Method and system for on-line prediction of skin forming quality for automotive applications

Abstract

This invention belongs to the field of automotive skin continuous forming quality prediction technology, and discloses an online prediction method and system for automotive skin forming quality. The method acquires online operating condition data and downstream quality inspection data for each forming section on the automotive skin continuous forming production line. Based on the corresponding transmission path length parameters, linear velocity parameters in the online operating condition data, and section operating state parameters, it determines the correction amount for the material segment migration position of each forming section, constructs the equivalent migration length of each forming section, and establishes a segmented equivalent migration correspondence between the quality inspection data and the upstream material segment position through reverse cumulative backtracking. Combining state change trigger information, it determines the operating condition boundary, forming residual influence segments and core influence segments. It extracts and fuses attenuation features and steady-state features respectively to generate prediction results and outputs risk warning information or process adjustment instructions, improving the accuracy of the prediction and the targeted nature of production line adjustments.

Description

Technical Field

[0001] This invention belongs to the field of continuous forming quality prediction technology for automotive skins, specifically a method and system for online prediction of automotive skin forming quality. Background Technology

[0002] In the continuous production process of automotive skins, they typically undergo a series of processes including hot pressing, hot air treatment, cooling and shaping, and traction output to achieve the desired surface texture, thickness, and surface quality. To ensure product consistency, production lines usually collect data online on operating parameters such as temperature, tension, speed, pressure, airflow, and cooling status. Downstream, quality inspection stations are set up for thickness detection, surface image detection, and surface temperature detection to monitor and assess the condition of the formed skin.

[0003] Existing automotive skin quality monitoring or early warning solutions typically involve statistical analysis of upstream operating conditions based on downstream inspection results, or aligning operating condition data and quality inspection data with time delays based on production line speed and workstation distance to make quality judgments. While these methods are applicable to general continuous conveying scenarios, they still have the following problems in continuous forming scenarios for automotive skins:

[0004] Inaccurate material segment correspondence: During the hot pressing, hot air action, and cooling and shaping processes of automotive skin, factors such as heat-induced length changes, speed differences between adjacent transmission or traction parts, and cooling retraction can easily affect the material segment's migration along the production line, causing it to not always satisfy a simple linear displacement relationship. If the existing solution only calculates time delays based on linear velocity and path length, it can easily lead to a mismatch between downstream quality inspection results and upstream actual operating conditions, thus affecting the accuracy of subsequent predictions.

[0005] The impact of operating condition switching is difficult to separate: In actual production, parameters such as temperature, air volume, traction speed, and pressure are adjusted according to changes in operating conditions. The impact of such adjustments on skin quality is often delayed and residual; a certain quality result detected downstream may be affected by both steady-state and transitional operating conditions simultaneously. Existing solutions often use fixed time windows to extract operating condition segments, which easily mixes the operating history under different states, leading to distortion of the formed state characteristics and making it difficult to accurately reflect the true forming state of the corresponding material segment.

[0006] Insufficient connection between prediction results and production line adjustments: Existing technologies mostly focus on judging or alarming the quality results themselves, lacking further identification of the sources of process impact corresponding to quality risks. It is difficult to effectively match prediction results with specific forming sections and adjustment directions, resulting in insufficient guidance for actual production line adjustments from online prediction results. Summary of the Invention

[0007] The purpose of this invention is to provide an online prediction method and system for the forming quality of automotive skins, so as to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: an online prediction method and system for the forming quality of automotive skins, comprising the following steps:

[0009] S1. Obtain online operating condition data of each forming section on the continuous forming production line for automotive skin, as well as quality inspection data output by the quality inspection station located downstream of each forming section.

[0010] S2. For each forming section, based on the corresponding transmission path length parameter and the linear velocity parameter and section running status parameter in the online working condition data, determine the correction amount of each forming section to the material segment migration position, determine the equivalent migration length of the corresponding forming section based on each correction amount, and perform reverse accumulation and backtracking of each equivalent migration length according to the order of each forming section in the production line, so as to establish the segmented equivalent migration correspondence between the quality inspection data and the upstream material segment position;

[0011] S3. Based on the segmented equivalent migration correspondence and combined with the state change triggering information in the online working condition data, determine the working condition state boundary corresponding to each of the quality detection data, and use the working condition state boundary as the boundary to form a residual influence segment in the transition influence interval and a core influence segment in the steady state influence interval.

[0012] S4. Extract steady-state features from the core influence segment, extract attenuation features from the residual influence segment, and form state features characterizing the forming state of the corresponding material segment based on the steady-state features and attenuation features.

[0013] S5. Generate a forming quality prediction result for the corresponding material segment based on the state characteristics. The forming quality prediction result includes at least one of the following: the type of influence of the associated process, the quality risk level, and the quality deviation trend.

[0014] S6. Output quality risk warning information or process adjustment instruction information based on the molding quality prediction results.

[0015] Preferably, the forming section includes at least two of the following: a hot pressing forming section, a hot air action section, a cooling and shaping section, and a traction output section.

[0016] Preferably, the online operating condition data includes at least one of temperature parameters, tension parameters, speed parameters, pressure parameters, roll gap parameters, air volume parameters, and cooling parameters; the quality inspection data includes at least one of thickness inspection data, surface image data, and surface temperature data.

[0017] Preferably, the correction amount in step S2 includes at least one of temperature correction component, speed difference correction component, and cooling shrinkage correction component;

[0018] The temperature correction component is used to characterize the effect of thermally induced length change of the material segment in the hot pressing section or hot air action section; the speed difference correction component is used to characterize the effect of relative slippage between adjacent transmission or traction parts of the material segment; and the cooling shrinkage correction component is used to characterize the effect of length retraction of the material segment in the cooling and shaping section.

[0019] Preferably, the state change triggering information includes at least one of the following: the parameter change amplitude exceeds a preset threshold, the parameter change rate exceeds a preset threshold, and a process switching command is received.

[0020] The working condition boundary is determined by mapping the moment when the state change trigger information is satisfied to the position of the upstream material segment.

