A standard system applicability evaluation method and system based on multi-source data

Abstract

The application discloses a kind of standard system applicability evaluation method and system based on multi-source data, it is related to standardization field, including, the historical operation data of acquisition equipment, equipment is divided into experimental group equipment group and control group equipment group;The historical operation data of equipment is preprocessed, and the experimental group equipment group and control group equipment group are respectively constructed to reflect the performance index numerical sequence of target standard for technical target achievement situation;Based on the performance index numerical sequence of experimental group equipment group and control group equipment group, the net effect value and statistical confidence of target standard are calculated using double difference model, and unit type and geographical and climatic zone are added to double difference model as fixed effect control variable.The net effect value and intermediate centrality score of the present application are fused into a comprehensive index by correction, realizing the multi-dimensional dynamic evaluation of data-driven, causal identification and knowledge value combination of standard applicability.

Description

Technical Field

[0001] This invention relates to the field of standardization, and in particular to a method and system for evaluating the applicability of a standard system based on multi-source data. Background Technology

[0002] In the field of standardization, scientific evaluation of the applicability of standard systems is key to supporting technological progress and industrial development. Traditional evaluation methods mainly rely on consensus-building mechanisms such as expert seminars, the Delphi method, or questionnaires. By gathering expert experience, the technical advancement, market compliance, and feasibility of standards can be assessed in a qualitative or semi-quantitative manner. In the absence of large-scale objective data, these methods have certain reference value and can utilize tacit knowledge that is difficult to quantify directly.

[0003] With the acceleration of technological iteration and the increasing complexity of application environments, the aforementioned methods relying on subjective judgment have obvious limitations. They are difficult to quantify and accurately assess the causal effects of standards in real industrial scenarios, and cannot effectively eliminate the influence of confounding factors such as equipment attributes, regional differences, and macro-environment. Therefore, they cannot accurately answer how significant the actual effects of standard implementation are. The lack of causal identification capabilities makes the evaluation results susceptible to bias and interference, making it difficult to provide reliable data support for the applicability of standards, thereby affecting the scientific nature of subsequent decisions. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a standard system applicability evaluation method based on multi-source data to solve the problem of insufficient reliability of evaluation conclusions caused by the inability of traditional methods to effectively identify causal relationships.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a standard system applicability evaluation method based on multi-source data, which includes collecting historical operating data of equipment and dividing the equipment into an experimental group and a control group.

[0008] The historical operating data of the equipment was preprocessed, and numerical sequences of performance indicators that quantitatively reflect the achievement of the technical objectives were constructed for the experimental group and the control group of equipment, respectively.

[0009] Based on the performance index numerical series of the experimental group equipment group and the control group equipment group, the net effect value and statistical confidence level of the target standard are calculated using a difference-in-differences model. The unit model and geographical climate zone are added as fixed effect control variables to the difference-in-differences model to control the potential bias caused by the unit model and geographical environment attributes to the estimation of the net effect value.

[0010] A domain knowledge graph is constructed based on domain knowledge text data, and the betweenness centrality score of the target standard in the knowledge graph is calculated.

[0011] When the middle centrality score of the target standard exceeds the judgment threshold set by statistical quantile analysis based on historical operating data, and the net effect value fails the statistical confidence test, the associated entities not included in the difference-in-differences model are extracted from the knowledge graph as omitted variables. The net effect value is then corrected by recalculating the difference-in-differences model to obtain the comprehensive index number for evaluating the applicability of the target standard.

[0012] As a preferred embodiment of the standard system applicability evaluation method based on multi-source data described in this invention, the following steps are included: collecting historical operating data of the equipment and dividing the equipment into an experimental group and a control group:

[0013] The device's historical operating data is constructed by collecting device identifiers, timestamps, and operating parameters from data sources.

[0014] The device identifiers are matched with the adoption records of the target standards. The adoption records of the target standards record the time points when the device identifiers were adopted and the target standards were adopted. A list of devices marked with the target standards is generated. Based on the adoption marks in the device list, the set of device identifiers is divided into an experimental group of devices that adopted the target standards and a candidate set of a control group of devices that did not adopt the target standards.

[0015] Based on the candidate set of the control group equipment group, the unit model, commissioning year, geographical climate zone and annual average load rate attributes extracted from the historical operation data of the equipment are matched with the experimental group equipment group to form a control group equipment group that matches the experimental group equipment group in terms of the aforementioned attributes.

[0016] As a preferred embodiment of the standard system applicability evaluation method based on multi-source data described in this invention, the method includes: preprocessing the historical operating data of the equipment, and constructing numerical sequences of performance indicators that quantitatively reflect the achievement of the target standard's technical objectives for both the experimental group and the control group of equipment, including the following steps:

[0017] The historical operating data of the equipment is cleaned to generate clean historical operating data; the clean historical operating data is then combined with the equipment maintenance records according to equipment identification and timestamp.

[0018] Alignment is performed to create unified time-series data of equipment identifiers, timestamps, operating parameters, and maintenance records;

[0019] Based on the target criteria and targeting the technical objectives, performance indicators are defined and calculated from unified time series data.

[0020] Numerical value;

[0021] For the experimental group of equipment, extract the performance of the experimental group of equipment from the uniform time series data.

[0022] The indicators are arranged in chronological order to construct a numerical sequence of performance indicators for the experimental group of equipment.

