Control limit determination method, device, equipment and computer storage medium

Abstract

This application discloses a control limit determination method, apparatus, device, and computer storage medium. The control limit determination method includes: obtaining process data from a previous control cycle during a current control cycle of a statistical process control system; and inputting the process data from the previous control cycle into a trained control limit prediction model to obtain control limits for the current control cycle. This embodiment can automatically determine control limits based on the trained control limit prediction model, effectively improving the flexibility and reliability of control limit determination in statistical process control.

Description

Technical Field

[0001] The present application belongs to the field of data processing technology, and in particular relates to a control limit determination method, apparatus, device, and computer storage medium. Background Art

[0002] As we all know, Statistical Process Control (SPC) is a process control tool that utilizes mathematical statistics. Control charts are commonly used in SPC to analyze and determine whether a process is stable. The control limits within a control chart have a significant impact on the accuracy of this analysis and determination. In related technologies, control limits in control charts are typically manually defined by the user, resulting in limited flexibility. Summary of the Invention

[0003] The embodiments of the present application provide a control limit determination method, apparatus, device, and computer storage medium to address the problem that control limits in related art control charts are usually determined manually by users and have poor flexibility.

[0004] In a first aspect, an embodiment of the present application provides a method for determining a control limit, the method comprising:

[0005] In the current control cycle of statistical process control, process data of the previous control cycle is obtained;

[0006] The process data of the previous control cycle is input into the trained control limit prediction model to obtain the control limits of the current control cycle.

[0007] In a second aspect, an embodiment of the present application provides a method for determining a control limit, the method comprising:

[0008] In the process of statistical process control for an industrial production line, process data of the industrial production line in the previous control cycle is obtained;

[0009] The process data of the previous control cycle is input into the pre-trained control limit prediction model to obtain the control limits of the preset production parameters in the industrial production line in the current control cycle.

[0010] In a third aspect, an embodiment of the present application provides a control limit determination device, the device comprising:

[0011] A first acquisition module is used to acquire process data of a previous control cycle in a current control cycle of the statistical process control;

[0012] The first determination module is used to input the process data of the previous control cycle into the trained control limit prediction model to obtain the control limit of the current control cycle.

[0013] In a fourth aspect, an embodiment of the present application further provides a control limit determination device, the device comprising:

[0014] A third acquisition module is used to acquire process data of the industrial production line in the previous control cycle during the statistical process control of the industrial production line;

[0015] The third determination module is used to input the process data of the previous control cycle into a pre-trained control limit prediction model to obtain the control limits of the preset production parameters in the industrial production line in the current control cycle.

[0016] In a fifth aspect, an embodiment of the present application provides an electronic device, the device comprising:

[0017] a processor and a memory storing computer program instructions;

[0018] When the processor executes the computer program instructions, the control limit determination method as shown in the first aspect or the control limit determination method as shown in the second aspect is implemented.

[0019] In a sixth aspect, an embodiment of the present application provides a computer storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the control limit determination method as shown in the first aspect is implemented, or the control limit determination method as shown in the second aspect is implemented.

[0020] In the seventh aspect, an embodiment of the present application provides a computer program product. When the instructions in the computer program product are executed by the processor of an electronic device, the electronic device executes the control limit determination method as shown in the first aspect, or implements the control limit determination method as shown in the second aspect.

[0021] The control limit determination method provided in an embodiment of the present application obtains process data from the previous control cycle during the current SPC control cycle and inputs this data into a trained control limit prediction model to obtain control limits for the current control cycle. This embodiment can automatically determine control limits based on the trained control limit prediction model, effectively improving the flexibility and reliability of control limit determination in SPC. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0023] Figure 1 is an example diagram of a system architecture to which the control limit determination method according to an embodiment of the present application can be applied;

[0024] Figure 2 1 is a flow chart of a control limit determination method provided in an embodiment of the present application;

[0025] Figure 3 This is a flow chart of another control limit determination method provided in an embodiment of the present application;

[0026] Figure 4 This is a flow chart of the control limit determination method in a specific application example;

[0027] Figure 5 is a structural diagram of a control limit determination device provided in an embodiment of the present application;

[0028] Figure 6 This is a structural diagram of another control limit determination device provided in an embodiment of the present application;

[0029] Figure 7 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0030] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is merely to provide a better understanding of the present application by illustrating the examples of the present application.

[0031] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0032] Statistical Process Control (SPC) can apply statistical techniques to evaluate and monitor each stage of an enterprise's production or management process, establish and maintain the process at an acceptable and stable level, and thus ensure that the product or service meets the specified requirements.

[0033] SPC can be applied to production processes in a company, depending on the application scenario. For example, in vinyl chloride production, SPC can monitor vinyl chloride production capacity, mixer temperature, or other production parameters. Another example is that in beverage bottling, SPC can monitor production parameters such as the volume of each bottle filled.

[0034] In other application scenarios, SPC can also be applied in enterprise management processes. For example, SPC can monitor and evaluate parameters such as energy consumption or service scores.

[0035] To simplify the description, the above-mentioned production parameters or parameters such as energy consumption and service score can be collectively referred to as controlled parameters.

[0036] Control charts are often used in SPC. Control charts are charts with control limits used to analyze and determine whether a process is in a stable state. They usually have the function of distinguishing normal fluctuations from abnormal fluctuations in the process.

[0037] Control limits, also known as control limits, are the basis for determining process stability on control charts. When the value of a controlled parameter exceeds the control limit, it generally indicates that a special cause is affecting the process. While these are just a few examples of using control limits to determine process stability, in practice, process stability can be determined by combining the trend of the controlled parameter relative to the control limit at multiple time points.

[0038] In the related art, the control limits used in SPC are usually determined in the following manner.

[0039] The first is to make the determination based on the user's experience, which requires a high level of professionalism from the user.

[0040] The second method is to determine the control limits based on statistical methods such as the 3σ principle. Taking the 3σ principle (also known as the Laida criterion) as an example, when determining the control limits, the historical data of the controlled parameters can be statistically analyzed to obtain the mean μ and standard deviation σ of the controlled parameters, and μ+3σ is used as the upper control limit and μ-3σ is used as the lower control limit. However, this statistical method often fails to take into account the characteristics of the controlled parameters themselves, making it difficult to ensure the rationality of the control limits. In practical applications, on the basis of determining the control limits based on statistical methods, users are often required to further manually adjust the control limits.

[0041] To address the problems of the prior art, embodiments of the present application provide a control limit determination method, apparatus, device, and computer storage medium, which can be applied in scenarios where SPC is implemented.

[0042] The control limit determination method provided in the embodiments of the present application can be applied to electronic devices. For example, the electronic device can be a mobile electronic device such as a personal computer or a mobile terminal, or the electronic device can be a server, an industrial computer in a production scenario, an edge computing unit, or other types of non-mobile electronic devices. No specific limitation is made here.

[0043] like Figure 1 As shown, Figure 1 1 is an example diagram of a system framework to which the control limit determination method according to an embodiment of the present application can be applied. The system framework may include the electronic device 10, production line equipment 20, and sensor 30 described above.

