Method and device for generating PID (Proportion Integration Differentiation) of secondary piping of semiconductor

Abstract

The invention discloses a semiconductor secondary piping PID generation method. The method comprises the following steps: acquiring initial process data for a semiconductor secondary piping operation process; based on the distribution state of each parameter in the initial process data, matching a corresponding normalization strategy to obtain standard process data; obtaining the information amount of each parameter in the standard process data according to the analysis of the standard process data, and giving the feature weight of each parameter in the standard process data in combination with the initial weight of each parameter; performing clustering analysis on the standard process data in combination with the feature weight of each parameter in the standard process data to obtain process component matching data; and in combination with the process component matching data, analyzing and adjusting each component, and generating the semiconductor secondary piping PID. According to the semiconductor secondary piping PID generation scheme provided by the invention, the flexibility and accuracy of semiconductor secondary piping PID generation are improved, so that the drawing requirement of a complex industrial process is met.

Description

Technical Field

[0001] The present invention belongs to the technical field of semiconductor secondary piping, and in particular relates to a method and device for generating a semiconductor secondary piping PID. Background Art

[0002] In the field of semiconductor manufacturing, secondary piping engineering is a crucial link. It is the process of distributing various utility systems from the main system to the use points of various process equipment, involving the configuration and installation of multiple components such as equipment, pipelines, valves, instruments, control systems, etc. This process not only needs to ensure the stable, safe and efficient transportation of resources, but also has a direct impact on the stability of the production line, product quality and production costs.

[0003] PID (piping and instrumentation diagram) is a process piping and instrumentation flow chart, which is a drawing showing all the equipment, instruments, pipelines, valves and other related public engineering systems of the system. In the field of industrial automation and informatization, PID is an important tool for describing industrial processes and control systems. Traditional PID production relies on manual drawing, which is inefficient and prone to errors. The current automated drawing tools lack flexibility and accuracy in drawing PID and cannot meet the drawing needs of complex industrial processes.

[0004] At present, the research on semiconductor secondary piping PID is limited to the three-dimensional display of PID. For example, patent CN119442532A discloses a layout device and method for semiconductor secondary piping installation engineering, including: a file upload module, receiving the layout file of the semiconductor production line; a data parsing module, parsing the layout file, obtaining the attribute information and location information of the equipment, and transmitting it to the secondary piping database; a data analysis layout module, giving the secondary piping installation pipeline diagram, and then combining the component matching algorithm to obtain the corresponding pipeline instrument components in the secondary piping database to form a pipeline instrument flow chart; a three-dimensional space management module, combined with a three-dimensional model generation algorithm, generates a layout model of the semiconductor secondary piping installation engineering. Through the linkage of semiconductor layout information and the secondary piping database, combined with the component matching algorithm, the information of the secondary piping components is obtained, and according to the three-dimensional model generation algorithm, the automatic output of the semiconductor secondary piping layout is realized and visualized in a three-dimensional scene.

[0005] How to improve the flexibility and accuracy of semiconductor secondary piping PID generation to meet the drawing needs of complex industrial processes is a problem that needs to be solved at present. Summary of the invention

[0006] In view of the defects existing in the above-mentioned prior art, the present invention provides a method for generating semiconductor secondary piping PID, including: obtaining initial process data for semiconductor secondary piping operation process; matching corresponding normalization strategies based on the distribution status of each parameter in the initial process data to obtain standard process data; obtaining the amount of information of each parameter in the standard process data according to the analysis of the standard process data, and giving the characteristic weight of each parameter in the standard process data in combination with the initial weight of each parameter; clustering the standard process data in combination with the characteristic weight of each parameter in the standard process data to obtain process component matching data; analyzing and adjusting each component in combination with the process component matching data to generate semiconductor secondary piping PID. The semiconductor secondary piping PID generation scheme provided by the present invention improves the flexibility and accuracy of semiconductor secondary piping PID generation to meet the drawing requirements of complex industrial processes.

[0007] In a first aspect, the present invention provides a method for generating a semiconductor secondary piping PID, comprising: Acquire initial process data for semiconductor secondary piping work processes; Based on the distribution state of each parameter in the initial process data, the corresponding normalization strategy is matched to obtain the standard process data; According to the analysis of the standard process data, the information amount of each parameter in the standard process data is obtained, and the characteristic weight of each parameter in the standard process data is given in combination with the initial weight of each parameter; Combined with the characteristic weights of each parameter in the standard process data, cluster analysis is performed on the standard process data to obtain process component matching data; Combined with process component matching data, each component is analyzed and adjusted to generate semiconductor secondary piping PID.

[0008] Furthermore, obtaining the initial process data for the semiconductor secondary piping operation process specifically includes the following steps: Obtaining raw process data for semiconductor secondary piping operations; Convert the format of the original process data to obtain structured process data; Analyze the data status of structured process data, clean the structured process data, and obtain initial process data.

