Factory intelligent management and control system, method and device and storage medium

Abstract

The embodiment of the invention provides an intelligent factory management and control system, method and device and a storage medium, and the method comprises the steps: obtaining the field data of a machining machine for machining a workpiece, and the field data comprises field machining parameter information; performing feature extraction processing on the field processing parameter information to obtain a plurality of field processing feature values; outputting a field processing parameter quality index based on the plurality of field processing characteristic values; on the basis of the on-site processing parameter quality index and a preset processing index threshold, determining an on-site processing characteristic value with a qualified index; on the basis of the field processing characteristic value, outputting a field processing parameter similarity index; on the basis of the on-site processing parameter similarity index and a preset similarity threshold value, on-site processing characteristic values with qualified similarity are determined; and obtaining a first quality prediction result of the workpiece based on the on-site processing characteristic value with the qualified index and the on-site processing characteristic value with the qualified similarity. According to the embodiment of the invention, the precision of an edge calculation result is effectively improved, and the edge calculation efficiency and effect are improved.

Description

Technical Field

[0001] The present application relates to the field of smart factory technology, and in particular to a factory smart management and control system, method, device and storage medium. Background Art

[0002] The edge computing system is a distributed computing model that can move the processing power of computing and data storage to edge devices or edge nodes that are closer to the data source to reduce transmission delays, reduce network congestion, and improve system response speed and user experience. It has been widely used in the fields of Internet of Things, industrial automation, etc. The current edge computing system usually collects field data of the equipment, obtains prediction results based on the field data, and performs feedback control on the equipment according to the prediction results. The functions that can be realized are relatively simple, and there is a lack of quality analysis of field data, resulting in low accuracy of edge computing results, which affects the efficiency and effect of edge computing. Summary of the invention

[0003] In view of the above, it is necessary to provide a factory intelligent management and control system, method, device and storage medium, aiming to solve the technical problem that the above-mentioned edge computing system lacks quality analysis of field data, resulting in low accuracy of edge computing results.

[0004] In a first aspect, an embodiment of the present application provides a factory intelligent management and control system, including an edge platform, wherein the edge platform includes: a feature processing module, which is used to obtain field data of a workpiece processed by a processing machine, and perform feature extraction processing on field processing parameter information in the field data to obtain multiple field processing feature values; a processing parameter quality determination module, which has a processing parameter quality determination model, which is used to output a field processing parameter quality index based on multiple field processing feature values; and is used to determine the field processing feature value with a qualified index based on the field processing parameter quality index and a preset processing index threshold; a processing parameter similarity determination module, which has a processing parameter similarity determination model, which is used to output a field processing parameter similarity index based on the field processing feature value; and is used to determine the field processing feature value with a qualified similarity based on the field processing parameter similarity index and a preset similarity threshold; a quality prediction module, which has a quality prediction model, which is used to obtain a first quality prediction result of the workpiece based on the field processing feature value with a qualified index and the field processing feature value with a qualified similarity.

[0005] In a second aspect, an embodiment of the present application provides a factory intelligent management and control method, the method comprising: acquiring field data of a workpiece processed by a processing machine, the field data comprising field processing parameter information; performing feature extraction processing on the field processing parameter information to obtain a plurality of field processing feature values; outputting a field processing parameter quality index based on the plurality of field processing feature values; determining the field processing feature value with a qualified index based on the field processing parameter quality index and a preset processing index threshold; outputting a field processing parameter similarity index based on the field processing feature value; determining the field processing feature value with a qualified similarity based on the field processing parameter similarity index and a preset similarity threshold; obtaining a first quality prediction result of the workpiece based on the field processing feature value with a qualified index and the field processing feature value with a qualified similarity.

[0006] In a third aspect, an embodiment of the present application provides a factory intelligent management and control device, including a memory and a processor, the memory is used to store instructions, and the instructions stored in the memory are executed by the processor to implement the above-mentioned factory intelligent management and control method.

[0007] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored program, wherein when the program is run, the factory intelligent management and control device where the computer-readable storage medium is located is controlled to execute the above-mentioned factory intelligent management and control method.

[0008] The factory intelligent management and control system, method, device and storage medium provided in the embodiments of the present application obtain the field data of the processing machine and extract the characteristics of the field data, analyze the quality of the field data based on the quality determination model and the similarity determination model, and input the field data characteristics with qualified quality index and qualified similarity index into the edge computing model, which effectively improves the accuracy of the edge computing results and improves the efficiency and effect of edge computing. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0010] Figure 1 A schematic diagram of the structure of a factory intelligent management and control system provided in one embodiment of the present application.

[0011] Figure 2 A schematic diagram of the structure of an edge platform provided in one embodiment of the present application.

[0012] Figure 3 A schematic diagram of the structure of a central control platform provided in one embodiment of the present application.

[0013] Figure 4 A factory intelligent management and control method is provided in one embodiment of the present application.

[0014] Figure 5 A factory intelligent management and control method is provided in another embodiment of the present application.

[0015] Figure 6 A factory intelligent management and control method is provided in another embodiment of the present application.

[0016] Figure 7 A factory intelligent management and control method is provided in another embodiment of the present application.

[0017] Figure 8 A factory intelligent management and control method is provided in another embodiment of the present application.

[0018] Fig. 9 A factory intelligent management and control method is provided in another embodiment of the present application.

[0019] Fig.10 A factory intelligent management and control method is provided in another embodiment of the present application.

[0020] Fig.11 A schematic diagram of the structure of an intelligent factory management and control device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application is described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] It should be noted that in this application, "at least one" means one or more, and "more than one" means two or more than two. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The terms "first", "second", "third", "fourth", etc. (if any) in the specification, claims and drawings of this application are used to distinguish similar objects, rather than to describe a specific order or sequence.

[0023] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a concrete way. The following embodiments and features in the embodiments may be combined with each other without conflict.

[0024] The edge computing system is a distributed computing model that can move the processing power of computing and data storage to edge devices or edge nodes that are closer to the data source to reduce transmission delays, reduce network congestion, and improve system response speed and user experience. It has been widely used in the fields of Internet of Things, industrial automation, etc. The current edge computing system usually collects field data of the equipment, performs edge analysis on the field data to obtain prediction results, and performs feedback control on the equipment based on the prediction results. The functions that can be realized are relatively simple, and there is a lack of quality analysis of field data, resulting in low accuracy of edge computing results, which affects the efficiency of edge computing.

[0025] An embodiment of the present application provides a factory intelligent management and control system, which can analyze the quality of the field data based on a quality determination model and a similarity determination model after acquiring the field data of a processing machine and extracting the features of the field data, and input the field data features with qualified quality index and qualified similarity index into an edge computing model to obtain corresponding prediction results, thereby effectively improving the accuracy of the edge computing results and improving the edge computing efficiency.

[0026] See also Figure 1 , which is a schematic diagram of the structure of a factory intelligent management and control system provided in one embodiment of the present application. The factory intelligent management and control system 100 includes but is not limited to: multiple processing machines 10, an edge server 20, and a central control server 30. The edge server 20 has an edge platform 200, and the central control server 30 has a central control platform 300.

[0027] In one embodiment of the present application, the processing machine 10 can be a CNC (Computerized Numerical Control) device, a laser device, anodizing equipment, etc. The edge platform collects the field data of the processing workpiece of the processing machine through the kafka cluster architecture. The kafka cluster architecture is implemented based on a server. The kafka cluster architecture can collect the field data of the processing workpiece based on a switch, a hypertext transfer protocol (HTTP) or direct upload by the processing machine. The processing machine is the production end of the kafka cluster architecture, and the edge platform is the consumer end of the kafka cluster architecture. In another embodiment, further, the central control platform can also receive data transmitted by the edge platform through the kafka cluster architecture. At this time, the edge platform is the production end of the kafka cluster architecture, and the central control platform is the consumer end of the kafka cluster architecture.

[0028] See also Figure 2 , which is a schematic diagram of the structure of the edge platform provided in one embodiment of the present application. In one embodiment of the present application, the edge platform 200 includes, but is not limited to: a feature processing module 201, a processing parameter quality determination module 202, a processing parameter similarity determination module 203, a quality prediction module 204, a credibility module 205, a qualified quality determination module 206, and a quality information quality determination module 207.

[0029] In one embodiment of the present application, the feature processing module 201 is used to obtain the field data of the workpiece processed by the processing machine, and perform feature extraction processing on the field processing parameter information in the field data to obtain multiple field processing feature values. Among them, the field processing parameter information may include parameters, signals, images and basic information of the workpiece. Taking CNC equipment as an example, the field processing parameter information includes machine rotation signal, spindle runout value, ultrasonic signal, spindle vibration frequency, machine feeding time, etc. Taking laser equipment as an example, the field processing parameter information includes plasma, temperature, etc. For low-frequency data in the field processing parameter information, such as processing parameters (laser speed, laser temperature, etc.), basic information of the workpiece (length, width, height and other dimensional information of the workpiece), etc., feature extraction processing can be performed by calculating the average value to obtain the corresponding field processing feature value. For high-frequency data in the field processing parameter information, such as machine rotation signal, vibration signal, etc., feature extraction processing can be performed by Fourier transform to obtain the corresponding field processing feature value.

