A rule designer and method for metallurgical process and quality

Abstract

The present invention provides a rule designer and method for metallurgical process and quality, wherein the rule designer includes a rule engine module, a rule editing module, a rule evaluation module, and a data environment module; wherein: a rule engine is created in the rule engine module as a container of rules to support the operation of the rules; the rule editing module is used for the definition of rules; the rule evaluation module is used for the testing and verification of rules; and the data environment module provides variables for rule definition and virtual connections between variables and data. The traditional rule design method cannot meet the requirements of complex rules for multi-source heterogeneous big data across metallurgical processes, and can efficiently design complex rules based on multi-source heterogeneous big data, which can not only improve the efficiency of data use, but also realize real-time judgment of process quality, timely resolution of process failures, and early re-judgment and downgrade of problematic products through the application of rules, so as to avoid problematic products flowing to the next process.

Description

Technical Field

[0001] The present invention relates to the technical field of metallurgical process quality control, and in particular to a metallurgical process and quality rule designer and method. Background Art

[0002] At present, domestic steel enterprises still generally rely on post-inspection and testing to control product quality. There are conventional quality judgment rules at the enterprise ERP level, but because the data at the ERP level is a single-point statistical result data, the judgment rules are usually based on the threshold judgment of production rules, using the traditional rule design method.

[0003] With the rapid development of information technology such as big data and the further improvement of customer requirements for product quality, a few first-class steel companies have begun to implement the construction of a full-process and quality big data platform. The platform data comes from a wide range of sources and types, including both structured data and unstructured data. If the traditional rule design method is continued to be applied, the application effect of the data will inevitably be reduced, resulting in a waste of data resources. Summary of the invention

[0004] In order to solve the technical problems raised by the background technology, the present invention provides a rule designer and method for metallurgical process and quality, which solves the problem that traditional rule design methods cannot meet the requirements of complex rules for multi-source heterogeneous big data across metallurgical processes. It can efficiently design complex rules based on multi-source heterogeneous big data, which can not only improve data utilization efficiency, but also realize real-time judgment of process quality and timely resolution of process failures through the application of rules, and early re-judgment and downgrading of problematic products to avoid problematic products flowing to the next process.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A rule designer for metallurgical process and quality, comprising a rule engine module, a rule editing module, a rule evaluation module, and a data environment module; wherein:

[0007] Create a rule engine in the rule engine module as a container of rules to support the running of rules;

[0008] The rule editing module is used for rule definition;

[0009] The rule evaluation module is used for testing and verification of rules;

[0010] The data environment module provides variables for rule definition and virtual connections between variables and data.

[0011] Furthermore, a rule engine is created or edited in the rule engine module, and based on the rule engine, rule results are defined, rule groups are managed, global constants are managed, and global scripts are managed; the definition of rule results includes the name, priority, color, and description of the rule results; the rule group management has a multi-level grouping mechanism, which can be customized and flexibly grouped by production line, customer, product type, and purpose, or grouped using variables in the data environment module; global constant management is used to define constants and provide them to all rules in the rule editing module; global script management is used for processing script editing of all rules in the rule editing module before and after rule execution.

[0012] Furthermore, the configuration of the rule engine is stored in an XML format file. At the same time, the XML format file includes the definition of the rule in the rule editing module, and different versions of the rule engine configuration are stored in the rule library.

[0013] Furthermore, the rules defined in the rule editing module have one or more judgment conditions and produce a unique judgment result. When there are multiple judgment conditions in a rule, the judgment conditions are connected with "AND" or "OR" to indicate the "AND" and "OR" relationship between the conditions. If the judgment condition obtains "True" through logical operation, then the corresponding rule result will be obtained.

