Intelligent analysis and decision-making system for air conditioner plastic shell quality inspection

Abstract

The invention relates to the technical field of air conditioner shell inspection, in particular to an air conditioner plastic shell quality inspection intelligent analysis and decision-making system which comprises the steps of collecting historical production and quality inspection data, analyzing and determining a defect set, dividing elements in the defect set into subsets according to the defect position of each defect product in the defect set and the defect position, abnormal features and performance data are extracted based on the abnormal data in the subsets; then collecting production process data in real time, comparing expression data to mark abnormal or inspection-free products, positioning defect types and positions of the abnormal products, finally detecting the abnormal products in a targeted manner, accurately positioning defect characteristics and causes, and realizing real-time monitoring and intelligent judgment of the production process. According to the invention, abnormal products can be rapidly identified, defect types and positions can be determined, and quality inspection accuracy and efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of air conditioner casing inspection technology, and in particular to an intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings. Background Technology

[0002] In the air conditioning manufacturing industry, the quality of the plastic casing of air conditioners has a crucial impact on the overall performance, appearance, and lifespan of the product.

[0003] The prior art CN119804203A discloses an automated method for quality inspection of air conditioner casing products, which includes the following steps: collecting and preprocessing the temperature, humidity, and external stress of the air conditioner casing; calculating the thermo-elastic coupling stress of the air conditioner casing, and using an activation energy model combined with the coefficient of thermal expansion to construct the oxidation acceleration effect of the air conditioner casing, and analyzing and evaluating the durability and safety of the air conditioner casing; based on the external stress, using Ito calculus to establish a multi-order random vibration equation, constructing a time-varying stress field model to simulate the external dynamic load, and combining the material yield stress to evaluate the strength level of the air conditioner casing; based on the durability, safety, and strength level of the air conditioner casing, comprehensively judging whether the quality of the air conditioner casing meets the standards.

[0004] During the processing of air conditioner plastic casings, multiple quality inspection items are required to ensure the production quality of the plastic casings. However, if all quality inspection items of the air conditioner plastic casings are inspected sequentially in real time during the production process, it will reduce the production efficiency. Moreover, although some existing automated quality inspection systems have improved the inspection speed to a certain extent, they often lack the ability to deeply analyze complex data. For example, they can only identify a single type of defect in a simple way, and cannot comprehensively consider multiple factors in the production process to determine the root cause of the defect. They also find it difficult to monitor and adjust the production process in real time and comprehensively. Summary of the Invention

[0005] The purpose of this invention is to solve the problems in the background art by proposing an intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] An intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings includes:

[0008] The information acquisition module is used to collect production information of the target product;

[0009] The data processing module is used to obtain historical production data and quality inspection data of the target product based on the production information of the target product, and to perform inertial processing on the production data and quality inspection data to determine the defect set. At the same time, according to the defect location of the defective products in the defect set, a subset is set in the defect set according to the defect location.

[0010] The anomaly analysis module is used to obtain the processed defect set and corresponding subsets. Based on the production time of the defective products in the subset, the corresponding production information is obtained. The production information is compared with the production parameters of normal products to identify abnormal data. Then, the feature processing algorithm is used to extract the feature data of all abnormal data in the same subset to determine the manifestation data of the defect.

[0011] The real-time acquisition module is used to collect process production data of the target product during real-time production.

[0012] The quality inspection positioning module is used to compare process production data with performance data to determine the product label of the target product. The product label includes abnormal products and products exempt from inspection. If the target product is an abnormal product, the same performance data as the process production data of the abnormal product is obtained, and the defect type and defect location of the target product are determined based on this performance data.

[0013] The product inspection module is used to receive and identify the product label of the target product, and to perform quality inspection on the abnormal product according to the defect type and defect location.

[0014] As a further aspect of the present invention, the method for determining the defect set includes:

[0015] Using the current time as the time node, obtain the quality inspection data within the valid time period before the time node, identify the types of defects in the quality inspection data, and the valid time refers to the time when the process actually runs and generates valid data in the historical data.

[0016] The target products produced within the selected valid time period are retrieved again, and the target products with defects are identified and marked as defective products. Then, the defective products are classified according to their defect types, and defective products with the same defect type are integrated into the same dataset and this dataset is marked as a defect set. In a defect set, all defective products have the same defect type.

[0017] As a further aspect of the present invention, the method for setting the subset includes:

[0018] All defect sets are sequentially labeled as target sets. Based on the model of the target product, the appearance model of the target product is obtained. The target product is the plastic shell of the air conditioner that is currently undergoing production analysis. The appearance model is a 3D image of the target product. Then, the unit area is set, and the appearance model is divided into several individual regions according to the unit area rule. The area of ​​the individual region is the unit area.

