Axis product inspection data analysis method and system based on artificial intelligence

Abstract

The embodiment of the present invention provides a method and system for analyzing shaft product detection data based on artificial intelligence. First, the shaft product detection data is obtained, and then loaded into the abnormal node positioning network with optimized network parameters to accurately locate the abnormal node and generate the positioning result. This positioning network performs parameter learning through the first sample learning data sequence (including the detection data and the corresponding abnormal positioning annotation data) and the second sample learning data sequence (obtained by the deep learning algorithm processing the third sample learning data sequence), thereby improving the accuracy of abnormal node positioning. This method combines prior knowledge with deep learning technology, provides an efficient and accurate analysis method for shaft product detection, and helps to improve the level of product quality control.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based shaft product detection data analysis method and system. Background Art

[0002] In modern industrial production, shaft products are widely used in various mechanical equipment, and their quality directly affects the performance and life of the equipment. Therefore, the quality inspection of shaft products is particularly important. Traditional shaft product inspection methods mainly rely on manual work, which is not only inefficient, but also easily affected by human factors, resulting in the inability to fully guarantee the accuracy and reliability of the inspection results.

[0003] With the rapid development of artificial intelligence technology, its application in industrial production is becoming more and more extensive. The data analysis method of shaft product inspection based on artificial intelligence technology can greatly improve the accuracy and efficiency of inspection and reduce the interference of human factors. However, there are still some problems with the current artificial intelligence inspection method, such as low positioning accuracy of abnormal nodes and susceptibility to noise data. Summary of the invention

[0004] In view of this, an object of an embodiment of the present invention is to provide an artificial intelligence-based method and system for analyzing shaft product inspection data.

[0005] According to one aspect of an embodiment of the present invention, there is provided a method for analyzing shaft product inspection data based on artificial intelligence, the method comprising:

[0006] Obtain shaft product inspection data;

[0007] The shaft product detection data is loaded into the abnormal node location network that has completed network parameter optimization to locate the abnormal nodes, and the abnormal node location result of the shaft product detection data is generated. The abnormal node location network is generated by performing parameter learning on the abnormal prediction network based on the first sample learning data sequence and the second sample learning data sequence. The first sample learning data sequence includes multiple first template shaft product detection data and first template abnormal location annotation data corresponding to each of the first template shaft product detection data. The second sample learning data sequence is obtained by performing abnormal node location on the third sample learning data sequence scheduled for training the abnormal prediction network based on a deep learning algorithm.

[0008] In a possible implementation of the first aspect, the abnormal node location network includes an abnormal feature encoder and an abnormal node identification unit, the abnormal node location result includes an abnormal area prediction window and abnormal node data, and the shaft product detection data is loaded into the abnormal node location network after network parameter optimization to perform abnormal node location, and the abnormal node location result of the shaft product detection data is generated, including:

[0009] Performing regular feature encoding on the shaft product detection data according to the abnormal feature encoder to generate the abnormal area prediction window, wherein the abnormal area prediction window is used to predict the window partition where the abnormal node in the shaft product detection data is located;

[0010] Abnormal nodes are identified on the shaft product detection data based on the abnormal node identification unit to generate the abnormal node data.

[0011] In a possible implementation manner of the first aspect, after performing abnormal node identification on the shaft product detection data according to the abnormal node identification unit and generating the abnormal node data, the method further includes:

[0012] Verifying the abnormal node data according to the abnormal area prediction window;

[0013] The final abnormal node positioning result of the shaft product detection data is determined based on the verification result.

[0014] In a possible implementation of the first aspect, determining a final abnormal node positioning result of the shaft product detection data based on the verification result includes:

[0015] When the verification result is that the window segment corresponding to the abnormal node data exceeds the window segment corresponding to the abnormal area prediction window, the abnormal nodes exceeding the window segment corresponding to the abnormal area prediction window are deleted from the abnormal node data to generate the final abnormal node positioning result.

[0016] In a possible implementation of the first aspect, the second sample learning data sequence includes a plurality of second template shaft product detection data, second template abnormality location annotation data corresponding to each of the second template shaft product detection data, and training label data corresponding to each of the second template shaft product detection data. Before acquiring the shaft product detection data, the method further includes:

[0017] Performing parameter learning on the candidate neural network according to the third sample learning data sequence and the initial training cost layer to generate the abnormality prediction network, wherein the third sample learning data sequence includes a plurality of third template shaft product detection data and a target data window corresponding to each of the third template shaft product detection data;

[0018] According to the first sample learning data sequence, the second sample learning data sequence and the initial training cost layer, parameter learning is performed on the abnormality prediction network to generate the abnormal node location network.

