Hardware function parameter identification and diagnosis method and device based on big data technology

Abstract

The present application relates to the technical field of electrical digital data processing, and discloses a method and device for hardware function parameter identification and diagnosis based on big data technology. The method includes: performing two-way cross-validation identification and parameter sensitivity quantification processing on the function parameter feature matrix and the function parameter mapping rule set to obtain a function parameter identification model and a function parameter evaluation system; performing multi-level parameter diagnosis and dynamic threshold analysis processing on the output results of the function parameter identification model to obtain a function parameter diagnosis report and a parameter tuning strategy library; performing fault propagation path analysis and correction scheme generation processing on the abnormal parameters in the function parameter diagnosis report to obtain a function parameter correction scheme set and a diagnosis rule library, and performing parameter identification accuracy verification and diagnosis effectiveness evaluation processing to obtain a hardware function parameter identification and diagnosis conclusion report. The present application improves the accurate identification and effective diagnosis of hardware function parameters based on multi-source heterogeneous data.

Description

Technical Field

[0001] The present application relates to the field of electrical digital data processing technology, and in particular to a method and device for identifying and diagnosing hardware function parameters based on big data technology. Background Art

[0002] In the current hardware system management, the identification and diagnosis of functional parameters are the key links to ensure the stable operation of the system. The existing technology mainly adopts a single parameter monitoring method, which detects anomalies by setting fixed thresholds, or uses simple statistical analysis methods to study parameter correlation. Some systems have introduced machine learning algorithms for parameter identification, but most of them are limited to specific scenarios and fixed parameter combinations. At the same time, some systems try to use expert systems for fault diagnosis, and judge and process abnormal parameters through preset rule bases. These methods have achieved certain results in processing structured data, performing single parameter analysis and local fault diagnosis.

[0003] However, the relevant technology has obvious shortcomings: first, it lacks the ability to comprehensively process multi-source heterogeneous data, making it difficult to effectively integrate different types of information such as equipment operation data, technical documents and historical records; second, the complex correlation between parameters is not considered during the parameter identification process, resulting in low reliability of the identification results; third, the diagnosis process relies too much on preset rules and lacks adaptive learning and dynamic adjustment capabilities; finally, the analysis of the fault propagation path and impact range is not in-depth enough, making it difficult to accurately locate the fault and generate effective correction plans. Summary of the invention

[0004] The present application provides a hardware function parameter identification and diagnosis method and device based on big data technology, which is used to achieve accurate identification and effective diagnosis of hardware function parameters based on multi-source heterogeneous data.

[0005] In the first aspect, the present application provides a hardware function parameter identification and diagnosis method based on big data technology, and the hardware function parameter identification and diagnosis method based on big data technology includes: distributed collection and quality rating processing of operation data, technical documents and historical records of hardware equipment to obtain a standardized function parameter data set and a function parameter classification dictionary; feature decomposition and parameter correlation calculation processing of the standardized function parameter data set and the function parameter classification dictionary to obtain a function parameter feature matrix and a function parameter mapping rule set; two-way cross-validation identification and parameter sensitivity quantification processing of the function parameter feature matrix and the function parameter mapping rule set to obtain a function parameter identification model and a function parameter evaluation system; wherein the Bidirectional cross-validation identification includes forward function mapping and reverse parameter tracing; the output results of the function parameter identification model are subjected to multi-level parameter diagnosis and dynamic threshold analysis based on the function parameter evaluation system to obtain a function parameter diagnosis report and a parameter tuning strategy library; wherein the multi-level parameter diagnosis includes single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis; the abnormal parameters in the function parameter diagnosis report are subjected to fault propagation path analysis and correction plan generation based on the parameter tuning strategy library to obtain a function parameter correction plan set and a diagnosis rule library; the function parameter correction plan set and the diagnosis rule library are subjected to parameter identification accuracy verification and diagnosis effectiveness evaluation to obtain a hardware function parameter identification diagnosis conclusion report.

[0006] In a second aspect, the present application provides a hardware function parameter identification and diagnosis device based on big data technology, and the hardware function parameter identification and diagnosis device based on big data technology includes:

[0007] The rating module is used to perform distributed collection and quality rating processing on the operating data, technical documents and historical records of hardware devices to obtain a standardized function parameter data set and a function parameter classification dictionary;

[0008] A calculation module, used for performing feature decomposition and parameter correlation calculation processing on the standardized function parameter data set and the function parameter classification dictionary to obtain a function parameter feature matrix and a function parameter mapping rule set;

[0009] An identification module is used to perform two-way cross-validation identification and parameter sensitivity quantification processing on the functional parameter feature matrix and the functional parameter mapping rule set to obtain a functional parameter identification model and a functional parameter evaluation system; wherein the two-way cross-validation identification includes forward function mapping and reverse parameter tracing;

[0010] A diagnosis module, which is used to perform multi-level parameter diagnosis and dynamic threshold analysis on the output results of the functional parameter identification model based on the functional parameter evaluation system to obtain a functional parameter diagnosis report and a parameter tuning strategy library; wherein the multi-level parameter diagnosis includes single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis;

[0011] A generation module, used for performing fault propagation path analysis and correction scheme generation processing on the abnormal parameters in the function parameter diagnosis report based on the parameter tuning strategy library, and obtaining a function parameter correction scheme set and a diagnosis rule library;

[0012] The evaluation module is used to perform parameter identification accuracy verification and diagnostic effectiveness evaluation processing on the functional parameter correction solution set and the diagnostic rule base to obtain a hardware functional parameter identification diagnosis conclusion report.

[0013] In the technical solution provided by this application, through distributed collection and quality rating processing, efficient collection and preprocessing of hardware equipment operation data, technical documents and historical records are realized, ensuring the integrity and reliability of the data source. Through feature decomposition and parameter correlation calculation, the intrinsic connection between parameters is deeply excavated, and the functional parameter feature matrix and mapping rule set are effectively extracted, laying a solid foundation for subsequent analysis. Adopting a two-way cross-validation recognition mechanism, combined with parameter sensitivity quantitative analysis, an accurate functional parameter recognition model and a scientific evaluation system are established, in which the combination of forward function mapping and reverse parameter tracing improves the accuracy and reliability of recognition. On this basis, a multi-level parameter diagnosis strategy is adopted, including single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis, and combined with dynamic threshold analysis technology, a comprehensive diagnosis from local to overall is realized, and abnormal parameters and fault types are accurately identified. Based on the parameter tuning strategy library, fault propagation path analysis and correction scheme generation are performed, which not only finds the root cause of the fault, but also provides an effective correction scheme, greatly improving the efficiency of fault handling. Finally, through parameter recognition accuracy verification and diagnostic effectiveness evaluation, a complete diagnostic conclusion report is formed to ensure the reliability and practicality of the diagnostic results. The entire solution forms a closed-loop diagnostic system with each link closely linked and supporting each other, significantly improving the accuracy, real-time and reliability of hardware function parameter identification and fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0015] Figure 1This is a schematic diagram of an embodiment of a method for identifying and diagnosing hardware function parameters based on big data technology in an embodiment of the present application;

