An intelligent manufacturing anomaly monitoring method and system based on multi-source data

Abstract

The application discloses a kind of based on multi-source data's intelligent manufacturing exception monitoring method and system, it is related to manufacturing exception monitoring field, obtains and analyzes production manufacturing plan, determines industrial product, manufacturing product feature and production line task time sequence;According to manufacturing product feature, determine product feature structure parameter;By analyzing product feature structure parameter, determine product structure threshold value;Collect the position tolerance of product mechanism area, appearance angle and surface gray value;According to position tolerance, determine welding structure smoothness, generate structure exception data;According to structure exception data and production line task time sequence, determine production line exception threshold value;According to appearance angle and surface gray value, generate surface fitting data, combine product structure feature parameter, generate surface offset threshold value;Collect manufacturing process real-time data, through product structure threshold value, production line exception threshold value and surface offset threshold value, manufacturing process real-time data is monitored.

Description

Technical Field

[0001] The present invention relates to the field of anomaly monitoring, and in particular to a method and system for intelligent manufacturing anomaly monitoring based on multi-source data. Background Art

[0002] Intelligent manufacturing anomaly monitoring uses advanced technologies and processing methods to collect and dynamically analyze various data in the intelligent manufacturing system in real time. By setting reasonable thresholds, various deviations from normal operating conditions in the manufacturing system can be discovered in a timely manner. In intelligent manufacturing anomaly monitoring, multi-source data refers to data collected from multiple different sources or systems. These data include various types, formats and sources. By integrating data from different systems and sensors, it can provide more comprehensive and accurate information support, which is helpful in production process optimization, fault prediction, quality control and energy management.

[0003] The intelligent manufacturing conditions of automotive parts are complex and changeable. The operating parameters and status of the equipment will vary greatly when producing different products. Traditional methods are usually based on fixed thresholds or rules and cannot adaptively adjust monitoring strategies and parameters according to changes in actual working conditions. They are prone to false alarms or missed alarms. Traditional monitoring methods can usually only detect some obvious anomalies that are directly reflected in the monitoring parameters. It is difficult to accurately diagnose some potential and deep-seated causes of failures, which in turn leads to potential quality risks being ignored and the overall production capacity of the manufacturing process cannot be fully utilized. Summary of the Invention

[0004] In order to solve the above technical problems, a method and system for intelligent manufacturing anomaly monitoring based on multi-source data are provided. This technical solution solves the problem proposed in the above background technology that the monitoring strategy and parameters cannot be adaptively adjusted according to changes in actual working conditions, and false alarms or missed alarms are prone to occur. It is difficult to accurately diagnose some potential and deep-seated causes of failures, which leads to potential quality risks being ignored and the overall production capacity of the manufacturing process cannot be fully utilized.

[0005] In order to achieve the above objects, the technical solution adopted by the present invention is:

[0006] A method for monitoring abnormalities in intelligent manufacturing based on multi-source data, characterized by comprising:

[0007] Obtain and analyze production and manufacturing plans to determine industrial products, manufacturing product characteristics, and production line task time series;

[0008] Analyze industrial products according to the characteristics of manufactured products, generate product structure areas, and determine product characteristic structure parameters in combination with production and manufacturing plans;

[0009] Analyze product characteristic structure parameters and determine product structure thresholds;

[0010] Collecting product area structure information, the product area structure information including position tolerance, appearance angle and surface gray value;

[0011] According to the position tolerance, determining welding structure smoothness, generating structure abnormal data;

[0012] According to the structure abnormal data and the production line task time sequence, determining production line abnormal threshold;

[0013] According to the appearance angle and the surface gray value, generating surface fitting data, combining product structure feature parameters, generating surface offset threshold;

[0014] Obtaining manufacturing process real-time data, and based on the product structure threshold, the production line abnormal threshold and the surface offset threshold, comparing the manufacturing process real-time data in turn, realizing monitoring.

[0015] Preferably, the product feature structure parameters are determined according to the production manufacturing plan, specifically including:

[0016] According to the manufacturing product features, analyzing the industrial products and performing edge processing, generating structure edges;

[0017] Taking the structure edges as the reference, performing contour extraction on the industrial products, generating product structure areas;

[0018] Analyzing the production manufacturing plan, determining the functional positioning of each product structure area in the industrial products and the actual application scene of the industrial products;

[0019] According to the actual application scene of the industrial products, collecting scene data, and based on digital modeling technology, generating a scene model;

[0020] Based on virtual prototyping technology, on the basis of the scene model, according to the functional positioning of each product structure area in the industrial products, performing stress analysis on the industrial products, generating industrial product stress analysis results;

[0021] Taking the industrial product stress analysis results as the boundary conditions of the structure edges, iteratively processing each product structure area, generating structure area stress results;

[0022] According to the related theories of structural mechanics and material mechanics, combining the structure area stress results, determining the product feature structure parameters.

[0023] Preferably, the product structure threshold is determined, specifically including:

[0024] Collect historical quality feedback values of industrial products and product feature structure parameters corresponding to each quality feedback value, analyze the historical quality feedback values of the industrial products, and generate quality statistical feature data, the quality statistical feature data including a quality-related parameter distribution range, a quality-related parameter mean value, and a quality-related parameter standard deviation;

[0025] According to the quality-related parameter standard deviation, the historical quality feedback values of the industrial products are preprocessed to generate preferred quality feedback values;

[0026] The preferred quality feedback values are summarized and arranged in descending order to generate quality feedback sequence data;

[0027] According to the quality-related parameter distribution range, the quality feedback sequence data is analyzed to obtain upper half distribution interval data;

[0028] According to the product feature structure parameters corresponding to each quality feedback value, the upper half distribution interval data is analyzed to obtain a maximum feature structure parameter value and a minimum feature structure parameter value, and a difference between the maximum feature structure parameter value and the minimum feature structure parameter value is taken as a first safety margin;

[0029] Combined with the product feature structure parameters corresponding to each quality feedback value and the quality-related parameter mean value, a product feature structure parameter corresponding to the quality-related parameter mean value is obtained, denoted as a standard structure parameter;

[0030] The standard structure parameter is added to the first safety margin to generate a product structure threshold value.

