Intelligent manufacturing factory data processing method and system and medium
By building a distributed data collaborative processing network in an intelligent manufacturing factory, processing and sharing factory data, and establishing a prediction model for industrial product qualifications, it solves the problem of inadequate prediction of product qualification in the existing technology, and improves the intelligent manufacturing factory's Management efficiency.
Patent Information
- Application Number
- CN202411105364.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-13
- Publication Date
- 2024-12-03
AI Technical Summary
The existing data processing methods of intelligent manufacturing plants cannot accurately predict the qualified status of industrial products manufactured by manufacturing plants in advance, resulting in inefficient intelligent management.
Build a distributed data collaborative processing network formed by a coordination center and multiple manufacturing plant processing centers, share factory data through data sharing protocols, collect and process factory data sets, form qualified and unqualified data sets, and establish qualified industrial products Predict the model and make product status prediction based on real-time data processing capability indicators.
It realizes accurate prediction of the qualified status of industrial products in the manufacturing plant, and improves the intelligent management efficiency of the manufacturing plant.
Smart Images

Figure CN119067372A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent manufacturing, and in particular to an intelligent manufacturing factory data processing method, system and medium. Background Art
[0002] In the traditional industrial product manufacturing process, manufacturing plants often rely more on the technical proficiency and experience of manufacturing workers to timely grasp the relevant status of the industrial product processing process, so as to timely judge whether the working status of the manufacturing equipment is normal according to the status of the manufacturing equipment in the process of manufacturing industrial products, and predict whether the industrial products manufactured by the manufacturing equipment are qualified or unqualified. With the continuous development of industrial processing technology towards intelligence, existing manufacturing plants have also begun to gradually transform into smart factories, so that they can timely grasp various status data in the working status of the equipment, so as to process the status data generated by the manufacturing plant, and provide maintenance prompts when the equipment is working abnormally.
[0003] However, the existing data processing methods for smart manufacturing factories have shortcomings: although they can monitor abnormal equipment status during the product manufacturing process, they cannot make accurate predictions in advance about the qualification of industrial products manufactured by the manufacturing factory. As a result, the intelligent needs of manufacturing factories cannot be met, which is not conducive to manufacturers predicting in advance the qualification status of their own industrial products, and reduces the intelligent management efficiency of manufacturing factories. Summary of the invention
[0004] The first technical problem to be solved by the present invention is to provide a smart manufacturing factory data processing method for the above-mentioned prior art, which can accurately predict the qualified status of industrial products manufactured by the manufacturing factory in advance.
[0005] The second technical problem to be solved by the present invention is to provide an intelligent manufacturing factory data processing system for realizing the above-mentioned intelligent manufacturing factory data processing method in view of the above-mentioned prior art.
[0006] The third technical problem to be solved by the present invention is to provide a readable storage medium for implementing the above-mentioned intelligent manufacturing factory data processing method in response to the above-mentioned prior art.
[0007] The technical solution adopted by the present invention to solve the first technical problem is: a data processing method for an intelligent manufacturing factory, characterized in that it comprises the following steps:
[0008] Step 1: construct a distributed data collaborative processing network formed by a coordination center and multiple manufacturing plant processing centers based on a data sharing agreement; wherein, in the distributed data collaborative processing network, a manufacturing plant processing center corresponds to a manufacturing plant one by one, and each manufacturing work processing center equally shares the factory data generated by each of them according to the data sharing agreement;
[0009] Step 2, each manufacturing plant processing center pre-collects the factory data generated in the process of manufacturing multiple industrial products of the same type in its corresponding manufacturing plant, and forms a factory data set when each manufacturing plant manufactures different products; wherein the factory data set includes a raw material data subset, a manufacturing equipment status data subset and a product status data subset when the industrial product is manufactured by the manufacturing plant, and the manufacturing equipment status data in the manufacturing equipment status data subset and the product status data in the product status data subset correspond one to one in time during the manufacturing process of the industrial product; the number of copies of each data subset in the factory data set is consistent with the number of industrial products of the same type manufactured by the manufacturing plant;
[0010] Step 3: Each manufacturing plant processing center classifies the data in each factory data set of the manufacturing plant to which it belongs, and obtains the qualified factory data set and the unqualified factory data set corresponding to the qualified industrial products and unqualified industrial products manufactured by each manufacturing plant respectively;
[0011] Among them, the qualified factory data set is formed by the product raw material data subset, manufacturing equipment status data subset and product manufacturing stage status data subset corresponding to the obtained industrial product being a qualified product; the unqualified factory data set is formed by the raw material data subset, manufacturing equipment status data subset and product manufacturing stage status data subset corresponding to the obtained industrial product being an unqualified product; the product manufacturing stage status data subset is formed by all the status data corresponding to the product being manufactured to different stages;
[0012] Step 4: Each manufacturing plant processing center obtains an industrial product qualified state prediction model for determining whether the product obtained when the manufacturing plant manufactures the corresponding industrial product is qualified or not based on data processing of the qualified state factory data set and the unqualified state factory data set of the manufacturing plant to which it belongs; wherein the industrial product qualified state prediction model has a one-to-one correspondence with the manufacturing plant and the industrial product;
[0013] Step 5, each manufacturing plant processing center sends the industrial product qualified state prediction model of each manufacturing plant to all manufacturing plant processing centers in the distributed data processing network for encrypted storage processing;
[0014] Step 6, each manufacturing plant processing center collects a set of real-time factory data during the process of manufacturing industrial products of its own manufacturing plant, and processes it to obtain the real-time data processing capability index of each manufacturing plant processing center itself;
[0015] Step 7: The manufacturing plant processing center determines whether it has completed the industrial product status prediction of the industrial products currently manufactured by the manufacturing plant in a timely manner based on its own real-time data processing capability indicators:
[0016] When the industrial product status prediction cannot be completed, the manufacturing plant processing center marks itself as a processing center to be assisted and sends a data scheduling processing request to the coordination center, and then goes to step 8; otherwise, the manufacturing plant processing center uses the corresponding industrial product qualified state prediction model to process the real-time factory data set of the current product to predict the product qualified state status of the current industrial product;
[0017] The data scheduling processing request includes the current real-time data processing capability index of the processing center to be assisted and the real-time factory data set for manufacturing the current product;
[0018] Step 8: The coordination center obtains the real-time data processing capability indicators of all manufacturing plant processing centers in the distributed data processing network, and selects the manufacturing plant processing center that meets the current data scheduling processing request as the collaborative processing center;
[0019] Step 9, the collaborative processing center retrieves the industrial product qualified state prediction model that has been stored in itself and sent in advance by the processing center to be assisted, and uses the industrial product qualified state prediction model to process the data scheduling processing request, and predicts the product qualified state status of the current industrial product manufactured by the manufacturing plant to which the processing center to be assisted belongs;
[0020] Step 10: The collaborative processing center sends the predicted product qualification status to the processing center to be assisted.