[0019] In a possible implementation of the first aspect, before performing parameter learning on the abnormality prediction network according to the first sample learning data sequence, the second sample learning data sequence, and the initial training cost layer to generate the abnormal node location network, the method further includes:

[0020] Based on the target data window corresponding to each of the third template shaft product detection data, determine the target abnormal window area in each of the third template shaft product detection data;

[0021] Performing abnormal location identification on each of the target abnormal window areas according to the deep learning algorithm, and generating an abnormal location identification result corresponding to each of the target abnormal window areas;

[0022] Based on each of the abnormal location identification results and each of the third template shaft product detection data, the second sample learning data sequence is generated.

[0023] In a possible implementation of the first aspect, during the process of training the abnormal node location network, if the sample learning data sequence is the first sample learning data sequence, the first weight in the initial training cost layer is updated; and / or, if the sample learning data sequence is the second sample learning data sequence, the second weight in the initial training cost layer is updated.

[0024] According to another aspect of an embodiment of the present invention, there is provided an artificial intelligence-based shaft product inspection data analysis system, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, communication interface and memory communicate with each other via the communication bus; the memory is used to store computer programs; the processor is used to implement any of the above-mentioned artificial intelligence-based shaft product inspection data analysis method steps when executing the computer program.

[0025] According to another aspect of an embodiment of the present invention, a readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned artificial intelligence-based shaft product inspection data analysis method can be executed.

[0026] In order to make the above-mentioned objects, features and advantages of the embodiments of the present invention more obvious and easy to understand, the embodiments will be described in detail below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0028] Figure 1 A schematic diagram of components of an artificial intelligence-based shaft product detection data analysis system provided by an embodiment of the present invention is shown;

[0029] Figure 2 A schematic flow chart of the artificial intelligence-based shaft product inspection data analysis method provided in an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field without creative work according to the embodiments of the present invention are within the scope of protection of the present invention.

[0031] The terms "first", "second", "third", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein, for example. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] Figure 1 An exemplary component diagram of an artificial intelligence-based shaft product detection data analysis system 100 is shown. The artificial intelligence-based shaft product detection data analysis system 100 may include one or more processors 104, such as one or more central processing units (CPUs), each of which may implement one or more hardware threads. The artificial intelligence-based shaft product detection data analysis system 100 may also include any storage medium 106 for storing any kind of information such as code, settings, data, etc. Non-limitingly, for example, the storage medium 106 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any storage medium may use any technology to store information. Further, any storage medium may provide volatile or non-volatile retention of information. Further, any storage medium may represent a fixed or removable component of the artificial intelligence-based shaft product detection data analysis system 100. In one case, when the processor 104 executes an associated instruction stored in any storage medium or a combination of storage media, the artificial intelligence-based shaft product detection data analysis system 100 may perform any operation of the associated instruction. The artificial intelligence-based shaft product inspection data analysis system 100 also includes one or more drive units 108 for interacting with any storage medium, such as a hard disk drive unit, an optical disk drive unit, etc.

[0033] The artificial intelligence-based shaft product inspection data analysis system 100 also includes an input / output 110 (I / O) for receiving various inputs (via an input unit 112) and for providing various outputs (via an output unit 114). A specific output mechanism may include a presentation device 116 and an associated graphical user interface (GUI) 118. The artificial intelligence-based shaft product inspection data analysis system 100 may also include one or more network interfaces 120, which are used to exchange data with other devices via one or more communication units 122. One or more communication buses 124 couple the components described above together.

[0034] The communication unit 122 can be implemented in any manner, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication unit 122 can include any combination of hardwired links, wireless links, routers, gateway functions, name artificial intelligence-based shaft product detection data analysis system 100, etc., governed by any protocol or combination of protocols.

[0035] Figure 2 The flowchart of the method and system for analyzing shaft product detection data based on artificial intelligence provided by the embodiment of the present invention is shown. The method and system for analyzing shaft product detection data based on artificial intelligence can be Figure 1 The artificial intelligence-based shaft product detection data analysis system 100 shown in the figure is executed, and the detailed steps of the artificial intelligence-based shaft product detection data analysis method are introduced as follows.

[0036] Step S110, obtaining shaft product detection data.

[0037] Step S120, loading the shaft product detection data into the abnormal node location network that has completed network parameter optimization to perform abnormal node location, and generating abnormal node location results for the shaft product detection data, wherein the abnormal node location network is generated by performing parameter learning on the abnormal prediction network based on the first sample learning data sequence and the second sample learning data sequence, wherein the first sample learning data sequence includes multiple first template shaft product detection data and first template abnormal location annotation data corresponding to each of the first template shaft product detection data, and the second sample learning data sequence is obtained by performing abnormal node location on the third sample learning data sequence scheduled for training the abnormal prediction network based on a deep learning algorithm.

