A method for online detection of turbine volute machining defects

Abstract

The present application provides a method for online detection of turbine volute machining defects, which relates to the field of defect detection technology. The method includes: the turbine volute production line determines K machining interest areas through non-rework turbine volute analysis, and constructs K machining defect online detection modules using non-rework machining images; pre-configures an image acquisition device in the production line, connects it to the production steps, monitors the images of semi-finished products in real time and detects defects; and controls the track transmission system based on the detection results. The present application can solve the problem that the prior art lacks real-time defect detection, cannot promptly dispose of defective semi-finished products, resulting in low efficiency and quality of turbine volute machining defect detection, further affecting the overall performance of the product, realizes intelligent and flexible production management, reduces the generation of unqualified products, and improves production efficiency and product quality.

Description

Technical Field

[0001] The present application relates to the technical field of defect detection, and in particular to an online detection method for turbine volute machining defects. Background Art

[0002] The turbine volute is an important component of the turbocharger, and its processing quality directly affects the performance and life of the turbocharger. During the processing of the turbine volute, problems such as dimensional deviation, shape error, and surface defects may occur, which will affect the sealing performance and airflow performance of the turbine volute, and thus affect the performance of the entire turbocharger. Traditional methods for detecting defects in turbine volute processing mainly rely on manual detection and offline detection, which have certain limitations.

[0003] At present, the existing turbine volute processing defect detection technology mainly relies on manual inspection and offline inspection. Manual inspection is easily affected by subjective factors and has low efficiency, while offline inspection requires the turbine volute to be removed from the production line for special inspection, which will prolong the production cycle and increase production costs.

[0004] In summary, the existing technology lacks real-time defect detection and cannot promptly process defective semi-finished products, resulting in low efficiency and quality of turbine volute processing defect detection, further affecting the overall performance of the product. Summary of the invention

[0005] The purpose of this application is to provide an online detection method for turbine volute machining defects, so as to solve the problem that the prior art lacks real-time defect detection and cannot promptly dispose of defective semi-finished products, resulting in low efficiency and quality of turbine volute machining defect detection, further affecting the overall performance of the product.

[0006] In view of the above problems, the present application provides a method for online detection of turbine volute machining defects.

[0007] The present application provides a method for online detection of turbine volute machining defects, wherein the method comprises: performing a finished product call of a conceptual turbine volute on a turbine volute production line to obtain a plurality of turbine volutes without rework; analyzing and determining K volute machining interest regions of K volute production steps in the turbine volute production line, wherein the conceptual turbine volute is transferred to the K volute production steps through a track transmission system during the production process; performing machining backtracking on the plurality of turbine volutes without rework to obtain a plurality of groups of machining images without rework; dividing the plurality of machining images without rework with the K volute machining interest regions as constraints to obtain K groups of images of regions without rework; and determining the K groups of images of regions without rework based on the K groups of images of regions without rework and the plurality of images of regions without rework. K online processing defect detection modules are obtained by image construction, and the K online processing defect detection modules are communicatively connected according to the order relationship of the K volute production steps; K image acquisition devices are pre-configured at the output ends of the K volute production steps, and the K image acquisition devices are mapped and communicatively connected to the input ends of the K online processing defect detection modules; during the production process of the conceptual turbine volute by the turbine volute production line, image acquisition of semi-finished products in the production process is performed based on the K image acquisition devices, and the collected semi-finished product images are analyzed by the K online processing defect detection modules to obtain online detection results; and the cross-production step transmission of the track transmission system is controlled according to the online detection results.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] By calling the finished product of the concept turbine volute on the turbine volute production line, a plurality of turbine volutes without rework are obtained; K volute processing interest areas of the K volute production steps in the turbine volute production line are analyzed and determined, wherein the concept turbine volute is transferred to the K volute production steps through a track transmission system during the production process; processing backtracking is performed on the plurality of turbine volutes without rework to obtain a plurality of groups of processing images without rework; the plurality of groups of processing images without rework are divided based on the K volute processing interest areas as constraints to obtain K groups of images of areas without rework; K processing defect online detection modules are constructed based on the K groups of images of areas without rework and the plurality of groups of processing images without rework, and the K processing defect online detection modules are communicatively connected according to the order relationship of the K volute production steps; and the processing defect online detection modules are pre-processed at the output end of the K volute production steps. K image acquisition devices are configured, and the K image acquisition devices are mapped and communicatively connected to the input ends of the K online processing defect detection modules; during the production process of the conceptual turbine volute by the turbine volute production line, image acquisition of semi-finished products in the production process is performed based on the K image acquisition devices, and the acquired semi-finished product images are analyzed by the K online processing defect detection modules to obtain online detection results; the cross-production step transmission of the track transmission system is controlled according to the online detection results, which effectively solves the problem that the prior art cannot promptly handle defective semi-finished products due to the lack of real-time defect detection, resulting in low efficiency and quality of turbine volute processing defect detection, further affecting the overall performance of the product, realizing intelligent and flexible production management, reducing the generation of unqualified products, and improving production efficiency and product quality.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0012] Figure 1 A schematic flow chart of an online detection method for turbine volute machining defects in this application;

