Inter-plant standard specification conversion and application management system

Abstract

The invention relates to the technical field of data management and specification conversion, and provides an inter-plant standard specification conversion and application management system, a standardized configuration module is used for uniformly formulating data specifications of a plant, the data specifications comprise a unit system and a parameter naming rule, and a multi-source data conversion module is used for converting collected and input heterogeneous data of all departments into a data storage module; the data processing module automatically converts the data into effective data with the same standard and specification as the factory area and the factory building, the data verification module monitors the data quality in real time, repairs errors, establishes a data quality rule base and generates a data quality report with credibility scores, and the tracing module supports backtracking of any version of data based on data operation traces; according to the method, heterogeneous parameter semantic matching is realized by combining a Hungary algorithm, abnormal data intelligent repair is realized by low-rank matrix completion of a data verification module and a graph Laplacian regular term repair technology, and multi-plant data specification and unification, cross-department data format compatibility and continuous optimization of production data quality are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of data management and specification conversion, and in particular to a standard specification conversion and application management system between factories. Background Art

[0002] With the continuous development of science and technology, as of 2024, in the automated and semi-automated technology-based industries, the configuration, integration, application and planning of process production machinery and equipment information, including subsequent machinery and equipment installation procedures and mutual coordination, adopt a combination of manual management and corresponding word processing software. Although a certain degree of information management can be achieved, there are still many problems that cannot be effectively avoided.

[0003] Among them, core elements such as information collection, calculation, conversion, confirmation of results, accounting and subsequent application planning still rely mainly on manpower, which is prone to problems such as misjudgment, loss, omission of information and other human errors.

[0004] Defects in the prior art: 1. Misjudgment of information: The information sources of each factory area and factory building are different, and there is a lack of effective integration and standardization. For example, the personnel responsible for information collection may misjudge or misinterpret the original information due to their working habits, resulting in a series of serious problems such as information errors and subsequent use.

[0005] 2. Information loss: Since the inherent data information formats and items are different between various factory areas and factory buildings, if the existing operating mode and habits are followed, it is easy to cause blank information to be obtained by subsequent personnel, including the importance of information and the shortage of original factory information, which will affect the comprehensiveness, accuracy and timeliness of the information, and may increase communication costs and even cause substantial economic losses.

[0006] 3. Information omission: Traditional manual management may lead to data errors or data loss, and may even lead to deliberate human sabotage, which will mislead the use of information within the system and the judgment and decision-making of management levels, thus causing irreparable economic losses. Summary of the invention

[0007] In order to solve the above technical problems, the present invention provides a standard specification conversion and application management system between factories, which realizes semantic matching of heterogeneous parameters through a semantic analysis unit combined with the Hungarian algorithm, and realizes intelligent repair of abnormal data through low-rank matrix completion and graph Laplace regularization term repair technology of the data verification module, thereby realizing the unification of data specifications of multiple factories, cross-departmental data format compatibility, and continuous optimization of production data quality, so as to solve the problems in the prior art.

[0008] An inter-factory standard specification conversion and application management system, comprising: A standardized configuration module, which is used to uniformly formulate data specifications for the factory area, including unit system and parameter naming rules, based on industry specifications and expert experience; A multi-source data conversion module, which is connected to the standardization configuration module for automatically converting the collected and input heterogeneous data of various departments into valid data with the same standards and specifications as the factory area and factory building based on the configuration of the standardization configuration module, and generating a conversion log during the conversion for users to trace the original data and conversion logic; A data verification module, which is data-connected to the multi-source data conversion module, is used to monitor data quality in real time and fix errors, establish a data quality rule base, and generate a data quality report with a credibility score; The tracing module is connected with the multi-source data conversion module and the data verification module. Based on the data operation traces, the operation log is solidified by the blockchain hash algorithm. Based on the timestamp tracing function, it supports the backtracking of any version of data. Preferably, the standardized configuration module includes a Java-based visual configuration interface, allowing experts to set standard templates, integrating existing databases, automatically mapping device parameter naming, a built-in international standard formula library, and using the Drools rule engine to achieve dynamic loading and effectiveness of standard rules; Preferably, the multi-source data conversion module includes a data acquisition unit, a semantic analysis unit and a conversion unit; The data collection unit establishes a connection with the data sources of each department and uses data collection technology to collect data in different formats and from different sources, thereby realizing the collection and input of heterogeneous data from multiple departments; The semantic analysis unit pre-processes the collected data, removes null values ​​and normalizes the numerical range, then uses the pre-trained word vector model Word2Vec to generate the semantic vector of the parameter name, and then calculates the cosine similarity matrix and uses the Hungarian algorithm to solve the maximum weight matching to complete the mapping of the parameter name; The conversion unit defines a linear transformation matrix, applies transformation to each data point, converts different units to a unified standard, records original values, matching rules, and operation timestamps, and finally outputs a standardized data matrix that complies with unified naming and unit specifications.

