Key metal quality processing method and device based on rule conflict detection graph

Abstract

The application discloses a key metal quality processing method and device based on a rule conflict detection graph, relates to the technical field of key metal industry quality management, and comprises the following steps: analyzing external standard clauses and internal quality constraints into minimum executable rule atoms; modeling historical quality distribution based on a specific context to generate a dynamic threshold value; constructing a rule conflict detection graph to identify conflicts and prioritize a candidate rule set; performing composite space-time rule matching on a real-time event stream; and realizing continuous operation and historical recalculation through collaborative management of rule set versions, threshold value versions and graph mapping versions. Thus, the technical problems of heterogeneous rule sources, parallel validity, obvious context dependency of quality threshold values, time and space composite characteristics of abnormal judgment, and cascading mismatch of rules and graph structures caused by standard iteration in the existing quality management process of the key metal industry chain are solved.

Description

Technical Field

[0001] This application relates to the field of quality management technology in the critical metal industry, and in particular to a critical metal quality processing method and apparatus based on a rule conflict detection chart. Background Technology

[0002] The critical metal supply chain typically encompasses multiple stages, including raw material supply, smelting and separation, refining and purification, deep processing, inspection and testing, and downstream delivery. While companies generally possess some data collection capabilities in actual quality management, the following technical bottlenecks still exist: 1. Heterogeneous source of rules and concurrent effect: The same quality indicator is often subject to multiple rules such as national standards, industry standards, enterprise internal control standards, customer contract terms and temporary quality notices. There may be differences in thresholds, scope of application and even conflicting conclusions between different rules.

[0003] 2. Quality thresholds are highly context-dependent: the reasonable distribution range of the same quality indicator is not constant under different metal categories, different process routes, different equipment, different versions of testing methods, and different time periods. Using a fixed threshold strategy can easily lead to an increase in false alarm rate or an increase in the risk of missed alarm.

[0004] 3. Anomaly detection often has spatiotemporal complex characteristics: In actual business, anomalies are not always manifested as a single record going out of bounds, but often manifested as complex event patterns such as "multiple batches of a certain supplier are close to the threshold", "same type of deviation occurs in a certain region in a specific season", and "continuous anomalies are superimposed across processes". Traditional single-point rules are difficult to identify effectively.

[0005] 4. Standard iteration can cause cascading mismatches between rules and graph structure: When standard clauses add detection items, modify indicator names, switch unit calibers, or update judgment formulas, the original rules, field mappings and historical judgment results may lose consistency, affecting the sustainable operation of the system.

[0006] In summary, in the existing quality management process of the critical metal industry chain, the rules are heterogeneous and have a parallel effect, the quality threshold has obvious context dependence, and the anomaly judgment often has spatiotemporal composite characteristics. Furthermore, the standard iteration will cause a cascade mismatch between the rules and the graph structure, which urgently needs to be addressed. Summary of the Invention

[0007] This application provides a method and apparatus for quality processing of critical metals based on a rule conflict detection graph, in order to solve the problems in the existing quality management process of critical metal industry chain, such as heterogeneous rule sources and parallel effects, obvious context-dependent quality thresholds, spatiotemporal composite characteristics of anomaly judgment, and cascading mismatch between rules and graph structure caused by standard iteration.

[0008] The first aspect of this application provides a method for processing the quality of critical metals based on a rule conflict detection graph, comprising the following steps: acquiring target rule representation data for a target critical metal industry quality management scenario, and determining the context fragments and corresponding quality data stream information corresponding to the target rule representation data, to perform preset robust quantile modeling, generating upper and lower limit thresholds corresponding to context binding relationships, and performing a preset threshold re-estimation operation based on the upper and lower limit thresholds to obtain a corresponding context binding threshold set and threshold version object; constructing a corresponding rule conflict detection graph based on the candidate rule set of the target rule representation data, the context binding threshold set, and the context fragments, and performing a preset priority adjudication operation on the rule conflict detection graph to generate a valid rule set and a valid rule set version corresponding to the context fragments; standardizing the online detection data in the target rule representation data into event objects based on the target rule representation data, the threshold version object, and the valid rule set version, and performing a preset composite spatiotemporal rule matching operation based on the event objects to generate corresponding evidence information, and combining the valid rule set version, the threshold version object, and the target rule representation data to perform a preset version governance and rollback operation to obtain the corresponding critical metal quality processing result.

[0009] Optionally, in one embodiment of this application, obtaining the target rule representation data for the target key metal industry quality management scenario includes: obtaining various quality management data for the target key metal industry quality management scenario, wherein the various quality management data includes multi-source standard clause data, master data, graph basic data, historical testing data, and online testing data; performing preset field-level parsing operations on the multi-source standard clause data to extract corresponding multiple field information, wherein the multiple field information includes indicator name, judgment relationship, threshold or formula, scope of application, effective date, source level, and unit caliber; performing preset unit normalization and caliber verification on the multiple field information to obtain corresponding multiple rule information, and decomposing each rule information into the smallest executable rule atom, and constructing a set of rule atoms corresponding to the multiple rule information; constructing a corresponding candidate rule set based on the set of rule atoms and combined with preset validity domain constraints; constructing a corresponding field mapping registry based on the multiple field information, and determining the target rule representation data based on the field mapping registry, the set of rule atoms, and the set of candidate rules.

[0010] Optionally, in one embodiment of this application, determining the context slices and corresponding quality data stream information corresponding to the target rule representation data for performing preset robust quantile modeling, generating upper and lower bound thresholds corresponding to the context binding relationship, and performing a preset threshold reestimation operation based on the upper and lower bound thresholds to obtain the corresponding context binding threshold set and threshold version object includes: obtaining multiple context slices and corresponding quality data stream information corresponding to the target rule representation data, determining the indicator sequence corresponding to each context slice, and constructing a corresponding quantile estimation function based on the quality data stream information, a preset target score point, and a learning rate; performing preset robust quantile modeling on the indicator sequence through the quantile estimation function to generate the context binding relationship and threshold version object corresponding to each context slice. The context binding relationship corresponds to upper and lower limit thresholds; based on the context binding relationship and the upper and lower limit thresholds, the recent window distribution and the baseline window distribution corresponding to each context slice are determined, and the corresponding distribution drift metric is calculated according to the recent window distribution and the baseline window distribution; it is determined whether the distribution drift metric is greater than a preset context binding threshold, and in response to the distribution drift metric being greater than the context binding threshold, a threshold re-estimation operation is performed to generate a new context binding threshold; it is determined whether the new context binding threshold meets the preset detection requirements within a preset grayscale observation period, and in response to the new context binding threshold meeting the preset detection requirements, the context binding threshold set is obtained, and the context binding threshold set and the context slice are bound to generate the corresponding threshold version object.

[0011] Optionally, in one embodiment of this application, the step of constructing a corresponding rule conflict detection graph based on the candidate rule set of the target rule representation data, the context binding threshold set, and the context fragments, and performing a preset priority adjudication operation on the rule conflict detection graph to generate a valid rule set and a valid rule set version corresponding to the context fragments, includes: determining a candidate rule subset for each context fragment based on the candidate rule set, and constructing the rule conflict detection graph according to the candidate rule subsets; calculating the rule source level score, timeliness score, scope of application specificity score, and historical execution reliability score for each rule in the candidate rule subsets, and performing a weighted summation operation on the rule source level score, the timeliness score, the scope of application specificity score, and the historical execution reliability score to obtain a priority score corresponding to each rule; determining whether there is a cycle in the rule conflict detection graph, and in response to the presence of a cycle in the rule conflict detection graph, performing a preset minimum cost cycle breaking operation to generate a corresponding acyclic graph, and performing topological sorting on the remaining rules in the candidate rule subsets corresponding to the acyclic graph to generate the valid rule set and the valid rule set version.