[0009] Furthermore, based on the distribution state of each parameter in the initial process data, the corresponding normalization strategy is matched to obtain the standard process data, which specifically includes: Filter multiple groups of core parameters in the initial process data, and construct corresponding data sets to be partitioned based on the values ​​of each group of core parameters; Determine the basic parameters in each set of core parameters, and sort each data set to be partitioned according to the basic parameters; According to the partition granularity, each data set to be partitioned is partitioned, and the values ​​of the core parameters of each partition are screened to form intermediate process data. The partition granularity is obtained by analyzing the basic parameters, operation process coefficients and other core parameters. According to the value of each parameter in the intermediate process data, the first parameter and the second parameter are divided, wherein the value of the first parameter is a finite number or a finite range, and the value of the second parameter is an unfixed range; Based on the value analysis of the first parameter, a corresponding normalization strategy is matched to obtain corresponding standard process data; Perform a normality test on the second parameter and obtain the test result; According to the test result, a corresponding normalization strategy is matched for the second parameter to obtain corresponding standard process data.

[0010] Further, according to the test result, a corresponding normalization strategy is matched for the second parameter to obtain corresponding standard process data, specifically including: If the test result is normally distributed, the second parameter is screened in combination with the preset normal distribution range to obtain the standard second parameter, and the standard second parameter is normalized using the Z-score normalization strategy to obtain the corresponding standard process data; If the test result is non-normal distribution, the second parameter is screened in combination with the preset non-normal distribution range to obtain the standard second parameter, the corresponding normalization strategy is matched to the standard second parameter and normalization is performed to obtain the corresponding standard process data.

[0011] Furthermore, based on the analysis of the standard process data, the information content of each parameter in the standard process data is obtained, and the feature weights of each parameter in the standard process data are given in combination with the initial weight of each parameter, including: Based on the preset process standards, sort the parameters in the standard process data and give the initial weight of each parameter; Combined with the information content of each parameter in the standard process data, the initial weight is corrected to obtain the feature weight.

[0012] Furthermore, by combining the feature weights of each parameter in the standard process data, cluster analysis is performed on the standard process data to obtain process component matching data, including: Based on preset physical constraints, the initial cluster centers are determined from the standard process data; According to the initial cluster center, combined with the characteristic weights of each parameter in the standard process data, the standard process data is clustered and analyzed, and the cluster center is updated and iterated until convergence to obtain the clustering result; Judge the clustering results and obtain constraint verification results; According to the constraint verification results, the feature weights of each parameter in the clustering results and standard process data are adjusted to provide process component matching data.

[0013] Furthermore, based on the initial cluster center and the characteristic weights of each parameter in the standard process data, the standard process data is clustered and analyzed, and the cluster center is updated and iterated until convergence to obtain the clustering results, which specifically include: According to the initial cluster center, combined with each parameter in the standard process data and the characteristic weight of each parameter, the distance between the initial cluster center and each standard process data is given; Analyze the distance between the initial cluster center and each standard process data, update the cluster center iteratively, and give the distance between the new cluster center and each standard process data, where the new cluster center meets the physical constraints; Based on the distance between the new cluster center and each standard process data, the mean distance between the standard process data in each cluster and the cluster center is analyzed to give the clustering result.

[0014] Furthermore, combined with the process component matching data, each component is analyzed and adjusted to generate the semiconductor secondary piping PID, including: Based on the process component matching data, draw the components corresponding to each standard process data to generate the initial PID; Construct a reward function based on the component relationships of each component in the initial PID; Combined with the reward function, the component relationship of each component in the initial PID is adjusted to give a standard PID; Render the standard PID to generate the semiconductor secondary piping PID.

[0015] Furthermore, according to the component relationship of each component in the initial PID, a reward function is constructed, including: According to the connection relationship between each component in the initial PID, the gap between the corresponding connection indicators of each component is analyzed, and the difference ranking of the connection indicators is given; Based on the difference ranking of connection indicators, corresponding weights are assigned to each connection indicator, and a reward weight matrix is ​​given; Combined with the reward weight matrix, a reward function is constructed.

[0016] In a second aspect, the present invention further provides a semiconductor secondary piping PID generation device, which adopts any of the semiconductor secondary piping PID generation methods described above, comprising: A data acquisition module, used to acquire initial process data for a semiconductor secondary piping operation process; The data processing module is used to match the corresponding normalization strategy based on the distribution state of each parameter in the initial process data to obtain the standard process data; The weight determination module is used to obtain the information amount of each parameter in the standard process data based on the analysis of the standard process data, and to give the characteristic weight of each parameter in the standard process data in combination with the initial weight of each parameter; The component matching module is used to combine the feature weights of each parameter in the standard process data, perform cluster analysis on the standard process data, and obtain process component matching data; The PID generation module is used to combine the process component matching data, analyze and adjust each component, and generate the semiconductor secondary piping PID.

[0017] The present invention provides a method and device for generating a semiconductor secondary piping PID, which at least has the following beneficial effects: (1) By acquiring and analyzing the distribution status of each parameter in the initial process data, matching the corresponding normalization strategy, and obtaining the standard process data, the information content of each parameter in the standard process data is combined to assign feature weights to each parameter in the standard process data. Based on the feature weights, the standard process data is clustered and analyzed to obtain process component matching data. Combined with the process component matching data, each component is analyzed and adjusted to generate the semiconductor secondary piping PID, which improves the flexibility and accuracy of semiconductor secondary piping PID generation and meets the drawing requirements of complex industrial processes.