[0009] Preferably, the specific analysis process of the semantic analysis unit is as follows: Assume that the original data matrix of each department acquired and input by the data acquisition unit is ;in, Represents the elements in the original data matrix, m represents the number of data entries, and n represents the characteristic dimension of the data, that is, the heterogeneous parameters of different departments. The data is then preprocessed to remove null values ​​and normalize the numerical range.

[0010] Parameter semantic matching: Set i and j as local index variables, embed word vectors, use the pre-trained word vector model Word2Vec to generate semantic vectors of parameter names, and map heterogeneous parameter names to a unified naming convention; in, Indicates the parameter name The word vector of Represents the dimension of the word vector; Similarity matrix, calculate the cosine similarity matrix between source parameters and target parameters , where k is the number of standard parameters; in, Represents the original data The unified data after mapping, express Standardized form of Optimal matching, using the Hungarian algorithm to solve the maximum weight matching; The optimization problem is: Constraints: The Hungarian algorithm is used to calculate the output matching matrix A and record the parameter mapping relationship.

[0011] Preferably, the conversion unit converts different units into a unified standard, and the specific process is as follows: Unit transformation matrix, defines the linear transformation matrix , each element Represents the conversion coefficient from i to j, and then performs batch conversion for each data point Apply the transformation: Among them, A is the parameter matching matrix, b is the bias vector, and finally, the output standardized data matrix , which complies with unified naming and unit specifications, and the conversion log records the original value, matching rules, and operation timestamp.

[0012] Preferably, the data verification module includes a verification unit, a repair unit and an evaluation unit; The inspection unit inputs a real-time standardized data stream that changes over time, uses a sliding window to calculate the mean and covariance to build a dynamic baseline, then measures the distance between each data point and the dynamic baseline by calculating the Mahalanobis distance, and sets a threshold based on the chi-square distribution to identify data anomalies in real time; The repair unit decomposes the data matrix into a low-rank matrix and sparse noise, obtains normal mode and noise through constraint decomposition, adds graph Laplace regularization terms to the low-rank matrix for spatiotemporal correction, uses conjugate gradient method and other iterative solutions, and finally outputs the repaired data; The evaluation unit defines five dimensions of quality indicators, calculates the weight of each indicator by entropy weight method, calculates the comprehensive score, normalizes the score to [0,1] by Sigmoid function, and finally outputs a quality report including the changes of scores, abnormal points, and repair records over time.

[0013] Preferably, the inspection unit inputs Changing real-time standardized data streams , and then input includes sliding window calculation of the mean of the real-time standardized data stream and covariance Dynamic baseline of in, Represents the attenuation factor, EMA is the exponential moving average, which is used to update the covariance matrix; Perform anomaly detection on input data to identify data anomalies in real time; measure the distance between each data point and the dynamic baseline by calculating the Mahalanobis distance : Dynamic Threshold: Setting Threshold Based on Chi-Square Distribution : in, Indicates based on k and confidence The chi-square distribution of This means a 99% confidence level. , marking this data point as an anomaly.