[0044] In one example, the production line equipment 20 may include a mixer 21 and a cooler 22 used in vinyl chloride production. The mixer 21 may be used for mixing and reacting acetylene and hydrogen chloride gases, and the cooler 22 may be used for cooling the mixer 21 .

[0045] The sensor can be used to collect sensor data related to the production line equipment 20, for example, Figure 1 As shown, sensor S1 can be used to collect the vinyl chloride content in the output gas of mixer 21, sensor S2 can be used to collect the temperature and humidity of mixer 21, sensor S3 can be used to collect the internal pressure of mixer 21, sensor S4 can collect the cooling fluid flow rate, sensor S5 can collect the temperature and humidity of cooler 22, sensor S6 can collect the internal pressure of cooler 22, and sensor S7 can collect the cooling fluid temperature, etc. Of course, the above are some examples of the configuration of sensor 30. In actual applications, sensor 30 can be used to collect other data from production line equipment 20, which will not be explained here one by one.

[0046] The electronic device 10 may include a processor and a communication module. The communication module may be used to communicate with the sensor 30 to receive sensor data collected by the sensor 30, while the processor may process the sensor data.

[0047] In some application scenarios, the electronic device 10 may include a display device for displaying the SPC control chart. The electronic device 10 may also include a user input device to facilitate user intervention in the SPC, such as forcing adjustment of the control limits or disabling the control limit adjustment function.

[0048] The data collected by the sensor 30 can be used as process data used in the control limit determination method. In some examples, the process data can include controlled parameters in the SPC control chart, such as the temperature of the mixer 21.

[0049] After the electronic device 10 obtains the process data, it can process the process data, for example, determine whether the process data meets the conditions for adjusting the control limits, or predict the process data based on a pre-stored control limit prediction model to obtain new control limits to achieve automatic adjustment of the control limits.

[0050] The following first introduces the control limit determination method provided in the embodiment of the present application.

[0051] Figure 2 A flow chart of a control limit determination method provided in one embodiment of the present application is shown.

[0052] like Figure 2 As shown, the method includes:

[0053] Step 201, in the current control cycle of SPC, obtain process data of the previous control cycle;

[0054] Step 202: Input the process data of the previous control cycle into the trained control limit prediction model to obtain the control limits of the current control cycle.

[0055] Control cycles can be pre-set in SPC. In some examples, one control cycle can correspond to a preset time period; or, in the product production process, one control cycle can correspond to the time it takes to produce a preset number of products, etc.

[0056] In one control cycle, process data of the control cycle can be obtained.

[0057] For example, process data within a control cycle can include the control limits used within that control cycle. For another example, in the aforementioned vinyl chloride production scenario, process data within a control cycle can include data collected by sensors at multiple sampling points within that control cycle, such as mixer temperature or humidity.

[0058] Accordingly, in step 201 , in the current control cycle of the SPC, the electronic device may obtain the previous control cycle, ie, the process data in the previous control cycle.

[0059] In some examples, the process data in the previous control cycle may be acquired from a related storage device, or may be obtained in response to a user's data input operation, and so on.

[0060] In step 202 , process data of the previous control cycle may be input into a trained control limit prediction model. To simplify the description, the trained control limit prediction model may be referred to as a target control limit prediction model hereinafter.

[0061] The control limit prediction model can be a deep learning model, such as a neural network model. The target control limit prediction model can be obtained by fully training the control limit prediction model using a relevant training sample set. By setting the training sample set, the target control limit prediction model can predict control limits based on the input process data.

[0062] For example, the training samples in the training sample set may include historical process data and corresponding annotation results, wherein the historical process data may be process data obtained in a historical control cycle, and the annotation results may be the control limits used in the next control cycle of the historical control cycle.

[0063] Of course, the above are some examples of training sample sets used to train control limit prediction models. In actual applications, training sample sets can also be established based on manual experience. For example, process data can be set based on manual experience and control limits can be manually labeled on the set process data.

[0064] In general, when training a control limit prediction model, the input data can be the sample process data, and the prediction results can converge towards the labeled control limits. Correspondingly, the input of the target control limit prediction model can be the process data of a control cycle, and the output can be the predicted control limits.

[0065] Specifically in step 202 , the input of the target control limit prediction model may be the process data of the previous control cycle, and the output may be the control limit of the current control cycle.

[0066] The control limit determination method provided in an embodiment of the present application obtains process data from the previous control cycle during the current SPC control cycle and inputs this data into a trained control limit prediction model to obtain control limits for the current control cycle. This embodiment can automatically determine control limits based on the trained control limit prediction model, effectively improving the flexibility and reliability of control limit determination in SPC.

[0067] Generally speaking, control limits with higher reliability can effectively guide production and achieve the goal of energy saving and efficiency improvement.

[0068] According to one embodiment of the present application, the process data of the previous control cycle includes control limits of the previous control cycle;

[0069] Input the process data of the previous control cycle into the trained control limit prediction model to obtain the control limits of the current control cycle, including:

[0070] Based on the trained control limit prediction model, a first feature is extracted from the process data of the previous control cycle, and the control limit adjustment rate is determined based on the first feature. The control limits of the current control cycle are determined based on the control limit adjustment rate and the control limits of the previous control cycle; wherein the first feature is a feature that affects the adjustment rate of the control limits.

[0071] To simplify the description, the process data of the last control cycle may be referred to as first process data.

[0072] In the vinyl chloride production scenario described above, the first process data may include mixer temperature, humidity and pressure, exhaust gas temperature, and specific gas component content. The mixer temperature can be the controlled parameter in SPC. The first process data may also include the control limits set for the controlled parameters in the previous control cycle, corresponding to the control limits of the previous control cycle.

[0073] In some examples, in the first process data, the mixer temperature may be time series data, that is, the first process data may include the mixer temperature obtained at P sampling time points, which is recorded as Here, i is an integer greater than 1, and i-1 can represent the previous control cycle.

[0074] After the first process data is input into the trained control limit prediction model (hereinafter referred to as the target control limit prediction model), the target control limit prediction model can perform feature extraction on the data.

[0075] For example, the first process data may include a first feature extraction network that, when trained, can extract The first feature of the type of numerical distribution characteristics, numerical change characteristics, etc.

[0076] Of course, the first feature extraction network can also extract features from other types of data in the first process data. For example, in practical applications, factors such as mixer cooler pressure and exhaust gas temperature may indirectly affect the determination of control limits. During training, the control limit prediction model can learn the impact of other types of data on temperature control limits. At the application level, the target control limit prediction model can extract the aforementioned first feature based on data such as mixer cooler pressure and exhaust gas temperature in the first process data.

[0077] In this embodiment, the first feature may be a feature that affects the adjustment rate of the control limit, and the target control limit prediction model may determine the control limit adjustment rate according to the first feature.

[0078] With some examples, the target control limit prediction model may include a first feature extraction network and a first classification network, etc., wherein the first feature extraction network can extract the first feature, and the first classification network can classify the first feature to obtain a control limit adjustment rate, denoted as g.