[0018] (2) By assigning feature weights to each parameter in the standard process data and performing cluster analysis on the standard process data based on the feature weights, the process component matching data obtained by clustering is made closer to the project requirements of semiconductor secondary piping, thereby improving the accuracy of semiconductor secondary piping PID. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flow chart of a method for generating a semiconductor secondary piping PID according to an embodiment of the present invention; Figure 2 A flowchart for determining initial process data provided by an embodiment of the present invention; Figure 3 A flowchart for obtaining standard process data provided by an embodiment of the present invention; Figure 4 A flowchart for obtaining process component matching data provided by an embodiment of the present invention; Figure 5 A flow chart of obtaining clustering results provided by an embodiment of the present invention; Figure 6 A flowchart of generating a semiconductor secondary piping PID according to an embodiment of the present invention; Figure 7 A flowchart of constructing a reward function provided by an embodiment of the present invention; Figure 8An example diagram of components connected by reward function adjustment provided by an embodiment of the present invention; Fig. 9 A structural block diagram of a device for generating a semiconductor secondary piping PID provided in an embodiment of the present invention.

[0020] Among them, 201 is a data acquisition module; 202 is a data processing module; 203 is a weight determination module; 204 is a component matching module; 205 is a PID generation module. DETAILED DESCRIPTION

[0021] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. 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.

[0022] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. The singular forms "a", "said" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings, and "multiple" generally includes at least two.

[0023] It should also be noted that the term "includes", "comprising" or any other variation thereof is intended to cover non-exclusive inclusion, so that a commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprising a ..." do not exclude the existence of other identical elements in the commodity or device including the elements.

[0024] PID stands for Process Piping and Instrumentation Diagram, which refers to a drawing that shows in detail all the equipment, instruments, pipelines, valves and other related utility systems of the system according to the requirements of the process flow diagram (PFD). The focus of PID is the flow of the pipeline and how the process is controlled. It can show how the pipeline system connects industrial processing equipment together. PID will also show the instruments and valves used to monitor the flow of materials in the pipeline.

[0025] Secondary piping is a re-matching or upgrading project based on the original equipment, system or building, aiming to improve performance or expand functions, involving multiple fields such as electrical, communication, weak current, etc. The core of secondary piping is to meet new needs or solve existing deficiencies by optimizing existing infrastructure.

[0026] In the semiconductor field, secondary piping projects include the configuration and installation of multiple components such as piping systems, valves and instruments, and control systems. These components need to be reasonably configured according to the properties of the conveying medium and the process requirements to ensure the stable operation of the process equipment. For example, the piping system needs to use corrosion-resistant, low-pollution and high-cleanliness materials, while the control system needs to have high reliability, easy operation and scalability. The design and installation requirements of secondary piping projects are strict. Safety, cleanliness and flexibility are the three core requirements in the design and installation process. Safety requires that the system design and installation follow relevant safety specifications and standards, cleanliness requires that the system maintain a high degree of cleanliness to avoid pollution of the production environment, and flexibility requires that the system be able to adapt to possible changes and expansion needs in the future.

[0027] With the rapid development of semiconductor technology, the technical requirements of secondary piping engineering are also constantly increasing. Traditional drawing methods rely on manual work, are inefficient, prone to errors, and cannot respond quickly to design changes. With the development of technology, some automated drawing tools have emerged, but these tools often lack flexibility and are difficult to adapt to changing industrial needs. In addition, the current level of intelligence of automated drawing tools is insufficient, and they cannot effectively handle complex process flows and equipment interfaces, resulting in unreasonable drawing layouts, waste of resources, and lack of targeted optimal drawing selection and layout solutions. They cannot accurately identify and handle complex process flows, and the drawings are poorly accurate, which may lead to errors and delays in the production process.

[0028] In view of these shortcomings of the prior art, the present invention proposes a method for generating semiconductor secondary piping PID, which optimizes the drawing process, improves drawing efficiency through automation and intelligent means, reduces human errors, and quickly responds to design changes. The level of automation and intelligence is improved, so that the automatic drawing system can more effectively handle complex process flows and equipment interfaces, optimize drawing layouts, and save resources. Develop targeted drawing selection and layout plans to ensure that drawings are clear and accurate, and meet the requirements of production safety, maintenance convenience, and cost-effectiveness. Improve the recognition and processing capabilities of the automatic drawing system, reduce errors and delays in the production process, and improve overall production efficiency.

[0029] like Figure 1 As shown, an embodiment of the present invention provides a method for generating a semiconductor secondary piping PID, and the specific steps are as follows: S101: Acquire initial process data for semiconductor secondary piping operation process, wherein the initial process data is the original process data obtained after format conversion and data cleaning.

[0030] Reference Figure 2 , S101 specifically includes: Obtaining raw process data for semiconductor secondary piping operations; Convert the format of the original process data to obtain structured process data; Analyze the data status of structured process data, clean the structured process data, and obtain initial process data.

