Data verification method, system, terminal and medium for the whole process of power transmission and transformation equipment manufacturing

Abstract

The present invention belongs to the field of data quality management technology and specifically discloses a data verification method, system, terminal, and medium for the entire manufacturing process of power transmission and transformation equipment. The method comprises: collecting raw business data from each stage of power transmission and transformation equipment manufacturing, and performing standardization, format unification, and preliminary anomaly removal; uploading the processed data to a cloud processing module to perform semantic fusion and field mapping processing; invoking the corresponding verification rule set based on the manufacturing status to perform field-level and record-level quality verification on structured data; generating anomaly identifiers for unqualified data and feeding them back to the client interface, while storing qualified data in a blockchain module for trusted evidence storage; and finally generating a data quality verification report and performing a structured visual display. The present invention solves problems such as inconsistent semantics of existing manufacturing data, the inability to dynamically adapt rules, the difficulty in identifying anomalies with high confidence, and the lack of data evidence.

Description

Technical Field

[0001] The present invention belongs to the technical field of data quality management, and in particular relates to a method, system, terminal and medium for verifying data throughout the entire manufacturing process of power transmission and transformation equipment. Background Art

[0002] As a typical discrete industrial process with high complexity and high reliability requirements, power transmission and transformation equipment manufacturing requires data traceability, quality transparency and process controllability throughout the entire product process, which are becoming core technical demands.

[0003] Currently, the industry generally adopts a multi-system collaborative manufacturing model, including design systems, production execution systems (MES), quality management systems (QMS), equipment monitoring systems, etc. However, there are problems such as heterogeneous data sources, inconsistent semantics, lack of standards, and redundant records among these systems, which makes it impossible to directly integrate and apply data, especially forming an obvious bottleneck when performing quality inspection tasks.

[0004] The existing data verification model for power transmission and transformation equipment manufacturing still relies primarily on manual verification and static rule-based proofreading. While some companies have deployed some automated verification tools, these tools typically only support field-level validity verification and fail to perceive key quality factors such as semantic correlations between fields, process stage dependencies, and historical trend changes. This can easily lead to misjudgments and omissions, and even serious information silos, such as the same batch of data being processed repeatedly and mistakenly entered into the system. Furthermore, existing systems struggle to flexibly schedule verification rules in dynamic manufacturing environments and lack the self-learning capabilities to adapt to complex operating conditions.

[0005] The identification of abnormal data often relies on empirical thresholds or fixed templates, lacking intelligent algorithm support and being nearly incapable of processing high-dimensional, multi-channel, and multi-process data matrices. Furthermore, the lack of a unified and reliable data archiving mechanism makes it difficult to trace abnormal events, and risks such as data falsification and version confusion are difficult to address at the source. Summary of the Invention

[0006] In response to the problems in the prior art, the present invention provides a data verification method, system, terminal and medium for the entire process of power transmission and transformation equipment manufacturing, which solves the problem that the judgment of abnormal data in the prior art mostly relies on empirical threshold methods or fixed templates, resulting in the risks of data falsification and version confusion.

[0007] The technical solution adopted in the present invention is as follows:

[0008] In a first aspect, the present application provides a method for verifying data throughout the entire manufacturing process of power transmission and transformation equipment, the method comprising the following steps:

[0009] Step S1: obtaining original business data generated in each link of power transmission and transformation equipment manufacturing;

[0010] Perform data type unification, preliminary elimination of abnormal fluctuations and standardization processing operations on the collected data;

[0011] Step S2: The pre-processed business data is transmitted to the cloud for processing. A field semantic graph is constructed based on the industry process ontology model. Semantic mapping, field normalization, and attribute unification operations are performed on heterogeneous fields through the field semantic graph to complete semantic fusion and obtain standardized business data. Based on the current manufacturing process state parameters, a verification rule set corresponding to the business context is dispatched from the rule warehouse;

[0012] Step S3: verify the obtained standardized business data using the corresponding verification rule set;

[0013] Step S4: Mark the data record with a failed verification result as abnormal, upload the field position, violation rule number, timestamp and device ID as abnormal meta information, and generate a prompt signal on the client interface;

[0014] Step S5: Mark the verified data records with the "Verified Pass" tag and write them into the blockchain data module. Calculate the hash value of each data to generate a digest, and store it on the chain to the distributed nodes to form a chain traceability structure.

