CN113487533B - Part assembly quality digital detection system and method based on machine learning - Google Patents
Part assembly quality digital detection system and method based on machine learning Download PDFInfo
- Publication number
- CN113487533B CN113487533B CN202110548147.4A CN202110548147A CN113487533B CN 113487533 B CN113487533 B CN 113487533B CN 202110548147 A CN202110548147 A CN 202110548147A CN 113487533 B CN113487533 B CN 113487533B
- Authority
- CN
- China
- Prior art keywords
- quality
- quality detection
- image
- riveting
- defect
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 197
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000010801 machine learning Methods 0.000 title claims abstract description 23
- 239000002245 particle Substances 0.000 claims abstract description 90
- 238000005457 optimization Methods 0.000 claims abstract description 36
- 238000004458 analytical method Methods 0.000 claims abstract description 15
- 238000010191 image analysis Methods 0.000 claims abstract description 8
- 230000007547 defect Effects 0.000 claims description 93
- 239000013598 vector Substances 0.000 claims description 70
- 238000003745 diagnosis Methods 0.000 claims description 56
- 238000004422 calculation algorithm Methods 0.000 claims description 55
- 230000006870 function Effects 0.000 claims description 39
- 238000012549 training Methods 0.000 claims description 38
- 238000007781 pre-processing Methods 0.000 claims description 14
- 238000000605 extraction Methods 0.000 claims description 12
- 238000003709 image segmentation Methods 0.000 claims description 10
- 238000007689 inspection Methods 0.000 claims description 7
- 238000009411 base construction Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 abstract description 28
- 238000005553 drilling Methods 0.000 description 11
- 238000012706 support-vector machine Methods 0.000 description 7
- 238000005259 measurement Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 230000008901 benefit Effects 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 238000001914 filtration Methods 0.000 description 4
- 238000009499 grossing Methods 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 238000012843 least square support vector machine Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 230000001149 cognitive effect Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 102100029469 WD repeat and HMG-box DNA-binding protein 1 Human genes 0.000 description 1
- 101710097421 WD repeat and HMG-box DNA-binding protein 1 Proteins 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011895 specific detection Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
Abstract
The invention provides a machine learning-based part assembly quality digital detection system and a machine learning-based part assembly quality digital detection method, wherein the machine learning-based part assembly quality digital detection system comprises the following components: the image acquisition module is used for acquiring image data of at least one part; the image analysis module is used for carrying out image information analysis on the image data to obtain a plurality of quality characteristic elements corresponding to the image data; the quality detection module is used for constructing a target quality detection model based on particle swarm parameter optimization, and carrying out hole making and/or riveting quality detection on the part according to the quality characteristic elements by utilizing the target quality detection model to obtain a quality detection result of the part. The quality detection of hole making and riveting in the process of assembling the parts in a flow, modularization, automation and high efficiency mode is realized, the accuracy of quality detection results is improved, and the labor and time cost is greatly saved.
Description
Technical Field
The invention relates to the technical field of detection, in particular to a part assembly quality digital detection system and method based on machine learning.
Background
At present, most of traditional manual detection methods for hole making and riveting quality in the aircraft assembly process are adopted, wherein the manual detection method is to detect the quality in a manual mode after hole making or riveting is completed, and traditional measuring tools such as calipers, plug gauges and roughness sample blocks are adopted to measure characteristic elements of the hole making and riveting quality in the aircraft part assembly process for a plurality of times so as to judge whether the hole making quality or the riveting quality is qualified or not. However, manual detection often results in low detection efficiency and difficult traceability of quality information.
Disclosure of Invention
In view of the above problems, the invention provides a machine learning-based digital detection system and method for the assembly quality of a part, which acquire image data of the part, analyze the image data to obtain quality characteristic elements corresponding to the part, detect the quality characteristic elements by using a quality detection model to obtain a quality detection result of the part, realize automatic detection of the hole making and riveting quality of the part, solve the problems of low detection efficiency and large error in the aircraft assembly process, save labor and time cost, and complete the flow and modularization of the hole making and riveting quality detection.
According to a first aspect of the present invention, there is provided a machine learning-based part assembly quality digital inspection system, comprising:
the image acquisition module is used for acquiring image data of at least one part;
the image analysis module is used for carrying out image information analysis on the image data to obtain a plurality of quality characteristic elements corresponding to the image data;
and the quality detection module is used for constructing a target quality detection model based on particle swarm parameter optimization, and carrying out hole making and/or riveting quality detection on the part according to the quality characteristic elements by utilizing the target quality detection model to obtain a quality detection result of the part.
Optionally, the image analysis module includes:
the image preprocessing unit is used for carrying out image preprocessing on the image data to obtain target image information; wherein the image preprocessing comprises at least one of image enhancement, image segmentation, image description, image recognition and image interpretation;
the characteristic element extraction unit is used for extracting quality characteristic elements in the target image information according to a detection algorithm, wherein the quality characteristic elements comprise a hole making quality characteristic element and a riveting quality characteristic element; the hole forming quality characteristic elements comprise at least one of aperture precision, flatness, perpendicularity, roundness, roughness, burrs and cracks, and the riveting quality characteristic elements comprise at least one of step differences and skin pits.
Optionally, the quality detection module includes:
the sample feature extraction unit is used for extracting a plurality of sample feature elements and constructing a sample feature library, wherein the sample feature library comprises a hole-making sample feature library and a riveting sample feature library;
the model building unit is used for building a least square vector machine model;
the model training unit is used for training and optimizing the least square vector machine model by utilizing sample feature elements in the sample feature library, and generating the target quality detection model based on particle swarm parameter optimization;
And the quality diagnosis unit is used for carrying out hole forming and/or riveting quality detection on a plurality of quality characteristic elements corresponding to the part by utilizing the target quality detection model to obtain a quality detection result of the part.
Optionally, the model training unit is further configured to:
generating a quality prediction sample set according to sample feature elements in the sample feature library;
carrying out optimization training on the least square vector machine model based on the quality prediction sample set by using a particle swarm algorithm to obtain kernel function parameters and regularization parameters corresponding to the least square vector machine model optimized based on the particle swarm algorithm;
and generating the target quality detection model based on particle swarm parameter optimization by using the kernel function parameters and regularization parameters.
Optionally the model training unit is further configured to:
taking the position vector of each particle in the particle swarm algorithm as a quality parameter vector representing a kernel function parameter and a regularization parameter;
and iteratively calculating a particle position optimal solution by using the particle swarm algorithm to obtain a target quality parameter vector, and generating a kernel function parameter and a regularization parameter corresponding to a least square vector machine model optimized based on the particle swarm algorithm according to the target quality parameter vector.
