Aluminum profile surface defect analysis method based on machine vision

Abstract

The invention relates to the technical field of surface defect detection, and discloses an aluminum profile surface defect analysis method based on machine vision, and the method comprises the steps: obtaining equipment operation parameters, environment parameters, aluminum bar processing procedure parameters and aluminum profile surface snapshots of processing equipment of a target aluminum profile in real time; obtaining a corresponding equipment operation risk value according to the equipment operation parameter; according to the aluminum bar machining procedure parameters, the equipment operation risk values and the environment parameters, procedure score values of the corresponding aluminum bar machining key procedures are obtained; the aluminum profile surface snapshots, obtained from multiple angles, of the target aluminum profile are combined and then sent to the pre-trained surface recognition model for evaluation, the quality inspection defect value output by the surface recognition model is obtained, and the surface quality defects can be continuously, accurately and rapidly found in a non-contact detection mode. The aluminum profile production efficiency and quality are fully improved, and compared with manual detection, a machine vision system has lower cost and higher efficiency.

Description

Technical Field

[0001] The present application relates to the technical field of surface defect detection, and in particular to a method for analyzing surface defects of aluminum profiles based on machine vision. Background Art

[0002] Aluminum profile processing is the process of processing aluminum ingots or aluminum bars into aluminum materials with specific cross-sectional shapes and sizes through a series of process steps. It has excellent mechanical properties and is widely used in industry and life. The aluminum profile processing production line mainly includes the aluminum bar heating part and the aluminum bar extrusion forming part. After the processing production line is repaired or used to a certain extent, it is necessary to evaluate its execution risk to ensure that the processed aluminum profile can meet the requirements.

[0003] The aluminum profile processing process includes raw material preparation, melting, extrusion, surface treatment and other links. Raw material preparation Select high-quality aluminum ingots or aluminum bars as raw materials. These raw materials usually use industrial-grade aluminum with high purity because of its good ductility, plasticity and corrosion resistance. In the melting and casting process, aluminum ingots, master alloys and other raw materials are first prepared in corresponding proportions, and the aluminum ingots are put into the smelting furnace. Natural gas is used to burn and melt the aluminum ingots, and the composition and temperature are uniformly stirred. The temperature of the aluminum liquid is raised to prepare for adjusting the composition and refining. The aluminum liquid composition ratio is adjusted according to the alloy requirements. The aluminum liquid is deslagging and degassing to improve the quality of the aluminum liquid. Continuous casting is carried out to produce aluminum bars of various specifications. The extrusion process is more critical. The aluminum bar needs to be heated to a certain temperature within a certain period of time, and the heated aluminum bar is extruded into a profile of the required cross-sectional shape and size through a die using an extruder. During the extrusion process, factors such as die design, extrusion temperature, extrusion speed and lubrication conditions will affect the molding quality and production efficiency of the profile.

[0004] Types of surface defects produced during the aluminum profile processing may include:

[0005] Rough texture: Surface roughness does not meet the requirements.

[0006] Color separation: uneven surface color or color difference.

[0007] Dark bands and light bands: stripes or band-like areas of varying light and dark appear on the surface.

[0008] Scratches: Scratches or line-like defects appear on the surface.

[0009] Bone protrusion (concave bone, convex bone): a bone-like defect with an uneven surface.

[0010] Poor finish: The overall surface smoothness does not meet the standards.

[0011] Slag inclusion: foreign matter or impurities are included on the surface.

[0012] Bending, twisting, deformation: the shape of the profile does not meet the requirements.

[0013] Bubbles and peeling: Bubbles or peeling appear on the surface.

[0014] Pitting or roughness: small pits or roughness appear on the surface.

[0015] In view of the surface defects that may occur during the processing of the above-mentioned aluminum profiles, most of the existing technologies currently use the touch method, pencil scratch method, paint spray test method and alkali washing test method. Although the above-mentioned detection methods have a certain degree of accuracy, there is a large lag in the detection results, and most of them are contact detection methods, and quality problems cannot be discovered in time through non-contact detection methods. Summary of the invention

[0016] In order to promptly discover aluminum profile surface defects through a non-contact detection method, the present application provides an aluminum profile surface defect analysis method based on machine vision.