[0023] For the control group of equipment, the performance index values ​​of the control group of equipment were extracted from the unified time series data and arranged in chronological order to construct the performance index value sequence of the control group of equipment.

[0024] As a preferred embodiment of the standard system applicability evaluation method based on multi-source data described in this invention, the method includes the following steps: Based on the performance index numerical sequences of the experimental group equipment group and the control group equipment group, a difference-in-differences model is used to calculate the net effect value and statistical confidence level of the target standard;

[0025] The performance index numerical sequences of the experimental group equipment group and the performance index numerical sequences of the control group equipment group were integrated to form a panel dataset.

[0026] Based on panel datasets, a difference-in-differences model was designed, with the performance index values ​​as the dependent variable and the interaction term between group labels and time points as the explanatory variable.

[0027] In the difference-in-differences model, the unit type and geographical climate zone are loaded as fixed effect control variables. The least squares method is used to fit the difference-in-differences model and calculate the coefficient estimates of the interaction term between the group label and the time point to form the net effect value of the target standard. Then, the standard error of the coefficient of the interaction term between the group label and the time point is calculated. Based on the standard error and the coefficient estimates, the statistic is derived to obtain the probability value of the net effect value of the target standard.

[0028] The probability value of the net effect value of the target standard is compared with the level threshold set by the variation analysis of the performance indicator numerical series of the control group of equipment in the historical operation data of the equipment before the adoption of the target standard, to form the statistical confidence level of the net effect value of the target standard.

[0029] As a preferred embodiment of the standard system applicability evaluation method based on multi-source data described in this invention, the method includes: incorporating the generator unit model and geographical climate zone as fixed-effect control variables into the difference-in-differences model to control for potential biases caused by the generator unit model and geographical environment attributes in the estimation of net effect values, comprising the following steps:

[0030] Extract the unit model attribute and geographic climate zone attribute from the panel dataset, encode the unit model attribute and geographic climate zone attribute as classification variables, and generate a dummy variable set for the unit model attribute and a dummy variable set for the geographic climate zone attribute.

[0031] The set of dummy variables for unit model attributes and the set of dummy variables for geographical climate zone attributes are added as fixed-effect control variables to the design matrix of the difference-in-differences model.

[0032] A matrix-fit difference-in-differences model with fixed-effects control variables is used to estimate the net effect value of the target standard by controlling the influence of unit type attribute and geographic climate zone attribute, and to control the potential bias caused by unit type attribute and geographic climate zone attribute on the estimation of net effect value.

[0033] As a preferred embodiment of the standard system applicability evaluation method based on multi-source data described in this invention, the following steps are included: constructing a domain knowledge graph based on domain knowledge text data and calculating the betweenness centrality score of the target standard in the knowledge graph:

[0034] Named entity recognition is performed on domain knowledge text data to identify and extract standard clause entities from target standard text, fault mode entities from equipment failure case library, and solution entities from maintenance solution manual.

[0035] Relationships are extracted from domain knowledge text data to establish diagnostic relationships between standard clause entities and failure mode entities, and disposal relationships between failure mode entities and solution entities.

[0036] Based on standard clause entities, failure mode entities, solution entities, diagnostic relationships, and handling relationships, a domain knowledge graph is constructed with entities as nodes and relationships as edges.

[0037] On the domain knowledge graph, the proportion of target standard nodes appearing on the shortest path from failure mode nodes to solution nodes is calculated to obtain the betweenness centrality score of the target standard in the knowledge graph.

[0038] As a preferred embodiment of the standard system applicability evaluation method based on multi-source data described in this invention, the method includes the following steps: when the middleness centrality score of the target standard exceeds the judgment threshold set by statistical quantile analysis based on historical operating data, and the net effect value fails the statistical confidence test, related entities not included in the difference-in-differences model are extracted from the knowledge graph as omitted variables. The net effect value is then corrected by recalculating the difference-in-differences model to obtain a comprehensive index for evaluating the applicability of the target standard.

[0039] The middleness centrality score of the target standard is compared with the judgment threshold set by statistical quantile analysis based on historical operating data. The probability value of the net effect value of the target standard is compared with the level threshold. If the probability value of the net effect value of the target standard is greater than the level threshold, the net effect value of the target standard fails the statistical confidence test.

[0040] When the middle centrality score of the target standard exceeds the judgment threshold and the net effect value of the target standard fails the statistical confidence test, the associated entities that are not included as control variables in the difference-in-differences model are extracted from the domain knowledge graph and are related to the nodes of the target standard, forming a set of potential omitted variables.

[0041] The quantifiable observational data corresponding to the potential missing variable set are added as new control variables to the difference-in-differences model;

[0042] The calculations were re-performed using a difference-in-differences model with new control variables to obtain the revised coefficient estimates, which formed the net effect value of the revised target standard.

[0043] The net effect value of the modified target standard is standardized and weighted and fused with the middleness centrality score of the target standard in the knowledge graph to calculate the comprehensive index number for evaluating the applicability of the target standard.

[0044] Secondly, the present invention provides a standard system applicability evaluation system based on multi-source data, including a data acquisition module for acquiring historical operating data of equipment and dividing the equipment into an experimental group and a control group.

[0045] The preprocessing module preprocesses the historical operating data of the equipment and constructs numerical sequences of performance indicators that quantitatively reflect the achievement of the target standard for the technical objectives for both the experimental group and the control group.