[0031] Preferably, the determination of the welding structure smoothness generates structure abnormal data, specifically including:

[0032] The product area structure information is analyzed to obtain a position tolerance of each product structure area;

[0033] According to the functional positioning of each product structure area in the industrial product, product structure areas with the same functional positioning are summarized to generate a same-type structure area;

[0034] The position tolerances of the same-type structure areas are compared in sequence to obtain a same-type position tolerance difference value;

[0035] The same-type position tolerance difference values are averaged, and the average value is analyzed to obtain same-type standard deviation data;

[0036] The same-type standard deviation data is analyzed to obtain a median standard deviation value of the same-type standard deviation data, and the same-type standard deviation data is analyzed based on the median standard deviation value to generate a data deviation degree;

[0037] The welding structure smoothness is determined according to the data deviation degree.

[0038] generating a structure judgment threshold according to the production plan and the product structure threshold;

[0039] Based on the structure judgment threshold, the data deviation degree and the welding structure smoothness are analyzed, and the abnormal data points are screened out. The data deviation degree and the welding structure smoothness corresponding to the abnormal data points are sorted and marked to generate structure abnormal data.

[0040] Preferably, the determination of the production line abnormal threshold specifically comprises:

[0041] Analyzing the production line task time sequence to obtain historical production line data, the historical production line data including the task types of different production lines, the historical production efficiency of the production line and the historical running state of the production line equipment in each time period;

[0042] According to the task types of different production lines in each time period, the historical production efficiency of the production line is analyzed to generate the production efficiency corresponding to the production line in each period;

[0043] According to the production line task time sequence, the historical running state of the production line equipment is analyzed to generate the equipment running fluctuation period data;

[0044] The production efficiency corresponding to the production line in each period is arranged and analyzed from high to low respectively to generate the upper quartile and the lower quartile. The difference between the upper quartile and the lower quartile is taken as the second safety margin;

[0045] The sum of the upper quartile and the second safety margin is taken as the upper limit of the production line efficiency, and the difference between the lower quartile and the second safety margin is taken as the lower limit of the production line efficiency, to determine the production line abnormal threshold.

[0046] Preferably, the generation of the surface fitting data comprises generating a surface offset threshold in combination with the product structure characteristic parameters, specifically comprising:

[0047] Based on the laser tracker, the appearance angle of adjacent product structure regions is measured respectively with the intersection points of each product structure region as the reference to obtain surface angle characteristic values;

[0048] The surface angle characteristic values of each product structure region are summarized and plotted to generate a position-angle characteristic value curve to obtain a curve ratio;

[0049] The curve ratio is arranged and sorted, and the abnormal curve ratio is marked to obtain the surface gray value of the position corresponding to the abnormal curve ratio;

[0050] The surface gray value of the product structure region containing the position corresponding to the abnormal curve ratio is analyzed to generate the surface fitting data;

[0051] According to the product structure characteristic parameters, the surface standard fitting value is determined.

[0052] According to the surface standard fitting value, the surface fitting data is analyzed to generate a surface offset threshold value.

[0053] Further, an intelligent manufacturing abnormality monitoring system based on multi-source data is proposed, which is suitable for the monitoring method described above, and is characterized in that it specifically comprises:

[0054] The acquisition module is used to acquire production manufacturing plans and manufacturing process real-time data, collect product area structure information, and collect industrial product historical quality feedback values, wherein the manufacturing process real-time data includes product structure values, production line real-time values, and surface flatness values, and the obtained data is transmitted to the analysis module and the monitoring module.

[0055] The analysis module is used to determine product feature structure parameters according to manufacturing product features, analyze product feature structure parameters, determine product structure threshold values, generate structure abnormality data through position tolerances, determine production line abnormality threshold values through structure abnormality data and production line task time sequences, generate surface offset threshold values according to appearance angles and surface gray scale values, and transmit the determined and generated data to the monitoring module and the management module.

[0056] The monitoring module is used to sort the received data, divide the data transmitted by the acquisition module into a category, divide the data transmitted by the analysis module into a category, compare the corresponding relationship between the two, determine whether an abnormal situation occurs in the intelligent manufacturing process, and transmit the data to the management module.

[0057] The management module is used to sort the received data and display the intelligent manufacturing process in real time according to the sorted results.

[0058] Preferably, the acquisition module specifically comprises:

[0059] The first acquisition unit is used to acquire production manufacturing plans, determine industrial product features, manufacturing product features, and production line task time sequences, and transmit the data to the analysis module.

[0060] The second acquisition unit is used to determine the position tolerance of the product through a three-coordinate measuring instrument, collect appearance angles through a laser tracker, collect surface gray scale values of the product through an industrial camera, and transmit the data to the analysis module.

[0061] The third acquisition unit is used to collect industrial product historical quality feedback values and transmit the data to the analysis module.

[0062] Preferably, the analysis module specifically comprises:

[0063] The first analysis unit is used for edge processing of the industrial product according to the product feature, generating a structure edge, and performing contour extraction to generate a product structure area, performing stress analysis on the industrial product according to the function positioning of the product structure area in the industrial product, determining a product feature structure parameter, and transmitting data to the monitoring module;

[0064] The second analysis unit is used for analyzing an industrial product historical quality feedback value, generating quality statistical feature data, determining a first safety margin and a standard structure parameter according to the quality statistical feature data, further generating a product structure threshold value, and transmitting data to the monitoring module.