[0013] Figure 2 A schematic diagram of the process of performing edge detection in an online detection method for turbine volute machining defects in the present application. DETAILED DESCRIPTION

[0014] The present application provides an online method for detecting defects in turbine volute processing, thereby solving the problem that the prior art lacks real-time defect detection and cannot promptly dispose of defective semi-finished products, resulting in low efficiency and quality of turbine volute processing defect detection, which further affects the overall performance of the product. The application realizes intelligent and flexible production management, reduces the generation of unqualified products, and improves production efficiency and product quality.

[0015] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them. Embodiment 1

[0016] Please refer to the attached Figure 1 The present application provides a method for online detection of turbine volute machining defects, wherein the method specifically comprises the following steps:

[0017] S1: The concept turbine volute production line is used to produce finished products of the turbine volute, and multiple turbine volutes without rework are obtained.

[0018] Specifically, the turbine volute production line is used to call the finished products of the concept turbine volute. During the production process of the turbine volute, some samples are selected from the final qualified finished products for research and analysis. These finished turbine volutes are not reworked, that is, they have no defects that need to be reworked during the manufacturing process. Such turbine volutes can be regarded as standard or ideal reference samples, and multiple finished turbine volutes without rework are collected from the turbine volute production line. These turbine volutes should be strictly quality inspected and confirmed to be defect-free.

[0019] S2: Analyze and determine K volute processing interest areas of K volute production steps in the turbine volute production line, wherein the conceptual turbine volute flows through the K volute production steps through a track transmission system during the production process.

[0020] Specifically, during the production of turbine volutes, each volute undergoes a series of processing steps, including but not limited to casting, machining, heat treatment, cleaning, inspection, etc. A detailed analysis of the production process of turbine volutes is conducted to understand the specific content of each step and its impact on the performance of the volute. A risk assessment is conducted to determine which steps are most likely to cause defects. These steps are usually candidate steps for machining areas of interest. Analyze historical quality data and rework records to find out which areas in each step most often have defects, and these areas will also be listed as machining areas of interest.

[0021] S3: Performing processing backtracking on the plurality of turbine volutes without rework to obtain a plurality of groups of processing images without rework.

[0022] Specifically, a number of turbine volutes without rework are selected from the production line as samples. These turbine volutes should be finished products that have been confirmed to be defect-free after quality inspection. For each selected turbine volute, its entire processing history from raw materials to finished products is recorded, including the parameters of each processing step, the equipment and tools used, the processing time, etc. In each key processing step, a high-resolution industrial camera is used to capture images of specific areas of the turbine volute. These areas are the processing areas of interest determined by analysis. The collected images are preprocessed, including denoising, contrast enhancement, correction, etc., to improve the quality and analyzability of the images. Features related to the processing quality of the turbine volute, such as geometric dimensions, surface texture, shape, etc., are extracted from the preprocessed images. Multiple groups of images of non-rework processing are obtained, wherein a group of semi-finished products without rework processing includes multiple semi-finished volute images corresponding to multiple volute production steps. These images provide the appearance and characteristics of the turbine volute under normal processing conditions.