Optionally, the quality detection module further comprises:
an error analysis unit for utilizing a quality diagnostic error calculated for the target quality detection model;
dividing the quality prediction sample set into a plurality of sub-sample sets, and performing optimization training on the least square vector machine model by using the plurality of sub-sample sets based on a particle swarm algorithm to generate a plurality of corresponding quality detection sub-models;
calculating an average quality diagnostic error for the plurality of quality detection sub-models;
the model training unit is further configured to: comparing the quality diagnosis error with the average quality diagnosis error, and when the quality diagnosis error is larger than the average quality diagnosis error, carrying out optimization training on the least square vector machine model again by using a particle swarm algorithm until a target quality detection model with the quality diagnosis error smaller than the average quality diagnosis error is obtained.
Optionally, the system further comprises:
the characteristic knowledge base construction module is used for constructing a quality characteristic knowledge base according to defect characteristics, defect types and defect reasoning rules corresponding to unqualified hole making and/or riveting of a plurality of quality detection results;
and the intelligent tracing module is used for tracing defects of the hole making and/or riveting of the part by utilizing the quality characteristic knowledge base when the quality detection result is unqualified.
Optionally, the intelligent traceability module includes:
the quality intelligent traceability unit is used for identifying defect characteristics and defect types corresponding to unqualified hole making and/or riveting according to defect reasoning rules in the quality characteristic knowledge base;
wherein the defect feature comprises at least one of a shape feature, a location feature, and a surface feature.
Optionally, the intelligent traceability module further includes:
the self-learning unit is used for carrying out online learning by simulating an artificial defect recognition method, generating a defect reasoning rule and storing the defect reasoning rule into the quality characteristic knowledge base.
According to a second aspect of the present invention, a machine learning-based digital inspection method for assembly quality of parts includes:
acquiring image data of at least one part;
performing image information analysis on the image data to obtain a plurality of quality feature elements corresponding to the image data;
and constructing a target quality detection model based on particle swarm parameter optimization, and carrying out hole making and/or riveting quality detection on the part according to the quality characteristic elements by utilizing the target quality detection model to obtain a quality detection result of the part.
According to the machine learning-based part assembly quality digital detection system and method, the image data of the part are preprocessed, the quality characteristic elements of the part are extracted, the least square vector machine model with optimized particle swarm parameters is constructed as a quality detection model, the error of the quality detection model is analyzed, the quality detection model with smaller error is selected as a target quality detection model, the quality characteristic elements are detected by the target quality detection model, automatic identification and diagnosis of hole making and riveting quality in the part assembly process are realized, efficient, flow-through and modularized automatic detection is realized, automatic analysis and intelligent traceability of quality defects are realized by establishing a quality characteristic knowledge base, and the quality defect decision-making reasoning capacity is improved by self-learning of the detection system, so that the labor cost is greatly saved, and the error rate of quality detection is reduced.
The foregoing description is only an overview of the present invention, and is intended to be implemented in accordance with the teachings of the present invention in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present invention more readily apparent.
The above, as well as additional objectives, advantages, and features of the present invention will become apparent to those skilled in the art from the following detailed description of a specific embodiment of the present invention when read in conjunction with the accompanying drawings.
Drawings
Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to designate like parts throughout the figures. In the drawings:
FIG. 1 is a schematic diagram of a machine learning-based digital inspection system for quality of assembly of parts according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a machine learning-based digital inspection system for quality of assembly of parts according to another embodiment of the present invention;
FIG. 3 shows an effect schematic diagram of a quality feature knowledge base management interface provided by an embodiment of the present invention;
FIG. 4 is a schematic flow chart of a machine learning-based method for digitally detecting quality of assembly of a part according to an embodiment of the present invention;
fig. 5 shows a schematic flow chart of a particle swarm algorithm according to an embodiment of the invention.
Detailed Description
In order that those skilled in the art will better understand the present application, a technical solution in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present application without making any inventive effort, shall fall within the scope of the present application. It should be noted that, without conflict, the embodiments of the present application and features of the embodiments may be combined with each other.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the application described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The embodiment of the invention provides a machine learning-based part assembly quality digital detection system, as shown in fig. 1, which can comprise: the device comprises an image acquisition module 110, an image analysis module 120 and a quality detection module 130.
An image acquisition module 110 for acquiring image data of at least one part.
The image acquisition module 110 may be provided with an illumination system, a CCD camera and an image acquisition card, and performs image data acquisition on the parts through the illumination system and the CCD camera, and transmits the acquired image data to a computer device through the image acquisition card for subsequent processing.
The image analysis module 120 is configured to perform image information analysis on the image data to obtain a plurality of quality feature elements corresponding to the image data.
The image information analysis can comprise preprocessing an image, recovering image distortion caused by transferring image data between different media, and dividing, describing and identifying the image data so as to facilitate the extraction of subsequent quality characteristic elements.
The quality characteristic elements refer to image characteristics capable of reflecting the drilling and riveting quality in the part assembly process, the measurement is required to be carried out through a high-precision dimension measuring instrument, an operator can select geometric elements required to be measured by a workpiece through a man-machine interaction interface arranged in the system, the measurement elements can comprise points, straight lines, circles, arcs, ellipses, rectangles, key grooves, circular rings, planes, cylinders, cones, balls, open curves, closed curves, focal planes and the like, and the extracted quality characteristic elements can comprise aperture precision, roundness, flatness, burrs, cracks, skin level differences, skin pits and the like.
The quality detection module 130 is configured to construct a target quality detection model based on particle swarm parameter optimization, and perform hole forming and/or riveting quality detection on the part according to the quality feature element by using the target quality detection model to obtain a quality detection result of the part.
The quality characteristic elements of the hole making and riveting in the embodiment of the invention have the characteristics of multiple types, small samples and uncertainty, the hole making and riveting quality diagnosis is difficult to realize by a general method, the theoretical basis of the model performance of the least square vector machine is optimized by adopting a particle swarm algorithm, the hole making and riveting quality diagnosis model of the least square support vector machine based on the particle swarm parameter optimization is constructed, and the hole making and riveting quality of the part is detected by the model, so that the detection result of the hole making and riveting quality of the part is obtained.