[0015] Step S6: Generate a quality check report for the manufacturing data of this batch based on the verification result vector and the statistical parameters.

[0016] Preferably, in step S3, a structured matrix is ​​constructed for the standardized business data:

[0017]

[0018] in Indicates the value of the jth field in the i-th business record. The check rule set is expressed as:

[0019]

[0020] in An expression function representing a rule, is the corresponding threshold, is a logical operation relationship; call the rule engine module to match the corresponding rule expression for all data matrix elements one by one to obtain the verification result vector:

[0021]

[0022] in Indicates that the i-th record satisfies all scheduling rules, Indicates that at least one rule is not satisfied.

[0023] Preferably, the data preprocessing operation in step S1 includes the following sub-steps:

[0024] Step S11: perform field alignment and null value interpolation operations on each type of original data, and use the interpolation strategy based on the K-nearest neighbor method to handle missing fields, where the missing field values The calculation method is:

[0025]

[0026] in, Represents the jth field value of the kth similar record;

[0027] Step S12: Smoothing time series fluctuations based on the sliding window method, with a window length of , the weighted moving average method is used for processing, and the output result is:

[0028]

[0029] Among them is Preset weight factors to meet , and is used to weaken the impact of sampling errors.

[0030] Preferably, the semantic fusion in step S2 includes field normalization based on ontology mapping, wherein the semantic ontology is constructed by domain experts and uses OWL language to represent entity-attribute-relationship triples, wherein each field Mapping to target ontology classes via SPARQL query interface And do the following:

[0031] Identify field synonyms and abbreviation mappings;

[0032] Normalized units and dimension conversion;

[0033] Merge logically equivalent fields;

[0034] The above operation results in a field-attribute mapping set ;

[0035] The rule scheduling method builds a rule activation graph based on the current business context , where the vertex set Represents rule set, edge set Indicates dependency or mutual exclusion between rules;

[0036] By constructing the state variable vector , using the matching function:

[0037]

[0038] Filter the currently applicable rule set and activate the path to trigger the rule matrix Indicates that it is used by the subsequent rule engine to execute scheduling;

[0039] in, To activate a rule set, is the i-th rule, is the rule trigger matrix.

[0040] Preferably, the rule checking engine of the S3 step uses an expression tree structure to store all rule expressions, wherein each rule expression is parsed into a Boolean tree structure, whose leaf nodes are field values ​​or constant nodes, and the intermediate nodes are logical operators; during the execution process, a depth-first traversal strategy is used to evaluate each branch expression in turn, and if the calculation result of the root node is true, the rule is considered to have hit; this mechanism supports expression-level dynamic compilation, and registers an abstract syntax tree inside the rule engine for runtime call.

[0041] Preferably, the step S3 further includes preprocessing the high-dimensional business data by reducing the dimensionality based on principal component analysis to discover potential abnormal samples, and the specific steps are as follows:

[0042] Calculate the covariance matrix of the original data matrix X ;

[0043] Solve the principal component vector P and construct the principal component space ;

[0044] Calculate the Mahalanobis distance:

[0045]

[0046] like > , the sample is considered abnormal and the process goes to step S4 for abnormal feedback processing;

[0047] in, is the principal component vector of the i-th sample, is the principal component mean vector, is the abnormality judgment threshold.

[0048] Preferably, the abnormal feedback module in step S4 adopts a structured log mechanism to number, locate the field position, and archive the rule trigger ID and original value of each business data record that fails the verification, and generate a hash summary for its abnormal event:

[0049]

[0050] Among them, ID is the business primary key, t is the timestamp of the event, f is the field name, and v is the original value of the field.

[0051] In a second aspect, the present application provides a data verification system for the entire process of power transmission and transformation equipment manufacturing, the system comprising:

[0052] The data acquisition module is used to obtain raw business data in real time from the raw materials, processing, and testing stages of power transmission and transformation equipment manufacturing, and pre-process the acquired data, including field standardization, null value interpolation, format normalization, and preliminary outlier removal.