[0016] Figure 2 This is a schematic diagram of an embodiment of a hardware function parameter identification and diagnosis device based on big data technology in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The embodiment of the present application provides a method and device for identifying and diagnosing hardware function parameters based on big data technology. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0018] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 , an embodiment of the hardware function parameter identification and diagnosis method based on big data technology in the embodiment of the present application includes:

[0019] Step S101, performing distributed collection and quality rating processing on the operation data, technical documents and historical records of the hardware equipment to obtain a standardized function parameter data set and a function parameter classification dictionary;

[0020] Step S102, performing feature decomposition and parameter correlation calculation processing on the standardized function parameter data set and the function parameter classification dictionary to obtain a function parameter feature matrix and a function parameter mapping rule set;

[0021] Step S103, performing two-way cross-validation identification and parameter sensitivity quantification processing on the function parameter feature matrix and the function parameter mapping rule set to obtain a function parameter identification model and a function parameter evaluation system; wherein the two-way cross-validation identification includes forward function mapping and reverse parameter tracing;

[0022] Step S104: Perform multi-level parameter diagnosis and dynamic threshold analysis on the output results of the function parameter identification model based on the function parameter evaluation system to obtain a function parameter diagnosis report and a parameter tuning strategy library; wherein the multi-level parameter diagnosis includes single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis;

[0023] Step S105: Perform fault propagation path analysis and correction scheme generation processing on the abnormal parameters in the function parameter diagnosis report based on the parameter tuning strategy library to obtain a function parameter correction scheme set and a diagnosis rule library;

[0024] Step S106: perform parameter recognition accuracy verification and diagnostic effectiveness evaluation on the function parameter correction solution set and the diagnostic rule base to obtain a hardware function parameter recognition diagnostic conclusion report.

[0025] It is understandable that the execution subject of the present application can be a hardware function parameter identification and diagnosis device based on big data technology, or a terminal or a server, which is not limited here. The present application embodiment is described by taking the server as the execution subject as an example.

[0026] Specifically, data is collected from hardware devices through distributed nodes, including real-time data streams generated by device operation, device technical description documents, and historical operation records. During the collection process, large-scale data is distributed to multiple computing nodes through sharding, and each node is responsible for processing specific data segments. Specifically, the operation data is collected by data sharding, and the data stream is divided into time windows. The size of each shard is 100MB, and the overlap rate of adjacent shards is set to 10% to ensure data continuity. For technical documents, text parsing technology is used to extract structured information, including key information such as device parameter specifications and operating condition restrictions. Historical records are extracted by time series data extraction, and historical parameter values ​​are extracted according to timestamp indexes. In the data quality rating stage, the collected data is evaluated in multiple dimensions, including indicators such as completeness, accuracy, and timeliness. By setting scoring rules, the quality score is calculated for each piece of data, and high-quality data is screened out for subsequent analysis. For example, the temperature parameter of a certain model of sensor has a normal value range of 20-80℃ and a sampling frequency of 1 time / second. Through outlier detection, it is found that there are outliers of 85℃ in the data. Such data will be marked and specially processed. The standardized functional parameter data is subjected to feature decomposition processing to decompose the complex parameter features into basic feature combinations. On this basis, the correlation between parameters is calculated and a parameter association network is constructed. The feature decomposition uses the principal component analysis method to extract the main feature vectors; the correlation calculation uses the Pearson correlation coefficient method to establish the correlation strength matrix between parameters. The two-way cross-validation recognition process includes two links: forward function mapping and reverse parameter tracing. The forward mapping process starts from the functional characteristics and tracks the corresponding parameter combination; the reverse tracing starts from the parameter changes and analyzes the affected functional characteristics. The recognition accuracy is improved through two-way validation. Parameter sensitivity quantification mainly analyzes the impact of parameter changes on the function and establishes sensitivity evaluation indicators.

[0027] Multi-level parameter diagnosis analyzes from three levels: single parameter, parameter combination, and system level. Single parameter diagnosis focuses on whether the parameter exceeds the threshold range, parameter combination diagnosis analyzes the synergistic relationship between parameters, and system-level diagnosis focuses on the impact of parameter changes on the overall system. Dynamic threshold analysis dynamically adjusts parameter thresholds according to the system operating status. Fault propagation path analysis tracks the spread process and impact range of faults by constructing a fault propagation network. Correction plans are generated based on fault propagation characteristics, and the optimal plan is selected through actual verification. Finally, the accuracy and effectiveness of the correction plan are evaluated and verified to form a complete diagnostic conclusion report.

[0028] Taking a certain model of industrial air-conditioning system as an example, functional parameter identification and fault diagnosis analysis are carried out: the operating data of core components such as compressors, condensers, and evaporators are collected through distributed nodes. 3-5 sensors are set for each component, and the sampling interval is 0.1 seconds. After data preprocessing and quality rating, more than 95% of valid data are retained. After feature decomposition, basic features such as temperature, pressure, and flow are obtained. Parameter association analysis finds that the correlation coefficient between the compressor outlet pressure and the condenser inlet temperature reaches 0.85, indicating a strong correlation. When the compressor outlet pressure is detected to be abnormally high, the fault propagation analysis finds that it is caused by poor heat dissipation of the condenser. Based on this, correction plans including cleaning the condenser and adjusting the fan speed are generated. The final diagnostic report shows that the accuracy of the correction plan reached 92%, which effectively solved the equipment failure problem.

[0029] In the embodiment of the present application, through distributed collection and quality rating processing, efficient collection and preprocessing of hardware equipment operation data, technical documents and historical records are achieved, ensuring the integrity and reliability of the data source. Through feature decomposition and parameter correlation calculation, the intrinsic connection between parameters is deeply excavated, and the functional parameter feature matrix and mapping rule set are effectively extracted, laying a solid foundation for subsequent analysis. Adopting a two-way cross-validation recognition mechanism, combined with parameter sensitivity quantitative analysis, an accurate functional parameter recognition model and a scientific evaluation system are established, in which the combination of forward function mapping and reverse parameter tracing improves the accuracy and reliability of recognition. On this basis, a multi-level parameter diagnosis strategy is adopted, including single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis, and dynamic threshold analysis technology is used to achieve a comprehensive diagnosis from local to overall, accurately identifying abnormal parameters and fault types. Based on the parameter tuning strategy library, fault propagation path analysis and correction scheme generation are performed, which not only finds the root cause of the fault, but also provides an effective correction scheme, greatly improving the efficiency of fault handling. Finally, through parameter recognition accuracy verification and diagnostic effectiveness evaluation, a complete diagnostic conclusion report is formed to ensure the reliability and practicality of the diagnostic results. The entire solution forms a closed-loop diagnostic system with each link closely linked and supporting each other, significantly improving the accuracy, real-time and reliability of hardware function parameter identification and fault diagnosis.