According to the embodiment of the invention, through the process of collecting, processing and detecting the part image, the automatic detection of the hole making and riveting quality of the part is realized, the problems of low detection efficiency and large error in the aircraft assembly process are solved, the labor and time cost are saved, and the detection efficiency and accuracy of the hole making and riveting quality are greatly improved.
In an alternative embodiment of the present invention, as shown in fig. 2, the image analysis module 120 may include: an image preprocessing unit 121 and a feature element extraction unit 122.
An image preprocessing unit 121, configured to perform image preprocessing on image data to obtain target image information.
Among other things, image preprocessing may include image enhancement, image segmentation, image description, image recognition, image interpretation, and so forth.
Image enhancement mainly performs processing such as enhancement, smoothing, filtering and the like on an image. Image quality is improved by taking image data of the part and transferring the image data between different media, which generally results in image distortion, which is somewhat different from the original image, using methods of image enhancement, smoothing, filtering.
Image segmentation is the process of dividing image data into objects of a certain meaning. The target area of the image data is analyzed and detected after being segmented, the scene with continuously changed gray scale on the image data can be changed into a relatively independent 'target geometry', the important characteristics of the image data are enhanced, redundant information is removed, and the target characteristics can be successfully extracted and classified.
Image description is the process of extracting features from image data for identification purposes. Ideally, the image description is independent of the size, orientation, and location of the object, and should contain enough information available for authentication to uniquely identify an object among a plurality of objects.
Image recognition is a marking process in which the function of a recognition algorithm is to identify each segmented object in the scene and to assign that object some sort of marking.
Image interpretation may be viewed as a higher level of cognitive behavior that a machine vision system has on its environment.
The target image information which is convenient for the subsequent extraction of quality characteristic elements is obtained by carrying out image enhancement, image segmentation, image description, image recognition, image interpretation and other processes on the image data.
The feature element extracting unit 122 is configured to extract quality feature elements in the target image information according to a detection algorithm, where the quality feature elements include a hole forming quality feature element and a caulking quality feature element.
The quality characteristic elements in the target image information are extracted mainly based on technical researches of image enhancement and image segmentation, and the hole making quality characteristic elements and the riveting quality characteristic elements are extracted according to detection algorithms of different quality characteristic elements. Wherein, the quality characteristic elements of the hole forming can comprise aperture precision, flatness, perpendicularity, roundness, roughness, burrs, cracks and the like; the rivet quality feature may include a step, skin recess, etc.
In practical application, the quality feature elements and the detection algorithm thereof can be flexibly set according to specific requirements, and the embodiment of the invention is not limited to the specific requirements.
In an alternative embodiment of the present invention, as shown in fig. 2, the quality detection module 130 may include: a sample feature extraction unit 131, a model construction unit 132, a model training unit 133, and a quality diagnosis unit 134.
The sample feature extraction unit 131 is configured to extract a plurality of sample feature elements, and construct a sample feature library, where the sample feature library includes a hole-making sample feature library and a riveting sample feature library.
Collecting quality characteristic elements of a plurality of parts, taking an extraction part as a sample characteristic element, and establishing a sample characteristic library, wherein the quality characteristic element comprises a drilling quality characteristic element and a riveting quality characteristic element, and the corresponding sample characteristic library can also comprise a drilling sample characteristic library and a riveting sample characteristic library.
The model building unit 132 is configured to build a least squares vector machine model.
The embodiment of the invention adopts a least square vector machine model, is a topological structure of a support vector machine determined according to a support vector, and takes structural risk minimization as a basic principle. The kernel functions in different forms are selected to generate the nonlinear classifier in different forms, and when the support vector machine solves the problem of nonlinear separability, the mapping relation is used for mapping the input low-dimensional space to the high-dimensional feature space, the nonlinear problem is converted into the linear problem, and then the mode for solving the linear classification can be used for solving the nonlinear classification.
Based on the characteristics of multiple types, small samples and uncertainty of the characteristic elements of the drilling and riveting quality, a radial basis function which is simple and easy to operate and has high fitting precision and is suitable for the small samples is selected, namely, a Gaussian kernel function is used as a kernel function, and the following formula is as follows:
wherein: delta 2 Is the width parameter of Gaussian kernel function, X is the center of kernel function, X i Representing a vector defines a nonlinear mapping from the precision property input space to the high dimensional precision feature space, thereby controlling the complexity of the final solution.
The least square support vector machine uses the secondary norm of the approximation error which allows the precision sample to exist as a loss function, and the application range of the traditional support vector machine is expanded. The least square support vector machine transforms the original quality diagnostic data into another new high-dimensional quality feature space, thereby improving the utilization rate of the quality diagnostic data. The optimization problem is converted into a solution linear equation set by using a regular least square equation with an equality constraint condition as a cost function, and the solution is carried out by using an iteration method such as a common rail gradient method, so that the training speed is faster than that of the traditional support vector machine.
The learning program for diagnosing the drilling and riveting quality by determining the least square vector machine model, namely the learning process of model training, and the preliminarily constructed least square vector machine model is obtained by writing the learning process of model training.
The model training unit 133 is configured to perform training optimization on the least squares vector machine model by using sample feature elements in the sample feature library, and generate a target quality detection model based on particle swarm parameter optimization.
Training and optimizing a least square vector machine model by using sample feature elements in a sample feature library, and firstly, generating a quality prediction sample set according to the sample feature elements in the sample feature library.
Sample feature elements in the sample feature library comprise a plurality of quality feature elements, and a quality prediction sample set is generated by setting corresponding quality detection results for the plurality of quality feature elements.
Further, the particle swarm optimization can be utilized to carry out optimization training on the least square vector machine model based on the quality prediction sample set, and kernel function parameters and regularization parameters corresponding to the least square vector machine model optimized based on the particle swarm optimization are obtained.
That is, the prediction sample data in the input quality prediction sample set can be divided into a training set and a testing set, and the initially constructed least squares vector machine model is optimally trained by utilizing a particle swarm algorithm based on the training set and the testing set, and the optimization training is research on optimizing the kernel function parameter delta and the regularization parameter C of the support vector machine by utilizing the particle swarm algorithm. The particle swarm algorithm has the advantages of simplicity, practicability, higher convergence speed, fewer adjustable parameters and the like, and is a rapid and effective optimization algorithm; the kernel function parameter delta affects whether the quality feature is separable in a high dimension; the size of the regularization parameter C mainly influences the intensity of the generalization capability of the support vector machine. The selection of different kernel parameters delta and regularization parameters C affects the classification ability of the support vector machine model. And (3) establishing an optimal drilling and riveting quality diagnosis model by iteratively searching an optimal regularization parameter C and a kernel function parameter delta based on a particle swarm algorithm.