[0017] This application provides a method for analyzing aluminum profile surface defects based on machine vision, which adopts the following technical solutions:

[0018] A method for analyzing surface defects of aluminum profiles based on machine vision, comprising:

[0019] Real-time acquisition of equipment operating parameters, environmental parameters, aluminum bar processing parameters, and aluminum profile surface snapshots of the target aluminum profile processing equipment;

[0020] Assessing the operation risk of the processing equipment according to the equipment operation parameters to obtain a corresponding equipment operation risk value;

[0021] Evaluate the key process of aluminum bar processing according to the aluminum bar processing process parameters, the equipment operation risk value, and the environmental parameters to obtain a process score value corresponding to the key process of aluminum bar processing;

[0022] Merging the surface snapshots of the target aluminum profile obtained from multiple angles and sending them to a pre-trained surface recognition model for evaluation, and obtaining the number of surface abnormal points and the corresponding confidence level and the corresponding quality inspection defect value output by the surface recognition model;

[0023] The surface defect evaluation value of the target aluminum profile is obtained by combining the equipment operation risk value, the process score value and the quality inspection defect value, and an early warning is issued when the surface defect evaluation value exceeds a specified threshold;

[0024] The surface recognition model is obtained through training based on YOLO.

[0025] By adopting the above technical solution, by monitoring various operating status data and operating environment data of important equipment in the aluminum profile production process, such as extrusion and stretching machines, and then cooperating with image capture devices, such as CMOS and CCD, to quickly obtain snapshots of the aluminum profile surface and send them to the surface recognition model designed based on the YOLO algorithm for rapid analysis and processing, it is possible to continuously and accurately detect surface quality defects through non-contact detection methods, fully improve the production efficiency and quality of aluminum profiles, and compared with manual inspection, the machine vision system has lower costs and higher efficiency.

[0026] Optionally, the method for obtaining the surface defect evaluation value includes:

[0027] C(t)=θ1*S(t) lnDp(t) +θ2*Cp(t)

[0028] Among them, C(t) is the surface defect assessment value at time t, θ1 and θ2 are the corresponding weight coefficients, Dp(t) is the equipment operation risk value at time t, S(t) is the process score value at time t, and Cp(t) is the quality inspection defect value at time t.

[0029] Optionally, the equipment operation parameters include the data of aluminum rod heating temperature changing over time, the data of vibration amplitude changing over time at key bearing positions, the data of vibration noise signal changing over time at key bearing positions, and the data of health risk value changing over time at lubricating oil samples:

[0030] The method for obtaining the equipment operation risk value Dp(t) includes:

[0031]

[0032] Among them, Dp(t) is the equipment operation risk value at time t, α 1 , α 2 are the preset weighting coefficients of the corresponding items, Δt is the specified time length, TA(t) is the aluminum bar heating temperature at time t, TAmp(t) is the ideal process heating temperature at time t, Ps(t) represents the vibration amplitude of the key bearing position at the speed s at time t, Psth represents the ideal vibration amplitude of the key bearing position at the speed s, Jn Δt is the number of abnormal pulses of the vibration noise signal at the key bearing position in the time interval (t-Δt, t), J Δt is the allowable number of abnormal pulses of the vibration noise signal at the key bearing position in the time interval (t-Δt, t).

[0033] Optionally, the method for obtaining the health risk value H(t) of the lubricating oil sample includes:

[0034]

[0035] Wherein, Xi(t) is the parameter value of the i-th lubricating oil monitoring item at time t, Xih(t) is the ideal value of the i-th lubricating oil monitoring item at time t, is the parameter average value of the i-th lubricating oil monitoring item in the time interval (t-Δt, t), and iS is the number of abnormal monitoring items.

[0036] Optionally, the method for obtaining the process score value S(t) includes:

[0037] Obtain the score value SE(t) of the aluminum bar heating process and the score value SL(t) of the aluminum bar stretching process respectively;

[0038]

[0039] Among them, β1 and β2 are the corresponding weight coefficients, and Dpst is the preset risk threshold.