[0030] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0031] (1) Perform data slicing and collection processing on the operation data of the hardware equipment to obtain the operation parameter data stream, and perform outlier detection processing on the operation parameter data stream to obtain the filtered operation data set;

[0032] (2) Perform text parsing on the technical documents of the hardware equipment to obtain a structured technical parameter table, and perform data format unification on the structured technical parameter table to obtain a standard technical parameter set;

[0033] (3) Perform time series data extraction on the historical records of the hardware equipment to obtain a historical parameter sequence, and perform data completion on the historical parameter sequence to obtain the target historical data set;

[0034] (4) Perform data fusion processing on the screening operation data set, standard technical parameter set and target historical data set to obtain a comprehensive parameter data table, and perform quality scoring processing on the comprehensive parameter data table to obtain a parameter quality rating table;

[0035] (5) Perform parameter classification and annotation processing on the parameter quality rating table to obtain a functional parameter classification dictionary, and perform hierarchical relationship construction processing on the functional parameter classification dictionary to obtain a parameter type hierarchy table;

[0036] (6) Perform distributed verification processing on the parameter type hierarchy table to obtain a verification parameter set, and perform data standardization processing on the verification parameter set to obtain a standardized functional parameter data set.

[0037] Specifically, the operation data is collected in pieces. Sliced ​​collection means that the continuous data stream is divided into several data blocks of fixed size. The size of each data block is 64MB, and there is a 10% overlap area between adjacent data blocks to ensure the continuity of the data. After the data is sliced, it is assigned to different computing nodes for parallel processing. Each node is responsible for the processing tasks of one or more data slices. The collected operation parameter data stream is detected for outliers. By calculating the mean and standard deviation of the data, a range of 3 times the standard deviation is set as the outlier judgment standard. Data points outside this range are marked as outliers. After screening, the cleaned operation data set is obtained. For the technical documents of hardware equipment, natural language processing technology is used for parsing to extract key information such as parameter type, parameter value range, and unit in the document. The text parsing process includes steps such as word segmentation, keyword extraction, and semantic analysis to convert unstructured text into a structured technical parameter table. The structured parameter table needs to be formatted in a unified manner, including unit conversion, numerical standardization, etc., to ensure that the expression of all parameters is consistent, which is convenient for subsequent analysis and processing.

[0038] For historical record data, it is extracted according to the time series characteristics. Each record contains information such as timestamp, parameter value, and status mark. Since there are often missing values ​​in historical data, data completion processing is required. Data completion uses linear interpolation. For missing data with a short time interval, the values ​​of adjacent time points are used for interpolation calculation; for missing data with a long time interval, historical data of the same type of equipment is used for filling. The data fusion link integrates the three types of data sources into a unified data table. The fusion process needs to solve problems such as data alignment and time synchronization. Data alignment refers to matching data of the same parameter from different sources, and time synchronization ensures that the time tags of different data sources are consistent. The fused comprehensive parameter data table contains fields such as parameter name, parameter value, data source, and timestamp. The comprehensive parameter data table is scored for quality. The scoring indicators include data completeness, accuracy, timeliness, etc. The weights of each indicator are 0.4, 0.3, and 0.3 respectively. The final quality score is obtained through weighted calculation.

[0039] The parameter classification and annotation process classifies parameters according to their functional attributes, such as performance parameters, state parameters, environmental parameters, etc. Each parameter can belong to one or more categories. On this basis, a hierarchical relationship between parameters is constructed to form a tree-like hierarchy. The upper nodes represent the major parameter categories, and the lower nodes represent the specific parameter types. The construction of hierarchical relationships helps to understand the subordinate and influence relationships between parameters. Finally, the parameter type hierarchy table is verified through distributed nodes, and the verification content includes the accuracy of parameter classification and the rationality of hierarchical relationships. The parameters that pass the verification are standardized, and the parameters of different dimensions are converted to a unified scale space for subsequent analysis and comparison.

[0040] Taking the joint motor of a certain model of industrial robot as an example, during the data collection stage, the motor's current, voltage, speed, temperature and other parameters are collected every 0.1 second, and the original data flow rate is 1MB / s. The data is divided into multiple 64MB data blocks through slicing processing. The abnormal value detection finds that the current value fluctuates instantaneously at certain times, exceeding the normal range (3 times the standard deviation). These abnormal points are marked and filtered. Technical document analysis extracts the motor's rated voltage, rated current, maximum speed and other specification parameters, and converts the units uniformly (such as the speed is unified to rad / s). Historical records show that the motor has 27 start and stop records in the past week, of which 5 data missing are found and supplemented by linear interpolation method. After data fusion, a comprehensive parameter table containing real-time operation data, technical specifications and historical records is formed, and the quality score is calculated to be 0.85. Parameter classification divides motor parameters into three categories: electrical parameters, mechanical parameters and thermal parameters, and establishes a detailed hierarchical relationship diagram, and finally obtains a standardized functional parameter data set.

[0041] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0042] (1) Performing parameter stratification processing on the standardized functional parameter data set to obtain a stratified parameter set, and performing feature decomposition processing on the stratified parameter set to obtain a basic feature set;

[0043] (2) Cross-matching the basic feature set and the functional parameter classification dictionary to obtain a feature type table, and perform data dimensionality reduction on the feature type table to obtain the main feature set;

[0044] (3) Perform parameter correlation analysis on the main feature set to obtain a parameter correlation matrix, and perform threshold segmentation on the parameter correlation matrix to obtain a core correlation table;

[0045] (4) Perform feature aggregation processing on the core association table to obtain a function parameter feature matrix, and perform feature verification processing on the function parameter feature matrix to obtain a verification feature set;

[0046] (5) Perform association rule extraction on the verification feature set to obtain an initial rule set, and perform rule screening on the initial rule set to obtain a valid rule table;

[0047] (6) Perform rule integration processing on the effective rule table to obtain a functional parameter mapping rule set.

[0048] Specifically, in the feature processing process of hardware function parameters, the standardized function parameter data set is firstly subjected to parameter hierarchical processing. Parameter hierarchical processing refers to dividing the parameters according to the functional level into primary parameters (such as overall equipment performance parameters), secondary parameters (such as core component parameters) and tertiary parameters (such as specific component parameters). The hierarchical process is realized by establishing a parameter hierarchy tree, and each node contains information such as parameter name, hierarchy number, parent node identifier, etc., to form a complete parameter hierarchy structure. The obtained hierarchical parameter set is subjected to feature decomposition, and the complex function parameters are decomposed into basic feature elements, including numerical features, timing features, state features, etc. In the feature cross-matching stage, the basic feature set is associated with the function parameter classification dictionary for analysis. The cross-matching process establishes the correspondence between features and function types, and determines which function categories each feature belongs to by calculating the matching score. The matching calculation adopts the cosine similarity method. The closer the cosine value of the angle between two vectors is to 1, the higher the matching degree. The obtained feature type table contains fields such as feature ID, function type, and matching score. Then the data was reduced in dimension, and the principal component analysis method was used to extract the main feature components. During the dimension reduction, the feature components with a cumulative contribution rate of 85% were retained.

[0049] The main feature set is analyzed for parameter correlation and the degree of correlation between features is calculated. The Spearman rank correlation coefficient is used to calculate the correlation. This method is insensitive to data distribution and is suitable for the analysis of nonlinear relationships. In the obtained parameter correlation matrix, each element represents the correlation strength between the corresponding two features. The matrix is ​​segmented by setting the correlation threshold (usually 0.7), and the feature pairs with high correlation are retained to form a core correlation table. The feature aggregation process combines closely related features to form feature clusters, and each feature cluster represents a class of related functional features. The aggregation process uses a hierarchical clustering algorithm to gradually merge from bottom to top according to the correlation between features until the preset number of clusters is reached. After aggregation, the functional parameter feature matrix is ​​obtained, and each row of the matrix represents a feature cluster and each column represents a feature attribute. The feature matrix is ​​verified to check the feature discrimination, stability and other indicators.

