Intelligent differential gear production auxiliary method and system

Abstract

The invention discloses an intelligent differential gear production auxiliary method and system, and relates to the technical field of intelligent gear production, and the method comprises the steps of production data collection, quality evaluation, traceability chain construction and intelligent production management. The invention specifically relates to a power transmission and transformation system supervision method based on artificial intelligence, and the method employs a comprehensive intelligent integration method combining gear quality evaluation and quality traceability, can achieve the precise backtracking of a specific abnormal process, equipment or operation time, and improves the problem positioning and quality responsibility investigation efficiency. A gear quality evaluation method combining multi-modal fusion and dynamic edge perception is adopted to carry out quality evaluation, so that the detection accuracy of a system on a high-complexity texture structure region is effectively improved; the block chain-based dynamic quality traceability chain construction method is adopted to construct the traceability chain, so that only a specific process or station in an abnormal batch can be positioned, 'accurate recall 'rather than'whole batch discarding' is supported, and the reworking cost is greatly reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent gear production, and specifically relates to an intelligent differential gear production auxiliary method and system. Background Art

[0002] The intelligent differential gear production assistance method uses image recognition, artificial intelligence and blockchain technology to collect and analyze key data in the differential gear production process, realize intelligent identification of differential gear production quality, and aims to improve the accuracy of gear defect detection, achieve quality transparency and traceability throughout the process, assist enterprises in optimizing production processes, reducing quality risks, and promote the upgrading of differential gear manufacturing towards intelligence and high quality.

[0003] However, in the existing differential gear production process, there are technical problems such as difficulty in locating quality problems, unclear division of responsibilities, and low efficiency in tracking abnormal batches; in the gear quality assessment process, there are technical problems such as existing methods relying only on images or manual judgment, difficulty in identifying subtle defects, and susceptibility to image angle and lighting; in the process of building a production traceability chain, there are technical problems such as chaotic chain data hierarchy, difficulty in matching product level information, and easy tampering of traceability information. Summary of the Invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides an intelligent differential gear production auxiliary method and system. In view of the technical problems in the existing differential gear production process, such as difficulty in locating quality problems, unclear division of responsibilities, and low efficiency in tracking abnormal batches, this solution creatively adopts a comprehensive intelligent integration method that combines gear quality assessment and quality traceability. By establishing a multi-layer traceability chain block from the individual level to the batch level for each gear, it can accurately trace back to the specific process, equipment or operation time where the abnormality occurred, significantly improving the efficiency of problem location and quality accountability. In view of the technical problems in the gear quality assessment process, such as the existing methods only rely on images or manual judgment, have difficulty in identifying subtle defects, and are easily affected by image angles and lighting, this solution creatively adopts a method that combines multimodal fusion to In the process of gear quality assessment based on the combination of dynamic edge perception, quality assessment is carried out, which is realized through acoustic data reinforcement. For example, the abnormal spectral morphology of high-frequency howling signals in the finished product sampling stage can be used as indirect evidence of the existence of hidden cracks. At the same time, with the dynamic edge enhancement and multi-scale feature fusion mechanism, the system's detection accuracy for high-complexity texture structure areas is effectively improved; in view of the technical problems of chaotic chain data hierarchy, difficulty in matching product level information, and easy tampering of traceability information in the process of constructing the production traceability chain, this solution creatively adopts a dynamic quality traceability chain construction method based on blockchain to construct the traceability chain, and realizes the positioning of specific processes or workstations in abnormal batches, supports "precise recall" rather than "abandonment of the entire batch", greatly reduces rework costs, and improves production response efficiency.

[0005] The technical solution adopted by the present invention is as follows: The present invention provides an intelligent differential gear production auxiliary method, which includes the following steps:

[0006] Step S1: production data collection;

[0007] Step S2: quality assessment;

[0008] Step S3: traceability chain construction;

[0009] Step S4: Intelligent production management.