[0014] Furthermore, the rule types in the rule editing module are divided into range rules and script rules; range rules are used for quick definition of rules, and the judgment condition has only one variable; script rules are an extension of range rules, and complex rules are defined based on scripting languages. Furthermore, when defining range rules, the variables defined in the data environment module are selected as judgment conditions, and variable preprocessing is performed, including dimension conversion and four arithmetic operations between variables and constants; the maximum and minimum values ​​of the variables, the minimum change value of the variables, the number of decimal places of the variables, and the units of the variables are limited; the default judgment threshold is calculated by the number of rule results and the maximum and minimum values ​​of the variables in the judgment conditions; the minimum value of the variable is set to V min , the maximum value is V max , the number of rule results is N, then the default decision thresholds corresponding to the N rule results with the highest priority from the lowest are The decision threshold is dynamically adjusted through the boundary slider in the human-computer interaction interface of the rule designer.

[0015] Furthermore, when defining script rules, the script language edits Boolean expressions by acquiring variables in the data environment. The result returned by the expression is of Boolean type, True or False. The constants in the expression can use global constants defined in the rule engine or local constants defined within the rule. The script language creates complex rules by flexibly programming acquired variables and constants.

[0016] Furthermore, the scope rules and script rules have two modes, online and simulation; the online mode is set by the combined use of two switches, which are E and D. If only E is selected, the rule participates in the automatic judgment of process and quality; if only D is selected, the rule will not be executed; if E and D are selected at the same time, the rule will be executed, but will not affect the final judgment result; the simulation mode is set by two switches, which are M and S; as long as M is selected, the rule will not be executed; when S is selected and M is not selected, the rule will be executed alone.

[0017] Furthermore, in the rule evaluation module, the evaluated steel coil data set can be precisely limited; the limiting conditions include steel coil data selection, steel coil data sorting, steel coil data volume, and steel coil data advanced limitation, and each limiting condition needs to be met at the same time; among them, steel coil data selection sets limiting conditions within the steel coil master data range, including fuzzy query of steel coil number, steel type, steel coil width range, steel coil weight range, steel coil length range and other limiting conditions that can be used as steel coil data selection; steel coil data sorting is achieved by selecting variables in the steel coil master data as keywords, and setting them to ascending or descending order according to the selected keywords; the steel coil data volume defaults to 1,000 items and can be set manually; advanced steel coil data limitation uses the LINQ language to make more complex limitations on the scope of the steel coil data set.

[0018] Furthermore, in the rule evaluation module, the evaluation results and historical results are displayed in the form of hit percentage and correlation matrix; the rules are adjusted by comparing the hit situations of the evaluation results and historical results to avoid the situation where the window range of some rule judgment conditions in the rule set is too narrow, resulting in too low a hit rate of the rule set.

[0019] Furthermore, the data configuration in the data environment module is stored in an XML file. The data environment module establishes a virtual connection between the rule designer and the multi-source heterogeneous big data database, and manages cross-line data through a global unique identifier, and manages single-line data through a product unique identifier. The global unique identifier is used to associate cross-line data, thereby realizing the creation of cross-line rules. The variables defined in the data environment module can directly access the data in the database; when the rule designer is started, the data environment module automatically establishes a connection with the background database.

[0020] The rule design method of the rule designer for metallurgical process and quality comprises the following steps:

[0021] (1) Create or edit a rule engine in the rule engine module, and based on the rule engine, define or edit rule results, rule groups, global constants, and global scripts as required;

[0022] (2) Create or edit rules in the rule editing module and set the rule parameters, which are obtained from the data environment module;

[0023] (3) Setting groups for rules to limit the scope of application of the rules;

[0024] (4) Repeat steps (2) and (3) until the created rule set meets the process quality determination requirements;

[0025] (5) According to the process quality goal, the created rule set is evaluated by the rule evaluation module;

[0026] (6) If a rule in the rule set contributes more to the unqualified evaluation result, analyze the rule and adjust the judgment condition of the rule;

[0027] (7) Repeat (5) and (6);

[0028] (8) The rule engine configuration is saved.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] 1) A rule designer and method for metallurgical process and quality of the present invention solves the problem that traditional rule design methods cannot meet the requirements of complex rules for multi-source heterogeneous big data across metallurgical processes. It can efficiently design complex rules based on multi-source heterogeneous big data, which can not only improve the efficiency of data use, but also realize real-time judgment of process quality and timely resolution of process failures through the application of rules, and re-judge and downgrade problematic products as soon as possible to avoid problematic products flowing to the next process;

[0031] 2) After the application of the rule designer and method for metallurgical process quality of the present invention, it is possible to solve the problem that traditional rule design methods cannot meet the requirements of complex rules for process and quality judgment of metallurgical cross-process multi-source heterogeneous big data. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 A schematic diagram of a metallurgical process and quality rule designer of the present invention;

[0033] Figure 2The present invention is a flow chart of a metallurgical process and quality rule design method. DETAILED DESCRIPTION

[0034] The specific implementation modes provided by the present invention are described in detail below with reference to the accompanying drawings.