[0065] The third analysis unit is used for analyzing a production line equipment historical operation state according to a production line task time sequence, generating equipment operation fluctuation cycle data and a corresponding production efficiency of each time period, arranging the corresponding production efficiency of each time period, determining a production line abnormal threshold value, and transmitting data to the monitoring module.

[0066] Preferably, the monitoring module specifically comprises:

[0067] The first monitoring unit is used for analyzing data deviation and welding structure smoothness according to the structure determination threshold value, screening out abnormal data points, and transmitting data to the management module.

[0068] The second monitoring unit is used for comparing product structure values in the manufacturing process real-time data with the product structure threshold value as a reference to judge whether there is abnormal data value, and transmitting data to the management module.

[0069] The third monitoring unit is used for comparing production line real-time values in the manufacturing process real-time data with the production line abnormal threshold value as a reference to judge whether there is abnormal data value, and transmitting data to the management module.

[0070] The fourth monitoring unit is used for comparing surface flatness values in the manufacturing process real-time data with the surface offset threshold value as a reference to judge whether there is abnormal data value, and transmitting data to the management module.

[0071] Compared with the prior art, the present application has the following beneficial effects:

[0072] The application provides an intelligent manufacturing anomaly monitoring method and system based on multi-source data. BRIEF DESCRIPTION OF DRAWINGS

[0073] Figure 1 A flow chart of the intelligent manufacturing anomaly monitoring method based on multi-source data is provided in the application.

[0074] Figure 2 A flow chart of the method for determining product feature structure parameters is provided in the application.

[0075] Figure 3 A flow chart of the method for determining product structure threshold values is provided in the application.

[0076] Figure 4 A flow chart of the method for generating structure anomaly data is provided in the application.

[0077] Figure 5 A flow chart of the method for determining line anomaly threshold values is provided in the application.

[0078] Figure 6 A flow chart of the method for generating surface offset threshold values is provided in the application.

[0079] Figure 7 A structure diagram of the intelligent manufacturing anomaly monitoring system based on multi-source data is provided in the application. DETAILED DESCRIPTION

[0080] The following description is provided to enable any person skilled in the art to practice the application. The preferred embodiments described in the following description are only provided as examples and modifications to the preferred embodiments can be made by one skilled in the art without departing from the spirit of the application.

[0081] ReferenceFigure 1 As shown, an intelligent manufacturing anomaly monitoring method based on multi-source data, comprising:

[0082] Acquire and analyze production planning, determine industrial product, manufacturing product features and production line task time sequence;

[0083] According to the manufacturing product features, analyze the industrial products, generate product structure area, and determine the product feature structure parameters combined with the production planning;

[0084] Analyze the product feature structure parameters to determine the product structure threshold;

[0085] Collect product area structure information, including position tolerance, appearance angle and surface gray value;

[0086] According to the position tolerance, determine the welding structure smoothness, and generate structure anomaly data;

[0087] According to the structure anomaly data and the production line task time sequence, determine the production line anomaly threshold;

[0088] According to the appearance angle and the surface gray value, generate the surface fitting data, and generate the surface offset threshold combined with the product structure feature parameters;

[0089] Obtain manufacturing process real-time data, and compare the manufacturing process real-time data based on the product structure threshold, the production line anomaly threshold and the surface offset threshold in turn to realize monitoring.

[0090] The scheme determines the industrial product, the manufacturing product feature and the production line task time sequence by acquiring and analyzing the production manufacturing plan, analyzes the industrial product according to the manufacturing product feature, generates a product structure region, determines product feature structure parameters in combination with the production manufacturing plan, determines product structure threshold values by analyzing the product feature structure parameters, collects position tolerances, appearance angles and surface gray scale values of the product mechanism region, determines welding structure smoothness according to the position tolerances, generates structure abnormal data, determines production line abnormal threshold values according to the structure abnormal data and the production line task time sequence, generates surface fitting data according to the appearance angles and the surface gray scale values, generates surface offset threshold values in combination with the product structure feature parameters, and collects manufacturing process real-time data. The industrial product represents a product after completion of manufacturing, the manufacturing product feature includes product structure features and product appearance features, the product feature structure parameter represents structure parameters that need to be paid attention to when each structure region of the manufacturing product corresponds to a function in real time, and the product structure threshold value represents a data range of each structure parameter when the product achieves an expected effect. The position tolerance represents a difference between the upper surface height and the lower surface height in the vertical plane in the same product structure region, the appearance angle represents a surface curvature existing between adjacent product structure regions in the vertical plane, the surface gray scale value represents a numerical index of the light and dark degree of the product surface, the welding structure smoothness represents a degree of surface smoothness of a welding joint and a nearby region after welding is completed, the structure abnormal data represents a data deviation degree corresponding to an abnormal data point and the welding structure smoothness, and the surface fitting data represents a matching degree value between the surfaces of two objects.

[0091] Referring to Figure 2 In combination with the production manufacturing plan, the product feature structure parameters are determined, and specifically include:

[0092] According to the manufacturing product feature, the industrial product is analyzed and edge processing is performed to generate a structure edge.

[0093] The industrial product is profiled with the structure edge as a reference to generate each product structure region.

[0094] The production manufacturing plan is analyzed to determine the functional positioning of each product structure region in the industrial product and the actual application scenario of the industrial product.

[0095] According to the actual application scenario of the industrial product, scene data is collected, and a scene model is generated based on digital modeling technology.