[0010] Furthermore, in step S1, the production data collection specifically collects a multi-source production data set from the differential gear production line; the multi-source production data set specifically includes image data, acoustic signal data, process parameter data, process timing information data and product batch information data.

[0011] Furthermore, in step S2, the quality assessment is used to assess the quality of the differential gear. Specifically, the quality assessment is performed based on a multi-source production dataset, in combination with multimodal fusion and dynamic edge sensing in the gear quality assessment process, to obtain differential gear quality information, including the following steps:

[0012] Step S21: Multimodal data encoding, which is used to convert data from different sources into a multimodal feature representation with a unified structure. Specifically, a convolutional neural network is constructed to extract features from the differential gear image to obtain gear image features; a long-term and short-term neural network is constructed to encode the acoustic signal data to obtain acoustic signal time series features; a fully connected network is constructed to vector map the process parameters to obtain process parameter features; and then, the gear image features, acoustic signal time series features, and process parameter features are subjected to attention-weighted fusion to generate a differential gear fusion feature map.

[0013] Step S22: Gear defect identification, specifically, performing gear defect identification through convolution offset prediction, local micro-feature extraction, dynamic edge enhancement, multi-scale feature dynamic fusion, and defect probability generation to obtain a differential gear defect probability map;

[0014] The convolution offset prediction is specifically to obtain the convolution sampling offset from the differential gear fusion feature map through a small convolutional network;

[0015] The local micro-feature extraction is specifically to perform a deformation convolution operation on the differential gear fusion feature map based on the convolution sampling offset to capture the local micro-features and obtain a gear fine feature map;

[0016] The dynamic edge enhancement is specifically performed by calculating the gradient map of the differential gear fusion feature map and using a trainable neural network to dynamically generate edge weights. Based on the edge weights, the differential gear fusion feature map and the gear fine feature map are fused to obtain an edge enhanced feature map.

[0017] The multi-scale feature dynamic fusion is specifically carried out by constructing a multi-scale deformation convolution block to extract defect features of three different receptive fields from the edge enhancement feature map to obtain a multi-scale defect feature map; by performing linear transformation and softmax normalization on each multi-scale defect feature map, dynamic fusion weights are generated; and the multi-scale defect feature maps are fused with the dynamic fusion weights to obtain a gear multi-scale fusion feature map;

[0018] The defect probability generation is specifically to locate and identify defects on the gear multi-scale fusion feature map through a convolution layer to obtain a differential gear defect probability map;

[0019] Step S23: Quality grade assessment, specifically, by calculating the area of ​​the differential gear defect probability map that is greater than the set defect threshold, estimating the defect coverage area to obtain the total defect area, and then constructing a defect type weighted index based on the defect type; by splicing the total defect area, the defect type weighted index and the process parameter characteristics, producing quality assessment features; finally, by constructing a standard support vector machine, classifying the production quality assessment features to obtain the differential gear quality grade;

[0020] Step S24: differential gear quality information is generated, specifically by executing steps S21 to S23 to generate differential gear quality information, wherein the differential gear quality information includes a differential gear defect probability map and a differential gear quality grade.

[0021] Furthermore, in step S3, the traceability chain is constructed to construct a verifiable quality traceability chain. Specifically, based on the multi-source production data set and the differential gear quality information, a dynamic quality traceability chain construction method based on blockchain is adopted to construct the traceability chain to obtain the differential gear production quality traceability chain, including the following steps:

[0022] Step S31: hashing the production data for subsequent quality traceability, specifically by concatenating the differential gear quality grade, process ID, and timestamp, and then generating a differential gear quality hash value through a hash algorithm;

[0023] Step S32: Smart contract verification, specifically, determining the production quality of the differential gear using a preset smart contract function, and writing production data that meets the preset smart contract function standards into the traceability chain to obtain quality verification data;

[0024] The smart contract function specifically takes the quality grade and total defect area of ​​the differential gear as input variables and outputs the production quality judgment result;