[0038] In this embodiment, the server receives a set of shaft product inspection data from the connected production line inspection equipment. These data are collected in real time by high-precision sensors during the production process of shaft products, and include multiple key dimensions and surface quality parameters such as diameter, roundness, surface roughness, etc. The server organizes these data into a standard format for subsequent analysis and processing.

[0039] After acquiring the inspection data of the shaft products, the server inputs the data into an abnormal node location network that has been optimized for network parameters. This network has been trained and optimized through two sample learning data sequences (the first sample learning data sequence and the second sample learning data sequence).

[0040] The first sample learning data sequence here includes multiple first template shaft product detection data and corresponding abnormal location annotation data. These template data are typical cases selected from historical production records. They represent various possible abnormal situations and have been accurately annotated by experts. These annotated data provide a learning standard for the network.

[0041] The second sample learning data sequence is obtained by processing another set of data (the third sample learning data sequence) through a deep learning algorithm. This set of data is specifically used to train the anomaly prediction network. Through the processing of the deep learning algorithm, the network can automatically identify and mark the abnormal nodes therein, thereby generating the second sample learning data sequence.

[0042] Now, the server inputs the newly acquired shaft product inspection data into the trained abnormal node location network. The network will automatically analyze each node in the data, find out the parts similar to the abnormal nodes marked in the first sample learning data sequence, and make further judgments and optimizations based on the experience in the second sample learning data sequence, and finally generate the abnormal node location results of this inspection data. This result will specify in detail which nodes are abnormal, possible problems, and the degree of abnormality, etc., providing important basis for subsequent quality control.

[0043] Based on the above steps, by acquiring the detection data of shaft products, the subsequent data analysis has a detailed foundation. Secondly, the detection data of shaft products is processed by using the abnormal node positioning network that has completed the network parameter optimization, which realizes efficient abnormal node positioning, can quickly and accurately identify the abnormal parts in the product, and greatly improves the accuracy and efficiency of product detection. In addition, the abnormal node positioning network in the present invention is generated by parameter learning through the first sample learning data sequence and the second sample learning data sequence, so that the network has strong learning and generalization capabilities, and can cope with various complex shaft product detection data. Among them, the first sample learning data sequence contains multiple first template shaft product detection data and corresponding first template abnormal positioning annotation data, which provides the network with rich prior knowledge and annotation information; and the second sample learning data sequence is obtained by locating the abnormal nodes of the third sample learning data sequence through a deep learning algorithm, which further enhances the positioning accuracy of the network. Therefore, through artificial intelligence technology, accurate analysis of shaft product detection data is achieved, and abnormal nodes in products can be quickly and effectively identified.

[0044] In a possible implementation, the abnormal node location network includes an abnormal feature encoder and an abnormal node identification unit, the abnormal node location result includes an abnormal area prediction window and abnormal node data, and the shaft product detection data is loaded into the abnormal node location network after network parameter optimization to perform abnormal node location, and the abnormal node location result of the shaft product detection data is generated, including:

[0045] The shaft product detection data is encoded with common features according to the abnormal feature encoder to generate the abnormal area prediction window, and the abnormal area prediction window is used to predict the window partition where the abnormal node in the shaft product detection data is located.

[0046] Abnormal nodes are identified on the shaft product detection data based on the abnormal node identification unit to generate the abnormal node data.

[0047] In a possible implementation manner, after performing abnormal node identification on the shaft product detection data according to the abnormal node identification unit and generating the abnormal node data, the method further includes:

[0048] The abnormal node data is verified according to the abnormal area prediction window.

[0049] The final abnormal node positioning result of the shaft product detection data is determined based on the verification result.

[0050] In a possible implementation manner, determining the final abnormal node location result of the shaft product detection data based on the verification result includes:

[0051] When the verification result is that the window segment corresponding to the abnormal node data exceeds the window segment corresponding to the abnormal area prediction window, the abnormal nodes exceeding the window segment corresponding to the abnormal area prediction window are deleted from the abnormal node data to generate the final abnormal node positioning result.

[0052] In a possible implementation, the second sample learning data sequence includes a plurality of second template shaft product detection data, second template abnormality location annotation data corresponding to each second template shaft product detection data, and training label data corresponding to each second template shaft product detection data. Before acquiring the shaft product detection data, the method further includes:

[0053] The parameter learning of the candidate neural network is performed based on the third sample learning data sequence and the initial training cost layer to generate the abnormality prediction network. The third sample learning data sequence includes multiple third template shaft product detection data and target data windows corresponding to each of the third template shaft product detection data.

[0054] According to the first sample learning data sequence, the second sample learning data sequence and the initial training cost layer, parameter learning is performed on the abnormality prediction network to generate the abnormal node location network.