[0021] Improved, in the intelligent manufacturing factory data processing method, in step 4, the process of obtaining the industrial product qualified state prediction model includes the following steps a1 to a11:
[0022] Step a1, performing numerical data screening processing on the data in the factory data set to obtain a numerical factory data list that can be numerically represented; wherein the numerical data is data that can be represented by numbers;
[0023] Step a2, filtering the numerical data of the qualified factory data set and the unqualified factory data set according to the obtained numerical factory data list, to obtain a qualified factory numerical data set corresponding to the qualified factory data set and an unqualified factory numerical data set corresponding to the unqualified factory data set;
[0024] Step a3, with the goal of filtering the product manufacturing stage status data, the qualified state factory numerical data set and the unqualified state factory numerical data set are filtered respectively to obtain the product manufacturing stage status numerical data set corresponding to the qualified product state and the product manufacturing stage status numerical data set corresponding to the unqualified product state;
[0025] Step a4, normalizing each numerical data in the numerical data set of the qualified factory and the numerical data set of the unqualified factory, respectively, to obtain the normalized numerical data set of the qualified factory and the normalized numerical data set of the unqualified factory;
[0026] Step a5, calculating the correlation between each data in the obtained qualified state factory normalized numerical data set and the product manufacturing stage state numerical data in the set, and marking all the calculated correlations as the product qualified state data correlation sequence;
[0027] Step a6, calculating the correlation between each data in the normalized numerical data set of the unqualified factory and the numerical data of the product manufacturing stage status in the set, and marking all the calculated correlations as a product unqualified state data correlation sequence;
[0028] Step a7, selecting all the associations in the product qualified state data association sequence and having association values greater than a first preset association threshold, and using all qualified state factory data corresponding to the selected associations as a set of product qualified rate influencing factors affecting the qualified rate of products manufactured by the manufacturing factory;
[0029] Step a8, selecting all the correlations in the product unqualified state data correlation sequence and having correlation values greater than a second preset correlation threshold, and using all the unqualified state factory data corresponding to the selected all correlations as a set of product unqualified rate influencing factors affecting the unqualified rate of products manufactured by the manufacturing factory;
[0030] Step a9, training processing is performed based on the product qualification rate influencing factor set and the product manufacturing stage status data corresponding to the product being in a qualified state, to obtain a product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures qualified products;
[0031] Step a10, training processing is performed based on the product unqualified rate influencing factor set and the product manufacturing stage status data corresponding to the product being in an unqualified state, to obtain a product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures unqualified products;
[0032] Step a11, the obtained product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures qualified products and the product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures qualified products form the industrial product qualified state prediction model together.
[0033] Furthermore, in the intelligent manufacturing factory data processing method, in step 6, the process of obtaining the real-time data processing capability index of the manufacturing factory processing center itself includes the following steps b1 to b5:
[0034] Step b1, the manufacturing plant processing center pre-screens the corresponding factory data set when the manufacturing plant produces qualified products, and obtains a subset of manufacturing equipment status data when the manufacturing plant produces qualified products;
[0035] Step b2, the manufacturing plant processing center uses the state data in the obtained manufacturing equipment state data subset and consistent with the influencing factors in the product qualification rate influencing factor set as the targeted state data representing the data capability of the manufacturing plant processing center itself; wherein the total number boundary of the targeted state data is M, M≥2;
[0036] Step b3: the manufacturing plant processing center processes the same type of targeted state data of the equipment in the screened manufacturing equipment state data subset to obtain the targeted state data offset variance of each targeted state data of the corresponding equipment, and uses the obtained state data offset variance as the targeted state data offset reference value of each targeted state data of the corresponding equipment; wherein the targeted state data offset reference value corresponding to the mth targeted state data is marked as σ m,0 ; 1≤m≤M;
[0037] Step b4: The manufacturing plant processing center obtains all the targeted state data corresponding to the manufacturing plant when producing products, and calculates the real-time offset value of the current targeted state data of the manufacturing plant; wherein the real-time offset value of the current targeted state data of the manufacturing plant is marked as σ m (t); t>0;
[0038] Step b5, calculating the real-time data processing capability index of the manufacturing plant processing center itself according to the obtained target state data offset reference value and the target state data real-time offset value; wherein the real-time data processing capability index of the manufacturing plant processing center itself is marked as δ(t):
[0039]
[0040]
[0041] in, is a sequence of values.
[0042] Further improved, in the present invention, the intelligent manufacturing factory data processing method further includes monitoring the manufacturing status of the manufacturing factory when producing products; wherein, monitoring the manufacturing status of the manufacturing factory when producing products includes the following steps c1 to c6:
[0043] Step c1, based on the factory data set collected by the manufacturing factory processing center when the manufacturing factory produces the product, obtain the product status data value time series sequence corresponding to different moments when the manufacturing factory processes the product;
[0044] Step c2, obtaining product status data value reference baseline values of all data values in the product status data value time series sequence according to the obtained product status data value time series sequence, and generating a product status data reference baseline line representing the product status data based on the product status data value reference baseline value; wherein the product status data reference baseline line is a horizontal straight line with constant data values;
[0045] Step c3, forming a real-time curve of product status data based on a set of real-time factory data collected in real time by the manufacturing factory processing center when the manufacturing factory produces the product;
[0046] Step c4, making a judgment based on the obtained product status data real-time curve and the product status data reference baseline:
[0047] When the deviation value of the real-time curve of the product status data from the reference baseline of the product status data exceeds the preset deviation threshold, it is determined that the product status data has an abnormal trend, and the process goes to step c5; otherwise, the process goes to step c3;
[0048] Step c5, finding the moment when the abnormal trend appears, taking the moment as the abnormal moment, and monitoring the abnormal duration of the abnormal trend from the abnormal moment;
[0049] Step c6: Make a judgment based on the abnormal duration obtained through monitoring:
[0050] When the abnormal duration exceeds the preset duration threshold, it is determined that the manufacturing state of the manufacturing plant when producing the product is abnormal; otherwise, it is determined that the manufacturing state of the manufacturing plant when producing the product is normal, and the process goes to step c3.
[0051] Furthermore, in the intelligent manufacturing factory data processing method, in step 8, the selection process of the collaborative processing center includes the following steps:
[0052] Step d1, obtaining the real-time data processing capability index of each manufacturing plant processing center in the distributed data processing network;
[0053] Step d2, based on all the obtained real-time data processing capability indicators, select the manufacturing plant processing centers corresponding to all the real-time data processing capability indicators whose real-time data processing capability indicators exceed the preset indicator threshold, and use all the selected manufacturing plant processing centers as candidate collaborative processing centers;
[0054] Step d3, calculate the current real-time offset value of the targeted status data and the real-time offset rate of the targeted status data of the manufacturing plant corresponding to each manufacturing plant processing center, and select the manufacturing plant processing center corresponding to the manufacturing plant with the smallest real-time offset rate of the targeted status data as the final selected collaborative processing center.
[0055] Improved, in the intelligent manufacturing factory data processing method, the calculation method of the real-time deviation rate of the targeted state data is as follows:
[0056]
[0057] Among them, ε(t) is the real-time deviation rate of the targeted state data corresponding to a targeted state data in the process of manufacturing products in the manufacturing plant.