[0053] A cloud processing module is used to receive the standardized data transmitted by the data acquisition module and call the semantic fusion module to perform semantic mapping, attribute unification and context normalization on data fields from different sources, while scheduling the corresponding data verification rule set according to the manufacturing context;

[0054] The data verification module is used to perform logical comparison and rule matching between the integrated business data and the scheduled rule set, perform field-level and record-level data quality verification, and generate corresponding verification results;

[0055] The exception feedback module is used to receive data records that do not meet the verification rules output by the data verification module, extract their field location information, trigger rule information, time information and source identification, and submit the abnormal data to the alarm module and client interface for prompting;

[0056] The blockchain storage module is used to mark the data records that have passed the verification and generate their summary information. The summary information and data records are then written to the blockchain to achieve data traceability, secure evidence storage and version control;

[0057] The report generation module is used to generate a batch quality verification report based on the data verification results. The report includes indicators such as field compliance rate, exception type distribution, rule hit statistics, etc., and is presented in a structured form through the client display interface.

[0058] In a third aspect, the present application provides a terminal, including:

[0059] Memory, used to store data verification procedures for the entire process of power transmission and transformation equipment manufacturing;

[0060] A processor is used to implement the steps of the data verification method for the whole process of power transmission and transformation equipment manufacturing as described in the first aspect when executing the data verification system for the whole process of power transmission and transformation equipment manufacturing.

[0061] In a fourth aspect, the present application provides a computer-readable storage medium, which stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the full-process data verification method for power transmission and transformation equipment manufacturing as described in the first aspect.

[0062] It can be seen from the above technical solutions that the advantages of the present invention are:

[0063] (1) A field normalization mechanism based on domain ontology model and semantic graph is introduced, which can automatically identify the synonymy, unit inconsistency and structural differences of fields between different systems, ensuring that heterogeneous data can achieve unified semantic representation before being passed into the verification engine, so that data from multiple systems such as design, process, testing, and quality inspection can be integrated, evaluated, and collaboratively verified, avoiding the problem of "field mismatch and data mismatch" in traditional systems.

[0064] (2) Before performing data quality verification, principal component analysis and isolation forest algorithm are introduced, and abnormal samples are identified by combining the covariance relationship between fields and structural characteristics. This effectively improves the recognition accuracy of complex problems such as nonlinear anomalies, isolated data points, and cross-field conflicts. It can assist the verification rules in filtering systematic deviations and random anomalies, significantly reducing the misjudgment rate and missed detection rate.

[0065] (3) By constructing a rule graph containing rule nodes and their contextual dependencies and introducing a rule activation mechanism based on the business state vector, the system can dynamically call a subset of matching rules for verification based on factors such as the current process, equipment status, and process segment, avoiding the problems of traditional systems such as "one-size-fits-all rules" or "static rules that are out of touch with reality", and improving rule adaptability and verification efficiency.

[0066] (4) Utilize the structured exception log mechanism and blockchain distributed storage to generate summary records with metadata such as timestamp, field location, and source device for all data that fails verification, and complete hash evidence and node signature verification on the chain to ensure that abnormal events are verifiable and non-repudiable in subsequent quality problem analysis, review, and supervision scenarios.

[0067] (5) After completing the quality check, a data quality report covering multiple indicators such as field compliance rate, abnormal thermal distribution, rule hit frequency, abnormal cluster type, etc. can be automatically generated, and visualized through the client interface in the form of charts, heat maps, indicator cards, etc., to assist management personnel to quickly locate the source of the problem and optimize the process flow, which has decision-making support value. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for the description. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0069] Figure 1A flow chart of a method for verifying data throughout the entire manufacturing process of power transmission and transformation equipment provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0070] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0071] See also Figure 1 As shown, the present invention provides a data verification method for the entire process of power transmission and transformation equipment manufacturing, comprising the following steps:

[0072] Step S1: obtaining original business data generated in each link of power transmission and transformation equipment manufacturing;

[0073] Perform data type unification, preliminary elimination of abnormal fluctuations and standardization processing operations on the collected data;

[0074] Step S2: The pre-processed business data is transmitted to the cloud for processing. A field semantic graph is constructed based on the industry process ontology model. Semantic mapping, field normalization, and attribute unification operations are performed on heterogeneous fields to complete semantic fusion. Based on the current manufacturing process state parameters, a verification rule set corresponding to the business context is dispatched from the rule warehouse.