[0025] Step S33: constructing a multi-layer traceability chain, specifically constructing a multi-layer traceability chain, connecting the multi-layer traceability chain through a hash index, and automatically stratifying the quality verification data based on data relationships to obtain multi-layer traceability chain layered data;

[0026] The multi-layer traceability chain specifically includes individual level, process level and batch level;

[0027] The individual level layer is used as an independent chain block for each product, storing hash values, quality levels, and abnormality flags;

[0028] The process level layer is used as a summary chain block for all products with the same process number and stores the workstation status of each product;

[0029] The batch level is used as a summary chain block for all product processes of the same production batch, storing the average quality and defect statistics of the same batch of products;

[0030] Step S34: Data encryption storage, specifically using an improved symmetric encryption method to encrypt the multi-layer traceability chain layered data to obtain an encrypted traceability chain data set, and constructing the differential gear production quality traceability chain by obtaining the encrypted traceability chain data set.

[0031] Furthermore, in step S4, the intelligent production management is used to perform comprehensive management in combination with quality assessment and traceability chain, specifically, to perform comprehensive management of differential gear production based on the differential gear quality information and the differential gear production quality traceability chain, and obtain differential gear production process management feedback data.

[0032] The present invention provides an intelligent differential gear production auxiliary system, which includes a production data acquisition module, a quality assessment module, a traceability chain construction module and an intelligent production management module;

[0033] The production data acquisition module is used to acquire production data, obtain a multi-source production data set through production data acquisition, and send the multi-source production data set to the quality assessment module and the traceability chain construction module;

[0034] The quality assessment module is used for quality assessment, obtains differential gear quality information through quality assessment, and sends the differential gear quality information to the traceability chain construction module and the intelligent production management module;

[0035] The traceability chain construction module is used to construct a traceability chain, obtain a differential gear production quality traceability chain through the traceability chain construction, and send the differential gear production quality traceability chain to the intelligent production management module;

[0036] The intelligent production management module is used for intelligent production management, and obtains differential gear production process management feedback data through intelligent production management.

[0037] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0038] (1) In response to the technical problems in the existing differential gear production process, such as difficulty in locating quality problems, unclear division of responsibilities, and low efficiency in tracking abnormal batches, this solution creatively adopts a comprehensive intelligent integration method that combines gear quality assessment and quality traceability. By establishing a multi-level traceability chain from the individual level to the batch level for each gear, it can accurately trace back to the specific process, equipment or operation time where the abnormality occurred, significantly improving the efficiency of problem location and quality accountability;

[0039] (2) In order to address the technical problems in the gear quality assessment process, existing methods rely only on images or manual judgment, are difficult to identify subtle defects, and are easily affected by image angles and lighting, this solution creatively adopts a gear quality assessment process that combines multimodal fusion and dynamic edge perception to conduct quality assessment, and achieves acoustic data reinforcement. For example, the abnormal spectral morphology of high-frequency howling signals in the finished product sampling stage can be used as indirect evidence of the existence of hidden cracks. At the same time, combined with dynamic edge enhancement and multi-scale feature fusion mechanisms, the system effectively improves the detection accuracy of highly complex texture structure areas;

[0040] (3) In order to solve the technical problems of chaotic chain data hierarchy, difficulty in matching product level information, and easy tampering of traceability information during the construction of the production traceability chain, this solution creatively adopts a dynamic quality traceability chain construction method based on blockchain to construct the traceability chain, and achieves the positioning of specific processes or workstations in abnormal batches, supports "precise recall" rather than "whole batch abandonment", greatly reduces rework costs, and improves production response efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic flow chart of an intelligent differential gear production auxiliary method provided by the present invention;

[0042] Figure 2 A schematic diagram of an intelligent differential gear production auxiliary system provided by the present invention;

[0043] Figure 3 Schematic diagram of the process of step S2;

[0044] Figure 4 Schematic diagram of the process of step S3.