[0040] Optionally, the method for obtaining the score value SE(t) of the aluminum rod heating process includes:

[0041]

[0042] Among them, λ1 and λ2 are the corresponding weight coefficients respectively, Temp(t) represents the temperature value of the heating part at time t under no-load state, Temp0(t) represents the ideal temperature value of the heating part at time t under no-load state, t1 is the no-load heating time, t2 is the trial production heating time, and Tev(t) is the ambient temperature value.

[0043] Optionally, the method for obtaining the score value SL(t) of the aluminum bar stretching process comprises:

[0044]

[0045] Among them, η1 and η2 are the corresponding weight coefficients respectively, Vemp(t) represents the pushing speed of the aluminum bar at time t in the no-load state, Vemp0(t) represents the ideal pushing speed of the aluminum bar at time t in the no-load state, VT(t) is the pushing speed at time t in the trial production process, VT0(t) is the standard pushing speed at time t in the trial production process, tA is the end time of no-load pushing, t2 is the end time of trial pushing, ΔV1 and ΔV2 are the corresponding preset comparison thresholds respectively.

[0046] Optionally, the aluminum profile surface snapshots of the target aluminum profile acquired from multiple angles are merged and sent to a pre-trained surface recognition model for evaluation, and the process of obtaining the number of surface abnormal points and the corresponding confidence level and the corresponding quality inspection defect value output by the surface recognition model includes:

[0047] SA1, take a snapshot of the target aluminum profile, obtain m corresponding aluminum profile surface snapshots Am, merge them into blank pictures in chronological order to obtain the recognition sampling picture A, and then send the surface recognition model;

[0048] SA2, rotating the target aluminum profile by a specified angle, obtaining z corresponding aluminum profile surface snapshots Bz during the rotation process, merging them into a blank picture in chronological order to obtain a recognition sampling picture B, and then sending the surface recognition model;

[0049] SA3, restore the position of the target aluminum profile, obtain the corresponding y aluminum profile surface snapshots Cy during the restoration process, merge them into a blank picture in chronological order to obtain the recognition sampling map C, and then send the surface recognition model;

[0050] SA4, respectively obtain the quality inspection defect value Cp(A) corresponding to the identification sampling diagram A, the quality inspection defect value Cp(B) corresponding to the identification sampling diagram B, and the quality inspection defect value Cp(C) corresponding to the identification sampling diagram C;

[0051] SA5, if Then The result is taken as the quality inspection defect value Cp(t);

[0052] like or Then The result is taken as the quality inspection defect value Cp(t);

[0053] like Then return to step SA1.

[0054] Optionally, a method for obtaining a quality inspection defect value Cp(X) corresponding to the identification sampling graph X includes:

[0055]

[0056] Among them, k is the number of surface anomaly points, pj is the confidence of the jth surface anomaly point, and Spj is the area value of the jth surface anomaly point.

[0057] In summary, the present application includes at least one of the following beneficial technical effects:

[0058] 1. The present invention monitors various operating status data and operating environment data of important equipment in the aluminum profile production process, such as extrusion and stretching machines, and then cooperates with image capture devices, such as CMOS and CCD, to quickly obtain snapshots of the aluminum profile surface and send them to a surface recognition model designed based on the YOLO algorithm for rapid analysis and processing. It can continuously and accurately detect surface quality defects through non-contact detection methods, fully improving the production efficiency and quality of aluminum profiles. Compared with manual inspection, the machine vision system has lower cost and higher efficiency.

[0059] 2. By quantifying the potential risks in processing equipment and its processing procedures, the output results of the surface recognition model are supplemented accordingly, so that related problems can be discovered earlier and losses can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 It is a schematic diagram of the aluminum profile surface defect analysis method based on machine vision in the present invention. DETAILED DESCRIPTION

[0061] Embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings.