[0035] A rule designer for metallurgical process and quality, comprising a rule engine module, a rule editing module, a rule evaluation module, and a data environment module; wherein:

[0036] Create a rule engine in the rule engine module as a container of rules to support the running of rules;

[0037] The rule editing module is used for rule definition;

[0038] The rule evaluation module is used for testing and verification of rules;

[0039] The data environment module provides variables for rule definition and virtual connections between variables and data.

[0040] Furthermore, a rule engine is created or edited in the rule engine module, and based on the rule engine, rule results are defined, rule groups are managed, global constants are managed, and global scripts are managed; the definition of rule results includes the name, priority, color, and description of the rule results; the rule group management has a multi-level grouping mechanism, which can be customized and flexibly grouped by production line, customer, product type, and purpose, or grouped using variables in the data environment module; global constant management is used to define constants and provide them to all rules in the rule editing module; global script management is used for processing script editing of all rules in the rule editing module before and after rule execution.

[0041] Furthermore, the configuration of the rule engine is stored in an XML format file. At the same time, the XML format file includes the definition of the rule in the rule editing module, and different versions of the rule engine configuration are stored in the rule library.

[0042] Furthermore, the rules defined in the rule editing module have one or more judgment conditions and produce a unique judgment result. When there are multiple judgment conditions in a rule, the judgment conditions are connected with "AND" or "OR" to indicate the "AND" and "OR" relationship between the conditions. If the judgment condition obtains "True" through logical operation, then the corresponding rule result will be obtained.

[0043] Furthermore, the rule types in the rule editing module are divided into scope rules and script rules; scope rules are used for quick definition of rules, and the judgment condition has only one variable; script rules are an extension of scope rules, and define complex rules based on scripting language.

[0044] Furthermore, when defining the range rule, the variables defined in the data environment module are selected as the judgment conditions, and variable preprocessing is performed, including dimension conversion and four arithmetic operations between variables and constants; the maximum and minimum values ​​of the variables, the minimum change value of the variables, the number of decimal places of the variables, and the units of the variables are limited; the default judgment threshold is calculated by the number of rule results and the maximum and minimum values ​​of the variables in the judgment conditions; the minimum value of the variable is set to V min , the maximum value is V max , the number of rule results is N, then the default decision thresholds corresponding to the N rule results with the highest priority from the lowest are The decision threshold is dynamically adjusted through the boundary slider in the human-computer interaction interface of the rule designer.

[0045] Furthermore, when defining script rules, the scripting language edits Boolean expressions by acquiring variables in the data environment module. The result returned by the expression is of Boolean type, True or False. The constants in the expression can use global constants defined in the rule engine or local constants defined within the rule. The scripting language creates complex rules by flexibly programming acquired variables and constants.

[0046] Furthermore, the scope rules and script rules have two modes, online and simulation; the online mode is set by the combined use of two switches, which are E and D. If only E is selected, the rule participates in the automatic judgment of process and quality; if only D is selected, the rule will not be executed; if E and D are selected at the same time, the rule will be executed, but will not affect the final judgment result; the simulation mode is set by two switches, which are M and S; as long as M is selected, the rule will not be executed; when S is selected and M is not selected, the rule will be executed alone.