[0045] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0047] In the description of the present invention, it should be understood that terms such as "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction. Therefore, they should not be understood as limiting the present invention.

[0048] Example 1, see Figure 1 The present invention provides an intelligent differential gear production auxiliary method, which includes the following steps:

[0049] Step S1: production data collection;

[0050] Step S2: quality assessment;

[0051] Step S3: traceability chain construction;

[0052] Step S4: intelligent production management;

[0053] By performing the above operations, this solution creatively adopts a comprehensive intelligent integration method that combines gear quality assessment and quality traceability to address the technical problems in the existing differential gear production process, such as difficulty in locating quality problems, unclear division of responsibilities, and low efficiency in tracking abnormal batches. By establishing a multi-level traceability chain from the individual level to the batch level for each gear, it can accurately trace back to the specific process, equipment or operation time where the abnormality occurred, significantly improving the efficiency of problem location and quality accountability.

[0054] Example 2, see Figure 1 In step S1, the production data collection specifically collects a multi-source production data set from the differential gear production line; the multi-source production data set specifically includes image data, acoustic signal data, process parameter data, process timing information data and product batch information data.

[0055] Example 3, see Figure 1 and Figure 3 This embodiment is based on the above embodiment. Further, in step S2, the quality assessment is used to assess the quality of the differential gear. Specifically, the quality assessment is performed based on a multi-source production dataset, in combination with multimodal fusion and dynamic edge sensing in the gear quality assessment process to obtain differential gear quality information, including the following steps:

[0056] Step S21: Multimodal data encoding, which is used to convert data from different sources into a multimodal feature representation with a unified structure. Specifically, a convolutional neural network is constructed to extract features from the differential gear image to obtain gear image features; a long-term and short-term neural network is constructed to encode the acoustic signal data to obtain acoustic signal time series features; a fully connected network is constructed to vector map the process parameters to obtain process parameter features; and then, the gear image features, acoustic signal time series features, and process parameter features are subjected to attention-weighted fusion to generate a differential gear fusion feature map.

[0057] Step S22: Gear defect identification, specifically, performing gear defect identification through convolution offset prediction, local micro-feature extraction, dynamic edge enhancement, multi-scale feature dynamic fusion, and defect probability generation to obtain a differential gear defect probability map;

[0058] The convolution offset prediction is specifically to obtain the convolution sampling offset from the differential gear fusion feature map through a small convolutional network;

[0059] The local micro-feature extraction is specifically to perform a deformation convolution operation on the differential gear fusion feature map based on the convolution sampling offset to capture the local micro-features and obtain a gear fine feature map;

[0060] The dynamic edge enhancement is specifically performed by calculating the gradient map of the differential gear fusion feature map and using a trainable neural network to dynamically generate edge weights. Based on the edge weights, the differential gear fusion feature map and the gear fine feature map are fused to obtain an edge enhanced feature map.

[0061] The calculation formula for dynamically generating edge weights using a trainable neural network is:

[0062] ;

[0063] Where, is the edge weight, It is a trainable neural network, specifically a small convolutional network with parameters, and G is the gradient map of the differential gear fusion feature map;

[0064] The calculation formula for obtaining the edge enhancement feature map by fusing the differential gear fusion feature map and the gear fine feature map based on the edge weight is:

[0065] ;

[0066] Where, F enhanced is the edge enhancement feature map, F ACD is the gear fine characteristic map, X0 is the differential gear fusion characteristic map;

[0067] The multi-scale feature dynamic fusion is specifically carried out by constructing a multi-scale deformation convolution block to extract defect features of three different receptive fields from the edge enhancement feature map to obtain a multi-scale defect feature map; by performing linear transformation and softmax normalization on each multi-scale defect feature map, dynamic fusion weights are generated; and the multi-scale defect feature maps are fused with the dynamic fusion weights to obtain a gear multi-scale fusion feature map;