[0058] Further improved, in the intelligent manufacturing factory data processing method, in step a9, the process of obtaining the product manufacturing stage state data prediction model corresponding to when the manufacturing factory manufactures qualified products includes the following steps e1 to e5:
[0059] Step e1, constructing an initial BP neural network; wherein the initial BP neural network includes two input layers, multiple intermediate layers and an output layer; one of the two input layers is configured to input a set of factors influencing the product qualification rate, and the other input layer is configured to input product manufacturing stage status data corresponding to the product being in a qualified state; the number of intermediate layers is the same as the number of process steps required for the manufacturing plant to produce qualified products; the output layer is configured to output product manufacturing stage status data corresponding to the manufacturing plant manufacturing qualified products;
[0060] Step e2, dividing all the data in the obtained product qualification rate influencing factor set into a training product qualification rate influencing factor set and a test product qualification rate influencing factor set equally according to the number of data of the same type;
[0061] Step e3, dividing the product manufacturing stage status data corresponding to the product being in a qualified state into a training product manufacturing stage status data subset and a test product manufacturing stage status data subset in equal amounts according to the same type of data;
[0062] Step e4, inputting the obtained training product qualification rate influencing factor set as input data of one input layer, and the obtained training product manufacturing stage status data subset as input data of another input layer into the initial BP neural network, to obtain an initial product manufacturing stage status data prediction model;
[0063] Step e5, using the set of influencing factors of the test product qualification rate and the subset of the test product manufacturing stage status data to optimize and adjust the initial product manufacturing stage status data prediction model, to obtain the optimized product manufacturing stage status data prediction model, and use the optimized product manufacturing stage status data prediction model as the product manufacturing stage status data prediction model corresponding to the manufacturing plant when predicting the qualified products.
[0064] Improved, in the intelligent manufacturing factory data processing method, in step a10, the process of obtaining a product manufacturing stage state data prediction model corresponding to when the manufacturing factory manufactures unqualified products includes the following steps f1 to f5:
[0065] Step f1, constructing an initial BP neural network; wherein the initial BP neural network includes two input layers, multiple intermediate layers and an output layer; one of the two input layers is configured to input a set of factors influencing the product unqualified rate, and the other input layer is configured to input product manufacturing stage status data corresponding to the product being in an unqualified state; the number of intermediate layers is the same as the number of process steps required for the manufacturing plant to produce unqualified products; the output layer is configured to output product manufacturing stage status data corresponding to the manufacturing plant manufacturing unqualified products;
[0066] Step f2, dividing all the data in the obtained product failure rate influencing factor set into a training product failure rate influencing factor set and a test product qualification rate influencing factor set equally according to the number of data of the same type;
[0067] Step f3, dividing the product manufacturing stage status data corresponding to the product being in an unqualified state into a training product manufacturing stage status data subset and a test product manufacturing stage status data subset in equal amounts according to the same type of data;
[0068] Step f4, input the obtained training product non-conformity rate influencing factor set as input data of one input layer, and the obtained training product manufacturing stage status data subset as input data of another input layer, and input them together into the initial BP neural network to obtain the initial product manufacturing stage status data prediction model;
[0069] Step f5, using the set of influencing factors of the test product failure rate and the subset of the test product manufacturing stage status data to optimize and adjust the initial product manufacturing stage status data prediction model, to obtain the optimized product manufacturing stage status data prediction model, and use the optimized product manufacturing stage status data prediction model as the product manufacturing stage status data prediction model corresponding to when predicting the manufacturing plant to manufacture unqualified products.
[0070] The technical solution adopted by the present invention to solve the second technical problem is: an intelligent manufacturing factory data processing system, which implements the intelligent manufacturing factory data processing method, characterized in that the intelligent manufacturing factory data processing system includes:
[0071] A plurality of manufacturing plant processing centers are interconnected and communicated with each other, and the plurality of manufacturing plant processing centers form a distributed data collaborative processing network based on a data sharing protocol; wherein, in the distributed data collaborative processing network, a manufacturing plant processing center corresponds to a manufacturing plant one by one, and each manufacturing work processing center equally shares the factory data generated by each of them according to the data sharing protocol;
[0072] A plurality of data collection devices, each of which corresponds to a processing center of a manufacturing plant; wherein the data collection device is configured to collect in advance a set of factory data when the manufacturing plant produces different products;
[0073] A plurality of processors, each of which corresponds to a processing center of a manufacturing plant one-to-one; wherein the processor obtains an industrial product qualified state prediction model for whether the product obtained when the manufacturing plant manufactures the corresponding industrial product is qualified or not and a real-time data processing capability indicator of the processing center of the manufacturing plant itself based on data processing of a qualified state factory data set and an unqualified state factory data set of the manufacturing plant to which the processor belongs, and determines whether the processor itself has completed the industrial product state prediction of the industrial product currently manufactured by the manufacturing plant in a timely manner;
[0074] A coordination center is located in the distributed data collaborative processing network and is in communication connection with each manufacturing plant processing center. The coordination center obtains the real-time data processing capability indicators of all manufacturing plant processing centers in the distributed data processing network, and selects the manufacturing plant processing center that meets the current data scheduling processing request as the collaborative processing center;
[0075] Among them, each manufacturing plant processing center has a data storage device, and the data storage device is configured to encrypt and store the industrial product qualified state prediction model received by the manufacturing plant processing center to which it belongs.
[0076] The technical solution adopted by the present invention to solve the third technical problem is: a readable storage medium, on which a computer program is stored, characterized in that when the computer program is executed by a processor, any one of the smart manufacturing factory data processing methods is implemented.
[0077] Compared with the prior art, the advantages of the present invention are: the intelligent manufacturing factory data processing method of the invention constructs a distributed data collaborative processing network formed by a coordination center and multiple manufacturing factory processing centers, and each manufacturing factory processing center processes the factory data set collected when the manufacturing factory manufactures different products to obtain a qualified factory data set and an unqualified factory data set, and obtains an industrial product qualified state prediction model based on the data processing of the obtained qualified factory data set and the unqualified factory data set; once any manufacturing factory processing center determines that it cannot complete the industrial product status prediction, it sends a data scheduling processing request to the coordination center, and the coordination center selects the manufacturing factory processing center that meets the current data scheduling processing request as the collaborative processing center based on the real-time data processing capability indicators of each manufacturing factory processing center, so as to complete the product qualified state status prediction for the current product and send it to the processing center to be assisted, thereby realizing the accurate prediction of whether the industrial products manufactured by the manufacturing plant are qualified or not based on the factory data set of the manufacturing plant, and improving the intelligent management efficiency of the manufacturing plant. BRIEF DESCRIPTION OF THE DRAWINGS
[0078] Figure 1 It is a flowchart of a method for processing data in an intelligent manufacturing factory in an embodiment of the present invention;
[0079] Figure 2 Schematic diagram of a smart manufacturing factory data processing system in an embodiment of the present invention. DETAILED DESCRIPTION
[0080] The present invention is further described in detail below with reference to the accompanying drawings.