[0047] Furthermore, in the rule evaluation module, the evaluated steel coil data set can be precisely limited; the limiting conditions include steel coil data selection, steel coil data sorting, steel coil data volume, and steel coil data advanced limitation, and each limiting condition needs to be met at the same time; among them, steel coil data selection sets limiting conditions within the steel coil master data range, including fuzzy query of steel coil number, steel type, steel coil width range, steel coil weight range, steel coil length range and other limiting conditions that can be used as steel coil data selection; steel coil data sorting is achieved by selecting variables in the steel coil master data as keywords, and setting them to ascending or descending order according to the selected keywords; the steel coil data volume defaults to 1,000 items and can be set manually; advanced steel coil data limitation uses the LINQ language to make more complex limitations on the scope of the steel coil data set.

[0048] Furthermore, in the rule evaluation module, the evaluation results and historical results are displayed in the form of hit percentage and correlation matrix; the rules are adjusted by comparing the hit situations of the evaluation results and historical results to avoid the situation where the window range of some rule judgment conditions in the rule set is too narrow, resulting in too low a hit rate of the rule set.

[0049] Furthermore, the data configuration in the data environment module is stored in an XML file. The data environment module establishes a virtual connection between the rule designer and the multi-source heterogeneous big data database, and manages cross-line data through a global unique identifier, and manages single-line data through a product unique identifier. The global unique identifier is used to associate cross-line data, thereby realizing the creation of cross-line rules. The variables defined in the data environment module can directly access the data in the database; when the rule designer is started, the data environment module automatically establishes a connection with the background database.

[0050] The rule design method of the rule designer for metallurgical process and quality comprises the following steps:

[0051] (1) Create or edit a rule engine in the rule engine module. Based on the rule engine, define or edit rule results, rule groups, global constants, and global scripts according to requirements;

[0052] (2) Create or edit rules in the rule editing module and set the rule parameters, which are obtained from the data environment module;

[0053] (3) Setting groups for rules to limit the scope of application of the rules;

[0054] (4) Repeat steps (2) and (3) until the created rule set meets the process quality determination requirements;

[0055] (5) According to the process quality goal, the created rule set is evaluated by the rule evaluation module;

[0056] (6) If a rule in the rule set contributes more to the unqualified evaluation result, analyze the rule and adjust the judgment condition of the rule;

[0057] (7) Repeat (5) and (6);

[0058] (8) The rule engine configuration is saved. Specific embodiment:

[0060] As attached Figure 1As shown, in order to realize the four modules of a rule designer for metallurgical process and quality, namely, rule engine module, rule editing module, rule evaluation module, and data environment module, the architecture of the designer should include system application, configuration file, rule base, and database. The system application is the human-computer interface of the designer, which realizes the human-computer interaction function. The configuration file records the configuration of the rule engine, including the rules created in the rule engine, which can use but is not limited to the use of XML language. The rule base saves different versions of different rule engines. For the same rule engine, only one can be activated. The database provides background data support for the variables in the data environment module.

[0061] The file menu is used to create, open, save and activate the rule engine. The rule engine menu is used to define rule results, manage rule groups, manage global constants and manage global scripts. The rule evaluation coil selection is used to accurately limit the steel coil data set to be evaluated. The rule evaluation results will represent the evaluation judgment results and historical judgment results in the form of hit percentage and correlation matrix. The rule editor is used to add and edit rules. The data environment provides the environment for data selection for the rule designer. The rule grouping can represent the grouping of rules in the form of a tree, including the number of rules included in each group.

[0062] The rule base can use relational database software to create two data tables, namely RuleEngines and RuleEngineVersions.

[0063] RuleEngines stores the names of all defined rule engines, and RuleEngineVersions stores the rule engine version, the content of the rule engine configuration file, and whether it is activated. The definition of RuleEngines is shown in Table 1;

[0064] Table 1

[0065] Serial number Data Item Data Types illustrate 1 Id int Rule Engine ID 2 Name nvarchar(150) Rule Engine Name

[0066] The definition of RuleEngineVersions is shown in Table 2:

[0067] Table 2

[0068]

[0069] The Id in Table 1 is associated with the RuleEngineId in Table 2.

[0070] The database can be created according to the actual situation of the data source. The single production line data is indexed by the product identifier, such as slab number, steel coil number, etc. The cross-production line data is indexed by the global unique identifier as the keyword.