[0079] Of course, this is just an example of the network architecture of the target control limit prediction model. In actual applications, the network architecture of the target control limit prediction model can be adjusted as needed.

[0080] The first process data includes the control limits of the previous control cycle, denoted as X i-1 When the target control limit prediction model obtains the control limit adjustment rate g, it can further calculate the control limit adjustment rate g and the control limit X of the previous control cycle. i -1 , determine the control limits for the current control cycle.

[0081] The control limit of the current control cycle is denoted as X i In one example, X can be determined as follows i :

[0082] X i =X i-1 (1+g) (1)

[0083] It is easy to understand that the target control limit prediction model is a trained prediction model. The extraction of the first feature, the determination of the control limit adjustment rate g, and the control limit X of the current control cycle are i The determination of is automatically performed in the target control limit prediction model. This can be considered as an example of the data flow in the target control limit prediction model. The specific processing method for this data can be determined by the network architecture and parameters of the target control limit prediction model, which will not be explained in detail here.

[0084] In this embodiment, the target control limit prediction model can extract the first feature from the first feature data, determine a control limit adjustment rate based on the first feature, and use the control limit adjustment rate to adjust the control limits of the previous control cycle to obtain the control limits of the current control cycle. This helps to more reasonably adjust the control limits based on the performance of the process data in SPC, so that the adjusted control limits better meet actual SPC requirements.

[0085] In addition, in this embodiment, the target control limit prediction model is based on the feature extraction of the first process data, which helps to obtain some features that affect the control limit adjustment rate that are difficult to obtain through manual analysis, thereby helping to improve the rationality of the control limit adjustment rate and make the control limits determined for the current control cycle better meet actual needs.

[0086] According to one embodiment of the present application, determining a control limit adjustment rate according to the first feature includes:

[0087] determining the control limit adjustment step size based on the first characteristic;

[0088] The control limit adjustment rate is determined based on the control limit adjustment step and the preset adjustment rate limit. The preset adjustment rate limit is used to limit the numerical range of the control limit adjustment rate.

[0089] In some application scenarios, when the control limit needs to be adjusted in the current control cycle, the control limit X in the current control cycle in the above formula (1) is i If you need to compare with X i-1 When the control limit adjustment rate g is adjusted, it is usually necessary to ensure that it is not equal to 0.

[0090] Of course, in some special scenarios, the adjustment of the control limit is also affected by other interference factors. When the control limit adjustment rate g is not equal to 0, X i Still possible with X i-1 The main purpose of this embodiment is to adjust the control limits as much as possible when the control limits need to be adjusted.

[0091] In this embodiment, the control limit adjustment step can be determined according to the first feature, which is denoted as {β i}, as for the control limit adjustment step {β i The process of} can be realized based on the preset network architecture and parameters in the target control limit prediction model, which will not be described here.

[0092] The control limit adjustment rate is determined based on the control limit adjustment step and the preset adjustment rate limit. That is, in this embodiment, the determination of the control limit adjustment rate may be affected by multiple factors, including the control limit adjustment step {β i}.

[0093] For ease of explanation, the control limit adjustment rate g can be expressed as the result function g() of its influencing factors. If the preset adjustment rate limit is recorded as C, the control limit adjustment rate g can be further expressed as g({β i},C).

[0094] Accordingly, formula (1) can be changed to:

[0095] X i =X i-1 (1+g({β i},C)) (2)

[0096] Take an example, without considering the adjustment rate limit C, there is the following function calculation relationship: g({β i})=A{β i} 2 +B{β i}, where A and B are network parameters in the target control limit prediction model. If the control limit adjustment step size {β i} is 0, it may lead to the control limit adjustment rate g({β i})=0, which results in the control limits not changing in the current control cycle.

[0097] The adjustment rate limit C may correspond to a preset value, which may be slightly greater than 0. In consideration of the adjustment rate limit C, the following function calculation relationship exists: g({β i},C)=A{β i} 2 +B{β i}+C, in this case, it can effectively avoid g({β i},C) is equal to 0.

[0098] It is easy to understand that the above functional calculation relationship of the control limit adjustment rate g can be reflected in the network framework and parameters of the target control limit prediction model, which will not be described in detail here.

[0099] In practical applications, the adjustment rate limit can be used to prevent the control limit adjustment rate from being equal to 0, and can also be used for other functions.

[0100] For example, in another example, there may be the following relationship g({β i},C)=max(g({β i}), C), where g({β i}) is the control limit adjustment rate obtained without considering the adjustment rate limit, and C is the preset upper limit of the adjustment rate, which corresponds to the above-mentioned adjustment rate limit.

[0101] In general, in this embodiment, the preset adjustment rate limit can limit the numerical range of the control limit adjustment rate, avoid the target control limit prediction model from obtaining an abnormal control limit adjustment rate, and improve the rationality of the control limit determined in the current control cycle.

[0102] According to one embodiment of the present application, the process data of the previous control cycle is input into the trained control limit prediction model to obtain the control limits of the current control cycle, including:

[0103] Based on the trained control limit prediction model, a second feature is extracted from the process data of the previous control cycle, and the control limit adjustment amount is determined based on the second feature. The control limit of the current control cycle is determined based on the control limit adjustment amount and the control limit of the previous control cycle; wherein the second feature is a feature that affects the adjustment amount of the control limit.

[0104] In the above embodiment, the target control limit prediction model can further determine the control limits of the current control cycle by determining the control limit adjustment rate. This embodiment can be considered a variation of the above embodiment, or can also be considered a further improvement based on the above embodiment.

[0105] In the case that this embodiment is a variation of the above embodiment, this embodiment may replace the control limit adjustment rate in the above embodiment with a control limit adjustment amount.

[0106] For example, if the control limit adjustment is recorded as f, the following relationship can exist:

[0107] X i =X i-1 +f (3)

[0108] Specifically, in some examples, the target control limit prediction model may include a second feature extraction network and a second classification network. The second feature extraction network can extract second features from the first process data, and these second features are respectively recorded as Wherein, n is a positive integer, which can be one of the hyperparameters of the second feature extraction network.

[0109] The second classification network can classify the second feature and obtain the control limit adjustment f. In other words, f can be expressed as The function of Formula (3) can be further expressed as:

[0110]

[0111] The above are some examples of the network architecture of the target control limit prediction model.

[0112] In practical applications, by adjusting the network architecture or parameters of the target control limit prediction model, the target control limit prediction model can be configured to determine the control limits based on the control limit adjustment amount, instead of determining the control limits based on the control limit adjustment rate. For at least the specific implementation methods and corresponding effective effects, please refer to the description of the above embodiment and will not be repeated here.

[0113] As shown above, this embodiment can also be a further improvement based on the above embodiment. That is, in this embodiment, the target control limit prediction model can determine the control limit of the current control cycle based on the control limit adjustment rate and the control limit adjustment amount.