[0023] S4: Dividing the plurality of groups of non-rework processing images based on the K volute processing interest regions as constraints to obtain K groups of non-rework region images.

[0024] Specifically, according to the determined K regions of interest for volute processing, the collected multiple groups of non-rework processing images are classified. The image set corresponding to each region will become an independent group. In each processing image, the previously determined region of interest is located. Image processing technology is used to identify and extract specific geometric shapes or feature points. The located region is segmented to separate the region of interest from the background. Data enhancement can be performed on the segmented region images, including rotation, flipping, scaling, adding noise and other operations, thereby increasing the diversity of defect-free images. The segmented region images are annotated to mark which parts of the image are normal and defect-free. These annotations will be used to tell the model which features are normal when training the detection model. The segmented and annotated region images are stored in a database, and each processing region of interest corresponds to a database set. K groups of non-rework region images can be obtained, and each group of images corresponds to a specific region of interest for volute processing. These image sets will be used to train and verify the online detection model for processing defects in each region to ensure that the model can accurately identify and classify defects.

[0025] S5: constructing K processing defect online detection modules based on the K groups of non-rework area images and multiple groups of non-rework processing images, and communicating and connecting the K processing defect online detection modules according to the order relationship of the K volute production steps.

[0026] Specifically, K groups of images of non-reworked areas are analyzed in depth to extract features that can represent the normal processing state, such as edges, shapes, textures, etc. The extracted features are used to train K online detection models for processing defects. These models are based on machine learning algorithms, such as support vector machines, random forests, gradient boosting decision trees, etc., or they can be convolutional neural networks based on deep learning. The trained models are verified by multiple groups of non-reworked processing images to ensure that the models can accurately identify defect-free turbine volutes. The models are optimized based on the verification results to improve their accuracy and robustness. The trained models are integrated into the online detection system to construct K online detection modules for processing defects. Each module is responsible for detecting defects in the corresponding volute processing area of ​​interest. According to the order relationship of the K volute production steps, the K online detection modules for processing defects are connected through a communication network. This ensures that during the production process, the turbine volute passes through each detection module in the correct order, thereby achieving continuous online detection. Ensure data synchronization between each detection module and integration with the production line control system. In this way, once a defect is detected, the system can respond immediately, such as stopping the production line, marking defective products, etc.

[0027] S6: pre-configure K image acquisition devices at the output ends of the K volute production steps, and map and communicate the K image acquisition devices to the input ends of the K processing defect online detection modules.

[0028] Specifically, according to the characteristics and requirements of the turbine volute processing steps, suitable image acquisition devices such as industrial cameras, light sources and lenses are selected. These devices need to be able to adapt to the harsh conditions of the production environment, such as temperature, humidity, vibration, etc. Install an image acquisition device at the output end of each volute production step. Ensure that the position and angle of the camera and light source can cover the previously determined processing area of ​​interest. Adjust the camera parameters, such as exposure time, resolution, frame rate, etc., as well as the brightness and color of the light source, to obtain the best image quality. Connect each image acquisition device to the online detection module for processing defects via Ethernet, serial communication or other suitable communication protocols. Ensure the stability and speed of data transmission. Establish a mapping relationship between the image acquisition device and the corresponding online detection module for processing defects. In addition, perform camera calibration to ensure the geometric and color accuracy of the image. Enable the image acquisition device to automatically trigger image acquisition according to the production process and transmit the image data to the corresponding detection module in real time.

[0029] S7: During the production process of the conceptual turbine volute by the turbine volute production line, image acquisition of semi-finished products in the production process is performed based on the K image acquisition devices, and the acquired semi-finished product images are analyzed by the K processing defect online detection modules to obtain online detection results.

[0030] Specifically, after each key step in the production of the turbine volute, the image acquisition device is automatically triggered to capture images of the semi-finished product. These images contain specific processing areas of the volute, that is, the determined processing areas of interest. The acquired image data is quickly transmitted to the corresponding online processing defect detection module through a preset communication connection. After each online processing defect detection module receives the image, it uses the built-in image processing and analysis algorithm to pre-process and analyze the image. Including steps such as denoising, segmentation, and feature extraction. The analyzed image data is processed by a trained model to identify and classify possible defects. The model determines whether the volute is qualified based on the image features. The online processing defect detection module outputs the detection results, including information such as the location, type, and severity of the defect. This information can be used for further processing, such as automatically marking defective products, triggering alarms, and recording statistical data.