[0071] In order to realize a rule-based design method for metallurgical process and quality, the process is as follows: Figure 2 As shown. If the rule engine does not exist, create a new rule engine and set a name for the rule engine. The version number is automatically generated. Here, it is assumed that the rule engine name is EngineDemo. Group A1 is created according to the production line. The rule results are defined from high to low priority as closed, viewed, and qualified, and are represented by three different colors: red, yellow, and green. Create a new rule in the rule engine and select the rule type and name when creating a new rule.

[0072] First, create a range rule to determine whether the fluctuation range of the strip width meets the quality requirements. Name it Range_Width, select the fluctuation amount of the strip width in the data environment as the judgment condition, set the minimum value to -30, the maximum value to 30, the minimum change value to 1, the number of decimal places of the variable to 1, and the unit of the variable to mm. Since three judgment results are defined, the default judgment thresholds corresponding to the rule results of closed, viewed, and qualified are [-30.0,-10.0), [-10.0,10.0), [10.0,30.0] respectively. The default thresholds are dynamically adjusted to correspond to the rule results of closed, viewed, qualified, viewed, and closed, respectively [-30.0,-20.0), [-20.0,-10.0), [-10.0,10.0), [10.0,20.0), [20.0,30.0]. Limit the rule Range_Width to group A1. Select "Yes" to create other rules.

[0073] Create another script rule to implement the judgment of whether the number of defects whose surface defect area of ​​the strip is greater than a certain value is qualified, and name it Script_Defects. Define two local constants, $allowedDefectArea and $allowedDefectAreaViolation, which represent the judgment threshold of the defect area and the judgment threshold of the number of defects whose number of defects exceeds the judgment threshold of the defect area respectively. The specific value can be set according to the actual situation. Assume that a virtual connection with the surface defect data table Defects of the strip in the database is defined in the data environment module, and Defects contains defect width and defect height data, which are Width and Height respectively. Then, define the number of defects in the rule script as vardefectsCount, where var is the variable type and defectsCount is the variable name. Here, use LINQ language to obtain the number of defects whose surface defect area of ​​the strip is greater than $allowedDefectArea. Refer to the following script:

[0074] var

[0075] defectsCount=Defects.Where(def=>((double)(def.Width*def.Height))>$allowedDefectArea).ToList().Count();

[0076] If defectsCount is greater than $allowedDefectAreaViolation, the script execution result returns True, otherwise it returns False.

[0077] Limit the rule Script_Defects to the A1 group, select "No" for whether to create other rules, and enter the rule evaluation. Evaluate the steel coil selection through the rules, and limit the data range of the rule evaluation. By analyzing the evaluation results and historical judgment results through the hit percentage and correlation matrix, if the evaluation results are lower than the historical judgment results, it is necessary to analyze the rules with higher contribution to the unqualified evaluation results in the rule set, and adjust the judgment conditions of the rules. That is, in the rule design method process, select "No" for whether the rules meet the requirements, and continue to edit the rule engine. If the rule set meets the requirements, then in the rule design method process, select "Yes" for whether the rules meet the requirements, and save the rule engine configuration. At this point, the rule design process of a rule designer based on a metallurgical process and quality is completed.

[0078] The above embodiments are implemented based on the technical solution of the present invention, and detailed implementation methods and specific operation processes are given, but the protection scope of the present invention is not limited to the above embodiments. The methods used in the above embodiments are conventional methods unless otherwise specified.