[0019] Obtain the target set, and mark the defect locations of all target products in the target set in the corresponding individual regions of the appearance model in sequence. At the same time, count the number of times each individual region is marked to obtain the region quantification value.

[0020] Divide the quantified value of the region by the total defect value to obtain the region frequency Fi, where i represents a different individual region and the total defect value is the total number of all defect locations in the target set.

[0021] The region frequency Fi is compared with the defect threshold Fy. If Fi < Fy, the corresponding single-unit region i is marked as an accidental region. Conversely, if Fi ≥ Fy, the corresponding single-unit region is marked as an inertial region.

[0022] Identify the location of all inertial regions in the appearance model. Using the inertial region as the center, identify the accidental regions adjacent to the inertial region and merge these accidental regions directly with the inertial region at the center to obtain local regions. If there are accidental regions without adjacent inertial regions, merge the consecutive accidental regions to obtain local regions. At the same time, integrate the defective products existing in the local regions into a subset.

[0023] As a further aspect of the present invention, when two inertial regions are adjacent to each other, the adjacent inertial regions and the accidental regions adjacent to these inertial regions are directly merged into a local region. If an accidental region is adjacent to both inertial regions, the region frequencies of the inertial regions are compared, and the accidental region is merged with the inertial region with the larger region frequency. If the region frequencies of the two inertial regions are equal, the accidental region is merged with the two adjacent inertial regions into a local region.

[0024] As a further aspect of the present invention, a unique code is set on the target product, and the code contains batch production information. Then, during the production and processing of the target product, the product is bound to the production link through the code identification. The time data of equipment, personnel and raw materials are integrated by relying on the MES system to achieve product traceability. The coding forms include QR codes, barcodes and RFID tags.

[0025] As a further aspect of the present invention, the method for determining performance data includes:

[0026] SS1: Select a subset from the target set, obtain all defective products in this subset, identify the production time of the defective products, extract the production process parameters within the corresponding time based on the production time, and mark the production process parameters of the defective products as defective production parameters;

[0027] Based on the production time of defective products in the subset, this production time is marked as an abnormal time. Normal products produced in the time close to the abnormal time are identified, and the production process parameters of the normal products are extracted. The production process parameters of the normal products are marked as normal production parameters. Normal products are products that have passed quality inspection.

[0028] SS2: Compare adjacent defective production parameters with normal production parameters. If the defective production parameters are the same as the normal production parameters, the defective production parameters are marked as normal data. Conversely, if the defective production parameters are different from the normal production parameters, the defective production parameters are marked as abnormal data. When comparing defective production parameters with normal production parameters, only parameters of the same type are compared.

[0029] SS3: Obtain all abnormal data in the subset, integrate abnormal parameters of the same type to obtain similar data, and then use a clustering algorithm to calculate the feature data in the similar data. Integrate the feature data of all parameter types and mark them as the performance data of the corresponding defect location in this subset.

[0030] As a further aspect of the present invention, the method for determining the product label of the target product includes:

[0031] All performance data is acquired, and the process production data is compared with the performance data. If there is a match between the performance data and the process production data, the target product is marked as an abnormal product. At the same time, the defect type and defect location corresponding to this performance data are acquired and transmitted to the product inspection module. Conversely, if there is no data in the performance data that is the same as the process production data, the target product is marked as an inspection-exempt product.

[0032] As a further aspect of the present invention, when comparing process production data and performance data, data of the same type are compared. If any type of data is the same in the process production data and performance data, the corresponding target product is marked as an abnormal product. If all types of data are different in the process production data and performance data, the target product is marked as a normal product.

[0033] As a further aspect of the present invention, a terminal display module is also included, which is used to display the detection results of the product inspection module on the abnormal products on the display terminal.

[0034] Compared with existing technologies, the advantages of this invention are:

[0035] This invention collects historical production and quality inspection data, analyzes and determines defect sets and subsets, and extracts abnormal features and performance data. It then collects real-time production process data, compares performance data to mark abnormal or exempted products, locates the defect type and location of abnormal products, and finally conducts targeted testing on abnormal products to accurately locate defect features and causes. This enables real-time monitoring and intelligent judgment of the production process. This invention can quickly identify abnormal products and determine defect types and locations, improving the accuracy and efficiency of quality inspection, reducing missed and false detections, providing a scientific basis for production decisions, assisting in process improvement, reducing costs, and enhancing product quality control capabilities. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the system structure of the present invention. Detailed Implementation

[0037] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0038] Reference Figure 1 An intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings includes an information acquisition module, a data processing module, an anomaly analysis module, a real-time acquisition module, a quality inspection positioning module, a product inspection module, and a terminal display module.