[0096] Based on virtual prototyping technology, the industrial product is subjected to stress analysis according to the functional positioning of each product structure region in the industrial product based on the scene model to generate an industrial product stress analysis result.

[0097] The stress analysis result of the industrial product is taken as a boundary condition of the structure edge, and each product structure area is iterated to generate a stress result of the structure area;

[0098] According to the related theories of structural mechanics and material mechanics, the product feature structure parameters are determined in combination with the stress result of the structure area, the critical buckling load of the structure area is calculated through the Euler formula, and the critical pressure of the thin wall is determined, wherein the formula for determining the critical pressure of the thin wall is , The critical pressure of the thin wall is represented by Pcr, The elastic modulus of the material is represented by E, The Poisson's ratio is represented by μ, The position tolerance is represented by δ, , The product feature structure parameter value is represented by X.

[0099] It can be understood that in industrial production, firstly, the industrial product is analyzed in depth according to the characteristics of the manufactured product, the material properties, structural design, process requirements and functional performance of the product are comprehensively considered, and based on these analyses, the marginalization processing is carried out, the function is screened, the key core part is reserved, the unnecessary parts and complex design are simplified, the structure edge is generated, and by defining the core and non-core parts of the product in the manufacturing process, it is helpful to more efficiently exert the production capacity of product manufacturing. In the production and manufacturing plan analysis of automobile parts, different product structure regions have different functional positioning, and the scheme takes the door as an example. The door occupies a key position in the overall structure of the automobile, and its functional positioning has multiple attributes. From the safety point of view, the door is an important defense line to protect the life safety of the driver and passenger. In a collision accident, a solid door structure can effectively resist external impact force and prevent the door from deforming into the cabin, thereby providing sufficient living space for the people inside the vehicle. The inside of the door is usually equipped with high-strength anti-collision steel beams made of special steel, which can absorb and disperse collision energy in an instant, greatly reducing the harm of the accident to the people inside the vehicle. In terms of convenience, the door is the only access for the driver and passenger to get in and out of the vehicle, and the convenience of opening and closing directly affects the user experience. The modern automobile door design fully considers ergonomics, optimizes the position, shape and opening angle of the door handle, so that passengers can easily open and close the door. The door also bears the important functions of sound insulation, heat insulation and waterproofing. The inside of the door is filled with a large amount of sound insulation materials such as sound-absorbing cotton and rubber sealing strips, which can create a quiet driving environment for passengers. However, due to the complexity of these functions, each region of the door bears different responsibilities, and different requirements are placed on the structural parameters of each region. Through the related theories of structural mechanics and material mechanics, the stress region of the structure is analyzed, the stress type is judged to be tension, compression, bending or torsion, the stress size and distribution law are determined, the stress value calculated by stress is compared with the allowable stress of the material to ensure that the structure is within the safe range, and then the principles of structural stability and stiffness in structural mechanics, such as deformation calculation of simply supported beam under different loads, are combined to determine the product feature structural parameters, such as the size of the part, the selection of the material and the shape of the structure, to ensure that the product can operate stably and reliably under the given stress environment.

[0100] Referring to Figure 3 determining a product structure threshold, specifically comprising:

[0101] Collect historical quality feedback values of industrial products and product feature structure parameters corresponding to each quality feedback value, analyze the historical quality feedback values of the industrial products, and generate quality statistical feature data, wherein the quality statistical feature data includes a quality-related parameter distribution range, a quality-related parameter mean value, and a quality-related parameter standard deviation, the historical quality feedback values of the industrial products refer to customer use feedback data values of product quality in the use and after-sales process of the industrial products;

[0102] According to the quality-related parameter standard deviation, the historical quality feedback values of the industrial products are preprocessed to generate preferred quality feedback values;

[0103] The preferred quality feedback values are summarized and arranged in descending order to generate quality feedback sequence data;

[0104] According to the quality-related parameter distribution range, the quality feedback sequence data is analyzed to obtain upper half distribution interval data, wherein the upper half distribution interval data represents a quality-related parameter distribution range corresponding to the first half data in the quality feedback sequence data;

[0105] According to the product feature structure parameters corresponding to each quality feedback value, the upper half distribution interval data is analyzed to obtain a maximum feature structure parameter value and a minimum feature structure parameter value, and a difference between the maximum feature structure parameter value and the minimum feature structure parameter value is taken as a first safety margin;

[0106] Combined with the product feature structure parameters corresponding to each quality feedback value and the quality-related parameter mean value, a product feature structure parameter corresponding to the quality-related parameter mean value is obtained, which is denoted as a standard structure parameter;

[0107] The standard structure parameter is added to the first safety margin to generate a product structure threshold.

[0108] It can be understood that the historical quality feedback values are arranged in ascending order to form an ordered data set, the value at the 0.025(n+1)th position after sorting is taken as the lower limit of the distribution, the value at the 0.975(n+1)th position is taken as the upper limit of the distribution, the data between the upper limit and the lower limit of the distribution is taken as the quality-related parameter distribution range, the mean and the standard deviation of the data in the quality-related parameter distribution range are obtained, the standard deviation can reflect the dispersion degree of the quality data, the larger the value is, the more dispersed the data is, and the greater the quality fluctuation is; on the contrary, the data is more concentrated, and the quality is relatively stable, the quality-related parameter standard deviation can be used to set a reasonable threshold range, the range of 2 times the standard deviation floating up and down from the mean of the quality-related parameter is set as the reasonable range threshold, then the historical quality feedback values of the industrial products are compared with the reasonable range threshold one by one, and the abnormal values beyond the range are removed, because these abnormal values are caused by measurement errors, special production accidents and other accidental factors, and cannot represent the normal quality level of the products, after the screening process, the data remaining in the threshold range is the preferred quality feedback value, which can more accurately reflect the actual situation of the product quality.

