aa are sequentially concatenated to obtain a ciphertext conversion sequence of the current preprocessing cycle.

[0006] Furthermore, in step one, after the image preprocessing module transmits the ciphertext conversion sequence of the current preprocessing cycle to the cloud processing terminal, the processing count currently stored in the image preprocessing module is incremented by 1.

[0007] Beneficial effects of the present invention: The present invention scans the automobile supply chain bill to obtain its scanned image, then periodically pre-processes the scanned images to obtain a ciphertext conversion sequence, and converts the ciphertext conversion sequence. This method avoids the conventional key encryption method from being easily cracked by others, and on the other hand, gets rid of the dependence on the key, avoiding excessive loss of time and financial costs caused by storing and managing the key. The present invention determines the array ciphertext interval of each preprocessing cycle according to the number of preprocessing times and the quantity characteristics of the segmented array of the currently preprocessed scanned image during the preprocessing process of the scanned image, compresses the original 16 four-bit binary numbers into eight according to the array ciphertext interval, and uses ciphertext characters to increase the fault tolerance of the ciphertext conversion sequence after the forwarding, so that the ciphertext conversion sequence finally converted will have 16 four-bit binary numbers both as a whole and in groups, which increases the complexity and difficulty of three-party cracking and further ensures the security of the scanned image in transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The present invention will be further described below in conjunction with the accompanying drawings.

[0009] Figure 1 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0010] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0011] like Figure 1 As shown, a deep learning-driven intelligent analysis method for automobile supply chain bill data, the above method is performed by a deep learning-driven intelligent analysis system for automobile supply chain bill data, the system includes a bill image acquisition module, an image preprocessing module and a cloud processing terminal; The bill image acquisition module is used to acquire scanned images of a number of automobile supply chain bills and transmit them to the image preprocessing module. The automobile supply chain bills in this application include but are not limited to transaction amount, transaction date, parts information and transaction parties information data; An image preprocessing module is used to periodically preprocess a plurality of scanned images stored therein. The image preprocessing module temporarily stores the scanned images of a plurality of automobile supply chain documents transmitted thereto after receiving them. The image preprocessing module updates and stores a processing count; The image preprocessing module is used to preprocess all scanned images temporarily stored in the preprocessing cycle according to the preset preprocessing rules at each preprocessing cycle. The preprocessing rules are as follows: S11: Acquire all scanned images stored in the preprocessing cycle, perform binary conversion on all scanned images in turn to obtain base conversion data of the corresponding scanned images, and splice the base conversion data of all scanned images stored in the preprocessing cycle in the order of storage to obtain the data to be processed in the preprocessing cycle; S12: specifying a cutting step length of 4, cutting the data to be processed from left to right to obtain a plurality of groups of segmentation arrays, and marking all the groups of segmentation arrays as A1, A2, ..., Aa, according to the order of the segmentation arrays in the data to be processed before cutting, where a≥1; S13: Obtain the processing count number B1 currently stored in the image preprocessing module. If either B1 or a is an odd number in value, the array ciphertext interval of the current preprocessing cycle is determined to be [0,7]; otherwise, the array ciphertext interval of the current preprocessing cycle is determined to be [8,15]; S14: According to the array ciphertext interval of the current preprocessing cycle, the ciphertext array and ciphertext characters of the segmented arrays A1, A2, ..., Aa are obtained according to the preset first or second acquisition rule, wherein if the array ciphertext interval of the current preprocessing cycle is [0,7], the first acquisition rule for obtaining the ciphertext array and ciphertext characters of the segmented arrays A1, A2, ..., Aa is as follows: SS11: Determine the decimal number B1 of the segment array A1. If B1 is within the array ciphertext interval [0,7], then the segment array A1 itself is used as its ciphertext array, and the character 0 is used as the ciphertext character of the segment array A1. Otherwise, if B1 is not within the array ciphertext interval [0,7], first obtain the decimal number C1 of the segment array A1, and then use the formula D1=C1-8 to calculate the complement value D1 of the segment array A1, and use the complement of the complement value D1 as the ciphertext array of the segment array A1, and use the character 1 as the ciphertext character of the segment array A1. SS12: Obtain the ciphertext arrays and ciphertext characters of the segmentation arrays A2, A3, ..., Aa in sequence according to SS11; If the array ciphertext interval of the current preprocessing cycle is [8,15], the second acquisition rule for obtaining the ciphertext array and ciphertext feature characters of the segmented arrays A1, A2, ..., Aa is as follows: SS21: Determine the decimal number B1 of the segment array A1. If B1 is within the array ciphertext interval [8,15], then the segment array A1 itself is used as its ciphertext array, and the character 0 is used as the ciphertext character of the segment array A1. Otherwise, if B1 is not within the array ciphertext interval [8,15], first obtain the decimal number E1 of the segment array A1, and then use the formula F1=E1+8 to calculate and obtain the original code value F1 of the segment array A1, and use the four-bit binary number of the original code value F1 as the ciphertext array of the segment array A1, and use the character 1 as the ciphertext character of the segment array A1. SS22: Obtain the ciphertext arrays and ciphertext characters of the segmentation arrays A2, A3, ..., Aa in sequence according to SS21; S15: in the order of ciphertext characters and ciphertext array, the ciphertext characters and ciphertext array of the segmentation array A1 are concatenated to obtain the ciphertext sequence of the segmentation array A1. Similarly, the ciphertext sequences of the segmentation arrays A2, A3, ..., Aa can be obtained in sequence. S16: in the order of the segmented arrays A1, A2, ..., Aa, sequentially concatenating the ciphertext sequences of the segmented arrays A1, A2, ..., Aa to obtain the ciphertext conversion sequence of the preprocessing period; The image preprocessing module transmits the ciphertext conversion sequence of the preprocessing cycle to the cloud processing terminal, and after the transmission is completed, the value of the processing count currently stored in the image preprocessing module is incremented by 1; The cloud processing terminal is used to receive and process the scanned image of the automobile supply chain bill in the cloud, and the processing count amount is updated and stored in the cloud processing terminal; After receiving the transmitted ciphertext conversion sequence of the preprocessing period, the cloud processing terminal inversely executes the preprocessing rules to restore all scanned images of the preprocessing period, and stores all the restored scanned images for subsequent image processing, where the image processing includes but is not limited to operations such as denoising, grayscale, and binarization; After the storage is completed, the processing count currently stored in the cloud processing terminal is incremented by 1; In the description of the specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0012] The above contents are merely examples and explanations of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the invention or exceed the scope defined by the claims, they shall all fall within the protection scope of the present invention.

