Training method, prediction method, device, and equipment for production process prediction model

Abstract

Embodiments of the present application provide a training method and a prediction method for a production process prediction model, a device and equipment. The method comprises: for each production process, inputting a first process parameter of the production process and a second process parameter of a previous process before the production process into a prediction model corresponding to the production process; obtaining a first prediction result for predicting whether a product flows into a subsequent process according to an output result of the prediction model; and performing supervised training on the prediction model corresponding to the production process according to the first prediction result and a true result of whether the product flows into the subsequent process, to obtain a trained prediction model. By combining the process parameters of the production process and the previous process to predict the product flow direction, the interaction between processes can be better balanced to determine the optimal process parameter value range, improve the accuracy and flexibility of process parameter setting, thereby reducing the defective product rate and improving the stability of the manufacturing process.

Description

Technical Field

[0001] The present application relates to the field of production technology, and in particular to a training method, prediction method, device and equipment for a production process prediction model. Background Art

[0002] The manufacture of a product involves multiple production processes, and each production process involves the setting of one or more process parameters, such as temperature, pressure, speed, and so on. The range of values ​​of process parameters will affect the quality of the product. To ensure that the product meets the specified standards and requirements, thresholds for process parameters are usually introduced to limit the range of values ​​of process parameters. Typically, these thresholds are fixed ranges derived from experience or statistical data, also known as rigid thresholds, which represent the reasonable range of values ​​of process parameters within an acceptable range. By strictly controlling the process parameters within the thresholds, the stability and consistency of the product can be guaranteed.

[0003] However, in reality, the interplay between production processes and the dynamic changes in process parameters can lead to limitations in rigid thresholds. Furthermore, complex nonlinear relationships can exist between different process parameters. Therefore, relying solely on rigid thresholds to define process parameters may not meet the requirements for product quality and process stability, and may even lead to low product yields due to over-adjustment. Summary of the Invention

[0004] The purpose of the embodiments of the present application is to provide a training method, prediction method, device and equipment for a production process prediction model, so as to achieve the technical effect of improving the range of process parameter values.

[0005] In a first aspect, an embodiment of the present application provides a method for training a production process prediction model, wherein multiple production processes of a product are sequential in time, each production process corresponds to a prediction model, and the prediction model is used to predict whether the product will flow into a subsequent process after the corresponding production process; the method includes:

[0006] For each of the production processes, inputting a first process parameter of the production process and a second process parameter of a previous process before the production process into a prediction model corresponding to the production process;

[0007] Obtaining a first prediction result for predicting whether the product will flow into the subsequent process according to an output result of the prediction model;

[0008] The prediction model corresponding to the production process is supervisedly trained based on the first prediction result and the actual result of whether the product flows into the subsequent process to obtain a trained prediction model.

[0009] In the above implementation process, by combining the process parameters of the production process and its previous process to predict the product flow, the interaction between the processes can be better weighed to determine the optimal range of process parameter values, thereby improving the accuracy and flexibility of process parameter settings, thereby reducing the defective rate and improving the stability of the manufacturing process.

[0010] Furthermore, obtaining a first prediction result for predicting whether the product will flow into the subsequent process based on the output of the prediction model includes: obtaining a quality parameter representing the quality of the product and a second prediction result corresponding to the prior process; and determining the first prediction result based on the output, the quality parameter, and the second prediction result. By combining the process parameters of the production process and its prior process with product quality to predict product flow, the interaction between processes can be better balanced to determine the optimal range of process parameter values, thereby improving the accuracy and flexibility of process parameter setting, thereby reducing the defective product rate and improving the stability of the manufacturing process.

[0011] Furthermore, determining the first prediction result based on the output result, the quality parameter, and the second prediction result includes: determining a cumulative error of the output result based on the output result, the quality parameter, and the second prediction result; and determining the first prediction result based on the output result and the cumulative error. By calculating the cumulative error of the output result and combining the output result of the prediction model with the cumulative error to determine the first prediction result, the accuracy of the first prediction result is improved.

[0012] Furthermore, determining the first prediction result based on the output result and the cumulative error includes: if the sum of the output result and the cumulative error is less than or equal to a preset threshold, determining that the first prediction result indicates that the product will flow into the subsequent process; if the difference between the output result and the cumulative error is greater than the preset threshold, determining that the first prediction result indicates that the product will not flow into the subsequent process; if the difference between the output result and the cumulative error is less than or equal to the preset threshold, and the sum of the output result and the cumulative error is greater than the preset threshold, determining that the first prediction result indicates that the product will not flow into the subsequent process. By combining the output result and the cumulative error to jointly determine whether the product will continue to flow into the subsequent process after passing through the production process, the prediction accuracy of the prediction model is improved.

[0013] Furthermore, the prediction model is a binary classification model; the first prediction result includes a first result indicating that the product will flow into the subsequent process, and a second result indicating that the product will not flow into the subsequent process. Using the binary classification model as the prediction model, the model output can be used to intuitively determine whether the product will continue to flow into the subsequent process.