[0109] Referring to Figure 4 The welding structure smoothness is determined, and structure abnormal data is generated, and specifically includes:

[0110] The product area structure information is analyzed to obtain the position tolerance of each product structure area;

[0111] According to the functional positioning of each product structure area in the industrial product, the product structure areas with the same functional positioning are summarized to generate the same type of structure area;

[0112] The position tolerances of the same type of structure area are compared in sequence to obtain the same type of position tolerance difference value;

[0113] The average of each same type of position tolerance difference value is calculated, and the same type of standard deviation data is obtained by analyzing the average;

[0114] The same type of standard deviation data is analyzed to obtain the median standard deviation value of the same type of standard deviation data, and the same type of standard deviation data is analyzed based on the median standard deviation value to generate data deviation degree;

[0115] The welding structure smoothness is determined according to the data deviation degree;

[0116] The structure judgment threshold is generated according to the production and manufacturing plan and the product structure threshold;

[0117] Based on the structure determination threshold, the data deviation degree and the welding structure smoothness are analyzed, the abnormal data points are screened out, the data deviation degree and the welding structure smoothness corresponding to the abnormal data points are arranged and marked, and the structure abnormal data is generated.

[0118] It can be understood that when analyzing the product area structure information, first, the design drawings, technical documents and production process description materials of the product are comprehensively collected, which contain the detailed design requirements and size information of the product structure. Then, according to the relevant industry standards and tolerance specifications, the positional relationship and size parameters of each product structure area in the ideal state are determined. Next, professional measuring tools such as technical laser trackers are used to accurately measure each structure area of the actual produced product to obtain the actual position data. The actual position data obtained by measurement is compared with the ideal position required by the design, and the difference between the two is calculated. These differences are the positional tolerances of each product structure area. The standard deviation values in the same type of standard deviation data are subtracted from the median standard deviation value to generate deviation data, and the deviation data is absolute value processed to eliminate the influence of positive and negative directions. The deviation data after absolute value processing is divided by the median standard deviation value in turn to generate data deviation degree. The data deviation degree reflects the dispersion of the welding related data, which is closely related to the welding structure smoothness. When the data deviation degree is low, it means that the weld formation is more uniform, and the surface of the welding structure also has smaller fluctuations, thereby showing higher smoothness. On the contrary, if the data deviation degree is high, it means that the welding parameters fluctuate greatly, and the weld has problems such as different widths, obvious fluctuations, etc. The smoothness of the welding structure is low. By analyzing the historical quality feedback values of industrial products and the product feature structure parameters corresponding to each quality feedback value, historical welding sample data is obtained. The quantitative relationship between data deviation degree and welding structure smoothness is obtained through statistical methods. According to the quantitative relationship, the data deviation degree is analyzed to generate the welding structure smoothness. The production and manufacturing plan clearly defines the quality standards of product production, and the product structure threshold value defines the acceptable upper and lower limit ranges of the size, shape and performance of each structure of the product. In-depth analysis of the process capability in the production and manufacturing plan, such as machining precision and assembly error range, and combining with the specific requirements of the product structure threshold value for each structure, the fluctuations generated by the production process and the allowable deviation of the product structure are comprehensively considered. According to the function importance and stress characteristics of different structure areas of the product, the corresponding determination threshold is determined. According to the structure determination threshold, the data deviation degree and the welding structure smoothness are analyzed. First, the data deviation degree is compared with the deviation degree threshold in the structure determination threshold. If the data deviation degree exceeds the threshold, it means that the data dispersion degree is abnormal, and the corresponding welding process is unstable. Then, the welding structure smoothness is compared with the smoothness threshold. If the smoothness is lower than the standard, it means that the surface of the welding structure is not flat, and there is an anomaly. At the same time, if both of these two abnormal conditions are met, or one of them is abnormal and it is determined through further analysis that the data point has a significant impact on the welding quality, it is an abnormal data point. These abnormal data points reveal the problems in the welding process, such as equipment failure and operation error. They can provide a basis for subsequent targeted troubleshooting and problem solving.

[0119] Referring toFigure 5 As shown in FIG. 1, determining the production line abnormality threshold value, specifically comprising:

[0120] Analyzing the production line task time sequence to obtain historical production line data, the historical production line data including the task type of different production lines in each time period, the production line historical production efficiency and the production line equipment historical running state;

[0121] According to the task type of different production lines in each time period, analyzing the production line historical production efficiency to generate the production efficiency corresponding to each time period;

[0122] According to the production line task time sequence, analyzing the production line equipment historical running state to generate the equipment running fluctuation period data;

[0123] Arranging and analyzing the production efficiency corresponding to each time period respectively from high to low to generate the upper quartile and the lower quartile, and taking the difference between the upper quartile and the lower quartile as the second safety margin;

[0124] Taking the sum of the upper quartile and the second safety margin as the upper limit of the production line efficiency, and taking the difference between the lower quartile and the second safety margin as the lower limit of the production line efficiency, to determine the production line abnormality threshold value.