[0046] The bias module, based on the performance index numerical series of the experimental group equipment group and the control group equipment group, uses a difference-in-differences model to calculate the net effect value and statistical confidence of the target standard. The unit model and geographical climate zone are added as fixed effect control variables to the difference-in-differences model to control the potential bias caused by the unit model and geographical environment attributes to the estimation of the net effect value.

[0047] The module constructs a domain knowledge graph based on domain knowledge text data and calculates the betweenness centrality score of the target standard in the knowledge graph.

[0048] In the judgment module, when the middleness centrality score of the target standard exceeds the judgment threshold set by statistical quantile analysis based on historical operating data, and the net effect value fails the statistical confidence test, the associated entities not included in the difference-in-differences model are extracted from the knowledge graph as omitted variables. The net effect value is then corrected by recalculating the difference-in-differences model to obtain the comprehensive index number for evaluating the applicability of the target standard.

[0049] Thirdly, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the standard system applicability evaluation method based on multi-source data as described in the first aspect of the present invention.

[0050] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the standard system applicability evaluation method based on multi-source data as described in the first aspect of the present invention.

[0051] The beneficial effects of this invention are as follows: By collecting historical operating data of equipment and dividing the experimental group and matched control group according to the adoption of standards, a performance index sequence reflecting technical objectives is constructed. A difference-in-differences model is adopted and fixed effects such as unit model and geographical climate zone are controlled to calculate the net effect value and statistical confidence of the target standard, so as to quantify the causal effect of standard implementation. A knowledge graph is constructed based on domain knowledge text to calculate the middleness centrality score of the target standard and evaluate its hub value in the knowledge network. When the aforementioned analysis shows that the standard has high knowledge value but the statistical effect does not meet the confidence requirements, potential omitted variables are automatically extracted from the knowledge graph, the difference-in-differences model is recalculated to correct the net effect value, and the corrected net effect value and middleness centrality score are integrated into a comprehensive index to realize a multi-dimensional dynamic evaluation of the applicability of the standard that combines data-driven, causal identification and knowledge value. Attached Figure Description

[0052] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a flowchart of a standard system applicability evaluation method based on multi-source data.

[0054] Figure 2 This is a schematic diagram of a standard system applicability evaluation system based on multi-source data.

[0055] Figure 3 This is a flowchart of the calculation process for the difference-in-differences model.

[0056] Figure 4 A flowchart for constructing a domain knowledge graph and calculating the betweenness centrality score. Detailed Implementation

[0057] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0058] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0059] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0060] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a method for evaluating the applicability of a standard system based on multi-source data, comprising the following steps:

[0061] S1. Collect historical operating data of the equipment and divide the equipment into experimental group and control group.

[0062] S1.1 Collect device identifiers, timestamps, and operating parameters from the data source to form the device's historical operating data.

[0063] Furthermore, data records containing equipment identifiers, timestamps, and operating parameters are collected from the production monitoring unit, equipment management unit, and log database. Records from different data sources are associated and merged based on equipment identifiers and timestamps, duplicates are removed, and inconsistent formats are handled to form structured historical operating data of the equipment.

[0064] S1.2 Match the device identifier with the target standard adoption record. The target standard adoption record records the time point of the device identifier and the target standard adoption. Generate a device list marked with the target standard adoption. According to the adoption mark in the device list, divide the device identifier set into the experimental group of devices that adopted the target standard and the candidate set of the control group of devices that did not adopt the target standard.

[0065] Furthermore, the database of adoption records for the target standard is read, which clearly records the device identifiers and the specific time points when the target standard was adopted. The set of device identifiers in the historical operation data of the devices is matched with the device identifiers in the adoption records of the target standard. Each successfully matched device identifier is marked with a target standard adoption tag, generating a list of device identifiers and adoption statuses. Based on the adoption tags in the device list, the set of device identifiers is divided into an experimental group of devices that have adopted the target standard and a candidate set of a control group of devices that have not adopted the target standard.

[0066] S1.3. Based on the candidate set of the control group equipment group, the unit model, commissioning year, geographical climate zone and annual average load rate attributes extracted from the historical operation data of the equipment are matched with the experimental group equipment group to form a control group equipment group that matches the experimental group equipment group in the aforementioned attributes.

[0067] Furthermore, from the historical operating data of the equipment, the unit model, commissioning year, geographical climate zone, and annual average load rate attributes of each piece of equipment in the candidate set of the control group equipment group are extracted, as well as the unit model, commissioning year, geographical climate zone, and annual average load rate attributes of each piece of equipment in the experimental group equipment group. Using the propensity score matching method, a subset of equipment that is closest to the experimental group equipment group in terms of unit model, commissioning year, geographical climate zone, and annual average load rate attributes is selected from the candidate set of the control group equipment group, forming a control group equipment group that matches the experimental group equipment group in the above attributes.

[0068] S2. Preprocess the historical operating data of the equipment, and construct numerical sequences of performance indicators that quantitatively reflect the achievement of the technical objectives by the target standard for both the experimental group and the control group.

[0069] S2.1 Clean the historical operating data of the equipment to generate clean historical operating data of the equipment.

[0070] Furthermore, a data cleaning process is performed on the equipment's historical operating data, including identifying and processing missing values, detecting and eliminating abnormal data points that are outside the reasonable range, and correcting data format inconsistencies to generate clean historical operating data for the equipment.