[0014] Preferably, the specific repair process of the repair unit is as follows: Low-rank matrix completion: Decompose the data matrix into a low-rank matrix L and sparse noise E: Constraints: break down , we get the normal mode L and noise E, where represents the sparse noise regularization weight, which controls the tolerance to outliers and is adjusted according to the noise level and selected by cross-validation. Expressed as the nuclear norm, i.e., the low-rank constraint, As a sparse constraint, the low-rank matrix L represents the main part of the repaired data, and the sparse noise E is used to capture outliers and missing values; Perform spatiotemporal correction on L and add graph Laplace regularization terms to maintain spatial correlation between devices: Derivative the above formula and set the derivative to zero, then iterate and solve it by conjugate gradient method to obtain ; in, is the trace of the matrix, i.e. the sum of the diagonal elements, used for the graph Laplace regularization term, is the Laplace matrix of the device connection graph, is the weight of the graph Laplace regularization term, which controls the strength of spatial correlation between devices and needs to be reduced when devices are densely connected; Output repair data and merge results ,in is the sparse noise after threshold processing.

[0015] Preferably, the evaluation unit performs a credibility evaluation and generates a data quality report; The quality index is defined as 5 dimensions: integrity ,consistency , Timeliness ,precise and stability , the weight of each indicator is calculated by entropy weight method : in: Overall rating: in, Represents data based The quality index value after normalization of the z-th quality index, The information entropy of the z-th quality indicator indicates the data discreteness of the quality indicator. represents the indicator weight corresponding to the z-th quality indicator value, Indicates at time The comprehensive credibility score is normalized to [0,1] through the Sigmoid function; Finally, the output is a quality report showing the changes of the data over time, including the score S, abnormal points, and repair records.

[0016] Preferably, the system further includes a plant area collaboration module and a standard specification application module; The plant collaboration module is used to support online collaborative editing and version control among multiple departments, and divide data editing permissions by plant / equipment type; The standard specification application module is used to provide standardized data services for other systems, output equipment parameters and process data in a unified format, and provide a standard conversion SDK to support third-party systems to call data conversion services, connect to the factory MES / ERP system, and automatically push standardized production data.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention realizes dynamic loading of industry specifications and expert experience through the visual configuration interface of the standardized configuration module and the Drools rule engine, achieves unified formulation and dynamic update of multi-plant data specifications, effectively avoids parameter naming confusion and rule conflicts caused by differences in manual interpretation, and improves the consistency of data understanding.

[0018] 2. The present invention realizes automatic semantic matching and mapping of heterogeneous parameter names through the semantic analysis unit of the multi-source data conversion module combined with Word2Vec word vector embedding and the Hungarian algorithm, solves the problem of missing key information caused by incompatible cross-departmental data formats, and ensures data integrity and comprehensiveness.

[0019] 3. The present invention realizes the intelligent repair of abnormal data and the preservation of spatiotemporal correlation through the low-rank matrix completion and graph Laplace regularization term repair technology of the data verification module, achieves real-time monitoring and automatic compensation of production data quality, and avoids the decision-making risks caused by data omission or tampering in traditional manual management. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the connection flow between the modules of the present invention; Figure 2 It is a schematic diagram of a specific working process of the multi-source data conversion module of the present invention; Figure 3 It is a schematic diagram of the specific working process of the data verification module of the present invention. DETAILED DESCRIPTION

[0021] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0022] The present invention provides a plant-to-plant standard specification conversion and application management system, comprising: A standardized configuration module, which is used to uniformly formulate data specifications for the factory area, including unit system and parameter naming rules, based on industry specifications and expert experience; A multi-source data conversion module, which is connected to the standardization configuration module for automatically converting the collected and input heterogeneous data of various departments into valid data with the same standards and specifications as the factory area and factory building based on the configuration of the standardization configuration module, and generating a conversion log during the conversion for users to trace the original data and conversion logic; A data verification module, which is data-connected to the multi-source data conversion module, is used to monitor data quality in real time and fix errors, establish a data quality rule base, and generate a data quality report with a credibility score; The tracing module is connected with the multi-source data conversion module and the data verification module, solidifies the operation log based on the data operation traces, and supports the backtracking of any version of data based on the timestamp tracing function.

