Abnormal monitoring device, method and readable storage medium for machine process parameters

Abstract

A device, method and computer-readable storage medium for monitoring abnormalities of machine process parameters, the method comprising: obtaining target machine process parameters that affect the measured values ​​of preset products at each measurement point based on multiple product data sets extracted in advance; constructing a measurement value prediction model corresponding to each measurement point based on the target machine process parameters and the measured values ​​of multiple preset products at each measurement point, and calculating the fit of each model; integrating the fit, predicted value, target machine process parameter and parameter coefficient of the measurement value prediction model corresponding to each measurement point into a problem indicator set; calculating the influence degree index value of each machine process parameter based on the elements in the problem indicator set; and outputting warning information for machine process parameters that exceed the first preset influence degree index value. The present invention uses multi-point measurement data and process data to construct a problem indicator set and analyze it to detect abnormal machine process parameters as early as possible and reduce machine downtime.

Description

Technical Field

[0001] The present invention relates to the field of machine monitoring technology, and in particular to an abnormality monitoring device and method for machine process parameters, and a computer-readable storage medium. Background Art

[0002] In the process of product processing, in order to improve the yield and efficiency of the product process, it is of great significance to reduce the failure rate of product process equipment. In the existing product production process, although some machines have automatic control compensation mechanism functions, which can automatically correct some process parameters physically, it is often difficult to find the real abnormal process parameters, which affects the process quality and even requires shutdown for maintenance. Summary of the invention

[0003] In view of this, it is necessary to provide an abnormal monitoring device, method and computer-readable storage medium for machine process parameters, which can realize early detection of abnormal machine process parameters, reduce machine downtime and the time for finding abnormal process parameters, and improve process quality.

[0004] An embodiment of the present invention provides a method for monitoring anomalies of a machine process parameter, wherein a plurality of the machine process parameters are used to process a plurality of preset products, wherein the preset products are provided with a first measurement point to an Nth measurement point, and the N measurement points are used to measure the same product parameter of the preset products, where N is a positive integer greater than 1, and the method for monitoring anomalies comprises:

[0005] Based on the pre-extracted multiple product data sets, multiple target machine process parameters that affect the measurement value of the preset product at the first measurement point are screened, wherein the multiple target machine process parameters are part of or all of the multiple machine process parameters, the multiple product data sets correspond to the multiple preset products, and each product data set includes the multiple machine process parameters and the measurement value of the corresponding preset product at the first measurement point;

[0006] A measurement value prediction model corresponding to the first measurement point is constructed based on a plurality of the target machine process parameters and the measurement values ​​of a plurality of the preset products at the first measurement point, so as to calculate the fitting degree of the measurement value prediction model according to the predicted value of the first measurement point predicted by the measurement value prediction model, wherein the measurement value prediction model calculates the predicted value of the first measurement point according to each of the target machine process parameters and the parameter coefficient corresponding to each of the target machine process parameters;

[0007] Aggregating the fitting degree of the measurement value prediction model corresponding to the first measurement point, the prediction value of the first measurement point, the target tool process parameter corresponding to the first measurement point, and the parameter coefficient of the target tool process parameter corresponding to the first measurement point into a problem indicator set;

[0008] Repeat the above steps until the fitting degree of the measurement value prediction model corresponding to the Nth measurement point, the prediction value of the Nth measurement point, the target tool process parameter corresponding to the Nth measurement point, and the parameter coefficient of the target tool process parameter corresponding to the Nth measurement point are integrated into the problem indicator set;

[0009] Calculate the influence index value of each of the machine process parameters based on the elements in the problem index set; and

[0010] Outputting warning information of the machine process parameter exceeding the first preset impact level indicator value.

[0011] Optionally, the method further comprises:

[0012] Collecting the processing parameters of the machine for processing the preset product and the measurement parameters of the preset product;

[0013] Extracting designated data from the collected machine processing parameters and the measurement parameters and storing them in an analysis database, wherein the analysis database at least includes a first data table, a second data table, and a third data table, wherein the first data table is used to store a plurality of the machine process parameters in the designated data, the second data table is used to store the measurement values ​​of N measurement points in the designated data, and the third data table is used to store the mapping relationship between the plurality of the machine process parameters and the measurement value of each measurement point;

[0014] A plurality of product data sets are extracted from the analysis database.

[0015] Optionally, the step of constructing an analysis database based on the collected machine processing parameters and the measurement parameters includes:

[0016] A preset ETL tool is used to extract a plurality of the machine process parameters and the measurement values ​​of the N measurement points from the collected machine processing parameters and the measurement parameters, and load the values ​​into the analysis database.

[0017] Optionally, before the step of constructing a measurement value prediction model corresponding to the first measurement point based on the plurality of target machine process parameters and the measurement values ​​of the plurality of preset products at the first measurement point, the step includes:

[0018] If the measurement value of the first measurement point includes multiple dimensional values, the measurement value of the first measurement point is mapped into a one-dimensional value by using a preset dimensionality reduction function.

[0019] Optionally, the measurement value prediction model is a linear model.

[0020] Optionally, the linear model includes a plurality of linear coefficients, and the parameter coefficients of a plurality of the target machine process parameters correspond one-to-one to the plurality of linear coefficients.