[0114] For example, the control limits for the current control period can be determined as follows:

[0115]

[0116] Considering the adjustment rate limit, formula (5) can be further expressed as:

[0117]

[0118] The above methods for determining the control limits of the current control cycle can all be achieved through the network architecture and parameter settings of the target control limit prediction model, which will not be elaborated here.

[0119] Combined with the production scenario of vinyl chloride, the g function (for example, g({β i},C)) can be considered as a multi-granularity adaptive adjustment rate, which can be mainly composed of The distribution range of these mixer temperatures is determined by the f function (e.g. ) can represent the adjustment amount caused by various service influencing factors, which can be determined by factors such as mixer cooler pressure changes and reset process parameters.

[0120] Of course, the above are some examples of the meanings of the f function and the g function. In practical applications, the calculation methods corresponding to the f function and the g function can be reflected in the network architecture and parameters of the target control limit prediction model. Accordingly, the content that affects the values ​​of the f function and the g function can be the more intuitive data mentioned in the above examples, or it can be the deep features learned through the target control limit prediction model, etc.

[0121] As mentioned above, in some special scenarios, the adjustment of control limits is also affected by other interference factors. When the control limit adjustment rate g is not equal to 0, X i Still possible with X i-1 Specifically in this embodiment, when the numerical variation trends of the f function and the g function are opposite, for example, when the value of the f function is negative and the value of the g function is positive, X may appear. i With X i-1 Equal situation.

[0122] It can be seen that when determining the control limits of the current control cycle based on the control limit adjustment rate and the control limit adjustment amount, this embodiment can fully consider various factors causing the control limit adjustment, thereby helping to effectively improve the rationality of the control limits of the current control cycle.

[0123] According to one embodiment of the present application, the process data of the previous control cycle includes data of preset hyperparameters;

[0124] Determining a control limit adjustment amount based on the second characteristic includes:

[0125] Determine the control limit adjustment amount based on the second feature and the data of the preset hyperparameters.

[0126] In combination with the vinyl chloride production scenario mentioned above, when the mixer temperature is used as the controlled parameter, there may be some data in the first process data that directly affects the mixer temperature, such as ambient temperature, coolant flow rate of the cooler, etc. These data can be collectively referred to as the first category of data, which corresponds to the data of the above-mentioned preset hyperparameters.

[0127] The data in the first process data, such as mixer pressure and humidity, may indirectly affect or may not affect the mixer temperature. These data can be collectively referred to as second-category data.

[0128] The target control limit prediction model in this embodiment can process the above-mentioned types of data in corresponding manners.

[0129] For example, the target control limit prediction model can directly use the first type of data as a factor in determining the control limit adjustment amount. For the second type of data, it can learn the features that influence the control limit adjustment amount, namely the second features mentioned above. Of course, in practical applications, different data may have certain correlations. When extracting the second feature through the feature extraction network, the input data can be the complete first process data.

[0130] Combined with the expression of the f function above In the data, one part can be the data of the preset hyperparameters, and the other part can be the feature value of the second feature.

[0131] In one example, the target control limit prediction model may include an input network, a second feature extraction network, a fully connected network, and a second classification network. The input network can be used to receive first process data, the second feature extraction network can extract features from the first process data, the fully connected network can connect the feature extraction network and the nodes corresponding to preset hyperparameters in the input network, and the second classification network can perform classification based on the output of the fully connected network to obtain the control limit adjustment amount.

[0132] Of course, this is just an example of the implementation of the target control limit prediction model. In actual applications, the network architecture of the target control limit prediction model can be adjusted according to actual needs, and no examples will be given here.

[0133] In this embodiment, the target control limit prediction model is used to realize automatic adjustment of the control limit. Based on the learning ability of the target control limit prediction model, the shortcomings of the related technologies such as difficulty in integrating service influencing factors and needing expert assistance can be avoided, and the setting of the control limit can be better adapted to the actual service.

[0134] According to one embodiment of the present application, before obtaining the process data of the previous control cycle, the method further includes:

[0135] Acquire multiple training samples, where the training samples include process training data and first control limits corresponding to the process training data;

[0136] The control limit prediction model is trained using multiple training samples so that the trained control limit prediction model predicts the control limits based on the input process data.

[0137] This embodiment can be considered as an embodiment related to the training method of the control limit prediction model. Combined with the above, the trained control limit prediction model can be referred to as the target control limit prediction model, and accordingly, the control limit prediction model that has not been trained can be referred to as the initial control limit prediction model.

[0138] In this embodiment, the initial control limit prediction model may be trained using training samples, wherein the training samples may include process training data and corresponding first control limits.

[0139] By way of example, process training data can be historical process data of a related process. For example, in the process control of vinyl chloride production, process training data can be process data collected through sensors or other means during historical production activities, such as mixer temperature or humidity.

[0140] In other examples, the process training data may also be data set by the user based on experience, that is, the process training data may be obtained based on user input without being collected from devices such as sensors.

[0141] The process training data may correspond to a first control limit, and the first control limit may be considered as a labeling result of the process training data.

[0142] In the example above, the process training data can be historical process data, such as process data from a historical time period, and the corresponding first control limits can be control limits determined after the end of the historical time period. In practical applications, the control limits determined after the end of the historical time period may or may not be adjusted, but both can be used as the annotation results of the process training data.

[0143] In the case where the process training data is set by the user based on experience, the first control limit may also be determined by the user based on experience, and the first control limit determined by the user's experience may be used as the labeling result of the process training data.

[0144] In this embodiment, multiple training samples can be used to train the control limit prediction model. As described above regarding training samples, the process training data can serve as input to the initial control limit prediction model. Ideally, the output of the initial control limit prediction model should be consistent with or similar to the labeled results, namely, the first control limits described above. In other words, if the initial control limit prediction model is fully trained, when the process training data is input into the initial control limit prediction model, the output should be consistent with or similar to the first control limits.

[0145] The target control limit prediction model can be considered a fully trained initial control limit prediction model, and the input and output data types of the target control limit prediction model are often the same. Specifically, the input of the target control limit prediction model can be process data, and the output can be the predicted control limits.

[0146] In this embodiment, the control limit prediction model can be trained based on the training samples, so that the trained control limit prediction model can predict the control limit according to the input process data. In this way, in the subsequent SPC process, the control limit of the current control cycle can be determined using the pre-trained control limit prediction model, thereby improving the flexibility and convenience of adjusting the control limit in SPC.

[0147] In some embodiments, the multiple training samples obtained may be data-cleaned training samples. In conjunction with the above method of obtaining training samples based on historical process data, abnormal data may appear in the historical process data due to sensor abnormalities or process control abnormalities.

[0148] For example, when a sensor fails, the sensor data reported may be abnormal data. For another example, when the control limit set manually in the process control significantly exceeds the reaction temperature required for product preparation, the set control limit can be considered as abnormal data.

[0149] When establishing training samples, this part of abnormal data can be cleaned so that the target control limit prediction model obtained by training the training samples has higher reliability.