[0030] In one embodiment of the present application, the processing parameter quality determination module 202 has a processing parameter quality determination model, which is used to output a field processing parameter quality index based on multiple field processing characteristic values, and is used to determine a field processing characteristic value with a qualified index based on the field processing parameter quality index and a preset processing index threshold. The calculation formula (1) of the processing parameter quality determination model is:

[0031]

[0032] In the calculation formula (1), y is the set of on-site processing feature values; M is the dimension reduction matrix obtained by the PCA dimension reduction algorithm based on the historical processing feature values; is the p-th mean of the historical processing feature value set; y p is the pth processing feature value in the processing feature value set; p is the number of feature values ​​in the processing feature value set.

[0033] The on-site processing characteristic values ​​of various on-site processing parameter information are input into the processing parameter quality determination model, and after the on-site processing parameter quality index DQPx is calculated, the on-site processing parameter quality index DQPx is compared with the preset processing index threshold to determine whether the on-site processing parameter quality index DQPx is greater than or equal to the preset processing index threshold. If the on-site processing parameter quality index DQPx is greater than or equal to the preset processing index threshold, it can be determined that the quality index of the corresponding on-site processing characteristic value is qualified, and the corresponding on-site processing characteristic value is marked as qualified in quality index. If the on-site processing parameter quality index DQPx is less than the preset processing index threshold, it can be determined that the quality index of the corresponding on-site processing characteristic value is unqualified, and the corresponding on-site processing characteristic value is marked as unqualified in quality index.

[0034] In one embodiment of the present application, the processing parameter similarity determination module 203 has a processing parameter similarity determination model, which is used to output the on-site processing parameter similarity index based on the on-site processing feature value, and determine the on-site processing feature value with qualified similarity based on the on-site processing parameter similarity index and the preset similarity threshold. The calculation formula (2) of the processing parameter similarity determination model is:

[0035]

[0036] In the calculation formula (2), y is the set of on-site processing feature values, and M is the dimension reduction matrix obtained by the PCA dimension reduction algorithm based on the historical processing feature values; is the mean of historical processing characteristics.

[0037] The field processing characteristic values ​​of various field processing parameter information are input into the processing parameter similarity determination model, and after the field processing parameter similarity index SPI is calculated, the field processing parameter similarity index SPI is compared with the preset similarity threshold to determine whether the field processing parameter similarity index SPI is greater than or equal to the preset similarity threshold. If the field processing parameter similarity index SPI is greater than or equal to the preset similarity threshold, it can be determined that the similarity of the corresponding field processing characteristic value is unqualified, and the corresponding field processing characteristic value is marked as unqualified in similarity. If the field processing parameter similarity index SPI is less than the preset similarity threshold, it can be determined that the similarity of the corresponding field processing characteristic value is qualified, and the corresponding field processing characteristic value is marked as qualified in similarity.

[0038] In one embodiment of the present application, the quality prediction module 204 has a quality prediction model for obtaining a first quality prediction result of the workpiece based on the on-site processing feature value with a qualified index and the on-site processing feature value with a qualified similarity. The on-site processing feature value with a qualified index and a qualified similarity is input into the quality prediction model to obtain the first quality prediction result of the workpiece.

[0039] In one embodiment of the present application, the credibility module 205 has a reference prediction training model and a credibility index determination model. The credibility module 205 is used to input the index-qualified on-site processing feature value and the similarity-qualified on-site processing feature value into the reference prediction training model to output the second quality prediction result of the workpiece, and input the first quality prediction result of the workpiece and the second quality prediction result of the workpiece into the credibility index determination model to determine the credibility index of the quality prediction model. The quality prediction model can be an xgboost model, and the reference prediction training model can be a neural network model. The calculation formula (3) of the credibility index determination model is:

[0040]

[0041] The credibility index determination model obtains the average value μ and standard deviation σ of the workpiece quality prediction result based on the historical processing characteristic value, respectively, for the first quality prediction result F xgb And the second quality prediction result of the workpiece F nn Perform z-score processing to obtain the first quality prediction result F of the workpiece xgb The corresponding intermediate parameter Z xgb ,in, And get the first quality prediction result F of the workpiece xgb The corresponding intermediate parameter Z nn ,in, The intermediate parameter Z xgb and the intermediate parameter Z nn Input calculation formula (3) to obtain the reliability index RMI.

[0042] Determine whether the credibility index RMI is greater than or equal to the preset credibility threshold. If the credibility index RMI is greater than or equal to the preset credibility threshold, determine that the credibility of the first quality prediction result of the workpiece is qualified, and mark the first quality prediction result of the workpiece as having qualified credibility. If the credibility index RMI is less than the preset credibility threshold, determine that the credibility of the first quality prediction result of the workpiece is unqualified, and mark the first quality prediction result of the workpiece as having unqualified credibility. When the central control platform receives the first quality prediction result of the workpiece marked as unqualified credibility, it can trigger the training of the quality prediction model to update the quality prediction model in the edge platform.

[0043] In one embodiment of the present application, the field data also includes the field quality information of the workpiece. Taking CNC equipment as an example, the field quality information of the workpiece includes the spindle runout determination result, the deep melt drill abnormal index prediction result, etc. Taking laser equipment as an example, the field quality information of the workpiece includes the pull-out force, image appearance size, dirt characteristics, etc.

[0044] The qualified quality determination module 206 is used to determine qualified on-site quality information based on the on-site quality information and a preset quality standard.

[0045] On-site quality information that meets the preset quality standard is qualified on-site quality information. Taking laser equipment as an example, for example, a pull-out force greater than or equal to a preset threshold is a qualified pull-out force.

[0046] The quality information quality determination module 207 has a quality information quality determination model, which is used to determine the field quality information quality index of the field quality information based on the qualified field quality information, and determine the field quality information with a qualified index based on the field quality information quality index and the preset quality index threshold. The quality information quality determination module 207 determines whether the field quality information quality index DQMy is greater than or equal to the preset quality index threshold. If the field quality information quality index DQMy is greater than or equal to the preset quality index threshold, it is determined that the index of the corresponding field quality information is qualified, and the corresponding field quality information is marked as qualified. If the field quality information quality index DQMy is less than the preset quality index threshold, it is determined that the index of the corresponding field quality information is unqualified, and the corresponding field quality information is marked as unqualified. The processing logic of the quality information quality index DQMy is similar to that of the field processing parameter quality index DQPx, which will not be repeated here.

[0047] See also Figure 3 The central control platform 300 includes, but is not limited to: a data cleaning processing module 301, a feature processing solidification module 302, a processing parameter quality determination model training module 303, and a quality information quality determination model training module 304.

[0048] In one embodiment of the present application, the data cleaning processing module 301 is used to obtain on-site processing parameter information of the edge platform to determine high-frequency data and low-frequency data, determine whether the data volume of the high-frequency data reaches the high-frequency preset data volume, and analyze whether the high-frequency data is qualified data based on whether the data volume of the high-frequency data reaches the high-frequency preset data volume. The qualified high-frequency data and the low-frequency data are aligned with the timestamp and the workpiece identification to form aligned on-site processing parameter information.

[0049] In one embodiment of the present application, high-frequency data is, for example, signal data in the on-site processing parameter information, and the parameters set by the equipment, images, and basic information of the workpiece in the on-site processing parameter information are low-frequency data. The calculation formula (4) for the high-frequency preset data volume N is:

[0050] N = ∑ predicted target category × (number of predicted target detections per day × 5) (4);

[0051] In calculation formula (4), the predicted target category is the number of categories of high-frequency data, and the daily predicted target detection number is the number of detections of high-frequency data of each category.

[0052] In one embodiment of the present application, the unqualified data in the high-frequency data includes missing data and erroneous data. For missing data, the data cleaning processing module 301 can use the mean, median, mode or other estimated values ​​to fill in the missing data, or use a prediction model to predict the missing data. For erroneous data, the data cleaning processing module 301 can directly filter the erroneous data.

[0053] In one embodiment of the present application, the minimum and maximum timestamps in the high-frequency data set and the low-frequency data set are determined to determine the aligned timestamp range, the high-frequency data set and the low-frequency data set are resampled and converted to a uniform time frequency, and for timestamps without original data, interpolation methods such as linear interpolation and polynomial interpolation can be used to estimate the corresponding high-frequency data or low-frequency data, and then the high-frequency data and low-frequency data with completed timestamp alignment are aligned with the representation of the processed workpiece.