[0012] Optionally, in one embodiment of this application, the step of standardizing the online detection data in the target rule representation data into event objects based on the target rule representation data, the threshold version object, and the effective rule set version, and performing a preset composite spatiotemporal rule matching operation on the event objects to generate corresponding evidence information includes: standardizing the online detection data into event objects based on the graph base data in the target rule representation data, the threshold version object, and the effective rule set version; obtaining the entity relationships corresponding to the graph base data, and performing a preset association extension operation on the event objects based on the entity relationships to obtain corresponding association extension data; performing a complex event matching operation on the association extension data based on a preset sliding time window and spatial range constraints to determine the corresponding matching conditions, and generating corresponding evidence information based on the matching conditions, wherein the evidence information includes quality warning results, compliance judgment results, audit evidence packages, and rule hit logs.

[0013] Optionally, in one embodiment of this application, the step of combining the effective rule set version, the threshold version object, and the target rule representation data to perform preset version governance and rollback operations to obtain the corresponding key metal quality processing results includes: obtaining new version standard clauses or field structure change information corresponding to the target key metal industry quality management scenario, and mapping the new version standard clauses or field structure change information to the current graph of the target rule representation data to calculate the literal similarity, unit compatibility, and contextual adaptation between the terminology information and graph attributes in the graph base data; and calculating the comprehensive similarity between the terminology information and the graph attributes based on the literal similarity, unit compatibility, and contextual adaptation. The system determines whether the overall similarity is greater than a preset similarity threshold. If the overall similarity is greater than the similarity threshold, a preset mode patching operation is performed to generate a new map mapping version. A preset gray-scale release operation is then performed on the effective rule set version, the threshold version object, and the new map mapping version. The system also checks whether a preset false positive rate requirement is met during the preset gray-scale release operation. If the preset false positive rate requirement is met, a preset version rollback and recalculation operation is performed on the effective rule set version, the threshold version object, and the new map mapping version to obtain the corresponding key metal quality processing results. The key metal quality processing results include version release results, version audit logs, rollback records, and historical recalculation results.

[0014] A second aspect of this application provides a critical metal quality processing device based on a rule conflict detection graph, comprising: a threshold re-estimation module, configured to acquire target rule representation data for a target critical metal industry quality management scenario, and determine the context fragments and corresponding quality data stream information corresponding to the target rule representation data, to perform preset robust quantile modeling, generate upper and lower limit thresholds corresponding to context binding relationships, and perform a preset threshold re-estimation operation based on the upper and lower limit thresholds to obtain a corresponding context binding threshold set and threshold version object; and a priority adjudication module, configured to, based on the candidate rule set of the target rule representation data, the context binding threshold set, and the context fragments, ... A corresponding rule conflict detection graph is constructed, and a preset priority adjudication operation is performed on the rule conflict detection graph to generate the effective rule set and effective rule set version corresponding to the context segment; the quality processing module is used to standardize the online detection data in the target rule representation data into event objects based on the target rule representation data, the threshold version object, and the effective rule set version, and perform a preset composite spatiotemporal rule matching operation according to the event objects to generate corresponding evidence information, and combine the effective rule set version, the threshold version object, and the target rule representation data to perform preset version governance and rollback operations to obtain the corresponding key metal quality processing results.

[0015] Optionally, in one embodiment of this application, the threshold re-estimation module includes: a first acquisition unit, used to acquire various quality management data of the target key metal industry quality management scenario, wherein the various quality management data includes multi-source standard clause data, master data, graph basic data, historical testing data, and online testing data; a parsing unit, used to perform preset field-level parsing operations on the multi-source standard clause data to extract corresponding multiple field information, wherein the multiple field information includes indicator name, judgment relationship, threshold or formula, scope of application, effective date, source level, and unit caliber; a decomposition unit, used to perform preset unit normalization and caliber verification on the multiple field information to obtain corresponding multiple rule information, and decompose each rule information into the smallest executable rule atom, and construct a set of rule atoms corresponding to the multiple rule information; a first construction unit, used to construct a corresponding candidate rule set based on the set of rule atoms and combined with preset validity domain constraints; and a determination unit, used to construct a corresponding field mapping registry based on the multiple field information, and to determine the target rule representation data based on the field mapping registry, the set of rule atoms, and the set of candidate rules.

[0016] Optionally, in one embodiment of this application, the threshold re-estimation module further includes: a second acquisition unit, configured to acquire multiple context slices corresponding to the target rule representation data and quality data stream information corresponding to the multiple context slices, determine the indicator sequence corresponding to each context slice, and construct a corresponding fractional estimation function based on the quality data stream information, a preset target score point, and a learning rate; a generation unit, configured to perform preset robust quantile modeling on the indicator sequence through the fractional estimation function to generate the context binding relationship corresponding to each context slice and the upper and lower limit thresholds corresponding to the context binding relationship; and a first calculation unit, configured to determine the context binding relationship and the upper and lower limit thresholds for each context slice. The system identifies the recent window distribution and baseline window distribution corresponding to each slice, and calculates the corresponding distribution drift metric based on these distributions. A first judgment unit determines whether the distribution drift metric is greater than a preset context binding threshold. If the distribution drift metric is greater than the context binding threshold, a threshold re-estimation operation is performed to generate a new context binding threshold. A second judgment unit determines whether the new context binding threshold meets preset detection requirements within a preset grayscale observation period. If the new context binding threshold meets the preset detection requirements, the system obtains the context binding threshold set and binds the context binding threshold set and the context slice to generate a corresponding threshold version object.

[0017] Optionally, in one embodiment of this application, the priority adjudication module includes: a second construction unit, configured to determine a subset of candidate rules for each context slice based on the candidate rule set, and construct the rule conflict detection graph according to the subset of candidate rules; a weighted summation unit, configured to calculate the rule source level score, timeliness score, scope of application specificity score, and historical execution reliability score of each rule in the subset of candidate rules, and perform a weighted summation operation on the rule source level score, the timeliness score, the scope of application specificity score, and the historical execution reliability score to obtain the priority score corresponding to each rule; and a sorting unit, configured to determine whether there is a cycle in the rule conflict detection graph, and in response to the presence of a cycle in the rule conflict detection graph, perform a preset minimum cost cycle breaking operation to generate a corresponding acyclic graph, and perform topological sorting on the remaining rules in the subset of candidate rules corresponding to the acyclic graph to generate the effective rule set and the version of the effective rule set.

[0018] Optionally, in one embodiment of this application, the quality processing module includes: a standardization unit, used to standardize the online detection data into event objects based on the graph base data, the threshold version object, and the effective rule set version in the target rule representation data; an expansion unit, used to obtain the entity relationships corresponding to the graph base data, and perform a preset association expansion operation on the event objects according to the entity relationships to obtain corresponding association expansion data; and a matching unit, used to perform complex event matching operations on the association expansion data based on a preset sliding time window and spatial range constraints to determine the corresponding matching conditions, and generate corresponding evidence information according to the matching conditions, wherein the evidence information includes quality warning results, compliance judgment results, audit evidence packages, and rule hit logs.