[0050] Association rule extraction is to discover association rules between parameters from the verification feature set. The Apriori algorithm is used to mine frequent item sets, and the minimum support is set to 0.3 and the minimum confidence is set to 0.8. The rules that meet the conditions are extracted to form the initial rule set. The form of the rule is "if feature A meets the condition, then feature B meets the condition". The initial rule set is screened, redundant rules and contradictory rules are eliminated, and effective rules with practical significance are retained. Finally, the effective rules are integrated to form a complete functional parameter mapping rule set. The rule integration process includes steps such as rule merging, rule supplementation, and rule optimization to ensure the integrity and consistency of the rule set.

[0051] Taking a certain model of CNC machine tool as an example, the parameter identification process of its spindle system is as follows: First, the standardized data is layered to obtain a three-layer parameter structure including spindle speed, feed speed, cutting force, etc. Feature decomposition obtains basic features such as speed stability, load change rate, and temperature rise rate. Cross-matching found that the matching degree between spindle speed and cutting accuracy is 0.82, indicating a strong correlation. After data dimensionality reduction, 7 main features were retained, accounting for 87% of the total variance. Correlation analysis shows that the correlation between spindle temperature and bearing vibration is 0.75, and those exceeding the threshold are retained in the core association table. Feature aggregation forms a feature matrix including performance features, state features, and fault features. Finally, functional parameter mapping rules such as "when the spindle temperature rises by more than 40°C and lasts for more than 30 minutes, the bearing vibration value increases by 25%" are extracted.

[0052] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0053] (1) Performing forward function mapping processing on the function parameter feature matrix to obtain a function mapping feature table, and performing parameter tracking processing on the function mapping feature table to obtain a parameter mapping sequence;

[0054] (2) Perform reverse parameter tracing on the function parameter mapping rule set to obtain a parameter tracing relationship table, and perform rule verification on the parameter tracing relationship table to obtain a verification rule set;

[0055] (3) Perform cross-validation recognition processing on the parameter mapping sequence and the verification rule set to obtain a recognition result matrix, and perform validity verification processing on the recognition result matrix to obtain a verification result table;

[0056] (4) Perform parameter sensitivity analysis on the verification result table to obtain a sensitivity index set, and perform quantitative evaluation on the sensitivity index set to obtain a parameter quantification table;

[0057] (5) Extract association rules from the parameter quantification table to obtain a function parameter evaluation system, and construct evaluation criteria for the function parameter evaluation system to obtain a parameter evaluation index set;

[0058] (6) Perform parameter identification calculation on the parameter evaluation index set to obtain a functional parameter identification model.

[0059] Specifically, the function parameter feature matrix is ​​processed by forward function mapping. Forward function mapping refers to starting from the function attributes, tracking the parameter combination that affects the function, and decomposing each function indicator into a combination of multiple parameter features. In specific implementation, a parameter influence factor table is established for each function indicator to record the parameter influence direction and weight. For example, for the function indicator of equipment accuracy, it is necessary to map parameter features such as temperature change, vibration amplitude, and displacement error. By analyzing the contribution of these parameters to the function indicator, a mapping relationship is established. After obtaining the function mapping feature table, parameter tracking processing is performed to trace the change law and interaction relationship of each parameter to form a time-series parameter mapping sequence. Reverse parameter tracing is to reversely deduce the parameter change process from the rule set. For each rule in the function parameter mapping rule set, the constraint conditions and association relationship of the parameters in the rule are analyzed to establish a transmission link between the parameters. The tracing process needs to consider the temporal dependency of the parameters and determine the order and causal relationship of the parameter changes. The obtained parameter tracing relationship table contains information such as parameter ID, associated parameters, and constraint conditions. Perform rule verification on the parameter traceability relationship table, check the completeness and consistency of the rules, eliminate redundant and contradictory rules, and form a verification rule set.

[0060] Cross-validation recognition is to compare and verify the results of forward mapping and reverse tracing. The parameter mapping sequence is matched and analyzed with the verification rule set to verify whether the parameter changes meet the rule constraints and calculate the matching degree score. The matching analysis uses a fuzzy matching algorithm to allow a certain range of parameter fluctuations. In the obtained recognition result matrix, each element represents the matching degree between the parameter and the rule. The recognition result matrix is ​​validated, the reliability and stability of the results are analyzed, and a verification result table is generated. Parameter sensitivity analysis is used to quantify the impact of parameter changes on the function. For each parameter in the verification result table, the sensitivity of its change to the functional index is calculated. The sensitivity calculation uses the partial differential method to obtain the ratio of the parameter change rate to the function change rate. By setting the sensitivity threshold, the key influencing parameters are screened out to form a sensitivity index set. The sensitivity index set is quantitatively evaluated, a parameter importance scoring system is established, and a parameter quantification table is obtained.

[0061] Association rule extraction is to mine the association rules between parameters from the parameter quantization table. Based on the sensitivity and correlation of the parameters, a parameter evaluation index system is constructed, including evaluation dimensions such as parameter weight, threshold range, and change trend. The evaluation system is standardized, and the scoring criteria and judgment criteria are set to form a parameter evaluation index set. Finally, parameter identification calculation is performed on the parameter evaluation index set to establish a calculation model for parameter identification. The model includes functional modules such as parameter feature extraction, rule matching, and result evaluation to realize automatic identification of hardware function parameters.

[0062] Take the spindle system of a certain type of precision machining equipment as an example: First, through forward functional mapping, the parameter characteristics that affect the machining accuracy are analyzed, including spindle speed error (±0.01%), axial runout (<3μm), temperature rise rate (<0.5℃ / h), etc. Parameter tracking found that there was a time lag of 5-8 minutes between the spindle temperature rise and the bearing vibration. Reverse tracing determined the transmission link of "increased bearing vibration leads to decreased positioning accuracy". Cross-validation shows that this rule chain is established in 95% of cases. Sensitivity analysis shows that the temperature rise rate has the greatest impact on accuracy, and the sensitivity coefficient reaches 0.85. Finally, a functional parameter evaluation system containing 15 evaluation indicators was established to achieve accurate identification of spindle system parameters.

[0063] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0064] (1) Perform single parameter diagnosis processing on the output results of the functional parameter identification model to obtain a single parameter anomaly table, and perform parameter threshold analysis processing on the single parameter anomaly table to obtain a parameter threshold interval table;

[0065] (2) Perform parameter combination diagnosis processing on the parameter threshold interval table to obtain a combination parameter anomaly set, and perform combination rule verification processing on the combination parameter anomaly set to obtain a parameter combination rule table;

[0066] (3) Perform system-level linkage diagnosis processing on the parameter combination rule table to obtain a system abnormality feature set, and perform correlation analysis on the system abnormality feature set to obtain a system diagnostic feature table;

[0067] (4) Perform dynamic threshold analysis on the system diagnostic feature table to obtain a threshold change curve set, and perform parameter fluctuation analysis on the threshold change curve set to obtain a parameter fluctuation law table;

[0068] (5) The diagnostic results of the parameter fluctuation pattern table are integrated and processed to obtain a functional parameter diagnostic report, and the functional parameter diagnostic report is subjected to strategy extraction processing to obtain a parameter tuning strategy library.