Further, a position vector of each particle in the particle swarm algorithm may be taken as a quality parameter vector representing the kernel function parameter δ and the regularization parameter C.
That is, the position of each mass particle in the population represents a group of mass parameter vectors (regularization parameter C, kernel function parameter δ), a particle swarm algorithm is utilized to iteratively calculate a particle position optimal solution to obtain a target mass parameter vector, and the kernel function parameter δ and the regularization parameter C corresponding to a least squares vector machine model optimized based on the particle swarm algorithm are generated according to the target mass parameter vector.
Further, a target quality detection model optimized based on the particle swarm parameters is generated by using the kernel function parameters delta and the regularization parameters C.
That is, the obtained kernel function parameter δ and regularization parameter C may be substituted into the constructed least squares vector machine model to obtain a target quality detection model based on particle swarm parameter optimization.
And the quality diagnosis unit 134 is used for performing hole forming and/or riveting quality detection on a plurality of quality characteristic elements corresponding to the part by utilizing the target quality detection model to obtain a quality detection result of the part.
After the target quality detection model is determined, quality characteristic elements corresponding to the detected parts can be input into the target quality detection model, detection results of hole making and riveting quality are automatically output, whether the quality of the parts is qualified or not is indicated, and specific detection results, such as measurement results of whether the hole making has cracks, whether the diameter meets the requirement, whether the riveting has pits or not, and the like, can also be directly output.
In an alternative embodiment of the present invention, as shown in fig. 2, the quality detection module 130 further includes:
an error analysis unit 135 for calculating a quality diagnostic error of the target quality detection model; dividing a quality prediction sample set into a plurality of sub-sample sets, and carrying out optimization training on a least square vector machine model based on the sub-sample sets by using a particle swarm algorithm to generate a plurality of corresponding quality detection sub-models; an average quality diagnostic error for the plurality of quality detection sub-models is calculated.
That is, by error analysis, the detection error of the least squares vector machine model based on the particle swarm algorithm is corrected, and the target quality detection model with higher detection accuracy is obtained.
The calculation of the average quality diagnosis error is to divide a quality prediction sample set into a plurality of sub-sample sets, respectively utilize the plurality of sub-sample sets to optimize and train a least square vector machine model based on a particle swarm algorithm, generate a plurality of corresponding quality detection sub-models based on particle swarm parameter optimization, utilize the plurality of quality detection sub-models to carry out hole making and riveting quality detection on the part to obtain a plurality of quality detection sub-results, compare the plurality of quality detection sub-results with an actual measurement result measured manually to obtain a plurality of corresponding quality diagnosis errors, and calculate the average quality diagnosis error of the plurality of quality detection sub-models according to the plurality of quality diagnosis errors.
The calculation of the quality diagnosis error is to take the least square vector machine model which is based on the training and optimizing of the quality prediction sample set by utilizing the particle swarm algorithm as a target quality detection model; and performing hole making and riveting quality detection on the part by using the target quality detection model to obtain a quality detection result, and calculating a quality diagnosis error of the quality detection model according to the quality detection result and an actual measurement result measured manually.
Further, the model training unit 133 may be further configured to: comparing the quality diagnosis error with the average quality diagnosis error, and when the quality diagnosis error is smaller than the average quality diagnosis error, not updating the target quality detection model; when the quality diagnosis error is larger than the average quality diagnosis error, the target quality detection model can be finer, namely the least square vector machine model is optimized and trained again by utilizing a particle swarm algorithm until the target quality detection model with the quality diagnosis error smaller than the average quality diagnosis error is obtained.
That is, by comparing the quality diagnostic error of the quality detection model with the average quality diagnostic error of the plurality of quality detection sub-models, it is determined whether the detection error of the quality detection model is within the allowable range. If the quality diagnosis error is smaller than the average quality diagnosis error, taking the quality detection model as a final target quality detection model; if the quality diagnosis error of the quality detection model is larger than the average quality diagnosis error, the least square vector machine model is subjected to optimization training again by using a particle swarm algorithm, namely, the kernel function parameter delta and the regularization parameter C are reselected until a target quality detection model with the quality diagnosis error smaller than the average quality diagnosis error is obtained.
By error correction, the detection accuracy of the target quality detection model can be improved, the detection efficiency and property loss caused by detection errors are reduced, and the labor and time cost are greatly saved.
Optionally, as shown in fig. 2, a machine learning-based part assembly quality digital detection system according to an embodiment of the present invention may further include: a feature knowledge base construction module 140 and an intelligent traceback module 150.
The feature knowledge base construction module 140 is configured to establish a quality feature knowledge base according to defect features, defect types and defect reasoning rules corresponding to the unqualified hole making and/or riveting of the plurality of quality detection results.
That is, for quality feature elements and corresponding features that collect a plurality of quality features of hole making and riveting for which the quality detection result is not qualified, the defect features may be classified into three aspects of shape features, position features, surface features, and defect types may include roundness defects, aperture accuracy defects, space distance defects, perpendicularity defects, roughness defects, crack defects, burr defects, skin recess defects, rivet step defects, and the like. The defect reasoning rule is a reasoning rule for simulating the manual identification of the quality defects by the system, and can be used for automatically carrying out reasoning decision on defect characteristics and defect types according to quality detection results.
By storing the defect characteristics, defect types and defect reasoning rules into the corresponding knowledge base, a special quality characteristic knowledge base is formed, and analysis of quality defects and tracing of defect characteristic elements are supported.
In addition, the system can also set a management interface of the quality feature knowledge base, as shown in fig. 3, an operator can complete setting management of the quality feature knowledge base through man-machine interaction, add, modify, delete, retrieve and other operations on defect types and defect features, and can also perform manual operations such as auditing, checking, editing and the like on the quality feature knowledge base according to requirements.
And the intelligent tracing module 150 is used for tracing the defects of the hole making and/or riveting of the parts by utilizing the quality characteristic knowledge base when the quality detection result is unqualified.
In an alternative embodiment of the present invention, as shown in fig. 2, the intelligent traceback module 150 may include a quality intelligent traceback unit 151 and a self learning unit 152.
The quality intelligent traceability unit 151 is configured to identify defect characteristics and defect types corresponding to the unqualified hole making and/or riveting according to defect reasoning rules in the quality feature knowledge base.