Claims

1. A rule designer for metallurgical process and quality, characterized by: It includes rule engine module, rule editing module, rule evaluation module and data environment module; among which: Create a rule engine in the rule engine module as a container of rules to support the running of rules; The rule editing module is used for rule definition; The rule evaluation module is used for testing and verification of rules; The data environment module provides variables for rule definition and virtual connections between variables and data; Create or edit the rule engine in the rule engine module, and define the rule results, manage rule groups, manage global constants, and manage global scripts based on the rule engine; the definition of the rule results includes the name, priority, color, and description of the rule results; the rule group management has a multi-level grouping mechanism, which allows for flexible custom grouping by production line, customer, product type, and purpose, or grouping using variables in the data environment module; global constant management is used to define constants for use by all rules in the rule editing module; global script management is used for editing processing scripts for all rules in the rule editing module before and after rule execution; The rule types in the rule editing module are divided into scope rules and script rules. Scope rules are used to quickly define rules, and the judgment condition has only one variable. Script rules are an extension of scope rules, which define complex rules based on script language and are applicable to all application scenarios of metallurgical process quality rule design. When defining the range rule, the variables defined in the data environment module are selected as the judgment conditions, and variable preprocessing is performed, including dimension conversion and four arithmetic operations between variables and constants; the maximum and minimum values ​​of the variables, the minimum change value of the variables, the number of decimal places of the variables, and the units of the variables are limited; the default judgment threshold is calculated by the number of rule results and the maximum and minimum values ​​of the variables in the judgment conditions; the minimum value of the variable is set to V min , the maximum value is V max , the number of rule results is N, then the default decision thresholds corresponding to the N rule results with the highest priority from the lowest are The decision threshold is dynamically adjusted through the boundary slider in the human-computer interaction interface of the rule designer; Range rules and script rules have two modes: online and simulation. The online mode is set by combining two switches, E and D. If only E is selected, the rule participates in the automatic judgment of process and quality. If only D is selected, the rule will not be executed. If E and D are selected at the same time, the rule will be executed, but it will not affect the final judgment result. The simulation mode is set by two switches, M and S. As long as M is selected, the rule will not be executed. When S is selected and M is not selected, the rule is executed alone. In the rule evaluation module, the evaluated steel coil data set can be precisely defined; the limiting conditions include steel coil data selection, steel coil data sorting, steel coil data volume, and steel coil data advanced limitation, and each limiting condition needs to be met at the same time; among them, steel coil data selection sets limiting conditions within the steel coil master data range, including fuzzy query of steel coil number, steel type, steel coil width range, steel coil weight range, steel coil length range, and other limiting conditions that can be used as steel coil data selection; steel coil data sorting selects variables in the steel coil master data as keywords, and sets them to ascending or descending order according to the selected keywords; and can be set manually arbitrarily; advanced steel coil data limitation uses LINQ language to make more complex limitations on the steel coil data set range; The data configuration in the data environment module is stored in an XML file. The data environment module establishes a virtual connection between the rule designer and a multi-source heterogeneous big data database, and manages cross-production line data through a globally unique identifier. The product unique identifier manages single production line data, and the globally unique identifier is used to associate cross-production line data, thereby realizing the creation of cross-production line rules. The variables defined in the data environment module can directly access the data in the database; when the rule designer is started, the data environment module automatically establishes a connection with the background database.

2. A metallurgical process and quality rule designer according to claim 1, characterized in that: The configuration of the rule engine is stored in an XML format file. At the same time, the XML format file contains the definition of the rules in the rule editing module, and saves different versions of the rule engine configuration in the rule library.

3. A metallurgical process and quality rule designer according to claim 1, characterized in that: In the rule evaluation module, the evaluation results and historical results are displayed in the form of hit percentage and correlation matrix. The rules are adjusted by comparing the hit situations of the evaluation results and historical results to avoid the situation where the window range of some rule judgment conditions in the rule set is too narrow, resulting in too low a hit rate of the rule set.

4. A metallurgical process and quality rule designer according to claim 1, characterized in that: The rule design method of the rule designer comprises the following steps: (1) Create or edit a rule engine in the rule engine module, and based on the rule engine, define or edit rule results, rule groups, global constants, and global scripts as required; (2) Create or edit rules in the rule editing module and set the rule parameters, which are obtained from the data environment module; (3) Setting groups for rules to limit the scope of application of the rules; (4) Repeat steps (2) and (3) until the created rule set meets the process quality determination requirements; (5) According to the process quality goal, the created rule set is evaluated by the rule evaluation module; (6) If a rule in the rule set contributes most to the unqualified evaluation result, analyze the rule and adjust the judgment condition of the rule; (7) Repeat (5) and (6); (8) The rule engine configuration is saved.

Images