[0021] Preferably, the transition influence interval starts from the boundary of the operating condition and ends when the parameter change amplitude corresponding to the online operating condition data is lower than the steady-state threshold for several consecutive sampling periods;

[0022] The steady-state influence range is located after the transitional influence range;

[0023] The residual impact segment consists of online operating condition data within the transitional impact range, while the core impact segment consists of online operating condition data within the steady-state impact range.

[0024] Preferably, the steady-state characteristics include at least one characteristic quantity that characterizes the operating condition level, the operating condition fluctuation state, and the operating condition change trend;

[0025] The attenuation characteristics include at least one of the following features: the intensity of the trailing effect, the attenuation rate of the trailing effect, and the degree of influence of the boundary distance.

[0026] In step S4, the attenuation feature is obtained by weighting the boundary distance between the material segment position corresponding to each working condition data in the residual influence segment and the working condition boundary and the parameter fluctuation amplitude.

[0027] Preferably, the associated process influence types include at least one of hot pressing influence, hot air effect influence, and cooling and shaping influence;

[0028] The process adjustment instruction information includes at least one of the target forming section identifier and the corresponding working condition parameter adjustment direction.

[0029] An online prediction system for automotive skin forming quality includes:

[0030] The data acquisition module is used to acquire online operating condition data of each forming section on the continuous forming production line for automotive skin, as well as quality inspection data output from the quality inspection station located downstream of each forming section.

[0031] The corresponding module is used to determine the correction amount of each forming section to the material segment migration position based on the corresponding transmission path length parameter, the linear velocity parameter and the section running status parameter in the online working condition data, for each forming section, determine the equivalent migration length of the corresponding forming section based on the correction amount, and perform reverse accumulation and backtracking of each equivalent migration length according to the order of each forming section in the production line, so as to establish the segmented equivalent migration correspondence between the quality inspection data and the upstream material segment position;

[0032] The boundary segment construction module is used to determine the working state boundary corresponding to each of the quality detection data based on the segmented equivalent migration correspondence and combined with the state change triggering information in the online working condition data. The module uses the working condition boundary as the boundary to form a residual influence segment in the transition influence interval and a core influence segment in the steady state influence interval.

[0033] The feature processing module is used to extract steady-state features from the core influence segment, extract attenuation features from the residual influence segment, and form state features characterizing the forming state of the corresponding material segment based on the steady-state features and the attenuation features.

[0034] The prediction module is used to generate a prediction result of the forming quality of the corresponding material segment based on the state characteristics;

[0035] The output module is used to output quality risk warning information or process adjustment instruction information based on the forming quality prediction result.

[0036] Preferably, the corresponding module is used to determine the correction amount for each forming section based on at least one of temperature correction component, speed difference correction component, and cooling shrinkage correction component;

[0037] The boundary segment construction module is used to determine the operating condition boundary based on at least one of parameter change amplitude, parameter change rate, or process switching command.

[0038] The output module is used to output process adjustment indication information, including the target forming section identifier and the corresponding working condition parameter adjustment direction.

[0039] The beneficial effects of this invention are as follows:

[0040] This invention determines the correction amount of each forming section to the material segment migration position by combining the transmission path length parameter of each forming section with the linear velocity parameter and section operation status parameter in the online working condition data. Based on the correction amount of each forming section, the corresponding equivalent migration length is constructed. By backtracking the equivalent migration length of each forming section, a segmented equivalent migration correspondence between quality inspection data and upstream material segment position is established, so that the downstream quality results can correspond to the upstream material segment working condition history that is closer to the actual forming process.

[0041] Meanwhile, by determining the working condition boundary corresponding to the quality inspection data, and forming residual influence segments within the transitional influence interval and core influence segments within the steady-state influence interval, the steady-state process influence and transitional tailing influence are distinguished and processed to reduce the interference of historical mixing of different working conditions on the construction of state characteristics.

[0042] Furthermore, based on the status characteristics, a forming quality prediction result is generated, which includes the type of influence of related processes, the level of quality risk, and the trend of quality deviation. Quality risk warning information or process adjustment instructions are output, thereby improving the pertinence and usability of the online prediction result for production line adjustment. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the system structure of the present invention;

[0044] Figure 2 This is a schematic diagram of the production line and testing station layout of the present invention;

[0045] Figure 3 This is a flowchart of the method of the present invention;

[0046] Figure 4 This is a schematic diagram of the segmented equivalent migration and boundary division of the present invention. Detailed Implementation

[0047] The present invention will be further described below with reference to the accompanying drawings and embodiments. It should be understood that the following embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. Equivalent substitutions or conventional adjustments made by those skilled in the art to the sequence of related steps, parameter selection, module connection methods, and data processing methods without departing from the concept of the present invention should all fall within the scope of protection of the present invention.

[0048] The present invention proposes an online prediction method and system for automotive skin forming quality, applicable to continuous forming production lines for automotive skins. Automotive skins can be PVC automotive skins, PU synthetic skins, TPO-type automotive skins, and other flexible decorative materials with continuous strip forms. These materials typically undergo hot pressing, hot air treatment, cooling and shaping, and traction output processes sequentially during continuous processing. Downstream, a quality inspection station is set up to acquire at least one of thickness detection data, surface image data, and surface temperature data for online quality inspection of the formed skin. Addressing issues such as material segment migration distortion, delayed impact of operating condition switching, and insufficient connection between quality prediction and production line adjustment in this type of continuous forming scenario, the present invention establishes a segmented equivalent migration correspondence between quality inspection data and upstream material segment positions. It combines this with operating condition boundaries to form residual and core influence segments. Based on this, state features are constructed, and quality risk warning information or process adjustment instructions are output, thereby achieving online prediction of automotive skin forming quality.

[0049] Example 1:

[0050] like Figure 1 and Figure 2 As shown, the online prediction system for automotive skin forming quality includes an acquisition module, a corresponding module, a boundary segment construction module, a feature processing module, a prediction module, and an output module.