When the detection result of the hole making and/or riveting quality of the part is unqualified, the defect reasoning rule is directly utilized to automatically identify and output and display the defect characteristics and defect types of the part.
The self-learning unit 152 is configured to perform online learning by simulating an artificial defect recognition method, generate a defect inference rule, and store the defect inference rule in the quality feature knowledge base.
That is, the system can simulate the mode and method of manually solving the problem through the reasoning decision of the computer program, generate the defect reasoning rule which automatically identifies the defect characteristics and defect types based on the quality detection result, and store the defect reasoning rule in the quality characteristic knowledge base for subsequent calling and management.
The digital detection system for the assembly quality of the parts, which is provided by the invention, can be used for completing automatic identification and diagnosis of the drilling and riveting quality in the assembly process of the parts, realizing efficient, procedural and modularized automatic detection, automatically analyzing and intelligently tracing the quality defects by establishing a quality characteristic knowledge base, improving the decision-making and reasoning capacity of the quality defects by self-learning of the detection system, greatly saving labor cost, reducing the error rate of quality detection, providing multiple functions for calling and managing by establishing a man-machine interaction interface, and improving the convenience of use of users.
Based on the same inventive concept, the embodiment of the invention also provides a machine learning-based part assembly quality digital detection method, as shown in fig. 4, which at least can comprise the following steps S401 to S403.
Step S401, acquiring image data of at least one part.
And the lighting system and the CCD camera arranged by the system are used for collecting image data of the parts, and the collected image data is transmitted or stored to computer equipment by utilizing an image collecting card for subsequent processing.
Step S402, image information analysis is carried out on the image data, and a plurality of quality feature elements corresponding to the image data are obtained.
And carrying out preprocessing such as image enhancement, image segmentation, image description, image recognition, image interpretation and the like on the acquired image data to obtain target image information, and extracting quality feature elements in the target image information according to a detection algorithm.
Further, extracting quality characteristic elements in the target image information according to a detection algorithm, wherein the quality characteristic elements comprise a hole making quality characteristic element and a riveting quality characteristic element; the hole-making quality characteristic elements comprise at least one of aperture precision, flatness, perpendicularity, roundness, roughness, burrs and cracks, and the riveting quality characteristic elements comprise at least one of step differences and skin pits.
Step S403, a target quality detection model based on particle swarm parameter optimization is constructed, and the target quality detection model is utilized to carry out hole making and/or riveting quality detection on the parts according to quality characteristic elements, so that quality detection results of the parts are obtained.
The quality detection result can be whether the hole making and/or riveting conditions in the part assembling process are qualified or not.
According to the embodiment of the invention, the automatic and flow-based hole making and riveting quality detection is realized by constructing the hole making and riveting quality diagnosis model of the least square support vector machine based on particle swarm parameter optimization.
Further, in the machine learning-based part assembly quality digital detection method provided by the embodiment of the invention, the step S402 may specifically further include the following substeps S402-1 to S402-3:
step S402-1: and performing image preprocessing on the image data to obtain target image information.
Among other things, image preprocessing may include image enhancement, image segmentation, image description, image recognition, image interpretation, and so forth.
Image enhancement mainly performs processing such as enhancement, smoothing, filtering and the like on an image. Image quality is improved by taking image data of the part and transferring the image data between different media, which generally results in image distortion, which is somewhat different from the original image, using methods of image enhancement, smoothing, filtering.
Image segmentation is the process of dividing image data into objects of a certain meaning. The target area of the image data is analyzed and detected after being segmented, the scene with continuously changed gray scale on the image data can be changed into a relatively independent 'target geometry', the important characteristics of the image data are enhanced, redundant information is removed, and the target characteristics can be successfully extracted and classified.
Image description is the process of extracting features from image data for identification purposes. Ideally, the image description is independent of the size, orientation, and location of the object, and should contain enough information available for authentication to uniquely identify an object among a plurality of objects.
Image recognition is a marking process in which the function of a recognition algorithm is to identify each segmented object in the scene and to assign that object some sort of marking.
Image interpretation may be viewed as a higher level of cognitive behavior that a machine vision system has on its environment.
The target image information which is convenient for the subsequent extraction of quality characteristic elements is obtained by carrying out image enhancement, image segmentation, image description, image recognition, image interpretation and other processes on the image data.
Step S402-2: and extracting quality characteristic elements in the target image information according to a detection algorithm, wherein the quality characteristic elements comprise a drilling quality characteristic element and a riveting quality characteristic element.
The different quality characteristic elements are provided with corresponding detection algorithms, and the drilling quality characteristic elements and the riveting quality characteristic elements are extracted according to the detection algorithms. Wherein, the quality characteristic elements of the hole forming can comprise aperture precision, flatness, perpendicularity, roundness, roughness, burrs, cracks and the like; the rivet quality feature may include a step, skin recess, etc.
Further, in the machine learning-based part assembly quality digital detection method provided by the embodiment of the present invention, the step S403 may specifically further include the following substeps S403-1 to S403-4:
step S403-1: and extracting a plurality of sample feature elements to construct a sample feature library, wherein the sample feature library comprises a hole-making sample feature library and a riveting sample feature library.
Collecting quality characteristic elements of a plurality of parts, taking an extraction part as a sample characteristic element, and establishing a sample characteristic library, wherein the quality characteristic element comprises a drilling quality characteristic element and a riveting quality characteristic element, and the corresponding sample characteristic library can also comprise a drilling sample characteristic library and a riveting sample characteristic library.
Step S403-2: and constructing a least square vector machine model.
And constructing a least square vector machine model, namely determining a learning program for diagnosing hole making and riveting quality by the least square vector machine model, namely, a learning process of model training, and obtaining the initially constructed least square vector machine model by writing the learning process of model training.
Step S403-3: and training and optimizing the least square vector machine model by using sample feature elements in a sample feature library to generate a target quality detection model based on particle swarm parameter optimization.
Specifically, a quality prediction sample set may be generated from sample feature elements in a sample feature library; and carrying out optimization training on the least square vector machine model based on the quality prediction sample set by using a particle swarm algorithm to obtain a kernel function parameter delta and a regularization parameter C corresponding to the least square vector machine model optimized based on the particle swarm algorithm.
The quality prediction sample set comprises sample feature elements and corresponding quality detection results, the quality prediction sample set can be divided into a training set and a testing set, the initially constructed least square vector machine model is optimally trained based on the training set and the testing set, and kernel function parameters delta and regularization parameters C corresponding to the least square vector machine model can be obtained by using a particle swarm algorithm.