[0014] The second aspect of the embodiment of the present application provides a production process prediction method. A plurality of production processes of a product have time sequence, each production process corresponds to a prediction model, and the prediction model is used to predict whether the product flows into a subsequent process after the corresponding production process. The method comprises the following steps:

[0015] obtaining a third process parameter of a current production process and a fourth process parameter of a previous process before the current production process;

[0016] inputting the third process parameter and the fourth process parameter into a trained prediction model corresponding to the current production process to obtain an output prediction result; the prediction result is used to indicate whether the product flows into or out of the subsequent process.

[0017] In the above process, when predicting the flow direction of the product, the process parameters of the current production process and the previous process are considered together. The value range of the process parameter of each process is no longer a rigid threshold, but a flexible threshold that can be adjusted flexibly according to the process parameter of the previous process. Thus, the problem of product “overkill” caused by the rigid threshold is solved.

[0018] Further, the method is applied to an edge server; the edge server and a production device of the product satisfy a preset geographical relationship. By shortening the geographical distance between the production device and the server, the data transmission time can be reduced. Thus, when obtaining the input data required by the prediction model, the input data can be immediately sent to the edge server for calculation and prediction, the prediction response efficiency is improved, and the production timeliness is met.

[0019] The third aspect of the embodiment of the present application provides a production process prediction model training device. A plurality of production processes of a product have time sequence, each production process corresponds to a prediction model, and the prediction model is used to predict whether the product flows into a subsequent process after the corresponding production process. The device comprises the following steps:

[0020] an input module configured to input, for each production process, a first process parameter of the production process and a second process parameter of a previous process before the production process into a prediction model corresponding to the production process;

[0021] a first prediction module configured to obtain, according to an output result of the prediction model, a first prediction result for predicting whether the product flows into the subsequent process;

[0022] a training module configured to perform supervised training on the prediction model corresponding to the production process according to the first prediction result and a true result of whether the product flows into the subsequent process, to obtain a trained prediction model.

[0023] The fourth aspect of the embodiments of the present application provides a production process prediction device, a plurality of production processes of a product have time sequence, each of the production processes corresponds to a prediction model, and the prediction model is used to predict whether the product flows into a subsequent process after the corresponding production process; the device comprises:

[0024] an acquisition module, configured to acquire a third process parameter of a current production process and a fourth process parameter of a previous process before the current production process;

[0025] a second prediction module, configured to input the third process parameter and the fourth process parameter into a trained prediction model corresponding to the current production process to obtain an output prediction result; the prediction result is used to indicate whether the product flows into or out of the subsequent process.

[0026] The fifth aspect of the embodiments of the present application provides a computer program product, the computer program product comprises a computer program, wherein the computer program is executed by a processor to implement the method of any one of the first aspect or the second aspect.

[0027] The sixth aspect of the embodiments of the present application provides an electronic device, the electronic device comprises:

[0028] a processor;

[0029] a memory for storing processor executable instructions;

[0030] When the processor calls the executable instructions, the operations of the method of any one of the first aspect or the second aspect are implemented.

[0031] The seventh aspect of the embodiments of the present application provides a computer readable storage medium, which stores computer instructions, and the computer instructions are executed by a processor to implement the steps of the method of any one of the first aspect or the second aspect. BRIEF DESCRIPTION OF DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0033] Figure 1 A flowchart of a production process prediction model training method provided by the embodiments of the present application;

[0034] Figure 2A flowchart of another method for training a production process prediction model provided in an embodiment of the present application;

[0035] Figure 3 A flowchart of another method for training a production process prediction model provided in an embodiment of the present application;

[0036] Figure 4 This is a schematic diagram of outputting data in an embodiment of the present application;

[0037] Figure 5 A schematic diagram of a production process prediction method provided in an embodiment of the present application;

[0038] Figure 6 A flowchart illustrating a production process prediction model training process and a production process prediction process provided in an embodiment of the present application;

[0039] Figure 7 A structural block diagram of a training device for a production process prediction model provided in an embodiment of the present application;

[0040] Figure 8 A structural block diagram of a production process prediction device provided in an embodiment of the present application;

[0041] Figure 9 A hardware structure diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] The following embodiments of the technical solution of the present application will be described in detail with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present application and are therefore only examples and are not intended to limit the scope of protection of the present application.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0044] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0045] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0046] In the description of the embodiments of this application, the term "and / or" is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0047] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0048] In the description of the embodiments of the present application, the technical terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the embodiments of the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the embodiments of the present application.

[0049] In the description of the embodiments of the present application, unless otherwise expressly specified or limited, technical terms such as "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integration; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components or interactions between two components. Those skilled in the art can understand the specific meanings of the above terms in the embodiments of the present application based on specific circumstances.