[0069] Specifically, single parameter diagnosis is performed. This process analyzes each independent parameter output by the functional parameter identification model to check whether the parameter value exceeds the preset range or fluctuates abnormally. Single parameter diagnosis uses statistical analysis methods to calculate the statistical characteristics of the parameter, such as the mean, standard deviation, skewness, and kurtosis, and judges whether the parameter is abnormal by the 3σ criterion. A baseline value and an allowable fluctuation range are set for each parameter, and the parameter value that exceeds the range is marked as an abnormal point. The time, duration, degree of deviation, and other information of the abnormality are recorded to form a single parameter abnormality table. The data in the single parameter abnormality table are subjected to threshold analysis, and the adaptive threshold method is used to calculate the dynamic threshold interval according to the distribution characteristics of the historical data of the parameter. The determination of the threshold interval takes into account the time-varying characteristics of the parameter, and sets differentiated threshold ranges for the parameters under different time periods and different working conditions to obtain a parameter threshold interval table. Parameter combination diagnosis analyzes the combined state of multiple related parameters. The parameters in the parameter threshold interval table are combined and matched to identify the coordinated change relationship between the parameters. Combined diagnosis takes into account the correlation and constraint relationship between the parameters, and discovers the abnormal pattern of parameter combination through the association rule mining algorithm. For example, when a parameter exceeds the threshold, check whether the related parameters are also abnormal to determine whether there is a chain reaction between the parameters. Record the abnormal parameter combinations found, including the value range, change trend, degree of mutual influence of each parameter in the combination, and form a combination parameter abnormality set. Then, the rule verification of the combination parameter abnormality set is performed to verify whether the abnormal combination meets the preset rule constraints, ensure the reliability of the diagnosis results, and obtain the parameter combination rule table.

[0070] System-level linkage diagnosis is to further analyze the abnormal characteristics at the whole system level on the basis of parameter combination diagnosis. System-level analysis is performed on the parameter combination rule table to examine the impact of parameter changes on the overall function of the system. Linkage diagnosis focuses on the propagation path and impact range of parameter changes, and identifies key nodes and weak links. By establishing a parameter influence network diagram, the diffusion effect of parameter changes is analyzed to obtain a system abnormal feature set. Correlation analysis is performed on the system abnormal feature set to study the correlation strength and action mechanism between different features and form a system diagnosis feature table. Dynamic threshold analysis is to dynamically monitor the features in the system diagnosis feature table. A time series feature curve is established to analyze the change trend and periodicity of the feature value. The dynamic threshold is adjusted in real time according to the change rate and trend of the feature to adapt to the dynamic changes of the system state. The obtained threshold change curve set reflects the dynamic adjustment process of the threshold during the operation of the system. Parameter fluctuation analysis is performed on the threshold change curve set to study the fluctuation characteristics of the parameter value around the threshold, including fluctuation amplitude, frequency, duration, etc., to form a parameter fluctuation law table.

[0071] The integration of diagnostic results is to conduct a comprehensive analysis of the diagnostic results of the aforementioned levels. The information in the parameter fluctuation law table is classified and sorted, and classified according to the fault type, severity, processing priority and other dimensions to form a structured functional parameter diagnostic report. The diagnostic report is deeply analyzed, the processing strategies and optimization suggestions are extracted, and a parameter tuning strategy library is established. The strategy library contains processing solutions for different abnormal situations, and each solution includes adjustment steps, expected effects, precautions and other contents.

[0072] Take the spindle system of a certain model of CNC milling machine as an example: In the single parameter diagnosis stage, the spindle speed is monitored, the rated speed is set to 6000rpm, and the allowable fluctuation range is ±1%. Through real-time monitoring, it is found that the speed fluctuates more after 4 hours of continuous operation, and the fluctuation range reaches ±2%, which exceeds the threshold range and is recorded in the single parameter abnormality table. Parameter combination diagnosis found that the speed fluctuation abnormality is related to the spindle temperature and bearing vibration value. When the temperature rises to 45℃ (normal range <40℃), the vibration value increases by 30%. System-level linkage diagnosis shows that this abnormality will spread to the processing accuracy, resulting in a 15% increase in the surface roughness of the workpiece. Dynamic threshold analysis shows that as the equipment operation time increases, the allowable range of speed fluctuation needs to be appropriately relaxed and adjusted to ±1.5% after 8 hours. The final diagnosis report points out that the spindle system has an overheating problem, and it is recommended to add the processing strategy of "when the temperature exceeds 43℃, reduce the spindle speed by 20% and continue cooling for 30 minutes" in the strategy library.

[0073] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0074] (1) Perform parameter classification and screening on the abnormal parameters in the functional parameter diagnosis report to obtain an abnormal parameter classification table, and perform parameter correlation analysis on the abnormal parameter classification table to obtain a parameter correlation network diagram;

[0075] (2) Perform fault source location processing on the parameter association network diagram and parameter tuning strategy library to obtain the initial fault node set, and perform propagation risk assessment processing on the initial fault node set to obtain a fault risk level table;

[0076] (3) Perform fault propagation path tracing on the fault risk level table to obtain a fault diffusion sequence, and perform propagation feature extraction on the fault diffusion sequence to obtain a propagation feature vector group;

[0077] (4) Perform fault impact range analysis on the propagation feature vector group to obtain an impact range matrix, and perform critical value judgment on the impact range matrix to obtain a set of key impact parameters;

[0078] (5) Perform correction strategy generation processing on the key influencing parameter set to obtain a parameter correction strategy table, and perform correction effect simulation processing on the parameter correction strategy table to obtain a correction effect evaluation set;

[0079] (6) Performing scheme screening processing on the correction effect evaluation set to obtain a functional parameter correction scheme set, and performing rule extraction processing on the functional parameter correction scheme set to obtain an initial diagnosis rule table;

[0080] (7) Perform rule verification processing on the initial diagnostic rule table to obtain a rule verification result set, and perform rule integration processing on the rule verification result set to obtain a diagnostic rule base.

[0081] Specifically, the abnormal parameters in the functional parameter diagnosis report are classified and screened. The classification and screening adopts the hierarchical clustering method to classify the parameters according to their physical properties, functional characteristics and fault types. For example, temperature parameters, pressure parameters and vibration parameters are classified separately to form an abnormal parameter classification table. Then, the parameters in the abnormal parameter classification table are analyzed for correlation. The correlation coefficient and mutual information value between the parameters are calculated to determine the correlation strength between the parameters. The parameter pairs with correlation strength exceeding the threshold (usually set to 0.7) are connected to construct a parameter association network diagram. In the fault source location stage, the reverse reasoning method is used to locate the fault source in combination with the parameter association network diagram and the existing parameter tuning strategy library. Starting from the abnormal parameters, the most likely fault source node is found by tracing back along the edge of the association network. The fault probability of each node is calculated according to its abnormality degree and impact range, and the node set with the highest fault probability is selected as the initial fault node set. The initial fault node set is evaluated for propagation risk. The evaluation indicators include fault diffusion speed, impact range, and degree of harm. The faults are graded according to the risk level (from level 1 to level 5) to generate a fault risk level table.