[0125] It can be understood that in the production of automobile parts, the task type of different production lines in different time periods, the production line historical production efficiency and the production line equipment historical running state are different, mainly due to the following reasons. From the task type, the automobile production plan will be adjusted according to the market order and the vehicle model update plan, such as before the new vehicle model is put on the market, the interior production line will change from producing old interior parts to producing new interior parts, and the engine production line will adjust the production task according to the demand change of different vehicle engines. In terms of production line historical production efficiency, the training and proficiency of new employees will have an impact. New employees are less efficient when they first start work, and their efficiency will gradually improve with experience. In addition, process improvement will continuously improve production efficiency. For example, the optimization of welding process will greatly improve the production efficiency of the vehicle body production line. In terms of production line equipment historical running state, some old stamping equipment will have precision decline and fault increase due to long-term use, while new equipment will run more stably. And the production task amount is different in different time periods, and the frequency of use of the equipment is also different. High frequency of use will increase the wear and failure probability of the equipment, resulting in differences in equipment running state.

[0126] Referring to Figure 6 As shown in FIG. 1, generating surface fitting data, combining product structure feature parameters to generate surface offset threshold value, specifically comprising:

[0127] Based on the laser tracker, the intersection points of each product structure region are taken as the reference to measure the appearance angle of adjacent product structure regions to obtain surface angle feature values;

[0128] induction mapping of the surface angle characteristic values of each product structure region, generating a position-angle characteristic value curve, and deriving a curve ratio;

[0129] arranging and sorting the curve ratio, marking the abnormal curve ratio, and obtaining the surface gray value of the position corresponding to the abnormal curve ratio;

[0130] analyzing the surface gray value of the product structure region containing the position corresponding to the abnormal curve ratio, and generating surface fitting data;

[0131] determining the surface standard fitting value according to the product structure characteristic parameters;

[0132] analyzing the surface fitting data according to the surface standard fitting value, and generating a surface offset threshold.

[0133] It can be understood that in the monitoring link of industrial products, high-precision appearance angle measurement can be realized with the help of a laser tracker. First, accurately position the intersection points of each product structure region. These intersection points are the key parts of product structure connection and transition, and have a major impact on the overall performance and appearance quality of the product. Take these intersection points as the reference, adjust the laser tracker to the appropriate position and calibrate it to ensure the accuracy of the measurement. Then, for adjacent product structure regions, the laser tracker emits a laser beam that shines on the surface of the adjacent structure region. Use the built-in angle measurement system to accurately capture the laser reflection signal and calculate the spatial angle of the adjacent product structure region relative to the intersection point in real time. After measuring multiple different positions, collect and analyze these data to obtain comprehensive and accurate surface angle characteristic values. These characteristic values reflect the connection angle relationship between adjacent product structure regions. By induction and sorting of the surface angle characteristic values of each product structure region, a spatial coordinate system with the intersection point as the origin can be generated. By taking a plane of the spatial coordinate system, the surface characteristic angle values of each position can be obtained. By sorting the surface characteristic angle values of each position, a position-angle characteristic value curve can be generated. The position value can be mapped to a number by extending the Cantor pairing function. The product structure has continuity, so the curve ratio will change according to the increasing or decreasing rule. Take the surface standard fitting value as the reference, and sequentially subtract the surface fitting data from the surface standard fitting value to generate a surface offset threshold.

[0134] Further, as shown in Figure 7 proposed an intelligent manufacturing abnormality monitoring system based on multi-source data, which is suitable for the monitoring method described above, and is characterized by specifically comprising:

[0135] The acquisition module is used for acquiring production manufacturing plans and manufacturing process real-time data, collecting product area structure information, collecting industrial product historical quality feedback values, and transmitting the obtained data to the analysis module and the monitoring module.

[0136] The analysis module is used for determining product feature structure parameters according to manufacturing product features, analyzing the product feature structure parameters, determining product structure threshold values, generating structure abnormality data through position tolerances, determining production line abnormality threshold values through the structure abnormality data and production line task time sequences, generating surface offset threshold values according to appearance angles and surface gray value, and transmitting the determined and generated data to the monitoring module and the management module.

[0137] The monitoring module is used for classifying and arranging the received data, classifying the data transmitted by the acquisition module and the data transmitted by the analysis module, comparing the corresponding relationship between the two, determining whether an abnormal situation occurs in the intelligent manufacturing process, and transmitting the data to the management module.

[0138] The management module is used for arranging the received data and performing real-time display of the intelligent manufacturing process according to the arranged results.

[0139] Preferably, the acquisition module specifically comprises:

[0140] The first acquisition unit is used for acquiring production manufacturing plans, determining industrial products, manufacturing product features and production line task time sequences, and transmitting the data to the analysis module.

[0141] The second acquisition unit is used for determining the position tolerance of the product through a three-coordinate measuring instrument, collecting appearance angles through a laser tracker, collecting surface gray values of the product through an industrial camera, and transmitting the data to the analysis module.

[0142] The third acquisition unit is used for collecting industrial product historical quality feedback values and transmitting the data to the analysis module.

[0143] Preferably, the analysis module specifically comprises:

[0144] The first analysis unit is used for performing edge processing on the industrial product according to the manufacturing product features, generating structure edges, performing contour extraction, generating various product structure areas, performing stress analysis on the industrial product according to the functional positioning of the various product structure areas in the industrial product, determining product feature structure parameters, and transmitting the data to the monitoring module.

[0145] The second analysis unit is used for analyzing the historical quality feedback value of the industrial product, generating quality statistical feature data, determining the first safety margin and the standard structure parameter according to the quality statistical feature data, further generating the product structure threshold value, and transmitting data to the monitoring module.

[0146] The third analysis unit is used for analyzing the historical operation state of the production line equipment according to the production line task time sequence, generating equipment operation fluctuation period data and corresponding production efficiency of each time period of the production line, arranging the corresponding production efficiency of each time period of the production line, determining the production line abnormal threshold value, and transmitting data to the monitoring module.