[0021] Optionally, the step of calculating the influence index value of each of the machine process parameters based on the elements in the problem index set includes:

[0022] The influence index value of each of the tool process parameters is obtained by conversion based on the number of occurrences of each of the tool process parameters in the problem index set and the parameter coefficient of each of the tool process parameters.

[0023] Optionally, the step of converting the influence index value of each of the machine process parameters based on the number of occurrences of each of the machine process parameters in the problem index set and the parameter coefficient of each of the machine process parameters comprises:

[0024] Using a plurality of preset conversion methods to convert the number of occurrences of each of the machine process parameters in the problem indicator set and the parameter coefficient of each of the machine process parameters, to obtain a plurality of influence degree indicator values ​​of each of the machine process parameters;

[0025] When each of the machine process parameters has multiple influence level indicator values ​​exceeding the first preset influence level indicator value, an aggregate machine abnormality indicator is obtained based on the multiple influence level indicator values ​​exceeding the first preset influence level indicator value to describe the abnormality level of each of the machine process parameters.

[0026] An embodiment of the present invention provides an abnormality monitoring device for machine process parameters, the device comprising a processor and a memory, the memory storing a plurality of computer programs, the processor being used to implement the steps of the abnormality monitoring method for machine process parameters when executing the computer programs stored in the memory.

[0027] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a plurality of instructions, and the plurality of instructions can be executed by one or more processors to implement the steps of the above-mentioned method for abnormal monitoring of machine process parameters.

[0028] Compared with the prior art, the above-mentioned abnormal monitoring device, method and computer-readable storage medium for machine process parameters use multi-point measurement data and machine process data to construct a problem indicator set and analyze it, so as to discover abnormal machine process parameters as early as possible, reduce machine downtime, and improve process quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a functional module diagram of a device for monitoring abnormality of a machine process parameter according to an embodiment of the present invention.

[0030] Figure 2 It is a functional module diagram of an abnormality monitoring program for a machine process parameter according to an embodiment of the present invention.

[0031] Figure 3 It is a flow chart of a method for monitoring abnormality of a process parameter of a machine platform according to an embodiment of the present invention.

[0032] Main component symbols

[0033] Memory 10 processor 20 Abnormal monitoring program 30 Filter Module 101 Building Blocks 102 Archive Module 103 Computing Module 104 Output Module 105 Abnormal monitoring device 100 Machine 200

[0034] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0036] It should be further noted that, in this article, the terms "comprises", "includes" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0037] See also Figure 1 , which is a schematic diagram of a preferred embodiment of the abnormality monitoring device of the present invention.

[0038] The abnormality monitoring device 100 can monitor the machine process parameters of one or more machines 200 to detect abnormal machine process parameters in advance. The machine 200 may refer to a machine for processing parts and products, including but not limited to machine tools, CNC equipment, industrial robots, etc. The machine process parameters may be parameters related to the processing parts of the machine 200, such as the length, rotation speed, cutting parameters, etc. of the tool.

[0039] The abnormality monitoring device 100 may include a memory 10, a processor 20, and an abnormality monitoring program 30 stored in the memory 10 and executable on the processor 20. When the processor 20 executes the abnormality monitoring program 30, the steps in the abnormality monitoring method embodiment are implemented, for example Figure 3 Alternatively, the processor 20 executes the abnormality monitoring program 30 to implement Figure 2 The functions of each module in the module, such as modules 101 to 105.

[0040] The abnormality monitoring program 30 may be divided into one or more modules, which are stored in the memory 10 and executed by the processor 20 to implement the present invention. The one or more modules may be a series of computer program instruction segments that can implement specific functions, and the instruction segments are used to describe the execution process of the abnormality monitoring program 30 in the abnormality monitoring device 100. For example, the abnormality monitoring program 30 may be divided into Figure 2 The screening module 101, the construction module 102, the compilation module 103, the calculation module 104 and the output module 105 in the embodiment of the present invention are shown in the following table. Figure 2 The functions of each module in.

[0041] Those skilled in the art will understand that the schematic diagram is merely an example of the abnormality monitoring device 100 and does not constitute a limitation of the abnormality monitoring device 100. The abnormality monitoring device 100 may include more or fewer components than shown in the diagram, or a combination of certain components, or different components. For example, the abnormality monitoring device 100 may also include an input display device, a bus, etc.

[0042] The processor 20 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor 20 may also be any conventional processor, etc. The processor 20 may connect various parts of the abnormality monitoring device 100 using various interfaces and buses.

[0043] The memory 10 can be used to store the abnormality monitoring program 30 and / or modules. The processor 20 implements various functions of the abnormality monitoring device 100 by running or executing the computer program and / or modules stored in the memory 10, and calling the data stored in the memory 10. The memory 10 may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0044] Figure 2 It is a functional module diagram of a preferred embodiment of the abnormality monitoring program of the present invention.

[0045] See also Figure 2 As shown, the abnormal monitoring program 30 may include a screening module 101, a construction module 102, a compilation module 103, a calculation module 104, and an output module 105. In one embodiment, the above modules may be programmable software instructions stored in the memory 10 and can be called and executed by the processor 20. It is understood that in other embodiments, the above modules may also be program instructions or firmware fixed in the processor 20.