[0031] The above-mentioned original process data can be obtained by users uploading or inputting data required for the secondary piping operation process of a semiconductor factory. The original process data includes the parameters of the secondary piping components, and the parameters include node information, pipeline attribute information, and interface information, wherein the node information can be node labels, equipment dimensions, valve positions, component positions, component angles, etc., and the pipeline attribute information can be pipeline labels, diameters, paths, directions, connection points, pressure ranges, etc., and the interface information can be interface types (such as flanges, clamps), etc. Of course, the structured process data and initial process data after format conversion and data cleaning also include the parameters of the secondary piping components. The original process data can be in the form of an Excel data form, or in other forms that are convenient for users to upload or input, and there is no limitation on this.

[0032] In a specific implementation, the format of the original process data is converted to obtain structured process data, and the format conversion of the original process data facilitates the storage, query, and processing of the original process data. In this example, the input original process data is converted from an Excel data form into JSON (JavaScript Object Notation) data, and the JSON data is stored in a corresponding database as a business logic information library.

[0033] JSON data consists of key-value pairs and supports multiple data types, including strings, numbers, Boolean values, arrays, and objects. This structure makes JSON data well organized and predictable, making it easy for programs to parse and operate. In addition, the JSON format has good compatibility between different programming languages ​​and is easy to convert and process.

[0034] Structured data refers to data that has a clear data model and format and can be easily stored, queried, and processed. Common structured data types include data in relational databases, spreadsheet data, CSV files, XML files, JSON files, fixed-width files, database views, and HTML tables. In other implementations, the original process data can be converted into other types of structured data according to actual needs, and this is not limited.

[0035] Data cleaning of structured process data includes correcting and marking obviously erroneous values, filling missing values, and deleting duplicate values. In the embodiment provided by the present invention, the structured process data is traversed to identify obviously erroneous values, missing values, and duplicate data in the data. For example, for the component specification values ​​in the structured process data, if there are abnormal values ​​that are obviously not in the normal range, corrections or markings will be made, for example, the distribution of other values ​​in the component specifications is analyzed, and appropriate values ​​are selected for correction based on the distribution. If there are missing values, appropriate methods are used to fill them according to the overall characteristics of the component data and the secondary piping business logic, such as mean filling, interpolation, etc. Duplicate data will be identified and deleted to ensure the accuracy and consistency of the data.

[0036] S102: Based on the distribution state of each parameter in the initial process data, a corresponding normalization strategy is matched to obtain standard process data.

[0037] Specifically, refer to Figure 3 , get standard process data, including: Filter multiple groups of core parameters in the initial process data, and construct corresponding data sets to be partitioned based on the values ​​of each group of core parameters; Determine the basic parameters in each set of core parameters, and sort each data set to be partitioned according to the basic parameters; According to the partition granularity, each data set to be partitioned is partitioned, and the values ​​of the core parameters of each partition are screened to form intermediate process data. The partition granularity is obtained by analyzing the basic parameters, operation process coefficients and other core parameters. According to the value of each parameter in the intermediate process data, the first parameter and the second parameter are divided, wherein the value of the first parameter is a finite number or a finite range, and the value of the second parameter is an unfixed range; Based on the value analysis of the first parameter, a corresponding normalization strategy is matched to obtain corresponding standard process data; Perform a normality test on the second parameter and obtain the test result; According to the test result, a corresponding normalization strategy is matched for the second parameter to obtain corresponding standard process data.

[0038] For the process of converting initial process data into standard process data, we first determine the core parameters for screening, and then conduct normality tests and normalization strategy matching. This can not only improve data processing efficiency, but also greatly improve the accuracy of the acquired standard process data.

[0039] The core parameters are screened from the parameters of the secondary piping components. The core parameters of different scenarios and different secondary piping components may be different. Taking a pipeline used to accurately control the flow of gas and liquid as an example in a certain embodiment, the core parameters may be the pipe size, maximum pressure, interface type, and installation location, etc. The values ​​corresponding to the above core parameters are collected to construct a data set to be partitioned, that is, the data set to be partitioned includes a set of each data consisting of core parameter values. For example, as shown in Table 1, in the data set to be partitioned, there is a piece of data with a pipe size of 2 inches and a maximum pressure of 6kg / cm 2 , the interface type is PCWS-FL (process cooling return water supply pipe), the installation position is B6, etc.

[0040] Table 1 Dataset A to be partitioned

[0041] For the constructed data set to be partitioned, a core parameter is used as the basic parameter, and each piece of data of the data set to be partitioned is sorted according to the benchmark parameter, as shown in Table 2.

[0042] Table 2 Dataset B to be partitioned

[0043] The partition granularity is obtained by analyzing the basic parameters, the operation process coefficient and other core parameters. The operation process coefficient indicates the importance of this group of core parameters, which can be expressed by the number of historical failures of the secondary piping components under this group of core parameters.

[0044] Partition granularity, specifically expressed as:

[0045] Among them, G is the partition granularity, P0 is the basic parameter, N is the operation process coefficient, is the weight coefficient of the first core parameter, is the first core parameter, is the weight coefficient of the nth core parameter, is the nth core parameter. The basic parameters and the n core parameters constitute all the core parameters of the data set to be partitioned.