[0019] Optionally, in one embodiment of this application, the quality processing module further includes: a mapping unit, configured to acquire new version standard clauses or field structure change information corresponding to the target key metal industry quality management scenario, and map the new version standard clauses or field structure change information to the current graph of the target rule representation data, so as to calculate the literal similarity, unit compatibility and context adaptation between the terminology information and the graph attributes in the graph basic data; a second calculation unit, configured to calculate the comprehensive similarity between the terminology information and the graph attributes based on the literal similarity, the unit compatibility and the context adaptation, and determine whether the comprehensive similarity is greater than a preset similarity threshold; and a detection unit. The system is configured to, in response to a situation where the overall similarity is greater than the similarity threshold, perform a preset mode patching operation to generate a new map mapping version, and perform a preset gray-scale release operation on the effective rule set version, the threshold version object, and the new map mapping version, and detect whether a preset false alarm rate requirement is met during the preset gray-scale release operation; the rollback unit is configured to, if the preset false alarm rate requirement is met, perform a preset version rollback and recalculation operation on the effective rule set version, the threshold version object, and the new map mapping version to obtain the corresponding key metal quality processing result, wherein the key metal quality processing result includes version release result, version audit log, rollback record, and historical recalculation result.

[0020] A third aspect of this application provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the critical metal quality processing method based on the rule conflict detection graph as described in the above embodiments.

[0021] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described critical metal quality processing method based on a rule-based conflict detection graph.

[0022] A fifth aspect of this application provides a computer program product, including a computer program that is executed to implement the above-described critical metal quality processing method based on a rule-based conflict detection graph.

[0023] Therefore, the embodiments of this application have the following beneficial effects: The embodiments of this application can acquire target rule representation data for a key metal industry quality management scenario, determine the context fragments and corresponding quality data flow information corresponding to the target rule representation data, perform preset robust quantile modeling, generate upper and lower limit thresholds corresponding to context binding relationships, and perform preset threshold reestimation operations based on the upper and lower limit thresholds to obtain the corresponding context binding threshold set and threshold version object; based on the candidate rule set, context binding threshold set, and context fragments of the target rule representation data, construct the corresponding rule conflict detection graph, and perform preset priority adjudication operations on the rule conflict detection graph to generate the effective rule set and effective rule set version corresponding to the context fragment; based on the target rule representation data, threshold version object, and effective rule set version, standardize the online detection data in the target rule representation data into event objects, and perform preset composite spatiotemporal rule matching operations based on the event objects to generate corresponding evidence information; and combine the effective rule set version, threshold version object, and target rule representation data to execute preset version governance and rollback operations to obtain the corresponding key metal quality processing results. This application uses knowledge graphs as the basis for entity relationship organization and context carrying, and a rule engine as the specific method, thereby realizing unified quality judgment and quality early warning under multi-source quality standards. This solves the problems in the existing quality management process of key metal industry chains, such as heterogeneous rule sources and concurrent effects, obvious context-dependent quality thresholds, spatiotemporal complex characteristics of anomaly detection, and cascading mismatch between rules and graph structures caused by standard iteration.

[0024] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0025] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart of a critical metal quality processing method based on a rule conflict detection graph provided in an embodiment of this application; Figure 2 This is an example diagram of a critical metal quality processing apparatus based on a rule conflict detection graph according to an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0026] Among them, 10-key metal quality processing device based on rule conflict detection graph, 100-threshold re-evaluation module, 200-priority adjudication module, 300-quality processing module, 301-memory, 302-processor, and 303-communication interface. Detailed Implementation

[0027] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0028] The following describes, with reference to the accompanying drawings, a method and apparatus for critical metal quality processing based on a rule-based conflict detection graph, according to embodiments of this application. To address the problems mentioned in the background, this application provides a method for processing the quality of critical metals based on a rule conflict detection graph. In this method, target rule representation data for a target critical metal industry quality management scenario is acquired, and the context fragments and corresponding quality data stream information corresponding to the target rule representation data are determined. Pre-defined robust quantile modeling is then performed to generate upper and lower thresholds corresponding to context binding relationships. A pre-defined threshold re-estimation operation is then performed based on these upper and lower thresholds to obtain the corresponding context binding threshold set and threshold version object. Based on the candidate rule set, context binding threshold set, and context fragments of the target rule representation data, a corresponding rule conflict detection graph is constructed. A pre-defined priority adjudication operation is then performed on the rule conflict detection graph to generate the effective rule set and effective rule set version corresponding to the context fragments. Based on the target rule representation data, threshold version object, and effective rule set version, the online detection data in the target rule representation data is standardized into event objects. A pre-defined composite spatiotemporal rule matching operation is then performed based on the event objects to generate corresponding evidence information. Finally, combining the effective rule set version, threshold version object, and target rule representation data, a pre-defined version governance and rollback operation is executed to obtain the corresponding critical metal quality processing result. This application uses knowledge graphs as the foundation for organizing entity relationships and carrying context, and a rule engine as the specific method, to achieve unified quality judgment and quality early warning under multi-source quality standards. This solves the problems in the existing quality management process of the critical metal industry chain, such as heterogeneous rule sources and concurrent effects, significant context-dependent quality thresholds, spatiotemporal complex characteristics of anomaly judgment, and the cascading mismatch between rules and graph structures caused by standard iteration.

[0029] Specifically, Figure 1 This is a flowchart of a critical metal quality processing method based on a rule conflict detection graph, provided as an embodiment of this application.

[0030] like Figure 1 As shown, the key metal quality processing method based on the rule-based conflict detection map includes the following steps: In step S101, target rule representation data for the target key metal industry quality management scenario is obtained, and the context shards and corresponding quality data flow information corresponding to the target rule representation data are determined to perform preset robust quantile modeling, generate upper and lower limit thresholds corresponding to the context binding relationship, and perform preset threshold re-estimation operation based on the upper and lower limit thresholds to obtain the corresponding context binding threshold set and threshold version object.

[0031] This application embodiment first obtains target rule representation data (i.e., including multi-source standard clause data, master data, graph basic data, historical detection data, and online detection data) in the key metal industry quality management scenario, determines the corresponding context fragments, and matches the corresponding quality data stream information. Then, through a preset robust quantile modeling strategy, it calculates the upper and lower limit thresholds of the context binding relationship, performs a re-evaluation operation on the thresholds, and finally generates a context binding threshold set and a threshold version object.

[0032] Therefore, the embodiments of this application rely on scenario rules, context fragmentation, and data flow to complete modeling, thereby accurately generating thresholds that are adapted to the business, improving the reliability and applicability of the thresholds, forming a standardized and traceable set of thresholds and version objects, ensuring stable and accurate quality management monitoring, effectively adapting to complex industry scenarios, and improving the effectiveness of control.

[0033] Optionally, in one embodiment of this application, obtaining target rule representation data for a target key metal industry quality management scenario includes: obtaining various quality management data for the target key metal industry quality management scenario, wherein the various quality management data include multi-source standard clause data, master data, graph basic data, historical testing data, and online testing data; performing preset field-level parsing operations on the multi-source standard clause data to extract corresponding multiple field information, wherein the multiple field information includes indicator name, judgment relationship, threshold or formula, scope of application, effective date, source level, and unit caliber; performing preset unit normalization and caliber verification on the multiple field information to obtain corresponding multiple rule information, and decomposing each rule information into the smallest executable rule atom, and constructing a set of rule atoms corresponding to multiple rule information; constructing a corresponding candidate rule set based on the set of rule atoms and combined with preset validity domain constraints; constructing a corresponding field mapping registry based on the multiple field information, and determining the target rule representation data based on the field mapping registry, the set of rule atoms, and the set of candidate rules.