[0050] Each production process involves setting different process parameters, and each process can influence the others. For example, the output of a previous production process may affect the value range of process parameters in a subsequent production process. For example, the manufacturing process of a battery cell involves multiple production processes, including raw material cutting, component assembly, and welding. For example, the mechanical properties and dimensions of the raw materials may affect the value range of process parameters used in the cutting process. Another example is the gap size between components in the assembly process, which may affect the value range of process parameters used in the subsequent welding process, including welding speed, power, and temperature. Therefore, setting rigid thresholds for process parameters has limitations. Using rigid thresholds to over-adjust process parameters can cause products that otherwise meet quality requirements to be classified as defective, resulting in an excessively low product yield rate, a problem known as "overkill."

[0051] To this end, the embodiment of the present application proposes setting up a prediction model for each production process of the product to predict whether the product will continue to flow into the next production process after passing the current production process. Among them, in the training stage of the prediction model, the first process parameter known in the production process and the second process parameter known in the previous process are used as training data; in the use stage of the prediction model, the third process parameter of the current production process and the fourth process parameter of the previous process are also input into the prediction model. In this way, when predicting the flow direction of the product, the process parameters of the current production process and the previous process are taken into consideration together. In other words, for the process parameters of the production process that can make the product flow into the subsequent process, the value range will be affected by the process parameters of the previous process. Therefore, the value range of the process parameters of each process is no longer a rigid threshold, but a flexible threshold that can be flexibly adjusted according to the process parameters of the previous process. Thereby solving the problem of "overkilling" of products caused by rigid thresholds.

[0052] According to some embodiments of the present application, a first aspect provides a method for training a production process prediction model. A product may have multiple production processes, and each production process is sequential. That is, the production processes are executed in a certain time sequence. Furthermore, each production process has a corresponding prediction model, which is used to predict whether the product will continue to flow into a subsequent process after passing through the production process. The training method provided in the embodiments of the present application can be used to train a prediction model corresponding to each production process.

[0053] See also Figure 1 The training method of a production process prediction model provided in the embodiment of the present application includes the following steps: Figure 1 Steps 110 to 130 are shown. The prediction model corresponding to each production process can be trained by executing steps 110 to 130.

[0054] Step 110: For each of the production processes, input the first process parameter of the production process and the second process parameter of the previous process before the production process into a prediction model corresponding to the production process.

[0055] Model training requires training data. In this embodiment, historical products can be used as training data. These historical products can be final products or semi-finished products that exit the production line due to process issues. Furthermore, these historical products can include both good and / or defective products.

[0056] Specifically, known process parameters of historical products in each production process can be obtained as training data. The amount of training data can be set according to training requirements, and this embodiment does not limit the amount.

[0057] For each production process of a product, the known process parameters of that production process can be obtained as the first process parameters; and the known process parameters of the previous production process can be obtained as the second process parameters. The previous process refers to the production process corresponding to the current model training, and there can be one or more previous processes. The second process parameters include the known process parameters of all previous processes.

[0058] Subsequently, the first process parameter and the second process parameter are used as training data of a prediction model corresponding to the production process and input into the prediction model.

[0059] Step 120: Obtain a first prediction result for predicting whether the product will flow into the subsequent process based on the output result of the prediction model.

[0060] After inputting the training data into the prediction model, an output result is obtained. Within the output value range, the closer the output result is to the lower limit of the range, the fewer defects the product will have after undergoing that production process. Similarly, the closer the output result is to the upper limit of the range, the more defects the product will have after undergoing that production process. Based on the output result, a first prediction result is obtained, which is used to predict whether the product will flow into a subsequent process. The subsequent process refers to the production process corresponding to the prediction model.

[0061] Optionally, the output result of the prediction model may be determined to be the first prediction result.

[0062] Optionally, the output result of the prediction model may be subjected to a preset process to obtain a first prediction result.

[0063] Step 130: Perform supervised training on the prediction model corresponding to the production process based on the first prediction result and the actual result of whether the product flows into the subsequent process to obtain a trained prediction model.

[0064] For historical products, the actual results of whether the product flows into the subsequent process after passing through the production process can be determined based on the actual situation. In this way, the first prediction result and the actual result obtained by the prediction model can be used to conduct supervised training of the prediction model.

[0065] For example, a loss function can be used to characterize the difference between the actual result and the first predicted result. The specific form of the loss function can be selected by a technician based on actual needs and is not limited in this embodiment. Subsequently, the loss function can be used to update the parameters of the prediction model until the training end conditions are met, thereby obtaining a trained prediction model. For example, the training result conditions may include, but are not limited to, the number of training times exceeding a preset number threshold, the convergence of the loss function, and the like.

[0066] In addition, for the first process in the product manufacturing process, since there is no previous process relative to the first process, only the first process parameter of the first process can be input into the prediction model corresponding to the first process to obtain the first prediction result.