[0054] In one embodiment of the present application, the feature processing solidification module 302 is used to perform feature extraction processing on the aligned on-site processing parameter information according to a preset processing feature extraction processing method to obtain a training processing feature value. The preset processing feature extraction processing method can be to calculate the maximum value, minimum value, average value, median value, variance, standard deviation, quantile, fast Fourier transform, short-time Fourier transform, and wavelet transform.

[0055] In one embodiment of the present application, the processing parameter quality determination model training module 303 is used to: determine whether there is a processing parameter quality determination model corresponding to the training processing feature value in the preset database based on the training processing feature value; if the preset database does not have a processing parameter quality determination model corresponding to the training processing feature value, it means that the first training of the processing parameter quality training model is required to obtain the processing parameter quality determination model. Specifically, the processing parameter quality determination model training module 30 determines whether the accumulated data volume of the training processing feature value reaches the preset processing training data volume. For example, the preset processing training data volume is 20. If the preset processing training data volume is reached, the processing parameter quality determination model training module 303 inputs the training processing feature value of the preset processing training data volume into the processing parameter quality training model for training to obtain the initial training processing parameter quality index and the preset training processing threshold. The processing parameter quality determination model training module 303 further determines whether the initial training processing parameter quality index is greater than the preset training processing threshold. If the initial training processing parameter quality index is greater than the preset training processing threshold, the processing parameter quality training model is solidified to form a processing parameter quality determination model, and the preset training processing threshold is updated to the preset processing index threshold accordingly, so as to update the processing parameter quality determination model and the preset processing index threshold to the edge platform and store them in the preset database.

[0056] In one embodiment of the present application, the processing parameter quality determination model in the preset database carries a preset processing index threshold, and the processing parameter quality determination model training module 303 is also used to: if the preset database has a processing parameter quality determination model, input the training processing feature value into the processing parameter quality determination model in the preset database to obtain the training processing parameter quality index; determine whether the training processing parameter quality index is greater than the preset processing index threshold, if the training processing parameter quality index is greater than the preset processing index threshold, determine the training processing feature value corresponding to the training processing parameter quality index greater than the preset processing index threshold as a qualified training processing feature value; determine whether the accumulated data volume of the qualified training processing feature value reaches the preset processing retraining data volume. Among them, the preset processing retraining data volume is a preset multiple of the preset processing training data volume, and the preset multiple is, for example, 3 times. If the preset processing retraining data volume is reached, the processing parameter quality determination model training module 303 inputs the qualified training processing feature value of the preset processing retraining data volume into the processing parameter quality training model for retraining to obtain a newly trained processing parameter quality index and a newly trained processing threshold. The processing parameter quality determination model training module 303 determines whether the newly trained processing parameter quality index is greater than the newly trained processing threshold; if the newly trained processing parameter quality index is greater than the newly trained processing threshold, the re-trained processing parameter quality training model is solidified to form a new version of the processing parameter quality determination model, and the newly trained processing threshold is formed into a new version of the preset processing index threshold, so as to update the processing parameter quality determination model and the preset processing index threshold of the edge platform respectively.

[0057] In one embodiment of the present application, the quality information quality determination model training module 304 is used to search whether there is a quality information quality determination model in the preset database based on the on-site quality information with qualified indexes. If there is no quality information quality determination model in the preset database, it means that the quality information quality training model needs to be trained for the first time to obtain the quality information quality determination model. Specifically, the quality information quality determination model training module 304 determines whether the accumulated data volume of the on-site quality information with qualified indexes reaches the preset quality training data volume. For example, the preset quality training data volume is 20. If the preset quality training data volume is reached, the quality information quality determination model training module 304 inputs the on-site quality information with qualified indexes of the preset quality training data volume into the quality information quality training model for training to obtain the initial training quality information quality index and the preset training quality threshold. The quality information quality determination model training module 304 determines whether the initial training quality information index is greater than the preset training quality threshold; if the initial training quality index is greater than the preset training quality threshold, the quality information quality training model is solidified to form a quality information quality determination model, and the preset training quality threshold is updated to the preset quality index threshold accordingly, so as to update the quality information quality determination model and the preset quality index threshold to the edge platform and store them in the preset database.

[0058] In one embodiment of the present application, the quality information quality determination model in the preset database carries a preset quality index threshold, and the quality information quality determination model training module 304 is also used to: if the preset database has a quality information quality determination model, input the qualified on-site quality information into the quality information quality determination model in the preset database to obtain the training quality information quality index; determine whether the training quality information quality index is greater than the preset quality index threshold, and if the training quality information quality index is greater than the preset quality index threshold, determine the quality information corresponding to the training quality information quality index greater than the preset quality index threshold as qualified training quality information. Determine whether the accumulated data volume of qualified training quality information reaches the preset quality retraining data volume. The preset quality retraining data volume is a preset multiple of the preset quality training data volume, and the preset multiple is, for example, 3 times. If the preset processing retraining data volume is reached, the quality information quality determination model training module 304 inputs the qualified training quality information of the preset processing retraining data volume into the quality information quality training model for retraining to obtain the new training quality information quality index and the new training quality threshold. Determine whether the quality index of the newly trained quality information is greater than the new training quality threshold; if the quality index of the newly trained quality information is greater than the new training quality threshold, solidify the re-trained processing parameter quality training model to form a new version of the quality information quality determination model, and update the newly trained processing threshold to the new version of the preset processing index threshold, so as to update the quality information quality determination model and the preset processing index threshold of the edge platform respectively.

[0059] In one embodiment of the present application, the central control platform further includes a processing parameter similarity training module 305 , a quality prediction model training module 306 , and a credibility index determination model training module 307 .

[0060] In one embodiment of the present application, the processing parameter similarity training module 305 is used to obtain on-site processing feature values ​​with qualified edge platform similarity for training to obtain a new version of the processing parameter similarity determination model and a new preset similarity threshold so as to update the processing parameter similarity determination model in the edge platform.

[0061] Specifically, based on the edge platform similarity qualified field processing feature values ​​newly obtained from the edge platform, the updated dimension reduction matrix M and the historical processing feature mean in the calculation formula (2) are determined: Determine the maximum value of the processing parameter similarity SPI in the historical data, and determine 3 times the maximum value of the processing parameter similarity SPI as the new preset similarity threshold. In one embodiment of the present application, the quality prediction model training module 306 is used to: obtain qualified sample information of the edge platform, the qualified sample information includes the edge platform's index-qualified on-site processing feature value, the similarity-qualified on-site processing feature value, and the index-qualified on-site quality information, and determine whether the data volume of the qualified sample information reaches the preset quality prediction training data volume. If the data volume of the qualified sample information reaches the preset quality prediction training data volume, based on the qualified sample information, search in the preset database for a quality prediction model. If there is no quality prediction model in the preset database, it means that the quality prediction training model needs to be trained for the first time to obtain the quality prediction model. Specifically, the quality prediction model training module 306 inputs the qualified sample information of the preset quality prediction training data volume into the quality prediction training model for full training to obtain an initial quality prediction model; and verifies whether the initial quality prediction model is qualified; solidifies the qualified initial quality prediction model to form a new version of the quality prediction model to update the quality prediction model of the edge platform and store it in the preset database.

[0062] For example, the quality prediction training model can be xgboost. The k-fold method is used to randomly split the filtered historical data into 5 equal parts, and 4 equal parts are taken out as training sets. The remaining 1 equal part is used as a test set. Five quality prediction models are trained, and the model with the smallest loss is selected as the final model. During training, the modeling parameters are adjusted point by point, and the parameters with the lowest cross-validation values ​​are selected as the optimal parameters. At this time, the final model is determined as the quality prediction model.

[0063] In one embodiment of the present application, the quality prediction model training module 306 is also used to: if the preset database has a quality prediction model, input the qualified sample information of the preset quality prediction training data volume into the quality prediction model of the preset database for incremental training to obtain a new version of the quality prediction model; input the field processing feature value with qualified index and the field processing feature value with qualified similarity into the new version of the quality prediction model to obtain the third quality prediction result of the workpiece; based on the third quality prediction result of the workpiece, the first quality prediction result of the workpiece of the edge platform and the field quality information with qualified index, determine the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model, and determine whether the prediction accuracy of the new version of the quality prediction model is greater than the prediction accuracy of the quality prediction model in the preset database, so as to update the quality prediction model of the edge platform. Specifically, if the prediction accuracy of the new version of the quality prediction model is greater than the prediction accuracy of the quality prediction model in the preset database, send the new version of the quality prediction model to the edge platform to update the quality prediction model of the edge platform. If the prediction accuracy of the new version of the quality prediction model is less than or equal to the prediction accuracy of the quality prediction model in the preset database, there is no need to send the new version of the quality prediction model to the edge platform.