[0071] S2.2 Align the historical operating data of the clean equipment with the equipment maintenance record data according to the equipment identifier and timestamp to form a unified time series data of equipment identifier, timestamp, operating parameters and maintenance records.

[0072] Furthermore, the historical operating data of the cleanroom equipment is compared with the data collected from the equipment management database.

[0073] The maintenance record data is precisely correlated and aligned based on the equipment identifier and timestamp, forming a complete and synchronized unified time-series data that includes equipment identifier, timestamp, operating parameters, and maintenance record fields.

[0074] S2.3. Based on target standards and technical objectives, define and calculate performance from unified time series data.

[0075] Performance index values.

[0076] The numerical expression for performance indicators is:

[0077] ;

[0078] in, In time Next The performance indicators of the equipment. For the first The equipment is in

[0079] time The set of operating parameters, Because the time window is long, For time indexing, For the first Maintenance record data of the equipment, It is a unique identifier for the device.

[0080] Furthermore, based on the specific technical objectives targeted by the standard, the calculation logic for performance indicator values ​​is defined from unified time series data; performance indicator values The specific calculation process is to use the equipment In time Set of running parameters With the device in the time window Internal maintenance record data Both are used as input, through a mapping function Perform calculations; mapping function The specific form is determined by the technical objective. When the technical objective is to reduce the failure rate, the mapping function... It can be defined as a statistical time window Internal equipment The number of failures.

[0081] S2.4. For the experimental group of equipment, extract the experimental group of equipment from the unified time series data.

[0082] The performance indicators were determined and arranged in chronological order to construct a numerical sequence of performance indicators for the experimental group of equipment.

[0083] Furthermore, based on the list of equipment identifiers included in the divided experimental group equipment clusters, the performance indicator values ​​corresponding to all equipment identifiers belonging to the experimental group equipment clusters are selected from the unified time series data; the selected performance indicator values ​​are sorted according to the corresponding timestamps to form the performance indicator value sequence of the experimental group equipment clusters.

[0084] S2.5 For the control group of equipment, extract the performance index values ​​of the control group of equipment from the unified time series data and arrange them in chronological order to construct the performance index value sequence of the control group of equipment.

[0085] Furthermore, based on the list of equipment identifiers included in the divided control group equipment groups, the performance indicator values ​​corresponding to all equipment identifiers belonging to the control group equipment groups are screened from the unified time series data; the screened performance indicator values ​​are sorted according to the corresponding timestamps to form the performance indicator value sequence of the control group equipment groups.

[0086] S3. Based on the performance index numerical sequences of the experimental group equipment group and the control group equipment group, the net effect value and statistical confidence level of the target standard are calculated using a difference-in-differences model.

[0087] S3.1 Integrate the performance index numerical sequences of the experimental group equipment group with the performance index numerical sequences of the control group equipment group to form a panel dataset.

[0088] Furthermore, the key performance indicator (KPI) sequences of the experimental group equipment clusters and the KPI sequences of the control group equipment clusters were vertically merged to form a panel dataset containing equipment identifiers, time points, group labels, and KPI values.

[0089] S3.2 Based on the panel dataset, a difference-in-differences model is set up. The difference-in-differences model uses the performance index value as the dependent variable and the interaction term between the group label and the time point as the explanatory variable.

[0090] Furthermore, based on panel datasets, a difference-in-differences model is defined, with performance index values ​​as the dependent variable and the interaction term between group labels and time points as the core explanatory variables.

[0091] It should be noted that the performance index values ​​of each device in the panel dataset at each time point are used as the dependent variable; then, group-labeled dummy variables (values ​​of 1 for the experimental group and 0 for the control group) and time-point dummy variables (values ​​of 1 after standard adoption and 0 before adoption) are constructed; the group-labeled dummy variables and time-point dummy variables are multiplied to generate interaction terms as core explanatory variables; finally, a difference-in-differences model is established with performance index values ​​as the dependent variable and interaction terms as core explanatory variables, and the net effect of standard implementation is identified by comparing the differences in performance index changes between the experimental group and the control group before and after standard adoption.

[0092] S3.3. In the difference-in-differences model, the unit type and geographical climate zone are loaded as fixed effect control variables. The least squares method is used to fit the difference-in-differences model and calculate the coefficient estimates of the interaction terms between the group labels and time points to form the net effect value of the target standard. The standard error of the coefficients of the interaction terms between the group labels and time points is calculated. Based on the standard error and the coefficient estimates, the statistic is derived to obtain the probability value of the net effect value of the target standard.

[0093] The expression for the coefficient estimate is:

[0094] ;

[0095] in, For the first The device in time The estimated coefficients of the performance indicators are as follows. For the intercept term, For grouped dummy variable coefficients, For the first Grouped virtual variables for the devices The coefficients of the time dummy variable. For time dummy variables, The coefficient of the interaction term. For the first The device in time Random disturbance factors affecting performance indicators.

[0096] The standard error expression for the coefficient is:

[0097] ;

[0098] in, For the first The device in time The standard error of the coefficients for the performance indicators. For the first Control variables for the model of the equipment unit. These are the control variables for the geographical and climatic zones. This is a time-fixed effect. For individual fixed effects, These are the control variable coefficients for the geographical and climatic zones. These are the control variable coefficients for the unit model.