[0150] According to one embodiment of the present application, training a control limit prediction model using multiple training samples includes:

[0151] Input the process training data into the control limit prediction model and output the second control limit;

[0152] The loss value of the loss function in the control limit prediction model is determined according to the first control limit and the second control limit, and the parameters of the control limit prediction model are adjusted according to the loss value.

[0153] In this embodiment, the process training data may be specifically input into the aforementioned initial control limit prediction model, that is, input into a control limit prediction model that has not been fully trained.

[0154] As shown above, the control limit prediction model can be a neural network model, etc. When the network architecture is pre-built, the control limit prediction model can predict the input process training data based on the initial network parameters to obtain the second control limit.

[0155] Generally speaking, there will be a difference between the second control limits output by the initial control limit prediction model and the ideal control limits, that is, the first control limits as described above. This difference can be quantified by the loss value of the loss function in the control limit prediction model.

[0156] The loss function can calculate the loss value based on the first control limit and the second control limit. Generally, the greater the difference between the first control limit and the second control limit, the greater the loss value of the loss function; and vice versa.

[0157] In the example above where the control limit prediction model can be a neural network model, the loss value of the loss function can be fed back into the network architecture, such as a convolutional layer, a fully connected layer, or other type of network layer that may exist in the neural network model, to facilitate adjustment of network parameters in the network architecture so that the subsequently obtained second control limit is as close as possible to the corresponding first control limit. The purpose of adjusting the network parameters in the loss function can be to gradually converge the loss value of the loss function.

[0158] The loss function in the control limit prediction model can be set as needed and is not specifically limited here. The parameters of the control limit prediction model adjusted based on the loss value can be the aforementioned network parameters or related hyperparameters, etc., and are also not specifically limited here.

[0159] This embodiment adjusts the parameters of the control limit prediction model based on the loss value of the loss function in the control limit prediction model to achieve reliable training of the control limit prediction model.

[0160] According to some embodiments of the present application, adjusting parameters of a control limit prediction model according to a loss value includes:

[0161] Adjust the parameters of the control limit prediction model based on the gradient descent calculation of the loss function.

[0162] For example, the process training data may include multiple parameters that directly affect the first control limits. Alternatively, the initial control limit prediction model may extract features from the process training data to obtain high-level features that affect the first control limits. These parameters and high-level features can be considered factors that affect the output of the first control limits.

[0163] When the process training data is input into the initial control limit prediction model, the initial control limit prediction model can use the set multiple network parameters to process the above factors to obtain the second control limit.

[0164] In one example, if the loss function is defined as the difference between the first control limit and the second control limit, the purpose of training the initial control limit prediction model may be to adjust its parameters so that the loss value of the loss function is as close to 0 as possible.

[0165] Of course, in practical applications, the loss function can also be other types of loss functions. In general, the training purpose of the initial control limit prediction model can be to converge the loss value of the loss function by adjusting the parameters.

[0166] This embodiment can perform gradient descent calculation on the loss function, that is, use a gradient descent algorithm to process the loss function, and adjust parameters so that the loss value of the loss function gradually converges.

[0167] As for the gradient descent algorithm, it can be a stochastic gradient descent method or a batch gradient descent method, etc., which is not specifically limited here, and the specific implementation principles of these gradient descent algorithms belong to the existing technology and are not described here.

[0168] Generally speaking, when the loss value of the loss function converges, a trained control limit prediction model can be obtained. In some examples, when the loss value of the loss function converges to 0, the control limit prediction model can be considered to be fully trained, and the above-mentioned target control limit prediction model can be obtained.

[0169] Of course, in practical applications, affected by factors such as the type of loss function, the requirements for model training accuracy, or the quality limitations of training samples, the target control limit prediction model can also be obtained when the loss value converges to a value close to 0.

[0170] According to one embodiment of the present application, step 202 of inputting the process data of the previous control cycle into the trained control limit prediction model to obtain the control limits of the current control cycle may specifically include:

[0171] When the process data of the previous control cycle meets the preset optimization conditions, the process data of the previous control cycle is input into the trained control limit prediction model to obtain the control limit of the current control cycle.

[0172] In step 202 , the electronic device may input the process data of the previous control cycle into a trained control limit prediction model to obtain the control limits of the current control cycle if the process data of the previous control cycle meets a preset optimization condition.

[0173] Taking the process data of the previous control cycle including the mixer temperature and the temperature control limit adopted in the control cycle (hereinafter referred to as the temperature control limit) as an example, if the mixer temperature exceeds the temperature control limit of the previous control cycle at P sampling time points in the previous control cycle, and P is greater than the preset threshold, it may indicate that the temperature control limit is not set reasonably and needs to be readjusted.

[0174] For example, if the mixer temperature did not exceed the temperature control limit of the previous control cycle at P sampling time points, but was relatively close to the upper temperature control limit, then the electronic device can trigger the action of re-determining the temperature control limit using the trained control limit prediction model under user intervention.

[0175] In combination with the above examples, the preset optimization condition may be a value range or a change trend condition of some preset types of data in the process data, or the preset optimization condition may also refer to receiving a user intervention operation on the process data, etc.

[0176] Accordingly, when the numerical value of the preset type of data in the process data is within the value range specified in the preset optimization condition, or the change trend of the preset type of data meets the change trend condition, or when a user intervention operation on the process data is received, the process data can be considered to meet the preset optimization condition.

[0177] Of course, the above are some examples of preset optimization conditions. In actual applications, the preset optimization conditions can be set as needed, and no examples are given here one by one.

[0178] The electronic device can determine the control limits of the current control cycle based on the trained control limit prediction model when the process data of the previous control cycle meets the preset optimization conditions. To a certain extent, it can reduce the number of control limit adjustments in SPC and save the consumption of related computing resources.

[0179] According to one embodiment of the present application, the preset optimization condition includes at least one of the following:

[0180] Optimization conditions determined based on preset parameters in process data;

[0181] A preset operation is received, where the preset operation is used to instruct the control limit prediction model to determine the control limits of the current control cycle based on process data of the previous control cycle.

[0182] In this embodiment, the preset optimization conditions may mainly include two types of conditions: one type is a condition determined based on preset parameters in the first process data, and the other type is a condition determined by factors other than the data itself.

[0183] The former type of conditions is explained in the examples of the above embodiments.

[0184] In the case where the preset optimization conditions include optimization conditions determined for preset parameters in the process data, when the preset parameters meet the corresponding determined optimization conditions, it can be considered that the process data of the previous control cycle meets the preset optimization conditions.

[0185] In one example, the preset parameter may be a controlled parameter, such as mixer temperature. The preset optimization condition may be an optimization condition determined for the mixer temperature. The preset optimization condition may be that the absolute value of the difference between the mean of the temperature data in a control cycle and the midpoint of the control limits is greater than a preset difference threshold. Accordingly, the process data in the previous control cycle meeting the preset optimization condition may mean that the absolute value of the difference between the mean of the preset parameter in the previous control cycle and the midpoint of the control limits is greater than the preset difference threshold.