[0064] In one embodiment of the present application, the credibility index determination model training module 307 is used to input the on-site processing feature values ​​with qualified index and qualified similarity and the quality information with qualified index into the reference prediction training model, train the reference prediction training model to obtain a new version of the reference prediction model, and solidify the new version of the credibility index determination model to update the credibility index determination model in the edge platform.

[0065] For example, the reference prediction training model can be a neural network model. The on-site processing feature values ​​with qualified index and qualified similarity and the quality information with qualified index are input into the neural network training model, and the neural network training model is trained to obtain a new version of the neural network model to solidify a new version of the credibility index determination model.

[0066] In one embodiment of the present application, there are multiple sets of preset processing feature extraction methods, and the quality prediction model training module 306 is also used to determine whether the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model are both less than the preset accuracy. The feature processing solidification module 301 is also used to replace and solidify the processing feature extraction processing method if the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model are both less than the preset accuracy.

[0067] like Figure 1As shown, in one embodiment of the present application, the factory intelligent management and control system 100 further includes an application terminal 40, which is used to receive and display the first quality prediction result of the workpiece. The application terminal 40 is also used to receive the marked data and transmit the marked data to the central control platform.

[0068] See also Figure 4 FIG. 1 is a flow chart of a factory intelligent management and control method provided by an embodiment of the present application. The present application also provides a factory intelligent management and control method, which is applied to an edge platform.

[0069] S101, obtaining on-site data of a workpiece processed by a processing machine, wherein the on-site data includes on-site processing parameter information.

[0070] Specifically, the on-site processing parameter information may include parameters, signals, images and basic workpiece information.

[0071] S102, perform feature extraction processing on the on-site processing parameter information to obtain multiple on-site processing feature values. Specifically, taking CNC equipment as an example, the on-site processing parameter information includes machine rotation signals, spindle runout values, ultrasonic signals, spindle vibration frequencies, machine feed time, etc. Taking laser equipment as an example, the on-site processing parameter information includes plasma, temperature, etc. For low-frequency data in the on-site processing parameter information, such as processing parameters (laser speed, laser temperature, etc.), basic workpiece information (length, width, height and other dimensional information of the workpiece), etc., feature extraction processing can be performed by calculating the average value to obtain the corresponding on-site processing feature values. For high-frequency data in the on-site processing parameter information, such as machine rotation signals, vibration signals, etc., feature extraction processing can be performed by Fourier transform to obtain the corresponding on-site processing feature values.

[0072] S103, based on the multiple on-site processing characteristic values, outputting the on-site processing parameter quality index. S104, based on the on-site processing parameter quality index and the preset processing index threshold, determining the on-site processing characteristic value with a qualified index.

[0073] Specifically, the edge platform has a processing parameter quality determination model, which is used to output a field processing parameter quality index based on multiple field processing feature values, and is used to determine a field processing feature value with a qualified index based on the field processing parameter quality index and a preset processing index threshold. The calculation formula (1) of the processing parameter quality determination model is:

[0074]

[0075] In the calculation formula (1), y is the set of on-site processing feature values; M is the dimension reduction matrix obtained by the PCA dimension reduction algorithm based on the historical processing feature values; is the p-th mean of the historical processing feature value set; y pis the pth processing feature value in the processing feature value set; p is the number of feature values ​​in the processing feature value set.

[0076] The on-site processing characteristic values ​​of various on-site processing parameter information are input into the processing parameter quality determination model, and after the on-site processing parameter quality index DQPx is calculated, the on-site processing parameter quality index DQPx is compared with the preset processing index threshold to determine whether the on-site processing parameter quality index DQPx is greater than or equal to the preset processing index threshold. If the on-site processing parameter quality index DQPx is greater than or equal to the preset processing index threshold, it can be determined that the quality index of the corresponding on-site processing characteristic value is qualified, and the corresponding on-site processing characteristic value is marked as qualified in quality index. If the on-site processing parameter quality index DQPx is less than the preset processing index threshold, it can be determined that the quality index of the corresponding on-site processing characteristic value is unqualified, and the corresponding on-site processing characteristic value is marked as unqualified in quality index.

[0077] S105: Based on the on-site processing characteristic value, output the on-site processing parameter similarity index. S106: Based on the on-site processing parameter similarity index and a preset similarity threshold, determine the on-site processing characteristic value with qualified similarity.

[0078] Specifically, the edge platform also has a processing parameter similarity determination model, which is used to output the field processing parameter similarity index based on the field processing feature value, and determine the field processing feature value with qualified similarity based on the field processing parameter similarity index and the preset similarity threshold. The calculation formula (2) of the processing parameter similarity determination model is:

[0079]

[0080] In the calculation formula (2), y is the set of on-site processing feature values, and M is the dimension reduction matrix obtained by the PCA dimension reduction algorithm based on the historical processing feature values; Historical processing feature mean.

[0081] The field processing characteristic values ​​of various field processing parameter information are input into the processing parameter similarity determination model, and after the field processing parameter similarity index SPI is calculated, the field processing parameter similarity index SPI is compared with the preset similarity threshold to determine whether the field processing parameter similarity index SPI is greater than or equal to the preset similarity threshold. If the field processing parameter similarity index SPI is greater than or equal to the preset similarity threshold, it can be determined that the similarity of the corresponding field processing characteristic value is unqualified, and the corresponding field processing characteristic value is marked as unqualified in similarity. If the field processing parameter similarity index SPI is less than the preset similarity threshold, it can be determined that the similarity of the corresponding field processing characteristic value is qualified, and the corresponding field processing characteristic value is marked as qualified in similarity.

[0082] S107, obtaining a first quality prediction result of the workpiece based on the on-site processing feature value with a qualified index and the on-site processing feature value with a qualified similarity.

[0083] Specifically, the edge platform also has a quality prediction model for obtaining a first quality prediction result of the workpiece based on the on-site processing feature value with a qualified index and the on-site processing feature value with a qualified similarity. The on-site processing feature value with a qualified index and a qualified similarity is input into the quality prediction model to obtain the first quality prediction result of the workpiece.

[0084] See also Figure 5 FIG. 1 is a flow chart of a factory intelligent management and control method provided by another embodiment of the present application. In one embodiment of the present application, the factory intelligent management and control method further includes:

[0085] S201, inputting the on-site processing feature values ​​with qualified indexes and the on-site processing feature values ​​with qualified similarities into a reference prediction model to output a second quality prediction result of the workpiece.

[0086] In one embodiment of the present application, the quality prediction model may be an xgboost model, and the reference prediction training model may be a neural network model.

[0087] S202: Determine a credibility index of a quality prediction model based on a first quality prediction result of the workpiece and a second quality prediction result of the workpiece.

[0088] In one embodiment of the present application, the first quality prediction result of the workpiece and the second quality prediction result of the workpiece are input into a credibility index determination model to determine the credibility index of the quality prediction model.

[0089] The calculation formula (3) of the credibility index determination model is:

[0090]

[0091] The credibility index determination model obtains the average value μ and standard deviation σ of the workpiece quality prediction result based on the historical processing characteristic value, respectively, for the first quality prediction result F xgb And the second quality prediction result of the workpiece F nn Perform z-score processing to obtain the first quality prediction result F of the workpiece xgb The corresponding intermediate parameter Z xgb ,in, And get the first quality prediction result F of the workpiece xgb The corresponding intermediate parameter Z nn ,in, The intermediate parameter Z xgb and the intermediate parameter Z nn Input calculation formula (3) to obtain the reliability index RMI.

[0092] Determine whether the credibility index RMI is greater than or equal to the preset credibility threshold. If the credibility index RMI is greater than or equal to the preset credibility threshold, determine that the credibility of the first quality prediction result of the workpiece is qualified, and mark the first quality prediction result of the workpiece as having qualified credibility. If the credibility index RMI is less than the preset credibility threshold, determine that the credibility of the first quality prediction result of the workpiece is unqualified, and mark the first quality prediction result of the workpiece as having unqualified credibility. When the central control platform receives the first quality prediction result of the workpiece marked as unqualified credibility, it can trigger the training of the quality prediction model to update the quality prediction model in the edge platform.

[0093] See also Figure 6 FIG. 1 is a flow chart of a factory intelligent management and control method provided by another embodiment of the present application. In one embodiment of the present application, the field data also includes field quality information of the workpiece, and the factory intelligent management and control method also includes:

[0094] S301, determining qualified on-site quality information based on on-site quality information and preset quality standards.

[0095] Taking CNC equipment as an example, the on-site quality information of the workpiece includes the spindle runout judgment result, the deep melt drill abnormal index prediction result, etc. Taking laser equipment as an example, the on-site quality information of the workpiece includes the pull-out force, image appearance size, dirt characteristics, etc. The on-site quality information that meets the preset quality standards is qualified on-site quality information. For example, the pull-out force greater than or equal to the preset threshold is qualified.