[0051] The acquisition module is used to obtain online operating condition data of each forming section on the continuous forming production line for automotive skin, as well as quality inspection data output from the quality inspection station located downstream of each forming section. A corresponding module, connected to the acquisition module, receives the online operating condition data and quality inspection data output by the acquisition module. Based on the transmission path length parameters of each forming section, the linear velocity parameters in the online operating condition data, and the section operating status parameters, it determines the correction amount for the material segment migration position of each forming section, thereby establishing a segmented equivalent migration correspondence between the quality inspection data and the upstream material segment position. A boundary segment construction module, connected to the corresponding module, determines the operating condition state boundary corresponding to each quality inspection data based on the established segmented equivalent migration correspondence and the state change trigger information in the online operating condition data. It then uses the operating condition state boundary as the boundary to form residual influence segments within the transitional influence interval and core influence segments within the steady-state influence interval. A feature processing module, connected to the boundary segment construction module, extracts steady-state features from the core influence segments and attenuation features from the residual influence segments. Based on the steady-state and attenuation features, it forms state features characterizing the forming state of the corresponding material segment. The prediction module is connected to the feature processing module and is used to generate a prediction result for the forming quality of the corresponding material segment based on the state characteristics. The output module is connected to the prediction module and is used to output quality risk warning information or process adjustment instructions based on the forming quality prediction result.

[0052] Optionally, the acquisition module, corresponding module, boundary segment construction module, feature processing module, prediction module, and output module can be integrated into the same industrial control computer, edge computing device, or main control server; alternatively, they can be implemented collaboratively by an industrial controller, a host computer, and a field execution terminal. The acquisition module connects to the field detection unit via an industrial bus, Ethernet interface, analog acquisition interface, or digital acquisition interface; the corresponding module, boundary segment construction module, feature processing module, and prediction module interact with each other via a data bus or process call method; and the output module outputs the prediction results through a human-machine interface, alarm terminal, parameter suggestion interface, or an interface program with the production line control system.

[0053] In this embodiment, the connection between the acquisition module and the signal sources at each workstation can be achieved via industrial fieldbus, Ethernet, serial interface, analog input, or digital input. Optionally, temperature parameters are acquired through a contact temperature detection unit or infrared temperature measurement unit; tension parameters are acquired through a tension sensor; speed parameters are acquired through an encoder or speed measurement unit; pressure parameters are acquired through a pressure detection unit; roll gap parameters are acquired through a displacement detection unit or actuator feedback value; airflow parameters are acquired through an airflow detector; and cooling parameters are acquired through a cooling fan, cooling roller operating condition unit, or cooling medium temperature detection unit. Thickness detection data can be obtained through an online thickness gauge; surface image data can be obtained through a line scan or area array vision inspection unit; and surface temperature data can be obtained through an infrared temperature detection unit.

[0054] Through the above system structure, a complete data flow and control flow of "operating condition acquisition - position reconstruction - boundary recognition - feature processing - prediction output" can be formed. Moreover, it is not an isolated data calculation method, but can be directly connected to the online prediction system of continuous forming production line.

[0055] During production line operation, the acquisition module continuously acquires operating condition data of each section and downstream quality inspection data. The corresponding module continuously reconstructs the material segment position. The boundary segment construction module updates the state boundary when operating condition changes or parameter changes occur. The feature processing module forms the state features of the current material segment. The prediction module generates quality prediction results. The output module pushes risks and suggestions to the operation interface or control system in real time.

[0056] In one implementation, the specific sources of online operating condition data and quality inspection data are as follows.

[0057] The automotive skin continuous forming production line includes, along the material travel direction, at least two of the following: a hot pressing forming section, a hot air treatment section, a cooling and shaping section, and a traction output section. Optionally, in a simplified production line, it may include only the hot pressing forming section and the cooling and shaping section; in a more complete continuous processing production line, it may include four sections: a hot pressing forming section, a hot air treatment section, a cooling and shaping section, and a traction output section.

[0058] In the hot pressing section, the acquisition module collects at least one of the following as online working condition data: hot pressing temperature, rolling pressure, roll gap opening and linear speed.

[0059] In the hot air operating zone, the data acquisition module collects at least one of the hot air temperature, air volume, and linear velocity as online operating condition data.

[0060] In the cooling and shaping section, the acquisition module collects at least one of the following as online operating condition data: cooling roller temperature, cooling air temperature, cooling rate, and linear speed.

[0061] In the traction output section, the acquisition module collects at least one of the following as online operating condition data: traction speed, tension, and traction roller speed.

[0062] Optionally, the online operating condition data may include at least temperature and speed parameters; for example, only temperature and speed parameters may be collected, or temperature, tension and cooling parameters may be collected, or all data of temperature, tension, speed, pressure, roll gap, airflow and cooling parameters may be collected.

[0063] Downstream quality inspection stations are used to output quality inspection data. This data includes at least one of thickness measurement data, surface image data, and surface temperature data. Optionally, thickness measurement data is acquired by an online thickness gauge, surface image data is acquired by a line scan camera or an area array vision inspection unit, and surface temperature data is acquired by an infrared temperature measurement unit or a thermal imaging unit.

[0064] In one optional implementation, the quality inspection station is located downstream of the traction output section, and a clear path length relationship exists between the inspection station and each upstream forming section. This path length relationship can be measured by the equipment installation location or pre-stored in the system using equipment layout parameters. Optionally, the distance between the quality inspection station and the traction output section can be 0.5m, 1.5m, or 3.0m; these parameters correspond to shorter, medium, and longer inspection delay path layouts, respectively.

[0065] Furthermore, the first module used for location backtracking calculation Transmission path length parameter for each forming section The preferred definition is: the material segment in the first... The geometric length measured along the material travel path between the inlet and outlet reference positions of each forming section. The inlet and outlet reference positions can be determined by the centerline positions of adjacent rollers before and after the forming section, the tangent position of the guide roller, the measurement position of the tension roller, or the boundary position of the process section. The measurements can be obtained from the equipment installation drawings or by actual measurement along the material's center path while the equipment is stopped, and then stored in the system in advance.

[0066] Through the above data collection methods, the system can continuously obtain online operating condition data and downstream quality inspection data related to the skin forming process, providing a data foundation for establishing segmented equivalent migration correspondences; and the operating condition data and quality inspection data are implemented as real data sources that can be directly collected in the continuous forming production line, providing support for subsequent method steps.

[0067] The specific implementation method is as follows: Based on the length of the production line, the distance between workstations, and the conditions for deploying detection units, configure the data acquisition relationship between the forming section and the quality inspection workstation.