[0068] The calculation formula for generating the dynamic fusion weight by performing linear transformation and softmax normalization on each multi-scale defect feature map is as follows:

[0069] ;

[0070] Where, is the dynamic fusion weight of the i-th multi-scale defect feature map, i is the multi-scale defect feature map index, softmax(·) is the softmax normalization function, W f is the linear transformation weight, F i is the i-th multi-scale defect feature map;

[0071] The calculation formula for combining the dynamic fusion weight and fusing the multi-scale defect feature map to obtain the gear multi-scale fusion feature map is:

[0072] ;

[0073] Where, F fused is the multi-scale fusion feature map of gear;

[0074] The defect probability generation is specifically to locate and identify defects on the gear multi-scale fusion feature map through a convolution layer to obtain a differential gear defect probability map;

[0075] Step S23: Quality grade assessment, specifically, by calculating the area of ​​the differential gear defect probability map that is greater than the set defect threshold, estimating the defect coverage area to obtain the total defect area, and then constructing a defect type weighted index based on the defect type; by splicing the total defect area, the defect type weighted index and the process parameter characteristics, producing quality assessment features; finally, by constructing a standard support vector machine, classifying the production quality assessment features to obtain the differential gear quality grade;

[0076] The calculation formula for the total defect area is:

[0077] ;

[0078] Where A defectis the total defect area, x is the row index of the differential gear defect probability map, M is the number of rows of the differential gear defect probability map, y is the column index of the differential gear defect probability map, N is the number of columns of the differential gear defect probability map, is an indicator function, when When the indicator function takes the value of 1, otherwise it takes the value of 0. i,j is the defect probability at position (i, j) in the differential gear defect probability graph, is to set the defect threshold;

[0079] The calculation formula for constructing the defect type weighted index according to the defect type is:

[0080] ;

[0081] Where W type is the defect type weighted index, k is the defect type index, K is the number of defect types, w k is the weight of the k-th defect type, A k is the total area of ​​the kth defect type;

[0082] Step S24: generating differential gear quality information, specifically, generating differential gear quality information by executing steps S21 to S23, wherein the differential gear quality information includes a differential gear defect probability map and a differential gear quality grade;

[0083] By performing the above operations, in order to address the technical problems in the gear quality assessment process, such as the existing methods relying only on images or manual judgment, having difficulty identifying subtle defects, and being easily affected by image angles and lighting, this solution creatively adopts a gear quality assessment process that combines multimodal fusion and dynamic edge perception. This achieves acoustic data reinforcement, such as the abnormal spectral morphology of high-frequency howling signals in the finished product sampling stage, which can be used as indirect evidence of the existence of hidden cracks. At the same time, combined with dynamic edge enhancement and multi-scale feature fusion mechanisms, the system effectively improves the detection accuracy of high-complexity texture structure areas.

[0084] Example 4, see Figure 1 and Figure 4 This embodiment is based on the above embodiment. Further, in step S3, the traceability chain is constructed to construct a verifiable quality traceability chain. Specifically, based on the multi-source production data set and the differential gear quality information, a dynamic quality traceability chain construction method based on blockchain is adopted to construct the traceability chain to obtain the differential gear production quality traceability chain, including the following steps:

[0085] Step S31: The production data is hashed for subsequent quality traceability. Specifically, the differential gear quality grade, process ID, and timestamp are concatenated, and then a differential gear quality hash value is generated using a hash algorithm. The calculation formula is:

[0086] ;

[0087] ;

[0088] Where S is the differential gear quality data, Q is the differential gear quality grade, ID proc is the process ID, T stamp is the timestamp, H Q is the differential gear mass hash value, SHA256(·) is the SHA-256 hash algorithm;

[0089] Step S32: Smart contract verification, specifically, determining the production quality of the differential gear using a preset smart contract function, and writing production data that meets the preset smart contract function standards into the traceability chain to obtain quality verification data;