[0096] S302: Determine a field quality information quality index of the field quality information based on the qualified field quality information.

[0097] In one embodiment of the present application, qualified on-site quality information is input into a quality information quality determination model to obtain an on-site quality information quality index of the on-site quality information.

[0098] S303: Determine on-site quality information with a qualified index based on the on-site quality information quality index and a preset quality index threshold.

[0099] In one embodiment of the present application, the edge platform determines whether the quality index DQMy of the on-site quality information is greater than or equal to the preset quality index threshold. If the quality index DQMy of the on-site quality information is greater than or equal to the preset quality index threshold, the index of the corresponding on-site quality information is determined to be qualified, and the corresponding on-site quality information is marked as qualified. If the quality index DQMy of the on-site quality information is less than the preset quality index threshold, the index of the corresponding on-site quality information is determined to be unqualified, and the corresponding on-site quality information is marked as unqualified. The processing logic of the quality information quality index DQMy is similar to that of the on-site processing parameter quality index DQPx, which will not be repeated here.

[0100] See also Figure 7 , which is a flow chart of a factory intelligent management and control method provided by another embodiment of the present application. The factory intelligent management and control method of this embodiment is applied to a central control platform. In one embodiment of the present application, the factory intelligent management and control method further includes:

[0101] S401, obtaining on-site processing parameter information of the edge platform to determine high-frequency data and low-frequency data.

[0102] In one embodiment of the present application, the high-frequency data may be signal data in the on-site processing parameter information, and the low-frequency data may be parameters set by the equipment, images, basic information of the workpiece, etc. in the on-site processing parameter information.

[0103] S402, determining whether the amount of high-frequency data reaches a preset amount of high-frequency data.

[0104] The calculation formula (4) for the high-frequency preset data volume N is:

[0105] N = ∑ predicted target category × (number of predicted target detections per day × 5) (4);

[0106] In calculation formula (4), the predicted target category is the number of categories of high-frequency data, and the daily predicted target detection number is the number of detections of high-frequency data of each category.

[0107] S403, based on whether the amount of high-frequency data reaches a preset amount of high-frequency data, analyzing whether the high-frequency data is qualified data.

[0108] In one embodiment of the present application, the unqualified data in the high-frequency data includes missing data and erroneous data. For missing data, the mean, median, mode or other estimated values ​​can be used to fill the missing data, or a prediction model can be used to predict the missing data. For erroneous data, the edge platform can directly filter the erroneous data.

[0109] S404: align the qualified high-frequency data with the low-frequency data by using the timestamp and the identification of the workpiece to form aligned on-site processing parameter information.

[0110] In one embodiment of the present application, the minimum and maximum timestamps in the high-frequency data set and the low-frequency data set are determined to determine the aligned timestamp range, the high-frequency data set and the low-frequency data set are resampled and converted to a uniform time frequency, and for timestamps without original data, interpolation methods such as linear interpolation and polynomial interpolation can be used to estimate the corresponding high-frequency data or low-frequency data, and then the high-frequency data and low-frequency data with completed timestamp alignment are aligned with the representation of the processed workpiece.

[0111] See also Figure 8 FIG. 1 is a flow chart of a factory intelligent management and control method provided by another embodiment of the present application. In one embodiment of the present application, the factory intelligent management and control method further includes:

[0112] S501, performing feature extraction processing on the aligned on-site processing parameter information according to a preset processing feature extraction processing method to obtain a training processing feature value.

[0113] In one embodiment of the present application, the preset processing feature extraction processing method can be to calculate the maximum value, minimum value, average value, median value, variance, standard deviation, quantile, fast Fourier transform, short-time Fourier transform, and wavelet transform.

[0114] S502, based on the training processing feature value, determining whether there is a processing parameter quality determination model corresponding to the training processing feature value in a preset database.

[0115] S503: If the preset database does not have a processing parameter quality determination model corresponding to the training processing feature value, it is determined whether the accumulated data volume of the training processing feature value reaches the preset processing training data volume.

[0116] In one embodiment of the present application, if the preset database does not have a processing parameter quality determination model corresponding to the training processing feature value, it means that the first training needs to be performed on the processing parameter quality training model to obtain the processing parameter quality determination model.

[0117] S504: If the preset processing training data volume is reached, the training processing feature value of the preset processing training data volume is input into the processing parameter quality training model for training to obtain the initial training processing parameter quality index and the preset training processing threshold.

[0118] S505, determining whether the initial training processing parameter quality index is greater than a preset training processing threshold.

[0119] S506, if the initial training processing parameter quality index is greater than the preset training processing threshold, the processing parameter quality training model is solidified to form a processing parameter quality determination model, and the preset training processing threshold is updated to the preset processing index threshold accordingly, so as to update the processing parameter quality determination model and the preset processing index threshold to the edge platform.

[0120] In one embodiment of the present application, the processing parameter quality determination model in the preset database carries a preset processing index threshold, and the method further includes:

[0121] S507, if the preset database has a processing parameter quality determination model, input the training processing feature value into the processing parameter quality determination model in the preset database to obtain the training processing parameter quality index.

[0122] S508, determining whether the training processing parameter quality index is greater than a preset processing index threshold.

[0123] S509: If the training processing parameter quality index is greater than the preset processing index threshold, the training processing feature value corresponding to the training processing parameter quality index greater than the preset processing index threshold is determined as a qualified training processing feature value.

[0124] S510, determining whether the accumulated data volume of qualified training processing feature values ​​reaches a preset processing retraining data volume.

[0125] The preset processing retraining data volume is a preset multiple of the preset processing training data volume, and the preset multiple is, for example, 3 times.

[0126] S511, if the preset processing retraining data volume is reached, the training processing feature values ​​of the preset processing retraining data volume that meet the requirements are input into the processing parameter quality training model for retraining to obtain a new training processing parameter quality index and a new training processing threshold.

[0127] S512, determining whether the quality index of the new training processing parameter is greater than the new training processing threshold.

[0128] S513, if the newly trained processing parameter quality index is greater than the newly trained processing threshold, the re-trained processing parameter quality training model will be solidified to form a new version of the processing parameter quality determination model, and the newly trained processing threshold will be formed into a new version of the preset processing index threshold, so as to update the processing parameter quality determination model and the preset processing index threshold of the edge platform respectively.

[0129] See also Fig. 9 FIG. 1 is a flow chart of a factory intelligent management and control method provided by another embodiment of the present application. In one embodiment of the present application, the factory intelligent management and control method further includes:

[0130] S601, based on the on-site quality information with qualified indexes, searching in a preset database whether there is a quality information quality determination model.

[0131] S602: If the preset database does not have a quality information quality determination model, determine whether the accumulated data volume of the on-site quality information with qualified indexes reaches a preset quality training data volume. For example, the preset quality training data volume is 20.

[0132] S603: If the preset quality training data volume is reached, the on-site quality information with a qualified index of the preset quality training data volume is input into the quality information quality training model for training to obtain the initial training quality information quality index and the preset training quality threshold.

[0133] S604: Determine whether the initial training quality information quality index is greater than a preset training quality threshold.

[0134] S605, if the initial training quality index is greater than the preset training quality threshold, the quality information quality training model is solidified to form a quality information quality determination model, and the preset training quality threshold is updated to the preset quality index threshold accordingly, so as to update the quality information quality determination model and the preset quality index threshold to the edge platform.

[0135] In one embodiment of the present application, the quality information quality determination model in the preset database carries a preset quality index threshold, and the factory intelligent management and control method further includes:

[0136] S606: If the preset database has a quality information quality determination model, the on-site quality information with a qualified index is input into the quality information quality determination model in the preset database to obtain a training quality information quality index.

[0137] S607: Determine whether the quality index of the training quality information is greater than a preset quality index threshold.

[0138] S608: If the quality index of the training quality information is greater than a preset quality index threshold, the quality information corresponding to the quality index of the training quality information greater than the preset quality index threshold is determined as qualified training quality information.

[0139] S609, judging whether the accumulated data volume of qualified training quality information reaches the preset quality retraining data volume, wherein the preset quality retraining data volume is a preset multiple of the preset quality training data volume, and the preset multiple is, for example, 3 times.

[0140] S610: If the preset processing and retraining data volume is reached, the training quality information of the preset processing and retraining data volume that meets the requirements is input into the quality information quality training model for retraining to obtain a new training quality information quality index and a new training quality threshold.

[0141] S611, determining whether the new training quality information quality index is greater than the new training quality threshold.

[0142] S612, if the quality index of the newly trained quality information is greater than the newly trained quality threshold, the retrained processing parameter quality training model will be solidified to form a new version of the quality information quality determination model, and the newly trained processing threshold will be updated to the new version of the preset processing index threshold, so as to update the edge platform's quality information quality determination model and the preset processing index threshold respectively.