[0067] It can be seen that the training method of a production process prediction model provided in this embodiment uses the first process parameter known to the production process and the second process parameter known to the previous process as training data for the prediction model corresponding to each production process, so that when predicting whether the product will flow into the subsequent process, the influence of the process parameters of the previous process on the value of the process parameters of the production process is also taken into account. In other words, whether the product flows into the subsequent process is determined by the process parameters of the production process and the process parameters of its previous process. Therefore, for the value range of the process parameters of the production process that can make the product flow into the subsequent process, it will be affected by the process parameters of its previous process. Therefore, by combining the process parameters of the production process and its previous process to predict the flow of products, the interaction between the processes can be better balanced to determine the optimal range of process parameter values, improve the accuracy and flexibility of process parameter setting, thereby reducing the defective rate and improving the stability of the manufacturing process.

[0068] Based on the above embodiment, the first prediction result obtained according to the output result of the prediction model in step 120 may include: Figure 2 Steps 121-122 are shown.

[0069] Step 121: Acquire a quality parameter for characterizing the quality of the product and a second prediction result corresponding to the previous process;

[0070] Step 122: determining the first prediction result according to the output result, the quality parameter, and the second prediction result.

[0071] For the historical product, the quality of the historical product can be detected and evaluated after production to obtain the quality parameter representing the product quality.

[0072] Optionally, the value range of the quality parameter includes a first value and a second value. For example, the first value is 0 and the second value is 1. Optionally, the first value can represent that the product quality is qualified, and the second value can represent that the product quality is unqualified. Optionally, the first value can represent that the number of defects of the product is less than a preset number threshold, and thus the product quality is qualified, and the second value can represent that the number of defects of the product is greater than the number threshold, and thus the product quality is unqualified.

[0073] In addition, as described above, each production process corresponds to a prediction model. For the previous process, the second prediction result corresponding to the previous process can be obtained through the prediction model corresponding to the previous process. The second prediction result is used to predict whether the product flows into the subsequent process of the previous process. The obtaining process of the second prediction result can refer to the obtaining process of the first prediction result. For example, the output result of the prediction model corresponding to the previous process can be determined as the second prediction result. For another example, the output result of the prediction model corresponding to the previous process can be obtained by the method described in the embodiment, and then the second prediction result can be obtained after a preset processing. If the previous process includes multiple processes, the second prediction result described in the embodiment includes multiple second prediction results output by the prediction models corresponding to the multiple previous processes.

[0074] Subsequently, the output result of the prediction model of the production process, the quality parameter, and the second prediction result corresponding to the previous process are processed in a preset manner, and the first prediction result for predicting whether the product flows into the subsequent process can be obtained.

[0075] It can be known that, after obtaining the output result of the prediction model, the second prediction result of the previous process and the quality parameter of the product are introduced again to process the output result, so that the first prediction result for predicting whether the product flows into the subsequent process is obtained. Therefore, by combining the process parameters of the production process and the previous process with the product quality to predict the product flow direction, the interaction between the processes can be better balanced to determine the optimal process parameter value range, improve the accuracy and flexibility of the process parameter setting, thereby reducing the defective product rate and improving the stability of the manufacturing process.

[0076] On the basis of the above embodiments, the process of determining the first prediction result according to the output result, the quality parameter and the second prediction result in step 122 can specifically include steps 1221-1222 as shown in the table. Figure 3

[0077] Step 1221: determining the cumulative error of the output result according to the output result, the quality parameter and the second prediction result.

[0078] As an example, the cumulative error can be a root mean square error.

[0079] Exemplarily, the root of the ratio of the sum of the squares of the differences between the quality parameter and each second prediction result and the sum of the squares of the differences between the quality parameter and the output result can be the root mean square error. The root mean square error δ n of the nth production process can be represented as: n

[0080]

[0081] wherein y is the quality parameter; y is the output result of the prediction model to be trained this time, y, y … y are the second prediction results corresponding to the previous processes; and n is the execution order of the production process in the entire production process. obs model,i model,1 model,2 model,n model,n model,1 model,2 model,n-1

[0082] Step 1222: determining the first prediction result according to the output result and the cumulative error.

[0083] The cumulative error represents the cumulative prediction error from the first production process to the nth production process. Therefore, the first prediction result can be determined according to the output result of the prediction model and the cumulative error.

[0084] It can be known that, by calculating the cumulative error of the output result and determining the first prediction result according to the output result of the prediction model and the cumulative error, the accuracy of the first prediction result is improved.

[0085] On the basis of the above embodiments, the process of determining the first prediction result according to the output result and the cumulative error in step 1222 can specifically include the following three cases.

[0086] Case one: if the sum of the output result and the cumulative error is less than or equal to a preset threshold, it is determined that the first prediction result indicates flowing into the subsequent process. ​​​​​​​​​​​

[0087] According to the cumulative error, the error bar of the prediction model output can be determined. For example, Figure 4 As shown, it can be determined from the output result f(x1,x2,…,x n ) and the cumulative error δ n The difference f(x1,x2,…,x n )-δ n , to the sum of the output result and the cumulative error f(x1,x2,…,x n )+δ n The range between is the error bar of the prediction model output. That is, the difference between the output and the cumulative error f(x1,x2,…,x n )-δ n is the lower limit of the error bar, and the output result is the sum of the cumulative errors f(x1,x2,…,x n )+δ n is the upper limit of the error bar.