[0090] The smart contract function specifically takes the quality grade and total defect area of ​​the differential gear as input variables and outputs the production quality judgment result;

[0091] Preferably, the pseudo code of the smart contract function is expressed as:

[0092]

[0093] In the formula, A is the mark of the excellent quality grade, A_defect is the total defect area, B is the mark of the qualified quality grade, The overall identification product has quality risks;

[0094] Step S33: constructing a multi-layer traceability chain, specifically constructing a multi-layer traceability chain, connecting the multi-layer traceability chain through a hash index, and automatically stratifying the quality verification data based on data relationships to obtain multi-layer traceability chain layered data;

[0095] The multi-layer traceability chain specifically includes individual level, process level and batch level;

[0096] The individual level layer is used as an independent chain block for each product, storing hash values, quality levels, and abnormality flags;

[0097] The process level layer is used as a summary chain block for all products with the same process number and stores the workstation status of each product;

[0098] The batch level is used as a summary chain block for all product processes of the same production batch, storing the average quality and defect statistics of the same batch of products;

[0099] The calculation formula for the hash index connecting the multi-layer traceability chain is:

[0100] ;

[0101] Where, is the hash value of the a-th chain block, is the hash value of the a-1th chain block, R a It is the value of the hierarchical data of the multi-layer traceability chain in the ath chain block, where a is the chain block index;

[0102] The calculation formula for the automatic stratification is:

[0103] ;

[0104] Where, L type (m) is the chain-level classification result corresponding to the m-th quality verification data, L1 is the individual level, L2 is the process level, L3 is the batch level, It is the product code corresponding to the mth quality verification data, unique is the unique identification code identifier, The process number corresponding to the mth quality verification data, is the process number corresponding to the nth quality verification data, where m is the quality verification data index, n is the neighboring data index of quality verification data m, and BatchID (m) BatchID is the production batch number corresponding to the mth quality verification data. (n) is the production batch number corresponding to the nth quality verification data;

[0105] Step S34: Data encryption and storage, specifically, using an improved symmetric encryption method to encrypt the multi-layer traceability chain layered data to obtain an encrypted traceability chain data set, and constructing a differential gear production quality traceability chain based on the obtained encrypted traceability chain data set;

[0106] The improved symmetric encryption method generates a dynamic key by introducing a time random factor and a product code, and uses a symmetric encryption method to encrypt chain data. The calculation formula is:

[0107] ;

[0108] Where K ais a dynamic key, AESKeyGen(·) is an improved symmetric key generation function, ID is the product code, Hash(·) is a hash value calculation function used to generate data digest as a key dynamic perturbation source, E a It is the encrypted on-chain record data. Is to call the dynamic key K a Perform symmetric encryption function;

[0109] By performing the above operations, this solution creatively adopts a dynamic quality traceability chain construction method based on blockchain to construct the production traceability chain, addressing the technical problems of chaotic chain data hierarchy, difficulty in matching product hierarchical information, and easy tampering of traceability information during the construction process of the production traceability chain. It can locate the specific process or workstation in the abnormal batch, support "precise recall" instead of "whole batch abandonment", greatly reduce rework costs, and improve production response efficiency.

[0110] Example 5, see Figure 1 This embodiment is based on the above embodiment. Further, in step S4, the intelligent production management is used to combine quality assessment and traceability chain for comprehensive management. Specifically, based on the differential gear quality information and the differential gear production quality traceability chain, comprehensive management of differential gear production is performed to obtain differential gear production process management feedback data.