[0143] In one embodiment of the present application, the factory intelligent management and control method further includes:

[0144] The field processing feature values ​​with qualified similarity of the edge platform are obtained for training to obtain a new version of the processing parameter similarity determination model and a new preset similarity threshold, so as to update the processing parameter similarity determination model in the edge platform.

[0145] Specifically, based on the edge platform similarity qualified field processing feature values ​​newly obtained from the edge platform, the updated dimension reduction matrix M and the historical processing feature mean in the calculation formula (2) are determined: The maximum value of the processing parameter similarity SPI in the historical data is determined, and three times the maximum value of the processing parameter similarity SPI is determined as a new preset similarity threshold.

[0146] See also Fig.10 FIG. 1 is a flow chart of a factory intelligent management and control method provided by another embodiment of the present application. In one embodiment of the present application, the factory intelligent management and control method further includes:

[0147] S701, obtaining qualified sample information of the edge platform, where the qualified sample information includes the on-site processing feature value with a qualified index, the on-site processing feature value with a qualified similarity, and the on-site quality information with a qualified index of the edge platform.

[0148] S702, determining whether the data volume of qualified sample information reaches a preset quality prediction training data volume.

[0149] S703: If the data volume of the qualified sample information reaches the preset quality prediction training data volume, based on the qualified sample information, a preset database is searched for a quality prediction model.

[0150] S704: If the preset database does not have a quality prediction model, input qualified sample information of a preset quality prediction training data volume into the quality prediction model training to perform training, so as to obtain an initial quality prediction model.

[0151] S705, verify whether the initial quality prediction model is qualified.

[0152] S706: Solidify the qualified initial quality prediction model to form a new version of the quality prediction model to update the quality prediction model of the edge platform.

[0153] For example, the quality prediction training model can be xgboost, which uses the k-fold method to randomly split the filtered historical data into 5 equal parts, take out 4 equal parts as training sets, and the remaining 1 equal part as the test set. Five quality prediction models are trained, and the model with the smallest loss is selected as the final model. During training, the modeling parameters are adjusted point by point, and the parameters with the lowest cross-validation value are selected as the optimal parameters. At this time, the final model is determined as the quality prediction model.

[0154] In one embodiment of the present application, the factory intelligent management and control method further includes:

[0155] S707, if the preset database has a quality prediction model, input the qualified sample information of the preset quality prediction training data volume into the quality prediction model of the preset database for incremental training to obtain a new version of the quality prediction model.

[0156] S708, inputting the on-site processing feature values ​​with qualified indexes and the on-site processing feature values ​​with qualified similarities into the new version of the quality prediction model to obtain the third quality prediction result of the workpiece.

[0157] S709, based on the third quality prediction result of the workpiece, the first quality prediction result of the workpiece of the edge platform and the qualified on-site quality information, determine the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model.

[0158] S710, determining whether the prediction accuracy of the new version of the quality prediction model is greater than the prediction accuracy of the quality prediction model in the preset database, so as to update the quality prediction model of the edge platform.

[0159] Specifically, if the prediction accuracy of the new version of the quality prediction model is greater than the prediction accuracy of the quality prediction model in the preset database, the new version of the quality prediction model is sent to the edge platform to update the quality prediction model of the edge platform. If the prediction accuracy of the new version of the quality prediction model is less than or equal to the prediction accuracy of the quality prediction model in the preset database, there is no need to send the new version of the quality prediction model to the edge platform.

[0160] In one embodiment of the present application, the factory intelligent management and control method further includes:

[0161] The index-qualified and similarity-qualified on-site processing feature values ​​and index-qualified quality information are input into the reference prediction training model, the reference prediction training model is trained to obtain a new version of the reference prediction model, and a new version of the credibility index determination model is solidified to update the credibility index determination model in the edge platform.

[0162] For example, the reference prediction training model can be a neural network model. The on-site processing feature values ​​with qualified index and qualified similarity and the quality information with qualified index are input into the neural network training model, and the neural network training model is trained to obtain a new version of the neural network model to solidify a new version of the credibility index determination model.

[0163] In one embodiment of the present application, there are multiple preset processing modes for extracting processing features, and the method further includes:

[0164] Determine whether the prediction accuracy of the edge platform's quality prediction model and the prediction accuracy of the new version of the quality prediction model are both less than the preset accuracy. If the prediction accuracy of the edge platform's quality prediction model and the prediction accuracy of the new version of the quality prediction model are both less than the preset accuracy, replace and solidify the processing feature extraction processing method.

[0165] See also Fig.11 The figure is a schematic diagram of the hardware structure of the factory intelligent control device provided in one embodiment of the present application. The factory intelligent control method provided in any embodiment of the present application is applied to the factory intelligent control device 1, and the factory intelligent control device 1 includes, but is not limited to, a processor 110 and a memory 120 connected via a communication bus 130. Fig.11 This is merely an example of a factory intelligent control device and does not constitute a corresponding limitation. In other embodiments, the factory intelligent control device may include more components than shown in the figure.

[0166] The memory 120 may include one or more random access memories (RAM) and one or more non-volatile memories (NVM). The random access memory can be directly read and written by the processor 110, and can be used to store executable programs (such as machine instructions) of the operating system or other running programs, and can also be used to store user and application data. The random access memory may include static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), etc.

[0167] The non-volatile memory may also store executable programs and user and application data, etc., and may be pre-loaded into the random access memory for direct reading and writing by the processor 110. The non-volatile memory may include a disk storage device and a flash memory.

[0168] The memory 120 is used to store one or more computer programs. The one or more computer programs are configured to be executed by the processor 110. The one or more computer programs include multiple instructions, and when the multiple instructions are executed by the processor 110, the factory intelligent management and control method executed on the factory intelligent management and control device 1 can be implemented.

[0169] In other embodiments, the factory intelligent management and control device 1 also includes an external memory interface for connecting to an external memory to expand the storage capacity of the factory intelligent management and control device 1.

[0170] The processor 110 may include one or more processing units, for example, the processor 110 may include an application processor (AP), a modem processor, a graphics processor (GPU), an image signal processor (ISP), a controller, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Different processing units may be independent devices or integrated into one or more processors.

[0171] The processor 110 provides computing and control capabilities. For example, the processor 110 is used to execute a computer program stored in the memory 120 to implement the above-mentioned factory intelligent management and control method.

[0172] The communication bus 130 is at least used to provide a channel for mutual communication between the memory 120 and the processor 110 in the factory intelligent management and control device 1.

[0173] It is understood that the structure illustrated in the embodiment of the present application does not constitute a specific limitation on the factory intelligent management and control device 1. In other embodiments of the present application, the factory intelligent management and control device 1 may include more or fewer components than shown in the figure, or combine certain components, or split certain components, or arrange the components differently. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0174] An embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. The computer program includes program instructions. The method implemented when the program instructions are executed can refer to the methods in the above-mentioned embodiments of the present application.

[0175] The computer-readable storage medium may be the internal memory of the factory intelligent control device described in the above embodiment, such as the hard disk or memory of the factory intelligent control device. The computer-readable storage medium may also be an external storage device of the factory intelligent control device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the factory intelligent control device.

[0176] In one embodiment of the present application, the computer-readable storage medium may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function, etc.; the data storage area may store data created according to the use of the factory intelligent management and control device, etc.

[0177] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0178] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0179] In the embodiments provided in the present application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0180] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0181] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A factory intelligent management and control system, characterized in that: An edge platform is included, wherein the edge platform includes: A feature processing module is used to obtain the field data of the workpiece processed by the processing machine, and perform feature extraction processing on the field processing parameter information in the field data to obtain multiple field processing feature values; A processing parameter quality determination module having a processing parameter quality determination model, for outputting a field processing parameter quality index based on a plurality of field processing characteristic values; and for determining a field processing characteristic value with a qualified index based on the field processing parameter quality index and a preset processing index threshold; A processing parameter similarity determination module has a processing parameter similarity determination model, which is used to output a field processing parameter similarity index based on the field processing characteristic value; and determine the field processing characteristic value with qualified similarity based on the field processing parameter similarity index and a preset similarity threshold; The quality prediction module has a quality prediction model, and is used to obtain a first quality prediction result of the workpiece based on the on-site processing feature value with a qualified index and the on-site processing feature value with a qualified similarity.

2. The factory intelligent management and control system according to claim 1, characterized in that: The edge platform also includes: The credibility module has a reference prediction training model and a credibility index determination model for: Inputting the on-site processing feature value with qualified index and the on-site processing feature value with qualified similarity into the reference prediction training model to output a second quality prediction result of the workpiece; The first quality prediction result of the workpiece and the second quality prediction result of the workpiece are input into the credibility index determination model to determine the credibility index of the quality prediction model.