[0068] In one implementation, the segmented equivalent migration correspondence between quality inspection data and the location of upstream material segments is established as follows.

[0069] For quality inspection data output from the quality inspection station at a certain moment, the system does not simply backtrack from the inspection time. Instead, it first divides the migration process between the quality inspection station and the upstream material segment into several segments, each corresponding to a forming segment. For each forming segment, based on the corresponding transmission path length parameter, the linear velocity parameter in the online operating data, and the segment's operating status parameter, the correction amount for the material segment's migration position is determined. Then, the correction amounts are accumulated according to the order of each forming segment on the production line, thereby establishing a segmented equivalent migration correspondence between the quality inspection data and the position of the upstream material segment.

[0070] In the implementation where temperature correction component, speed difference correction component, and cooling shrinkage correction component are all involved in the calculation, the three correction components are not directly added together as position quantities at the same level. Instead, the equivalent migration length corresponding to each forming section is first determined, and then the equivalent migration length of each forming section is back-accumulated by the position of the quality inspection station to obtain the position of the upstream material section.

[0071] Specifically, no. Equivalent migration length of each forming section It can be represented as:

[0072]

[0073] in, For the first The geometric length measured along the material travel path between the inlet reference position and the outlet reference position of each forming section; For the first The equivalent increment caused by thermally induced length change within each forming section; For the first Compensation for belt directional slippage caused by the difference in speed between adjacent transmission or traction within each forming section; For the first The amount of shrinkage caused by cooling contraction within each cooling and shaping section.

[0074] Based on this, the upstream material section location It can be determined by the following formula:

[0075]

[0076] in, This refers to the location of the upstream material section. This is the location of the quality inspection station. The number of formed segments participating in the backtracking calculation.

[0077] Among them, the temperature correction component is used to characterize the effect of thermally induced length changes of the material segment in the hot pressing section or the hot air action section.

[0078] The coefficient of thermally induced length variation is The actual temperature is The reference temperature is Then the temperature correction component can be expressed as:

[0079]

[0080] in, This is used to characterize the degree of thermally induced length change of the material segment within the forming section relative to the reference operating condition.

[0081] The velocity difference correction component is used to characterize the relative slippage effect between adjacent transmission or traction sections of the material. Optionally, let the first... The geometric length measured along the material travel path between the inlet reference position and the outlet reference position of each forming section is: The linear velocity of the adjacent upstream section is The linear velocity of the current section or traction output section is The slip correction factor is Then the speed difference correction component can be expressed as:

[0082]

[0083] in, For directional compensation; when > hour, A positive value indicates an increase in the equivalent migration length of the material segment within the forming section; when... < hour, Taking a negative value indicates that the equivalent migration length of the material segment within the forming section is reduced.

[0084] Slip correction factor Used to characterize the The conversion ratio of the linear velocity difference within each forming section to the actual sliding displacement of the material section. Optionally, This can be determined through calibration: under reference material system and reference process conditions, at the [number]th [year]... Trackable markers are placed between the inlet and outlet reference positions of each forming section, or a vision inspection unit and an encoder are used to measure the actual sliding displacement of the material segment within that section. Synchronous data collection , and And calculate the displacement corresponding to the theoretical velocity difference:

[0085]

[0086] but It can be represented as:

[0087]

[0088] in, This represents the actual sliding displacement with direction. The displacement corresponds to the theoretical velocity difference with direction; the directions of both are defined as follows: The positive and negative directions remain consistent.

[0089] For the same material system, the same transmission structure, and the same tension range It can be pre-calibrated and stored in the system; it can be recalibrated when the material system, traction structure or tension window is changed.

[0090] The cooling shrinkage correction component is used to characterize the effect of length retraction of the material segment in the cooling and shaping section. Optionally, let the first... The geometric length measured along the material travel path between the inlet reference position and the outlet reference position of each cooling and shaping section is: The cooling shrinkage coefficient is The temperature difference before and after cooling is The cooling contraction correction component can then be expressed as:

[0091]

[0092] in, Taking a positive value represents the first The retraction amplitude within each cooling and shaping section; since retraction reduces the equivalent migration length of this section, therefore... The reduction of items is reflected in the middle.

[0093] In embodiments where only one or two of the temperature correction component, speed difference correction component, and cooling shrinkage correction component are used for calculation, the correction components not included in the calculation are not counted in the equivalent migration length of the corresponding forming section.

[0094] Optionally, the reference temperature of the hot pressing section can be 80℃, 110℃, or 140℃; the linear speed can be 5m / min, 15m / min, or 25m / min; and the temperature difference before and after cooling can be 10℃, 25℃, or 40℃. The above endpoint and intermediate values ​​are only used to illustrate the implementation under different operating conditions and are not intended to be a single limitation.

[0095] In this way, the quality inspection data output by the current quality inspection station can be mapped to the upstream material section position that is closer to the actual processing, thereby avoiding the mismatch problem caused by linear backtracking based solely on linear speed and station distance, and improving the accuracy of the correspondence between the downstream quality inspection results and the upstream actual working conditions.

[0096] The specific implementation method is as follows: after receiving a certain quality inspection data, the system backtracks and corrects it segment by segment according to the order of each forming section to obtain the position of the upstream material segment corresponding to the quality inspection data.

[0097] In one implementation, the method for determining the boundary of the operating condition and the method for constructing the residual influence segment and the core influence segment are as follows.

[0098] During continuous forming, when a state change trigger message appears in the online operating condition data, the system considers a change in operating condition to have occurred. Optionally, the state change trigger message includes at least one of the following: parameter change amplitude exceeding a preset threshold, parameter change rate exceeding a preset threshold, and receiving a process switching command. The parameter change amplitude threshold can optionally be 3%, 8%, or 15%; the parameter change rate threshold can optionally be 0.5% / s, 1.5% / s, or 3% / s.

[0099] When a state change trigger is detected, the starting position of the working condition boundary is determined by mapping the moment that triggers the state change to the upstream material segment position. After this starting position, the boundary segment construction module continuously monitors changes in the online working condition data. When the parameter change amplitude is below the steady-state threshold for several consecutive sampling periods, the end position of the transition effect is determined. Optionally, the consecutive sampling periods are 3, 5, or 8; the steady-state threshold is optionally 1%, 2%, or 5%.