[0034] It should be noted that, before performing standard clause parsing and rule atomization, the following data must be input in this application embodiment: 1. Multi-source standard clause data, including national standards, industry standards, enterprise standards, customer contract terms, testing specifications, and interim quality notices; 2. Master data and graph basic data, including entities such as products, batches, suppliers, processes, equipment, factories, regions, and testing method versions, and their relationships; 3. Historical testing data and online testing data.

[0035] Secondly, in the process of parsing standard clauses and atomizing rules, the embodiments of this application can first perform field-level parsing on multi-source standard clause data to extract various field information such as indicator name, judgment relationship, threshold or formula, scope of application, effective date, source level and unit caliber.

[0036] Furthermore, the embodiments of this application can perform unit normalization and caliber verification on various field information, such as converting expressions like ppm, mg / kg, and percentage content to a unified measurement system to obtain corresponding multiple rule information.

[0037] Subsequently, in this embodiment of the application, each rule can be broken down into the smallest executable rule atom, and denoted as... As shown in the following formula:

[0038] in, Indicates the target object; Indicates quality indicators; Indicates comparison symbols or interval relationships; Indicates the threshold; Indicates the unit; Represents the set of context constraints.

[0039] At the rule level, each rule The expression is:

[0040] in, It is a set of several regular atoms; For the domain of effectiveness constraints; This refers to the actions taken after a rule is triggered, such as issuing a warning, allowing passage, blocking, or escalating the review process.

[0041] Finally, the embodiments of this application can yield the corresponding set of rule atoms. Candidate rule set And the registry mapping for the terms-graph field-data source field.

[0042] Therefore, the embodiments of this application can transform quality clauses from different sources, with different structures and different definitions into a unified and executable rule representation through standard clause parsing and rule atomization operations, effectively reducing the rule access cost.

[0043] Optionally, in one embodiment of this application, determining the context slices and corresponding quality data stream information corresponding to the target rule representation data to perform preset robust quantile modeling, generating upper and lower bound thresholds corresponding to the context binding relationship, and performing a preset threshold reestimation operation based on the upper and lower bound thresholds to obtain the corresponding context binding threshold set and threshold version object, includes: obtaining multiple context slices and quality data stream information corresponding to the target rule representation data, determining the indicator sequence corresponding to each context slice, and constructing a corresponding fractional estimation function based on the quality data stream information, the preset target score point, and the learning rate; performing preset robust quantile modeling on the indicator sequence through the fractional estimation function to generate upper and lower bound thresholds corresponding to each context slice. The upper and lower thresholds corresponding to the context binding relationship and the context binding relationship are determined. Based on the context binding relationship and the upper and lower thresholds, the recent window distribution and the baseline window distribution corresponding to each context segment are determined, and the corresponding distribution drift metric is calculated according to the recent window distribution and the baseline window distribution. It is determined whether the distribution drift metric is greater than the preset context binding threshold. If the distribution drift metric is greater than the context binding threshold, a threshold re-estimation operation is performed to generate a new context binding threshold. It is determined whether the new context binding threshold meets the preset detection requirements within the preset grayscale observation period. If the new context binding threshold meets the preset detection requirements, a context binding threshold set is obtained, and the context binding threshold set and the context segment are bound to generate the corresponding threshold version object.

[0044] As one possible approach, embodiments of this application can obtain a subset of rules requiring dynamic thresholds from the candidate rule set, the context slices bound to the rules (e.g., "metal category = germanium, factory = factory A, process = refining, detection method version = v3, time window = last 90 days"), and the historical and real-time quality data streams under the corresponding contexts. Based on the online quantile update formula, each context slice is updated accordingly. The following index sequence Robust quantile modeling is performed to generate the context binding relationship and the upper and lower bound thresholds corresponding to each context slice. The online quantile update formula is as follows:

[0045] in, For a moment quantile estimates; The learning rate; The target quantile.

[0046] It should be noted that the above contextual segmentation can be determined based on at least one of the following contextual elements: metal category, product or steel grade, process route, production station, equipment, testing method version, operating condition range, and time window.

[0047] For example, taking the final sulfur content in the hot metal desulfurization process as the target quality indicator, the embodiments of this application can define "the steel grade is the target steel grade, the desulfurization station is the first desulfurization station, the detection method version is v3, the sulfur content before desulfurization is within a preset range, the inlet temperature is within a preset temperature range, and the time window is the last 90 days" as a context fragment. Context-compliant The final sulfur content of each furnace run is arranged according to the detection time, forming an index sequence under this context segment. .

[0048] Context binding relationships are used to represent context slices. The correspondence between the target quality indicators and their upper and lower thresholds. Specifically, this involves the relationship between real-time furnace steel grade, desulfurization station, detection method version, pre-desulfurization sulfur content, and inlet temperature, and the context information of the furnace segment. When a match is found, the upper limit threshold calculated based on the historical endpoint sulfur content index sequence under that context shard is invoked. and lower threshold It can perform quality judgment or abnormal warning on the final sulfur content of the real-time furnace batch.

[0049] In the embodiments of this application, the mathematical expressions for the upper and lower thresholds are:

[0050] in, The upper limit threshold; The lower limit threshold; This is a quantile estimation function; Indicates the target quantile above the upper threshold; The lower target quantiles represent the lower limit thresholds, both of which are located within the (0,1) interval.

[0051] It should be noted that, in order to determine whether the upper and lower thresholds need to be updated, embodiments of this application may introduce a distribution drift metric between the recent window and the baseline window. As shown in the following formula:

[0052] in, The Wasserstein distance; This represents the recent window distribution; The baseline window distribution is used.

[0053] when When the threshold (i.e., the context binding threshold) is reached, it indicates that the quality distribution of the context has drifted significantly, thus triggering threshold re-evaluation and the generation of a new threshold version.

[0054] In engineering implementation, the context threshold will not directly overwrite the old value, but will first enter the gray-scale observation period. If the new context threshold (i.e. the new context-bound threshold) can reduce the false alarm rate and not significantly increase the false negative rate during the gray-scale observation period, it will be fixed as the official version, and the context threshold will be bound to the context to generate the corresponding threshold version.

[0055] Therefore, the embodiments of this application can obtain a set of context binding thresholds. Define the threshold version object ThresholdSetVersion Its attributes include at least ,in, A unique identifier representing the threshold version object; This indicates the time when the threshold version takes effect; This indicates the expiration date of this threshold version; This indicates the identifier of the previous version that is being replaced by the current threshold version; This indicates the target threshold version identifier when a version rollback occurs.

[0056] Therefore, the embodiments of this application can automatically generate dynamic thresholds (i.e., context-bound thresholds) that are compatible with the current business context based on the distribution of historical quality data and real-time drift, and then solidify the versions, thereby reducing false alarms and false negatives caused by long-term reliance on fixed thresholds.

[0057] In step S102, based on the candidate rule set, context binding threshold set, and context segmentation of the target rule representation data, a corresponding rule conflict detection graph is constructed, and a preset priority adjudication operation is performed on the rule conflict detection graph to generate the effective rule set and effective rule set version corresponding to the context segment.

[0058] Furthermore, embodiments of this application may be based on a set of candidate rules. The rule instances that have been bound to dynamic thresholds (i.e., multiple context-bound thresholds or a set of context-bound thresholds) and the context shard set are used to construct the corresponding rule conflict detection graph and to make priority decisions on them, thereby generating the effective rule set and effective rule set version corresponding to the context shard.