3. The factory intelligent management and control system according to claim 2, characterized in that: The field data also includes field quality information of the workpiece, and the edge platform further includes: A qualified quality determination module, which determines qualified on-site quality information based on the on-site quality information and a preset quality standard; The quality information quality determination module has a quality information quality determination model for: Based on the qualified on-site quality information, a quality index of the on-site quality information is determined; based on the on-site quality information quality index and a preset quality index threshold, the on-site quality information with a qualified index is determined.

4. The factory intelligent management and control system according to claim 3, characterized in that: It also includes a central control platform, which includes a data cleaning processing module, and the data cleaning processing module is used to: Acquiring the on-site processing parameter information of the edge platform to determine high-frequency data and low-frequency data; Determining whether the amount of the high-frequency data reaches a preset amount of high-frequency data; Analyzing whether the high-frequency data is qualified data based on whether the data volume of the high-frequency data reaches the high-frequency preset data volume; The qualified high-frequency data and the low-frequency data are aligned with the timestamp and the identification of the workpiece to form aligned on-site processing parameter information.

5. The factory intelligent management and control system according to claim 4, characterized in that: The central control platform also includes: A feature processing and solidification module is used to perform feature extraction processing on the aligned on-site processing parameter information according to a preset processing feature extraction processing method to obtain a training processing feature value; Processing parameter quality determination model training module for: Based on the training processing feature value, determining whether there is a processing parameter quality determination model corresponding to the training processing feature value in a preset database; If the preset database does not have the processing parameter quality determination model corresponding to the training processing feature value, determining whether the accumulated data volume of the training processing feature value reaches the preset processing training data volume; If the preset processing training data volume is reached, the training processing feature value of the preset processing training data volume is input into the processing parameter quality training model for training to obtain the initial training processing parameter quality index and the preset training processing threshold value; Determining whether the initial training processing parameter quality index is greater than the preset training processing threshold; If the initial training processing parameter quality index is greater than the preset training processing threshold, the processing parameter quality training model is solidified to form the processing parameter quality determination model, and the preset training processing threshold is correspondingly updated to the preset processing index threshold, so as to update the processing parameter quality determination model and the preset processing index threshold to the edge platform.

6. The factory intelligent management and control system according to claim 5, characterized in that: The processing parameter quality determination model in the preset database carries the preset processing index threshold, The processing parameter quality determination model training module is also used for: If the preset database has the processing parameter quality determination model, input the training processing feature value into the processing parameter quality determination model in the preset database to obtain a training processing parameter quality index; Determining whether the training processing parameter quality index is greater than the preset processing index threshold; If the training processing parameter quality index is greater than the preset processing index threshold, the training processing feature value corresponding to the training processing parameter quality index greater than the preset processing index threshold is determined as a qualified training processing feature value; Determine whether the accumulated data volume of the qualified training processing feature values ​​reaches the preset processing retraining data volume; If the preset processing retraining data volume is reached, the training processing feature value of the preset processing retraining data volume that meets the requirements is input into the processing parameter quality training model for retraining to obtain a new training processing parameter quality index and a new training processing threshold; Determining whether the new training processing parameter quality index is greater than the new training processing threshold; If the newly trained processing parameter quality index is greater than the newly trained processing threshold, the re-trained processing parameter quality training model will be solidified to form a new version of the processing parameter quality determination model, and the newly trained processing threshold will be formed into a new version of the preset processing index threshold, so as to update the processing parameter quality determination model and the preset processing index threshold of the edge platform respectively.

7. The factory intelligent management and control system according to claim 6, characterized in that: The central control platform also includes: Quality information quality determination model training module, used for: Based on the on-site quality information with qualified index, searching the preset database for the quality information quality determination model; If the preset database does not have the quality information quality determination model, determine whether the accumulated data volume of the on-site quality information with the qualified index reaches the preset quality training data volume; If the preset quality training data volume is reached, the on-site quality information with the qualified index of the preset quality training data volume is input into the quality information quality training model for training to obtain the initial training quality information quality index and the preset training quality threshold; Determining whether the initial training quality information quality index is greater than the preset training quality threshold; If the initial training quality index is greater than the preset training quality threshold, the quality information quality training model is solidified to form the quality information quality determination model, and the preset training quality threshold is updated to the preset quality index threshold accordingly, so as to update the quality information quality determination model and the preset quality index threshold to the edge platform.

8. The factory intelligent management and control system according to claim 7, characterized in that: The quality information quality determination model in the preset database carries the preset quality index threshold, The quality information quality determination model training module is also used for: If the preset database has the quality information quality determination model, input the on-site quality information with the qualified index into the quality information quality determination model in the preset database to obtain the training quality information quality index; Determining whether the quality index of the training quality information is greater than the preset quality index threshold; If the quality index of the training quality information is greater than the preset quality index threshold, the quality information corresponding to the training quality information quality index greater than the preset quality index threshold is determined as qualified training quality information; Determine whether the accumulated data volume of the qualified training quality information reaches a preset quality retraining data volume; If the preset processing retraining data volume is reached, the training quality information of the preset processing retraining data volume that meets the requirements is input into the quality information quality training model for retraining to obtain a new training quality information quality index and a new training quality threshold; Determining whether the new training quality information quality index is greater than the new training quality threshold; If the quality index of the newly trained quality information is greater than the newly trained quality threshold, the re-trained processing parameter quality training model will be solidified to form a new version of the quality information quality determination model, and the newly trained processing threshold will be updated to the new version of the preset processing index threshold, so as to update the quality information quality determination model and the preset processing index threshold of the edge platform respectively.

9. The factory intelligent management and control system according to claim 4, characterized in that: The central control platform also includes: The processing parameter similarity training model is used to obtain the on-site processing feature values ​​with qualified similarity on the edge platform for training to obtain a new version of the processing parameter similarity determination model and a new preset similarity threshold so as to update the processing parameter similarity determination model in the edge platform.

10. The factory intelligent management and control system according to claim 5, characterized in that: The central control platform also includes: Quality prediction model training module for: Acquire qualified sample information of the edge platform, the qualified sample information including the on-site processing feature value of the edge platform with a qualified index, the on-site processing feature value with a qualified similarity, and the on-site quality information with a qualified index; Determine whether the data volume of the qualified sample information reaches a preset quality prediction training data volume; If the data amount of the qualified sample information reaches the preset quality prediction training data amount, based on the qualified sample information, searching in the preset database whether there is the quality prediction model; If the preset database does not have the quality prediction model, inputting the qualified sample information of the preset quality prediction training data amount into the quality prediction training model for training to obtain an initial quality prediction model; Verifying whether the initial quality prediction model is qualified; The qualified initial quality prediction model is solidified to form a new version of the quality prediction model to update the quality prediction model of the edge platform.

11. The factory intelligent management and control system according to claim 10, characterized in that: The quality prediction model training module is also used for: If the preset database has the quality prediction model, input the qualified sample information of the preset quality prediction training data volume into the quality prediction model of the preset database for training to obtain a new version of the quality prediction model; Inputting the on-site processing characteristic value with a qualified index and the on-site processing characteristic value with a qualified similarity into the new version of the quality prediction model to obtain a third quality prediction result of the workpiece; Determine the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model based on the third quality prediction result of the workpiece, the first quality prediction result of the workpiece of the edge platform, and the on-site quality information of the qualified index; Determine whether the prediction accuracy of the new version of the quality prediction model is greater than the prediction accuracy of the quality prediction model in the preset database, so as to update the quality prediction model of the edge platform.

12. The factory intelligent management and control system according to claim 9, characterized in that: The central control platform also includes: a credibility index determination model training module, which is used to input the on-site processing feature values ​​with qualified indexes and qualified similarities and the quality information with qualified indexes into a reference prediction training model, train the reference prediction training model to obtain a new version of the reference prediction model, and solidify the new version of the credibility index determination model to update the credibility index determination model in the edge platform.

13. The factory intelligent management and control system according to claim 11, characterized in that: The preset processing method for extracting processing features is multiple sets. The quality prediction model training module is also used for: Determine whether the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model are both less than a preset accuracy; The feature processing and curing module is also used If the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model are both lower than the preset accuracy, the processing feature extraction method is replaced and solidified.

14. A factory intelligent management and control method, characterized in that: The method comprises: Acquiring on-site data of a workpiece processed by a processing machine, wherein the on-site data includes on-site processing parameter information; Performing feature extraction processing on the on-site processing parameter information to obtain a plurality of on-site processing feature values; Based on the plurality of on-site processing characteristic values, outputting an on-site processing parameter quality index; Based on the on-site processing parameter quality index and a preset processing index threshold, determining the on-site processing characteristic value with a qualified index; Based on the on-site processing characteristic value, outputting an on-site processing parameter similarity index; Based on the on-site processing parameter similarity index and a preset similarity threshold, determining the on-site processing characteristic value with qualified similarity; A first quality prediction result of a workpiece is obtained based on the on-site processing feature value with a qualified index and the on-site processing feature value with a qualified similarity.