[0100] The online operating condition data from the starting position of the operating condition boundary to the ending position of the transitional influence constitutes the residual influence segment; the online operating condition data within the steady-state influence range after the ending position of the transitional influence constitutes the core influence segment.

[0101] It should be noted that the residual impact fragments and core impact fragments are not directly divided according to the detection time. Instead, they are first constructed by combining the previously established segmented equivalent migration correspondence, mapping the online operating condition data and boundary positions to the corresponding upstream material segment positions, and then constructing the fragments. In this way, the resulting residual impact fragments and core impact fragments can more accurately correspond to the actual material segment operating condition history represented by the current quality inspection data.

[0102] The specific implementation method is as follows: when process parameters such as hot pressing temperature, air volume, linear velocity, and tension are switched, the system identifies the boundary of the working condition and extracts the residual influence segment and the core influence segment respectively.

[0103] In one implementation, the state features are constructed as follows.

[0104] For the core impact segment, the feature processing module extracts steady-state features. Optionally, the steady-state features include at least one feature quantity characterizing the operating condition level, operating condition fluctuation state, and operating condition change trend. For example, for temperature parameters, the average temperature, temperature standard deviation, and temperature change slope within the steady-state range can be extracted; for speed parameters, the average linear velocity, speed fluctuation coefficient, and steady-state deviation can be extracted; for pressure parameters or roll gap parameters, the average value and fluctuation range can be extracted. Thus, the operating condition data in the core impact segment can form steady-state features used to characterize the steady-state process effect.

[0105] For the residual effect segment, the feature processing module extracts attenuation features. Optionally, the attenuation features include at least one feature quantity characterizing the intensity of the tailing effect, the attenuation rate of the tailing effect, and the degree of influence of the boundary distance. The intensity of the tailing effect can be obtained from the average deviation of the parameters in the residual effect segment relative to the steady-state reference; the attenuation rate of the tailing effect can be obtained from the decreasing trend of the parameter deviation in the residual effect segment with the change of boundary distance; the degree of influence of the boundary distance can be obtained from the distance between each sampling point and the boundary of the operating condition. Thus, the operating condition data in the residual effect segment can form attenuation features to characterize the transition tailing effect.

[0106] To reflect the stronger residual effects closer to the boundary of the operating condition, the system assigns weights to each sampling point in the residual effect segment.

[0107] In one implementation, the residual effect fragment is the first The weighting value of each sampling point can be determined by the following formula:

[0108]

[0109] in, The first residual effect fragment The weighting value of each sampling point For the first The parameter fluctuation range at each sampling point For the first The boundary distance between the material segment location corresponding to each sampling point and the boundary of the working condition. This is the attenuation coefficient. Therefore, the sampling point with the larger the parameter fluctuation amplitude and the closer it is to the boundary of the operating condition, the higher its corresponding weighting value.

[0110] To facilitate subsequent feature fusion, the above weights can be normalized. The normalized weights can be determined by the following formula:

[0111]

[0112] in, The normalized weighted values, The first residual effect fragment The weighting value of each sampling point For cumulative indexing, The number of sampling points in the residual influence segment is denoted as . Based on the normalized weighted values, the deviation of each sampling point in the residual influence segment from the steady-state baseline and its variation trend along the boundary distance direction are weighted to obtain the attenuation characteristics.

[0113] After obtaining the steady-state characteristics and attenuation characteristics, the feature processing module fuses them to form state characteristics representing the forming state of the corresponding material segment. In one embodiment, the state characteristics can be determined by the following formula:

[0114]

[0115] in, As state characteristics, Steady-state features extracted from core influencing fragments, The attenuation features of residual influence fragment extraction, The fusion coefficient is used. The steady-state features extracted from the core influence fragments are used to characterize the steady-state process influence, while the attenuation features extracted from the residual influence fragments are used to characterize the transition tailing influence. The two are fused to form the state features of the corresponding material segment.

[0116] Optionally, the weighting criteria include boundary distance and parameter fluctuation amplitude. The closer the boundary distance and the greater the parameter fluctuation amplitude, the higher the weighting value corresponding to the sampling point.

[0117] Optionally, attenuation coefficient It can be any one of 0.2, 0.5, or 0.8; fusion coefficient It can be any one of 0.3, 0.5, or 0.7, corresponding to implementation methods that emphasize residual effects, moderate balance, and steady-state effects, respectively.

[0118] Through the above method, the state features are no longer a unified extraction result of the entire operating history, but can separately reflect the steady-state process influence and the transition tailing influence, thus more closely reflecting the actual quality formation mechanism in the continuous forming process of automotive skins. At the same time, it also avoids the mixing of influences from different stages caused by uniformly processing the entire operating history, allowing the transition tailing influence and steady-state process influence to be distinguished and expressed in the state features, thereby improving the accuracy of the state features in representing the actual forming state.

[0119] The specific implementation method is as follows: the system extracts features from the core influence segment and the residual influence segment respectively, then assigns weights to the residual influence segment according to the boundary distance and parameter fluctuation amplitude, and fuses them with the steady-state features of the core influence segment to form the state features of the corresponding material segment.

[0120] In one embodiment, the method for generating the forming quality prediction result and process adjustment instruction information is as follows.

[0121] After receiving the aforementioned formed state characteristics, the prediction module generates a forming quality prediction result. Optionally, the forming quality prediction result includes at least one of the following: the type of influence of the associated process, the quality risk level, and the quality deviation trend. To improve the guidance of the prediction result for production line adjustment, the prediction module not only determines the quality risk level of the corresponding material segment, but also identifies which forming segment the risk is more likely to originate from, and provides a corresponding deviation trend judgment.

[0122] Optionally, the prediction module can be implemented using a preset prediction model or a rule-based judgment unit. When using a preset prediction model, the state characteristics can be used as input to output the quality risk score of the corresponding material segment, and the quality risk level can be determined based on the quality risk score and a preset grading threshold. When using a rule-based judgment unit, the quality risk level, quality deviation trend, and associated process influence type of the corresponding material segment can be determined based on the comparison results between each feature quantity in the state characteristics and the preset threshold.

[0123] Optionally, the quality risk level can be divided into three, four, or five levels. For example, under a three-level classification, it can be divided into low risk, medium risk, and high risk; under a four- or five-level classification, the warning level can be further refined to adapt to implementation scenarios under different quality control requirements.

[0124] The quality deviation trend can be represented by at least one of the following: an increasing trend in thickness deviation, a fluctuating trend in surface flatness, an abnormal trend in surface temperature, or an abnormal trend in surface image. Optionally, the prediction module can determine the quality deviation trend based on the changing direction of state characteristics within multiple consecutive detection cycles; when the relevant feature quantity characterizing the degree of abnormality continuously increases, it can be determined as an increasing trend in the corresponding quality deviation; when the relevant feature quantity continuously decreases, it can be determined as a decreasing trend in the corresponding quality deviation; when the relevant feature quantity changes within a preset fluctuation range, it can be determined as a basically stable trend.

[0125] The types of associated process influences may include at least one of hot pressing influence, hot air effect influence, and cooling and shaping influence. Optionally, the prediction module can group the characteristic quantities corresponding to different forming sections in the state characteristics and compare the contribution of each group of characteristic quantities to the current quality anomaly characterization to determine the type of associated process influence. For example, when the state characteristics formed by parameters such as temperature, pressure, and roll gap related to the hot pressing section contribute more, it can be determined that the current forming quality prediction result is mainly related to the hot pressing influence; when the state characteristics formed by cooling parameters, tension parameters, or surface temperature parameters related to the cooling and shaping section contribute more, it can be determined that the current forming quality prediction result is mainly related to the cooling and shaping influence.

[0126] The output module outputs quality risk warning information or process adjustment instructions based on the forming quality prediction results. The quality risk warning information may include at least one of the following: abnormal material segment identifier, abnormal section identifier, corresponding quality risk level, and quality deviation trend. The process adjustment instructions may include the target forming section identifier and the corresponding adjustment direction of the operating parameters. Optionally, when the associated process influence type points to the hot pressing forming influence, process adjustment instructions are output for the hot pressing forming section; when the associated process influence type points to the hot air effect influence, process adjustment instructions are output for the hot air effect section; when the associated process influence type points to the cooling and shaping influence, process adjustment instructions are output for the cooling and shaping section. The process adjustment instructions are preferably directional prompts, and are not limited to a fixed value.

[0127] Furthermore, when the quality deviation trend manifests as an increasing thickness deviation and the associated process influence type points to the effect of hot pressing, adjustment prompts can be output in the direction of increasing the intensity of hot pressing, reducing the roll gap, or decreasing the throughput speed. When the quality deviation trend manifests as an abnormal surface temperature trend and the associated process influence type points to the effect of cooling and shaping, adjustment prompts can be output in the direction of increasing the intensity of cooling, optimizing the cooling rhythm, or adjusting the traction speed. When the quality deviation trend manifests as a fluctuating surface flatness trend and the associated process influence type points to the effect of hot air, adjustment prompts can be output in the direction of adjusting the intensity of hot air or the distribution of hot air. The above adjustment prompts are used to provide operators with segmented adjustment directions corresponding to the current quality risk.

[0128] After the state characteristics are formed, the prediction module determines the quality risk level, quality deviation trend, and associated process impact type of the corresponding material segment based on the state characteristics. Then, the output module outputs the corresponding quality risk warning information or process adjustment instruction information. In this way, the prediction results can not only reflect the quality risk level, but also further characterize the quality deviation trend and its more likely associated process impact sources, thereby providing more targeted directional information for production line adjustment.

[0129] Optional implementation methods under different parameter ranges:

[0130] Optionally, the linear speed of the continuous forming production line can be 5 m / min, 15 m / min or 25 m / min;

[0131] The process temperature of the hot pressing section can be 80℃, 110℃ or 140℃;

[0132] The air volume in the hot air treatment zone can be 300 m³ / h, 600 m³ / h or 900 m³ / h;

[0133] The temperature gradient in the cooling and shaping section can be 10℃, 25℃, or 40℃;

[0134] The tension parameters can be 20N, 50N, or 80N.

[0135] The aforementioned endpoint and intermediate values ​​correspond to implementation methods under lower, medium, and higher process windows, respectively. Under lower speed and smaller cooling gradient conditions, the correction amount of each forming section to the material segment position is relatively small, and the length of the residual influence fragment is relatively short. Under medium speed and medium cooling gradient conditions, the influence of the residual influence fragment and the core influence fragment on the state characteristics is relatively balanced. Under higher speed and larger cooling gradient conditions, the material segment migration distortion and cooling retraction are more obvious, and the role of the residual influence fragment in the state characteristics will increase.

[0136] The aforementioned endpoint values ​​and intermediate values ​​correspond to implementation methods under different process windows and can be used to support parameter selection under different production conditions in this invention.

[0137] The specific implementation method is as follows: select an appropriate implementation method within the above parameter range according to the actual equipment capacity, material system and process window of the production line; and adapt to continuous forming production lines under different production conditions by setting parameter combinations under different process windows.

[0138] In one embodiment, the online prediction system for automotive skin forming quality includes a data acquisition module, a corresponding module, a boundary segment construction module, a feature processing module, a prediction module, and an output module. These modules are connected sequentially according to the data flow direction and are used to process online operating condition data and downstream quality inspection data during the continuous forming process of automotive skins, and to output corresponding quality risk warning information or process adjustment instructions.

[0139] The data acquisition module is used to acquire online operating condition data of each forming section on the continuous forming production line for automotive skins, as well as quality inspection data output from the quality inspection station located downstream of each forming section. Optionally, the online operating condition data includes at least one of temperature parameters, tension parameters, speed parameters, pressure parameters, roll gap parameters, airflow parameters, and cooling parameters; the quality inspection data includes at least one of thickness detection data, surface image data, and surface temperature data.

[0140] The corresponding module is used to determine the correction amount for the material segment migration position of each forming segment based on the corresponding transmission path length parameter, the linear velocity parameter and the segment operation status parameter in the online working condition data. Based on each correction amount, the equivalent migration length of the corresponding forming segment is determined, and the equivalent migration lengths are back-accumulated in reverse according to the order of each forming segment in the production line to establish a segmented equivalent migration correspondence between quality inspection data and the position of the upstream material segment. Optionally, the correction amount includes at least one of temperature correction component, speed difference correction component, and cooling shrinkage correction component to characterize the thermally induced length change, the relative slippage between adjacent transmission or traction parts, and the length retraction effect during the cooling and shaping process, respectively.

[0141] Among them, the slip correction coefficient used in the velocity difference correction component can be pre-stored by the system based on historical calibration results and called when the corresponding module performs position backtracking calculation; when the material system, traction structure or tension window changes, the corresponding module can receive the updated slip correction coefficient.

[0142] The boundary segment construction module is used to determine the operating state boundary corresponding to each quality inspection data based on the segmented equivalent migration correspondence and combined with the state change triggering information in the online operating condition data. Using the operating state boundary as the demarcation, residual influence segments are formed within the transitional influence interval, and core influence segments are formed within the steady-state influence interval. Optionally, the state change triggering information includes at least one of the following: parameter change amplitude exceeding a preset threshold, parameter change rate exceeding a preset threshold, and receiving a process switching command.

[0143] The feature processing module determines steady-state features based on at least one of the working condition mean, working condition dispersion, and working condition change slope in the core influence segment. It also determines attenuation features by weighting the deviation of each working condition data from the steady-state benchmark, boundary distance, and parameter fluctuation amplitude in the residual influence segment. Furthermore, it normalizes the steady-state and attenuation features and then weights and fuses them to form state features. Optionally, the feature processing module can first extract steady-state features representing the working condition level, working condition fluctuation state, and working condition change trend from the core influence segment, and then extract attenuation features representing the tailing influence intensity, tailing influence attenuation rate, and boundary distance influence degree from the residual influence segment. Finally, it fuses the two types of features according to a preset fusion coefficient to obtain state features representing the forming state of the corresponding material segment.

[0144] The prediction module is used to input state characteristics into a preset prediction model or rule-based judgment unit to output at least one of the following: quality risk level, quality deviation trend, and associated process influence type. Optionally, the preset prediction model can be a classification model, regression model, or a combination of both trained based on historical qualified samples, early warning samples, and abnormal samples; the rule-based judgment unit can be a set of judgment rules established based on preset thresholds, weight coefficients, and logical judgment relationships. Further, the prediction module can first obtain the quality risk score of the corresponding material segment based on the state characteristics, and then determine the quality risk level based on the preset grading threshold; and determine the quality deviation trend based on the direction of change of the state characteristics over multiple consecutive detection cycles; simultaneously, determine the associated process influence type based on the feature contribution magnitude of the corresponding hot pressing forming section, hot air action section, and cooling and shaping section, respectively.

[0145] The output module is used to determine the target forming section based on the type of influence of the associated process, and output the corresponding adjustment direction of the operating parameters in conjunction with the quality deviation trend. Optionally, when the type of influence of the associated process points to the influence of hot pressing, the output module outputs the adjustment direction of the operating parameters for the hot pressing section; when the type of influence of the associated process points to the influence of hot air, the output module outputs the adjustment direction of the operating parameters for the hot air section; when the type of influence of the associated process points to the influence of cooling and shaping, the output module outputs the adjustment direction of the operating parameters for the cooling and shaping section. The adjustment direction of the operating parameters is preferably a directional prompt, rather than limited to a certain fixed value.

[0146] In one implementation, the system operates as follows: The acquisition module first obtains online operating condition data for each forming section and quality inspection data output from the downstream quality inspection station; the corresponding module determines the correction amount for the material segment migration position of each forming section based on the corresponding transmission path length parameter, the linear velocity parameter in the online operating condition data, and the section running status parameter, determines the equivalent migration length of the corresponding forming section based on each correction amount, and establishes a segmented equivalent migration correspondence between the quality inspection data and the upstream material segment position by backtracking the reverse cumulative equivalent migration length of each forming section; the boundary segment construction module identifies the operating condition boundary and forms residual influence segments and core influence segments based on this; the feature processing module extracts steady-state features and attenuation features respectively, and forms state features through normalization and weighted fusion; the prediction module outputs the quality risk level, quality deviation trend, and associated process influence type based on the state features; finally, the output module generates the target forming section identifier and the corresponding operating condition parameter adjustment direction, and outputs quality risk warning information or process adjustment instruction information.

[0147] Furthermore, when the quality deviation trend output by the prediction module is characterized by an increasing thickness deviation, and the associated process influence type points to the influence of hot pressing, the output module can output adjustment prompts in the direction of increasing the intensity of hot pressing, reducing the roll gap, or reducing the throughput speed. When the quality deviation trend is characterized by an abnormal surface temperature, and the associated process influence type points to the influence of cooling and shaping, the output module can output adjustment prompts in the direction of increasing the intensity of cooling, optimizing the cooling rhythm, or adjusting the traction speed. When the quality deviation trend is characterized by a fluctuating surface flatness, and the associated process influence type points to the influence of hot air, the output module can output adjustment prompts in the direction of adjusting the intensity of hot air or the hot air distribution state.

[0148] In actual operation, the system first maps downstream quality inspection results to upstream historical operating conditions, then distinguishes between residual and core impact segments, constructs state characteristics accordingly, and outputs quality risk level, quality deviation trend, and associated process impact type. Finally, it generates target forming section identifiers and corresponding operating condition parameter adjustment directions. Through this approach, the system not only accurately maps quality inspection data to upstream historical operating conditions but also differentiates and expresses steady-state process impacts and transitional tailing effects at the system level, further outputting segment-oriented process adjustment directions. This improves the pertinence and usability of online prediction results for real-time production line adjustments.