[0046] Each data set to be partitioned is partitioned according to the partition granularity, the values ​​of the core parameters of each partition are screened, and unreasonable data in the initial process data is eliminated to form intermediate process data.

[0047] It is understandable that since the data of different parameters (i.e., nodes, pipeline attributes and / or interfaces) in the intermediate process data may have different dimensions and value ranges, it will affect the accuracy of the analysis of the intermediate process data and subsequent drawing. Therefore, it is necessary to standardize the intermediate process data, but considering that the value ranges and distribution of various parameters in the intermediate process data are quite different, if the same normalization strategy is used, the effect of data standardization may be reduced, thereby affecting the accuracy of the drawing.

[0048] In a specific implementation, according to the values ​​of different parameters in the intermediate process data, it is divided into a first parameter and a second parameter. The value of the first parameter is a finite number or a finite range, and the value of the second parameter is no fixed range. For example, the pipe diameter in the pipe attribute belongs to the second parameter, because the value range of the pipe diameter is not fixed, and the size of the pipe diameter is related to the actual required scene. In different requirements, the pipe diameter may be large or small. For another example, the interface type in the interface is the first parameter, because the value of the interface type is limited, and it can be a flange, a clamp, and the like. If the value of the first parameter is limited, the first parameter is normalized using a unique hot encoding to obtain the corresponding standard process data. For example, the code of the flange in the interface type is defined as 001, and the code of the clamp is defined as 100. If the value of the first parameter is a finite range, the first parameter is processed using the minimum-maximum normalization method to obtain the corresponding standard process data. All parameters except the first parameter are second parameters. It is necessary to perform a normality test on the second parameter. When the test result is consistent with the normal distribution, the Z-score normalization method is used to process the second parameter to obtain the corresponding standard process data. When the test result is inconsistent with the normal distribution, the corresponding normalization strategy is matched according to the distribution difference with the normal distribution to obtain the corresponding standard process data. In a specific example, if the test result is a right-skewed distribution, the second parameter is normalized using Log transformation. In other examples, the second parameter can also be processed using Box-Cox transformation, quantile normalization, Robust normalization, etc., without limitation.

[0049] In a specific example, a multidimensional data matrix corresponding to different parameters in the intermediate process data is input, including pipeline attributes (such as pipeline diameter, pressure), interfaces (flange / clamp), etc. Then, normality tests are performed on different parameters in the intermediate process data. The Shapiro-Wilk test is used to determine whether the feature conforms to the normal distribution, which is used to screen features within the normal value range. First, determine whether each parameter in the intermediate process data is within the corresponding fixed range, for example, the fixed range corresponding to the pressure data is 0-10MPa, and the fixed range corresponding to the pipeline diameter is 0.5mm-10mm. The data that is not within the corresponding fixed range is eliminated, and the normality test is performed on each parameter after the elimination is completed. For example, the p value of the pressure data is calculated, and the p value = 0.03, indicating that the pressure data is non-normally distributed and belongs to the right-skewed distribution. The p value of the pipeline diameter is calculated, and the p value = 0.45, indicating that the pipeline diameter is approximately normal. For the second parameter, although there is no fixed range, based on different process data, a corresponding fixed range will be preset for judgment. For the first parameter, a corresponding limited range is used for range analysis, for example, to determine whether the valve opening is within the range of 0%-100%. According to the test results, a corresponding normalization strategy is matched and standard process data is obtained.

[0050] Furthermore, the corresponding standard process data is obtained, including: If the test result is a normal distribution, the second parameter is screened in combination with a preset normal distribution range to obtain a standard second parameter; The Z-score normalization strategy is used to normalize the standard second parameter to obtain the corresponding standard process data.

[0051] If the test result is a non-normal distribution, the second parameter is screened in combination with a preset non-normal distribution range to obtain a standard second parameter; A corresponding normalization strategy is matched to the standard second parameter and normalization processing is performed to obtain corresponding standard process data.

[0052] In a specific implementation, after the normality test is performed on the second parameter, a corresponding test result will be obtained. The test result may be a normal distribution or a non-normal distribution, for example, a left-skewed distribution or a right-skewed distribution. Regardless of the distribution state, there is an area in the data set, and the second parameter is screened according to different data requirements. In a specific example, the preset normal distribution range is a standard deviation range (μ±σ), and the data within a standard deviation range in the second parameter is screened out to obtain a standard second parameter. For the right-skewed distribution, the preset non-normal distribution range is the range between the median and the first quartile, and the data between the median and the first quartile in the second parameter is screened out to obtain a standard second parameter. For the left-skewed distribution, the preset non-normal distribution range is the range between the median and the third quartile, and the data between the median and the third quartile in the second parameter is screened out to obtain a standard second parameter.

[0053] In other implementations, a dynamic and adaptive clustering algorithm model can be built into the data processing layer, which can automatically detect the distribution mode (such as normal distribution, long-tail distribution) of each feature for the nodes, pipeline attributes, interfaces and other multi-dimensional parameters in the initial process data of the semiconductor secondary piping operation process, and dynamically match the corresponding normalization strategy (such as Min-Max, Log transformation, Z-score) according to the distribution type. Combined with physical constraints (such as pipeline connection rules and safety spacing), the clustering results are optimized, the feature vectors that can reflect the overall structure are extracted, and more representative and relevant high-value features are screened to improve the rationality of the layout.