[0013] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A deep learning-driven intelligent analysis method for automobile supply chain bill data, characterized by: The following steps are involved: Step 1: The image preprocessing module preprocesses the scanned images of all automobile supply chain documents stored in the current preprocessing cycle according to the preset preprocessing rules to obtain the ciphertext conversion sequence of the current preprocessing cycle, and transmits the ciphertext conversion sequence of the current preprocessing cycle to the cloud processing terminal; Step 2: The cloud processing terminal restores and stores the received ciphertext conversion sequence of the current preprocessing cycle.

2. The deep learning driven automobile supply chain bill data intelligent analysis method according to claim 1 is characterized in that: The image preprocessing module updates and stores the processing count.

3. The deep learning driven automobile supply chain bill data intelligent analysis method according to claim 1 is characterized in that: In step 1, the preprocessing rules for the ciphertext conversion sequence of the current preprocessing cycle are obtained as follows: S11: Acquire all scanned images stored in the current preprocessing cycle, perform binary conversion on all scanned images in turn to obtain the base conversion data of the corresponding scanned images, and splice the base conversion data of all scanned images stored in the current preprocessing cycle in the order of storage to obtain the data to be processed in the current preprocessing cycle; S12: specifying a cutting step length of 4, cutting the data to be processed from left to right to obtain a plurality of groups of segmentation arrays, and marking all the groups of segmentation arrays as A1, A2, ..., Aa, according to the order of the segmentation arrays in the data to be processed before cutting, where a≥1; S13: Obtain the processing count number B1 currently stored in the image preprocessing module. If either B1 or a is an odd number in value, the array ciphertext interval of the current preprocessing cycle is determined to be [0,7]; otherwise, the array ciphertext interval of the current preprocessing cycle is determined to be [8,15]; S14: according to the array ciphertext interval of the current preprocessing cycle, the ciphertext array and ciphertext characters of the segmented arrays A1, A2, ..., Aa are obtained according to the preset first or second acquisition rule; S15: in the order of ciphertext characters and ciphertext array, the ciphertext characters and ciphertext array of the segmentation array A1 are concatenated to obtain the ciphertext sequence of the segmentation array A1. Similarly, the ciphertext sequences of the segmentation arrays A2, A3, ..., Aa can be obtained in sequence. S16: In the order of the segmented arrays A1, A2, ..., Aa, the ciphertext sequences of the segmented arrays A1, A2, ..., Aa are sequentially concatenated to obtain a ciphertext conversion sequence of the current preprocessing cycle.

4. The deep learning driven automobile supply chain bill data intelligent analysis method according to claim 3 is characterized in that: S14, if the array ciphertext interval of the current preprocessing cycle is [0,7], the first acquisition rule for obtaining the ciphertext array and ciphertext characters of the segmented arrays A1, A2, ..., Aa is as follows: SS11: Determine the decimal number B1 of the segmented array A1: If B1 is within the array ciphertext interval [0,7], then the segmentation array A1 itself is used as its ciphertext array, and character 0 is used as the ciphertext character of the segmentation array A1; If B1 is not within the array ciphertext interval [0,7], first obtain the decimal number C1 of the segment array A1, then use the formula D1=C1-8 to calculate the complement value D1 of the segment array A1, use the complement of the complement value D1 as the ciphertext array of the segment array A1, and use character 1 as the ciphertext character of the segment array A1; SS12: Obtain the ciphertext arrays and ciphertext characters of the segmented arrays A2, A3, ..., Aa in sequence according to SS11.

5. The deep learning-driven intelligent analysis method for automobile supply chain bill data according to claim 3 is characterized in that: S14, if the array ciphertext interval of the current preprocessing cycle is [8,15], the second acquisition rule for obtaining the ciphertext array and ciphertext characteristic characters of the segmented arrays A1, A2, ..., Aa is as follows: SS21: Determine the decimal number B1 of the partitioned array A1: If B1 is within the array ciphertext interval [8,15], then the segmentation array A1 itself is used as its ciphertext array, and character 0 is used as the ciphertext character of the segmentation array A1; If B1 is not within the array ciphertext interval [8,15], first obtain the decimal number E1 of the segment array A1, then use the formula F1=E1+8 to calculate and obtain the original code value F1 of the segment array A1, use the four-bit binary number of the original code value F1 as the ciphertext array of the segment array A1, and use character 1 as the ciphertext character of the segment array A1; SS22: Obtain the ciphertext arrays and ciphertext characters of the segmented arrays A2, A3, ..., Aa in sequence according to SS21.

6. The deep learning driven automobile supply chain bill data intelligent analysis method according to claim 1 is characterized in that: In step 1, after the image preprocessing module transmits the ciphertext conversion sequence of the current preprocessing cycle to the cloud processing terminal, the processing count currently stored in the image preprocessing module is incremented by 1.

Images