[0099] Furthermore, the design matrix of the difference-in-differences model is loaded with dummy variable sets representing unit model attributes and dummy variable sets representing geographic climate zone attributes as fixed effect control variables; the extended difference-in-differences model is fitted using the least squares method, and the coefficient estimates of the interaction terms between group labels and time points are calculated. The coefficient estimates are the net effect values ​​of the target standard; the standard errors of the interaction term coefficients are extracted from the variance-covariance matrix of the difference-in-differences model parameters; the statistics are calculated based on the coefficient estimates and standard errors, and the probability values ​​corresponding to the net effect values ​​of the target standard are obtained according to the distribution of the statistics.

[0100] It should be noted that the variance-covariance matrix is ​​one of the outputs of the parameter estimation in a difference-in-differences model. It is a symmetric square matrix. The elements on the main diagonal are the variances of the estimated parameters in the difference-in-differences model (such as coefficients of interaction terms, intercept terms, individual fixed effects, control variable coefficients for unit type, and control variable coefficients for geographic climate zones). The arithmetic square root of the variance is the standard error of the corresponding parameter, used to measure the accuracy of the parameter estimation. The elements on the off-diagonal correspond to the covariance between different parameter estimates, reflecting the degree of linear correlation between the estimators. For example, if the covariance between individual fixed effects and the control variable coefficient for a certain unit type is positive, it indicates that the estimation errors of both tend to move in the same direction. It is obtained by multiplying the estimated variance of the disturbance term in the regression model by the inverse of the cross product matrix of the design matrix.

[0101] S3.4 Compare the probability value of the net effect value of the target standard with the level threshold set by the variation analysis of the performance indicator numerical series of the control group equipment group in the historical operation data of the equipment before the adoption of the target standard, and form the statistical confidence level of the net effect value of the target standard.

[0102] Furthermore, the probability value of the net effect value of the target standard is compared with a statistical level threshold. The level threshold is set by performing variation analysis on the performance indicator value series of the control group equipment group in the panel data before the adoption time of the target standard. If the probability value of the net effect value of the target standard is less than the level threshold, the net effect value of the target standard is determined to pass the statistical confidence test. When the probability value of the net effect value of the target standard is greater than the level threshold, the net effect value of the target standard is determined to fail the statistical confidence test, thus forming a statistical confidence judgment on the net effect value of the target standard.

[0103] S4. Incorporate the generator unit model and geographical climate zone as fixed-effect control variables into the difference-in-differences model to control for potential biases caused by generator unit model and geographical environment attributes in the estimation of net effect value.

[0104] S4.1 Extract the unit model attribute and geographical climate zone attribute from the panel dataset, encode the unit model attribute and geographical climate zone attribute as classification variables, and generate a dummy variable set for the unit model attribute and a dummy variable set for the geographical climate zone attribute.

[0105] Furthermore, the generator set model attribute field and the geographic climate zone attribute field are read from the panel dataset; the generator set model attribute is treated as a categorical variable, generating a dummy variable with a value of 0 or 1 for each generator set model, and the dummy variables of all generator set models constitute the dummy variable set of the generator set model attribute; the geographic climate zone attribute is treated as a categorical variable, generating a dummy variable with a value of 0 or 1 for each geographic climate zone, and the dummy variables of all geographic climate zones constitute the dummy variable set of the geographic climate zone attribute.

[0106] S4.2. Add the set of dummy variables for unit model attributes and the set of dummy variables for geographical climate zone attributes as fixed effects control variables to the design matrix of the difference-in-differences model.

[0107] Furthermore, all dummy variables contained in the set of dummy variables for the generator model attribute are added as a set of column vectors to the design matrix of the difference-in-differences model; all dummy variables contained in the set of dummy variables for the geographic climate zone attribute are added as another set of column vectors to the design matrix of the difference-in-differences model; the newly added column vectors participate in the construction of the difference-in-differences model as fixed effects control variables.

[0108] S4.3. Use a matrix fitting double difference model with fixed-effect control variables to estimate the net effect value of the target standard by controlling the influence of unit model attribute and geographical climate zone attribute, and control the potential bias caused by unit model attribute and geographical climate zone attribute on the estimation of net effect value.

[0109] Furthermore, an extended design matrix containing dummy variables for generator model attributes and geographic climate zone attributes is used as the fixed-effects control variables. A difference-in-differences model is fitted using the least squares method. During model fitting, the dummy variables for generator model attributes and geographic climate zone attributes respectively control the systematic impact of inherent, time-invariant performance differences caused by different generator models and geographic climate zones on performance index values. By controlling the influence of inherent attributes, the final net effect value estimate of the target standard more accurately reflects the causal effects of standard implementation itself, thus effectively controlling the potential bias caused by generator model attributes and geographic climate zone attributes on the net effect value estimate.

[0110] S5. Construct a domain knowledge graph based on domain knowledge text data, and calculate the betweenness centrality score of the target standard in the knowledge graph.

[0111] S5.1 Perform named entity recognition on domain knowledge text data, identify and extract standard clause entities from target standard text, identify and extract fault mode entities from equipment failure case library, and identify and extract solution entities from maintenance solution manual.

[0112] Furthermore, named entity recognition is performed on the target standard text to identify and extract clauses describing specific technical requirements as standard clause entities; named entity recognition is performed on the equipment failure case library to identify and extract failure phenomena or causes as failure mode entities; and named entity recognition is performed on the maintenance solution manual to identify and extract specific maintenance measures or methods as solution entities.

[0113] S5.2 Extract relationships from domain knowledge text data, establish diagnostic relationships between standard clause entities and failure mode entities, and establish disposal relationships between failure mode entities and solution entities.