[0054] S103: Based on the analysis of the standard process data, the information amount of each parameter in the standard process data is obtained, and the feature weight of each parameter in the standard process data is given in combination with the initial weight of each parameter.

[0055] Furthermore, the characteristic weights of each parameter in the standard process data are given, including: Based on the preset process standards, sort the parameters in the standard process data and give the initial weight of each parameter; Combined with the information content of each parameter in the standard process data, the initial weight is corrected to obtain the feature weight.

[0056] In a specific embodiment, the preset process standard is the installation requirement and / or configuration requirement of the secondary piping assembly. It is understandable that for different secondary piping assemblies, the importance of corresponding parameters is different. For example, for pipelines, the parameter importance is ranked in the order of interface model, pipe pressure, pipe diameter, and coordinate position. The process standard is pre-set by experienced experts. After obtaining the ranking of different parameters in the standard process data, each parameter is assigned a corresponding initial weight. For example, the initial weight corresponding to the interface model is 0.4, the initial weight corresponding to the pipe pressure is 0.3, the initial weight corresponding to the pipe diameter is 0.2, and the initial weight corresponding to the coordinate position is 0.1.

[0057] The amount of information is a measure of the degree of discreteness, which can be characterized by standard deviation, variance, etc. In a certain embodiment provided by the present invention, the variance corresponding to each parameter in the standard process data is used as the amount of information of each parameter. For example, the variance of the pipe diameter is 0.8, and the variance of the pressure in the pipe is 0.5. According to the preset correspondence between the variance and the weight, the weight corresponding to the variance of the pipe diameter is adjusted to +0.1, and the weight corresponding to the variance of the pressure in the pipe is adjusted to -0.1, and the final feature weight is obtained, that is, the feature weight corresponding to the interface model is 0.4, the feature weight corresponding to the pressure in the pipe is 0.3-0.1=0.2, the feature weight corresponding to the pipe diameter is 0.2+0.1=0.3, and the feature weight corresponding to the coordinate position is 0.1. The sum of the feature weights corresponding to each parameter in the standard process data is 1.

[0058] S104: Based on the feature weights of various parameters in the standard process data, cluster analysis is performed on the standard process data to obtain process component matching data.

[0059] Specifically, refer to Figure 4 , get the process component matching data, including: Based on preset physical constraints, the initial cluster centers are determined from the standard process data; According to the initial cluster center, combined with the characteristic weights of each parameter in the standard process data, the standard process data is clustered and analyzed, and the cluster center is updated and iterated until convergence to obtain the clustering result; Judge the clustering results and obtain constraint verification results; According to the constraint verification results, the feature weights of each parameter in the clustering results and standard process data are adjusted to provide process component matching data.

[0060] Further, refer to Figure 5 , get the clustering results, including: According to the initial cluster center, combined with each parameter in the standard process data and the characteristic weight of each parameter, the distance between the initial cluster center and each standard process data is given; Analyze the distance between the initial cluster center and each standard process data, update the cluster center iteratively, and give the distance between the new cluster center and each standard process data, where the new cluster center meets the physical constraints; Based on the distance between the new cluster center and each standard process data, the mean distance between the standard process data in each cluster and the cluster center is analyzed to give the clustering result.

[0061] In a specific implementation, the above-mentioned preset physical constraints are the connection relationships between the components in the standard process data, and the connection relationships include multiple connection indicators, each of which includes the safety distance between the components, whether there is an intersection between the components, whether the interfaces of the components are matched, etc. The criteria for determining the initial cluster center are based on the preset physical constraints. When there are multiple initial cluster centers, the initial cluster centers can be screened according to actual conditions. For example, the standard process data corresponding to the component with the largest number of connected components is selected as the initial cluster center. In other implementations, other rules can be used to screen the initial cluster center, which is not limited to this.

[0062] Combined with the feature weights obtained in the previous process, each component is clustered, which is specifically expressed as follows:

[0063] in, is the distance between the i-th initial cluster center and the j-th component, is the feature weight corresponding to the kth parameter in the standard process data, m is the number of parameters in the standard process data, is the parameter value of the kth parameter in the i-th initial cluster center, is the parameter value of the kth parameter in the standard process data corresponding to the jth component.

[0064] In a specific example, the standard process data includes four parameters, namely, interface model, pipe pressure, pipe diameter, and coordinate position, and the corresponding feature weights are 0.4, 0.2, 0.3, and 0.1. The distance between the initial cluster center and each standard process data is calculated to obtain:

[0065] is the pipe diameter corresponding to the initial cluster center, is the pressure in the tube corresponding to the initial cluster center, is the interface type corresponding to the initial cluster center, is the coordinate position corresponding to the initial cluster center, is the pipe diameter corresponding to the jth standard process data, is the pipe pressure corresponding to the jth standard process data, is the interface type corresponding to the jth standard process data, is the coordinate position corresponding to the j-th standard process data.