[0186] Alternatively, the preset optimization condition may be that the variance of the temperature data in a control cycle is greater than a preset variance threshold. Accordingly, the process data of the previous control cycle meeting the preset optimization condition may mean that the variance of the preset parameters of the previous control cycle is greater than a preset variance threshold.

[0187] Alternatively, there are Q consecutive temperature data that continue to increase in the temperature data in one control cycle, etc. Accordingly, the process data of the previous control cycle meets the preset optimization condition, which may mean that there are Q consecutive values ​​that continue to increase in the preset parameters of the previous control cycle.

[0188] Of course, in actual applications, the preset parameters may also be parameters other than the controlled parameters. For example, in a vinyl chloride production scenario, the preset parameters may be the obtained vinyl chloride content, etc.

[0189] The latter type of conditions can be conditions for human intervention.

[0190] In an application scenario, a user may determine based on personal experience that the control limits need to be adjusted. In this case, the user may operate a relevant button or control of the electronic device to force the electronic device to adjust the control limits. In this case, the user's operation can be considered as the above-mentioned preset operation. Upon receiving the preset operation, after a control cycle begins, regardless of the first process data obtained, the electronic device may determine that the first process data in the control cycle meets the preset optimization condition.

[0191] In other words, when the electronic device receives the preset operation, it can be considered that the process data of the previous control cycle meets the preset optimization condition.

[0192] In other application scenarios, users can also turn off the electronic device's automatic adjustment function for control limits. On this basis, after each SPC control cycle starts, no matter what first process data is obtained, the electronic device can consider that the first process data in the control cycle does not meet the preset optimization conditions.

[0193] It can be seen that the preset optimization conditions in this embodiment can be selected as needed, which helps to flexibly adjust or maintain the control limits in various scenarios and expand the applicability of the control limit determination method.

[0194] According to one embodiment of the present application, after obtaining the process data of the previous control cycle, the method further includes:

[0195] When the process data of the previous control cycle does not meet the preset optimization conditions, the control limits of the previous control cycle are determined as the control limits of the current control cycle.

[0196] In combination with the above vinyl chloride production scenario, the process data of the previous control cycle, that is, the first process data mentioned above, may include the mixer temperature obtained at P sampling time points.

[0197] In a stable SPC, The control limits may be stable within the control limits determined in the previous control cycle. In this case, there may be no need to adjust the control limits when entering the current control cycle.

[0198] Whether the temperature data is stably within the control limits determined in the previous control cycle can be determined by the distribution or change trend of the temperature data. For example, whether the absolute value of the difference between the mean of the temperature data and the median of the control limit is greater than a preset difference threshold, or whether the variance of the temperature data is greater than a preset variance threshold, or whether there are Q consecutive temperature data that continue to increase (Q is an integer greater than 1) in the temperature data. As for the preset optimization conditions, they can correspond to the above-mentioned conditions of being greater than the preset difference threshold, greater than the preset variance threshold, or the presence of Q consecutive temperature data that continue to increase.

[0199] When the first process data does not meet the preset optimization conditions, the control limits of the previous control cycle can be determined as the control limits of the current control cycle. In this way, while ensuring the rationality of the control limits of the current control cycle, the consumption of computing resources caused by the adjustment of the control limits can be saved.

[0200] Of course, the above are some examples of situations where the preset optimization conditions are not met. In actual applications, the above preset optimization conditions can be set according to actual needs. When the preset optimization conditions are determined, it is also easy to determine whether the first process data meets the preset optimization conditions, which will not be repeated here.

[0201] like Figure 3 As shown, the embodiment of the present application also provides a control limit determination method, including:

[0202] Step 301, in the process of performing statistical process control on an industrial production line, obtaining process data of the industrial production line in a previous control cycle;

[0203] Step 302: Input the process data of the previous control cycle into a pre-trained control limit prediction model to obtain the control limits of the preset production parameters in the industrial production line in the current control cycle.

[0204] In this embodiment, the industrial production line can be a production line for compounds such as vinyl chloride, or an assembly line for various industrial products, or a canning line for industrial raw materials, etc., and no specific limitation is made here.

[0205] Taking an industrial production line for vinyl chloride as an example, the control limit determination method can also be applied to electronic devices such as industrial computers. These devices acquire real-time data from sensors in the production line, such as mixer temperature and humidity, vinyl chloride content, and cooling fluid temperature.

[0206] In step 301, the electronic device can perform SPC on the industrial production line, and the electronic device can be pre-set with control cycles. Based on these pre-set control cycles, the electronic device can obtain process data from the previous control cycle, such as the mixer temperature and humidity, vinyl chloride content, and cooling fluid temperature.

[0207] In the SPC control chart, control limits can be applied. The control limits can be set for one or more preset production parameters in the industrial production line to analyze and determine whether the industrial production line is in a stable state.

[0208] In some embodiments, the preset production parameters can be set as needed. For example, the preset production parameters can be mixer temperature, etc.

[0209] In step 302 , process data from a previous control cycle is input into a pre-trained control limit prediction model to obtain control limits of preset production parameters in the industrial production line in the current control cycle.

[0210] The pre-trained control limit prediction model may correspond to the target control limit prediction model mentioned above. The specific type and the principle of pre-setting the control limit will not be described in detail here.

[0211] In this embodiment, the pre-trained control limit prediction model can predict control limits for preset production parameters in an industrial production line. These predicted control limits can be used to compare and analyze the preset production parameters during the current control cycle, thereby determining whether the production process of the industrial production line is stable. The use of control limits on control charts in SPC to analyze and determine process stability has been described in detail above and will not be repeated here.

[0212] The control limit determination method provided in an embodiment of the present application obtains process data of the industrial production line in the previous control cycle during statistical process control of the industrial production line; inputs the process data of the previous control cycle into a pre-trained control limit prediction model to obtain control limits of preset production parameters in the industrial production line in the current control cycle. In this embodiment, the control limits of preset production parameters can be automatically determined in the industrial production line based on the trained control limit prediction model, thereby effectively improving the flexibility and reliability of determining the control limits of the preset production parameters, thereby facilitating the effective monitoring of the stability of the production process of the industrial production line.

[0213] According to some embodiments of the present application, step 302, inputting the process data of the previous control cycle into a pre-trained control limit prediction model to obtain the control limits of the preset production parameters of the industrial production line in the current control cycle, may specifically include:

[0214] When the process data of the previous control cycle meets the preset optimization conditions, the process data of the previous control cycle is input into the pre-trained control limit prediction model to obtain the control limits of the preset production parameters in the industrial production line in the current control cycle.

[0215] In some embodiments, the preset optimization condition includes at least one of the following:

[0216] Optimized conditions determined for preset production parameters;

[0217] A preset operation is received, where the preset operation is used to instruct the control limit prediction model to determine control limits of preset production parameters in a current control cycle based on process data from a previous control cycle.

[0218] With some examples, when judging whether the process data of the previous control cycle meets the preset optimization conditions, the electronic device can perform statistical analysis on the process data of the previous control cycle, for example, calculate the mean or variance of the numerical values ​​of the preset production parameters in the process data, and determine whether the preset optimization conditions are met based on the data obtained from the statistical analysis.