[0114] Furthermore, relation extraction is performed on domain knowledge text data to identify descriptions related to fault diagnosis in standard clause texts and establish diagnostic relationships between standard clause entities and fault mode entities; the association descriptions of fault causes and corresponding measures in maintenance documents are identified to establish disposal relationships between fault mode entities and solution entities.

[0115] S5.3. Based on standard clause entities, failure mode entities, solution entities, diagnostic relationships and handling relationships, construct a domain knowledge graph with entities as nodes and relationships as edges.

[0116] Furthermore, by using standard clause entities, failure mode entities, and solution entities as nodes, and diagnostic relationships and remediation relationships as directed edges, a domain knowledge graph is constructed.

[0117] S5.4 On the domain knowledge graph, calculate the proportion of the target standard node appearing on the shortest path from the failure mode node to the solution node, and obtain the betweenness centrality score of the target standard in the knowledge graph.

[0118] The expression for the betweenness centrality score is:

[0119] ;

[0120] in, Let j be the betweenness centrality score of node j. For fault mode nodes, As a solution node, This represents the total number of shortest paths from fault mode nodes to solution nodes.

[0121] Furthermore, on the domain knowledge graph, the proportion of times the target standard node appears in the shortest paths from all failure mode nodes to the corresponding solution nodes is calculated out of the total number of shortest paths. This proportion is the betweenness centrality score of the target standard in the knowledge graph.

[0122] S6. When the middle centrality score of the target standard exceeds the judgment threshold set by statistical quantile analysis based on historical operating data, and the net effect value fails the statistical confidence test, the associated entities not included in the difference-in-differences model are extracted from the knowledge graph as omitted variables. The net effect value is corrected by recalculating the difference-in-differences model to obtain a comprehensive index for evaluating the applicability of the target standard.

[0123] S6.1 Compare the middleness centrality score of the target standard with the judgment threshold set by statistical quantile analysis based on historical operating data, and compare the probability value of the net effect value of the target standard with the level threshold. If the probability value of the net effect value of the target standard is greater than the level threshold, the net effect value of the target standard fails the statistical confidence test.

[0124] S6.2 When the middleness centrality score of the target standard exceeds the judgment threshold and the net effect value of the target standard fails the statistical confidence test, extract the associated entities that are related to the target standard node but not included as control variables in the difference-in-differences model from the domain knowledge graph to form a set of potential omitted variables.

[0125] Furthermore, when the middle centrality score of the target standard exceeds the judgment threshold and the net effect value of the target standard fails the statistical confidence test, the failure mode entities and solution entities directly connected to the target standard node are traversed from the domain knowledge graph to screen out the related entities that have not yet been included as control variables in the difference-in-differences model, and the set of related entities is defined as the potential omission variable set.

[0126] S6.3. Add the quantifiable observation data corresponding to the potential missing variable set as new control variables to the difference-in-differences model.

[0127] Furthermore, each associated entity in the potential omission variable set is mapped to the corresponding quantifiable observation data. The monthly number of equipment failures corresponding to specific failure mode entities and the frequency of technology application corresponding to solution entities are added as new control variables to the design matrix of the difference-in-differences model.

[0128] S6.4. Recalculate using the difference-in-differences model with the new control variables to obtain the corrected coefficient estimates and form the net effect value of the corrected target standard.

[0129] The corrected expression for the coefficient estimate is:

[0130]

[0131] in, For the first The device in time The adjusted coefficient estimates for the following performance indicators To add a new set of control variables, for In a set of variables, the panel data sequence observations for each variable vary depending on the device. and different times Changes, This is the corrected intercept term. The coefficients of the corrected interaction terms. These are the control variable coefficients for the corrected unit model. This is the corrected time-fixed effect. This is the modified individual fixed effect. For the revised first The device in time Random disturbance factors in performance indicators For variable indexing.

[0132] Furthermore, by incorporating the quantifiable observational data corresponding to the potential omitted variable set as new control variables into the difference-in-differences model for refitting, the influence of previously ignored confounding factors on performance indicators can be effectively controlled, thereby separating out a purer standard treatment effect. The corrected interaction term coefficient estimate is used as the net effect value of the target standard, and the estimation result is closer to the true causal effect, reducing the estimation error caused by model specification bias and improving the accuracy and reliability of the evaluation results.

[0133] S6.5. Standardize and weight the net effect value of the modified target standard and the middleness centrality score of the target standard in the knowledge graph to calculate the comprehensive index number for evaluating the applicability of the target standard.

[0134] The comprehensive index expression for the applicability of the overall evaluation target standard is as follows:

[0135] ;

[0136] in, A comprehensive number of indicators to evaluate the applicability of target standards. The net effect value coefficient of the corrected target standard. For the standardized function, This represents the betweenness centrality score coefficient.

[0137] Furthermore, the net effect value of the modified target standard and the middleness centrality score of the target standard in the knowledge graph are standardized to eliminate the difference in dimensions. Then, the standardized values ​​are weighted and summed according to the preset weights. The result is the comprehensive index for evaluating the applicability of the target standard.

[0138] This embodiment also provides a standard system applicability evaluation system based on multi-source data, including: a data acquisition module, which collects historical operating data of the equipment and divides the equipment into an experimental group and a control group;

[0139] The preprocessing module preprocesses the historical operating data of the equipment and constructs numerical sequences of performance indicators that quantitatively reflect the achievement of the target standard for the technical objectives for both the experimental group and the control group.