[0046] The screening module 101 is used to screen, based on a plurality of pre-extracted product data sets, a plurality of target machine process parameters that affect a measurement value of a preset product at a first measurement point.

[0047] In one embodiment, the machine 200 is used to process the predetermined product. Assume that the machine 200 includes p machine process parameters X 1 ~X p , p is a positive integer greater than 1. The default product is set with N measurement points (defined as the first measurement point Y 1 To the Nth measuring point Y N ), N measurement points can be selected to measure the same product parameter of the preset product, and N is a positive integer greater than 1. For example, the preset product is a liquid crystal panel, and the preset product is set with N measurement points to measure the panel thickness. The multiple target machine process parameters are part or all of the multiple machine process parameters, for example, the multiple machine process parameters are Ω={X 1 ,X 2 ,X 3 ,…,X p}, multiple target machine process parameters are {X' 1 ,X' 2 ,…,X' k}, X' k ∈Ω and k≤p.

[0048] Multiple product data sets correspond to multiple preset products, and each product data set includes p machine process parameters {X 1 ,X 2 ,X 3 ,…,X p} and the corresponding preset product at the first measuring point Y 1 It can be understood that for a plurality of preset products, at the first measurement point Y 1 The measured values ​​may or may not be the same.

[0049] In one embodiment, the data (such as XML format files) thrown out by the machine 200 for processing the preset product and the measurement data of the preset product can be collected regularly or irregularly, and manually or automatically temporarily stored in a designated storage area. Then, the ETL tool is used to extract the specified data from the designated storage area and transfer it to the analysis database, and then the abnormal machine process parameters can be found out based on the analysis database. For example, the specified data may at least include the machine process parameters and the measurement values ​​of the measurement points. The analysis database may include at least a first data table, a second data table and a third data table, the first data table is used to store a plurality of the machine process parameters, the second data table is used to store the measurement values ​​of N measurement points, and the third data table is used to store the mapping relationship between the plurality of the machine process parameters and the measurement values ​​of each of the measurement points.

[0050] In one embodiment, the ETL tool can also extract the specified data from the specified storage area regularly or irregularly and transfer it to the analysis database. Multiple product data sets can be extracted from the analysis database.

[0051] In one embodiment, the screening module 101 can analyze the pre-extracted multiple product data sets based on a preset screening algorithm to screen multiple target machine process parameters that affect the measurement value of the preset product at the first measurement point. The preset screening algorithm screens the target machine process parameters by determining the difference degree index of the combination of multiple machine process parameters. The difference degree index may include mean-square error (MSE), and MSE may reflect the difference degree between the estimated value and the estimated value. For the screening of the target machine process parameters, the smaller the difference degree index, the better, and the fewer the number of machine process parameters, the better.

[0052] For example, there are 10 machine process parameters, namely, machine process parameters {X 1 ,X 2 ,X 3 ,…,X 10}, multiple machine process parameter sets can be obtained by arranging and combining 10 machine process parameters, each of which can include at least one machine process parameter, and the difference index value corresponding to each machine process parameter set is determined according to the difference between the value of the first measurement point predicted by each machine process parameter set and the actual measurement value of the first measurement point. For multiple preset products, if there are one or more differences in the measurement values ​​at the first measurement point, the average value of the measurement value of the first measurement point can be calculated first, and then the difference between the average value and the value of the first measurement point predicted by each machine process parameter set can be calculated. Assume that the first machine process parameter set {X 1 ,X 2} corresponds to a difference index value of 0.3, and the second machine process parameter set {X 1 ,X 2 ,X 3} corresponds to a difference index value of 0.7, and the third machine process parameter set {X 2 ,X 4 ,X 7} corresponds to a difference index value of 0.1, and the fourth machine process parameter set {X 2 ,X 4 ,X 7 ,X 9} corresponds to a difference index value of 0.1. Assume that after comparative analysis, the minimum difference index value of multiple machine process parameter sets is 0.1, and the machine process parameter sets with a difference index value of 0.1 are the third machine process parameter set and the fourth machine process parameter set. Since the number of elements in the third machine process parameter set is less than the number of elements in the fourth machine process parameter set, the three machine process parameter sets {X 2 ,X 4 ,X 7} is defined as the target tool process parameter.

[0053] The construction module 102 is used to construct a measurement value prediction model M1 corresponding to the first measurement point based on multiple target machine process parameters and multiple measurement values ​​of the preset products at the first measurement point, so as to calculate the fitting degree of the measurement value prediction model M1 according to the predicted value of the first measurement point predicted by the measurement value prediction model M1.

[0054] In one embodiment, the plurality of target machine process parameters are machine process parameters that can more accurately predict the measurement value of the first measurement point. The measurement value prediction model M1 is preferably a linear model, and the measurement value prediction model M1 can calculate the predicted value of the first measurement point according to each target machine process parameter and the parameter coefficient corresponding to each target machine process parameter.