[0031] S8: Controlling the cross-production step transmission of the track transmission system according to the online detection result.

[0032] Specifically, the online inspection module is integrated with the track transmission control system of the production line. According to the production process and quality standards, the control logic is defined. For example, if a minor defect is detected, the product is allowed to continue to flow to the next process, but the defect information is recorded; if a serious defect is detected, the product flow to the next process is immediately stopped and directed to the isolation or rework area. Ensure that there is real-time communication capability between the online inspection module and the track transmission control system. Once a defect is detected, the inspection module immediately sends a signal to the control system. After receiving the defect signal, the control system automatically adjusts the track transmission system according to the preset control logic. This includes stopping the conveyor belt, changing the product flow direction, and starting a warning.

[0033] Furthermore, step S2 of the present application also includes:

[0034] Interactively obtain a first adjacent volute semi-finished product set of a first volute production step in the turbine volute production line, wherein the first volute production step is an initial production step among the K volute production steps; perform image acquisition on the first adjacent volute semi-finished product set to obtain a plurality of first image comparison groups; perform image comparison to locate processing areas based on the plurality of first image comparison groups to obtain a plurality of first processing areas; perform coverage analysis on the plurality of first processing areas to obtain a first volute processing area of ​​interest; and so on, analyze and determine the K volute processing areas of interest for the K volute production steps in the turbine volute production line.

[0035] Specifically, these semi-finished products should be in the state after the completion of this step and before entering the next production step. A group of adjacent volute semi-finished products are obtained from the first volute production step of the turbine volute production line. Including the semi-finished products before and after this step, this group of semi-finished products is imaged to obtain a first image comparison group, and multiple first image comparison groups are obtained using the same method. These images should contain sufficient information to characterize the processing state of the volute in this production step. Use image processing techniques, such as feature matching, template matching, etc., to compare multiple first image comparison groups to locate the processing areas. These processing areas are places where defects may occur during the production process or have a greater impact on product quality. Perform coverage analysis on the multiple first processing areas located. The purpose of coverage analysis is to find areas that appear consistently in multiple semi-finished products, which are key processing areas. Based on the results of coverage analysis, determine the first volute processing area of ​​interest. These areas will be the focus of subsequent online detection of processing defects. For the subsequent volute production steps in the turbine volute production line, the above process is repeated, that is, the adjacent volute semi-finished product set is obtained, image acquisition and comparison are performed, and coverage analysis is performed to determine the volute processing interest area for each step. Through the above steps, K volute processing interest areas for K volute production steps in the turbine volute production line can be analyzed and determined. These areas will be used to build the online detection module for processing defects and guide the configuration and positioning of the image acquisition device.

[0036] Furthermore, the present application also includes:

[0037] The first adjacent volute semi-finished product set includes multiple groups of first adjacent volute semi-finished products, and the multiple groups of first adjacent volute semi-finished products are split to obtain multiple pre-processed volute semi-finished products and multiple processed volute semi-finished products; the multiple pre-processed volute semi-finished products and the multiple processed volute semi-finished products are combined and enumerated to obtain multiple groups of updated adjacent volute semi-finished products; predefine workpiece positioning parameters and image acquisition angle sets; use the workpiece positioning parameters to control the standardized fixture to clamp the first group of updated adjacent volute semi-finished products, use the image acquisition angle set to perform image acquisition of the first group of updated adjacent volute semi-finished products in the clamping state, and obtain a first adjacent volute comparison image sequence, wherein the first group of updated adjacent volute semi-finished products is any group of the multiple groups of updated adjacent volute semi-finished products; obtain a conceptual volute model based on the conceptual turbine volute modeling, and use the conceptual volute model as a constraint to stereoscopically stitch the first adjacent volute comparison image sequence based on the image acquisition order to obtain a first image comparison group; and so on, perform image acquisition on the first adjacent volute semi-finished product set to obtain the multiple first image comparison groups.