[0023] Example: like Figure 1-Figure 3 As shown, in this embodiment, a manufacturing enterprise has multiple production plants, and the naming rules of equipment parameters in each plant are not unified. For example, Plant A uses "Temperature (℃)", Plant B uses "TEMP (F)", etc. At the same time, the data formats are significantly different.

[0024] Under the traditional manual processing mode, cross-factory data integration takes a long time and causes considerable losses every month due to parameter misjudgment and omission. Therefore, an inter-factory standard specification conversion and application management system of the present invention is introduced.

[0025] First, the plant collaboration module organizes online collaboration among engineers from the process, equipment, and IT departments. Through the visual interface of the standardized configuration module, it defines a unified unit system and parameter naming rules based on some standard documents of the plant industry and transmits them to the multi-source data conversion module.

[0026] Through standardized configuration modules, based on industry norms and expert experience, the company has uniformly formulated plant data specifications, including unit systems and parameter naming rules. This makes the data of each plant consistent and comparable, allowing staff to understand and use data more easily, improving work efficiency.

[0027] The multi-source data conversion module includes a data acquisition unit, a semantic analysis unit and a conversion unit; The data collection unit establishes connections with data sources of various departments and uses data collection technology to collect data in different formats and from different sources, thus realizing the collection and input of heterogeneous data from multiple departments; The semantic analysis unit pre-processes the collected data, removes null values ​​and normalizes the numerical range, then uses the pre-trained word vector model Word2Vec to generate the semantic vector of the parameter name, and then calculates the cosine similarity matrix and uses the Hungarian algorithm to solve the maximum weight matching to complete the mapping of the parameter name; The specific analysis process of the semantic analysis unit is as follows: Assume that the original data matrix of each department acquired and input by the data acquisition unit is ;in, Represents the elements in the original data matrix, m represents the number of data entries, and n represents the characteristic dimension of the data, that is, the heterogeneous parameters of different departments. The data is then preprocessed to remove null values ​​and normalize the numerical range.

[0028] Parameter semantic matching: Set i and j as local index variables, embed word vectors, use the pre-trained word vector model Word2Vec to generate semantic vectors of parameter names, and map heterogeneous parameter names to a unified naming convention; in, Indicates the parameter name The word vector of Represents the dimension of the word vector; Similarity matrix, calculate the cosine similarity matrix between source parameters and target parameters , where k is the number of standard parameters; in, Represents the original data The unified data after mapping, express Standardized form of Optimal matching, using the Hungarian algorithm to solve the maximum weight matching; The optimization problem is: Constraints: The Hungarian algorithm is used to calculate the output matching matrix A and record the parameter mapping relationship.

[0029] The conversion unit converts different units into a unified standard by defining a linear transformation matrix and applying transformation to each data point, while recording the original value, matching rules, and operation timestamps, and finally outputs a standardized data matrix that conforms to unified naming and unit specifications.

[0030] Convert units to a unified standard. The specific process is as follows: Unit transformation matrix, defines the linear transformation matrix , each element Represents the conversion coefficient from i to j, and then performs batch conversion for each data point Apply the transformation: Among them, A is the parameter matching matrix, b is the bias vector, and finally, the output standardized data matrix , which complies with unified naming and unit specifications, and the conversion log records the original value, matching rules, and operation timestamp.

[0031] The multi-source data conversion module automatically converts the heterogeneous data of various departments into valid data with unified standards and specifications, and generates conversion logs. This greatly reduces the workload of manual data processing, while improving the accuracy and traceability of data conversion. Staff can easily trace the original data and conversion logic to ensure the reliability of the data.