[0075] Step S3: Verify the obtained standardized business data using the corresponding verification rule set, including:

[0076] Construct a structured matrix for standardized business data:

[0077]

[0078] in Indicates the value of the jth field in the i-th business record. The check rule set is expressed as:

[0079]

[0080] in An expression function representing a rule, is the corresponding threshold, is a logical operation relationship; call the rule engine module to match the corresponding rule expression for all data matrix elements one by one to obtain the verification result vector:

[0081]

[0082] in Indicates that the i-th record satisfies all scheduling rules, Indicates that at least one rule is not satisfied;

[0083] Step S4: Mark the data record with a failed verification result as abnormal, upload the field position, violation rule number, timestamp and device ID as abnormal meta information, and generate a prompt signal on the client interface;

[0084] Step S5: Mark the verified data records with the "Verified Pass" tag and write them into the blockchain data module. Calculate the hash value of each data to generate a digest, and store it on the chain to the distributed nodes to form a chain traceability structure.

[0085] Step S6: Generate a quality check report for this batch of manufacturing data based on the verification result vector and statistical parameters. The report is pushed to the client interface and supports interface call download.

[0086] In some embodiments, the data preprocessing operation in step S1 includes the following sub-steps:

[0087] Step S11: perform field alignment and null value interpolation operations on each type of original data, and use the interpolation strategy based on the K-nearest neighbor method to handle missing fields, where the missing field values The calculation method is:

[0088]

[0089] in, Represents the jth field value of the kth similar record;

[0090] Step S12: Smoothing time series fluctuations based on the sliding window method, with a window length of , the weighted moving average method is used for processing, and the output result is:

[0091]

[0092] Among them is Preset weight factors to meet , and is used to weaken the impact of sampling errors.

[0093] In some embodiments, the semantic fusion in step S2 includes field normalization based on ontology mapping, wherein the semantic ontology is constructed by domain experts and uses OWL language to represent entity-attribute-relationship triples, wherein each field Mapping to target ontology classes via SPARQL query interface And do the following:

[0094] Identify field synonyms and abbreviation mappings;

[0095] Normalized units and dimension conversion;

[0096] Merge logically equivalent fields;

[0097] The above operation results in a field-attribute mapping set ;

[0098] The rule scheduling method builds a rule activation graph based on the current business context , where the vertex set Represents rule set, edge set Indicates dependency or mutual exclusion between rules;

[0099] By constructing the state variable vector , using the matching function:

[0100]

[0101] Filter the currently applicable rule set and activate the path to trigger the rule matrix Indicates that it is used by the subsequent rule engine to execute scheduling;

[0102] in, To activate a rule set, is the i-th rule, is the rule trigger matrix.

[0103] In some embodiments, the rule checking engine of the S3 step uses an expression tree structure to store all rule expressions, where each rule expression is parsed into a Boolean tree structure, whose leaf nodes are field values ​​or constant nodes, and the intermediate nodes are logical operators; during the execution process, a depth-first traversal strategy is used to evaluate each branch expression in turn, and if the calculation result of the root node is true, the rule is considered to have hit; this mechanism supports expression-level dynamic compilation, and registers an abstract syntax tree inside the rule engine for runtime call.

[0104] In some embodiments, the step S3 further includes performing dimensionality reduction preprocessing on the high-dimensional business data based on principal component analysis to discover potential abnormal samples. The specific steps are as follows:

[0105] Calculate the covariance matrix of the original data matrix X ;

[0106] Solve the principal component vector P and construct the principal component space ;

[0107] Calculate the Mahalanobis distance:

[0108]

[0109] like > , the sample is considered abnormal and the process goes to step S4 for abnormal feedback processing;

[0110] in, is the principal component vector of the i-th sample, is the principal component mean vector, is the abnormality judgment threshold.

[0111] In some embodiments, the exception feedback module in step S4 uses a structured log mechanism to number, locate the field position, and archive the rule trigger ID and original value of each business data record that fails verification, and generates a hash summary for its exception event:

[0112]

[0113] Among them, ID is the business primary key, t is the timestamp of the event, f is the field name, and v is the original value of the field.

[0114] In some embodiments, the data that has passed the verification in step S5 is organized in a Merkle tree structure and forms a root hash , the root hash is encapsulated into a new block together with the timestamp and batch identifier, generating the following structure:

[0115]

[0116] in, is the data structure of the tth block, Generate timestamps for blocks, is the current batch identifier, Sign the node.