Taking the position vector of each particle in a particle swarm algorithm as a quality parameter vector representing a kernel function parameter and a regularization parameter; iteratively calculating a particle position optimal solution by using a particle swarm algorithm to obtain a target quality parameter vector, and generating a kernel function parameter and a regularization parameter corresponding to a least square vector machine model optimized based on the particle swarm algorithm according to the target quality parameter vector; and generating a target quality detection model based on particle swarm parameter optimization by using the kernel function parameters and the regularization parameters.
That is, the position vector of each particle in the particle swarm algorithm may be taken as a mass parameter vector representing the kernel function parameter δ and the regularization parameter C, and the position of each mass particle in the swarm represents a set of mass parameter vectors (regularization parameter C, kernel function parameter δ).
When the least square vector machine model is optimized, the predicted sample data in the quality predicted sample set can be normalized, namely, the predicted sample data is scaled by equal proportion, so that the test data is arranged between-1 and-1, and the subsequent particle swarm optimization calculation is facilitated.
As shown in fig. 5, the basic flow of the particle swarm algorithm includes steps S501 to S505:
step S501: the algorithm initializes all particle velocities and positions.
Randomly initializing the speeds and positions of all particles in a particle swarm, wherein the optimal position of each particle is set as an initial position, and the optimal position of the population is set as a global optimal position of the initial particle;
step S502: the velocities and positions of all particles are updated.
Adjusting the current speed and position of the particles according to the following speed v and position x formula;
v i (t+1)＝ωv i (t)+C 1 r 1 (P i,best )(t)+C 2 r 2 [g i,best (t)-x i (t)]
x i (t+1)＝x i (t)+v i (k+1)
wherein i=1, 2,3 … m; c (C) 1 And C 2 As the quality acceleration factor, a value range of (0, 2) is usually adopted; r is (r) 1 And r 2 Is a relatively independent mass random function varying between 0 and 1.
Step S503: and selecting the current optimal particle position.
Comparing the current position of each particle with the historical position of each particle, if the current position is better than the historical optimal position, taking the current position as the optimal position of the individual, otherwise, continuing iteration until the optimal solution is updated along the historical optimal position. The optimal solution of individual particles, i.e. the individual extremum (P i Best); the optimal solution of the whole population, i.e. global extremum (g i Best), the particle i updates the speed and position according to the speed and position formula, and continuously updates the individual extremum and the global extremum.
Step S504: and selecting the optimal particle position of the population.
And comparing the local optimal position of each particle with the group optimal position, and if the local optimal position is better than the group optimal position, replacing, otherwise, keeping the group optimal position unchanged.
Step S505: and judging whether a termination condition is satisfied.
When the optimal group position where the position fix is no longer replaced is obtained, the algorithm termination condition is detected.
And (3) iteratively calculating a particle position optimal solution by using a particle swarm algorithm to obtain a target quality parameter vector, a kernel function parameter delta and a regularization parameter C.
Step S403-4: calculating a quality diagnosis error of the target quality detection model; dividing a quality prediction sample set into a plurality of sub-sample sets, and carrying out optimization training on a least square vector machine model based on the sub-sample sets by using a particle swarm algorithm to generate a plurality of corresponding quality detection sub-models; an average quality diagnostic error for the plurality of quality detection sub-models is calculated.
And calculating a quality diagnosis error of a quality detection model corresponding to the quality prediction sample set. And dividing the quality prediction sample set into a plurality of sub-sample sets, performing hole making and riveting quality detection on the part by utilizing a quality detection sub-model corresponding to the plurality of sub-sample sets to obtain a plurality of quality detection sub-results, comparing the plurality of quality detection sub-results with an actual measurement result manually measured to obtain a plurality of corresponding quality diagnosis errors, and calculating an average value to obtain an average quality diagnosis error. And comparing the quality detection sub-model with the average quality diagnosis error, and analyzing the detection error of the quality detection model.
Comparing the quality diagnosis error with the average quality diagnosis error, and when the quality diagnosis error is smaller than the average quality diagnosis error, not updating the target quality detection model; when the quality diagnosis error is larger than the average quality diagnosis error, the least square vector machine model is optimized and trained again by using the particle swarm algorithm until a target quality detection model with the quality diagnosis error smaller than the average quality diagnosis error is obtained.
Further, the method for digitally detecting the assembly quality of the parts based on the machine learning provided by the embodiment of the invention can further comprise steps S404 to S405:
Step S404: and carrying out hole forming and/or riveting quality detection on a plurality of quality characteristic elements corresponding to the part by utilizing the target quality detection model to obtain a quality detection result of the part.
The quality detection result can be whether the hole making and/or riveting conditions in the part assembling process are qualified or not.
Optionally, a quality feature knowledge base may be established according to defect features, defect types and defect reasoning rules corresponding to the unqualified hole making and/or riveting of the plurality of quality detection results.
Wherein the defect feature comprises at least one of a shape feature, a location feature, and a surface feature; the defect types include at least one of roundness defect, aperture accuracy defect, space distance defect, perpendicularity defect, roughness defect, crack defect, burr defect, skin dent defect, and rivet step defect.
The defect reasoning rule can be generated by online learning by simulating an artificial defect recognition method, and can also be stored in a quality characteristic knowledge base.
Step S405: and when the quality detection result is unqualified, performing defect tracing on the hole making and/or riveting of the part by utilizing the quality characteristic knowledge base.
That is, when the detection result of the hole making and/or riveting quality of the part is unqualified, the system can directly identify the defect characteristics and defect types corresponding to the unqualified hole making and/or riveting by utilizing the defect reasoning rule in the quality characteristic knowledge base, and automatically identify and output and display the defect characteristics and defect types of the part.
According to the machine learning-based part assembly quality digital detection system and method, the image data of the part are preprocessed, the quality characteristic elements of the part are extracted, the least square vector machine model with optimized particle swarm parameters is constructed as the quality detection model, the error of the quality detection model is analyzed, the quality detection model with smaller error is selected as the target quality detection model, the quality characteristic elements are detected by the target quality detection model, automatic identification and diagnosis of hole making and riveting quality in the part assembly process are realized, efficient, flow-through and modularized automatic detection is realized, automatic analysis and intelligent traceability of quality defects are realized by establishing a quality characteristic knowledge base, the quality defect decision-making capability is improved by self-learning of the detection system, the labor cost is greatly saved, the quality detection error rate is reduced, and multiple functions are called and managed by establishing a man-machine interaction interface, so that convenience and comfort of use of users are improved.