[0032] The data verification module includes a verification unit, a repair unit and an evaluation unit; The inspection unit inputs a real-time standardized data stream that changes over time, uses a sliding window to calculate the mean and covariance to build a dynamic baseline, then measures the distance between each data point and the dynamic baseline by calculating the Mahalanobis distance, and sets a threshold based on the chi-square distribution to identify data anomalies in real time; Verify unit input over time Changing real-time standardized data streams , and then input includes sliding window calculation of the mean of the real-time standardized data stream and covariance Dynamic baseline of in, Represents the attenuation factor, EMA is the exponential moving average, which is used to update the covariance matrix; Perform anomaly detection on input data to identify data anomalies in real time; measure the distance between each data point and the dynamic baseline by calculating the Mahalanobis distance : Dynamic Threshold: Setting Threshold Based on Chi-Square Distribution : in, Indicates based on k and confidence The chi-square distribution of This means a 99% confidence level. , marking this data point as an anomaly.

[0033] The repair unit decomposes the data matrix into a low-rank matrix and sparse noise, obtains the normal mode and noise through constraint decomposition, adds the graph Laplace regularization term to the low-rank matrix for spatiotemporal correction, uses the conjugate gradient method and other iterative solutions, and finally outputs the repaired data; The specific repair process of the repair unit is as follows: Low-rank matrix completion: Decompose the data matrix into a low-rank matrix L and sparse noise E: Constraints: break down , we get the normal mode L and noise E, where represents the sparse noise regularization weight, which controls the tolerance to outliers and is adjusted according to the noise level and selected by cross-validation. Expressed as the nuclear norm, i.e., the low-rank constraint, As a sparse constraint, the low-rank matrix L represents the main part of the repaired data, and the sparse noise E is used to capture outliers and missing values; Perform spatiotemporal correction on L and add graph Laplace regularization terms to maintain spatial correlation between devices: Derivative the above formula and set the derivative to zero, then iterate and solve it by conjugate gradient method to obtain ; in, is the trace of the matrix, i.e. the sum of the diagonal elements, used for the graph Laplace regularization term, is the Laplace matrix of the device connection graph, is the weight of the graph Laplace regularization term, which controls the strength of spatial correlation between devices and needs to be reduced when devices are densely connected; Output repair data and merge results ,in is the sparse noise after threshold processing.

[0034] Through the low-rank matrix completion and graph Laplace regularization term repair technology of the data verification module, intelligent repair of abnormal data and maintenance of spatiotemporal correlation are achieved, real-time monitoring and automatic compensation of production data quality are achieved, and decision-making risks caused by data omission or tampering in traditional manual management are avoided.

[0035] The evaluation unit defines five dimensions of quality indicators, calculates the weight of each indicator by using the entropy weight method, and then calculates the comprehensive score. The score is normalized to [0,1] using the Sigmoid function, and finally outputs a quality report containing the score, abnormal points, and repair records over time. The quality index is defined as 5 dimensions: integrity ,consistency , Timeliness ,precise and stability , the weight of each indicator is calculated by entropy weight method : in: Overall rating: in, Represents data based The quality index value after normalization of the z-th quality index, The information entropy of the z-th quality indicator indicates the data discreteness of the quality indicator. represents the indicator weight corresponding to the z-th quality indicator value, Indicates at time The comprehensive credibility score is normalized to [0,1] through the Sigmoid function; The data verification module monitors data quality in real time, discovers and fixes errors in a timely manner. The established data quality rule base and the generated data quality report with credibility score enable enterprises to fully understand the quality status of data and provide reliable data support for production decisions.

[0036] Finally, the output is a quality report showing the changes of the data over time, including the score S, abnormal points, and repair records.

[0037] At the same time, the system also includes a plant collaboration module and a standard specification application module. The plant collaboration module is used to support multi-department online collaborative editing and version control, and divides data editing permissions by plant / equipment type; The standard specification application module is used to provide standardized data services for other systems, output equipment parameters and process data in a unified format, and provide a standard conversion SDK to support third-party systems to call data conversion services, connect to the factory MES / ERP system, and automatically push standardized production data.