[0062] In the description of this specification, the description with reference to the terms "certain embodiments", "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0063] The present application embodiment discloses a method for analyzing surface defects of aluminum profiles based on machine vision, comprising:

[0064] Real-time acquisition of equipment operating parameters, environmental parameters, aluminum bar processing parameters, and aluminum profile surface snapshots of the target aluminum profile processing equipment;

[0065] Assessing the operation risk of the processing equipment according to the equipment operation parameters to obtain a corresponding equipment operation risk value;

[0066] Evaluate the key process of aluminum bar processing according to the aluminum bar processing process parameters, the equipment operation risk value, and the environmental parameters to obtain a process score value corresponding to the key process of aluminum bar processing;

[0067] Merging the surface snapshots of the target aluminum profile obtained from multiple angles and sending them to a pre-trained surface recognition model for evaluation, and obtaining the number of surface abnormal points and the corresponding confidence level and the corresponding quality inspection defect value output by the surface recognition model;

[0068] The surface defect evaluation value of the target aluminum profile is obtained by combining the equipment operation risk value, the process score value and the quality inspection defect value, and an early warning is issued when the surface defect evaluation value exceeds a specified threshold;

[0069] The surface recognition model is obtained through training based on YOLO.

[0070] By adopting the above technical scheme, the present invention monitors various operating status data and operating environment data of important equipment in the aluminum profile production process, such as extrusion and stretching machines, and then cooperates with image capture devices, such as CMOS and CCD, to quickly obtain snapshots of the aluminum profile surface and send them to a surface recognition model designed based on the YOLO algorithm for rapid analysis and processing. It can continuously and accurately detect surface quality defects through non-contact detection methods, fully improve the production efficiency and quality of aluminum profiles, and compared with manual inspection, the machine vision system has lower cost and higher efficiency.

[0071] Optionally, the method for obtaining the surface defect evaluation value includes:

[0072] C(t)=θ1*S(t) lnDp(t) +θ2*Cp(t)

[0073] Among them, C(t) is the surface defect assessment value at time t, θ1 and θ2 are the corresponding weight coefficients, Dp(t) is the equipment operation risk value at time t, S(t) is the process score value at time t, and Cp(t) is the quality inspection defect value at time t.

[0074] In this embodiment of the present invention, the potential risks existing in the processing equipment and its processing procedures can be quantified, and the output results of the surface recognition model can be supplemented accordingly, so that related problems can be discovered earlier and losses can be reduced.

[0075] Optionally, the equipment operation parameters include the data of aluminum rod heating temperature changing over time, the data of vibration amplitude changing over time at key bearing positions, the data of vibration noise signal changing over time at key bearing positions, and the data of health risk value changing over time at lubricating oil samples:

[0076] The method for obtaining the equipment operation risk value Dp(t) includes:

[0077]

[0078] Among them, Dp(t) is the equipment operation risk value at time t, α 1 , α 2 are the preset weighting coefficients of the corresponding items, Δt is the specified time length, TA(t) is the aluminum bar heating temperature at time t, TAmp(t) is the ideal process heating temperature at time t, Ps(t) represents the vibration amplitude of the key bearing position at the speed s at time t, Psth represents the ideal vibration amplitude of the key bearing position at the speed s, Jn Δt is the number of abnormal pulses of the vibration noise signal at the key bearing position in the time interval (t-Δt, t), J Δt is the allowable number of abnormal pulses of the vibration noise signal at the key bearing position in the time interval (t-Δt, t).

[0079] In this embodiment, the working state of the stretch extruder can be evaluated. It is a key equipment in industrial production, and its operating state directly affects production efficiency and product quality. Since the extruder has a complex structure, difficult process, and is often in a high-speed, heavy-load working environment, it is very important to monitor its operating state effectively. Through monitoring, potential faults can be discovered in time, and major equipment accidents can be prevented, thereby ensuring the continuity and stability of production.

[0080] Specifically, the equipment status can be diagnosed by collecting vibration signals and noise signals of mechanical equipment. However, in the early wear stage of such equipment, the changes in vibration signals may be minimal and difficult to capture accurately. However, the vibration and noise signals at key bearing positions based on the shock pulse method (SPM) can accurately determine the operating status of rolling bearings, such as fatigue failure, wear failure, poor lubrication, etc., with a high diagnosis rate.