[0059] Optionally, in one embodiment of this application, a corresponding rule conflict detection graph is constructed based on the candidate rule set, context binding threshold set, and context fragments of the target rule representation data. A preset priority adjudication operation is then performed on the rule conflict detection graph to generate a valid rule set and a valid rule set version corresponding to each context fragment. This includes: determining a subset of candidate rules for each context fragment based on the candidate rule set, and constructing a rule conflict detection graph based on the candidate rule subsets; calculating the rule source level score, timeliness score, scope of application specificity score, and historical execution reliability score for each rule in the candidate rule subset, and performing a weighted summation operation on the rule source level score, timeliness score, scope of application specificity score, and historical execution reliability score to obtain a priority score for each rule; determining whether a cycle exists in the rule conflict detection graph, and in response to the presence of a cycle, performing a preset minimum cost cycle breaking operation to generate a corresponding acyclic graph, and performing topological sorting on the remaining rules in the candidate rule subset corresponding to the acyclic graph to generate a valid rule set and a valid rule set version.

[0060] Specifically, for each context fragment Construct the corresponding subset of candidate rules As shown in the following formula:

[0061] in, This represents the subset of candidate rules corresponding to context segment c. The nth candidate rule in the list.

[0062] And establish a rule conflict detection graph corresponding to the subset of candidate rules, as shown in the following formula:

[0063] in, = , This represents a set of nodes, where each node represents a rule. This indicates a conflict between rules. If two rules intersect in their domain and cannot both be true simultaneously, a conflict edge is established, as shown in the following equation:

[0064] in, Indicates the type of conflict, which may include mutual exclusion conflict, overriding conflict, and circular dependency conflict.

[0065] Secondly, embodiments of this application can define a priority scoring function for each rule, as shown in the following formula:

[0066] in, Indicates the score based on the rule's source level; Indicates the timeliness score; The score indicates the specificity of the applicable scope. This indicates the historical execution credibility score; This indicates the score weighting coefficient of the rule source level; This represents the weighting coefficient for the timeliness score; This represents the score weighting coefficient for the specificity of the applicable scope; This represents the weighting coefficient of the historical execution credibility score. For conflicting rules, the one with higher priority is retained.

[0067] If a cycle exists in the rule conflict detection graph, a minimum-cost cycle-breaking strategy is adopted to remove the edge with the lowest conflict weight until an acyclic graph is formed. Then, the remaining rules corresponding to the acyclic graph are topologically sorted to generate valid rules, as shown in the following formula:

[0068] in, This represents a conflict resolution and adjudication function that automatically adjudicates based on the source, timeliness, and authority of the rules.

[0069] Therefore, the embodiments of this application can obtain an effective rule set under each context segment. Valid rule set version Meanwhile, during the process of constructing the rule conflict detection graph and prioritizing the decision, this application embodiment can record a conflict decision log, including conflict rule pairs, conflict types, priority scores, and decision results.

[0070] Therefore, the embodiments of this application construct a rule conflict detection graph and determine priority to automatically identify mutual exclusion, overriding and dependency conflicts between rules within the same context domain, thereby generating an executable and consistent rule set in a multi-standard parallel scenario.

[0071] In step S103, based on the target rule representation data, threshold version object, and effective rule set version, the online detection data in the target rule representation data is standardized into event objects, and a preset composite spatiotemporal rule matching operation is performed according to the event objects to generate corresponding evidence information. In addition, combined with the effective rule set version, threshold version object, and target rule representation data, a preset version governance and rollback operation is performed to obtain the corresponding key metal quality processing results.

[0072] Subsequently, embodiments of this application can standardize online detection data into event objects based on online detection data and graph base data in the target rule representation data, as well as threshold version objects and effective rule set versions, and perform composite spatiotemporal rule matching on them to generate corresponding evidence information (i.e. evidence packages), and perform pre-version governance and rollback operations to obtain the corresponding key metal quality processing results.

[0073] Optionally, in one embodiment of this application, based on the target rule representation data, threshold version object, and effective rule set version, the online detection data in the target rule representation data is standardized into event objects, and a preset composite spatiotemporal rule matching operation is performed on the event objects to generate corresponding evidence information. This includes: standardizing the online detection data into event objects based on the graph base data, threshold version object, and effective rule set version in the target rule representation data; obtaining the entity relationships corresponding to the graph base data, and performing a preset association extension operation on the event objects based on the entity relationships to obtain corresponding association extension data; performing a complex event matching operation on the association extension data based on a preset sliding time window and spatial range constraints to determine the corresponding matching conditions, and generating corresponding evidence information based on the matching conditions. The evidence information includes quality warning results, compliance judgment results, audit evidence packages, and rule hit logs.

[0074] Specifically, embodiments of this application can first input online detection data streams and batch event streams, entity relationships and contextual relationships in the knowledge graph, and threshold versions. and effective rule set version .

[0075] Secondly, embodiments of this application can standardize online detection data streams and batch event streams (i.e., online data) into event objects. :

[0076] in, Corresponding spectral entities such as batches, samples, and process records; Indicates spatial location; Indicates the business context; The measured value of a quality indicator; This represents the timestamp used to collect the event object.

[0077] Furthermore, embodiments of this application can leverage the entity relationships provided by a knowledge graph to extend the associations of event objects, thereby obtaining corresponding extended association data, as described below: batch supplier batch Process Process equipment factory area Test items Detection method version At the rule enforcement layer, embodiments of this application can use sliding time windows and spatial range constraints for complex event matching. For example, "a supplier within a time window..." At least one Taking "one batch is unqualified" as an example, the matching condition can be expressed as:

[0078] For cross-regional consistency anomalies, this can be further extended to:

[0079] in, This represents the i-th quality event object matched by the rule; This indicates a mapping from the event object to its corresponding supplier entity; Indicates the supplier entity; This represents the function for determining whether an event is unqualified. This represents a function to extract the event occurrence time. A counting operator that counts the number of event objects that meet a given condition; This represents the mapping from an event object to its corresponding domain function; Indicates the length of the sliding time window; , Indicates the start and end times of the statistical time interval; For a specified set of regions.

[0080] After a match is found, this embodiment of the application can output a judgment result and an evidence package, which includes: ,in, A unique identifier representing the hit rule. This represents the sequence of event objects that triggered the rule, thus forming an evidence output format of "rule version, threshold version, and hit event sequence".

[0081] Therefore, the embodiments of this application can obtain quality warning results, compliance judgment results, audit evidence packages, and rule hit logs that can be consumed by upper-layer systems.

[0082] Therefore, the embodiments of this application can support quality anomaly determination of complex event patterns such as time windows, spatial windows, continuous batches and cross-entity associations through spatiotemporal rule matching and evidence output operations. This enables the use of knowledge graphs to complete composite rule matching across batches, processes, regions and suppliers, thereby improving the ability to identify complex anomaly patterns.

[0083] Optionally, in one embodiment of this application, by combining the effective rule set version, the threshold version object, and the target rule representation data, a preset version governance and rollback operation is performed to obtain the corresponding key metal quality processing result. This includes: obtaining information on the changes in the structure of new version standard clauses or fields corresponding to the target key metal industry quality management scenario, and mapping the information on the changes in the structure of new version standard clauses or fields to the current graph of the target rule representation data, so as to calculate the literal similarity, unit compatibility, and contextual adaptation between the terminology information in the graph base data and the graph attributes; and calculating the comprehensive similarity between the terminology information and the graph attributes based on the literal similarity, unit compatibility, and contextual adaptation, and determining... The system checks whether the overall similarity is greater than a preset similarity threshold. If the overall similarity is greater than the threshold, a preset mode patching operation is performed to generate a new map mapping version. A preset gray-scale release operation is then performed on the valid rule set version, the threshold version object, and the new map mapping version. The system also checks whether the preset false alarm rate requirement is met during the gray-scale release operation. If the preset false alarm rate requirement is met, a preset version rollback and recalculation operation is performed on the valid rule set version, the threshold version object, and the new map mapping version to obtain the corresponding key metal quality processing results. The key metal quality processing results include version release results, version audit logs, rollback records, and historical recalculation results.