[0088] Continue to see Figure 4 , if the output result is equal to the sum of the cumulative errors f(x1,x2,…,x n )+δ n Less than or equal to the preset threshold indicates that the error bar of the output result is less than or equal to the preset threshold. In this case, the output result with the error bar is closer to the lower limit of the value range, indicating that the product has fewer defects after passing through this production process. Therefore, it can be determined that the first prediction result indicates that the product will flow into the subsequent process.

[0089] The preset threshold value may be determined by those skilled in the art according to the value range of the output result, and the present invention does not limit the value of the preset threshold value.

[0090] Case 2: If the difference between the output result and the accumulated error is greater than the preset threshold, it is determined that the first prediction result indicates that the first prediction result does not flow into the subsequent process.

[0091] Continue to see Figure 4 , if the difference between the output result and the cumulative error f(x1,x2,…,x n )-δ n If the error bar is greater than the preset threshold, it indicates that the error bar of the output result is greater than the preset threshold. In this case, the output result with the error bar is closer to the upper limit of the value range, indicating that the product has more defects after passing through this production process. Therefore, it can be determined that the first prediction result indicates that the product will not flow into the subsequent process.

[0092] Case 3: If the difference between the output result and the cumulative error is less than or equal to the preset threshold, and the sum of the output result and the cumulative error is greater than the preset threshold, it is determined that the first prediction result indicates that it does not flow into the subsequent process.

[0093] Continue to see Figure 4 , if the difference between the output result and the cumulative error f(x1,x2,…,x n )-δ n is less than the preset threshold, and the output result is equal to the sum of the cumulative errors f(x1,x2,…,x n )+δ n If the output is greater than the preset threshold, it indicates that the number of defects in the product after passing through this production process is close to the threshold that determines whether it will be sent to the next process. In this case, to avoid "missing" products, that is, sending potentially unqualified products to the next process, this situation can be set to not be sent to the next process.

[0094] It can be seen that this embodiment combines the output result and the accumulated error to jointly determine whether the product continues to flow into the subsequent process after passing through the production process, thereby improving the prediction accuracy of the prediction model.

[0095] Based on any of the above embodiments, as a feasible example, the prediction model can be a binary classification model. The output of the prediction model can include a first value and a second value. For example, the first value is 0 and the second value is 1. Alternatively, the first value can represent that the product flows into a subsequent process, and the second value can represent that the product does not flow into a subsequent process.

[0096] In this way, after obtaining the output results of the prediction model, according to Figure 2-Figure 3 The embodiment described can calculate the cumulative error of the output result. Figure 4 In the described embodiment, a first prediction result can be determined based on the output of the prediction model and the accumulated error, wherein the first prediction result includes a first result indicating that the product will flow into a subsequent process and a second result indicating that the product will not flow into the subsequent process.

[0097] It can be seen that this embodiment uses a binary classification model as a prediction model, and based on the output of the model, it can be intuitively determined whether the product continues to flow into the subsequent process.

[0098] In addition, according to some embodiments of the present application, the second aspect further provides a method for predicting a production process. There are multiple production processes for a product, and each production process has a time sequence. That is, the production processes are executed in a certain time sequence. In addition, each production process corresponds to a prediction model, which is used to predict whether the product will continue to flow into the subsequent process after passing through the production process. The prediction method provided in this embodiment can be used to predict whether the product will continue to flow into the subsequent process after passing through the current production process. That is, every time the product passes through a production process, the prediction method can be executed to determine whether the product will continue to flow into the next production process.

[0099] Referring to Figure 5 The production process prediction method provided by the embodiment of the present application comprises steps 510-520 as shown in the figure. Figure 5

[0100] Step 510: Obtain third process parameters of a current production process and fourth process parameters of a previous process before the current production process.

[0101] In this step, the third process parameters of the product used in the current production process and the fourth process parameters of the previous process are obtained. The previous process is relative to the current production process, and the previous process includes one or more. The fourth process parameters include the process parameters of all previous processes.

[0102] Step 520: Input the third process parameters and the fourth process parameters into a trained prediction model corresponding to the current production process to obtain an output prediction result.

[0103] The prediction result is used to indicate that the product flows into or out of the subsequent process.

[0104] In this step, the third process parameters and the fourth process parameters are input as input data of the model into the trained prediction model corresponding to the current production process, and the prediction model can output a prediction result.

[0105] The prediction result indicates that the product flows into or out of the subsequent process. The subsequent process is relative to the current production process, which means the next production process of the current production process. According to the indication of the prediction result, the product completing the current production process can be processed to flow into the subsequent process or out of the production line.