[0081] Optionally, the method for obtaining the health risk value H(t) of the lubricating oil sample includes:

[0082]

[0083] Wherein, Xi(t) is the parameter value of the i-th lubricating oil monitoring item at time t, Xih(t) is the ideal value of the i-th lubricating oil monitoring item at time t, is the parameter average value of the i-th lubricating oil monitoring item in the time interval (t-Δt, t), and iS is the number of abnormal monitoring items.

[0084] By adopting the above technical solution, in this embodiment of the present invention, data of lubricating oil samples can be collected in real time, including multiple parameters such as temperature, viscosity, moisture, metal particles, etc., to comprehensively reflect the health status of the lubricating oil and the operating status of the equipment. Because the reaction indicators are relatively subtle, they are more sensitive and effective in capturing early wear changes inside processing equipment, and can often provide early warning before the vibration monitoring sensor detects equipment wear.

[0085] Optionally, the method for obtaining the process score value S(t) includes:

[0086] Obtain the score value SE(t) of the aluminum bar heating process and the score value SL(t) of the aluminum bar stretching process respectively;

[0087]

[0088] Among them, β1 and β2 are the corresponding weight coefficients, and Dpst is the preset risk threshold.

[0089] By adopting the above technical solution, The ratio is incorporated into the quantitative formula of the process score of aluminum rod processing, which can amplify the potential risks of processing equipment and reflect them in the process, so that problems can be discovered more quickly.

[0090] Optionally, the method for obtaining the score value SE(t) of the aluminum rod heating process includes:

[0091]

[0092] Among them, λ1 and λ2 are the corresponding weight coefficients respectively, Temp(t) represents the temperature value of the heating part at time t under no-load state, Temp0(t) represents the ideal temperature value of the heating part at time t under no-load state, t1 is the no-load heating time, t2 is the trial production heating time, and Tev(t) is the ambient temperature value.

[0093] The processing temperature rise control of the aluminum rod needs to be relatively precise. In this embodiment, the heating temperature curve change of the aluminum rod and the temperature curve change of the equipment components responsible for heating are comprehensively considered. If the temperature is low, it will affect the subsequent extrusion molding process. If the temperature process is high, it will cause defects in the molded product. If there is a problem with the equipment components responsible for heating, it will inevitably affect the actual temperature of the final aluminum rod. The larger SE(t) is, the greater the probability that there will be a problem with the equipment components responsible for heating.

[0094] Optionally, the method for obtaining the score value SL(t) of the aluminum bar stretching process comprises:

[0095]

[0096] Among them, η1 and η2 are the corresponding weight coefficients respectively, Vemp(t) represents the pushing speed of the aluminum bar at time t in the no-load state, Vemp0(t) represents the ideal pushing speed of the aluminum bar at time t in the no-load state, VT(t) is the pushing speed at time t in the trial production process, VT0(t) is the standard pushing speed at time t in the trial production process, tA is the end time of no-load pushing, t2 is the end time of trial pushing, ΔV1 and ΔV2 are the corresponding preset comparison thresholds respectively.

[0097] By adopting the above technical solution, the no-load process is combined with the trial production process to determine whether there are potential failure problems in the push component, thereby ensuring the continuous operation of the production line after it is officially put into production.

[0098] Optionally, the aluminum profile surface snapshots of the target aluminum profile acquired from multiple angles are merged and sent to a pre-trained surface recognition model for evaluation, and the process of obtaining the number of surface abnormal points and the corresponding confidence level and the corresponding quality inspection defect value output by the surface recognition model includes:

[0099] SA1, take a snapshot of the target aluminum profile, obtain m corresponding aluminum profile surface snapshots Am, merge them into blank pictures in chronological order to obtain the recognition sampling picture A, and then send the surface recognition model;

[0100] SA2, rotating the target aluminum profile by a specified angle, obtaining z corresponding aluminum profile surface snapshots Bz during the rotation process, merging them into a blank picture in chronological order to obtain a recognition sampling picture B, and then sending the surface recognition model;