[0084] It should be noted that the embodiments of this application can first obtain a valid rule set version. Threshold version The terms-map fields-data source fields map to the registry, as well as information on changes in the structure of new version standard terms or fields.

[0085] When standard clauses are updated, this application embodiment needs to determine whether new terms, fields, and units can be mapped to the current graph schema, and define the comprehensive similarity between terms in the graph's basic data and graph attributes (props):

[0086] in, For literal similarity; Unit compatibility; For context adaptation; This is the weighting coefficient for literal similarity. This is a unit compatibility weighting coefficient; This is the context fit weighting coefficient. If ,in, Representation of terms The maximum value of the overall similarity between the current atlas and the attributes of each candidate atlas in the current atlas schema. If a preset similarity threshold is specified, a Schema Patch operation is triggered, performing at least one of the following operations: 1. Added new attribute fields; 2. Added new entity types; 3. Added new relationship types.

[0087] Meanwhile, this application embodiment can generate a new schema mapping version: .

[0088] Furthermore, this application embodiment can employ a canary release mechanism to collaboratively distribute the new effective rule set version, the new threshold version object, and the new schema mapping version. If an abnormally high false positive rate, unit conversion conflict, field interpretation failure, or incompatibility with old queries is detected during the runtime phase, a version rollback is triggered, including the rule version, threshold version, and schema mapping version, as shown in the following formula:

[0089] in, These represent the new effective rule set version, the threshold version object, and the graph mapping version, respectively. These represent the effective rule set version, threshold version object, and graph mapping version after rollback, respectively.

[0090] Furthermore, embodiments of this application can perform recalculation on specified historical data ranges based on older version rules. The rule set, thresholds, and schema mappings have been versioned and released in the draft; if false positives / missed positives or field changes cause anomalies, a rollback mechanism will be triggered.

[0091] Therefore, the embodiments of this application can obtain version release results, version audit logs, rollback records, and historical recalculation results.

[0092] Therefore, the embodiments of this application achieve versioned mapping (i.e., continuous rule operation), compatibility, issue rollback, and historical recalculation between rules and quality graphs through the collaborative governance of rule versions, threshold versions, and schema mapping versions, when standard iterations, schema changes, and field caliber changes occur.

[0093] In summary, the embodiments of this application introduce four supporting functions of a knowledge graph: it carries entity relationships between products, batches, suppliers, processes, equipment, factories, regions, and testing method versions; it determines relationship constraints based on rule context; it supports cross-entity associations in composite spatiotemporal rule matching; and it supports versioned mapping of standard clauses to schemas. This allows for the construction of a rule conflict detection graph (where nodes are rules, edges are conflict relationships, and the goal is to complete rule adjudication) to perform version governance and rollback of rules, thresholds, and schema mappings. This solves the problems of unified execution, continuous evolution, and audit rollback of multi-source quality rules.

[0094] According to the key metal quality processing method based on rule conflict detection graph proposed in this application, the method acquires target rule representation data of the target key metal industry quality management scenario, determines the context fragments and corresponding quality data flow information corresponding to the target rule representation data, performs preset robust quantile modeling, generates upper and lower limit thresholds corresponding to context binding relationships, and performs preset threshold re-estimation operation based on the upper and lower limit thresholds to obtain the corresponding context binding threshold set and threshold version object; based on the candidate rule set, context binding threshold set and context fragments of the target rule representation data, the corresponding rule conflict detection graph is constructed, and a preset priority adjudication operation is performed on the rule conflict detection graph to generate the effective rule set and effective rule set version corresponding to the context fragment; based on the target rule representation data, threshold version object and effective rule set version, the online detection data in the target rule representation data is standardized into event objects, and a preset composite spatiotemporal rule matching operation is performed based on the event objects to generate corresponding evidence information; and combined with the effective rule set version, threshold version object and target rule representation data, a preset version governance and rollback operation is performed to obtain the corresponding key metal quality processing result. This application uses knowledge graphs as the basis for organizing entity relationships and carrying context, and rule engines as the specific method, thereby achieving unified quality judgment and quality early warning under multi-source quality standards.

[0095] Secondly, the critical metal quality processing apparatus based on a rule conflict detection map proposed according to an embodiment of this application is described with reference to the accompanying drawings.

[0096] Figure 2 This is a block diagram of a key metal quality processing device based on a rule conflict detection graph, according to an embodiment of this application.

[0097] like Figure 2 As shown, the critical metal quality processing device 10 based on the rule conflict detection map includes: a threshold re-evaluation module 100, a priority adjudication module 200, and a quality processing module 300.

[0098] Among them, the threshold re-estimation module 100 is used to obtain the target rule representation data of the target key metal industry quality management scenario, and determine the context fragment and the corresponding quality data flow information corresponding to the target rule representation data, so as to perform preset robust quantile modeling, generate the upper and lower limit thresholds corresponding to the context binding relationship, and perform preset threshold re-estimation operation based on the upper and lower limit thresholds to obtain the corresponding context binding threshold set and threshold version object. The priority adjudication module 200 is used to construct a corresponding rule conflict detection graph based on the candidate rule set, context binding threshold set, and context segmentation of the target rule representation data, and to perform a preset priority adjudication operation on the rule conflict detection graph to generate the effective rule set and effective rule set version corresponding to the context segmentation. The quality processing module 300 is used to standardize the online detection data in the target rule representation data into event objects based on the target rule representation data, threshold version objects, and effective rule set versions. It then performs a preset composite spatiotemporal rule matching operation based on the event objects to generate corresponding evidence information. In addition, it combines the effective rule set version, threshold version objects, and target rule representation data to perform preset version governance and rollback operations to obtain the corresponding key metal quality processing results.

[0099] Optionally, in one embodiment of this application, the threshold re-estimation module 100 includes: a first acquisition unit, a parsing unit, a disassembly unit, a first construction unit, and a determination unit.

[0100] The first acquisition unit is used to acquire various quality management data in the target key metal industry quality management scenario. These various quality management data include multi-source standard clause data, master data, graph basic data, historical testing data, and online testing data.

[0101] The parsing unit is used to perform preset field-level parsing operations on multi-source standard clause data to extract various field information, including indicator name, judgment relationship, threshold or formula, scope of application, effective date, source level and unit scope.

[0102] The decomposition unit is used to normalize and validate multiple fields of information using preset units to obtain multiple rule information, and to decompose each rule information into the smallest executable rule atom, and to construct a set of rule atoms corresponding to multiple rule information.

[0103] The first building unit is used to construct a corresponding set of candidate rules based on the set of rule atoms and in combination with preset validity domain constraints.

[0104] The determination unit is used to construct a corresponding field mapping registry based on multiple field information, so as to determine the target rule representation data based on the field mapping registry, the rule atom set, and the candidate rule set.

[0105] Optionally, in one embodiment of this application, the threshold re-evaluation module 100 further includes: a second acquisition unit, a generation unit, a first calculation unit, a first judgment unit, and a second judgment unit.

[0106] The second acquisition unit is used to acquire multiple context slices corresponding to the target rule representation data and the quality data stream information corresponding to the multiple context slices, determine the index sequence corresponding to each context slice, and construct the corresponding score estimation function based on the quality data stream information, the preset target score points and the learning rate.