[0117] In some embodiments, the data item to be uploaded to the chain in step S5 is compressed and encoded using the CBOR format, and the hash value is calculated before and after encoding to verify the integrity. The hash value before compression is:

[0118]

[0119] in This is the i-th verified business record. The hash is used to verify whether structural damage occurs during the CBOR compression process. The compressed data is used for on-chain storage and a posteriori reconstruction.

[0120] In some embodiments, the abnormality feedback module in S4 is linked to the client interface to display a field-level abnormality heat map, and each heat value in the image rendering is calculated by the function:

[0121]

[0122] Calculated, where Representation field The number of anomalies in the time window t, Indicates the maximum anomaly count per unit time for any field; the range of the color gradient mapping function C is [0,1].

[0123] In some embodiments, the step S3 further includes identifying structural anomalies using an isolation forest algorithm and executing the following scoring function:

[0124]

[0125] in is the average path length of sample x isolated in multiple random cut trees, c(n) is the expected path normalization coefficient, and n is the total number of samples; if , then the sample x is judged to be structurally abnormal.

[0126] In some embodiments, the rule engine includes a reinforcement learning module that uses a Q-learning algorithm to dynamically optimize rule selection weights. The update formula is:

[0127]

[0128] in It represents the current valuation of taking action a in state s, r is the actual feedback result, α is the learning rate, γ is the discount coefficient, s′ is the next state, and a′ is the next action, which is used to optimize the activation order of rules.

[0129] The inter-field semantic completion process is modeled using a graph neural network (GNN), and the graph structure is defined as:

[0130]

[0131] in is a collection of field nodes. Indicates field co-occurrence or dependency relationships, For fields The embedding vector in this figure is subsequently used for missing value prediction and field verification through similarity measurement with missing fields.

[0132] In some embodiments, the structured verification report generated in step S6 includes the abnormality type label after clustering by the DBSCAN algorithm, and the clustering core parameter is the density radius With the minimum number of points minPts, and output the silhouette coefficient S to evaluate the clustering effectiveness:

[0133]

[0134] Where a(i) is the average distance from sample i to other points in the same cluster, and b(i) is the average distance from sample i to the nearest cluster.

[0135] In some embodiments, the present application provides a data verification system for the entire manufacturing process of power transmission and transformation equipment, the system comprising:

[0136] The data acquisition module is used to obtain raw business data in real time from the raw materials, processing, and testing stages of power transmission and transformation equipment manufacturing, and pre-process the acquired data, including field standardization, null value interpolation, format normalization, and preliminary outlier removal.

[0137] A cloud processing module is used to receive the standardized data transmitted by the data acquisition module and call the semantic fusion module to perform semantic mapping, attribute unification and context normalization on data fields from different sources, while scheduling the corresponding data verification rule set according to the manufacturing context;

[0138] The data verification module is used to perform logical comparison and rule matching between the integrated business data and the scheduled rule set, perform field-level and record-level data quality verification, and generate corresponding verification results;

[0139] The exception feedback module is used to receive data records that do not meet the verification rules output by the data verification module, extract their field location information, trigger rule information, time information and source identification, and submit the abnormal data to the alarm module and client interface for prompting;

[0140] The blockchain storage module is used to mark the data records that have passed the verification and generate their summary information. The summary information and data records are then written to the blockchain to achieve data traceability, secure evidence storage and version control;

[0141] The report generation module is used to generate a batch quality verification report based on the data verification results. The report includes indicators such as field compliance rate, exception type distribution, rule hit statistics, etc., and is presented in a structured form through the client display interface.

[0142] In some embodiments, the present application provides a terminal, including:

[0143] Memory, used to store data verification procedures for the entire process of power transmission and transformation equipment manufacturing;

[0144] A processor is used to implement the steps of the data verification method for the entire process of power transmission and transformation equipment manufacturing when executing the data verification system for the entire process of power transmission and transformation equipment manufacturing.

[0145] In some embodiments, the present application provides a computer-readable storage medium that stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the method for verifying data throughout the entire manufacturing process of power transmission and transformation equipment.

[0146] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the described specific embodiments or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they should all fall within the scope of protection of the present invention.