[0101] SA3, restore the position of the target aluminum profile, obtain the corresponding y aluminum profile surface snapshots Cy during the restoration process, merge them into a blank picture in chronological order to obtain the recognition sampling map C, and then send the surface recognition model;

[0102] SA4, respectively obtain the quality inspection defect value Cp(A) corresponding to the identification sampling diagram A, the quality inspection defect value Cp(B) corresponding to the identification sampling diagram B, and the quality inspection defect value Cp(C) corresponding to the identification sampling diagram C;

[0103] SA5, if Then The result is taken as the quality inspection defect value Cp(t);

[0104] like or Then The result is taken as the quality inspection defect value Cp(t);

[0105] like Then return to step SA1.

[0106] In this embodiment of the present invention, it is designed in this way, taking into account that the machine vision system converts the captured target into an image signal through a machine vision product, transmits it to a dedicated image processing system, obtains the morphological information of the captured target, and converts it into a digital signal based on pixel distribution, brightness, color and other information. The image system performs various operations on these signals to extract the characteristics of the target. Some surface defects may not be caused by the process, such as impurities and foreign matter, or digital noise caused by light influence. Therefore, when sampling, the target aluminum profile can be shaken or vibrated to try to eliminate impurities and foreign matter, or the shooting angle can be changed during the rotation process to change the lighting environment. In order to achieve reliable recognition, the surface recognition model also needs to obtain samples in the same way during training, so the sample size and disturbance amount can be increased, thereby ultimately improving the accuracy of defect recognition.

[0107] Optionally, a method for obtaining a quality inspection defect value Cp(X) corresponding to the identification sampling graph X includes:

[0108]

[0109] Among them, k is the number of surface anomaly points, pj is the confidence of the jth surface anomaly point, and Spj is the area value of the jth surface anomaly point.

[0110] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for analyzing aluminum profile surface defects based on machine vision, characterized in that: include: Real-time acquisition of equipment operating parameters, environmental parameters, aluminum bar processing parameters, and aluminum profile surface snapshots of the target aluminum profile processing equipment; Assessing the operation risk of the processing equipment according to the equipment operation parameters to obtain a corresponding equipment operation risk value; Evaluate the key process of aluminum bar processing according to the aluminum bar processing process parameters, the equipment operation risk value, and the environmental parameters to obtain a process score value corresponding to the key process of aluminum bar processing; Merging the surface snapshots of the target aluminum profile obtained from multiple angles and sending them to a pre-trained surface recognition model for evaluation, and obtaining the number of surface abnormal points and the corresponding confidence level and the corresponding quality inspection defect value output by the surface recognition model; The surface defect evaluation value of the target aluminum profile is obtained by combining the equipment operation risk value, the process score value and the quality inspection defect value, and an early warning is issued when the surface defect evaluation value exceeds a specified threshold; The surface recognition model is obtained through training based on YOLO.

2. The method for analyzing aluminum profile surface defects based on machine vision according to claim 1, characterized in that: The method for obtaining the surface defect evaluation value comprises: C(t)=θ1*S(t) lnDp(t) +θ2*Cp(t) Among them, C(t) is the surface defect assessment value at time t, θ1 and θ2 are the corresponding weight coefficients, Dp(t) is the equipment operation risk value at time t, S(t) is the process score value at time t, and Cp(t) is the quality inspection defect value at time t.

3. The method for analyzing aluminum profile surface defects based on machine vision according to claim 2, characterized in that: The equipment operation parameters include the data of aluminum rod heating temperature changing over time, the data of vibration amplitude changing over time at key bearing positions, the data of vibration noise signal changing over time at key bearing positions, and the data of health risk value changing over time at lubricating oil samples: The method for obtaining the equipment operation risk value Dp(t) includes: Where Dp(t) is the equipment operation risk value at time t, α1 and α2 are the preset weighting coefficients of the corresponding items, Δt is the specified time length, TA(t) is the aluminum bar heating temperature at time t, TAmp(t) is the ideal process heating temperature at time t, Ps(t) represents the vibration amplitude of the key bearing position with a speed of s at time t, Psth represents the ideal vibration amplitude of the key bearing position with a speed of s, and Jn Δt is the number of abnormal pulses of the vibration noise signal at the key bearing position in the time interval (t-Δt, t), J Δt is the allowable number of abnormal pulses of the vibration noise signal at the key bearing position in the time interval (t-Δt, t).