[0147] Preferably, the monitoring module specifically comprises:

[0148] The first monitoring unit is used for analyzing the data deviation degree and the welding structure smoothness according to the structure judgment threshold value, screening out abnormal data points, and transmitting data to the management module.

[0149] The second monitoring unit is used for comparing the product structure value in the manufacturing process real-time data with the product structure threshold value as the reference, judging whether there is abnormal data value, and transmitting data to the management module.

[0150] The third monitoring unit is used for comparing the production line real-time value in the manufacturing process real-time data with the production line abnormal threshold value as the reference, judging whether there is abnormal data value, and transmitting data to the management module.

[0151] The fourth monitoring unit is used for comparing the surface flatness value in the manufacturing process real-time data with the surface offset threshold value as the reference, judging whether there is abnormal data value, and transmitting data to the management module.

[0152] In summary, the advantages of the present application are that the adaptability of adjusting the monitoring strategy and parameters according to the changes of the actual working conditions can be effectively improved, the occurrence of false positives or false negatives is reduced, the difficulty of accurately diagnosing some potential and deep-seated fault causes is reduced, potential quality risks are reduced, and the overall productivity of the intelligent manufacturing process is fully utilized.

[0153] The basic principles, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection claimed by the present application is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring abnormalities in intelligent manufacturing based on multi-source data, characterized in that: include: Obtain and analyze production and manufacturing plans to determine industrial products, manufacturing product characteristics, and production line task time series; Analyze industrial products according to the characteristics of manufactured products, generate product structure areas, and determine product characteristic structure parameters in combination with production and manufacturing plans; Analyze product characteristic structure parameters and determine product structure thresholds; Collecting product area structure information, wherein the product area structure information includes position tolerance, appearance angle, and surface grayscale value; Determine the smoothness of the weld structure based on position tolerance and generate structural anomaly data; Determine the production line anomaly threshold based on structural anomaly data and production line task time series; Generate surface fitting data based on appearance angle and surface grayscale value, and generate surface offset threshold based on product structural characteristic parameters; Acquire real-time data of the manufacturing process and compare it in sequence based on product structure thresholds, production line anomaly thresholds, and surface deviation thresholds to achieve monitoring.

2. The method for monitoring abnormalities in intelligent manufacturing based on multi-source data according to claim 1, characterized in that: The determination of product characteristic structural parameters in combination with the production and manufacturing plan specifically includes: Analyze industrial products and perform marginalization based on the characteristics of manufactured products to generate structural edges; Extract the contours of industrial products based on the structural edge and generate the structural areas of each product; Analyze the production and manufacturing plan to determine the functional positioning of each product structure area in the industrial product and the actual application scenarios of the industrial product; According to the actual application scenarios of industrial products, scenario data is collected and scenario models are generated based on digital modeling technology; Based on virtual prototype technology and scenario models, the stress analysis of industrial products is carried out according to the functional positioning of each product structure area in the industrial product, and the stress analysis results of industrial products are generated; The stress analysis results of industrial products are used as the boundary conditions of the structural edge, and each product structural area is iterated to generate the stress results of the structural area; According to the relevant theories of structural mechanics and material mechanics, combined with the stress results of the structural area, the characteristic structural parameters of the product are determined.

3. The method for monitoring abnormalities in intelligent manufacturing based on multi-source data according to claim 2, characterized in that: Determining the product structure threshold specifically includes: Collect historical quality feedback values ​​of industrial products and product characteristic structure parameters corresponding to each quality feedback value, analyze the historical quality feedback values ​​of industrial products, and generate quality statistical characteristic data, wherein the quality statistical characteristic data includes the distribution range of quality-related parameters, the mean of quality-related parameters, and the standard deviation of quality-related parameters; Preprocess the historical quality feedback values ​​of industrial products according to the standard deviation of quality-related parameters to generate optimal quality feedback values; Summarize and organize the preferred quality feedback values, arrange them in descending order, and generate quality feedback sequence data; According to the distribution range of quality-related parameters, the quality feedback sequence data is analyzed to obtain the data in the upper half of the distribution interval; According to the product characteristic structure parameters corresponding to each quality feedback value, the upper half distribution interval data is analyzed to obtain the maximum characteristic structure parameter value and the minimum characteristic structure parameter value, and the difference between the maximum characteristic structure parameter value and the minimum characteristic structure parameter value is used as the first safety margin; Combine the product characteristic structure parameters corresponding to each quality feedback value and the mean value of the quality-related parameters to obtain the product characteristic structure parameters corresponding to the mean value of the quality-related parameters, and record them as standard structure parameters; The standard structural parameters are added to the first safety margin to generate the product structure threshold.

4. The method for monitoring abnormalities in intelligent manufacturing based on multi-source data according to claim 3, characterized in that: The determining of the smoothness of the welding structure and generating the structural abnormality data specifically includes: Analyze product area structure information to obtain the position tolerance of each product structure area; According to the functional positioning of each product structure area in industrial products, the product structure areas with the same functional positioning are summarized to generate the same type of structure areas; Compare the position tolerances of the same type of structural areas in sequence to obtain the position tolerance difference values ​​of each same type; Calculate the average value of the position tolerance difference of the same type, and obtain the standard deviation data of the same type by analyzing the average value; Analyze the same type of standard deviation data, obtain the median standard deviation value of the same type of standard deviation data, and analyze the same type of standard deviation data based on the median standard deviation value to generate data deviation degree; Determine the smoothness of the welding structure based on the data deviation; Generate structure determination thresholds based on production and manufacturing plans and product structure thresholds; Based on the structural judgment threshold, the data deviation and welding structure smoothness are analyzed to filter out abnormal data points. The data deviation and welding structure smoothness corresponding to the abnormal data points are sorted and marked to generate structural abnormality data.