[0038] The factory collaboration module supports online collaborative editing and version control among multiple departments, and divides data editing permissions by factory area / equipment type. This promotes communication and collaboration among departments and avoids data conflicts and duplications.

[0039] The embodiments of the present invention are provided for the purpose of illustration and description. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations of the present invention. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. An inter-factory standard specification conversion and application management system, characterized in that: include: A standardized configuration module, which is used to uniformly formulate data specifications for the factory area, including unit system and parameter naming rules, based on industry specifications and expert experience; A multi-source data conversion module, which is connected to the standardization configuration module for automatically converting the collected and input heterogeneous data of various departments into valid data with the same standards and specifications as the factory area and factory building based on the configuration of the standardization configuration module, and generating a conversion log during the conversion for users to trace the original data and conversion logic; A data verification module, which is data-connected to the multi-source data conversion module, is used to monitor data quality in real time and fix errors, establish a data quality rule base, and generate a data quality report with a credibility score; The tracing module is connected with the multi-source data conversion module and the data verification module, solidifies the operation log based on the data operation traces, and supports the backtracking of any version of data based on the timestamp tracing function.

2. The inter-factory standard specification conversion and application management system as claimed in claim 1, characterized in that: The standardized configuration module includes a Java-based visual configuration interface, allowing experts to set standard templates, integrate existing databases, automatically map device parameter names, have a built-in international standard formula library, and use the Drools rule engine to dynamically load and validate standard rules.

3. The inter-factory standard specification conversion and application management system as claimed in claim 1, characterized in that: The multi-source data conversion module includes a data acquisition unit, a semantic analysis unit and a conversion unit; The data collection unit establishes a connection with the data sources of each department and uses data collection technology to collect data in different formats and from different sources, thereby realizing the collection and input of heterogeneous data from multiple departments; The semantic analysis unit pre-processes the collected data, removes null values ​​and normalizes the numerical range, then uses the pre-trained word vector model Word2Vec to generate the semantic vector of the parameter name, and then calculates the cosine similarity matrix and uses the Hungarian algorithm to solve the maximum weight matching to complete the mapping of the parameter name; The conversion unit defines a linear transformation matrix, applies transformation to each data point, converts different units to a unified standard, records original values, matching rules, and operation timestamps, and finally outputs a standardized data matrix that complies with unified naming and unit specifications.

4. The inter-factory standard specification conversion and application management system as claimed in claim 3, characterized in that: The specific analysis process of the semantic analysis unit is as follows: Assume that the original data matrix of each department acquired and input by the data acquisition unit is ;in, represents the elements in the original data matrix, m represents the number of data entries, and n represents the characteristic dimension of the data, that is, the heterogeneous parameters of different departments. Then the data is preprocessed to remove null values ​​and normalize the value range; Parameter semantic matching: Set i and j as local index variables, embed word vectors, use the pre-trained word vector model Word2Vec to generate semantic vectors of parameter names, and map heterogeneous parameter names to a unified naming convention; in, Indicates the parameter name The word vector of Represents the dimension of the word vector; Similarity matrix, calculate the cosine similarity matrix between source parameters and target parameters , where k is the number of standard parameters; in, Represents the original data The unified data after mapping, express Standardized form of Optimal matching, using the Hungarian algorithm to solve the maximum weight matching; The optimization problem is: Constraints: The Hungarian algorithm is used to calculate the output matching matrix A and record the parameter mapping relationship.

5. The inter-factory standard specification conversion and application management system as claimed in claim 4, characterized in that: The conversion unit converts different units into a unified standard. The specific process is as follows: Unit transformation matrix, defines the linear transformation matrix , each element Represents the conversion coefficient from i to j, and then performs batch conversion for each data point Apply the transformation: Among them, A is the matching matrix, b is the bias vector, and finally, the output is the standardized data matrix , which complies with unified naming and unit specifications, and the conversion log records the original value, matching rules, and operation timestamp.