4. The method for analyzing aluminum profile surface defects based on machine vision according to claim 3, characterized in that: The method for obtaining the health risk value H(t) of the lubricating oil sample includes: Wherein, Xi(t) is the parameter value of the i-th lubricating oil monitoring item at time t, Xih(t) is the ideal value of the i-th lubricating oil monitoring item at time t, is the parameter average value of the i-th lubricating oil monitoring item in the time interval (t-Δt, t), and iS is the number of abnormal monitoring items.

5. The method for analyzing aluminum profile surface defects based on machine vision according to claim 2, characterized in that: The method for obtaining the process score value S(t) includes: Obtain the score value SE(t) of the aluminum bar heating process and the score value SL(t) of the aluminum bar stretching process respectively; Among them, β1 and β2 are the corresponding weight coefficients, and Dpst is the preset risk threshold.

6. The method for analyzing aluminum profile surface defects based on machine vision according to claim 5, characterized in that: The method for obtaining the score value SE(t) of the aluminum rod heating process includes: Among them, λ1 and λ2 are the corresponding weight coefficients respectively, Temp(t) represents the temperature value of the heating part at time t under no-load state, Temp0(t) represents the ideal temperature value of the heating part at time t under no-load state, t1 is the no-load heating time, t2 is the trial production heating time, and Tev(t) is the ambient temperature value.

7. The method for analyzing aluminum profile surface defects based on machine vision according to claim 5, characterized in that: The method for obtaining the score value SL(t) of the aluminum bar stretching process comprises: Among them, η1 and η2 are the corresponding weight coefficients respectively, Vemp(t) represents the pushing speed of the aluminum bar at time t in the no-load state, Vemp0(t) represents the ideal pushing speed of the aluminum bar at time t in the no-load state, VT(t) is the pushing speed at time t in the trial production process, VT0(t) is the standard pushing speed at time t in the trial production process, tA is the end time of no-load pushing, t2 is the end time of trial pushing, ΔV1 and ΔV2 are the corresponding preset comparison thresholds respectively.

8. The method for analyzing aluminum profile surface defects based on machine vision according to claim 2, characterized in that: The process of merging the surface snapshots of the target aluminum profile obtained from multiple angles and sending them to a pre-trained surface recognition model for evaluation, and obtaining the number of surface abnormal points and the corresponding confidence level and the corresponding quality inspection defect value output by the surface recognition model includes: SA1, take a snapshot of the target aluminum profile, obtain m corresponding aluminum profile surface snapshots Am, merge them into blank pictures in chronological order to obtain the recognition sampling picture A, and then send the surface recognition model; SA2, rotating the target aluminum profile by a specified angle, obtaining z corresponding aluminum profile surface snapshots Bz during the rotation process, merging them into a blank picture in chronological order to obtain a recognition sampling picture B, and then sending the surface recognition model; SA3, restore the position of the target aluminum profile, obtain the corresponding y aluminum profile surface snapshots Cy during the restoration process, merge them into a blank picture in chronological order to obtain the recognition sampling map C, and then send the surface recognition model; SA4, respectively obtain the quality inspection defect value Cp(A) corresponding to the identification sampling diagram A, the quality inspection defect value Cp(B) corresponding to the identification sampling diagram B, and the quality inspection defect value Cp(C) corresponding to the identification sampling diagram C; SA5, if Then The result is taken as the quality inspection defect value Cp(t); like or Then The result is taken as the quality inspection defect value Cp(t); like Then return to step SA1.

9. The method for analyzing aluminum profile surface defects based on machine vision according to claim 8, characterized in that: The method for obtaining the quality inspection defect value Cp(X) corresponding to the identification sampling diagram X includes: Among them, k is the number of surface anomaly points, pj is the confidence of the jth surface anomaly point, and Spj is the area value of the jth surface anomaly point.

Images