[0055] In one embodiment, the construction module 102 can construct a measurement value prediction model M1 corresponding to the first measurement point based on multiple target machine process parameters and multiple measurement values ​​of the preset products at the first measurement point. For example, the multiple target machine process parameters are machine process parameters {X 2 ,X 4 ,X 7}, the measurement value prediction model M1 of the first measurement point is Y 1 =a*X 2 +b*X 4 +c*X 7 The construction module 102 can be based on the target machine process parameters {X 2 ,X 4 ,X 7} and the measured values ​​of multiple preset products at the first measurement point for regression analysis to determine the values ​​of parameter coefficients a, b, and c. After determining the values ​​of parameter coefficients a, b, and c, the machine process parameter {X 2 ,X 4 ,X 7 A set of parameter values ​​of} is input into the measurement value prediction model M1, and the predicted value of the first measurement point corresponding to the set of parameter values ​​is calculated.

[0056] Furthermore, when the value of the first measurement point can be predicted by the measurement value prediction model M1, the degree of fit (the degree of agreement between the prediction result of the model and the actual result) of the measurement value prediction model M1 can be calculated according to the prediction value of the first measurement point predicted by the measurement value prediction model M1. Specifically, the prediction value of the first measurement point predicted by the measurement value prediction model M1 and the actual measurement value of the first measurement point can be analyzed to obtain the degree of fit Rs1 of the measurement value prediction model M1.

[0057] In one embodiment, multiple linear models may be pre-constructed, and the construction module 102 may select a suitable linear model according to the model index of each linear model to construct the measurement value prediction model M1 corresponding to the first measurement point. The model index may include mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE).

[0058] In one embodiment, since the output of the linear model is generally a one-dimensional value, when each of the N measurement points contains measurement values ​​of multiple dimensions, a preset dimensionality reduction function can be used to map the measurement value of each measurement point to a one-dimensional value. For example, the measurement value of the first measurement point is the deviation value Z in the three-axis direction. 1 , Z 2 , Z3 , that is, it includes three dimensional values. The preset dimensionality reduction function can be set according to actual needs. For example, the preset dimensionality reduction function is g(x), After reducing the dimension of the measurement value of the first measurement point to one dimension, the construction module 102 constructs a measurement value prediction model M1 based on a plurality of target machine process parameters and a plurality of measurement values ​​of a plurality of preset products at the first measurement point.

[0059] The aggregation module 103 is used to aggregate the fitting degree of the measurement value prediction model M1 corresponding to the first measurement point, the prediction value of the first measurement point, the target tool process parameter corresponding to the first measurement point and the parameter coefficient of the target tool process parameter corresponding to the first measurement point into a problem indicator set.

[0060] In one embodiment, for example, the measurement value prediction model M1 corresponding to the first measurement point is Y 1 =0.2*X 2 +0.3*X 4 +0.1*X 7 , then the fitting degree of the measurement value prediction model M1 corresponding to the first measurement point is the fitting degree Rs1, the predicted value of the first measurement point is the value predicted by the measurement value prediction model M1, and the target machine process parameter corresponding to the first measurement point is {X 2 ,X 4 ,X 7}, the parameter coefficient of the target tool process parameter corresponding to the first measurement point is {0.2, 0.3, 0.1}.

[0061] It can be understood that for the second measurement point, the same processing method as the first measurement point is adopted to obtain the measurement value prediction model M2 corresponding to the second measurement point. The aggregation module 103 can aggregate the degree of fit of the measurement value prediction model M2 corresponding to the second measurement point, the predicted value of the second measurement point, the target machine process parameter corresponding to the second measurement point, and the parameter coefficient of the target machine process parameter corresponding to the second measurement point into the problem indicator set. For the Nth measurement point, the same processing method as the first measurement point can be adopted to obtain the measurement value prediction model MN corresponding to the Nth measurement point. The aggregation module 103 can aggregate the degree of fit of the measurement value prediction model MN corresponding to the Nth measurement point, the predicted value of the Nth measurement point, the target machine process parameter corresponding to the Nth measurement point, and the parameter coefficient of the target machine process parameter corresponding to the Nth measurement point into the problem indicator set.

[0062] The calculation module 104 is used to calculate the influence index value of each machine process parameter based on the elements in the problem index set.

[0063] In one embodiment, the influence index value can reflect the importance of the machine process parameter that affects the processing of the preset product by the machine 200. The larger the calculated influence index value, the more important the machine process parameter is, and the machine process parameter may be abnormal. The calculation module 104 can select a preset conversion function and the elements in the problem index set to calculate the influence index value of each machine process parameter.

[0064] For example, with 10 machine process parameters {X 1 ,X 2 ,X 3 ,…,X 10}, 5 measuring points {Y 1 ,Y 2 ,Y 3 ,…,Y 5} is used as an example to illustrate that the screening module 101 screens and obtains the target machine process parameters that affect the measurement values ​​of the preset product at the five measurement points as shown in Table 1 below.