5. The method for monitoring abnormalities in intelligent manufacturing based on multi-source data according to claim 4, characterized in that: Determining the production line abnormality threshold specifically includes: Analyze the time series of production line tasks to obtain historical production line data, including the task types of different production lines in each time period, the historical production efficiency of the production lines, and the historical operating status of the production line equipment; Analyze the historical production efficiency of production lines based on the task types of different production lines in each time period and generate the corresponding production efficiency of production lines in each time period; Analyze the historical operating status of production line equipment based on the production line task time series and generate equipment operation fluctuation cycle data; The production efficiency of the production lines in each period is sorted and analyzed from high to low to generate the upper quartile and lower quartile. The difference between the upper quartile and the lower quartile is recorded as the second safety margin. The sum of the upper quartile and the second safety margin is taken as the upper limit of the production line efficiency, and the difference between the lower quartile and the second safety margin is taken as the lower limit of the production line efficiency to determine the production line abnormality threshold.

6. The method for monitoring abnormalities in intelligent manufacturing based on multi-source data according to claim 5, characterized in that: The generating of surface fitting data and the generating of surface offset threshold value in combination with product structural characteristic parameters specifically include: Based on the laser tracker, the intersection points of each product structure area are used as the reference to measure the appearance angles of adjacent product structure areas and obtain the surface angle characteristic values; The surface angle characteristic values ​​of each product structure area are summarized and mapped to generate a position-angle characteristic value curve and obtain the curve ratio; Arrange and sort the curve ratios, mark abnormal curve ratios, and obtain the surface grayscale value of the position corresponding to the abnormal curve ratio; Analyze the surface grayscale value of the product structure area corresponding to the position of the abnormal curve ratio to generate surface fitting data; Determine the surface standard fitting value based on the product structural characteristic parameters; The surface fitting data is analyzed according to the surface standard fitting value to generate a surface deviation threshold.

7. An intelligent manufacturing anomaly monitoring system based on multi-source data, applicable to the monitoring method according to any one of claims 1 to 6, characterized in that: Specifically include: An acquisition module, which is used to obtain real-time data on production and manufacturing plans and manufacturing processes, collect product regional structure information, and collect historical quality feedback values ​​of industrial products. The real-time manufacturing process data includes product structure values, production line real-time values, and surface flatness values, and transmit the obtained data to the analysis module and the monitoring module; An analysis module, the analysis module being used to determine product characteristic structural parameters based on manufactured product characteristics, analyze the product characteristic structural parameters, determine product structural thresholds, generate structural anomaly data based on position tolerances, determine production line anomaly thresholds based on the structural anomaly data and production line task time series, generate surface offset thresholds based on appearance angles and surface grayscale values, and transmit the determined and generated data to the monitoring module and the management module; The monitoring module is used to classify and organize the received data, classify the data transmitted by the acquisition module into one category, classify the data transmitted by the analysis module into another category, compare the two according to the corresponding relationship between them, determine whether there is any abnormality in the intelligent manufacturing process, and transmit the data to the management module; The management module is used to organize the received data and display the intelligent manufacturing process in real time based on the organized results.

8. The intelligent manufacturing anomaly monitoring system based on multi-source data according to claim 7 is characterized in that: The acquisition module specifically includes: a first acquisition unit, which is used to obtain the production and manufacturing plan, determine the characteristics of industrial products, manufactured products and the time sequence of production line tasks, and transmit the data to the analysis module; a second acquisition unit, configured to determine the position tolerance of the product using a three-dimensional coordinate measuring machine, acquire the appearance angle using a laser tracker, acquire the surface grayscale value of the product using an industrial camera, and transmit the data to the analysis module; The third collection unit is used to collect historical quality feedback values ​​of industrial products and transmit the data to the analysis module.

9. The intelligent manufacturing anomaly monitoring system based on multi-source data according to claim 8, characterized in that: The analysis module specifically includes: a first analysis unit, configured to perform edge processing on the industrial product according to the characteristics of the manufactured product, generate structural edges, perform contour extraction, generate structural regions of each product, perform stress analysis on the industrial product according to the functional positioning of each structural region of the product in the industrial product, determine characteristic structural parameters of the product, and transmit the data to the monitoring module; a second analysis unit, the second analysis unit being configured to analyze historical quality feedback values ​​of industrial products, generate quality statistical characteristic data, determine a first safety margin and standard structural parameters based on the quality statistical characteristic data, and further generate a product structure threshold value, and transmit the data to the monitoring module; The third analysis unit is used to analyze the historical operating status of the production line equipment according to the production line task time series, generate equipment operation fluctuation cycle data and the corresponding production efficiency of the production line in each time period, arrange the corresponding production efficiency of the production line in each time period, determine the production line abnormality threshold, and transmit the data to the monitoring module.

10. The intelligent manufacturing anomaly monitoring system based on multi-source data according to claim 9, characterized in that: The monitoring module specifically includes: a first monitoring unit, configured to analyze data deviation and welding structure smoothness according to a structural determination threshold, screen out abnormal data points, and transmit the data to a management module; a second monitoring unit, the second monitoring unit being configured to compare the product structure value in the real-time manufacturing process data with the product structure threshold value, determine whether there is an abnormal data value, and transmit the data to the management module; a third monitoring unit, configured to compare the production line abnormality threshold with the production line real-time value in the manufacturing process real-time data, determine whether there is an abnormal data value, and transmit the data to the management module; The fourth monitoring unit is used to compare the surface flatness value in the real-time data of the manufacturing process with the surface deviation threshold as a reference, determine whether there is an abnormal data value, and transmit the data to the management module.

Images