[0066] After completing the distance between the initial cluster center and each standard process data, the cluster center is updated and iterated, and the distance between the new cluster center and each standard process data is repeatedly calculated until the clustering result is obtained. The cluster center update process can be adjusted randomly, or it can be selected and determined based on the distance from the previous cluster center, or it can be determined by various types of optimization algorithms, which are not limited. After each time the distance between the new cluster center and each standard process data is obtained, the mean distance between the standard process data and the cluster center in each cluster is analyzed. When the mean distance corresponding to all clusters is less than the distance threshold, it means that the clustering has reached the convergence stage and the clustering result is given.

[0067] After obtaining the clustering results, it is necessary to further judge the clustering results based on the preset physical constraints to obtain the constraint verification results. According to the specific situation of the constraint verification results, the clustering results are adjusted to give the process component matching data. In a specific example, in a certain category, pipeline A and pipeline B intersect, indicating that there is a large deviation in the coordinate positions of pipeline A and pipeline B. In this case, the feature weight corresponding to the coordinate position is increased, and the coordinate position of pipeline A and / or pipeline B is adjusted.

[0068] S105: Analyze and adjust each component in combination with the process component matching data to generate a semiconductor secondary piping PID.

[0069] Further, refer to Figure 6 , generate semiconductor secondary piping PID, including: Based on the process component matching data, draw the components corresponding to each standard process data to generate the initial PID; Construct a reward function based on the connection relationship between the components in the initial PID; Combined with the reward function, each component in the initial PID is adjusted to give the standard PID; Render the standard PID to generate the semiconductor secondary piping PID.

[0070] Further, refer to Figure 7 , construct a reward function, including: According to the connection relationship between each component in the initial PID, the gap between the corresponding connection indicators of each component is analyzed, and the difference ranking of the connection indicators is given; Based on the difference ranking of connection indicators, corresponding weights are assigned to each connection indicator, and a reward weight matrix is ​​given; Combined with the reward weight matrix, a reward function is constructed.

[0071] Although the process component matching data is adjusted based on the constraint verification results, the initial PID still deviates from the physical constraints. In a specific implementation, first, the components in the standard process data are drawn and connected according to the process component matching data to obtain the initial PID. Then, the connection relationship between the components in the initial PID is analyzed to construct a reward function, for example, Figure 8 As shown in the figure, when there is no problem with the connection between component A and component B and they fully comply with the physical constraints, there is no need to build a reward function for adjustment. If there is a position deviation between component A and component B, it is necessary to obtain the difference between each connection index in component A and component B, and use different reward functions according to the difference in connection index.

[0072] In different implementations, weights corresponding to different drawing indicators are preliminarily set according to different requirements of semiconductor secondary piping.

[0073] After each round of drawing is completed and the evaluation results are obtained, the reward weight matrix is ​​adjusted according to the evaluation results. Then, the PID is adjusted using the reward function corresponding to the adjusted reward weight matrix, and the new reward function is used to calculate the reward value of each drawing step and result, continuously optimizing the drawing process.

[0074] After completing component matching and property adjustment, the various components are reasonably laid out and connected, and finally a visually rendered PID diagram, namely the semiconductor secondary piping PID, is generated, which can clearly and intuitively display the detailed structure and component relationship of the semiconductor secondary piping.

[0075] Reference Fig. 9 The embodiment of the present invention provides a device for generating a semiconductor secondary piping PID, comprising: The data acquisition module 201 is used to acquire the initial process data for the semiconductor secondary piping operation process; The data processing module 202 is used to match the corresponding normalization strategy based on the distribution state of each parameter in the initial process data to obtain standard process data; The weight determination module 203 is used to obtain the information amount of each parameter in the standard process data based on the analysis of the standard process data, and to give the characteristic weight of each parameter in the standard process data in combination with the initial weight of each parameter; The component matching module 204 is used to perform cluster analysis on the standard process data by combining the feature weights of various parameters in the standard process data to obtain process component matching data; The PID generation module 205 is used to analyze and adjust each component in combination with the process component matching data to generate the semiconductor secondary piping PID.

[0076] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0077] Although preferred embodiments of the present invention have been described, additional changes and modifications may be made to these embodiments by those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for generating a semiconductor secondary piping PID, characterized in that: include: Acquire initial process data for semiconductor secondary piping work processes; Based on the distribution state of each parameter in the initial process data, the corresponding normalization strategy is matched to obtain the standard process data; According to the analysis of the standard process data, the information content of each parameter in the standard process data is obtained, and the characteristic weight of each parameter in the standard process data is given in combination with the initial weight of each parameter; Combined with the characteristic weights of each parameter in the standard process data, cluster analysis is performed on the standard process data to obtain process component matching data; Combined with process component matching data, each component is analyzed and adjusted to generate semiconductor secondary piping PID.

2. The method for generating a semiconductor secondary piping PID according to claim 1, wherein: Obtaining initial process data for semiconductor secondary piping operation process includes the following steps: Obtaining raw process data for semiconductor secondary piping operations; Convert the format of the original process data to obtain structured process data; Analyze the data status of structured process data, clean the structured process data, and obtain initial process data.