[0106] It can be known that the embodiment proposes to set a prediction model for each production process of the product to predict whether the product continues to flow into the subsequent process after passing through the current production process. The third process parameters of the current production process and the fourth process parameters of the previous process are input into the prediction model at the time of prediction. In this way, when predicting the flow direction of the product, the process parameters of the current production process and the previous process are considered together. In other words, the value range of the process parameters of the production process that can make the product flow into the subsequent process will be affected by the process parameters of the previous process. Therefore, the value range of the process parameters of each process is no longer a rigid threshold, but a flexible threshold that can be adjusted flexibly according to the process parameters of the previous process. Thus, the problem of "overkill" of the product caused by the rigid threshold is solved.

[0107] Based on the above embodiment, optionally, the prediction method can be applied to an edge server. The edge server and the production equipment of the product satisfy a preset geographical relationship. ​

[0108] Among them, the production equipment can be communicatively connected with the server cluster. The server cluster may include multiple servers, each server is used to implement different functions. According to the geographical distribution of each server in the server cluster, the servers that meet the preset geographical relationship with the production equipment can be divided as edge servers. Exemplarily, the preset geographical relationship can be that the distance between the production equipment and the server is less than a preset distance threshold. In this way, by shortening the geographical distance between the production equipment and the server, the data transmission time can be reduced. In the production process, when the input data required by the prediction model is obtained, it can be immediately sent to the edge server for calculation and prediction, thereby improving the prediction response efficiency and meeting the production timeliness.

[0109] According to some embodiments of the present application, a method for training a production process prediction model and a method for predicting a production process are also provided. Figure 6 As shown in the figure, each production process of a product corresponds to a prediction model. The prediction model is a binary classification model. For the first process of a product, the known process parameter x1 of the historical product in the first production process can be obtained as the first process parameter and input into the corresponding prediction model. Then, the output result f(x1) of the prediction model is obtained, and the cumulative error δ1=y of the output result f(x1) is calculated. obs -f(x1). The output value range includes the first value and the second value. For example, the first value is 0, indicating that the product has fewer defects; the second value is 1, indicating that the product has more defects. obs is the quality parameter of the historical product, and its value range includes the first value and the second value. For example, the first value is 0, which means the product quality is qualified; the second value is 1, which means the product quality is unqualified. It can be seen that if the output result f(x1) of the prediction model is consistent with the product quality parameter y obs The values ​​of are consistent, for example, they are all 1, which means that the prediction model has predicted that the product will have many defects after the current production process, and in fact the final quality of the product is also unqualified. At this time, the prediction accuracy of the prediction model is high and the cumulative error is small, for example, 0. On the contrary, if the output result f(x1) of the prediction model is consistent with the quality parameter y obs The values ​​of are inconsistent, indicating that the prediction accuracy of the prediction model is low, and the corresponding cumulative error is large, for example, 1. It can be seen that the value range of the cumulative error is [0, 1].

[0110] Then, it can be determined which of the cases 1 to 3 is satisfied between the output result f(x1) and the accumulated error δ1. If case 1 is satisfied, i.e., the sum of the output result and the accumulated error f(x1) + δ1 is less than or equal to a preset threshold value, for example, 0.5, it is determined that the first prediction result indicates flowing into the subsequent process, i.e., the second production process. If case 2 is satisfied, i.e., the sum of the output result and the accumulated error f(x1) - δ1 is greater than the preset threshold value, or if case 3 is satisfied, i.e., the sum of the output result and the accumulated error f(x1) + δ1 is greater than the preset threshold value and the sum of the output result and the accumulated error f(x1) - δ1 is less than the preset threshold value, it is determined that the first prediction result indicates not flowing into the subsequent process.

[0111] Subsequently, according to the first prediction result and the true result of whether the historical product flows into the subsequent process, the prediction model can be supervised trained. In this way, by taking a large number of historical products in the first process parameter of the first production process as training data, the training work of the prediction model corresponding to the first production process can be completed.

[0112] For the non-first production process of the product, the known process parameters x n of the historical product in the non-first production process can be obtained as the first process parameters, and the known process parameters x1, x2,..., x n-1 of the previous process can be obtained as the second process parameters. The first process parameters x n and the second process parameters x1, x2,..., x n-1 are input as input data into the prediction model of the non-first production process. Then, the output result f(x1, x2,..., x n of the prediction model is obtained, and the accumulated error δ n of the output result f(x1, x2,..., x n is calculated. Then, it can be determined which of the cases 1 to 3 is satisfied between the output result f(x1, x2,..., x n ) and the accumulated error δ n , and the corresponding first prediction result is determined.

[0113] Subsequently, according to the first prediction result and the true result of whether the historical product flows into the subsequent process, the prediction model can be supervised trained. In this way, by taking a large number of historical products in the first process parameter of the non-first production process and the second process parameter of the previous process as training data, the training work of the prediction model corresponding to the non-first production process can be completed.