[0039] The information acquisition module is used to collect production information and quality inspection standard information of the target product. The production information refers to the key data generated and recorded during the production process of the target product, including production process parameters and production process records, such as production date and production batch. The quality inspection standard information is used to determine the specifications, indicators and requirements for whether the air conditioner plastic shell is qualified, such as the dimensional deviation of key parts and appearance defects. Furthermore, in this embodiment, the target product refers to the air conditioner plastic shell. Then, a one-way communication connection is established between the information acquisition module and the data processing module.

[0040] The data processing module is used to acquire historical production and quality inspection data of the target product based on its production information, and to perform inertial processing on the production and quality inspection data. The inertial processing methods include:

[0041] S1: Using the current time as the time node, obtain the quality inspection data within the valid time period before the time node, and identify the defect types existing in the quality inspection data. Among them, the defect types include scratches, color difference, burrs, missing materials, deformation, and physical performance defects. Furthermore, the valid time refers to the time when the process actually runs and generates valid data in the historical data. The specific time range of the valid time is set by those skilled in the art based on big data experience.

[0042] The target products produced within the selected valid time period are retrieved again, the defective target products are identified and marked as defective products. Then, the defective products are classified according to the defect type. Defective products with the same defect type are integrated into the same dataset and this dataset is marked as a defect set. In a defect set, all defective products have the same defect type, and this defect type is used as the label of the defect set.

[0043] S2: Select any defect set, take this defect set as an example, mark this defect set as the target set, identify the defect location of all defective products in the target set, and set a subset in the target set based on the defect location, where one subset corresponds to one defect location;

[0044] Then, based on the production information, the production time of each defective product is obtained, and the production times of the defective products in each subset are integrated to obtain a set of times.

[0045] It should be further explained that each target product is assigned a unique code, which contains information such as batch number and production line number. During the production and processing of the target product, the product is linked to the production process through the code. By relying on systems such as MES to integrate the time data of equipment, personnel and raw materials, product traceability can be achieved. The coding form includes QR code, barcode or RFID tag, etc.

[0046] Furthermore, the methods for setting subsets in the target set include:

[0047] Based on the model of the target product, obtain the appearance model of the target product, wherein the appearance model is a three-dimensional image of the target product. Then, set the unit area and divide the appearance model into several individual regions according to the unit area rules. Furthermore, the area of ​​the individual region is the unit area. The specific value of the unit area is set by those skilled in the art based on big data experience.

[0048] Obtain the target set, and mark the defect locations of all target products in the target set in the corresponding individual regions of the appearance model in sequence. At the same time, count the number of times each individual region is marked to obtain the region quantification value.

[0049] Then, the region quantification value is divided by the total defect value to obtain the region frequency Fi, where i represents different individual regions and the total defect value is the total number of all defect locations in the target set.

[0050] The region frequency Fi is compared with the defect threshold Fy. If Fi < Fy, the corresponding single region i is marked as an accidental region. Conversely, if Fi ≥ Fy, the corresponding single region is marked as an inertial region. The specific value of the defect threshold Fy is set by those skilled in the art based on big data experience. In this embodiment, the defect threshold Fy is set to 5%.

[0051] Identify the location of all inertial regions in the appearance model. Using the inertial region as the center, identify the accidental regions adjacent to the inertial region and merge these accidental regions directly with the inertial region at the center to obtain a local region. If there is an accidental region without an adjacent inertial region, merge the consecutive accidental regions to obtain a local region. At the same time, integrate the defective products existing in the local region into a subset.

[0052] It should be further explained that when there are two inertial regions that are adjacent to each other, the adjacent inertial regions and the accidental regions adjacent to them are directly merged into a local region. If there is an accidental region that is adjacent to two inertial regions, the region frequencies of the inertial regions are compared, and the accidental region is merged with the inertial region with the larger region frequency. If the region frequencies of the two inertial regions are equal, the accidental region is merged with the two adjacent inertial regions into a local region.

[0053] A one-way communication connection is established between the data processing module and the anomaly analysis module, and the target set is transmitted to the anomaly analysis module.