15. The factory intelligent management and control method according to claim 14, characterized in that: Also includes: Inputting the on-site processing characteristic value with the qualified index and the on-site processing characteristic value with the qualified similarity into a reference prediction model to output a second quality prediction result of the workpiece; The first quality prediction result of the workpiece and the second quality prediction result of the workpiece are input into a credibility index determination model to determine the credibility index of the quality prediction model.

16. The factory intelligent management and control method according to claim 15, characterized in that: The field data also includes field quality information of the workpiece, and the method further includes: Based on the on-site quality information and a preset quality standard, determining qualified on-site quality information; Determining a field quality information quality index of the quality information based on the qualified field quality information; Based on the quality index of the on-site quality information and a preset quality index threshold, the on-site quality information with a qualified index is determined.

17. The factory intelligent management and control method according to claim 16, characterized in that: Also includes: Acquire the on-site processing parameter information of the edge platform to determine high-frequency data and low-frequency data; Determining whether the amount of the high-frequency data reaches a preset amount of high-frequency data; Analyzing whether the high-frequency data is qualified data based on whether the data volume of the high-frequency data reaches the high-frequency preset data volume; The qualified high-frequency data and the low-frequency data are aligned with the timestamp and the identification of the workpiece to form aligned on-site processing parameter information.

18. The factory intelligent management and control method according to claim 17, characterized in that: Also includes: Performing feature extraction processing on the aligned on-site processing parameter information according to a preset processing feature extraction processing method to obtain a training processing feature value; Based on the training processing feature value, determining whether there is a processing parameter quality determination model corresponding to the training processing feature value in a preset database; If the preset database does not have the processing parameter quality determination model corresponding to the training processing feature value, determining whether the accumulated data volume of the training processing feature value reaches the preset processing training data volume; If the preset processing training data volume is reached, the training processing feature value of the preset processing training data volume is input into the processing parameter quality training model for training to obtain the initial training processing parameter quality index and the preset training processing threshold value; Determining whether the initial training processing parameter quality index is greater than the preset training processing threshold; If the initial training processing parameter quality index is greater than the preset training processing threshold, the processing parameter quality training model is solidified to form the processing parameter quality determination model, and the preset training processing threshold is correspondingly updated to the preset processing index threshold, so as to update the processing parameter quality determination model and the preset processing index threshold to the edge platform.

19. The factory intelligent management and control method according to claim 18, characterized in that: The processing parameter quality determination model in the preset database carries the preset processing index threshold, and the method further includes: If the preset database has the processing parameter quality determination model, input the training processing feature value into the processing parameter quality determination model in the preset database to obtain a training processing parameter quality index; Determining whether the training processing parameter quality index is greater than the preset processing index threshold; If the training processing parameter quality index is greater than the preset processing index threshold, the training processing feature value corresponding to the training processing parameter quality index greater than the preset processing index threshold is determined as a qualified training processing feature value; Determine whether the accumulated data volume of the qualified training processing feature values ​​reaches the preset processing retraining data volume; If the preset processing retraining data volume is reached, the training processing feature value of the preset processing retraining data volume that meets the requirements is input into the processing parameter quality training model for retraining to obtain a new training processing parameter quality index and a new training processing threshold; Determining whether the new training processing parameter quality index is greater than the new training processing threshold; If the newly trained processing parameter quality index is greater than the newly trained processing threshold, the re-trained processing parameter quality training model will be solidified to form a new version of the processing parameter quality determination model, and the newly trained processing threshold will be formed into a new version of the preset processing index threshold, so as to update the processing parameter quality determination model and the preset processing index threshold of the edge platform respectively.

20. The factory intelligent management and control method according to claim 19, characterized in that: The method further comprises: Based on the on-site quality information with qualified index, searching the preset database for the quality information quality determination model; If the preset database does not have the quality information quality determination model, determine whether the accumulated data volume of the on-site quality information with the qualified index reaches the preset quality training data volume; If the preset quality training data volume is reached, the on-site quality information with the qualified index of the preset quality training data volume is input into the quality information quality training model for training to obtain the initial training quality information quality index and the preset training quality threshold; Determining whether the initial training quality information quality index is greater than the preset training quality threshold; If the initial training quality index is greater than the preset training quality threshold, the quality information quality training model is solidified to form the quality information quality determination model, and the preset training quality threshold is updated to the preset quality index threshold accordingly, so as to update the quality information quality determination model and the preset quality index threshold to the edge platform.

21. The factory intelligent management and control method according to claim 20, characterized in that: The quality information quality determination model in the preset database carries the preset quality index threshold, and the method further includes: If the preset database has the quality information quality determination model, input the on-site quality information with the qualified index into the quality information quality determination model in the preset database to obtain the training quality information quality index; Determining whether the quality index of the training quality information is greater than the preset quality index threshold; If the quality index of the training quality information is greater than the preset quality index threshold, the quality information corresponding to the training quality information quality index greater than the preset quality index threshold is determined as qualified training quality information; Determine whether the accumulated data volume of the qualified training quality information reaches a preset quality retraining data volume; If the preset processing retraining data volume is reached, the training quality information of the preset processing retraining data volume that meets the requirements is input into the quality information quality training model for retraining to obtain a new training quality information quality index and a new training quality threshold; Determining whether the new training quality information quality index is greater than the new training quality threshold; If the quality index of the newly trained quality information is greater than the newly trained quality threshold, the re-trained processing parameter quality training model will be solidified to form a new version of the quality information quality determination model, and the newly trained processing threshold will be updated to the new version of the preset processing index threshold, so as to update the quality information quality determination model and the preset processing index threshold of the edge platform respectively.

22. The factory intelligent management and control method according to claim 17, characterized in that: The method further comprises: The on-site processing feature values ​​with qualified similarity of the edge platform are obtained for training to obtain a new version of the processing parameter similarity determination model and a new preset similarity threshold, so as to update the processing parameter similarity determination model in the edge platform.

23. The factory intelligent management and control method according to claim 18, characterized in that: The method further comprises: Acquire qualified sample information of the edge platform, the qualified sample information including the on-site processing feature value of the edge platform with a qualified index, the on-site processing feature value with a qualified similarity, and the on-site quality information with a qualified index; Determine whether the data volume of the qualified sample information reaches a preset quality prediction training data volume; If the data amount of the qualified sample information reaches the preset quality prediction training data amount, based on the qualified sample information, searching in the preset database whether there is the quality prediction model; If the preset database does not have the quality prediction model, inputting the qualified sample information of the preset quality prediction training data volume into the quality prediction model training to perform training, so as to obtain an initial quality prediction model; Verifying whether the initial quality prediction model is qualified; The qualified initial quality prediction model is solidified to form a new version of the quality prediction model to update the quality prediction model of the edge platform.

24. The factory intelligent management and control method according to claim 23, characterized in that: The method further comprises: If the preset database has the quality prediction model, input the qualified sample information of the preset quality prediction training data volume into the quality prediction model of the preset database for training to obtain a new version of the quality prediction model; Inputting the on-site processing characteristic value with a qualified index and the on-site processing characteristic value with a qualified similarity into the new version of the quality prediction model to obtain a third quality prediction result of the workpiece; Determine the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model based on the third quality prediction result of the workpiece, the first quality prediction result of the workpiece of the edge platform, and the on-site quality information of the qualified index; Determine whether the prediction accuracy of the new version of the quality prediction model is greater than the prediction accuracy of the quality prediction model in the preset database, so as to update the quality prediction model of the edge platform.

25. The factory intelligent management and control method according to claim 22, characterized in that: The method also includes: inputting the on-site processing feature values ​​with qualified index and qualified similarity and the quality information with qualified index into a reference prediction training model, training the reference prediction training model to obtain a new version of the reference prediction model, and solidifying the new version of the credibility index determination model to update the credibility index determination model in the edge platform.

26. The factory intelligent management and control method according to claim 24, characterized in that: The preset processing method for extracting and processing the processing features is multiple sets, and the method further includes: Determine whether the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model are both less than a preset accuracy; If the prediction accuracy of the quality prediction model of the edge platform and the prediction accuracy of the new version of the quality prediction model are both lower than the preset accuracy, the processing feature extraction method is replaced and solidified.

27. A factory intelligent management and control device, characterized in that: It includes a memory and a processor, the memory is used to store instructions, and the instructions stored in the memory are executed by the processor to implement the factory intelligent management and control method as described in any one of claims 14 to 26.

28. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein, when the program is executed, the factory intelligent management and control device where the computer-readable storage medium is located is controlled to execute the factory intelligent management and control method as described in any one of claims 14 to 26.

Images