[0111] Example 6, see Figure 2 , this embodiment is based on the above embodiment, and the present invention provides an intelligent differential gear production auxiliary system, including a production data acquisition module, a quality assessment module, a traceability chain construction module and an intelligent production management module;

[0112] The production data acquisition module is used to acquire production data, obtain a multi-source production data set through production data acquisition, and send the multi-source production data set to the quality assessment module and the traceability chain construction module;

[0113] The quality assessment module is used for quality assessment, obtains differential gear quality information through quality assessment, and sends the differential gear quality information to the traceability chain construction module and the intelligent production management module;

[0114] The traceability chain construction module is used to construct a traceability chain, obtain a differential gear production quality traceability chain through the traceability chain construction, and send the differential gear production quality traceability chain to the intelligent production management module;

[0115] The intelligent production management module is used for intelligent production management, and obtains differential gear production process management feedback data through intelligent production management.

[0116] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0117] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

[0118] The present invention and its embodiments are described above. This description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs structures and embodiments similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. An intelligent differential gear production auxiliary method, characterized by: The method comprises the following steps: Step S1: Production data collection to obtain a multi-source production data set; Step S2: Quality assessment, used to assess the quality of the differential gear. Specifically, based on a multi-source production dataset, a gear quality assessment method combining multimodal fusion and dynamic edge sensing is used to perform quality assessment and obtain differential gear quality information. The method includes the following steps: Step S21: Multimodal data encoding; Step S22: Gear defect identification; Step S23: Quality grade assessment; Step S24: Differential gear quality information generation. Step S3: Traceability chain construction, which is used to build a verifiable quality traceability chain. Specifically, based on multi-source production data sets and differential gear quality information, a dynamic quality traceability chain construction method based on blockchain is used to construct the traceability chain to obtain the differential gear production quality traceability chain. The process includes the following steps: Step S31: Production data hashing; Step S32: Smart contract verification; Step S33: Multi-layer traceability chain construction; Step S34: Data encryption and storage; Step S4: Intelligent production management.

2. The intelligent differential gear production auxiliary method according to claim 1, characterized in that: In step S2, the quality assessment is specifically performed based on a multi-source production data set, in combination with a gear quality assessment method that combines multimodal fusion and dynamic edge perception, to obtain differential gear quality information, including the following steps: Step S21: Multimodal data encoding, which is used to convert data from different sources into a multimodal feature representation with a unified structure. Specifically, a convolutional neural network is constructed to extract features from the differential gear image to obtain gear image features; a long-term and short-term neural network is constructed to encode the acoustic signal data to obtain acoustic signal time series features; a fully connected network is constructed to vector map the process parameters to obtain process parameter features; and then, the gear image features, acoustic signal time series features, and process parameter features are subjected to attention-weighted fusion to generate a differential gear fusion feature map. Step S22: Gear defect identification, specifically, performing gear defect identification through convolution offset prediction, local micro-feature extraction, dynamic edge enhancement, multi-scale feature dynamic fusion, and defect probability generation to obtain a differential gear defect probability map; Step S23: Quality grade assessment, specifically, by calculating the area of ​​the differential gear defect probability map that is greater than the set defect threshold, estimating the defect coverage area to obtain the total defect area, and then constructing a defect type weighted index based on the defect type; by splicing the total defect area, the defect type weighted index and the process parameter characteristics, producing quality assessment features; finally, by constructing a standard support vector machine, classifying the production quality assessment features to obtain the differential gear quality grade; Step S24: differential gear quality information is generated, specifically by executing steps S21 to S23 to generate differential gear quality information, wherein the differential gear quality information includes a differential gear defect probability map and a differential gear quality grade.