[0107] The generation unit is used to perform a preset robust quantile model on the index sequence through a quantile estimation function, so as to generate the context binding relationship and the upper and lower limit thresholds corresponding to each context segment.

[0108] The first calculation unit is used to determine the recent window distribution and the baseline window distribution corresponding to each context slice based on the context binding relationship and upper and lower thresholds, and to calculate the corresponding distribution drift metric based on the recent window distribution and the baseline window distribution.

[0109] The first judgment unit is used to determine whether the distribution drift metric is greater than the preset context binding threshold. In response to the distribution drift metric being greater than the context binding threshold, a threshold re-evaluation operation is performed to generate a new context binding threshold.

[0110] The second judgment unit is used to determine whether the new context binding threshold meets the preset detection requirements within the preset grayscale observation period. In response to the new context binding threshold meeting the preset detection requirements, a context binding threshold set is obtained, and the context binding threshold set and context fragment are bound together to generate the corresponding threshold version object.

[0111] Optionally, in one embodiment of this application, the priority decision module 200 includes: a second construction unit, a weighted summation unit, and a sorting unit.

[0112] The second construction unit is used to determine a subset of candidate rules for each context segment based on the set of candidate rules, and to construct a rule conflict detection graph based on the subset of candidate rules.

[0113] The weighted summation unit is used to calculate the rule source level score, timeliness score, scope of application specificity score, and historical execution credibility score for each rule in the candidate rule subset, and to perform a weighted summation operation on the rule source level score, timeliness score, scope of application specificity score, and historical execution credibility score to obtain the priority score corresponding to each rule.

[0114] The sorting unit is used to determine whether there is a cycle in the rule conflict detection graph. In response to the presence of a cycle in the rule conflict detection graph, it performs a preset minimum cost cycle breaking operation to generate the corresponding acyclic graph. Then, it performs topological sorting on the remaining rules in the candidate rule subset corresponding to the acyclic graph to generate a valid rule set and a version of the valid rule set.

[0115] Optionally, in one embodiment of this application, the quality processing module 300 includes: a standardization unit, an expansion unit, and a matching unit.

[0116] The standardization unit is used to standardize online detection data into event objects based on the graph base data, threshold version objects, and effective rule set versions in the target rule representation data.

[0117] The extension unit is used to obtain the entity relationships corresponding to the basic data of the graph, and to perform preset association extension operations on the event objects according to the entity relationships to obtain the corresponding association extension data.

[0118] The matching unit is used to perform complex event matching operations on associated extended data based on preset sliding time windows and spatial range constraints to determine the corresponding matching conditions and generate corresponding evidence information based on the matching conditions. The evidence information includes quality warning results, compliance judgment results, audit evidence packages, and rule hit logs.

[0119] Optionally, in one embodiment of this application, the quality processing module 300 further includes: a mapping unit, a second calculation unit, a detection unit, and a rollback unit.

[0120] The mapping unit is used to obtain information on the changes in the structure of new version standard clauses or fields corresponding to the quality management scenario of the target key metal industry, and to map the information on the changes in the structure of new version standard clauses or fields to the current graph of the target rule representation data, so as to calculate the literal similarity, unit compatibility and contextual adaptation between the terminology information in the graph base data and the graph attributes.

[0121] The second calculation unit is used to calculate the comprehensive similarity between term information and graph attributes based on literal similarity, unit compatibility and contextual adaptation, and to determine whether the comprehensive similarity is greater than a preset similarity threshold.

[0122] The detection unit is used to perform a preset mode patching operation in response to the case where the overall similarity is greater than the similarity threshold, so as to generate a new map mapping version, and to perform a preset gray-scale release operation on the effective rule set version, the threshold version object and the new map mapping version, and to detect whether the preset false alarm rate requirement is met during the preset gray-scale release operation.

[0123] The rollback unit is used to perform preset version rollback and recalculation operations on the effective rule set version, threshold version object and new map mapping version when the preset false alarm rate requirement is met, so as to obtain the corresponding key metal quality processing results. The key metal quality processing results include version release results, version audit logs, rollback records and historical recalculation results.

[0124] It should be noted that the foregoing explanation of the embodiment of the critical metal quality processing method based on the rule conflict detection map also applies to the critical metal quality processing device based on the rule conflict detection map in this embodiment, and will not be repeated here.

[0125] The critical metal quality processing device based on rule conflict detection graph proposed in this application includes a threshold re-estimation module 100, used to acquire target rule representation data of the target critical metal industry quality management scenario, and determine the context fragments and corresponding quality data stream information corresponding to the target rule representation data, so as to perform preset robust quantile modeling, generate upper and lower limit thresholds corresponding to the context binding relationship, and perform preset threshold re-estimation operation based on the upper and lower limit thresholds to obtain the corresponding context binding threshold set and threshold version object; and a priority adjudication module 200, used to perform priority adjudication based on the candidate rule set, context binding threshold set and upper limit threshold of the target rule representation data. The following steps involve segmenting the context, constructing corresponding rule conflict detection graphs, and performing preset priority adjudication operations on these graphs to generate valid rule sets and versions corresponding to the context segments. The quality processing module 300, based on the target rule representation data, threshold version objects, and valid rule set versions, standardizes the online detection data in the target rule representation data into event objects. It then performs preset composite spatiotemporal rule matching operations based on the event objects to generate corresponding evidence information. Finally, combining the valid rule set versions, threshold version objects, and target rule representation data, it executes preset version governance and rollback operations to obtain the corresponding key metal quality processing results. This application uses a knowledge graph as the foundation for entity relationship organization and context carrying, and a rule engine as the specific method, thereby achieving unified quality judgment and quality early warning under multi-source quality standards.

[0126] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device may include: The memory 301, the processor 302, and the computer program stored on the memory 301 and capable of running on the processor 302.

[0127] When the processor 302 executes the program, it implements the key metal quality processing method based on the rule conflict detection graph provided in the above embodiments.

[0128] Furthermore, electronic devices also include: Communication interface 303 is used for communication between memory 301 and processor 302.

[0129] The memory 301 is used to store computer programs that can run on the processor 302.

[0130] The memory 301 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0131] If the memory 301, processor 302, and communication interface 303 are implemented independently, then the communication interface 303, memory 301, and processor 302 can be interconnected via a bus to complete communication between them. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0132] Optionally, in a specific implementation, if the memory 301, processor 302, and communication interface 303 are integrated on a single chip, then the memory 301, processor 302, and communication interface 303 can communicate with each other through an internal interface.

[0133] Processor 302 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application.

[0134] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described key metal quality processing method based on a rule-based conflict detection graph.

[0135] This application also provides a computer program product, including a computer program, which, when executed, is used to implement the above-described key metal quality processing method based on a rule conflict detection graph.

[0136] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0137] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0138] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0139] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0140] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0141] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0142] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0143] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for processing critical metal quality based on a rule-based conflict detection map, characterized in that, Includes the following steps: Acquire target rule representation data for the quality management scenario of the target key metal industry, and determine the context shards and corresponding quality data stream information corresponding to the target rule representation data, so as to perform preset robust quantile modeling, generate upper and lower limit thresholds corresponding to the context binding relationship, and perform preset threshold re-estimation operation based on the upper and lower limit thresholds to obtain the corresponding context binding threshold set and threshold version object. Based on the candidate rule set of the target rule representation data, the context binding threshold set, and the context segment, a corresponding rule conflict detection graph is constructed, and a preset priority adjudication operation is performed on the rule conflict detection graph to generate the effective rule set and effective rule set version corresponding to the context segment; Based on the target rule representation data, the threshold version object, and the effective rule set version, the online detection data in the target rule representation data is standardized into event objects, and a preset composite spatiotemporal rule matching operation is performed according to the event objects to generate corresponding evidence information. In addition, combined with the effective rule set version, the threshold version object, and the target rule representation data, a preset version governance and rollback operation is performed to obtain the corresponding key metal quality processing results.