It will be clear to those skilled in the art that the specific working processes of the above-described systems, modules and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein for brevity.
In addition, each functional unit in the embodiments of the present invention may be physically independent, two or more functional units may be integrated together, or all functional units may be integrated in one processing unit. The integrated functional units may be implemented in hardware or in software or firmware.
Those of ordinary skill in the art will appreciate that: the integrated functional units, if implemented in software and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in essence or in whole or in part in the form of a software product stored in a storage medium, comprising instructions for causing a computing device (e.g., a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the embodiments of the present invention when the instructions are executed. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a read-only memory (ROM), a random-access memory (RAM), a magnetic disk, or an optical disk, etc.
Alternatively, all or part of the steps of implementing the foregoing method embodiments may be implemented by hardware (such as a personal computer, a server, or a computing device such as a network device) associated with program instructions, where the program instructions may be stored on a computer-readable storage medium, and where the program instructions, when executed by a processor of the computing device, perform all or part of the steps of the method according to the embodiments of the present invention.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all technical features thereof can be replaced by others within the spirit and principle of the present invention; such modifications and substitutions do not depart from the scope of the invention.
Claims (9)
1. A machine learning-based digital inspection system for assembly quality of a part, comprising:
the image acquisition module is used for acquiring image data of at least one part;
The image analysis module is used for carrying out image information analysis on the image data to obtain a plurality of quality characteristic elements corresponding to the image data;
the quality detection module is used for constructing a target quality detection model based on particle swarm parameter optimization, and carrying out hole making and/or riveting quality detection on the part according to the quality characteristic elements by utilizing the target quality detection model to obtain a quality detection result of the part;
the characteristic knowledge base construction module is used for constructing a quality characteristic knowledge base according to defect characteristics, defect types and defect reasoning rules corresponding to unqualified hole making and/or riveting of a plurality of quality detection results;
and the intelligent tracing module is used for tracing defects of the hole making and/or riveting of the part by utilizing the quality characteristic knowledge base when the quality detection result is unqualified.
2. The system of claim 1, wherein the image analysis module comprises:
the image preprocessing unit is used for carrying out image preprocessing on the image data to obtain target image information; wherein the image preprocessing comprises at least one of image enhancement, image segmentation, image description, image recognition and image interpretation;
The characteristic element extraction unit is used for extracting quality characteristic elements in the target image information according to a detection algorithm, wherein the quality characteristic elements comprise a hole making quality characteristic element and a riveting quality characteristic element; the hole forming quality characteristic elements comprise at least one of aperture precision, flatness, perpendicularity, roundness, roughness, burrs and cracks, and the riveting quality characteristic elements comprise at least one of step differences and skin pits.
3. The system of claim 1, wherein the quality detection module comprises:
the sample feature extraction unit is used for extracting a plurality of sample feature elements and constructing a sample feature library, wherein the sample feature library comprises a hole-making sample feature library and a riveting sample feature library;
the model building unit is used for building a least square vector machine model;
the model training unit is used for training and optimizing the least square vector machine model by utilizing sample feature elements in the sample feature library, and generating the target quality detection model based on particle swarm parameter optimization;
and the quality diagnosis unit is used for carrying out hole forming and/or riveting quality detection on a plurality of quality characteristic elements corresponding to the part by utilizing the target quality detection model to obtain a quality detection result of the part.
4. A system according to claim 3, wherein the model training unit is further configured to:
generating a quality prediction sample set according to sample feature elements in the sample feature library;
carrying out optimization training on the least square vector machine model based on the quality prediction sample set by using a particle swarm algorithm to obtain kernel function parameters and regularization parameters corresponding to the least square vector machine model optimized based on the particle swarm algorithm;
and generating the target quality detection model based on particle swarm parameter optimization by using the kernel function parameters and regularization parameters.
5. The system of claim 4, wherein the model training unit is further configured to:
taking the position vector of each particle in the particle swarm algorithm as a quality parameter vector representing a kernel function parameter and a regularization parameter;
and iteratively calculating a particle position optimal solution by using the particle swarm algorithm to obtain a target quality parameter vector, and generating a kernel function parameter and a regularization parameter corresponding to a least square vector machine model optimized based on the particle swarm algorithm according to the target quality parameter vector.
6. The system of claim 4, wherein the quality detection module further comprises:
An error analysis unit for calculating a quality diagnostic error of the target quality detection model;
dividing the quality prediction sample set into a plurality of sub-sample sets, and carrying out optimization training on the least square vector machine model based on the sub-sample sets by using a particle swarm algorithm to generate a plurality of corresponding quality detection sub-models;
calculating an average quality diagnostic error for the plurality of quality detection sub-models;
the model training unit is further configured to: comparing the quality diagnosis error with the average quality diagnosis error, and when the quality diagnosis error is larger than the average quality diagnosis error, carrying out optimization training on the least square vector machine model again by using a particle swarm algorithm until a target quality detection model with the quality diagnosis error smaller than the average quality diagnosis error is obtained.
7. The system of claim 1, wherein the intelligent traceback module comprises:
the quality intelligent traceability unit is used for identifying defect characteristics and defect types corresponding to unqualified hole making and/or riveting according to defect reasoning rules in the quality characteristic knowledge base;
wherein the defect feature comprises at least one of a shape feature, a location feature, and a surface feature.
8. The system of claim 1, wherein the intelligent traceback module further comprises:
the self-learning unit is used for carrying out online learning by simulating an artificial defect recognition method, generating a defect reasoning rule and storing the defect reasoning rule into the quality characteristic knowledge base.