3. The intelligent differential gear production auxiliary method according to claim 2, characterized in that: In step S22, the convolution offset is predicted by learning the convolution sampling offset from the differential gear fusion feature map through a small convolution network; The local micro-feature extraction is specifically to perform a deformation convolution operation on the differential gear fusion feature map based on the convolution sampling offset to capture the local micro-features and obtain a gear fine feature map; The dynamic edge enhancement is specifically performed by calculating the gradient map of the differential gear fusion feature map and using a trainable neural network to dynamically generate edge weights. Based on the edge weights, the differential gear fusion feature map and the gear fine feature map are fused to obtain an edge enhanced feature map. The multi-scale feature dynamic fusion is specifically to extract defect features of three different receptive fields from the edge enhancement feature map by constructing a multi-scale deformation convolution block to obtain a multi-scale defect feature map; generate dynamic fusion weights by performing linear transformation and softmax activation on each multi-scale defect feature map; combine the dynamic fusion weights to fuse the multi-scale defect feature maps to obtain a gear multi-scale fusion feature map; The defect probability generation is specifically to locate and identify defects on the gear multi-scale fusion feature map through a convolution layer to obtain a differential gear defect probability map.

4. The intelligent differential gear production auxiliary method according to claim 3, characterized in that: In step S3, the traceability chain is constructed to construct a verifiable quality traceability chain. Specifically, based on the multi-source production data set and the differential gear quality information, a dynamic quality traceability chain construction method based on blockchain is adopted to construct the traceability chain to obtain the differential gear production quality traceability chain, including the following steps: Step S31: hashing the production data for subsequent quality traceability, specifically by concatenating the differential gear quality grade, process ID, and timestamp, and then generating a differential gear quality hash value through a hash algorithm; Step S32: Smart contract verification, specifically, determining the production quality of the differential gear using a preset smart contract function, and writing production data that meets the preset smart contract function standards into the traceability chain to obtain quality verification data; The smart contract function specifically takes the quality grade and total defect area of ​​the differential gear as input variables and outputs the production quality judgment result; Step S33: constructing a multi-layer traceability chain, specifically constructing a multi-layer traceability chain, connecting the multi-layer traceability chain through a hash index, and automatically stratifying the quality verification data based on data relationships to obtain multi-layer traceability chain layered data; The multi-layer traceability chain specifically includes individual level, process level and batch level; The individual level layer is used as an independent chain block for each product, storing hash values, quality levels, and abnormality flags; The process level layer is used as a summary chain block for all products with the same process number and stores the workstation status of each product; The batch level is used as a summary chain block for all product processes of the same production batch, storing the average quality and defect statistics of the same batch of products; Step S34: Data encryption storage, specifically using an improved symmetric encryption method to encrypt the multi-layer traceability chain layered data to obtain an encrypted traceability chain data set, and constructing the differential gear production quality traceability chain by obtaining the encrypted traceability chain data set.

5. The intelligent differential gear production auxiliary method according to claim 4, characterized in that: In step S4, the intelligent production management is used to perform comprehensive management in combination with quality assessment and traceability chain. Specifically, comprehensive management of differential gear production is performed based on the differential gear quality information and the differential gear production quality traceability chain to obtain differential gear production process management feedback data.

6. The intelligent differential gear production auxiliary method according to claim 5, characterized in that: In step S1, the production data collection specifically collects a multi-source production data set from a differential gear production line; the multi-source production data set specifically includes image data, acoustic signal data, process parameter data, process timing information data, and product batch information data.

7. An intelligent differential gear production assistance system, used to implement an intelligent differential gear production assistance method according to any one of claims 1 to 6, characterized in that: It includes production data collection module, quality assessment module, traceability chain construction module and intelligent production management module.

8. The intelligent differential gear production auxiliary system according to claim 7, characterized in that: The production data acquisition module is used to acquire production data, obtain a multi-source production data set through production data acquisition, and send the multi-source production data set to the quality assessment module and the traceability chain construction module; The quality assessment module is used for quality assessment, obtains differential gear quality information through quality assessment, and sends the differential gear quality information to the traceability chain construction module and the intelligent production management module; The traceability chain construction module is used to construct a traceability chain, obtain a differential gear production quality traceability chain through the traceability chain construction, and send the differential gear production quality traceability chain to the intelligent production management module; The intelligent production management module is used for intelligent production management, and obtains differential gear production process management feedback data through intelligent production management.

Images