[0082] Fault propagation path tracing is to analyze the propagation path of the fault in the fault risk level table. Using the breadth-first search algorithm, starting from the fault source node, it expands layer by layer according to the correlation between parameters, records the propagation path and affected range of the fault, and forms a fault diffusion sequence. Extract features from each propagation path, including propagation speed, attenuation coefficient, branching factor, etc., to form a propagation feature vector group. In the fault impact range analysis stage, cluster analysis is performed on the propagation feature vector group, the impact intensity of different propagation paths is calculated, and the impact range matrix is ​​established. Each element in the matrix represents the degree of influence of the fault on a certain parameter. By setting the critical value (generally taking the 80% quantile of the impact intensity), the parameters with greater impact are screened out to form a set of key impact parameters.

[0083] The correction strategy generation process designs correction schemes based on the key influencing parameter set. Formulate adjustment strategies for each key parameter, including adjustment direction, step size, constraints, etc., and generate a parameter correction strategy table. Perform simulation verification on the correction strategy, simulate the system response after parameter adjustment, evaluate the correction effect, including parameter regression degree, stability, response time and other indicators, and form a correction effect evaluation set. The scheme screening stage conducts a comprehensive evaluation of the schemes in the correction effect evaluation set, considers factors such as correction effect, implementation difficulty, resource consumption, etc., screens out the optimal correction scheme, and forms a functional parameter correction scheme set. Extract correction rules from the correction scheme, record the conditions, methods and effects of parameter adjustment, and generate an initial diagnostic rule table.

[0084] Finally, the initial diagnostic rule table is verified to verify the accuracy and reliability of the rules, check whether there are conflicts between the rules, and obtain the rule verification result set. The rules that pass the verification are integrated and classified according to the fault type and processing strategy to form a complete diagnostic rule library.

[0085] Taking a certain type of industrial chiller as an example: the analysis found that the evaporator outlet temperature increased abnormally (exceeding the set value by 5°C). Through parameter correlation analysis, it was found that it was related to parameters such as refrigerant pressure, compressor power, and condenser temperature, with correlation coefficients of 0.85, 0.78, and 0.72, respectively. The fault source was located as refrigerant leakage, with a risk level of 4. The fault propagation path shows that refrigerant leakage causes the system pressure to drop (a drop of 20% within 15 minutes), causing the compressor power to increase (an increase of 15%), and ultimately leading to a decrease in the evaporator's cooling effect. Impact range analysis shows that the fault affects 8 key parameters of the refrigeration system. Correction strategies include: replenishing refrigerant, adjusting the expansion valve opening, optimizing compressor operating parameters, etc. Simulation verification shows that the system can return to normal operation within 30 minutes after replenishing refrigerant. Finally, the fault correction experience is refined into diagnostic rules and integrated into the diagnostic rule library.

[0086] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0087] (1) Perform parameter recognition accuracy verification on the functional parameter correction solution set to obtain a recognition accuracy table, and perform deviation statistics on the recognition accuracy table to obtain a recognition deviation set;

[0088] (2) Perform parameter correlation analysis on the identification deviation set to obtain a correlation parameter group, and perform accuracy distribution processing on the correlation parameter group to obtain an accuracy distribution graph;

[0089] (3) Performing diagnostic effectiveness evaluation on the diagnostic rule base to obtain a diagnostic evaluation table, and performing rule verification on the diagnostic evaluation table to obtain a rule verification set;

[0090] (4) Calculate and process the evaluation index of the rule verification set to obtain an evaluation index table, and perform diagnostic capability analysis on the evaluation index table to obtain a diagnostic capability matrix;

[0091] (5) Integrate the results of the accuracy distribution graph and the diagnostic capability matrix to obtain a hardware function parameter identification and diagnostic conclusion report.

[0092] Specifically, the parameter recognition accuracy of the functional parameter correction scheme set was verified. The accuracy verification adopted the cross-validation method, and the historical data was divided into a training set and a verification set at a ratio of 8:2. The recognition accuracy was calculated by comparing the difference between the predicted value of the correction scheme and the actual value. During the verification process, multiple evaluation indicators were set, including recognition rate, false alarm rate, missed alarm rate, etc., and the verification results were recorded in the recognition accuracy table. The data in the recognition accuracy table were statistically analyzed for deviations, and the statistical characteristics of each indicator were calculated, including mean deviation, standard deviation, maximum deviation, minimum deviation, etc., to form a complete recognition deviation set. The parameters in the recognition deviation set were analyzed for correlation, and the relationship between different parameter deviations was studied by using the correlation analysis method. By calculating the correlation coefficient between the parameters, highly correlated parameter combinations were identified, and the parameters with a correlation coefficient exceeding 0.8 were combined to form an associated parameter group. The accuracy distribution analysis of the associated parameter group was performed, and the kernel density estimation method was used to draw the accuracy distribution curve of each parameter group, analyze the central tendency and dispersion of the accuracy, and generate an accuracy distribution diagram, which shows the accuracy distribution characteristics of different parameter groups.

[0093] The effectiveness evaluation of the diagnostic rule base is a systematic verification of the established diagnostic rules. The evaluation process uses the rule coverage analysis method to check the coverage of the rule base for various fault conditions, calculate the scope of application and restrictions of the rules, and record the evaluation results in the diagnostic evaluation table. The diagnostic evaluation table is rule-verified to verify the applicability of each rule under different working conditions, check whether there are conflicts or overlaps between the rules, and filter out the effective rule set to form a rule verification set. The evaluation index calculation of the rule verification set is performed, and evaluation indicators such as rule accuracy, recall rate, and F1 score are set. The accuracy reflects the accuracy of the rule judgment, the recall rate indicates the completeness of the rule coverage, and the F1 score comprehensively considers the accuracy and recall rates. The calculation results are recorded in the evaluation index table, and each rule corresponds to a set of evaluation index values. The diagnostic capability analysis of the evaluation index table is performed to construct a diagnostic capability matrix. Each element in the matrix represents the diagnostic capability of a certain rule for a specific type of fault.

[0094] Finally, the accuracy distribution map and the diagnostic capability matrix are integrated to form a complete hardware function parameter identification diagnostic conclusion report. The report content includes the accuracy analysis of parameter identification, the effectiveness evaluation of diagnostic rules, existing problems and improvement suggestions, etc.

[0095] Take the joint servo system of a certain type of industrial robot as an example: in the parameter recognition accuracy verification stage, by analyzing 1000 sets of historical data, the position error recognition accuracy is calculated to be 92%, the speed error recognition accuracy is 88%, and the current error recognition accuracy is 85%. Deviation statistics show that the average deviation of the position error is 0.03mm, and the standard deviation is 0.01mm; the average deviation of the speed error is 0.05rad / s, and the standard deviation is 0.02rad / s; the average deviation of the current error is 0.1A, and the standard deviation is 0.05A. Correlation analysis found that the correlation coefficient between the position error and the speed error is 0.85, forming a highly correlated parameter group. The accuracy distribution diagram shows that the probability of the position error recognition accuracy being in the range of 90%-95% is 75%. Diagnostic rule verification shows that in 100 test cases, the rule base has a coverage rate of 95% for position-related faults, 92% for speed-related faults, and 88% for current-related faults. The evaluation index calculation shows that the overall accuracy of the rule base is 90%, the recall rate is 87%, and the F1 score is 88.5%. The final diagnosis conclusion report points out that the parameter identification and fault diagnosis performance of the servo system meet the requirements of industrial applications, and it is recommended to further optimize the recognition accuracy of current parameters under high-speed conditions.