[0081] This embodiment provides a method for processing data in an intelligent manufacturing factory. Figure 1 As shown, the intelligent manufacturing factory data processing method of this embodiment includes the following steps 1 to 10:
[0082] Step 1: construct a distributed data collaborative processing network formed by a coordination center and multiple manufacturing plant processing centers based on a data sharing agreement; wherein, in the distributed data collaborative processing network, a manufacturing plant processing center corresponds to a manufacturing plant one-to-one, and each manufacturing work processing center equally shares the factory data formed by each of them according to the data sharing agreement; the factory data shared here is a set of factory data corresponding to when the manufacturing plant manufactures different industrial products; of course, the premise that these factory data can be shared is that all manufacturing plant processing centers in the distributed data collaborative processing network are based on the sharing agreement, and the factory data can only be circulated in the distributed data collaborative processing network, and it is not allowed to be released to the outside without authorization;
[0083] Step 2, each manufacturing plant processing center pre-collects the factory data generated in the process of manufacturing multiple industrial products of the same type in its corresponding manufacturing plant, and forms a factory data set when each manufacturing plant manufactures different products; wherein the factory data set includes a raw material data subset, a manufacturing equipment status data subset and a product status data subset when the industrial product is manufactured by the manufacturing plant, and the manufacturing equipment status data in the manufacturing equipment status data subset and the product status data in the product status data subset correspond one to one in time during the manufacturing process of the industrial product; the number of copies of each data subset in the factory data set is consistent with the number of industrial products of the same type manufactured by the manufacturing plant;
[0084] That is to say, when manufacturing an industrial product, each manufacturing plant needs to produce multiple copies of the same industrial product in advance, so that multiple factory data sets for the product can be obtained, which is convenient for subsequent processing of these factory data sets to obtain an industrial product qualification prediction model that predicts whether the industrial product is qualified or not;
[0085] Step 3: Each manufacturing plant processing center classifies the data in each factory data set of the manufacturing plant to which it belongs, and obtains the qualified factory data set and the unqualified factory data set corresponding to the qualified industrial products and unqualified industrial products manufactured by each manufacturing plant respectively;
[0086] Among them, the qualified factory data set is formed by the product raw material data subset, manufacturing equipment status data subset and product manufacturing stage status data subset corresponding to the obtained industrial product being a qualified product; the unqualified factory data set is formed by the raw material data subset, manufacturing equipment status data subset and product manufacturing stage status data subset corresponding to the obtained industrial product being an unqualified product; the product manufacturing stage status data subset is formed by all the status data corresponding to the product being manufactured to different stages;
[0087] Step 4: Each manufacturing plant processing center obtains an industrial product qualified state prediction model for determining whether the product obtained when the manufacturing plant manufactures the corresponding industrial product is qualified or not based on data processing of the qualified state factory data set and the unqualified state factory data set of the manufacturing plant to which it belongs; wherein the industrial product qualified state prediction model has a one-to-one correspondence with the manufacturing plant and the industrial product;
[0088] Step 5, each manufacturing plant processing center sends the industrial product qualified state prediction model of each manufacturing plant to all manufacturing plant processing centers in the distributed data processing network for encrypted storage processing; that is, during the execution of step 5, all industrial product qualified state prediction models formed by each manufacturing plant processing center will be shared to the distributed data processing network, so that each manufacturing plant processing center in the distributed data processing network can call it according to actual needs;
[0089] Step 6, each manufacturing plant processing center collects a set of real-time factory data during the process of manufacturing industrial products of its own manufacturing plant, and processes it to obtain the real-time data processing capability index of each manufacturing plant processing center itself;
[0090] Step 7: The manufacturing plant processing center determines whether it has completed the industrial product status prediction of the industrial products currently manufactured by the manufacturing plant in a timely manner based on its own real-time data processing capability indicators:
[0091] When the industrial product status prediction cannot be completed, the manufacturing plant processing center marks itself as a processing center to be assisted and sends a data scheduling processing request to the coordination center, and then goes to step 8; otherwise, the manufacturing plant processing center uses the corresponding industrial product qualified state prediction model to process the real-time factory data set of the current product to predict the product qualified state status of the current industrial product;
[0092] The data scheduling processing request includes the current real-time data processing capability index of the processing center to be assisted and the real-time factory data set for manufacturing the current product;
[0093] Step 8: The coordination center obtains the real-time data processing capability indicators of all manufacturing plant processing centers in the distributed data processing network, and selects the manufacturing plant processing center that meets the current data scheduling processing request as the collaborative processing center;
[0094] Step 9, the collaborative processing center retrieves the industrial product qualified state prediction model that has been stored in itself and sent in advance by the processing center to be assisted, and uses the industrial product qualified state prediction model to process the data scheduling processing request, and predicts the product qualified state status of the current industrial product manufactured by the manufacturing plant to which the processing center to be assisted belongs;
[0095] Step 10: The collaborative processing center sends the predicted product qualification status to the processing center to be assisted.
[0096] Specifically in this embodiment, in the above step 4, the process of obtaining the industrial product qualified state prediction model includes the following steps a1 to a11:
[0097] Step a1, performing numerical data screening processing on the data in the factory data set to obtain a numerical factory data list that can be numerically represented; wherein the numerical data is data that can be represented by numbers;
[0098] Step a2, filtering the numerical data of the qualified factory data set and the unqualified factory data set according to the obtained numerical factory data list, to obtain the qualified factory numerical data set corresponding to the qualified factory data set and the unqualified factory numerical data set corresponding to the unqualified factory data set;
[0099] Step a3, with the goal of filtering the product manufacturing stage status data, the qualified state factory numerical data set and the unqualified state factory numerical data set are filtered respectively to obtain the product manufacturing stage status numerical data set corresponding to the qualified product state and the product manufacturing stage status numerical data set corresponding to the unqualified product state;
[0100] Since the manufacturing process of each industrial product is usually divided into multiple sub-manufacturing stages, the factory data corresponding to each sub-manufacturing stage is usually different. Of course, the factory data corresponding to each sub-manufacturing stage of qualified industrial products and each sub-manufacturing stage of unqualified industrial products are also usually different;
[0101] Step a4, normalizing each numerical data in the qualified state factory numerical data set and the unqualified state factory numerical data set, respectively, to obtain the normalized numerical data set of the qualified state factory and the normalized numerical data set of the unqualified state factory; wherein:
[0102] For the qualified factory numerical data set, the qualified factory numerical data set is marked as X, X = {X (i)}, 1 ≤ i ≤ I; the corresponding qualified factory normalized numerical data set is marked as x, x = {x (i)}:
[0103]
[0104] The corresponding program code snippet is as follows:
[0105] #include <iostream>
[0106] using namespace std;
[0107] int main() {
[0108] int n;
[0109] cin >> n;
[0110] int sum = 0; / / Define the initial value of the counting variable sum for cumulative summation as 0
[0111] int x = X(1); / / X(1) is the first numerical data value in the qualified state factory numerical data set
[0112] for (int i = 1; i <= n; i++) / / Traverse each item in the qualified state factory numerical data set. i represents the current item number in the qualified state factory numerical data set, and the value of n is the total number of numerical data I in the qualified state factory numerical data set
[0113] sum = sum + i;
[0114] x = x / sum;
[0115] cout << x << endl; / / Output the result
[0116] return 0;
[0117] }
[0118] For the unqualified state factory numerical data set, the qualified state factory numerical data set is marked as Y, Y = {Y(j)}, 1 ≤ j ≤ J; the corresponding normalized unqualified state factory numerical data set is marked as y, y = {y(j)}:
[0119]
[0120] The corresponding program code segment is as follows:
[0121] #include <iostream>
[0122] using namespace std;
[0123] int main() {
[0124] int n;
[0125] cin >> n;
[0126] int sum = 0; / / Define the initial value of the counting variable sum for cumulative summation as 0
[0127] int y = Y(1); / / Y(1) is the first numerical data value in the numerical data set of the unqualified state factory
[0128] for (int i = 1; i <= n; i++) / / Traverse each item in the numerical data set of the unqualified state factory. i represents the current item number in the numerical data set of the unqualified state factory, and the value of n is the total number J of numerical data in the numerical data set of the unqualified state factory
[0129] sum = sum + i;
[0130] y = y / sum;
[0131] cout << y << endl; / / Output the result
[0132] return 0;
[0133] }
[0134] Step a5, calculate the correlation degree between each data in the numerical data set of the qualified state factory after normalization and the numerical data of the product manufacturing stage state in this set, and mark all the calculated correlation degrees as the product qualified state data correlation degree sequence; among them, as needed, when calculating the correlation degree between data here, each data in the numerical data set of the qualified state factory after normalization can be calculated successively with each numerical data in the numerical data of the product manufacturing stage state in this set, so as to obtain the correlation coefficient sequence corresponding to different numerical data in the numerical data of the product manufacturing stage state in this set;
[0135] For a certain data in the numerical data set of the qualified state factory after normalization, all the data in the numerical data set of the qualified state factory after normalization are presented as the original data in tabular form. Among them, the program code segment corresponding to calculating the correlation degree between data is as follows:
[0136] import pandas as pd
[0137] import numpy as np
[0138] import matplotlib.pyplot as plt
[0139] import matplotlib as mpl
[0140] mpl.rcParams['font.sans-serif']=['SimSun']
[0141] mpl.rcParams['axes.unicode_minus']=False
[0142] data=pd.read_excel('.\data.xlsx')
[0143] label_need = data.keys()[1:]
[0144] data1=data[label_need].values
[0145] [m,n] = data1.shape
[0146] data2 = data1.astype('float')
[0147] data3=data2
[0148] for j in range(0,n):
[0149] d_max=max(data2[:,j])
[0150] d_min=min(data2[:,j])
[0151] data3[:,j]=(data2[:,j]-d_min) / (d_max-d_min)
[0152] print(data3)
[0153] for iin range(1,I): / / I is the total number of all numerical data in the normalized numerical data set of the qualified factory
[0154] data3[:,i]=np.abs(data3[:,i]-data3[:,0])
[0155] data4=data3[:,1:I]
[0156] d_max = np.max(data4)
[0157] d_min = np.min(data4)
[0158] c=0.5 / / c is the set resolution coefficient
[0159] data4=(d_min+c*d_max) / (data4+c*d_max)
[0160] xishu=np.mean(data4,axis=0)
[0161] print('The grey correlation between each data in the normalized numerical data set of qualified factories and the numerical data of the product manufacturing stage status in the set is:')
[0162] In this embodiment, by calculating the correlation between each data in the normalized numerical data set of the qualified factory and the numerical data of the product manufacturing stage status in the set, the manufacturing factory data (or factors) that affect whether the manufactured industrial products are qualified or not can be mined, thereby ensuring a more comprehensive and accurate comprehensive analysis, so as to provide comprehensive data support for the subsequent prediction model of the product manufacturing stage status data corresponding to the qualified products manufactured by the manufacturing factory;
[0163] Step a6, calculating the correlation between each data in the normalized numerical data set of the unqualified factory and the numerical data of the product manufacturing stage status in the set, and marking all the calculated correlations as the product unqualified state data correlation sequence; the correlation calculation here can be processed according to the existing mature technology, and will not be repeated here; as for the relevant program code, refer to the description in step a5;
[0164] Step a7, selecting all the associations in the product qualified state data association sequence and having association values greater than a first preset association threshold, and using all qualified state factory data corresponding to the selected associations as a set of product qualified rate influencing factors affecting the qualified rate of products manufactured by the manufacturing factory;
[0165] Step a8, selecting all the correlations in the product unqualified state data correlation sequence and having correlation values greater than a second preset correlation threshold, and using all the unqualified state factory data corresponding to the selected all correlations as a set of product unqualified rate influencing factors affecting the unqualified rate of products manufactured by the manufacturing factory;
[0166] Step a9, training processing is performed based on the product qualification rate influencing factor set and the product manufacturing stage status data corresponding to the product being in a qualified state, to obtain a product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures qualified products;
[0167] Step a10, training processing is performed based on the product unqualified rate influencing factor set and the product manufacturing stage status data corresponding to the product being in an unqualified state, to obtain a product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures unqualified products;
[0168] Step a11, the obtained product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures qualified products and the product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures qualified products form the industrial product qualified state prediction model together.
[0169] In addition, in step 6 of this embodiment, the process of obtaining the real-time data processing capability index of the manufacturing plant processing center itself includes the following steps b1 to b5:
[0170] Step b1, the manufacturing plant processing center pre-screens the corresponding factory data set when the manufacturing plant produces qualified products, and obtains a subset of manufacturing equipment status data when the manufacturing plant produces qualified products;
[0171] Step b2, the manufacturing plant processing center uses the state data that is located in the obtained manufacturing equipment state data subset and is consistent with the influencing factors in the product qualification rate influencing factor set as the targeted state data that characterizes the data capability of the manufacturing plant processing center itself; wherein the total number boundary of the targeted state data is M, M≥2;
[0172] Step b3: the manufacturing plant processing center processes the same type of targeted state data of the equipment in the screened manufacturing equipment state data subset to obtain the targeted state data offset variance of each targeted state data of the corresponding equipment, and uses the obtained state data offset variance as the targeted state data offset reference value of each targeted state data of the corresponding equipment; wherein the targeted state data offset reference value corresponding to the mth targeted state data is marked as σ m,0 ; 1≤m≤M;
[0173] In the process of forming a subset of manufacturing equipment status data, a certain type of status data of manufacturing equipment will continuously generate a large amount of status data along with the manufacturing process of industrial products. These status data usually show certain characteristics or abnormal characteristics. Here, these abnormal characteristics need to be processed;
[0174] Step b4: The manufacturing plant processing center obtains all the targeted state data corresponding to the manufacturing plant when producing products, and calculates the real-time offset value of the current targeted state data of the manufacturing plant; wherein the real-time offset value of the current targeted state data of the manufacturing plant is marked as σ m (t); t>0;
[0175] Step b5, calculating the real-time data processing capability index of the manufacturing plant processing center itself according to the obtained target state data offset reference value and the target state data real-time offset value; wherein the real-time data processing capability index of the manufacturing plant processing center itself is marked as δ(t):
[0176]
[0177]
[0178] in, is a sequence of values.
[0179] In order to monitor the process of manufacturing industrial products in a manufacturing plant in real time, the intelligent manufacturing plant data processing method of this embodiment also includes monitoring the manufacturing status of the manufacturing plant when producing products. Specifically, the monitoring operation of the manufacturing status of the manufacturing plant when producing products includes the following steps c1 to c6:
[0180] Step c1, based on the factory data set collected by the manufacturing factory processing center when the manufacturing factory produces the product, obtain the product status data value time series sequence corresponding to different moments when the manufacturing factory processes the product;
[0181] Step c2, obtaining product status data value reference baseline values of all data values in the product status data value time series sequence according to the obtained product status data value time series sequence, and generating a product status data reference baseline line representing the product status data based on the product status data value reference baseline value; wherein the product status data reference baseline line is a horizontal straight line with constant data values; for example, for a state of a product, the product status data value reference baseline value corresponding to the state is the average value of all data values in the product status data value time series sequence corresponding to the state;
[0182] Step c3, forming a real-time curve of product status data based on a set of real-time factory data collected in real time by the manufacturing factory processing center when the manufacturing factory produces the product;
[0183] Step c4, making a judgment based on the obtained product status data real-time curve and the product status data reference baseline:
[0184] When the deviation value of the real-time curve of the product status data from the reference baseline of the product status data exceeds the preset deviation threshold, it is determined that the product status data has an abnormal trend, and the process goes to step c5; otherwise, the process goes to step c3;
[0185] Step c5, finding the moment when the abnormal trend appears, taking the moment as the abnormal moment, and monitoring the abnormal duration of the abnormal trend from the abnormal moment;
[0186] Step c6: Make a judgment based on the abnormal duration obtained through monitoring:
[0187] When the abnormal duration exceeds the preset duration threshold, it is determined that the manufacturing state of the manufacturing plant when producing the product is abnormal; otherwise, it is determined that the manufacturing state of the manufacturing plant when producing the product is normal, and the process goes to step c3.
[0188] Specifically in this embodiment, in the aforementioned step 8, the selection process of the collaborative processing center includes the following steps:
[0189] Step d1, obtaining the real-time data processing capability index of each manufacturing plant processing center in the distributed data processing network;
[0190] Step d2, based on all the obtained real-time data processing capability indicators, select the manufacturing plant processing centers corresponding to all the real-time data processing capability indicators whose real-time data processing capability indicators exceed the preset indicator threshold, and use all the selected manufacturing plant processing centers as candidate collaborative processing centers;
[0191] Step d3, calculate the current real-time offset value of the targeted state data and the real-time offset rate of the targeted state data of the manufacturing plant corresponding to each manufacturing plant processing center, and select the manufacturing plant processing center corresponding to the manufacturing plant with the smallest real-time offset rate of the targeted state data as the final selected collaborative processing center. The calculation method of the real-time offset rate of the targeted state data is as follows:
[0192]
[0193] Among them, ε(t) is the real-time deviation rate of the targeted state data corresponding to a targeted state data in the process of manufacturing products in the manufacturing plant.
[0194] It should be noted that in the intelligent manufacturing factory data processing method of this embodiment, in step a9, the process of obtaining the product manufacturing stage state data prediction model corresponding to when the manufacturing factory manufactures qualified products includes the following steps e1 to e5:
[0195] Step e1, constructing an initial BP neural network; wherein the initial BP neural network includes two input layers, multiple intermediate layers and an output layer; one of the two input layers is configured to input a set of factors influencing the product qualification rate, and the other input layer is configured to input product manufacturing stage status data corresponding to the product being in a qualified state; the number of intermediate layers is the same as the number of process steps required for the manufacturing plant to produce qualified products; the output layer is configured to output product manufacturing stage status data corresponding to the manufacturing plant manufacturing qualified products;
[0196] Step e2, dividing all the data in the obtained product qualification rate influencing factor set into a training product qualification rate influencing factor set and a test product qualification rate influencing factor set equally according to the number of data of the same type;
[0197] Step e3, dividing the product manufacturing stage status data corresponding to the product being in a qualified state into a training product manufacturing stage status data subset and a test product manufacturing stage status data subset in equal amounts according to the same type of data;
[0198] Step e4, inputting the obtained training product qualification rate influencing factor set as input data of one input layer, and the obtained training product manufacturing stage status data subset as input data of another input layer into the initial BP neural network, to obtain an initial product manufacturing stage status data prediction model;
[0199] Step e5, using the set of influencing factors of the test product qualification rate and the subset of the test product manufacturing stage status data to optimize and adjust the initial product manufacturing stage status data prediction model, to obtain the optimized product manufacturing stage status data prediction model, and use the optimized product manufacturing stage status data prediction model as the product manufacturing stage status data prediction model corresponding to the manufacturing plant when predicting the qualified products.
[0200] Specifically in this embodiment, in the above step a10, the process of obtaining the product manufacturing stage state data prediction model corresponding to the prediction of the manufacturing plant manufacturing unqualified products includes the following steps f1 to f5:
[0201] Step f1, constructing an initial BP neural network; wherein the initial BP neural network includes two input layers, multiple intermediate layers and an output layer; one of the two input layers is configured to input a set of factors affecting the product unqualified rate, and the other input layer is configured to input product manufacturing stage status data corresponding to the product being in an unqualified state; the number of intermediate layers is the same as the number of process steps required when the manufacturing plant produces unqualified products; the output layer is configured to output product manufacturing stage status data corresponding to when the manufacturing plant manufactures unqualified products; wherein, in this embodiment, the constructed initial BP neural network is as follows:
[0202]
[0203] Among them, y i is the output result after convolution calculation, x 1,i is the input product failure rate influencing factor, w 1,i is the weight value on the convolution kernel, b is the bias, x 2,i is the product manufacturing stage status data corresponding to the input product being in an unqualified state, w 2,i is the weight value on the corresponding convolution kernel; f(·) is the activation function after convolution calculation; n is the total number of influencing factors in the input product failure rate influencing factors;
[0204] Step f2, dividing all data in the obtained product failure rate influencing factor set into a training product failure rate influencing factor set and a test product qualification rate influencing factor set equally according to the number of data of the same type;
[0205] Step f3, dividing the product manufacturing stage status data corresponding to the product being in an unqualified state into a training product manufacturing stage status data subset and a test product manufacturing stage status data subset in equal amounts according to the same type of data;
[0206] Step f4, input the obtained training product non-conformity rate influencing factor set as input data of one input layer, and the obtained training product manufacturing stage status data subset as input data of another input layer, and input them together into the initial BP neural network to obtain the initial product manufacturing stage status data prediction model;
[0207] Step f5, using the set of influencing factors of the test product failure rate and the subset of the test product manufacturing stage status data to optimize and adjust the initial product manufacturing stage status data prediction model, to obtain the optimized product manufacturing stage status data prediction model, and use the optimized product manufacturing stage status data prediction model as the product manufacturing stage status data prediction model corresponding to when predicting the manufacturing plant to manufacture unqualified products.
[0208] This embodiment also provides a smart manufacturing factory data processing system to implement the smart manufacturing factory data processing method. Figure 2 As shown, the intelligent manufacturing factory data processing system includes:
[0209] A plurality of manufacturing plant processing centers are interconnected and communicated with each other, and the plurality of manufacturing plant processing centers form a distributed data collaborative processing network based on a data sharing protocol; wherein, in the distributed data collaborative processing network, a manufacturing plant processing center corresponds to a manufacturing plant one by one, and each manufacturing work processing center equally shares the factory data generated by each of them according to the data sharing protocol;
[0210] The multiple manufacturing plant processing centers include manufacturing plant processing center 1, manufacturing plant processing center 2, ..., manufacturing plant processing center Q; correspondingly, the manufacturing plants include manufacturing plant 1, manufacturing plant 2, ..., manufacturing plant Q; Q ≥ 3;
[0211] A plurality of data collection devices, each of which corresponds to a processing center of a manufacturing plant; wherein the data collection device is configured to pre-collect a set of factory data when the manufacturing plant produces different products; the data collection devices are respectively data collection device 1, data collection device 2, ..., data collection device Q;
[0212] A plurality of processors, each of which corresponds to a processing center of a manufacturing plant one-to-one; wherein the processor obtains an industrial product qualified state prediction model for whether the product obtained when the manufacturing plant manufactures the corresponding industrial product is qualified or not and a real-time data processing capability indicator of the processing center of the manufacturing plant itself based on data processing of a qualified state factory data set and an unqualified state factory data set of the manufacturing plant to which the processor belongs, and determines whether the processor itself has completed the industrial product state prediction of the industrial product currently manufactured by the manufacturing plant in a timely manner; wherein the plurality of processors include processor 1, processor 2, ..., processor Q;
[0213] A coordination center is located in the distributed data collaborative processing network and is in communication connection with each manufacturing plant processing center. The coordination center obtains the real-time data processing capability indicators of all manufacturing plant processing centers in the distributed data processing network, and selects the manufacturing plant processing center that meets the current data scheduling processing request as the collaborative processing center;
[0214] Among them, each manufacturing plant processing center has a data storage device, and the data storage device is configured to encrypt and store the industrial product qualified state prediction model received by the manufacturing plant processing center to which it belongs.
[0215] This embodiment also provides a readable storage medium. Specifically, the readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned intelligent manufacturing factory data processing method is implemented.
[0216] Although the preferred embodiments of the present invention are described in detail above, it should be clearly understood that various modifications and variations are possible for those skilled in the art. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.< / iostream> < / iostream>
Claims
1. A method for processing data in an intelligent manufacturing factory, characterized in that: The steps include: Step 1: construct a distributed data collaborative processing network formed by a coordination center and multiple manufacturing plant processing centers based on a data sharing agreement; wherein, in the distributed data collaborative processing network, a manufacturing plant processing center corresponds to a manufacturing plant one by one, and each manufacturing work processing center equally shares the factory data generated by each of them according to the data sharing agreement; Step 2, each manufacturing plant processing center pre-collects the factory data generated in the process of manufacturing multiple industrial products of the same type in its corresponding manufacturing plant, and forms a factory data set when each manufacturing plant manufactures different products; wherein the factory data set includes a raw material data subset, a manufacturing equipment status data subset and a product status data subset when the industrial product is manufactured by the manufacturing plant, and the manufacturing equipment status data in the manufacturing equipment status data subset and the product status data in the product status data subset correspond one to one in time during the manufacturing process of the industrial product; the number of copies of each data subset in the factory data set is consistent with the number of industrial products of the same type manufactured by the manufacturing plant; Step 3: Each manufacturing plant processing center classifies the data in each factory data set of the manufacturing plant to which it belongs, and obtains the qualified factory data set and the unqualified factory data set corresponding to the qualified industrial products and unqualified industrial products manufactured by each manufacturing plant respectively; Among them, the qualified factory data set is formed by the product raw material data subset, manufacturing equipment status data subset and product manufacturing stage status data subset corresponding to the obtained industrial product being a qualified product; the unqualified factory data set is formed by the raw material data subset, manufacturing equipment status data subset and product manufacturing stage status data subset corresponding to the obtained industrial product being an unqualified product; the product manufacturing stage status data subset is formed by all the status data corresponding to the product being manufactured to different stages; Step 4: Each manufacturing plant processing center obtains an industrial product qualified state prediction model for determining whether the product obtained when the manufacturing plant manufactures the corresponding industrial product is qualified or not based on data processing of the qualified state factory data set and the unqualified state factory data set of the manufacturing plant to which it belongs; wherein the industrial product qualified state prediction model has a one-to-one correspondence with the manufacturing plant and the industrial product; Step 5, each manufacturing plant processing center sends the industrial product qualified state prediction model of each manufacturing plant to all manufacturing plant processing centers in the distributed data processing network for encrypted storage processing; Step 6, each manufacturing plant processing center collects a set of real-time factory data during the process of manufacturing industrial products of its own manufacturing plant, and processes it to obtain the real-time data processing capability index of each manufacturing plant processing center itself; Step 7: The manufacturing plant processing center determines whether it has completed the industrial product status prediction of the industrial products currently manufactured by the manufacturing plant in a timely manner based on its own real-time data processing capability indicators: When the industrial product status prediction cannot be completed, the manufacturing plant processing center marks itself as a processing center to be assisted and sends a data scheduling processing request to the coordination center, and then goes to step 8; otherwise, the manufacturing plant processing center uses the corresponding industrial product qualified state prediction model to process the real-time factory data set of the current product to predict the product qualified state status of the current industrial product; The data scheduling processing request includes the current real-time data processing capability index of the processing center to be assisted and the real-time factory data set for manufacturing the current product; Step 8: The coordination center obtains the real-time data processing capability indicators of all manufacturing plant processing centers in the distributed data processing network, and selects the manufacturing plant processing center that meets the current data scheduling processing request as the collaborative processing center; Step 9, the collaborative processing center retrieves the industrial product qualified state prediction model that has been stored in itself and sent in advance by the processing center to be assisted, and uses the industrial product qualified state prediction model to process the data scheduling processing request, and predicts the product qualified state status of the current industrial product manufactured by the manufacturing plant to which the processing center to be assisted belongs; Step 10: The collaborative processing center sends the predicted product qualification status to the processing center to be assisted.
2. The intelligent manufacturing factory data processing method according to claim 1, characterized in that: In step 4, the process of obtaining the industrial product qualified state prediction model includes the following steps a1 to a11: Step a1, performing numerical data screening processing on the data in the factory data set to obtain a numerical factory data list that can be numerically represented; wherein the numerical data is data that can be represented by numbers; Step a2, filtering the numerical data of the qualified factory data set and the unqualified factory data set according to the obtained numerical factory data list, to obtain a qualified factory numerical data set corresponding to the qualified factory data set and an unqualified factory numerical data set corresponding to the unqualified factory data set; Step a3, with the goal of filtering the product manufacturing stage status data, the qualified state factory numerical data set and the unqualified state factory numerical data set are filtered respectively to obtain the product manufacturing stage status numerical data set corresponding to the qualified product state and the product manufacturing stage status numerical data set corresponding to the unqualified product state; Step a4, normalizing each numerical data in the numerical data set of the qualified factory and the numerical data set of the unqualified factory, respectively, to obtain the normalized numerical data set of the qualified factory and the normalized numerical data set of the unqualified factory; Step a5, calculating the correlation between each data in the obtained qualified state factory normalized numerical data set and the product manufacturing stage state numerical data in the set, and marking all the calculated correlations as the product qualified state data correlation sequence; Step a6, calculating the correlation between each data in the normalized numerical data set of the unqualified factory and the numerical data of the product manufacturing stage status in the set, and marking all the calculated correlations as a product unqualified state data correlation sequence; Step a7, selecting all the associations in the product qualified state data association sequence and having association values greater than a first preset association threshold, and using all qualified state factory data corresponding to the selected associations as a set of product qualified rate influencing factors affecting the qualified rate of products manufactured by the manufacturing factory; Step a8, selecting all the correlations in the product unqualified state data correlation sequence and having correlation values greater than a second preset correlation threshold, and using all the unqualified state factory data corresponding to the selected all correlations as a set of product unqualified rate influencing factors affecting the unqualified rate of products manufactured by the manufacturing factory; Step a9, training processing is performed based on the product qualification rate influencing factor set and the product manufacturing stage status data corresponding to the product being in a qualified state, to obtain a product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures qualified products; Step a10, training processing is performed based on the product unqualified rate influencing factor set and the product manufacturing stage status data corresponding to the product being in an unqualified state, to obtain a product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures unqualified products; Step a11, the obtained product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures qualified products and the product manufacturing stage status data prediction model corresponding to when the manufacturing plant manufactures qualified products form the industrial product qualified state prediction model together.
3. The intelligent manufacturing factory data processing method according to claim 2, characterized in that: The method further includes monitoring the manufacturing status of the product produced by the manufacturing plant; wherein the monitoring the manufacturing status of the product produced by the manufacturing plant includes the following steps c1 to c6: Step c1, based on the factory data set collected by the manufacturing factory processing center when the manufacturing factory produces the product, obtain the product status data value time series sequence corresponding to different moments when the manufacturing factory processes the product; Step c2, obtaining product status data value reference baseline values of all data values in the product status data value time series sequence according to the obtained product status data value time series sequence, and generating a product status data reference baseline line representing the product status data based on the product status data value reference baseline value; wherein the product status data reference baseline line is a horizontal straight line with constant data values; Step c3, forming a real-time curve of product status data based on a set of real-time factory data collected in real time by the manufacturing factory processing center when the manufacturing factory produces the product; Step c4, making a judgment based on the obtained product status data real-time curve and the product status data reference baseline: When the deviation value of the real-time curve of the product status data from the reference baseline of the product status data exceeds the preset deviation threshold, it is determined that the product status data has an abnormal trend, and the process goes to step c5; otherwise, the process goes to step c3; Step c5, finding the moment when the abnormal trend appears, taking the moment as the abnormal moment, and monitoring the abnormal duration of the abnormal trend from the abnormal moment; Step c6: Make a judgment based on the abnormal duration obtained through monitoring: When the abnormal duration exceeds the preset duration threshold, it is determined that the manufacturing state of the manufacturing plant when producing the product is abnormal; otherwise, it is determined that the manufacturing state of the manufacturing plant when producing the product is normal, and the process goes to step c3.
4. The intelligent manufacturing factory data processing method according to claim 2, characterized in that: In step a9, the process of obtaining the product manufacturing stage state data prediction model corresponding to the prediction of the manufacturing plant manufacturing qualified products includes the following steps: Step e1, constructing an initial BP neural network; wherein the initial BP neural network includes two input layers, multiple intermediate layers and an output layer; one of the two input layers is configured to input a set of factors influencing the product qualification rate, and the other input layer is configured to input product manufacturing stage status data corresponding to the product being in a qualified state; the number of intermediate layers is the same as the number of process steps required for the manufacturing plant to produce qualified products; the output layer is configured to output product manufacturing stage status data corresponding to the manufacturing plant manufacturing qualified products; Step e2, dividing all the data in the obtained product qualification rate influencing factor set into a training product qualification rate influencing factor set and a test product qualification rate influencing factor set equally according to the number of data of the same type; Step e3, dividing the product manufacturing stage status data corresponding to the product being in a qualified state into a training product manufacturing stage status data subset and a test product manufacturing stage status data subset in equal amounts according to the same type of data; Step e4, input the obtained training product qualification rate influencing factor set as input data of one input layer, and the obtained training product manufacturing stage status data subset as input data of another input layer, and input them together into the initial BP neural network to obtain the initial product manufacturing stage status data prediction model; Step e5, using the set of influencing factors of the test product qualification rate and the subset of the test product manufacturing stage status data to optimize and adjust the initial product manufacturing stage status data prediction model, to obtain the optimized product manufacturing stage status data prediction model, and use the optimized product manufacturing stage status data prediction model as the product manufacturing stage status data prediction model corresponding to the manufacturing plant when predicting the qualified products.
5. The intelligent manufacturing factory data processing method according to claim 2, characterized in that: In step a10, the process of obtaining the product manufacturing stage state data prediction model corresponding to the prediction of the manufacturing plant manufacturing unqualified products includes the following steps: Step f1, constructing an initial BP neural network; wherein the initial BP neural network includes two input layers, multiple intermediate layers and an output layer; one of the two input layers is configured to input a set of factors influencing the product unqualified rate, and the other input layer is configured to input product manufacturing stage status data corresponding to the product being in an unqualified state; the number of intermediate layers is the same as the number of process steps required for the manufacturing plant to produce unqualified products; the output layer is configured to output product manufacturing stage status data corresponding to the manufacturing plant manufacturing unqualified products; Step f2, dividing all the data in the obtained product failure rate influencing factor set into a training product failure rate influencing factor set and a test product qualification rate influencing factor set equally according to the number of data of the same type; Step f3, dividing the product manufacturing stage status data corresponding to the product being in an unqualified state into a training product manufacturing stage status data subset and a test product manufacturing stage status data subset in equal amounts according to the same type of data; Step f4, input the obtained training product non-conformity rate influencing factor set as input data of one input layer, and the obtained training product manufacturing stage status data subset as input data of another input layer, and input them together into the initial BP neural network to obtain the initial product manufacturing stage status data prediction model; Step f5, using the set of influencing factors of the test product failure rate and the subset of the test product manufacturing stage status data to optimize and adjust the initial product manufacturing stage status data prediction model, to obtain the optimized product manufacturing stage status data prediction model, and use the optimized product manufacturing stage status data prediction model as the product manufacturing stage status data prediction model corresponding to when predicting the manufacturing plant to manufacture unqualified products.
6. An intelligent manufacturing factory data processing system, which implements the intelligent manufacturing factory data processing method according to claim 1, characterized in that: The intelligent manufacturing factory data processing system includes: A plurality of manufacturing plant processing centers are interconnected and communicated with each other, and the plurality of manufacturing plant processing centers form a distributed data collaborative processing network based on a data sharing protocol; wherein, in the distributed data collaborative processing network, a manufacturing plant processing center corresponds to a manufacturing plant one by one, and each manufacturing work processing center equally shares the factory data generated by each of them according to the data sharing protocol; A plurality of data collection devices, each of which corresponds to a processing center of a manufacturing plant; wherein the data collection device is configured to collect in advance a set of factory data when the manufacturing plant produces different products; A plurality of processors, each of which corresponds to a processing center of a manufacturing plant one-to-one; wherein the processor obtains an industrial product qualified state prediction model for whether the product obtained when the manufacturing plant manufactures the corresponding industrial product is qualified or not and a real-time data processing capability indicator of the processing center of the manufacturing plant itself based on data processing of a qualified state factory data set and an unqualified state factory data set of the manufacturing plant to which the processor belongs, and determines whether the processor itself has completed the industrial product state prediction of the industrial product currently manufactured by the manufacturing plant in a timely manner; A coordination center is located in the distributed data collaborative processing network and is in communication connection with each manufacturing plant processing center. The coordination center obtains the real-time data processing capability indicators of all manufacturing plant processing centers in the distributed data processing network, and selects the manufacturing plant processing center that meets the current data scheduling processing request as the collaborative processing center; Among them, each manufacturing plant processing center has a data storage device, and the data storage device is configured to encrypt and store the industrial product qualified state prediction model received by the manufacturing plant processing center to which it belongs.
7. A readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the intelligent manufacturing factory data processing method according to any one of claims 1 to 5 is implemented.