[0038] Specifically, multiple groups of first adjacent volute semi-finished products are split into volute semi-finished products before and after processing. In this way, the states of the same volute at different processing stages can be compared. Combination enumeration is performed on multiple volute semi-finished products before processing and multiple volute semi-finished products after processing to obtain multiple groups of updated adjacent volute semi-finished products. In this way, sufficient sample data can be ensured in the subsequent image acquisition and comparison process. In order to ensure the standardization and consistency of image acquisition, predefined workpiece positioning parameters, such as the position and pressure of the fixture and the image acquisition angle set, such as the position and direction of the camera, are predefined. The standardized fixture is controlled to clamp the first group of updated adjacent volute semi-finished products using the predefined workpiece positioning parameters, and the image acquisition angle set is used to perform image acquisition in the clamping state. In this way, the images collected at different time points are consistent and comparable. Through the above steps, a comparison image sequence of the first group of updated adjacent volute semi-finished products is obtained. These images will be used for subsequent image comparison and processing area positioning. The conceptual volute model is obtained according to the conceptual turbine volute modeling, and the model is used as a constraint to stereo stitch the first adjacent volute comparison image sequence based on the image acquisition sequence, thereby obtaining the first image comparison group. Repeat the above process for other volute semi-finished products in the first adjacent volute semi-finished product set to obtain a plurality of first image comparison groups, which will be used to determine the first volute processing region of interest.

[0039] Further, if Figure 2 As shown, this application also includes:

[0040] Grayscale processing is performed on the multiple first image comparison groups and image noise reduction is performed based on Gaussian blur to obtain multiple grayscale image comparison groups; image alignment and overlap are performed on the multiple grayscale image comparison groups, and the difference between overlapping pixels is calculated to generate multiple overlapping differential images; binarization processing is performed on the multiple overlapping differential images, and edge detection is performed based on the processing results to locate the multiple first processing areas.

[0041] Specifically, grayscale processing is performed on multiple first image comparison groups to convert color images into grayscale images. Grayscale processing helps to simplify image data, making it more suitable for subsequent image analysis and processing. Noise reduction is performed on grayscale images based on Gaussian blur. Gaussian blur is to reduce the influence of noise by blurring the image while maintaining the main features of the image. Image alignment and overlap are performed on the grayscale image comparison groups after noise reduction. Image alignment ensures that different images are in the same coordinate system. The difference between overlapping pixels is calculated for the aligned overlapping images. This method can highlight the differences between images, thereby helping to identify the locations where changes have occurred during the processing process. Based on the calculation of the difference between pixels, multiple overlapping differential images are generated. The differential images show the change areas between the images, which are key areas that need to be paid attention to during the processing process. Binarization is performed on multiple overlapping differential images. Binarization converts the image into a black and white image, making the image analysis simpler and clearer. Based on the results of the binarization processing, edge detection is performed. Edge detection helps to identify the contours and structural changes of objects in the image, thereby locating the processing area. Through the above steps, multiple first processing areas can be located. These areas are where defects may occur during the turbine volute machining process or have a greater impact on product quality.

[0042] Furthermore, the present application also includes:

[0043] The multiple groups of non-rework processing images are split and reorganized according to the K volute production steps to obtain K groups of step processing images; the K groups of step processing images are divided based on the K volute processing interest areas as constraints to obtain the K groups of non-rework area images; K detection area positioning units are constructed based on the K groups of step processing images and the K groups of non-rework area images; K groups of defect area images are interactively obtained, and K processing defect recognition units are constructed based on the K groups of defect area images and the K groups of non-rework area images after being marked; the K detection area positioning units and the K processing defect recognition units are cascaded according to the K volute production step mapping to obtain the K processing defect online detection modules.