[0096] The above describes the hardware function parameter identification and diagnosis method based on big data technology in the embodiment of the present application. The following describes the hardware function parameter identification and diagnosis device based on big data technology in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the hardware function parameter identification and diagnosis device based on big data technology includes:

[0097] Rating module 201, used for distributed collection and quality rating of operation data, technical documents and historical records of hardware devices to obtain a standardized function parameter data set and a function parameter classification dictionary;

[0098] A calculation module 202 is used to perform feature decomposition and parameter correlation calculation on the standardized function parameter data set and the function parameter classification dictionary to obtain a function parameter feature matrix and a function parameter mapping rule set;

[0099] Identification module 203, used to perform two-way cross-validation identification and parameter sensitivity quantification processing on the functional parameter feature matrix and the functional parameter mapping rule set to obtain a functional parameter identification model and a functional parameter evaluation system; wherein the two-way cross-validation identification includes forward function mapping and reverse parameter tracing;

[0100] Diagnosis module 204, used for performing multi-level parameter diagnosis and dynamic threshold analysis on the output result of the functional parameter identification model based on the functional parameter evaluation system to obtain a functional parameter diagnosis report and a parameter tuning strategy library; wherein the multi-level parameter diagnosis includes single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis;

[0101] A generation module 205 is used to perform fault propagation path analysis and correction scheme generation processing on the abnormal parameters in the function parameter diagnosis report based on the parameter tuning strategy library to obtain a function parameter correction scheme set and a diagnosis rule library;

[0102] The evaluation module 206 is used to perform parameter identification accuracy verification and diagnosis effectiveness evaluation processing on the functional parameter correction solution set and the diagnosis rule base to obtain a hardware functional parameter identification diagnosis conclusion report.

[0103] Through the cooperation of the above components, distributed collection and quality rating processing, efficient collection and preprocessing of hardware equipment operation data, technical documents and historical records are achieved, ensuring the integrity and reliability of the data source. Through feature decomposition and parameter correlation calculation, the intrinsic relationship between parameters is deeply explored, and the functional parameter feature matrix and mapping rule set are effectively extracted, laying a solid foundation for subsequent analysis. Adopting a two-way cross-validation recognition mechanism, combined with parameter sensitivity quantitative analysis, an accurate functional parameter identification model and a scientific evaluation system are established, in which the combination of forward function mapping and reverse parameter tracing improves the accuracy and reliability of identification. On this basis, a multi-level parameter diagnosis strategy is adopted, including single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis, combined with dynamic threshold analysis technology, to achieve comprehensive diagnosis from local to overall, accurately identify abnormal parameters and fault types. Based on the parameter tuning strategy library, fault propagation path analysis and correction plan generation are carried out, which not only finds the root cause of the fault, but also provides an effective correction plan, greatly improving the efficiency of fault handling. Finally, through parameter identification accuracy verification and diagnostic effectiveness evaluation, a complete diagnostic conclusion report is formed to ensure the reliability and practicality of the diagnostic results. The entire solution forms a closed-loop diagnostic system with each link closely linked and supporting each other, significantly improving the accuracy, real-time and reliability of hardware function parameter identification and fault diagnosis.

[0104] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A hardware function parameter identification and diagnosis method based on big data technology, characterized in that: The hardware function parameter identification and diagnosis method based on big data technology includes: Distributed collection and quality rating of hardware equipment operation data, technical documents and historical records are performed to obtain standardized function parameter data sets and function parameter classification dictionaries; Performing feature decomposition and parameter correlation calculation processing on the standardized function parameter data set and the function parameter classification dictionary to obtain a function parameter feature matrix and a function parameter mapping rule set; The functional parameter feature matrix and the functional parameter mapping rule set are subjected to two-way cross-validation identification and parameter sensitivity quantification processing to obtain a functional parameter identification model and a functional parameter evaluation system; wherein the two-way cross-validation identification includes forward functional mapping and reverse parameter tracing; The output result of the functional parameter identification model is subjected to multi-level parameter diagnosis and dynamic threshold analysis based on the functional parameter evaluation system to obtain a functional parameter diagnosis report and a parameter tuning strategy library; wherein the multi-level parameter diagnosis includes single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis; Performing fault propagation path analysis and correction scheme generation processing on the abnormal parameters in the function parameter diagnosis report based on the parameter tuning strategy library to obtain a function parameter correction scheme set and a diagnosis rule library; The function parameter correction scheme set and the diagnosis rule base are subjected to parameter identification accuracy verification and diagnosis effectiveness evaluation processing to obtain a hardware function parameter identification diagnosis conclusion report.

2. The hardware function parameter identification and diagnosis method based on big data technology according to claim 1 is characterized in that: The distributed collection and quality rating processing of the operation data, technical documents and historical records of the hardware equipment to obtain a standardized function parameter data set and a function parameter classification dictionary include: Performing data slicing collection processing on the operation data of the hardware device to obtain an operation parameter data stream, and performing outlier detection processing on the operation parameter data stream to obtain a screened operation data set; Performing text parsing processing on the technical document of the hardware device to obtain a structured technical parameter table, and performing data format unification processing on the structured technical parameter table to obtain a standard technical parameter set; Performing time series data extraction processing on the historical records of the hardware device to obtain a historical parameter sequence, and performing data completion processing on the historical parameter sequence to obtain a target historical data set; Performing data fusion processing on the screening operation data set, the standard technical parameter set and the target historical data set to obtain a comprehensive parameter data table, and performing quality scoring processing on the comprehensive parameter data table to obtain a parameter quality rating table; Performing parameter classification and annotation processing on the parameter quality rating table to obtain a function parameter classification dictionary, and performing hierarchical relationship construction processing on the function parameter classification dictionary to obtain a parameter type hierarchy table; The parameter type hierarchy table is subjected to distributed verification processing to obtain a verification parameter set, and the verification parameter set is subjected to data standardization processing to obtain a standardized function parameter data set.

3. The hardware function parameter identification and diagnosis method based on big data technology according to claim 1 is characterized in that: The step of performing feature decomposition and parameter correlation calculation on the standardized function parameter data set and the function parameter classification dictionary to obtain a function parameter feature matrix and a function parameter mapping rule set includes: Performing parameter stratification processing on the standardized functional parameter data set to obtain a stratified parameter set, and performing feature decomposition processing on the stratified parameter set to obtain a basic feature set; Cross-matching the basic feature set and the function parameter classification dictionary to obtain a feature type table, and performing data dimension reduction processing on the feature type table to obtain a main feature set; Performing parameter correlation analysis on the main feature set to obtain a parameter correlation matrix, and performing threshold segmentation on the parameter correlation matrix to obtain a core correlation table; Performing feature aggregation processing on the core association table to obtain a function parameter feature matrix, and performing feature verification processing on the function parameter feature matrix to obtain a verification feature set; Performing association rule extraction processing on the verification feature set to obtain an initial rule set, and performing rule screening processing on the initial rule set to obtain a valid rule table; The effective rule table is subjected to rule integration processing to obtain a function parameter mapping rule set.