[0065] Table 1

[0066] <![CDATA[Y 1 ]]> <![CDATA[Y 2 ]]> <![CDATA[Y 3 ]]> <![CDATA[Y 4 ]]> <![CDATA[Y 5 ]]> <![CDATA[X 1 ]]> V <![CDATA[X 2 ]]> V V V <![CDATA[X 3 ]]> V <![CDATA[X 4 ]]> V <![CDATA[X 5 ]]> <![CDATA[X 6 ]]> V V V V <![CDATA[X 7 ]]> V V <![CDATA[X 8 ]]> <![CDATA[X 9 ]]> <![CDATA[X 10 ]]>

[0067] As shown in Table 1 above, the first measurement point Y 1 The corresponding target machine process parameter is machine process parameter X 2 , X 6 The second measuring point Y 2 The corresponding target machine process parameter is machine process parameter X 3 , X 4 , X 7 . Fifth measuring point Y 5 The corresponding target machine process parameter is machine process parameter X 2 , X 6 When the construction is obtained with each measurement point {Y 1 ,Y 2 ,Y 3 ,…,Y 5 After the measurement value prediction models M1~M5 of} are constructed, a linear model is established for each measurement point or the target machine process parameter corresponding to each measurement point, and an R-squared (coefficient of determination) operation is performed to obtain the R-squared value shown in Table 2 below.

[0068] Table 2

[0069]

[0070]

[0071] As shown in Table 2 above, for Table X2 , Y 1 For example, using X 2 and Y 1 A linear model is established and R-squared = 0.43 is calculated. 0.43 is then stored in the corresponding column of Table 2. 6 , Y 1 For example, using X 6 and Y 1 A linear model is established and R-squared = 0.1 is calculated, and 0.1 is stored in the corresponding column of Table 2. For other machine process parameters that are not the target machine process parameters corresponding to the measurement point, 0 can be added.

[0072] It can be understood that the above example takes the measurement value prediction model as a linear model, so the corresponding decision coefficient is R-squared. If the measurement value prediction model is not a linear model, the corresponding decision coefficient is not necessarily R-squared, but can be other coefficients of determination.

[0073] The calculation module 104 can convert the influence index value of each machine process parameter based on the number of times each machine process parameter appears in Table 2, the R-squared value and the parameter coefficient of each machine process parameter, as shown in Table 3. In actual situations, for each machine process parameter, multiple influence index values ​​may be included.

[0074] Table 3

[0075]

[0076]

[0077] As shown in Table 3 above, the machine process parameters X 1 Only occurs once, and the machine process parameter X 1 The R-squared value is 0.08. The machine process parameter X can be converted through the preset conversion function. 1 The influence index value of the machine process parameter X is 0.08. 2 Appears three times, and the machine process parameter X 2 The corresponding R-squared values ​​include 0.43, 0.28, and 0.22. The machine process parameter X can be converted through the preset conversion function. 2 The influence index value of the machine process parameter X is 1.597. 5 Occurs zero times, and the machine process parameter X can be converted through the preset conversion function 5 The influence index value of the machine process parameter X7 Occurs twice, and the machine process parameter X 7 The corresponding R-squared values ​​include 0.1 and 0.11. The machine process parameter X can be converted through the preset conversion function. 7 The influence index value of is 0.398. The preset conversion function is set based on the number of times the machine process parameter appears and the degree to which the machine process parameter can affect the measurement value.

[0078] The output module 105 is used to output warning information of the machine process parameter exceeding the first preset impact level indicator value.

[0079] In one embodiment, the first preset impact index value can be set and adjusted according to actual needs. For example, the first preset impact index value is set to 1.0, then in Table 3, the output module 105 outputs the machine process parameter X 2 , X 6 The warning information can be used to remind the machine user to perform machine maintenance in time to reduce machine downtime. For example, the machine process parameter X can be output on the display device of the abnormal monitoring device 100. 2 , X 6 Warning message.

[0080] In one embodiment, a plurality of preset conversion methods can be used to convert the number of occurrences of each of the machine process parameters in the problem indicator set and the parameter coefficient of each of the machine process parameters to obtain a plurality of influence degree index values ​​of each of the machine process parameters. When there are a plurality of influence degree index values ​​of each of the machine process parameters that exceed a first preset influence degree index value, an aggregate machine abnormality index can be output according to the plurality of influence degree index values ​​that exceed the first preset influence degree index value. The aggregate machine abnormality index can be used to describe the degree of abnormality of each of the machine process parameters so that the machine user can understand the cause and degree of the abnormality. For example, the aggregate machine abnormality index indicates an abnormality caused by a processing part of the machine 200.

[0081] Figure 3 The flowchart of the abnormality monitoring method in one embodiment of the present invention is shown in FIG. According to different requirements, the order of the steps in the flowchart can be changed, and some steps can be omitted.

[0082] Step S300 , based on a plurality of pre-extracted product data sets, a plurality of target machine process parameters that affect the measurement value of the preset product at the first measurement point are screened and obtained.

[0083] In one embodiment, the machine 200 is used to process the predetermined product. Assume that the machine 200 includes p machine process parameters X 1 ~X p, p is a positive integer greater than 1. The default product is set with N measurement points (defined as the first measurement point Y 1 To the Nth measuring point Y N ), N measurement points can be selected to measure the same product parameter of the preset product, and N is a positive integer greater than 1. For example, the preset product is a liquid crystal panel, and the preset product is set with N measurement points to measure the panel thickness. The multiple target machine process parameters are part or all of the multiple machine process parameters, for example, the multiple machine process parameters are Ω={X 1 ,X 2 ,X 3 ,…,X p}, multiple target machine process parameters are {X' 1 ,X' 2 ,…,X' k}, X' k ∈Ω and k≤p.