[0219] In some embodiments, when process data of a previous control cycle does not satisfy a preset optimization condition, the control limits of the preset production parameters in the previous control cycle are determined as the control limits of the preset production parameters in the current control cycle.

[0220] like Figure 4 As shown, Figure 4 This is a flowchart of a control limit determination method in a specific application example. This specific application example may be in the production scenario of vinyl chloride. Figure 4 As shown, the method may include steps 401 to 406.

[0221] Step 401: Obtain the operation data of the previous cycle.

[0222] This step corresponds to the step of obtaining the process data of the previous control cycle. In the vinyl chloride production scenario, the process data can be the operating data of the industrial production line.

[0223] Step 402, statistical analysis;

[0224] This step performs statistical analysis on the operating data of the previous cycle.

[0225] For example, in the previous cycle, parameters such as mixer temperature or vinyl chloride content at multiple time points can be obtained. Through statistical analysis, the mean or variance of these parameters can be obtained, or operational statistical data such as time series determination of change trends and distribution patterns of these parameters can be obtained.

[0226] Step 403, determine whether the optimization conditions are met; if so, execute step 404, if not, execute step 406.

[0227] The optimization conditions and the method of determining whether the relevant parameters meet the optimization conditions have been described in detail above and will not be repeated here.

[0228] Step 404: input the previous cycle operation data into the trained control limit prediction model.

[0229] Step 405: Obtain new control limit output based on the trained control limit prediction model.

[0230] In some examples, the new control limits may include an upper control limit and a lower control limit. Based on the above formula (6), the upper control limit and the lower control limit can be determined as follows:

[0231]

[0232]

[0233] Among them, the subscript up corresponds to the upper control limit, and the subscript low corresponds to the lower control limit. As for other parameters and the meanings of superscripts and subscripts, they are all explained above and will not be repeated here.

[0234] In practical applications, the network architectures of the control limit prediction model for determining the upper control limit and the control limit prediction model for determining the lower control limit may be similar, and the difference may mainly lie in the network parameters or hyperparameters of the two.

[0235] Step 406: Output the original control limits.

[0236] In this step, the control limits used in the previous cycle can be used as the control limits of this cycle.

[0237] The control limit determination method provided in the embodiments of this application uses data intelligence technology combined with statistical analysis methods to set reasonable upper and lower limits for control charts. On the one hand, it absorbs the advantages of existing statistical methods and uses their results as input to the prediction model to better integrate the characteristics of data operation. On the other hand, by adopting data intelligence technology and utilizing deep learning models, it can circumvent the shortcomings of traditional methods such as the inability to incorporate service-related factors and the need for expert assistance, thereby enabling the control limits of the control chart to be set more closely in line with service reality.

[0238] like Figure 5 As shown, the embodiment of the present application further provides a control limit determination device 500, which includes:

[0239] The first acquisition module 501 is used to acquire process data of the previous control cycle in the current control cycle of the statistical process control (SPC);

[0240] The first determination module 502 is configured to input the process data of the previous control cycle into the trained control limit prediction model to obtain the control limits of the current control cycle.

[0241] In one embodiment of the present application, the process data of the previous control cycle includes the control limits of the previous control cycle;

[0242] The first determining module 502 includes:

[0243] The first determination unit is used to extract a first feature from the process data of the previous control cycle based on the trained control limit prediction model, determine the control limit adjustment rate according to the first feature, and determine the control limit of the current control cycle based on the control limit adjustment rate and the control limit of the previous control cycle; wherein the first feature is a feature that affects the adjustment rate of the control limit.

[0244] In one embodiment of the present application, the first determining unit includes:

[0245] a first determining subunit, configured to determine a control limit adjustment step size according to the first feature;

[0246] The second determining subunit is configured to determine the control limit adjustment rate according to the control limit adjustment step and a preset adjustment rate limit, wherein the preset adjustment rate limit is used to limit a numerical range of the control limit adjustment rate.

[0247] In one embodiment of the present application, the first determining module 502 includes:

[0248] The second determination unit is used to extract a second feature from the process data of the previous control cycle based on the trained control limit prediction model, determine the control limit adjustment amount according to the second feature, and determine the control limit of the current control cycle based on the control limit adjustment amount and the control limit of the previous control cycle; wherein the second feature is a feature that affects the adjustment amount of the control limit.

[0249] In one embodiment of the present application, the process data of the previous control cycle includes data of preset hyperparameters;

[0250] The second determining unit includes:

[0251] The third determining subunit is used to determine the control limit adjustment amount according to the second feature and the data of the preset hyperparameter.

[0252] In one embodiment of the present application, the control limit determination device 500 may further include:

[0253] A second acquisition module is used to acquire multiple training samples, where the training samples include process training data and first control limits corresponding to the process training data;

[0254] The training module is used to train the control limit prediction model using multiple training samples, so that the trained control limit prediction model can predict the control limit according to the input process data.

[0255] In one embodiment of the present application, the training module includes:

[0256] An input-output unit, used for inputting process training data into the control limit prediction model and outputting a second control limit;

[0257] An adjustment unit is used to determine a loss value of a loss function in a control limit prediction model according to the first control limit and the second control limit, and to adjust parameters of the control limit prediction model according to the loss value.

[0258] In one embodiment of the present application, the adjustment unit may specifically include:

[0259] Adjust the parameters of the control limit prediction model based on the gradient descent calculation of the loss function.

[0260] In one embodiment of the present application, the first determining module 502 may be specifically configured to:

[0261] When the process data of the previous control cycle meets the preset optimization conditions, the process data of the previous control cycle is input into the trained control limit prediction model to obtain the control limit of the current control cycle;

[0262] The preset optimization conditions include at least one of the following:

[0263] Optimization conditions determined based on preset parameters in process data;

[0264] A preset operation is received, where the preset operation is used to instruct the control limit prediction model to determine the control limits of the current control cycle based on process data of the previous control cycle.

[0265] In one embodiment of the present application, the control limit determination device 500 may further include:

[0266] The second determining module is configured to determine the control limits of the previous control cycle as the control limits of the current control cycle when the process data of the previous control cycle does not meet the preset optimization condition.

[0267] like Figure 6 As shown, the embodiment of the present application further provides a control limit determination device 600, comprising:

[0268] The third acquisition module 601 is used to acquire process data of the industrial production line in the previous control cycle during the statistical process control of the industrial production line;

[0269] The third determination module 602 is configured to input the process data of the previous control cycle into a pre-trained control limit prediction model to obtain the control limits of the preset production parameters in the industrial production line in the current control cycle.

[0270] It should be noted that the control limit determination device is a device corresponding to the above-mentioned control limit determination method. All implementation methods in the above-mentioned method embodiments are applicable to the embodiments of the device and can achieve the same technical effects.

[0271] Figure 7 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application is shown.

[0272] The electronic device may include a processor 701 and a memory 702 storing computer program instructions.