[0044] Specifically, according to the K volute production steps, multiple groups of non-rework processing images are split and recombined to obtain K groups of step processing images corresponding to each production step. In this way, the image data can be closely matched with the production steps. Using the determined K volute processing interest regions as constraints, the K groups of step processing images are divided to obtain K groups of non-rework area images. These images will be used to train and verify the detection model. Based on the K groups of step processing images and the K groups of non-rework area images, K detection area positioning units are constructed. These units are responsible for locating the processing area of ​​interest in the image. K groups of defect area images are obtained through an interactive process. These images contain actual processing defects and are used to train the defect recognition model. K processing defect recognition units are constructed using the marked K groups of defect area images and K groups of non-rework area images. These units will learn how to distinguish between normal and defective processing areas. According to the mapping relationship of the K volute production steps, K detection area positioning units and K processing defect recognition units are cascaded. This ensures that defects are detected in the processing area of ​​interest in the correct production step. Through the above steps, K processing defect online detection modules are obtained. Each module is dedicated to a specific processing step and processing area of ​​interest, and can accurately identify and classify defects.

[0045] Furthermore, step S6 of the present application also includes:

[0046] Perform response tests on the K online processing defect detection modules to obtain K response delays; locate K waste diversion nodes of the K volute production steps on the track transmission system; reversely locate K image acquisition nodes at the output end of the K volute production steps according to the standard transmission speed of the track transmission system, the K response delays and the K waste diversion nodes, and configure the K image acquisition devices at the K image acquisition nodes.

[0047] Specifically, a response test is performed on K processing defect online detection modules to measure the time from receiving the image to outputting the detection results for each module. These times are called response delays. K scrap diversion nodes for K volute production steps are located on the track transmission system. These nodes are specific locations on the production line for isolating and processing defective products. K image acquisition nodes are reversely located based on the standard transmission speed of the track transmission system, K response delays, and the locations of K scrap diversion nodes. These nodes should be located at appropriate locations on the production line to ensure that there is enough time for image acquisition and processing when defects are detected. Corresponding K image acquisition devices are configured on the located K image acquisition nodes. These devices can capture images of each production step in real time and ensure that the image quality meets the requirements of online detection. The configured image acquisition devices are tested to ensure that they can work properly and communicate with the online detection modules. Make necessary adjustments and optimizations to the system based on the test results.

[0048] Furthermore, step S7 of the present application also includes:

[0049] During the production process of the conceptual turbine volute on the turbine volute production line, after the semi-finished turbine volute is processed and output through the first volute production step, the first image acquisition device is activated to perform image acquisition on the semi-finished turbine volute to obtain a first real-time semi-finished image, wherein the first image acquisition device is an image acquisition device arranged at the output end of the first volute production step among the K image acquisition devices; the first real-time semi-finished image is input into the first online processing defect detection module to perform online defect detection to obtain a first online detection result; if the first online detection result is unqualified, the track transmission system clamps and controls the semi-finished turbine volute to enter the waste branch.

[0050] Specifically, after the first volute production step is completed, a semi-finished turbine volute is produced. The first image acquisition device arranged at the output end of the first volute production step is activated to perform real-time image acquisition on the semi-finished turbine volute. The first real-time semi-finished image of the semi-finished turbine volute is obtained by the image acquisition device. The first real-time semi-finished image is input into the first online processing defect detection module. The first online processing defect detection module performs online defect detection on the input image to identify possible defects in the semi-finished turbine volute. The detection module outputs the first online detection result, which will indicate whether the semi-finished turbine volute is qualified. If the first online detection result is unqualified, the track transmission system will clamp and control the semi-finished turbine volute, separate it from the qualified product flow, and guide it to the waste branch. If the first online detection result is qualified, the semi-finished turbine volute will continue to flow to the next production step.

[0051] Furthermore, the present application also includes:

[0052] If the first online inspection result is qualified, the track transmission system clamps and controls the semi-finished turbine volute to enter the second volute production step; after the semi-finished turbine volute is processed and produced by the second volute production step, the first processing defect online detection module communicates and activates the second image acquisition device and the second processing defect online detection module; the second image acquisition device performs image acquisition on the semi-finished turbine volute to obtain a second real-time semi-finished product image; the second processing defect online detection module performs defect online detection on the second real-time semi-finished product image to obtain a second online inspection result; if the second online inspection result is qualified, the track transmission system clamps and controls the semi-finished turbine volute to enter the third volute production step, wherein the first volute production step, the second volute production step and the third volute production step are adjacent in order among the K volute production steps; and so on, based on the K image acquisition devices, image acquisition of semi-finished products in the production process is performed, and the collected semi-finished product images are analyzed by the K processing defect online detection modules, and online inspection of the entire production process is performed.