[0084] Multiple product data sets correspond to multiple preset products, and each product data set includes p machine process parameters {X 1 ,X 2 ,X 3 ,…,X p} and the corresponding preset product at the first measuring point Y 1 It can be understood that for a plurality of preset products, at the first measurement point Y 1 The measured values ​​may or may not be the same.

[0085] In one embodiment, the data (such as XML format files) thrown out by the machine 200 for processing the preset product and the measurement data of the preset product can be collected regularly or irregularly, and manually or automatically temporarily stored in a designated storage area. Then, the ETL tool is used to extract the specified data from the designated storage area and transfer it to the analysis database, and then the abnormal machine process parameters can be found out based on the analysis database. For example, the specified data may at least include the machine process parameters and the measurement values ​​of the measurement points. The analysis database may include at least a first data table, a second data table and a third data table, the first data table is used to store a plurality of the machine process parameters, the second data table is used to store the measurement values ​​of N measurement points, and the third data table is used to store the mapping relationship between the plurality of the machine process parameters and the measurement values ​​of each of the measurement points.

[0086] In one embodiment, the ETL tool can also extract the specified data from the specified storage area regularly or irregularly and transfer it to the analysis database. Multiple product data sets can be extracted from the analysis database.

[0087] In one embodiment, a plurality of pre-extracted product data sets may be analyzed based on a preset pre-screening algorithm to screen out a plurality of target machine process parameters that affect the measurement value of the preset product at the first measurement point. The preset pre-screening algorithm screens the target machine process parameters by determining a difference index of a combination of the plurality of machine process parameters. The difference index may include a mean-square error (MSE), which may reflect the difference between the estimated value and the estimated value. For the screening of the target machine process parameters, the smaller the difference index, the better, and the fewer the number of machine process parameters, the better.

[0088] For example, there are 10 machine process parameters, namely, machine process parameters {X 1 ,X 2 ,X 3 ,…,X 10}, multiple machine process parameter sets can be obtained by arranging and combining 10 machine process parameters, each of which can include at least one machine process parameter, and the difference index value corresponding to each machine process parameter set is determined according to the difference between the value of the first measurement point predicted by each machine process parameter set and the actual measurement value of the first measurement point. For multiple preset products, if there are one or more differences in the measurement values ​​at the first measurement point, the average value of the measurement value of the first measurement point can be calculated first, and then the difference between the average value and the value of the first measurement point predicted by each machine process parameter set can be calculated. Assume that the first machine process parameter set {X 1 ,X 2} corresponds to a difference index value of 0.3, and the second machine process parameter set {X 1 ,X 2 ,X 3} corresponds to a difference index value of 0.7, and the third machine process parameter set {X 2 ,X 4 ,X 7} corresponds to a difference index value of 0.1, and the fourth machine process parameter set {X 2 ,X 4 ,X 7 ,X 9} corresponds to a difference index value of 0.1. Assume that after comparative analysis, the minimum difference index value of multiple machine process parameter sets is 0.1, and the machine process parameter sets with a difference index value of 0.1 are the third machine process parameter set and the fourth machine process parameter set. Since the number of elements in the third machine process parameter set is less than the number of elements in the fourth machine process parameter set, the three machine process parameter sets {X 2 ,X 4 ,X 7} is defined as the target tool process parameter.

[0089] Step S302, constructing a measurement value prediction model corresponding to the first measurement point based on multiple target machine process parameters and multiple measurement values ​​of the preset products at the first measurement point, and calculating the fitting degree of the measurement value prediction model according to the predicted value of the first measurement point predicted by the measurement value prediction model.

[0090] In one embodiment, the plurality of target machine process parameters are machine process parameters that can more accurately predict the measurement value of the first measurement point. The measurement value prediction model M1 is preferably a linear model, and the measurement value prediction model M1 can calculate the predicted value of the first measurement point according to each target machine process parameter and the parameter coefficient corresponding to each target machine process parameter.

[0091] In one embodiment, a measurement value prediction model M1 corresponding to the first measurement point can be constructed based on multiple target machine process parameters and multiple measurement values ​​of the preset products at the first measurement point. For example, the multiple target machine process parameters are machine process parameters {X 2 ,X 4 ,X 7}, the measurement value prediction model M1 of the first measurement point is Y 1 =a*X 2 +b*X 4 +c*X 7 , can be based on the target machine process parameters {X 2 ,X 4 ,X 7} and the measured values ​​of multiple preset products at the first measurement point for regression analysis to determine the values ​​of parameter coefficients a, b, and c. After determining the values ​​of parameter coefficients a, b, and c, the machine process parameter {X 2 ,X 4 ,X 7 A set of parameter values ​​of} is input into the measurement value prediction model M1, and the predicted value of the first measurement point corresponding to the set of parameter values ​​is calculated.

[0092] Furthermore, when the value of the first measurement point can be predicted by the measurement value prediction model M1, the degree of fit (the degree of agreement between the prediction result of the model and the actual result) of the measurement value prediction model M1 can be calculated according to the prediction value of the first measurement point predicted by the measurement value prediction model M1. Specifically, the prediction value of the first measurement point predicted by the measurement value prediction model M1 and the actual measurement value of the first measurement point can be analyzed to obtain the degree of fit Rs1 of the measurement value prediction model M1.