[0273] Specifically, the processor 701 may include a processor (CPU), or an application-specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0274] The memory 702 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 702 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 702 may include removable or non-removable (or fixed) media. Where appropriate, the memory 702 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 702 is a non-volatile solid-state memory.

[0275] The memory may include read-only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical or other physical / tangible memory storage devices. Thus, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to the present disclosure.

[0276] The processor 701 reads and executes computer program instructions stored in the memory 702 to implement any one of the control limit determination methods in the above embodiments.

[0277] In one example, the electronic device may further include a communication interface 703 and a bus 710. Figure 7 As shown, the processor 701, the memory 702, and the communication interface 703 are connected via a bus 710 and communicate with each other.

[0278] The communication interface 703 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0279] Bus 710 comprises hardware, software or both.For example, and not limitation, bus can comprise accelerated graphics port (AGP) or other graphics bus, enhanced industry standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industry standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations.In suitable cases, bus 710 can comprise one or more buses.Although the present application embodiment describes and shows specific bus, the application considers any suitable bus or interconnection.

[0280] In addition, in conjunction with the control limit determination method in the above embodiments, embodiments of the present application may provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any of the control limit determination methods in the above embodiments is implemented.

[0281] It should be understood that the present application is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, a detailed description of known methods is omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order of the steps after understanding the spirit of the present application.

[0282] The functional blocks shown in the above block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. Programs or code segments can be stored in machine-readable media, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable media" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROMs, flash memories, erasable ROMs (EROMs), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0283] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps. In other words, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0284] Aspects of the present disclosure have been described above with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It is also understood that each box in the block diagram and / or flowchart and the combination of the boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0285] The above is only a specific implementation method of the present application. Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the scope of protection of the present application is not limited to this. Any technician familiar with this technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the scope of protection of this application.

Claims

1. A method for determining control limits, comprising: In the current control cycle of statistical process control, process data of the previous control cycle is obtained; The process data of the last control cycle includes the control limits of the last control cycle; Inputting the process data of the previous control cycle into the trained control limit prediction model to obtain the control limits of the current control cycle; Inputting the process data of the previous control cycle into the trained control limit prediction model to obtain the control limits of the current control cycle includes: Based on the trained control limit prediction model, a first feature is extracted from the process data of the previous control cycle, a control limit adjustment rate is determined according to the first feature, and the control limit of the current control cycle is determined based on the control limit adjustment rate and the control limit of the previous control cycle; wherein the first feature is a feature that affects the adjustment rate of the control limit.

2. The method according to claim 1, wherein The determining of the control limit adjustment rate according to the first characteristic includes: determining a control limit adjustment step size according to the first characteristic; The control limit adjustment rate is determined according to the control limit adjustment step and a preset adjustment rate limit, and the preset adjustment rate limit is used to limit the numerical range of the control limit adjustment rate.

3. The method according to claim 1, wherein Inputting the process data of the previous control cycle into the trained control limit prediction model to obtain the control limits of the current control cycle includes: Based on the trained control limit prediction model, a second feature is extracted from the process data of the previous control cycle, a control limit adjustment amount is determined according to the second feature, and the control limit of the current control cycle is determined based on the control limit adjustment amount and the control limit of the previous control cycle; wherein the second feature is a feature that affects the adjustment amount of the control limit.

4. The method according to claim 3, wherein: The process data of the previous control cycle includes data of preset hyperparameters; The determining of the control limit adjustment amount according to the second characteristic includes: The control limit adjustment amount is determined based on the second feature and the data of the preset hyperparameter.

5. The method according to claim 1, wherein Before obtaining the process data of the previous control cycle, the method further includes: Acquire a plurality of training samples, wherein the training samples include process training data and first control limits corresponding to the process training data; The control limit prediction model is trained using the multiple training samples, so that the trained control limit prediction model predicts the control limit according to the input process data.

6. The method according to claim 1, wherein Inputting the process data of the previous control cycle into the trained control limit prediction model to obtain the control limits of the current control cycle includes: In a case where the process data of the previous control cycle meets a preset optimization condition, inputting the process data of the previous control cycle into a trained control limit prediction model to obtain the control limit of the current control cycle; The preset optimization condition includes at least one of the following: Optimization conditions determined for preset parameters in the process data; A preset operation is received, where the preset operation is used to instruct the control limit prediction model to determine the control limit of the current control cycle based on the process data of the previous control cycle.

7. The method according to claim 6, wherein: After obtaining the process data of the previous control cycle, the method further includes: When the process data of the previous control cycle does not satisfy the preset optimization condition, the control limit of the previous control cycle is determined as the control limit of the current control cycle.

8. A method for determining control limits, comprising: In a process of performing statistical process control on an industrial production line, obtaining process data of the industrial production line in a previous control cycle; The process data of the last control cycle includes the control limits of the last control cycle; Inputting the process data of the previous control cycle into a pre-trained control limit prediction model to obtain control limits of preset production parameters in the industrial production line in the current control cycle; Inputting the process data of the previous control cycle into a pre-trained control limit prediction model to obtain control limits of preset production parameters in the industrial production line in the current control cycle, including: Based on the trained control limit prediction model, a first feature is extracted from the process data of the previous control cycle, a control limit adjustment rate is determined according to the first feature, and the control limit of the current control cycle is determined based on the control limit adjustment rate and the control limit of the previous control cycle; wherein the first feature is a feature that affects the adjustment rate of the control limit.

9. A control limit determination device, comprising: A first acquisition module is used to acquire process data of a previous control cycle in a current control cycle of the statistical process control; The process data of the last control cycle includes the control limits of the last control cycle; a first determining module, configured to input the process data of the previous control cycle into a trained control limit prediction model to obtain the control limits of the current control cycle; The first determination module is specifically configured to extract a first feature from the process data of the previous control cycle based on the trained control limit prediction model, determine a control limit adjustment rate based on the first feature, and determine the control limit of the current control cycle based on the control limit adjustment rate and the control limit of the previous control cycle; wherein the first feature is a feature that affects the adjustment rate of the control limit.

10. A control limit determination device, comprising: A third acquisition module is used to acquire process data of the industrial production line in a previous control cycle during the statistical process control of the industrial production line; The process data of the last control cycle includes the control limits of the last control cycle; a third determination module, configured to input the process data of the previous control cycle into a pre-trained control limit prediction model to obtain control limits of preset production parameters in the industrial production line in the current control cycle; The third determination module is specifically used to extract a first feature from the process data of the previous control cycle based on the trained control limit prediction model, determine the control limit adjustment rate according to the first feature, and determine the control limit of the current control cycle based on the control limit adjustment rate and the control limit of the previous control cycle; wherein the first feature is a feature that affects the adjustment rate of the control limit.

11. An electronic device, wherein: The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the control limit determination method according to any one of claims 1 to 7, or implements the control limit determination method according to claim 8.

12. A computer storage medium, wherein: The computer storage medium stores computer program instructions, and when the computer program instructions are executed by the processor, the control limit determination method according to any one of claims 1 to 7 is implemented, or the control limit determination method according to claim 8 is implemented.

Images