[0053] Specifically, if the first online inspection result is qualified, the track transmission system clamps and controls the semi-finished turbine volute and sends it to the second volute production step. The first processing defect online detection module activates the second image acquisition device and the second processing defect online detection module through communication. The second image acquisition device performs image acquisition on the semi-finished turbine volute to obtain a second real-time semi-finished product image. The second processing defect online detection module performs online defect detection on the second real-time semi-finished product image to obtain a second online detection result. If the second online detection result is qualified, the track transmission system clamps and controls the semi-finished turbine volute and sends it to the third volute production step. The first volute production step, the second volute production step and the third volute production step are adjacent in order among the K volute production steps. Similarly, based on K image acquisition devices, image acquisition of semi-finished products in the production process is performed, and the collected semi-finished product images are analyzed by K processing defect online detection modules to perform online detection of the entire production process.

[0054] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0055] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A method for online detection of turbine volute machining defects, characterized in that: The method comprises: The turbine volute production line was used to produce finished products of the concept turbine volute, and multiple turbine volutes without rework were obtained; Analyze and determine K volute processing interest areas of K volute production steps in the turbine volute production line, wherein the conceptual turbine volute is transferred to the K volute production steps through a track transmission system during the production process; Performing machining backtracking on the plurality of turbine volutes without rework to obtain a plurality of sets of machining images without rework; Dividing the plurality of groups of non-rework processing images based on the K volute processing interest regions as constraints to obtain K groups of non-rework region images; Constructing K processing defect online detection modules based on the K groups of non-rework area images and multiple groups of non-rework processing images, and communicating and connecting the K processing defect online detection modules according to the order relationship of the K volute production steps; Pre-configure K image acquisition devices at the output ends of the K volute production steps, and map and communicate the K image acquisition devices to the input ends of the K processing defect online detection modules; During the production of the conceptual turbine volute by the turbine volute production line, the image of the semi-finished product in the production process is acquired based on the K image acquisition devices, and the acquired semi-finished product images are analyzed by the K processing defect online detection modules to obtain online detection results; Controlling the cross-production step transmission of the track transmission system according to the online detection result; Wherein, K volute processing interest areas of K volute production steps in the turbine volute production line are analyzed and determined, and the method further comprises: Interactively obtain a first adjacent volute semi-finished product set of a first volute production step in the turbine volute production line, wherein the first volute production step is an initial production step among the K volute production steps; Performing image acquisition on the first adjacent volute semi-finished product set to obtain a plurality of first image comparison groups; Performing image comparison to locate the processing area based on the multiple first image comparison groups to obtain multiple first processing areas; Performing coverage analysis on the plurality of first processing regions to obtain a first volute processing region of interest; By analogy, the K volute processing interest areas of the K volute production steps in the turbine volute production line are analyzed and determined.

2. The method for online detection of turbine volute machining defects according to claim 1, characterized in that: Performing image acquisition on the first adjacent volute semi-finished product set to obtain a plurality of first image comparison groups, the method further comprising: The first adjacent volute semi-finished product set includes a plurality of groups of first adjacent volute semi-finished products, and the plurality of groups of first adjacent volute semi-finished products are split to obtain a plurality of pre-processed volute semi-finished products and a plurality of processed volute semi-finished products; Performing combined enumeration on the plurality of pre-processed volute semi-finished products and the plurality of post-processed volute semi-finished products to obtain a plurality of groups of updated adjacent volute semi-finished products; Predefined workpiece positioning parameters and image acquisition angle sets; The workpiece positioning parameters are used to control the standardized fixture to clamp the first group of updated adjacent volute semi-finished products, and the image acquisition angle set is used to perform image acquisition of the first group of updated adjacent volute semi-finished products in the clamping state to obtain a first adjacent volute comparison image sequence, wherein the first group of updated adjacent volute semi-finished products is any group of the multiple groups of updated adjacent volute semi-finished products; A conceptual volute model is obtained according to the conceptual turbine volute modeling, and the first adjacent volute comparison image sequence is stereoscopically stitched based on the image acquisition order to obtain a first image comparison group using the conceptual volute model as a constraint; By analogy, image acquisition is performed on the first adjacent volute semi-finished product set to obtain the plurality of first image comparison groups.