3. The method for generating a semiconductor secondary piping PID according to claim 1, wherein: Based on the distribution state of each parameter in the initial process data, the corresponding normalization strategy is matched to obtain the standard process data, including: Filter multiple groups of core parameters in the initial process data, and construct corresponding data sets to be partitioned based on the values ​​of each group of core parameters; Determine the basic parameters in each set of core parameters, and sort each data set to be partitioned according to the basic parameters; According to the partition granularity, each data set to be partitioned is partitioned, and the values ​​of the core parameters of each partition are screened to form intermediate process data. The partition granularity is obtained by analyzing the basic parameters, operation process coefficients and other core parameters. According to the value of each parameter in the intermediate process data, the first parameter and the second parameter are divided, wherein the value of the first parameter is a finite number or a finite range, and the value of the second parameter is an unfixed range; Based on the value analysis of the first parameter, a corresponding normalization strategy is matched to obtain corresponding standard process data; Perform a normality test on the second parameter and obtain the test result; According to the test result, a corresponding normalization strategy is matched for the second parameter to obtain corresponding standard process data.

4. The method for generating a semiconductor secondary piping PID according to claim 3, wherein: According to the test results, the corresponding normalization strategy is matched for the second parameter to obtain the corresponding standard process data, including: If the test result is normally distributed, the second parameter is screened in combination with the preset normal distribution range to obtain the standard second parameter, and the standard second parameter is normalized using the Z-score normalization strategy to obtain the corresponding standard process data; If the test result is non-normal distribution, the second parameter is screened in combination with the preset non-normal distribution range to obtain the standard second parameter, the corresponding normalization strategy is matched to the standard second parameter and normalization is performed to obtain the corresponding standard process data.

5. The method for generating a semiconductor secondary piping PID according to claim 1, wherein: Based on the analysis of the standard process data, the information content of each parameter in the standard process data is obtained. Combined with the initial weight of each parameter, the characteristic weight of each parameter in the standard process data is given, including: Based on the preset process standards, sort the parameters in the standard process data and give the initial weight of each parameter; Combined with the information content of each parameter in the standard process data, the initial weight is corrected to obtain the feature weight.

6. The method for generating a semiconductor secondary piping PID according to claim 1, wherein: Combined with the feature weights of each parameter in the standard process data, cluster analysis is performed on the standard process data to obtain process component matching data, including: Based on preset physical constraints, the initial cluster centers are determined from the standard process data; According to the initial cluster center, combined with the characteristic weights of each parameter in the standard process data, the standard process data is clustered and analyzed, and the cluster center is updated and iterated until convergence to obtain the clustering result; Judge the clustering results and obtain constraint verification results; According to the constraint verification results, the feature weights of each parameter in the clustering results and standard process data are adjusted to provide process component matching data.

7. The method for generating a semiconductor secondary piping PID according to claim 6, wherein: According to the initial cluster center, combined with the feature weights of each parameter in the standard process data, the standard process data is clustered and analyzed, and the cluster center is updated and iterated until convergence to obtain the clustering results, which include: According to the initial cluster center, combined with each parameter in the standard process data and the characteristic weight of each parameter, the distance between the initial cluster center and each standard process data is given; Analyze the distance between the initial cluster center and each standard process data, update the cluster center iteratively, and give the distance between the new cluster center and each standard process data; Based on the distance between the new cluster center and each standard process data, the mean distance between the standard process data in each cluster and the cluster center is analyzed to give the clustering result.

8. The method for generating a semiconductor secondary piping PID according to claim 1, wherein: Combined with the process component matching data, each component is analyzed and adjusted to generate the semiconductor secondary piping PID, including: Based on the process component matching data, draw the components corresponding to each standard process data to generate the initial PID; Construct a reward function based on the component relationships of each component in the initial PID; Combined with the reward function, the component relationship of each component in the initial PID is adjusted to give a standard PID; Render the standard PID and generate the semiconductor secondary piping PID.

9. The method for generating a semiconductor secondary piping PID according to claim 8, wherein: According to the component relationship of each component in the initial PID, a reward function is constructed, including: According to the connection relationship between each component in the initial PID, the gap between the corresponding connection indicators of each component is analyzed, and the difference ranking of the connection indicators is given; Based on the difference ranking of connection indicators, corresponding weights are assigned to each connection indicator, and a reward weight matrix is ​​given; Combined with the reward weight matrix, a reward function is constructed.

10. A semiconductor secondary piping PID generation device, characterized in that: The method for generating a semiconductor secondary piping PID as claimed in any one of claims 1 to 9 comprises: A data acquisition module, used to acquire initial process data for a semiconductor secondary piping operation process; The data processing module is used to match the corresponding normalization strategy based on the distribution state of each parameter in the initial process data to obtain the standard process data; The weight determination module is used to obtain the information amount of each parameter in the standard process data based on the analysis of the standard process data, and to give the characteristic weight of each parameter in the standard process data in combination with the initial weight of each parameter; The component matching module is used to combine the feature weights of each parameter in the standard process data, perform cluster analysis on the standard process data, and obtain process component matching data; The PID generation module is used to combine the process component matching data, analyze and adjust each component, and generate the semiconductor secondary piping PID.

Images