2. The method for processing key metal quality based on rule-based conflict detection maps according to claim 1, characterized in that, The target rule representation data for the quality management scenario of the target key metal industry includes: Acquire various quality management data for the target key metal industry quality management scenario, including multi-source standard clause data, master data, graph basic data, historical testing data, and online testing data; The multi-source standard clause data is subjected to a preset field-level parsing operation to extract various field information, including indicator name, judgment relationship, threshold or formula, scope of application, effective date, source level and unit scope. The various field information is normalized to a preset unit and validated to obtain multiple corresponding rule information. Each rule information is then decomposed into the smallest executable rule atom, and a set of rule atoms corresponding to the multiple rule information is constructed. Based on the set of rule atoms and combined with the preset domain of effectiveness constraints, a corresponding set of candidate rules is constructed. A corresponding field mapping registry is constructed based on the various field information, and the target rule representation data is determined based on the field mapping registry, the set of rule atoms, and the set of candidate rules.

3. The method for processing key metal quality based on rule-based conflict detection maps according to claim 2, characterized in that, The process of determining the context shards and corresponding quality data stream information corresponding to the target rule representation data, performing preset robust quantile modeling, generating upper and lower bound thresholds corresponding to the context binding relationship, and performing a preset threshold re-estimation operation based on the upper and lower bound thresholds to obtain the corresponding context binding threshold set and threshold version object includes: Obtain multiple context segments corresponding to the target rule representation data and the quality data stream information corresponding to the multiple context segments, determine the indicator sequence corresponding to each context segment, and construct the corresponding score estimation function based on the quality data stream information, the preset target score points and the learning rate; The index sequence is modeled using a preset robust quantile model through the fractional estimation function to generate the context binding relationship corresponding to each context slice and the upper and lower limit thresholds corresponding to the context binding relationship; Based on the context binding relationship and the upper and lower thresholds, the recent window distribution and the baseline window distribution corresponding to each context slice are determined, and the corresponding distribution drift metric is calculated based on the recent window distribution and the baseline window distribution. Determine whether the distribution drift metric is greater than a preset context binding threshold. If the distribution drift metric is greater than the context binding threshold, perform a threshold re-evaluation operation to generate a new context binding threshold. Determine whether the new context binding threshold meets the preset detection requirements within the preset grayscale observation period. In response to the new context binding threshold meeting the preset detection requirements, obtain the context binding threshold set and bind the context binding threshold set and the context fragment to generate the corresponding threshold version object.

4. The critical metal quality processing method based on rule conflict detection map according to claim 3, characterized in that, The process of constructing a corresponding rule conflict detection graph based on the candidate rule set of the target rule representation data, the context binding threshold set, and the context segmentation, and performing a preset priority adjudication operation on the rule conflict detection graph to generate a valid rule set and a valid rule set version corresponding to the context segment, includes: Based on the candidate rule set, a subset of candidate rules for each context segment is determined, and the rule conflict detection graph is constructed according to the subset of candidate rules. Calculate the rule source level score, timeliness score, scope of application specificity score, and historical execution reliability score for each rule in the candidate rule subset, and perform a weighted summation operation on the rule source level score, the timeliness score, the scope of application specificity score, and the historical execution reliability score to obtain the priority score corresponding to each rule; Determine whether a cycle exists in the rule conflict detection graph. If a cycle exists in the rule conflict detection graph, perform a preset minimum cost cycle breaking operation to generate a corresponding acyclic graph. Then, perform topological sorting on the remaining rules in the candidate rule subset corresponding to the acyclic graph to generate the effective rule set and the version of the effective rule set.

5. The method for processing key metal quality based on rule-based conflict detection maps according to claim 4, characterized in that, Based on the target rule representation data, the threshold version object, and the effective rule set version, the online detection data in the target rule representation data is standardized into event objects, and a preset composite spatiotemporal rule matching operation is performed on the event objects to generate corresponding evidence information, including: Based on the graph base data in the target rule representation data, the threshold version object, and the effective rule set version, the online detection data is standardized into event objects; Obtain the entity relationships corresponding to the basic data of the graph, and perform a preset association extension operation on the event object based on the entity relationships to obtain the corresponding association extension data; Based on a preset sliding time window and spatial range constraints, a complex event matching operation is performed on the associated extended data to determine the corresponding matching conditions, and corresponding evidence information is generated according to the matching conditions. The evidence information includes quality warning results, compliance judgment results, audit evidence packages, and rule hit logs.

6. The method for processing key metal quality based on rule-based conflict detection maps according to claim 5, characterized in that, The step of combining the effective rule set version, the threshold version object, and the target rule representation data to perform preset version governance and rollback operations to obtain the corresponding key metal quality processing results includes: Obtain information on the new version of standard clauses or field structure changes corresponding to the target key metal industry quality management scenario, and map the new version of standard clauses or field structure changes to the current graph of the target rule representation data, so as to calculate the literal similarity, unit compatibility and contextual adaptation between the terminology information and graph attributes in the graph base data; Based on the literal similarity, the unit compatibility, and the contextual adaptation, calculate the comprehensive similarity between the terminology information and the graph attributes, and determine whether the comprehensive similarity is greater than a preset similarity threshold; In response to the overall similarity being greater than the similarity threshold, a preset mode patching operation is performed to generate a new map mapping version, and a preset gray-scale release operation is performed on the effective rule set version, the threshold version object, and the new map mapping version, and it is detected whether the preset false alarm rate requirement is met during the preset gray-scale release operation. If the preset false alarm rate requirement is met, preset version rollback and recalculation operations are performed on the effective rule set version, the threshold version object, and the new map mapping version to obtain the corresponding key metal quality processing results. The key metal quality processing results include version release results, version audit logs, rollback records, and historical recalculation results.

7. A critical metal quality processing device based on a rule-based conflict detection map, characterized in that, include: The threshold re-estimation module is used to acquire target rule representation data of the target key metal industry quality management scenario, and determine the context fragment and corresponding quality data stream information corresponding to the target rule representation data, so as to perform preset robust quantile modeling, generate upper and lower limit thresholds corresponding to the context binding relationship, and perform preset threshold re-estimation operation according to the upper and lower limit thresholds to obtain the corresponding context binding threshold set and threshold version object. The priority adjudication module is used to construct a corresponding rule conflict detection graph based on the candidate rule set of the target rule representation data, the context binding threshold set, and the context segment, and to perform a preset priority adjudication operation on the rule conflict detection graph to generate a valid rule set and a valid rule set version corresponding to the context segment; The quality processing module is used to standardize the online detection data in the target rule representation data into event objects based on the target rule representation data, the threshold version object, and the effective rule set version, and to perform a preset composite spatiotemporal rule matching operation based on the event objects to generate corresponding evidence information. In addition, it combines the effective rule set version, the threshold version object, and the target rule representation data to perform preset version governance and rollback operations to obtain the corresponding key metal quality processing results.

8. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the program to implement the critical metal quality processing method based on a rule-based conflict detection graph as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the critical metal quality processing method based on the rule conflict detection map as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, The computer program is executed to implement the critical metal quality processing method based on the rule conflict detection map as described in any one of claims 1-6.

Images