4. The hardware function parameter identification and diagnosis method based on big data technology according to claim 1 is characterized in that: The function parameter feature matrix and the function parameter mapping rule set are subjected to two-way cross-validation identification and parameter sensitivity quantification processing to obtain a function parameter identification model and a function parameter evaluation system; wherein the two-way cross-validation identification includes forward function mapping and reverse parameter tracing, including: Performing forward function mapping processing on the function parameter feature matrix to obtain a function mapping feature table, and performing parameter tracking processing on the function mapping feature table to obtain a parameter mapping sequence; Performing reverse parameter tracing processing on the function parameter mapping rule set to obtain a parameter tracing relationship table, and performing rule verification processing on the parameter tracing relationship table to obtain a verification rule set; Performing cross-validation recognition processing on the parameter mapping sequence and the verification rule set to obtain a recognition result matrix, and performing validity verification processing on the recognition result matrix to obtain a verification result table; Performing parameter sensitivity analysis on the verification result table to obtain a sensitivity index set, and performing quantitative evaluation on the sensitivity index set to obtain a parameter quantification table; Performing association rule extraction processing on the parameter quantization table to obtain a function parameter evaluation system, and performing evaluation standard construction processing on the function parameter evaluation system to obtain a parameter evaluation index set; The parameter evaluation index set is subjected to parameter identification calculation processing to obtain a functional parameter identification model.

5. The hardware function parameter identification and diagnosis method based on big data technology according to claim 1 is characterized in that: Performing multi-level parameter diagnosis and dynamic threshold analysis on the output results of the functional parameter identification model based on the functional parameter evaluation system to obtain a functional parameter diagnosis report and a parameter tuning strategy library; The multi-level parameter diagnosis includes single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis, including: Performing single parameter diagnosis processing on the output result of the functional parameter identification model to obtain a single parameter abnormality table, and performing parameter threshold analysis processing on the single parameter abnormality table to obtain a parameter threshold interval table; Performing parameter combination diagnosis processing on the parameter threshold interval table to obtain a combination parameter anomaly set, and performing combination rule verification processing on the combination parameter anomaly set to obtain a parameter combination rule table; Performing system-level linkage diagnosis processing on the parameter combination rule table to obtain a system abnormality feature set, and performing correlation analysis processing on the system abnormality feature set to obtain a system diagnosis feature table; Performing dynamic threshold analysis processing on the system diagnostic feature table to obtain a threshold change curve set, and performing parameter fluctuation analysis processing on the threshold change curve set to obtain a parameter fluctuation law table; The parameter fluctuation rule table is processed by diagnostic result integration to obtain a function parameter diagnostic report, and the function parameter diagnostic report is processed by strategy extraction to obtain a parameter tuning strategy library.

6. The hardware function parameter identification and diagnosis method based on big data technology according to claim 1 is characterized in that: The abnormal parameters in the function parameter diagnosis report are analyzed for fault propagation paths and correction schemes are generated based on the parameter tuning strategy library to obtain a function parameter correction scheme set and a diagnosis rule library, including: Performing parameter classification and screening processing on the abnormal parameters in the functional parameter diagnosis report to obtain an abnormal parameter classification table, and performing parameter correlation analysis processing on the abnormal parameter classification table to obtain a parameter correlation network diagram; Performing fault source location processing on the parameter association network diagram and the parameter tuning strategy library to obtain an initial fault node set, and performing propagation risk assessment processing on the initial fault node set to obtain a fault risk level table; Performing fault propagation path tracing processing on the fault risk level table to obtain a fault diffusion sequence, and performing propagation feature extraction processing on the fault diffusion sequence to obtain a propagation feature vector group; Performing fault impact range analysis processing on the propagation feature vector group to obtain an impact range matrix, and performing critical value judgment processing on the impact range matrix to obtain a key impact parameter set; Performing correction strategy generation processing on the key influencing parameter set to obtain a parameter correction strategy table, and performing correction effect simulation processing on the parameter correction strategy table to obtain a correction effect evaluation set; Performing scheme screening processing on the correction effect evaluation set to obtain a function parameter correction scheme set, and performing rule extraction processing on the function parameter correction scheme set to obtain an initial diagnosis rule table; The initial diagnosis rule table is subjected to rule verification processing to obtain a rule verification result set, and the rule verification result set is subjected to rule integration processing to obtain a diagnosis rule base.

7. The hardware function parameter identification and diagnosis method based on big data technology according to claim 1 is characterized in that: The parameter identification accuracy verification and diagnostic effectiveness evaluation processing are performed on the functional parameter correction scheme set and the diagnostic rule base to obtain a hardware functional parameter identification diagnostic conclusion report, including: Performing parameter recognition accuracy verification processing on the functional parameter correction solution set to obtain a recognition accuracy table, and performing deviation statistical processing on the recognition accuracy table to obtain a recognition deviation set; Performing parameter correlation analysis on the identification deviation set to obtain a correlation parameter group, and performing accuracy distribution processing on the correlation parameter group to obtain an accuracy distribution graph; Performing a diagnosis validity evaluation process on the diagnosis rule base to obtain a diagnosis evaluation table, and performing a rule verification process on the diagnosis evaluation table to obtain a rule verification set; Performing evaluation index calculation processing on the rule verification set to obtain an evaluation index table, and performing diagnostic capability analysis processing on the evaluation index table to obtain a diagnostic capability matrix; The accuracy distribution graph and the diagnostic capability matrix are integrated and processed to obtain a hardware function parameter identification and diagnostic conclusion report.

8. A hardware function parameter identification and diagnosis device based on big data technology, used to implement the hardware function parameter identification and diagnosis method based on big data technology as described in any one of claims 1 to 7, characterized in that: The hardware function parameter identification and diagnosis device based on big data technology includes: The rating module is used to perform distributed collection and quality rating processing on the operating data, technical documents and historical records of hardware devices to obtain a standardized function parameter data set and a function parameter classification dictionary; A calculation module, used for performing feature decomposition and parameter correlation calculation processing on the standardized function parameter data set and the function parameter classification dictionary to obtain a function parameter feature matrix and a function parameter mapping rule set; An identification module is used to perform two-way cross-validation identification and parameter sensitivity quantification processing on the functional parameter feature matrix and the functional parameter mapping rule set to obtain a functional parameter identification model and a functional parameter evaluation system; wherein the two-way cross-validation identification includes forward function mapping and reverse parameter tracing; A diagnosis module, which is used to perform multi-level parameter diagnosis and dynamic threshold analysis on the output results of the functional parameter identification model based on the functional parameter evaluation system to obtain a functional parameter diagnosis report and a parameter tuning strategy library; wherein the multi-level parameter diagnosis includes single parameter diagnosis, parameter combination diagnosis and system-level linkage diagnosis; A generation module, used for performing fault propagation path analysis and correction scheme generation processing on the abnormal parameters in the function parameter diagnosis report based on the parameter tuning strategy library, and obtaining a function parameter correction scheme set and a diagnosis rule library; The evaluation module is used to perform parameter identification accuracy verification and diagnostic effectiveness evaluation processing on the functional parameter correction solution set and the diagnostic rule base to obtain a hardware functional parameter identification diagnosis conclusion report.

Images