[0055] In a possible implementation, before performing parameter learning on the abnormality prediction network based on the first sample learning data sequence, the second sample learning data sequence, and the initial training cost layer to generate the abnormal node location network, the method further includes:

[0056] Based on the target data window corresponding to each of the third template shaft product detection data, a target abnormal window area is determined in each of the third template shaft product detection data.

[0057] Anomaly location and identification are performed on each of the target abnormal window areas according to the deep learning algorithm to generate an abnormality location and identification result corresponding to each of the target abnormal window areas.

[0058] Based on each of the abnormal location identification results and each of the third template shaft product detection data, the second sample learning data sequence is generated.

[0059] In a possible implementation, during the training of the abnormal node location network, if the sample learning data sequence is the first sample learning data sequence, the first weight in the initial training cost layer is updated. And / or, if the sample learning data sequence is the second sample learning data sequence, the second weight in the initial training cost layer is updated.

[0060] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0061] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, the embodiments should be considered exemplary and non-restrictive in all respects, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the present invention.

Claims

1. A method for analyzing shaft product inspection data based on artificial intelligence, characterized in that: The method comprises: Obtain shaft product inspection data; The shaft product detection data is loaded into the abnormal node location network that has completed network parameter optimization to locate the abnormal nodes, and the abnormal node location result of the shaft product detection data is generated, wherein the abnormal node location network is generated by performing parameter learning on the abnormal prediction network based on the first sample learning data sequence and the second sample learning data sequence, wherein the first sample learning data sequence includes a plurality of first template shaft product detection data and first template abnormal location annotation data corresponding to each of the first template shaft product detection data, and the second sample learning data sequence is obtained by performing abnormal node location on the third sample learning data sequence scheduled for training the abnormal prediction network based on a deep learning algorithm; The abnormal node location network includes an abnormal feature encoder and an abnormal node identification unit, the abnormal node location result includes an abnormal area prediction window and abnormal node data, the shaft product detection data is loaded into the abnormal node location network after network parameter optimization to locate the abnormal node, and the abnormal node location result of the shaft product detection data is generated, including: Performing regular feature encoding on the shaft product detection data according to the abnormal feature encoder to generate the abnormal area prediction window, wherein the abnormal area prediction window is used to predict the window partition where the abnormal node in the shaft product detection data is located; Performing abnormal node identification on the shaft product detection data according to the abnormal node identification unit to generate the abnormal node data; After performing abnormal node identification on the shaft product detection data according to the abnormal node identification unit and generating the abnormal node data, the method further includes: Verifying the abnormal node data according to the abnormal area prediction window; Determine the final abnormal node positioning result of the shaft product detection data based on the verification result; The determining of the final abnormal node positioning result of the shaft product detection data based on the verification result includes: When the verification result is that the window segment corresponding to the abnormal node data exceeds the window segment corresponding to the abnormal area prediction window, the abnormal nodes exceeding the window segment corresponding to the abnormal area prediction window are deleted from the abnormal node data to generate the final abnormal node positioning result; The second sample learning data sequence includes a plurality of second template shaft product detection data, second template abnormality location annotation data corresponding to each second template shaft product detection data, and training label data corresponding to each second template shaft product detection data. Before acquiring the shaft product detection data, the method further includes: Performing parameter learning on the candidate neural network according to the third sample learning data sequence and the initial training cost layer to generate the abnormality prediction network, wherein the third sample learning data sequence includes a plurality of third template shaft product detection data and a target data window corresponding to each of the third template shaft product detection data; According to the first sample learning data sequence, the second sample learning data sequence and the initial training cost layer, performing parameter learning on the abnormality prediction network to generate the abnormal node location network; Before performing parameter learning on the abnormality prediction network based on the first sample learning data sequence, the second sample learning data sequence and the initial training cost layer to generate the abnormal node location network, the method further includes: Based on the target data window corresponding to each of the third template shaft product detection data, determine the target abnormal window area in each of the third template shaft product detection data; Performing abnormal location identification on each of the target abnormal window areas according to the deep learning algorithm, and generating an abnormal location identification result corresponding to each of the target abnormal window areas; Based on each of the abnormal location identification results and each of the third template shaft product detection data, generate the second sample learning data sequence; During the process of training the abnormal node positioning network, if the sample learning data sequence is the first sample learning data sequence, the first weight in the initial training cost layer is updated; and / or, if the sample learning data sequence is the second sample learning data sequence, the second weight in the initial training cost layer is updated.

2. An artificial intelligence-based shaft product detection data analysis system, characterized in that: include: A processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; The memory is used to store computer programs; the processor is used to implement the steps of the artificial intelligence-based shaft product inspection data analysis method according to claim 1 when executing the computer program.

3. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for analyzing shaft product inspection data based on artificial intelligence described in claim 1 are implemented.

Images