Claims

1. A data verification method for the entire process of power transmission and transformation equipment manufacturing, characterized in that: The following steps are involved: Step S1: obtaining original business data generated in each link of power transmission and transformation equipment manufacturing; Perform data type unification, preliminary elimination of abnormal fluctuations and standardization processing operations on the collected data; Step S2: The pre-processed business data is transmitted to the cloud for processing. A field semantic graph is constructed based on the industry process ontology model. Semantic mapping, field normalization, and attribute unification operations are performed on heterogeneous fields through the field semantic graph to complete semantic fusion and obtain standardized business data. Based on the current manufacturing process state parameters, a verification rule set corresponding to the business context is dispatched from the rule warehouse; The semantic fusion in step S2 includes field normalization based on ontology mapping. The semantic ontology is constructed by domain experts and uses OWL language to represent entity-attribute-relationship triples, where each field Mapping to target ontology classes via SPARQL query interface And do the following: Identify field synonyms and abbreviation mappings; Normalized units and dimension conversion; Merge logically equivalent fields; The above operation results in a field-attribute mapping set ; The rule scheduling method builds a rule activation graph based on the current business context , where the vertex set Represents rule set, edge set Indicates dependency or mutual exclusion between rules; By constructing the state variable vector , using the matching function: Filter the currently applicable rule set and activate the path to trigger the rule matrix Indicates that it is used for subsequent verification of rule set execution scheduling; in, To activate a rule set, is the i-th rule, is the rule trigger matrix; Step S3: verify the obtained standardized business data using the corresponding verification rule set; Construct a structured matrix for standardized business data: in Indicates the value of the jth field in the i-th business record. The check rule set is expressed as: in An expression function representing a rule, is the corresponding threshold, is a logical operation relationship; call the verification rule set to match the corresponding rule expression for all data matrix elements one by one to obtain the verification result vector: in Indicates that the i-th record satisfies all scheduling rules, Indicates that at least one rule is not satisfied; The check rule set in step S3 uses an expression tree structure to store all rule expressions, where each rule expression is parsed into a Boolean tree structure, with leaf nodes being field values ​​or constant nodes and intermediate nodes being logical operators. During execution, a depth-first traversal strategy is used to sequentially evaluate each branch expression. If the root node evaluates to true, the rule is considered a hit. The check rule set supports expression-level dynamic compilation during execution, and registers an abstract syntax tree within the check rule set for runtime invocation. Step S4: Mark the data record with a failed verification result as abnormal, upload the field position, violation rule number, timestamp and device ID as abnormal meta information, and generate a prompt signal on the client interface; Step S5: Mark the verified data records with the "Verified Pass" tag and write them into the blockchain data module. Calculate the hash value of each data to generate a digest, and store it on the chain to the distributed nodes to form a chain traceability structure. Step S6: Generate a quality check report for the manufacturing data of this batch based on the verification result vector and the statistical parameters.

2. The data verification method for the entire process of power transmission and transformation equipment manufacturing according to claim 1 is characterized in that: The S3 step also includes preprocessing the high-dimensional business data by reducing its dimensionality based on principal component analysis to discover potential abnormal samples. The specific steps are as follows: Calculate the covariance matrix of the original data matrix X ; Solve the principal component vector P and construct the principal component space ; Calculate the Mahalanobis distance: like > , the sample is considered abnormal and the process goes to step S4 for abnormal feedback processing; in, is the principal component vector of the i-th sample, is the principal component mean vector, is the abnormality judgment threshold.

3. The data verification method for the entire process of power transmission and transformation equipment manufacturing according to claim 1 is characterized in that: The exception feedback module in step S4 uses a structured log mechanism to number, locate the field position, and archive the rule trigger ID and original value of each business data record that fails verification, and generates a hash summary for its exception event: Among them, ID is the business primary key, t is the timestamp of the event, f is the field name, and v is the original value of the field.

4. A terminal, characterized in that: include: Memory, used to store data verification procedures for the entire process of power transmission and transformation equipment manufacturing; A processor is used to implement the steps of the data verification method for the whole process of power transmission and transformation equipment manufacturing as claimed in claim 1 when executing the data verification system for the whole process of power transmission and transformation equipment manufacturing.

5. A computer-readable storage medium, characterized in that The storage medium stores computer instructions. When the computer reads the computer instructions in the storage medium, the computer executes the data verification method for the entire process of power transmission and transformation equipment manufacturing as described in claim 1.

Images