6. The inter-factory standard specification conversion and application management system as claimed in claim 5, characterized in that: The data verification module includes a verification unit, a repair unit and an evaluation unit; The inspection unit inputs a real-time standardized data stream that changes over time, uses a sliding window to calculate the mean and covariance to build a dynamic baseline, then measures the distance between each data point and the dynamic baseline by calculating the Mahalanobis distance, and sets a threshold based on the chi-square distribution to identify data anomalies in real time; The repair unit decomposes the data matrix into a low-rank matrix and sparse noise, obtains normal mode and noise through constraint decomposition, adds graph Laplace regularization terms to the low-rank matrix for spatiotemporal correction, uses conjugate gradient method and other iterative solutions, and finally outputs the repaired data; The evaluation unit defines five dimensions of quality indicators, calculates the weight of each indicator by entropy weight method, calculates the comprehensive score, normalizes the score to [0,1] by Sigmoid function, and finally outputs a quality report including the changes of scores, abnormal points, and repair records over time.

7. The inter-factory standard specification conversion and application management system as claimed in claim 6, characterized in that: The inspection unit inputs over time Changing real-time standardized data streams , and then input includes sliding window calculation of the mean of the real-time standardized data stream and covariance Dynamic baseline of in, Represents the attenuation factor, EMA is the exponential moving average, which is used to update the covariance matrix; Perform anomaly detection on input data to identify data anomalies in real time; measure the distance between each data point and the dynamic baseline by calculating the Mahalanobis distance : Dynamic Threshold: Setting Threshold Based on Chi-Square Distribution : in, Represents the k-based and confidence The chi-square distribution of This means a 99% confidence level. , marking this data point as an anomaly.

8. The inter-factory standard specification conversion and application management system as claimed in claim 7, characterized in that: The specific repair process of the repair unit is as follows: Low-rank matrix completion: Decompose the data matrix into a low-rank matrix L and sparse noise E: Constraints: break down , we get the normal mode L and noise E, where represents the sparse noise regularization weight, which controls the tolerance to outliers and is adjusted according to the noise level and selected by cross-validation. Expressed as the nuclear norm, i.e., the low-rank constraint, As a sparse constraint, the low-rank matrix L represents the main part of the repaired data, and the sparse noise E is used to capture outliers and missing values; Perform spatiotemporal correction on L and add graph Laplace regularization terms to maintain spatial correlation between devices: Derivative the above formula and set the derivative to zero, then iterate and solve it by conjugate gradient method to obtain ; in, is the trace of the matrix, i.e. the sum of the diagonal elements, used for the graph Laplace regularization term, is the Laplace matrix of the device connection graph, is the weight of the graph Laplace regularization term, which controls the strength of spatial correlation between devices and needs to be reduced when devices are densely connected; Output repair data and merge results ,in is the sparse noise after threshold processing.

9. The inter-factory standard specification conversion and application management system as claimed in claim 8, characterized in that: The evaluation unit performs credibility evaluation and generates a data quality report; The quality index is defined as 5 dimensions: integrity ,consistency , Timeliness ,precise and stability , the weight of each indicator is calculated by entropy weight method : in: Overall rating: in, Represents data based The quality index value after normalization of the z-th quality index, The information entropy of the z-th quality indicator indicates the data discreteness of the quality indicator. represents the indicator weight corresponding to the z-th quality indicator value, Indicates at time The comprehensive credibility score is normalized to [0,1] through the Sigmoid function; Finally, the output is a quality report showing the changes of the data over time, including the score S, abnormal points, and repair records.

10. The inter-factory standard specification conversion and application management system as claimed in claim 1, characterized in that: The system also includes a plant area collaboration module and a standard specification application module; The plant collaboration module is used to support online collaborative editing and version control among multiple departments, and divide data editing permissions by plant / equipment type; The standard specification application module is used to provide standardized data services for other systems, output equipment parameters and process data in a unified format, and provide a standard conversion SDK to support third-party systems to call data conversion services, connect to the factory MES / ERP system, and automatically push standardized production data.

Images