[0140] The bias module, based on the performance index numerical series of the experimental group equipment group and the control group equipment group, uses a difference-in-differences model to calculate the net effect value and statistical confidence of the target standard. The unit model and geographical climate zone are added as fixed effect control variables to the difference-in-differences model to control the potential bias caused by the unit model and geographical environment attributes to the estimation of the net effect value.

[0141] The module constructs a domain knowledge graph based on domain knowledge text data and calculates the betweenness centrality score of the target standard in the knowledge graph.

[0142] In the judgment module, when the middleness centrality score of the target standard exceeds the judgment threshold set by statistical quantile analysis based on historical operating data, and the net effect value fails the statistical confidence test, the associated entities not included in the difference-in-differences model are extracted from the knowledge graph as omitted variables. The net effect value is then corrected by recalculating the difference-in-differences model to obtain the comprehensive index number for evaluating the applicability of the target standard.

[0143] This embodiment also provides a computer device applicable to the standard system applicability evaluation method based on multi-source data, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the standard system applicability evaluation method based on multi-source data as proposed in the above embodiment.

[0144] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0145] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the standard system applicability evaluation method based on multi-source data as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0146] In summary, this invention collects historical operating data of equipment and divides experimental and matched control groups based on standard adoption. It constructs a performance indicator sequence reflecting technical objectives, employs a difference-in-differences model, and controls for fixed effects such as unit model and geographical climate zone to calculate the net effect value and statistical confidence level of the target standard. This quantifies the causal effect of standard implementation. A knowledge graph is constructed based on domain knowledge text to calculate the middle centrality score of the target standard and assess its pivotal value in the knowledge network. When the aforementioned analysis indicates that the standard's knowledge value is high but the statistical effect does not meet the confidence requirements, latent omitted variables are automatically extracted from the knowledge graph, and the difference-in-differences model is recalculated to correct the net effect value. The corrected net effect value is then integrated with the middle centrality score to form a comprehensive index, achieving a multi-dimensional dynamic evaluation of standard applicability that combines data-driven approaches, causal identification, and knowledge value.

[0147] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating the applicability of a standard system based on multi-source data, characterized in that: include, Collect historical operating data of the equipment and divide the equipment into experimental group and control group, including the following steps: The device's historical operating data is constructed by collecting device identifiers, timestamps, and operating parameters from data sources. The device identifiers are matched with the adoption records of the target standards. The adoption records of the target standards record the time points when the device identifiers were adopted and the target standards were adopted. A list of devices marked with the target standards is generated. Based on the adoption marks in the device list, the set of device identifiers is divided into an experimental group of devices that adopted the target standards and a candidate set of a control group of devices that did not adopt the target standards. Based on the candidate set of the control group equipment group, the unit model, years of operation, geographical climate zone and annual average load rate attributes extracted from the historical operation data of the equipment are matched with the experimental group equipment group to form a control group equipment group that matches the experimental group equipment group in the aforementioned attributes. Historical operating data of the equipment is preprocessed, and performance indicator numerical sequences that quantify the achievement of the technical objective are constructed for the experimental group and the control group, respectively. The technical objective is to reduce the failure rate, and the performance indicator numerical sequences are the number of failures. Based on the performance index numerical sequences of the experimental group and the control group, a difference-in-differences model was used to calculate the net effect value and statistical confidence level of the target standard; including the following steps: The performance index numerical sequences of the experimental group equipment group and the performance index numerical sequences of the control group equipment group were integrated to form a panel dataset. Based on the panel dataset, a difference-in-differences model is set up. The difference-in-differences model uses the performance index value as the dependent variable and the interaction term between the group label and the time point as the explanatory variable. The time point is the time point when the target standard is adopted. In the difference-in-differences model, the generator type and geographical climate zone are loaded as fixed effect control variables. The least squares method is used to fit the difference-in-differences model and calculate the coefficient estimates of the interaction term between the group label and the time point to form the net effect value of the target standard. Then, the standard error of the coefficient of the interaction term between the group label and the time point is calculated. Based on the standard error and the coefficient estimates, the statistic is derived to obtain the probability value of the net effect value of the target standard. The probability value of the net effect value of the target standard is compared with the level threshold set by the variation analysis of the performance indicator numerical series of the control group of equipment in the historical operation data of the equipment before the adoption of the target standard, to form the statistical confidence level of the net effect value of the target standard. The generator unit model and geographical climate zone are added as fixed-effect control variables to the difference-in-differences model to control the potential bias caused by the generator unit model and geographical environment attributes to the estimation of net effect value. A domain knowledge graph is constructed based on domain knowledge text data, and the betweenness centrality score of the target standard in the knowledge graph is calculated. When the middle centrality score of the target standard exceeds the judgment threshold set by statistical quantile analysis based on historical operating data, and the net effect value fails the statistical confidence test, the associated entities related to the target standard node that are not included in the difference-in-differences model are extracted from the knowledge graph as omitted variables. The net effect value is corrected by recalculating the difference-in-differences model. The corrected net effect value of the target standard is then standardized and weighted and fused with the middle centrality score of the target standard in the knowledge graph to calculate the comprehensive index number for evaluating the applicability of the target standard. Constructing a domain knowledge graph based on domain knowledge text data, and calculating the betweenness centrality score of the target standard in the knowledge graph, includes the following steps: Named entity recognition is performed on domain knowledge text data to identify and extract standard clause entities from target standard text, fault mode entities from equipment failure case library, and solution entities from maintenance solution manual. Relationships are extracted from domain knowledge text data to establish diagnostic relationships between standard clause entities and failure mode entities, and disposal relationships between failure mode entities and solution entities. Based on standard clause entities, failure mode entities, solution entities, diagnostic relationships, and handling relationships, a domain knowledge graph is constructed with entities as nodes and relationships as edges. On the domain knowledge graph, the proportion of target standard nodes appearing on the shortest path from failure mode nodes to solution nodes is calculated to obtain the betweenness centrality score of the target standard in the knowledge graph.