3. The method for online detection of turbine volute machining defects according to claim 1, characterized in that: Based on the plurality of first image comparison groups, image comparison is performed to locate the processing area to obtain a plurality of first processing areas, and the method further includes: Performing grayscale processing on the plurality of first image comparison groups and performing image noise reduction based on Gaussian blur to obtain a plurality of grayscale image comparison groups; Performing image alignment and overlapping on the plurality of grayscale image comparison groups, and calculating the difference between overlapping pixels to generate a plurality of overlapping difference images; Binarization processing is performed on the plurality of overlapping difference images, and edge detection is performed based on the processing results to locate the plurality of first processing areas.

4. The method for online detection of turbine volute machining defects according to claim 1, characterized in that: The method further comprises: According to the K volute production steps, the multiple groups of non-rework processing images are split and reorganized to obtain K groups of step processing images; Dividing the K groups of step processing images based on the K volute processing interest areas as constraints to obtain the K groups of non-rework area images; Constructing K detection area positioning units based on the K groups of step processing images and the K groups of non-rework area images; Interactively obtain K groups of defective area images, and construct K processing defect recognition units based on the K groups of defective area images and K groups of non-rework area images after adopting the marking; The K detection area positioning units and the K processing defect identification units are cascaded according to the K volute production step mappings to obtain the K processing defect online detection modules.

5. The method for online detection of turbine volute machining defects according to claim 1, characterized in that: Pre-configuring K image acquisition devices at the output ends of the K volute production steps, and mapping and communicating the K image acquisition devices to the input ends of the K processing defect online detection modules, the method further includes: Performing a response test on the K processing defect online detection modules to obtain K response delays; Positioning K waste diversion nodes of the K volute production steps on the track transmission system; According to the standard transmission speed of the track transmission system, the K response delays and the K waste diversion nodes, the K image acquisition nodes are reversely located at the output ends of the K volute production steps, and the K image acquisition devices are configured at the K image acquisition nodes.

6. The method for online detection of turbine volute machining defects according to claim 1, characterized in that: During the production of the conceptual turbine volute by the turbine volute production line, the image of the semi-finished product in the production process is acquired based on the K image acquisition devices, and the acquired semi-finished product images are analyzed by the K processing defect online detection modules to obtain online detection results. The method further includes: During the production process of the conceptual turbine volute on the turbine volute production line, after the semi-finished turbine volute is produced through the first volute production step, the first image acquisition device is activated to perform image acquisition on the semi-finished turbine volute to obtain a first real-time semi-finished image, wherein the first image acquisition device is an image acquisition device arranged at the output end of the first volute production step among the K image acquisition devices; Inputting the first real-time semi-finished product image into a first processing defect online detection module to perform online defect detection to obtain a first online detection result; If the first online detection result is unqualified, the track transmission system clamps and controls the semi-finished turbine volute to enter the waste branch.

7. The method for online detection of turbine volute machining defects according to claim 6, characterized in that: The method further comprises: If the first online detection result is qualified, the track transmission system clamps and controls the semi-finished turbine volute to enter the second volute production step; After the semi-finished turbine volute is produced by the second volute production step, the first processing defect online detection module communicates and activates the second image acquisition device and the second processing defect online detection module; Performing image acquisition on the semi-finished turbine volute via the second image acquisition device to obtain a second real-time semi-finished product image; Performing online defect detection on the second real-time semi-finished product image via the second processing defect online detection module to obtain a second online detection result; If the second online inspection result is qualified, the track transmission system clamps and controls the semi-finished turbine volute to enter the third volute production step, wherein the first volute production step, the second volute production step and the third volute production step are adjacent in order among the K volute production steps; and so on, based on the K image acquisition devices, image acquisition of semi-finished products in the production process is performed, and the acquired semi-finished product images are analyzed by the K processing defect online detection modules to perform online inspection of the entire production process.

Images