[0054] The anomaly analysis module is used to acquire the processed target set. Based on the production time of the defective products in the target set, it obtains the corresponding production information, performs feature analysis on the production information and the target set, and determines the manifestation data of defects on the target products. Further, the specific methods for determining the manifestation data include:

[0055] SS1: Select a subset from the target set, obtain all defective products in this subset, identify the production time of the defective products, extract the production process parameters within the corresponding time according to the production time, and mark the production process parameters of the defective products as defective production parameters. In this embodiment, the production process parameters include temperature, equipment vibration value, voltage, and injection speed, etc.

[0056] At the same time, based on the production time of defective products in the subset, this production time is marked as an abnormal time. Normal products produced in the time close to the abnormal time are identified, and the production process parameters of the normal products are extracted. The production process parameters of the normal products are marked as normal production parameters. Among them, normal products are products that have passed quality inspection.

[0057] SS2: Compare adjacent defective production parameters with normal production parameters. If the defective production parameter is the same as the normal production parameter, mark the defective production parameter as normal data. Conversely, if the defective production parameter is different from the normal production parameter, mark the defective production parameter as abnormal data. When comparing defective production parameters with normal production parameters, only parameters of the same type are compared. For example, only normal production parameters and defective production parameters that are both temperature data are compared.

[0058] SS3: Obtain all abnormal data in the subset, integrate the abnormal parameters of the same type to obtain similar data, and then use the feature processing algorithm to calculate the feature data in the similar data. Integrate the feature data of all parameter types and mark it as the performance data of the corresponding defect location of this subset. Furthermore, in this embodiment, the feature processing algorithm is a clustering algorithm. The specific clustering algorithm for selecting feature data is existing technology and will not be described in detail here.

[0059] Following the above method, calculate the performance data of all subsets in the target set at corresponding positions, and then establish a one-way communication connection between the anomaly analysis module and the quality inspection positioning module.

[0060] The real-time acquisition module is used to collect production data during the real-time production process of the target product, obtain process production data, and transmit it to the quality inspection positioning module.

[0061] The quality inspection positioning module is used to receive real-time collected process production data and acquire all performance data. It compares the process production data with the performance data. If there is a performance data that matches the process production data, the target product is marked as an abnormal product. At the same time, the defect type and defect location corresponding to this performance data are acquired and transmitted to the product inspection module. Conversely, if there is no data in the performance data that matches the process production data, the target product is marked as an inspection-exempt product.

[0062] It should be further explained that when comparing process production data with performance data, data of the same type are compared. If any type of data is the same in the process production data and performance data, the corresponding target product is marked as an abnormal product and processed according to the above method. If all types of data are different in the process production data and performance data, the target product is marked as a normal product.

[0063] Then, a one-way communication connection is established between the quality inspection positioning module and the product inspection module;

[0064] The product inspection module is used to receive and identify the product label of the target product. The product label includes defective products and products exempt from inspection. If the target product is an exempt product, the product inspection module will not perform quality inspection on the target product. If the target product is a defective product, the module will obtain the defect type and defect location of the defective product. Then, the product inspection module will perform quality inspection on the defective product according to the defect type and defect location and obtain the inspection results.

[0065] The product inspection module then transmits the test results to the terminal display module, which displays the received test results on the display terminal, making it convenient for staff to view the quality inspection results of the target product.

[0066] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A smart analysis and decision-making system for quality inspection of air conditioner plastic casings, characterized in that, include: The information acquisition module is used to collect production information of the target product; The data processing module is used to obtain historical production data and quality inspection data of the target product based on the production information of the target product, and to perform inertial processing on the production data and quality inspection data to determine the defect set. At the same time, according to the defect location of the defective products in the defect set, a subset is set in the defect set according to the defect location. The anomaly analysis module is used to obtain the processed defect set and corresponding subsets. Based on the production time of the defective products in the subset, the corresponding production information is obtained. The production information is compared with the production parameters of normal products to identify abnormal data. Then, the feature processing algorithm is used to extract the feature data of all abnormal data in the same subset to determine the manifestation data of the defect. The real-time acquisition module is used to collect process production data of the target product during real-time production. The quality inspection positioning module is used to compare process production data with performance data to determine the product label of the target product. The product label includes abnormal products and products exempt from inspection. If the target product is an abnormal product, the same performance data as the process production data of the abnormal product is obtained, and the defect type and defect location of the target product are determined based on this performance data. The product inspection module is used to receive and identify the product label of the target product, and to perform quality inspection on the abnormal product according to the defect type and defect location.