[0114] During the application phase of the prediction model, similarly, if the product's current production process is the first process, the process parameters of the first process can be input into the trained prediction model as the third process parameters. Based on the prediction results output by the prediction model, it is determined whether the product that has passed through the first process will flow into a subsequent production process or out of the production line. If the product continues to flow into a second process, the process parameters of the first process can be input into the trained prediction model as the fourth process parameters, and the process parameters of the second process can be input into the trained prediction model as the third process parameters. Based on the prediction results output by the prediction model, it is determined whether the product that has passed through the second process will flow into a subsequent production process or out of the production line. This process continues in this manner until the product has flowed out of the production line or completed the last process.

[0115] It can be seen that the prediction result f(x1,x2,…,x n ), since the process parameters x1, x2, ... x n-1 is known, so f(x1,x2,…,x n )+δ n So x n is a function of the only variable. For the above situation 1, that is, the product can flow into the next process, if the preset threshold is 0.5, then let f(x1,x2,…,x n )+δ n ≤0.5, we can calculate the x that allows the product to flow into the subsequent process n From this we can see that the process parameter x of the current production process n The value range of is not fixed; instead, it is influenced by the process parameters of the previous step, creating a flexible threshold that changes in real time. Compared to a rigid threshold, a flexible threshold can significantly reduce the problem of overkill in industrial manufacturing.

[0116] According to some embodiments of the present application, a third aspect further provides a training device for a production process prediction model. In which, the multiple production processes of a product are sequential, and each production process corresponds to a prediction model, which is used to predict whether the product will flow into the subsequent process after the corresponding production process. Figure 7 As shown, the training device 700 includes:

[0117] An input module 710 is configured to input, for each production process, a first process parameter of the production process and a second process parameter of a previous process before the production process into a prediction model corresponding to the production process;

[0118] A first prediction module 720 is configured to obtain a first prediction result for predicting whether the product will flow into the subsequent process based on an output result of the prediction model;

[0119] a training module 730, configured to perform supervised training on the prediction model corresponding to the production process according to the first prediction result and a true result of whether the product flows into the subsequent process, to obtain a trained prediction model.

[0120] In some embodiments, the first prediction module 720 is specifically configured to:

[0121] obtain a quality parameter for representing a quality of the product and a second prediction result corresponding to the previous process;

[0122] determine the first prediction result according to the output result, the quality parameter and the second prediction result.

[0123] In some embodiments, the first prediction module 720 is specifically configured to:

[0124] determine a cumulative error of the output result according to the output result, the quality parameter and the second prediction result.

[0125] determine the first prediction result according to the output result and the cumulative error.

[0126] In some embodiments, the first prediction module 720 is specifically configured to:

[0127] if a sum of the output result and the cumulative error is less than or equal to a preset threshold, determine that the first prediction result indicates that the product flows into the subsequent process.

[0128] if a difference between the output result and the cumulative error is greater than the preset threshold, determine that the first prediction result indicates that the product does not flow into the subsequent process.

[0129] if the difference between the output result and the cumulative error is less than or equal to the preset threshold, and a sum of the output result and the cumulative error is greater than the preset threshold, determine that the first prediction result indicates that the product does not flow into the subsequent process.

[0130] In some embodiments, the prediction model is a binary classification model; and the first prediction result includes a first result for indicating that the product flows into the subsequent process, and a second result for indicating that the product does not flow into the subsequent process.

[0131] The implementation process of the functions and roles of each module in the above apparatus is specifically described in the implementation process of the corresponding steps in the above method, which will not be repeated here.

[0132] According to some embodiments of the present application, the fourth aspect further provides a prediction device of a production process. Wherein, a plurality of production processes of a product have a time sequence, and each of the production processes corresponds to a prediction model, and the prediction model is used to predict whether the product flows into a subsequent process after the corresponding production process. As shown in Figure 8 The prediction device 800 includes:

[0133] The acquisition module 810 is configured to acquire a third process parameter of a current production process and a fourth process parameter of a previous process before the current production process.

[0134] The second prediction module 820 is configured to input the third process parameter and the fourth process parameter into a trained prediction model corresponding to the current production process to obtain an output prediction result. The prediction result is used to indicate whether the product flows into or out of the subsequent process.

[0135] In some embodiments, the prediction device 800 is applied to an edge server, and the edge server and a production device of the product satisfy a preset geographical relationship.

[0136] The functions and effects of each module in the above device are specifically described in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0137] Based on the method of any of the above embodiments, the embodiments of the present application further provide a computer program product, which includes a computer program. When the computer program is executed by a processor, the method of any of the above embodiments can be implemented.

[0138] Based on the method of any of the above embodiments, the embodiments of the present application further provide an electronic device as shown in Figure 9 As shown in Figure 9 At the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and of course can also include other hardware required by the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs to implement the method of any of the above embodiments.

[0139] The embodiments of the present application further provide a computer storage medium, and the storage medium stores a computer program. When the computer program is executed by a processor, the method of any of the above embodiments can be executed.

[0140] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to the multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0141] In addition, the functional modules in each embodiment of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0142] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0143] The above merely provides an example of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, and thus, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

[0144] The above merely provides an example of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, and thus, once an item is defined in one drawing, it need not be further defined and explained in subsequent drawings.

[0145] It should be noted that the relational terms herein such as first and second and the like are used solely to distinguish one entity or action from another, without necessarily requiring or implying any actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

Claims

1. A training method for a production process prediction model, characterized in that: The multiple production processes of a product are sequential in time, and each production process corresponds to the prediction model, which is used to predict whether the product will flow into a subsequent process after the corresponding production process; the method includes: For each of the production processes, inputting a first process parameter of the production process and a second process parameter of a previous process before the production process into a prediction model corresponding to the production process; Obtaining a first prediction result for predicting whether the product will flow into the subsequent process according to an output result of the prediction model; Performing supervised training on a prediction model corresponding to the production process based on the first prediction result and a real result of whether the product flows into the subsequent process, to obtain a trained prediction model; Wherein, obtaining a first prediction result for predicting whether the product flows into the subsequent process based on the output result of the prediction model includes: Obtaining a quality parameter for characterizing the quality of the product and a second prediction result corresponding to the prior process; determining the first prediction result according to the output result, the quality parameter, and the second prediction result; The determining of the first prediction result according to the output result, the quality parameter, and the second prediction result includes: determining a cumulative error of the output result based on the output result, the quality parameter, and the second prediction result; The first prediction result is determined according to the output result and the accumulated error.

2. The method according to claim 1, characterized in that Determining the first prediction result according to the output result and the accumulated error includes: If the sum of the output result and the accumulated error is less than or equal to a preset threshold, determining that the first prediction result indicates that the output will flow into the subsequent process; If the difference between the output result and the accumulated error is greater than the preset threshold, determining that the first prediction result indicates that the first prediction result does not flow into the subsequent process; If the difference between the output result and the accumulated error is less than or equal to the preset threshold, and the sum of the output result and the accumulated error is greater than the preset threshold, it is determined that the first prediction result indicates that the first prediction result does not flow into the subsequent process.

3. The method according to any one of claims 1-2, characterized in that: The prediction model is a binary classification model; the first prediction result includes a first result for indicating that the product will flow into the subsequent process, and a second result for indicating that the product will not flow into the subsequent process.

4. A method for predicting a production process, characterized in that: The multiple production processes of a product are sequential in time, and each production process corresponds to a prediction model according to any one of claims 1 to 3, wherein the prediction model is used to predict whether the product will flow into a subsequent process after the corresponding production process; the method comprises: Acquiring a third process parameter of a current production process and a fourth process parameter of a previous process before the current production process; The third process parameter and the fourth process parameter are input into a trained prediction model corresponding to the current production process to obtain an output prediction result; the prediction result is used to indicate whether the product flows into or out of the subsequent process.

5. The method according to claim 4, characterized in that The method is applied to an edge server; the edge server and the production equipment of the product meet a preset geographical relationship.

6. A training device for a production process prediction model, characterized in that: The multiple production processes of a product are sequential in time, and each production process corresponds to the prediction model, which is used to predict whether the product will flow into a subsequent process after the corresponding production process; the device includes: an input module, configured to input, for each of the production processes, a first process parameter of the production process and a second process parameter of a previous process before the production process into a prediction model corresponding to the production process; a first prediction module, configured to obtain, based on an output result of the prediction model, a first prediction result for predicting whether the product will flow into the subsequent process; wherein obtaining, based on the output result of the prediction model, the first prediction result for predicting whether the product will flow into the subsequent process comprises: obtaining a quality parameter for characterizing the quality of the product and a second prediction result corresponding to the prior process; determining the first prediction result based on the output result, the quality parameter, and the second prediction result; wherein determining the first prediction result based on the output result, the quality parameter, and the second prediction result comprises: determining a cumulative error of the output result based on the output result, the quality parameter, and the second prediction result; and determining the first prediction result based on the output result and the cumulative error; The training module is used to perform supervised training on the prediction model corresponding to the production process according to the first prediction result and the actual result of whether the product flows into the subsequent process to obtain a trained prediction model.

7. A production process prediction device, characterized in that: The multiple production processes of a product are sequential in time, and each production process corresponds to a prediction model according to any one of claims 1 to 3, and the prediction model is used to predict whether the product will flow into a subsequent process after the corresponding production process; the device includes: an acquisition module, configured to acquire a third process parameter of a current production process and a fourth process parameter of a previous process before the current production process; The second prediction module is used to input the third process parameter and the fourth process parameter into a trained prediction model corresponding to the current production process to obtain an output prediction result; the prediction result is used to indicate whether the product flows into or out of the subsequent process.

8. A computer program product, characterized in that The computer program product comprises a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 5 can be implemented.

9. An electronic device, characterized in that: The electronic device comprises: processor; a memory for storing processor-executable instructions; Wherein, when the processor calls the executable instruction, the operation of the method according to any one of claims 1 to 5 is implemented.

10. A computer-readable storage medium, characterized in that Computer instructions are stored thereon, and when the computer instructions are executed by a processor, the steps of any one of the methods of claims 1-5 are implemented.

Images