[0093] In one embodiment, multiple linear models may be pre-constructed, and a suitable linear model may be selected according to the model index of each linear model to construct the measurement value prediction model M1 corresponding to the first measurement point. The model index may include mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE).

[0094] In one embodiment, since the output of the linear model is generally a one-dimensional value, when each of the N measurement points contains measurement values ​​of multiple dimensions, a preset dimensionality reduction function can be used to map the measurement value of each measurement point to a one-dimensional value. For example, the measurement value of the first measurement point is the deviation value Z in the three-axis direction. 1 , Z 2 , Z 3 , that is, it includes three dimensional values. The preset dimensionality reduction function can be set according to actual needs. For example, the preset dimensionality reduction function is g(x), After the measurement value of the first measurement point is reduced to one dimension, a measurement value prediction model M1 is constructed based on a plurality of target machine process parameters and the measurement values ​​of a plurality of preset products at the first measurement point.

[0095] In step S304, the fitting degree of the measurement value prediction model corresponding to the first measurement point, the prediction value of the first measurement point, the target tool process parameter corresponding to the first measurement point, and the parameter coefficient of the target tool process parameter corresponding to the first measurement point are aggregated into a problem indicator set.

[0096] In one embodiment, for example, the measurement value prediction model M1 corresponding to the first measurement point is Y 1 =0.2*X 2 +0.3*X 4 +0.1*X 7 , then the fitting degree of the measurement value prediction model M1 corresponding to the first measurement point is the fitting degree Rs1, the predicted value of the first measurement point is the value predicted by the measurement value prediction model M1, and the target machine process parameter corresponding to the first measurement point is {X 2 ,X 4 ,X 7}, the parameter coefficient of the target tool process parameter corresponding to the first measurement point is {0.2, 0.3, 0.1}.

[0097] It can be understood that for the second measurement point, the same processing method as the first measurement point is adopted to obtain the measurement value prediction model M2 corresponding to the second measurement point. The degree of fit of the measurement value prediction model M2 corresponding to the second measurement point, the predicted value of the second measurement point, the target machine process parameter corresponding to the second measurement point, and the parameter coefficient of the target machine process parameter corresponding to the second measurement point can be summarized into the problem indicator set. For the Nth measurement point, the same processing method as the first measurement point can be adopted to obtain the measurement value prediction model MN corresponding to the Nth measurement point. The degree of fit of the measurement value prediction model MN corresponding to the Nth measurement point, the predicted value of the Nth measurement point, the target machine process parameter corresponding to the Nth measurement point, and the parameter coefficient of the target machine process parameter corresponding to the Nth measurement point can be summarized into the problem indicator set.

[0098] Step S306, repeat the above steps until the fitting degree of the measurement value prediction model corresponding to the Nth measurement point, the prediction value of the Nth measurement point, the target machine process parameter corresponding to the Nth measurement point, and the parameter coefficient of the target machine process parameter corresponding to the Nth measurement point are integrated into the problem indicator set.

[0099] Step S308 , calculating the influence index value of each of the machine process parameters based on the elements in the problem index set.

[0100] In one embodiment, the influence index value can reflect the importance of the machine process parameter that affects the processing of the preset product by the machine 200. The larger the calculated influence index value, the more important the machine process parameter is, and the machine process parameter may be abnormal. The influence index value of each machine process parameter can be calculated by selecting a preset conversion function and the elements in the problem index set.

[0101] Step S310 , outputting warning information of a machine process parameter exceeding a first preset impact level indicator value.

[0102] In one embodiment, the first preset impact index value can be set and adjusted according to actual needs. For example, the first preset impact index value is set to 1.0, then in Table 3, the machine process parameter X can be output. 2 , X 6 The warning information can be used to remind the machine user to perform machine maintenance in time to reduce machine downtime. For example, the machine process parameter X can be output on the display device of the abnormal monitoring device 100. 2 , X 6 Warning message.

[0103] In one embodiment, a plurality of preset conversion methods can be used to convert the number of occurrences of each of the machine process parameters in the problem indicator set and the parameter coefficient of each of the machine process parameters to obtain a plurality of influence degree index values ​​of each of the machine process parameters. When there are a plurality of influence degree index values ​​exceeding a first preset influence degree index value for each of the machine process parameters, an aggregated machine abnormality index can be output according to the plurality of influence degree index values ​​exceeding the first preset influence degree index value. The aggregated machine abnormality index can be used to describe the degree of abnormality of each of the machine process parameters so that the machine user can understand the cause and degree of the abnormality. For example, the aggregated machine abnormality index indicates an abnormality caused by a processing part of the machine 200. For example, when the aggregated machine abnormality indicators of a plurality of machine process parameters are all related to the processing part A of the machine 200, the aggregated machine abnormality index can be information reflecting that the processing part A may have an abnormality.

[0104] The above-mentioned abnormal monitoring device, method and computer-readable storage medium for machine process parameters use multi-point measurement data and machine process data to construct a problem indicator set and analyze it, so as to discover abnormal machine process parameters as soon as possible, reduce machine downtime, and improve process quality.

[0105] For those skilled in the art, other corresponding changes or adjustments can be made based on the inventive scheme and inventive concept of the present invention in combination with actual production needs, and these changes and adjustments should all fall within the scope disclosed by the present invention.