9. A machine learning-based part assembly quality digital inspection method, characterized in that the machine learning-based part assembly quality digital inspection system according to any one of claims 1 to 8 is employed, the method comprising:
acquiring image data of at least one part;
performing image information analysis on the image data to obtain a plurality of quality feature elements corresponding to the image data;
constructing a target quality detection model based on particle swarm parameter optimization, and carrying out hole making and/or riveting quality detection on the part according to the quality characteristic elements by utilizing the target quality detection model to obtain a quality detection result of the part;
establishing a quality characteristic knowledge base according to defect characteristics, defect types and defect reasoning rules corresponding to unqualified hole making and/or riveting according to a plurality of quality detection results;
and when the quality detection result is unqualified, performing defect tracing on the hole making and/or riveting of the part by utilizing the quality characteristic knowledge base.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110548147.4A CN113487533B (en) | 2021-05-19 | 2021-05-19 | Part assembly quality digital detection system and method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110548147.4A CN113487533B (en) | 2021-05-19 | 2021-05-19 | Part assembly quality digital detection system and method based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113487533A CN113487533A (en) | 2021-10-08 |
| CN113487533B true CN113487533B (en) | 2023-11-14 |
Family
ID=77932907
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110548147.4A Active CN113487533B (en) | 2021-05-19 | 2021-05-19 | Part assembly quality digital detection system and method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113487533B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110096772B (en) * | 2019-04-17 | 2022-10-25 | 大连理工大学 | Method for establishing shape and position deviation feature library for aerospace thin shell structure |
| CN114777651B (en) * | 2022-05-05 | 2024-01-26 | 东北大学 | Digital twinning-based aircraft surface assembly quality detection method |
| CN116611850B (en) * | 2023-07-14 | 2023-10-24 | 浙江春风动力股份有限公司 | System for detecting and tracing engine assembly quality curve |
| CN117236809B (en) * | 2023-11-13 | 2024-02-27 | 宁德市天铭新能源汽车配件有限公司 | Automobile part production management method and system based on artificial intelligence |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020434A (en) * | 2012-11-30 | 2013-04-03 | 南京航空航天大学 | Particle swarm optimization-based least square support vector machine combined predicting method |
| CN107741606A (en) * | 2017-08-31 | 2018-02-27 | 江苏宜润智能装备产业创新研究院有限公司 | A kind of plastics hot riveting assembling monitoring method and device |
| CN108830304A (en) * | 2018-05-29 | 2018-11-16 | 深圳明创自控技术有限公司 | A kind of image identification system based on support vector machines |
| CN110340728A (en) * | 2019-06-13 | 2019-10-18 | 浙江大学 | Quality control system and method towards the processing of tractor manufacturing process hole group |
| CN111222669A (en) * | 2018-11-27 | 2020-06-02 | 国网新疆电力有限公司经济技术研究院 | Overhead line cost prediction method and device |
| CN111681214A (en) * | 2020-05-24 | 2020-09-18 | 哈尔滨理工大学 | Aviation bearing surface rivet detection method based on U-net network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10181185B2 (en) * | 2016-01-11 | 2019-01-15 | Kla-Tencor Corp. | Image based specimen process control |
-
2021
- 2021-05-19 CN CN202110548147.4A patent/CN113487533B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020434A (en) * | 2012-11-30 | 2013-04-03 | 南京航空航天大学 | Particle swarm optimization-based least square support vector machine combined predicting method |
| CN107741606A (en) * | 2017-08-31 | 2018-02-27 | 江苏宜润智能装备产业创新研究院有限公司 | A kind of plastics hot riveting assembling monitoring method and device |
| CN108830304A (en) * | 2018-05-29 | 2018-11-16 | 深圳明创自控技术有限公司 | A kind of image identification system based on support vector machines |
| CN111222669A (en) * | 2018-11-27 | 2020-06-02 | 国网新疆电力有限公司经济技术研究院 | Overhead line cost prediction method and device |
| CN110340728A (en) * | 2019-06-13 | 2019-10-18 | 浙江大学 | Quality control system and method towards the processing of tractor manufacturing process hole group |
| CN111681214A (en) * | 2020-05-24 | 2020-09-18 | 哈尔滨理工大学 | Aviation bearing surface rivet detection method based on U-net network |
Non-Patent Citations (4)
| Title |
|---|
| " Percussion-based Detection of Bolt Looseness Using Speech Recognition Technology and Least Square Support Vector Machine";Furui Wang;《2020 IEEE International Conference on Networking, Sensing and Control (ICNSC)》;全文 * |
| "Application of soft sensor in welding seam tracking prediction based on LSSVM and PSO with compression factor";Jianhui Wang;《2013 25th Chinese Control and Decision Conference (CCDC)》;全文 * |
| "基于PSO-LS-SVM的发动机多工序装配质量预测模型研究";吕旭泽;《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》;正文第3.5，4.4节 * |
| "基于图像识别的飞机蒙皮铆接质量检测技术研究";王海燕;《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》;第5章 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113487533A (en) | 2021-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113487533B (en) | Part assembly quality digital detection system and method based on machine learning | |
| EP3452959B1 (en) | Model construction in a neural network for object detection | |
| EP3364341A1 (en) | Analyzing digital holographic microscopy data for hematology applications | |
| CN108711148B (en) | Tire defect intelligent detection method based on deep learning | |
| CN107622277B (en) | Bayesian classifier-based complex curved surface defect classification method | |
| CN110927171A (en) | Bearing roller chamfer surface defect detection method based on machine vision | |
| CN109284779A (en) | Object detecting method based on the full convolutional network of depth | |
| CN112149758B (en) | Hyperspectral open set classification method based on Euclidean distance and deep learning | |
| CN111400667A (en) | Aviation blade profile detection method and system based on variable tolerance zone constraint | |
| CN111860106B (en) | Unsupervised bridge crack identification method | |
| CN108416373A (en) | A kind of unbalanced data categorizing system based on regularization Fisher threshold value selection strategies | |
| CN114694178A (en) | Method and system for monitoring safety helmet in power operation based on fast-RCNN algorithm | |
| CN115147363A (en) | Image defect detection and classification method and system based on deep learning algorithm | |
| CN110516920B (en) | Gyroscope quality grade evaluation method based on index fusion | |
| CN116822342A (en) | Process identification and performance prediction method for zirconium alloy laser cutting | |
| US20150242676A1 (en) | Method for the Supervised Classification of Cells Included in Microscopy Images | |
| García-Pérez et al. | CNN-based in situ tool wear detection: A study on model training and data augmentation in turning inserts | |
| CN109614512B (en) | Deep learning-based power equipment retrieval method | |
| CN115019294A (en) | Pointer instrument reading identification method and system | |
| CN111402236A (en) | Hot-rolled strip steel surface defect grading method based on image gray value | |
| CN115082713B (en) | Method, system and equipment for extracting target detection frame by introducing space contrast information | |
| US20230268078A1 (en) | Method and system for generating a visual representation | |
| CN115106615A (en) | Welding deviation real-time detection method and system based on intelligent working condition identification | |
| CN114637893A (en) | FMEA database system and processing method thereof | |
| CN114358279A (en) | Image recognition network model pruning method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |