Automobile decoration parts quality optimization method and system based on production big data

Abstract

The present invention provides a method and system for optimizing the quality of automobile decorative parts based on production big data, which relates to the field of data processing technology. The method extracts production big data according to a preset cycle and constructs an abnormal evaluation model; the abnormal evaluation model is used to perform abnormal evaluation on the current production data of automobile decorative parts to obtain abnormal evaluation information, and then locate production control parameters; the production control parameters are used as optimization variables to perform optimization to obtain a production control strategy for production optimization control. The method solves the technical problem that the production adjustment of automobile decorative parts in the prior art has a lag, and manual analysis and optimization of production control are often performed after defective parts are produced, resulting in unstable production quality of automobile decorative parts. The method achieves the technical effect of dynamically adjusting and controlling the production parameters of automobile decorative parts, improving the production control accuracy of automobile decorative parts, and ensuring the production quality of automobile decorative parts.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for optimizing the quality of automobile decorative parts based on production big data. Background Art

[0002] The current production of automobile assembly parts suffers from a production adjustment lag, which is mainly reflected in the fact that manual analysis and optimization are only carried out after defective parts are produced.

[0003] This practice leads to unstable production quality of automobile decorative parts. For example, suppose an automobile assembly line produces a batch of decorative parts with defects in appearance. According to traditional practices, adjustments are made only after problems occur. The consequence of this adjustment method is that a certain number of defective products are produced, which wastes resources and increases production costs.

[0004] In summary, the prior art has the problem that the production adjustment of automobile decorative parts has a lag, and manual analysis and optimization of production control is often performed after defective parts are produced, resulting in technical problems such as unstable production quality of automobile decorative parts. Summary of the invention

[0005] The present application provides a method and system for optimizing the quality of automobile decorative parts based on production big data, which is used to solve the technical problems in the prior art that the production adjustment of automobile decorative parts has a lag, and manual analysis and optimization of production control are often performed after the production of defective parts, resulting in unstable production quality of automobile decorative parts.

[0006] In view of the above problems, the present application provides a method and system for optimizing the quality of automobile decorative parts based on production big data.

[0007] The first aspect of the present application provides a method for optimizing the quality of automobile decorative parts based on production big data, the method comprising: extracting production big data according to a preset cycle, the preset cycle corresponding to the production cycle of automobile decorative parts; performing abnormal case analysis and fitting according to the production big data, and constructing an abnormal evaluation model; performing abnormal evaluation on the current production data of automobile decorative parts through the abnormal evaluation model to obtain abnormal evaluation information; locating production control parameters based on the abnormal evaluation information; using the production control parameters as optimization variables, iteratively optimizing the optimization variables through a simulation model, and obtaining a production control strategy when the quality target is met; generating production adjustment instructions based on the production control strategy, and performing production optimization control.

[0008] The second aspect of the present application provides an automobile decorative parts quality optimization system based on production big data, the system comprising: a data extraction execution unit, used to extract production big data according to a preset cycle, wherein the preset cycle corresponds to the production cycle of automobile decorative parts; an evaluation model construction unit, used to perform abnormal case analysis and fitting according to the production big data, and construct an abnormal evaluation model; an abnormal evaluation execution unit, used to perform abnormal evaluation on the current production data of automobile decorative parts through the abnormal evaluation model, and obtain abnormal evaluation information; a control parameter positioning unit, used to locate the production control parameters based on the abnormal evaluation information; a control strategy optimization unit, used to use the production control parameters as optimization variables, perform iterative optimization of the optimization variables through a simulation model, and obtain a production control strategy when the quality target is met; a production optimization control unit, used to generate production adjustment instructions based on the production control strategy, and perform production optimization control.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] The method provided in the embodiment of the present application extracts production big data according to a preset cycle, and the preset cycle corresponds to the production cycle of automobile decorative parts; performs abnormal case analysis and fitting according to the production big data, and constructs an abnormal evaluation model; performs abnormal evaluation on the current production data of automobile decorative parts through the abnormal evaluation model to obtain abnormal evaluation information; locates production control parameters based on the abnormal evaluation information; uses the production control parameters as optimization variables, performs iterative optimization of optimization variables through a simulation model, and obtains a production control strategy when the quality target is met; generates production adjustment instructions based on the production control strategy, and performs production optimization control. The technical effect of dynamically adjusting and controlling the production parameters of automobile decorative parts, improving the production control accuracy of automobile decorative parts, and ensuring the production quality of automobile decorative parts is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic diagram of the process flow of the automobile decoration parts quality optimization method based on production big data provided in this application;

[0012] Figure 2 A schematic diagram of the process of locating production control parameters in the automobile decoration parts quality optimization method based on production big data provided in this application;

[0013] Figure 3 A schematic diagram of the structure of the automobile decoration parts quality optimization system based on production big data provided in this application.

[0014] Explanation of the accompanying drawings: data extraction execution unit 1, evaluation model building unit 2, abnormality assessment execution unit 3, control parameter positioning unit 4, control strategy optimization unit 5, production optimization control unit 6. DETAILED DESCRIPTION

[0015] This application provides a method and system for optimizing the quality of automobile decorative parts based on production big data, which is used to solve the technical problem in the prior art that the production adjustment of automobile decorative parts is lagging, and manual analysis and optimization of production control are often performed after the production of defective parts, resulting in unstable production quality of automobile decorative parts. The technical effect of dynamically adjusting and controlling the production parameters of automobile decorative parts, improving the production control accuracy of automobile decorative parts, and ensuring the production quality of automobile decorative parts is achieved.

[0016] Below, the technical solutions in the present invention will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments described herein. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. It should also be noted that, for the convenience of description, only the parts related to the present invention are shown in the accompanying drawings, rather than all of them.

[0017] Embodiment 1

[0018] like Figure 1 As shown, the present application provides a method for optimizing the quality of automobile decorative parts based on production big data, the method comprising:

[0019] A100: Extract production big data according to a preset cycle, where the preset cycle corresponds to a production cycle of automobile decorative parts.

[0020] Specifically, in this embodiment, the production cycle of a single batch of the automotive decorative parts is interactively obtained, and production big data is extracted based on the production cycle as a constraint. The production big data is a large amount of diverse data collected during the product manufacturing and production process of a single batch of the automotive decorative parts.

[0021] The production big data described in this embodiment specifically includes detailed records of various production activities corresponding to the production cycle of automotive decorative parts, all operating parameters of production equipment during the production process, product quality inspection results, the number of accidents during the production process, customer satisfaction feedback on product quality and other information, which are used for the subsequent construction of an abnormality evaluation model.

[0022] A200: Based on the production big data, perform abnormal case analysis and fitting, and build an abnormal evaluation model.

[0023] In one embodiment, based on the production big data, abnormal case analysis and fitting are performed to construct an abnormal evaluation model. The method step A200 provided in the present application also includes:

[0024] A210: Extract an abnormal case set from the production big data and set a top event, where the top event is the abnormal event with the highest quality impact selected according to the accident rate or customer satisfaction.

[0025] A220: Based on the top event, the impact relationship of the top event is analyzed through the abnormal case set to obtain intermediate events and basic events, wherein the intermediate event is an abnormal problem existing in the production process, and the basic event is an abnormal operating parameter of the production equipment.

[0026] A230: Construct an accident tree structure based on the logical relationship between the top event, intermediate event, and basic event.

[0027] A240: According to the fitting parameters of the occurrence probabilities of basic events, intermediate events and top events in the abnormal case set, the logical relationship of each tree structure is established, and the abnormal evaluation model is constructed by fitting them into the accident tree structure through logic gates.

[0028] In one embodiment, according to the fitting parameters of the occurrence probabilities and impact relationships of the basic events, intermediate events, and top events in the abnormal case set, the method step A240 provided by the present application further includes:

[0029] A241: According to the abnormal case set, taking the basic event as the influencing factor, the intermediate event relationship is fitted to obtain the fitting parameters of the basic event-intermediate event, and the fitting parameters are used to characterize the probability coefficient of the basic event for the occurrence of the intermediate event.

[0030] A242: Taking the intermediate event as the influencing factor, fitting the probability relationship of the top event is performed to obtain the fitting parameters of the intermediate event-top event.

[0031] Specifically, in this embodiment, an abnormal case set is extracted from the production big data, and the abnormal case set includes production failure events, quality problem events, equipment failure events, etc., and then the abnormal event with the greatest impact on the quality of the automotive decorative parts is screened from the abnormal case set according to the accident rate or customer satisfaction, and is set as the top event.

[0032] Based on the top event, the impact relationship of the top event is analyzed through the abnormal case set to identify the intermediate events and basic events that lead to the top event. The intermediate event is an abnormal problem that can be observed in the production process, such as production delays, increased defective rate, etc. The basic event leads to abnormal operating parameters of the production equipment where the intermediate event occurs, such as excessive temperature, insufficient pressure, etc. It should be understood that the logical relationship between the basic event, the intermediate event and the top event (abnormal event) is that the intermediate event is a direct manifestation of the failure to resolve the basic event in time, and the final result is the top event (abnormal event), but the basic event does not necessarily lead to the intermediate event, and the occurrence of the intermediate event does not necessarily lead to the occurrence of the top event (abnormal event).

[0033] It should be understood that, in this embodiment, based on the above process, multiple abnormal events are finally obtained as multiple intermediate events and multiple basic events corresponding to multiple top events.

[0034] By using the graphical tool of accident tree, a plurality of top events, a plurality of intermediate events and a plurality of basic events are graphically displayed according to the logical relationship (cause and effect relationship) among the top events, intermediate events and basic events, so as to complete the construction of the tree structure of the accident tree.

[0035] According to the abnormal case set, the production quantity of the automobile-generated decorative parts is obtained, and the multiple occurrence frequencies of the multiple top events are obtained. The multiple occurrence frequencies are used as numerators and the production quantity is used as the denominator to calculate the multiple occurrence probabilities of the multiple top events. The same method is used to obtain the multiple occurrence probabilities of multiple basic events and the occurrence probabilities of multiple intermediate events.

[0036] Based on the accident tree structure, multiple occurrence probabilities of multiple top events, multiple occurrence probabilities of multiple basic events and multiple occurrence probabilities of multiple intermediate events are divided into multiple groups of top event occurrence probability-basic event occurrence probability-intermediate event occurrence probability.

[0037] In each group of top event occurrence probability - basic event occurrence probability - intermediate event occurrence probability, the basic event occurrence probability is taken as the numerator and the intermediate event occurrence probability is taken as the denominator to calculate the fitting parameters of the basic event-intermediate event, which are used to characterize the probability coefficient of the basic event for the intermediate event.

[0038] Using the same method, in each group of top event occurrence probability-basic event occurrence probability-intermediate event occurrence probability, the intermediate event is used as the influencing factor to fit the top event occurrence probability relationship, and the fitting parameters of the intermediate event-top event are obtained. The fitting parameters are used to characterize the probability coefficient of the intermediate event for the occurrence of the top event.

[0039] The accident tree structure includes multiple branches, each branch is a group of top events, intermediate events, and basic events with a logical relationship. Then, the fitting parameters of multiple intermediate events-top events corresponding to the multiple top events obtained in the previous text and the probability of occurrence of multiple intermediate events are mapped and stored in the corresponding multiple intermediate events, the fitting parameters of multiple basic events-intermediate events and the probability of occurrence of multiple basic events are mapped and stored in the corresponding multiple basic events, and the probability of occurrence of multiple top events is mapped and stored in the corresponding multiple top events, so as to complete the construction of the abnormal evaluation model. The abnormal evaluation model can evaluate the probability of occurrence of the top event when a specific basic event occurs.

[0040] A300: Perform an abnormality evaluation on the current production data of automobile decorative parts through the abnormality evaluation model to obtain abnormality evaluation information.

[0041] Specifically, in this embodiment, the current production data is the real-time operating parameters of the production equipment of the automobile decorative parts.

[0042] Based on step A200, it can be known that the basic event is the operating parameter of the production equipment. Based on this, this embodiment uses the real-time operating parameter as the input of the abnormal evaluation model, traverses and compares multiple abnormal operating parameters of multiple basic events of the abnormal evaluation model, and calculates multiple abnormal similarities between the real-time operating parameter and the multiple abnormal operating parameters based on the Euclidean distance.

[0043] A similarity threshold is preset, and when the abnormal similarity between the real-time operation parameter and a certain abnormal operation parameter meets (falls within) the similarity threshold, the real-time operation parameter is considered to be a basic event that may induce an intermediate event and indirectly induce a top event.

[0044] The similarity threshold is used to traverse the multiple anomaly similarities, and a set of intermediate events, top events, fitting parameters of intermediate events-top events, fitting parameters of basic events-intermediate events, top event occurrence probabilities, basic event occurrence probabilities, and intermediate event occurrence probabilities corresponding to the anomaly similarities falling into the similarity threshold in the anomaly evaluation model are extracted as the anomaly evaluation information.

[0045] A400: Based on the abnormal evaluation information, locate the production control parameters.

[0046] In one embodiment, Figure 2 As shown, based on the abnormal evaluation information, the production control parameters are located, and the method step A400 provided in the present application also includes:

[0047] A410: Obtain abnormal events and abnormal probabilities based on the abnormal evaluation information.

[0048] A420: Based on the abnormal event, trace and locate the abnormal operating parameters according to the fault tree structure.

[0049] A430: According to the abnormal operation parameters and abnormal probability, the production control parameters are decomposed and located based on the current production data to obtain the production control parameters.

[0050] Specifically, in this embodiment, based on step A300, it can be known that the data structure of the abnormal evaluation information includes top events and the occurrence probability of top events, and it is known that the top events are actually abnormal events.

[0051] Based on this, this embodiment directly calls and obtains abnormal events from the abnormal evaluation information. The abnormal events include but are not limited to production failure events, quality problem events, and equipment failure events. The abnormal probability is the probability of occurrence of abnormal events.

[0052] The abnormal event is input as a top event into the fault tree structure for tracing and locating the source, and is traced back to the specific abnormal operation parameters that caused the abnormal event. The abnormal operation parameters are composed of multiple equipment operation parameters.

[0053] The abnormal operation parameters and the current production data are compared item by item with the same control index, and the same numerical items are retained, so as to obtain a total of multiple abnormal control index values, which constitute the candidate production control parameters.

[0054] The fitting parameters of the basic event-intermediate event obtained from the abnormal evaluation information are multiplied by the number of parameter indicators of the alternative production control parameters to obtain the number of primary screening indicators, and the primary screening indicator number is used to randomly extract from the multiple abnormal control indicator values ​​to obtain several abnormal control indicator values.

[0055] The fitting parameters of the intermediate event-abnormal event obtained from the abnormal evaluation information are multiplied by the number of parameter indicators of the several abnormal control indicator values ​​to obtain the number of secondary screening indicators, and the number of secondary screening indicators is used to randomly extract K abnormal control indicator values ​​from the several abnormal control indicator values, and the K abnormal control indicator values ​​constitute the production control parameters, and the production control parameters are adjustable indicators to be adjusted to avoid the occurrence of abnormal events.

[0056] A500: Taking the production control parameters as optimization variables, the optimization variables are iteratively optimized through the simulation model, and when the quality target is met, the production control strategy is obtained.

[0057] In one embodiment, the production control parameters are used as optimization variables, and the optimization variables are iteratively optimized through a simulation model. When the quality target is met, a production control strategy is obtained. The method step A500 provided in the present application also includes:

[0058] A510: The simulation model includes a first simulation channel and a second simulation channel arranged in parallel, and the first simulation channel and the second simulation channel include different optimization step lengths and optimization directions.

[0059] A520: Obtaining evaluation time, and based on the evaluation time, using the simulation model to perform simulation control with the production control parameters as optimization variables, to obtain optimization results for the first simulation channel and optimization results for the second simulation channel respectively.

[0060] A530: Taking the quality target as the evaluation value, the optimization results of the first simulation channel and the optimization results of the second simulation channel are evaluated, and the optimization step size and optimization direction of the first simulation channel and the second simulation channel are adjusted according to the optimization evaluation results.

[0061] A540: Utilize the adjusted first simulation channel and second simulation channel to iteratively optimize the production control parameters until the quality target is met.

[0062] A550: The preferred control strategy selected according to the optimization evaluation result is spliced ​​according to the temporal relationship of the evaluation time to obtain the production control strategy.

[0063] In one embodiment, the method step A520 of obtaining the evaluation time provided by the present application further includes:

[0064] A521: Perform timing matching based on the current production data and production control parameters to determine the abnormal positioning production cycle.

[0065] A522: Configure the evaluation time step according to the abnormal positioning production cycle.

[0066] A523: Based on the evaluation time step, obtain the evaluation time.

[0067] In one embodiment, the optimization step length and optimization direction of the first simulation channel and the second simulation channel are adjusted according to the optimization evaluation result. The method step A530 provided by the present application also includes:

[0068] A531: Based on the optimization evaluation result, the currently best control parameter of the optimization evaluation result is used as the preferred control strategy, and the corresponding simulation channel is set as the preferred channel, wherein the optimization step size and optimization direction of the preferred channel remain unchanged.

[0069] A532: The simulation channel other than the preferred channel is used as an adjustment channel. The optimization step length and optimization direction of the adjustment channel are adjusted based on the optimization step length and optimization direction of the preferred channel. At the same time, the optimization step length and optimization direction of the adjustment channel are added to the taboo space. The optimization direction and optimization step length in the taboo space will no longer be used in the adjustment of subsequent simulation channels.

[0070] In one embodiment, the production control strategy is obtained by splicing according to the temporal relationship of the evaluation time. The method step A550 provided by the present application further includes:

[0071] A551: Calculate the node strategy change difference of the splicing control strategy to obtain the node difference value.

[0072] A552: Perform smoothing processing based on the node difference value to obtain smoothing control parameters.

[0073] A553: The smooth splicing control parameters are evaluated with the quality target as the target value. When the evaluation result meets the quality target, the smooth splicing control parameters are used as the production control strategy.

[0074] A554: When the evaluation result does not meet the quality target, the adjustment weight is configured in the order of the node difference value and the evaluation time period.

[0075] A555: Adjust the smoothing splicing control parameters based on the adjustment weights to obtain the production control strategy.

[0076] Specifically, in this embodiment, a simulation model is pre-constructed, and the simulation model includes a first simulation channel and a second simulation channel arranged in parallel, and the first simulation channel and the second simulation channel include different optimization step lengths and optimization directions.

[0077] The construction methods and specific uses of the first simulation channel and the second simulation channel are the same, and both predict the quality performance of the produced automobile decorative parts based on the production control parameters.

[0078] Interactively obtain multiple sets of sample production control parameters and multiple sample quality performance measured information, build the first simulation channel based on the back propagation neural network, and divide the multiple sets of sample production control parameters and multiple sample quality performance measured information into training set, test set and verification set according to the data volume requirement of 17:1:2, and then use the conventional training method of the back propagation neural network to train and test the model of the first simulation channel based on the training set and the test set, and use the verification set to verify the model quality performance prediction accuracy of the first simulation channel until the prediction accuracy of the first simulation channel is higher than 98%. Copy the trained first simulation channel to obtain the second simulation channel.

[0079] The production control parameter adjustment direction and the single adjustment step length are set for the first simulation channel as the first optimization step length and the first optimization direction. The production control parameter adjustment direction and the single adjustment step length are set for the second simulation channel as the second optimization step length and the second optimization direction. The optimization direction is the numerical adjustment direction of the production control parameter, and the optimization step length is the adjustment amount of the single numerical adjustment. At the same time, the first optimization step length is 1 / 2 of the second optimization step length. The optimization direction of multiple adjustment indicators in the first optimization direction and the optimization direction of multiple adjustment indicators in the second optimization direction are mapped oppositely.

[0080] It should be understood that the earlier the risk of production abnormality is detected within a production cycle, the longer the time left for error correction. Conversely, the shorter the time left for error correction, the error correction time is the available time for optimizing production control parameters based on the simulation model.

[0081] This embodiment obtains the start time of the production of the current batch of automobile decorative parts and the production cycle of a single batch of automobile decorative parts, and then performs time matching in combination with the collection time of the current production data to determine the abnormal positioning production cycle, which is the remaining production time for the production of the current batch of automobile assembly parts.

[0082] 1 / 10 of the abnormality positioning production cycle is used as the evaluation time. The evaluation time is the longest time consumed for optimizing production control parameters based on the simulation model to determine the production control strategy.

[0083] A standard optimization step length is preset, the abnormal positioning production cycle is used as the numerator, and the single batch automobile decorative parts production cycle is used as the denominator to perform percentage calculation, and then the percentage calculation result is multiplied by the standard optimization step length to obtain the evaluation time step length.

[0084] The evaluation time is used as the optimization time constraint, the evaluation time step is used as the second optimization step, 1 / 2 of the second optimization step is used as the first optimization step, the first optimization direction and the second optimization direction are used as constraints, the production control parameters are used as the optimization starting point, the production control parameters are adjusted for multiple rounds based on the first optimization step and the first optimization direction, and then input into the first simulation channel, and multiple prediction qualities of multiple optimized control parameters are obtained as the optimization result of the first simulation channel, and the production control parameters are adjusted for multiple rounds based on the second optimization step and the second optimization direction, and then input into the second simulation channel, and multiple prediction qualities of multiple optimized control parameters are obtained as the optimization result of the second simulation channel.

[0085] Among the multiple prediction qualities of the optimization result, each prediction quality includes multiple performance indicator prediction values, and the quality target is multiple performance indicator target values.

[0086] Taking the quality target as the evaluation value, the optimization results of the first simulation channel and the optimization results of the second simulation channel are optimized and evaluated. The specific method of the optimization evaluation is to calculate the Euclidean distance between multiple performance indicator target values ​​of the quality target and each predicted quality including multiple performance indicator prediction values, and obtain multiple quality similarities corresponding to the optimization results of the first simulation channel and multiple quality similarities corresponding to the optimization results of the second simulation channel.

[0087] The two groups of quality similarities are grouped and the similarities are serialized, and after serialization, the optimization control parameter corresponding to the minimum similarity value is called as the optimization evaluation result.

[0088] The calculation result of 1 minus the minimum value of similarity is multiplied by the second optimization step length of the second simulation channel as the second updated optimization step length, 1 / 2 of the second updated optimization step length is used as the first updated optimization step length, the second optimization direction is used as the first updated optimization direction, and the first optimization direction is used as the second updated optimization direction, to complete the adjustment of the optimization step length and optimization direction of the first simulation channel and the second simulation channel according to the optimization evaluation results.

[0089] It should be understood that the above updated information is all conceptual data, and the specific call and use judgment process is as follows:

[0090] According to the optimization evaluation result, the optimized control parameters corresponding to the optimization evaluation result are used as the current best control parameters as the preferred control strategy, and the corresponding simulation channel is set as the preferred channel, and the optimization step size and optimization direction of the preferred channel remain unchanged.

[0091] The simulation channel other than the preferred channel is used as an adjustment channel. The optimization step length and optimization direction of the adjustment channel are adjusted based on the optimization step length and optimization direction of the preferred channel. At the same time, the optimization step length and optimization direction of the adjustment channel are added to the taboo space. The optimization direction and optimization step length in the taboo space are no longer used in the adjustment of subsequent simulation channels. The production control parameters are optimized iteratively using the adjusted first simulation channel and second simulation channel until the quality target is met, and the optimized control parameters that meet the quality target are used as the preferred control strategy.

[0092] It should be understood that the time consumed to obtain the preferred control strategy is the evaluation time. Therefore, this embodiment splices the preferred control strategy into the original production control strategy according to the evaluation time. At the same time, directly performing large-span adjustments to the production control parameters of the current production data risks affecting the operational stability of the production equipment. Therefore, this embodiment smoothes the control parameters to allow the current production data to smoothly and stably transition to the preferred control strategy. The specific method is as follows.

[0093] The current production data and the preferred control strategy are used as the splicing control strategy, and then the similarity of the two groups of equipment control data in the splicing control strategy is calculated based on the Euclidean distance, the node strategy change difference calculation of the splicing control strategy is completed, and the node difference value is obtained.

[0094] The node difference value is multiplied by 10 as the integer value as the number of intermediate transition control parameters between the current production data and the preferred control strategy, and then the smooth splicing control parameters are obtained. The smooth splicing control parameters include M transition control parameters, and the M transition control parameters have the same spacing, which evenly divide the control parameter intervals between the current production data and the preferred control strategy.

[0095] Taking the quality target as the target value, the simulation model is used to perform performance prediction, the smooth splicing control parameters are evaluated, and M performance index prediction values ​​are obtained. When the evaluation result (M performance index predictions) meets the quality target, the smooth splicing control parameters are used as the production control strategy.

[0096] When the evaluation result does not meet the quality target, the integer value of the node difference value multiplied by 10 divided by the evaluation time multiplied by the time proportion of the production cycle of a single batch of automobile decorative parts is used as the configuration adjustment weight.

[0097] The configuration adjustment weight is multiplied by M as the number of intermediate transition control parameters between the current production data and the preferred control strategy, and the smooth splicing control parameters are adjusted to obtain multiple transition control parameters. The multiple transition control parameters are then evaluated with the quality target as the target value, and the production control strategy is output when the evaluation result meets the quality target.

[0098] This embodiment achieves the technical effect of optimizing the production control strategy based on the model, improving the adjustment reliability when using the obtained production control strategy for production control adjustment, and improving the effectiveness of the obtained production control strategy in avoiding abnormal events.

[0099] A600: Generate production adjustment instructions based on the production control strategy to perform production optimization control.

[0100] Specifically, in this embodiment, a production adjustment instruction is generated based on the production control strategy, and the production adjustment instruction is used to perform optimization control of the automobile decoration parts production equipment.

[0101] This embodiment achieves the technical effect of dynamically adjusting and controlling the production parameters of automobile decorative parts, improving the production control accuracy of automobile decorative parts, and ensuring the production quality of automobile decorative parts.

[0102] Embodiment 2

[0103] Based on the same inventive concept as the automobile decoration parts quality optimization method based on production big data in the aforementioned embodiment, Figure 3 As shown, the present application provides an automobile decoration parts quality optimization system based on production big data, wherein the system includes:

[0104] The data extraction execution unit 1 is used to extract production big data according to a preset cycle, and the preset cycle corresponds to the production cycle of automobile decorative parts.

[0105] The evaluation model building unit 2 is used to perform abnormal case analysis and fitting based on the production big data to build an abnormal evaluation model.

[0106] The abnormality evaluation execution unit 3 is used to perform abnormality evaluation on the current production data of the automobile decoration parts through the abnormality evaluation model to obtain abnormality evaluation information.

[0107] The control parameter positioning unit 4 is used to position the production control parameters based on the abnormal evaluation information.

[0108] The control strategy optimization unit 5 is used to use the production control parameters as optimization variables, perform iterative optimization of the optimization variables through a simulation model, and obtain a production control strategy when the quality target is met.

[0109] The production optimization control unit 6 is used to generate production adjustment instructions based on the production control strategy to perform production optimization control.

[0110] In one embodiment, the evaluation model building unit 2 further includes:

[0111] A set of abnormal cases is extracted from the production big data, and a top event is set, wherein the top event is the abnormal event with the highest quality impact screened according to the accident rate or customer satisfaction; based on the top event, the impact relationship of the top event is analyzed through the abnormal case set to obtain intermediate events and basic events, wherein the intermediate event is an abnormal problem existing in the production process, and the basic event is an abnormal operating parameter of the production equipment; according to the logical relationship between the top event, the intermediate event, and the basic event, an accident tree structure is constructed; according to the fitting parameters of the occurrence probabilities of the basic events, intermediate events, and top events in the abnormal case set, the logical relationship of each tree structure is established, and the accident tree structure is fitted through logic gates to construct the abnormal evaluation model.

[0112] In one embodiment, the evaluation model building unit 2 further includes:

[0113] According to the abnormal case set, taking the basic event as the influencing factor, the intermediate event relationship is fitted to obtain the fitting parameters of the basic event-intermediate event, and the fitting parameters are used to characterize the probability coefficient of the basic event for the occurrence of the intermediate event; taking the intermediate event as the influencing factor, the top event occurrence probability relationship is fitted to obtain the fitting parameters of the intermediate event-top event.

[0114] In one embodiment, the control parameter positioning unit 4 further includes:

[0115] According to the abnormal evaluation information, abnormal events and abnormal probabilities are obtained; based on the abnormal events, abnormal operation parameters are traced and located according to the accident tree structure; according to the abnormal operation parameters and abnormal probabilities, production control parameters are decomposed and located based on the current production data to obtain the production control parameters.

[0116] In one embodiment, the control strategy optimization unit 5 further includes:

[0117] The simulation model includes a first simulation channel and a second simulation channel arranged in parallel, wherein the first simulation channel and the second simulation channel include different optimization steps and optimization directions; obtaining an evaluation time, and according to the evaluation time, using the simulation model to perform simulation control with the production control parameters as optimization variables, and obtaining optimization results of the first simulation channel and the second simulation channel respectively; taking the quality target as the evaluation value, performing optimization evaluation on the optimization results of the first simulation channel and the second simulation channel, and adjusting the optimization steps and optimization directions of the first simulation channel and the second simulation channel according to the optimization evaluation results; using the adjusted first simulation channel and the second simulation channel to iteratively optimize the production control parameters until the quality target is met; splicing the preferred control strategy selected according to the optimization evaluation result according to the timing relationship of the evaluation time to obtain the production control strategy.

[0118] In one embodiment, the control strategy optimization unit 5 further includes:

[0119] Perform time matching according to the current production data and production control parameters to determine the abnormality positioning production cycle; configure the evaluation time step according to the abnormality positioning production cycle; and obtain the evaluation time based on the evaluation time step.

[0120] In one embodiment, the control strategy optimization unit 5 further includes:

[0121] According to the optimization evaluation result, the currently best control parameter of the optimization evaluation result is used as the preferred control strategy, and the corresponding simulation channel is set as the preferred channel, wherein the optimization step length and optimization direction of the preferred channel remain unchanged; the simulation channel outside the preferred channel is used as an adjustment channel, and the optimization step length and optimization direction of the adjustment channel are adjusted based on the optimization step length and optimization direction of the preferred channel, and the optimization step length and optimization direction of the adjustment channel are added to the taboo space, and the optimization direction and optimization step length in the taboo space are no longer used in the adjustment of subsequent simulation channels.

[0122] In one embodiment, the control strategy optimization unit 5 further includes:

[0123] The node strategy change difference of the splicing control strategy is calculated to obtain the node difference value; smoothing is performed based on the node difference value to obtain the smooth splicing control parameter; the smooth splicing control parameter is evaluated with the quality target as the target value, and when the evaluation result meets the quality target, the smooth splicing control parameter is used as the production control strategy; when the evaluation result does not meet the quality target, the adjustment weight is configured in the order of the node difference value and the period of the evaluation time; the smooth splicing control parameter is adjusted based on the adjustment weight to obtain the production control strategy.

[0124] Any of the methods or steps described above may be stored as computer instructions or programs in various types of computer memories, and the computer instructions or programs may be recognized by various types of computer processors to implement any of the methods or steps described above.

[0125] Based on the above specific embodiments of the present invention, any improvements and modifications made to the present invention by those skilled in the art without departing from the principles of the present invention shall fall within the patent protection scope of the present invention.

Claims

1. A quality optimization method for automobile decoration parts based on production big data, characterized in that: The method comprises: Extracting production big data according to a preset cycle, wherein the preset cycle corresponds to a production cycle of automobile decorative parts; Based on the production big data, abnormal case analysis and fitting are performed to build an abnormal evaluation model; Performing abnormal evaluation on current production data of automobile decoration parts by using the abnormal evaluation model to obtain abnormal evaluation information; Locating production control parameters based on the abnormal evaluation information; Taking the production control parameters as optimization variables, the optimization variables are iteratively optimized through the simulation model, and when the quality target is met, the production control strategy is obtained; Generate production adjustment instructions based on the production control strategy to perform production optimization control; According to the production big data, abnormal case analysis and fitting are performed to build an abnormal evaluation model, including: Extracting an abnormal case set from the production big data and setting a top event, wherein the top event is an abnormal event with the highest quality impact selected according to accident rate or customer satisfaction; Based on the top event, the top event impact relationship analysis is performed through the abnormal case set to obtain intermediate events and basic events, wherein the intermediate event is an abnormal problem existing in the production process, and the basic event is an abnormal operating parameter of the production equipment; According to the logical relationship between the top event, the intermediate event and the basic event, an accident tree structure is constructed; According to the fitting parameters of the occurrence probabilities of basic events, intermediate events, and top events in the abnormal case set, the logical relationship of each tree structure is established, and the abnormal evaluation model is constructed by fitting the fault tree structure through logic gates; The fitting parameters of the occurrence probability and impact relationship of the basic events, intermediate events and top events in the abnormal case set include: According to the abnormal case set, taking the basic event as the influencing factor, performing intermediate event relationship fitting, obtaining basic event-intermediate event fitting parameters, wherein the fitting parameters are used to characterize the probability coefficient of the basic event for the intermediate event; Taking the intermediate event as the influencing factor, fitting the probability relationship of the top event is performed to obtain the fitting parameters of the intermediate event-top event; Based on the abnormal evaluation information, locate the production control parameters, including: Acquire abnormal events and abnormal probabilities according to the abnormal evaluation information; Based on the abnormal event, tracing and locating abnormal operation parameters according to the fault tree structure; According to the abnormal operation parameters and abnormal probability, production control parameters are decomposed and located based on the current production data to obtain the production control parameters.

2. The method according to claim 1, characterized in that The production control parameters are used as optimization variables, and the optimization variables are iteratively optimized through the simulation model. When the quality target is met, the production control strategy is obtained, including: The simulation model includes a first simulation channel and a second simulation channel arranged in parallel, wherein the first simulation channel and the second simulation channel include different optimization step lengths and optimization directions; Obtaining an evaluation time, and performing simulation control using the simulation model and the production control parameter as an optimization variable according to the evaluation time, to obtain an optimization result of the first simulation channel and an optimization result of the second simulation channel respectively; Taking the quality target as the evaluation value, performing optimization evaluation on the optimization results of the first simulation channel and the second simulation channel, and adjusting the optimization step length and optimization direction of the first simulation channel and the second simulation channel according to the optimization evaluation results; Iteratively optimizing production control parameters using the adjusted first simulation channel and second simulation channel until the quality target is met; The preferred control strategies selected according to the optimization evaluation results are spliced ​​according to the temporal relationship of the evaluation time to obtain the production control strategy.

3. The method according to claim 2, characterized in that The time for obtaining the evaluation includes: Perform time sequence matching based on the current production data and production control parameters to determine the abnormal positioning production cycle; According to the abnormal positioning production cycle, configure the evaluation time step; Based on the evaluation time step, the evaluation time is obtained.

4. The method according to claim 2, characterized in that According to the optimization evaluation result, the optimization step length and optimization direction of the first simulation channel and the second simulation channel are adjusted, including: According to the optimization evaluation result, the currently best control parameter of the optimization evaluation result is used as the preferred control strategy, and the corresponding simulation channel is set as the preferred channel, wherein the optimization step length and optimization direction of the preferred channel remain unchanged; The simulation channel outside the preferred channel is used as an adjustment channel. The optimization step length and optimization direction of the adjustment channel are adjusted based on the optimization step length and optimization direction of the preferred channel. At the same time, the optimization step length and optimization direction of the adjustment channel are added to the taboo space. The optimization direction and optimization step length in the taboo space are no longer used in the adjustment of subsequent simulation channels.

5. The method according to claim 2, characterized in that The production control strategy is obtained by splicing according to the temporal relationship of the evaluation time, including: Calculate the node strategy change difference of the splicing control strategy to obtain the node difference value; Perform smoothing processing based on the node difference value to obtain smoothing splicing control parameters; The smooth splicing control parameter is evaluated with the quality target as the target value, and when the evaluation result meets the quality target, the smooth splicing control parameter is used as the production control strategy; When the evaluation result does not meet the quality target, the adjustment weight is configured in the order of the node difference value and the evaluation time period; The smoothing splicing control parameters are adjusted based on the adjustment weights to obtain the production control strategy.

6. The automobile decoration parts quality optimization system based on production big data is characterized by: The steps for implementing the method according to any one of claims 1 to 5 include: A data extraction execution unit, used to extract production big data according to a preset cycle, wherein the preset cycle corresponds to a production cycle of automobile decorative parts; An evaluation model building unit, used to perform abnormal case analysis and fitting based on the production big data, and build an abnormal evaluation model; an abnormality evaluation execution unit, used to perform abnormality evaluation on current production data of automobile decorative parts through the abnormality evaluation model to obtain abnormality evaluation information; A control parameter locating unit, used for locating production control parameters based on the abnormal evaluation information; A control strategy optimization unit is used to use the production control parameters as optimization variables, perform iterative optimization of the optimization variables through a simulation model, and obtain a production control strategy when the quality target is met; The production optimization control unit is used to generate production adjustment instructions based on the production control strategy to perform production optimization control.

Images