Claims

1. A method for monitoring abnormality of a machine process parameter, characterized in that: The plurality of machine process parameters are used to process a plurality of preset products, the preset products are provided with a first measurement point to an Nth measurement point, the N measurement points are used to measure the same product parameter of the preset products, N is a positive integer greater than 1, and the abnormality monitoring method comprises: Based on the pre-extracted multiple product data sets, multiple target machine process parameters that affect the measurement value of the preset product at the first measurement point are screened, wherein the multiple target machine process parameters are part of or all of the multiple machine process parameters, the multiple product data sets correspond to the multiple preset products, and each product data set includes the multiple machine process parameters and the measurement value of the corresponding preset product at the first measurement point; A measurement value prediction model corresponding to the first measurement point is constructed based on a plurality of the target machine process parameters and the measurement values ​​of a plurality of the preset products at the first measurement point, so as to calculate the fitting degree of the measurement value prediction model according to the predicted value of the first measurement point predicted by the measurement value prediction model, wherein the measurement value prediction model calculates the predicted value of the first measurement point according to each of the target machine process parameters and the parameter coefficient corresponding to each of the target machine process parameters; Aggregating the fitting degree of the measurement value prediction model corresponding to the first measurement point, the prediction value of the first measurement point, the target tool process parameter corresponding to the first measurement point, and the parameter coefficient of the target tool process parameter corresponding to the first measurement point into a problem indicator set; Repeat the above steps until the fitting degree of the measurement value prediction model corresponding to the Nth measurement point, the prediction value of the Nth measurement point, the target tool process parameter corresponding to the Nth measurement point, and the parameter coefficient of the target tool process parameter corresponding to the Nth measurement point are integrated into the problem indicator set; Calculate the influence index value of each of the machine process parameters based on the elements in the problem index set; and Outputting warning information of the machine process parameter exceeding the first preset impact level indicator value.

2. The method for monitoring abnormality of a machine process parameter according to claim 1, wherein: The method further comprises: Collecting the processing parameters of the machine for processing the preset product and the measurement parameters of the preset product; Extracting designated data from the collected machine processing parameters and the measurement parameters and storing them in an analysis database, wherein the analysis database at least includes a first data table, a second data table, and a third data table, wherein the first data table is used to store a plurality of the machine process parameters in the designated data, the second data table is used to store the measurement values ​​of N measurement points in the designated data, and the third data table is used to store the mapping relationship between the plurality of the machine process parameters and the measurement value of each measurement point; A plurality of product data sets are extracted from the analysis database.

3. The method for monitoring abnormality of a machine process parameter according to claim 2, wherein: The step of constructing an analysis database based on the collected machine processing parameters and the measurement parameters includes: A preset ETL tool is used to extract a plurality of the machine process parameters and the measurement values ​​of the N measurement points from the collected machine processing parameters and the measurement parameters, and load the values ​​into the analysis database.

4. The method for monitoring abnormality of a machine process parameter according to claim 1, wherein: Before the step of constructing a measurement value prediction model corresponding to the first measurement point based on the plurality of target machine process parameters and the measurement values ​​of the plurality of preset products at the first measurement point, the method includes: If the measurement value of the first measurement point includes multiple dimensional values, the measurement value of the first measurement point is mapped into a one-dimensional value by using a preset dimensionality reduction function.

5. The method for monitoring abnormality of a machine process parameter according to claim 1, wherein: The measurement value prediction model is a linear model.

6. The method for monitoring abnormality of a machine process parameter according to claim 5, wherein: The linear model includes a plurality of linear coefficients, and the parameter coefficients of a plurality of target tool process parameters correspond to the plurality of linear coefficients in a one-to-one manner.

7. The method for monitoring abnormality of a machine process parameter according to claim 1, wherein: The step of calculating the influence index value of each of the machine process parameters based on the elements in the problem index set includes: The influence index value of each of the tool process parameters is obtained by conversion based on the number of occurrences of each of the tool process parameters in the problem index set and the parameter coefficient of each of the tool process parameters.

8. The method for monitoring abnormality of a machine process parameter according to claim 7, wherein: The step of converting the influence index value of each of the machine process parameters based on the number of occurrences of each of the machine process parameters in the problem index set and the parameter coefficient of each of the machine process parameters comprises: Using a plurality of preset conversion methods to convert the number of occurrences of each of the machine process parameters in the problem indicator set and the parameter coefficient of each of the machine process parameters, to obtain a plurality of influence degree indicator values ​​of each of the machine process parameters; When each of the machine process parameters has multiple influence level indicator values ​​exceeding the first preset influence level indicator value, an aggregate machine abnormality indicator is obtained based on the multiple influence level indicator values ​​exceeding the first preset influence level indicator value to describe the abnormality level of each of the machine process parameters.

9. A device for monitoring abnormality of process parameters of a machine platform, the device comprising a processor and a memory, wherein the memory stores a plurality of computer programs, wherein: The processor is used to implement the steps of the method for abnormal monitoring of machine process parameters as described in any one of claims 1 to 8 when executing the computer program stored in the memory.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a plurality of instructions, and the plurality of instructions can be executed by one or more processors to implement the steps of the method for abnormal monitoring of process parameters of a machine platform as described in any one of claims 1-8.

Images