Claims

1. A method for online prediction of automotive skin forming quality, characterized in that, Includes the following steps: S1. Obtain online operating condition data of each forming section on the continuous forming production line for automotive skin, as well as quality inspection data output by the quality inspection station located downstream of each forming section. S2. For each forming section, based on the corresponding transmission path length parameter and the linear velocity parameter and section running status parameter in the online working condition data, determine the correction amount of each forming section to the material segment migration position, determine the equivalent migration length of the corresponding forming section based on each correction amount, and perform reverse accumulation and backtracking of each equivalent migration length according to the order of each forming section in the production line, so as to establish the segmented equivalent migration correspondence between the quality inspection data and the upstream material segment position; S3. Based on the segmented equivalent migration correspondence and combined with the state change triggering information in the online working condition data, determine the working condition state boundary corresponding to each of the quality detection data, and use the working condition state boundary as the boundary to form a residual influence segment in the transition influence interval and a core influence segment in the steady state influence interval. S4. Extract steady-state features from the core influence segment, extract attenuation features from the residual influence segment, and form state features characterizing the forming state of the corresponding material segment based on the steady-state features and attenuation features. S5. Generate a forming quality prediction result for the corresponding material segment based on the state characteristics. The forming quality prediction result includes at least one of the following: the type of influence of the associated process, the quality risk level, and the quality deviation trend. S6. Output quality risk warning information or process adjustment instruction information based on the molding quality prediction results.

2. The method for online prediction of automotive skin forming quality according to claim 1, characterized in that: The forming section includes at least two of the following: a hot pressing forming section, a hot air action section, a cooling and shaping section, and a traction output section.

3. The method for online prediction of automotive skin forming quality according to claim 1, characterized in that: The online operating condition data includes at least one of temperature parameters, tension parameters, speed parameters, pressure parameters, roll gap parameters, air volume parameters, and cooling parameters; the quality inspection data includes at least one of thickness inspection data, surface image data, and surface temperature data.

4. The method for online prediction of automotive skin forming quality according to claim 1, characterized in that: The correction amount in step S2 includes at least one of temperature correction component, speed difference correction component, and cooling shrinkage correction component; The temperature correction component is used to characterize the effect of thermally induced length change of the material segment in the hot pressing section or hot air action section; the speed difference correction component is used to characterize the effect of relative slippage between adjacent transmission or traction parts of the material segment; and the cooling shrinkage correction component is used to characterize the effect of length retraction of the material segment in the cooling and shaping section.

5. The method for online prediction of automotive skin forming quality according to claim 1, characterized in that: The state change triggering information includes at least one of the following: parameter change amplitude exceeding a preset threshold, parameter change rate exceeding a preset threshold, and receiving a process switching command. The working condition boundary is determined by mapping the moment when the state change trigger information is satisfied to the position of the upstream material segment.

6. The method for online prediction of automotive skin forming quality according to claim 1, characterized in that: The transitional influence interval starts from the boundary of the operating condition state and ends when the amplitude of parameter change corresponding to the online operating condition data is lower than the steady-state threshold for several consecutive sampling periods. The steady-state influence range is located after the transitional influence range; The residual impact segment consists of online operating condition data within the transitional impact range, while the core impact segment consists of online operating condition data within the steady-state impact range.

7. The method for online prediction of automotive skin forming quality according to claim 1, characterized in that: In step S4, for each operating condition parameter in the core influence segment, at least one of the following is calculated: the mean characteristic representing the operating condition level, the discrete characteristic representing the operating condition fluctuation state, and the slope characteristic representing the operating condition change trend, so as to form a steady-state characteristic. For each working condition parameter in the residual influence segment, weights are assigned based on the deviation of each sampling point from the steady-state benchmark, the boundary distance between the material segment position corresponding to each sampling point and the working condition boundary, and the parameter fluctuation amplitude. Then, attenuation characteristics are formed based on the weighted deviation and attenuation trend. After normalizing the steady-state features and the attenuation features, they are weighted and fused according to a preset fusion coefficient to obtain the state features that characterize the forming state of the corresponding material segment.

8. The method for online prediction of automotive skin forming quality according to claim 1, characterized in that: Step S5 includes: inputting the state characteristics into a preset prediction model or rule judgment unit to obtain a quality risk score for the corresponding material segment; determining the quality risk level based on the quality risk score and a preset grading threshold; determining the quality deviation trend based on the direction of change of the state characteristics in multiple consecutive detection cycles; determining the associated process influence type based on the feature contribution of the corresponding hot pressing forming section, hot air action section, and cooling and shaping section; and generating a target forming section identifier and corresponding working condition parameter adjustment direction based on the associated process influence type and the quality deviation trend.

9. An online prediction system for the forming quality of automotive skins, characterized in that, include: The data acquisition module is used to acquire online operating condition data of each forming section on the continuous forming production line for automotive skin, as well as quality inspection data output from the quality inspection station located downstream of each forming section. The corresponding module is used to determine the correction amount of each forming section to the material segment migration position based on the corresponding transmission path length parameter, the linear velocity parameter and the section running status parameter in the online working condition data, for each forming section, determine the equivalent migration length of the corresponding forming section based on the correction amount, and perform reverse accumulation and backtracking of each equivalent migration length according to the order of each forming section in the production line, so as to establish the segmented equivalent migration correspondence between the quality inspection data and the upstream material segment position; The boundary segment construction module is used to determine the working state boundary corresponding to each of the quality detection data based on the segmented equivalent migration correspondence and combined with the state change triggering information in the online working condition data. The module uses the working condition boundary as the boundary to form a residual influence segment in the transition influence interval and a core influence segment in the steady state influence interval. The feature processing module is used to extract steady-state features from the core influence segment, extract attenuation features from the residual influence segment, and form state features characterizing the forming state of the corresponding material segment based on the steady-state features and the attenuation features. The prediction module is used to generate a prediction result of the forming quality of the corresponding material segment based on the state characteristics; The output module is used to output quality risk warning information or process adjustment instruction information based on the forming quality prediction result.

10. The online prediction system for automotive skin forming quality according to claim 9, characterized in that: The feature processing module is used to determine steady-state features based on at least one of the working condition mean, working condition dispersion, and working condition change slope in the core influence segment, and to determine attenuation features based on the deviation of each working condition data in the residual influence segment from the steady-state benchmark, boundary distance, and parameter fluctuation amplitude after weighting processing. It is also used to normalize the steady-state features and the attenuation features and then weight-fuse them to form state features. The prediction module is used to input the state characteristics into a preset prediction model or rule judgment unit to output at least one of the following: quality risk level, quality deviation trend, and associated process influence type; the output module is used to determine the target forming section according to the associated process influence type, and output the corresponding working condition parameter adjustment direction in combination with the quality deviation trend.

Images