2. The method for evaluating the applicability of a standard system based on multi-source data as described in claim 1, characterized in that: The historical operating data of the equipment was preprocessed, and numerical sequences of performance indicators reflecting the achievement of the technical objectives were constructed for both the experimental and control group equipment. These included the following steps: The historical operating data of the equipment is cleaned to generate clean historical operating data. The clean historical operating data is then aligned with the equipment maintenance record data according to the equipment identifier and timestamp to form a unified time series data of equipment identifier, timestamp, operating parameters and maintenance records. Based on the target standards and technical objectives, performance indicator values ​​are defined and calculated from the unified time series data. For the experimental group of equipment, the performance indicators of the experimental group of equipment are extracted from the unified time series data and arranged in chronological order to construct the numerical sequence of the performance indicators of the experimental group of equipment. For the control group of equipment, the performance index values ​​of the control group of equipment were extracted from the unified time series data and arranged in chronological order to construct the performance index value sequence of the control group of equipment.

3. The method for evaluating the applicability of a standard system based on multi-source data as described in claim 1, characterized in that: The generator unit model and geographical climate zone are added as fixed-effects control variables to the difference-in-differences model to control for potential biases in the estimation of net effect values ​​caused by generator unit model and geographical environmental attributes. This includes the following steps: Extract the unit model attribute and geographic climate zone attribute from the panel dataset, encode the unit model attribute and geographic climate zone attribute as classification variables, and generate a dummy variable set for the unit model attribute and a dummy variable set for the geographic climate zone attribute. The set of dummy variables for unit model attributes and the set of dummy variables for geographical climate zone attributes are added as fixed-effect control variables to the design matrix of the difference-in-differences model. A matrix-fit difference-in-differences model with fixed-effects control variables is used to estimate the net effect value of the target standard by controlling the influence of unit type attribute and geographic climate zone attribute, and to control the potential bias caused by unit type attribute and geographic climate zone attribute on the estimation of net effect value.

4. The method for evaluating the applicability of a standard system based on multi-source data as described in claim 1, characterized in that: When the middle centrality score of the target standard exceeds the threshold set by statistical quantile analysis based on historical operating data, and the net effect value fails the statistical confidence test, the associated entities not included in the difference-in-differences model are extracted from the knowledge graph as omitted variables. The net effect value is then corrected by recalculating the difference-in-differences model to obtain a comprehensive index for evaluating the applicability of the target standard. This includes the following steps: The middleness centrality score of the target standard is compared with the judgment threshold set by statistical quantile analysis based on historical operating data. The probability value of the net effect value of the target standard is compared with the level threshold. If the probability value of the net effect value of the target standard is greater than the level threshold, the net effect value of the target standard fails the statistical confidence test. When the middle centrality score of the target standard exceeds the judgment threshold and the net effect value of the target standard fails the statistical confidence test, the associated entities that are not included as control variables in the difference-in-differences model are extracted from the domain knowledge graph and are related to the nodes of the target standard, forming a set of potential omitted variables. The quantifiable observational data corresponding to the potential missing variable set are added as new control variables to the difference-in-differences model; The calculations were recalculated using a difference-in-differences model with new control variables to obtain revised coefficient estimates, which formed the revised net effect value of the target standard.

5. A standard system applicability evaluation system based on multi-source data, based on the standard system applicability evaluation method based on multi-source data as described in any one of claims 1 to 4, characterized in that: This includes a data acquisition module that collects historical operating data from the equipment and divides the equipment into an experimental group and a control group. The preprocessing module preprocesses the historical operating data of the equipment and constructs numerical sequences of performance indicators that quantitatively reflect the achievement of the target standard for the technical objectives for both the experimental group and the control group. The bias module, based on the performance index numerical series of the experimental group equipment group and the control group equipment group, uses a difference-in-differences model to calculate the net effect value and statistical confidence of the target standard. The unit model and geographical climate zone are added as fixed effect control variables to the difference-in-differences model to control the potential bias caused by the unit model and geographical environment attributes to the estimation of the net effect value. The module constructs a domain knowledge graph based on domain knowledge text data and calculates the betweenness centrality score of the target standard in the knowledge graph. In the judgment module, when the middleness centrality score of the target standard exceeds the judgment threshold set by statistical quantile analysis based on historical operating data, and the net effect value fails the statistical confidence test, the associated entities not included in the difference-in-differences model are extracted from the knowledge graph as omitted variables. The net effect value is then corrected by recalculating the difference-in-differences model to obtain the comprehensive index number for evaluating the applicability of the target standard.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the standard system applicability evaluation method based on multi-source data as described in any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the standard system applicability evaluation method based on multi-source data as described in any one of claims 1 to 4.

Images