2. The intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings according to claim 1, characterized in that, Methods for determining the defect set include: Using the current time as the time node, obtain the quality inspection data within the valid time period before the time node, identify the types of defects in the quality inspection data, and the valid time refers to the time when the process actually runs and generates valid data in the historical data. The target products produced within the selected valid time period are retrieved again. Defective target products are identified and marked as defective products. Then, defective products are classified according to defect type. Defective products with the same defect type are integrated into the same dataset and this dataset is marked as a defect set. In a defect set, all defective products have the same defect type.

3. The intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings according to claim 2, characterized in that, Methods for setting subsets include: All defect sets are sequentially labeled as target sets. Based on the model of the target product, the appearance model of the target product is obtained. The target product is the plastic shell of the air conditioner that is currently undergoing production analysis. The appearance model is a 3D image of the target product. Then, the unit area is set, and the appearance model is divided into several individual regions according to the unit area rule. The area of ​​the individual region is the unit area. Obtain the target set, and mark the defect locations of all target products in the target set in the corresponding individual regions of the appearance model in sequence. At the same time, count the number of times each individual region is marked to obtain the region quantification value. Divide the quantified value of the region by the total defect value to obtain the region frequency Fi, where i represents a different individual region and the total defect value is the total number of all defect locations in the target set. The region frequency Fi is compared with the defect threshold Fy. If Fi < Fy, the corresponding single-unit region i is marked as an accidental region. Conversely, if Fi ≥ Fy, the corresponding single-unit region is marked as an inertial region. Identify the location of all inertial regions in the appearance model. Using the inertial region as the center, identify the accidental regions adjacent to the inertial region and merge these accidental regions directly with the inertial region at the center to obtain local regions. If there are accidental regions without adjacent inertial regions, merge the consecutive accidental regions to obtain local regions. At the same time, integrate the defective products existing in the local regions into a subset.

4. The intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings according to claim 3, characterized in that, When two inertial regions are adjacent to each other, the adjacent inertial regions and the accidental regions adjacent to them are directly merged into a local region. If an accidental region is adjacent to two inertial regions, the regional frequencies of the inertial regions are compared, and the accidental region is merged with the inertial region with the larger regional frequency. If the regional frequencies of the two inertial regions are equal, the accidental region is merged with the two adjacent inertial regions into a local region.

5. The intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings according to claim 3, characterized in that, Each target product is assigned a unique code containing batch production information. During the production and processing of the target product, the product is linked to the production process through the code. The MES system integrates the time data of equipment, personnel, and raw materials to achieve product traceability. The coding formats include QR codes, barcodes, and RFID tags.

6. The intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings according to claim 1, characterized in that, Methods for determining performance data include: SS1: Select a subset from the target set, obtain all defective products in this subset, identify the production time of the defective products, extract the production process parameters within the corresponding time based on the production time, and mark the production process parameters of the defective products as defective production parameters; Based on the production time of defective products in the subset, this production time is marked as an abnormal time. Normal products produced in the time close to the abnormal time are identified, and the production process parameters of the normal products are extracted. The production process parameters of the normal products are marked as normal production parameters. Normal products are products that have passed quality inspection. SS2: Compare adjacent defective production parameters with normal production parameters. If the defective production parameters are the same as the normal production parameters, the defective production parameters are marked as normal data. Conversely, if the defective production parameters are different from the normal production parameters, the defective production parameters are marked as abnormal data. When comparing defective production parameters with normal production parameters, only parameters of the same type are compared. SS3: Obtain all abnormal data in the subset, integrate abnormal parameters of the same type to obtain similar data, and then use a clustering algorithm to calculate the feature data in the similar data. Integrate the feature data of all parameter types and mark them as the performance data of the corresponding defect location in this subset.

7. The intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings according to claim 1, characterized in that, Methods for determining the product label of the target product include: All performance data is acquired, and the process production data is compared with the performance data. If there is a match between the performance data and the process production data, the target product is marked as an abnormal product. At the same time, the defect type and defect location corresponding to this performance data are acquired and transmitted to the product inspection module. Conversely, if there is no data in the performance data that is the same as the process production data, the target product is marked as an inspection-exempt product.

8. The intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings according to claim 7, characterized in that, When comparing process production data with performance data, data of the same type are compared. If any type of data is the same in the process production data and performance data, the corresponding target product is marked as an abnormal product. If all types of data are different in the process production data and performance data, the target product is marked as a normal product.

9. The intelligent analysis and decision-making system for quality inspection of air conditioner plastic casings according to claim 1, characterized in that, It also includes a terminal display module, which is used to display the test results of the product inspection module on abnormal products on the display terminal.

Images