A method and system for analyzing production data anomalies

Abstract

This invention provides a method and system for anomaly analysis of production data, relating to the field of data processing technology. The method includes: acquiring historical cable crimping process records and dividing them into a core sample dataset and a boundary sample dataset; determining a crimping reference state based on the core sample dataset; extracting state deviation features from the boundary sample dataset to generate multiple state offset sequences for the boundary sample dataset; constructing deviation propagation datasets corresponding to multiple crimping stages and training multiple deviation propagation analysis models; performing deviation propagation correction on the boundary sample dataset to generate a propagation correction dataset; constructing a stage deviation evolution map based on the crimping reference state and the propagation correction dataset; and performing anomaly analysis on the collected real-time cable crimping data using the stage deviation evolution map to generate production data anomaly analysis results. This invention improves the stability and intelligence level of crimping process quality monitoring.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for analyzing production data anomalies. Background Technology

[0002] Cable crimping is a key process in the manufacturing of power equipment, widely used in the internal connections of high and low voltage switchgear, prefabricated substations, and other equipment. The quality of the crimping directly affects the safety and stability of the electrical connection. Some technologies analyze key parameters in process record data item by item based on static threshold judgment and standard template matching. Essentially, these methods statically slice the crimping process, dividing continuously changing data into discrete points or segments for judgment.

[0003] Cable crimping is actually a process where pressure is gradually increased and the contact state continuously evolves, resulting in highly process-oriented parameter changes. Deviations at a single point or in a localized period may not fully reflect the evolution of the entire crimping process, thus missing potential anomalies. In this continuous process, factors such as equipment response and material deformation may cause deviations in the early stages to be partially offset or compensated by subsequent processes, making the overall curve appear normal when in fact there is a systematic deviation. Such intra-process compensation phenomena are difficult to identify using traditional point-by-point judgment methods. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a production data anomaly analysis method and system. By identifying the actual offset characteristics during the continuous processing of cable crimping, it enables early identification of abnormal trends and reveals potential risks from the overall evolution trend.

[0005] The first aspect of this invention provides a method for analyzing production data anomalies, comprising:

[0006] Obtain historical cable crimping process records for the target power equipment, including historical crimping process records and historical crimping quality test data, and divide the historical cable crimping process record data into core sample datasets and boundary sample datasets;

[0007] Based on the core sample dataset, the crimping reference state is determined. Based on the crimping reference state, the state deviation feature of the boundary sample dataset is extracted, and the state offset sequence corresponding to each group of boundary sample data in the boundary sample dataset at multiple crimping stages is generated.

[0008] Construct deviation transmission datasets corresponding to multiple pressing stages, train multiple deviation transmission analysis models based on the deviation transmission datasets, and use the trained deviation transmission analysis models to perform deviation transmission correction on the boundary sample datasets to generate a transmission correction dataset.

[0009] Based on the crimping reference state and conduction correction dataset, a stage deviation evolution map is constructed. Anomaly analysis is performed on the collected real-time cable crimping data using the stage deviation evolution map to generate production data anomaly analysis results.

[0010] Preferably, the boundary sample dataset is subjected to state deviation feature extraction based on the pressing reference state, including:

[0011] The crimping reference state includes the reference crimping state sequence corresponding to multiple crimping stages in the cable crimping process of the target power equipment. This includes constructing the core crimping state sequence corresponding to each core sample data in the core sample dataset for multiple crimping stages, and fusing multiple core crimping state sequences of the crimping stages to generate the reference crimping state sequence for the crimping stage.

[0012] Construct boundary pressing state sequences for each group of boundary sample data in the boundary sample dataset at multiple pressing stages. Calculate the deviation of the boundary pressing state sequences based on the reference pressing state sequences to generate multiple state offset sequences for the boundary sample data.

[0013] Preferably, multiple deviation transmission datasets corresponding to different pressing stages are constructed, and multiple deviation transmission analysis models are trained based on these datasets, including:

[0014] Based on the state offset sequence, the state deviation values ​​of each group of boundary sample data at multiple stages are calculated. The deviation transmission path of each pressing stage except the first pressing stage is determined. Based on the state deviation values ​​of multiple stages, multiple sets of transmission sample data of each deviation transmission path are extracted to generate multiple deviation transmission datasets.

[0015] Based on the deviation transmission path, a deviation transmission analysis model is constructed for each deviation transmission dataset. The deviation transmission analysis model is a linear regression model. The deviation transmission analysis model is trained using the deviation transmission dataset to obtain the deviation transmission analysis models corresponding to the remaining pressing stages, except for the first pressing stage.

[0016] Preferably, a stage deviation evolution map is constructed based on the press-fit reference state and the conduction correction dataset, including:

[0017] Multiple boundary sample data are corrected based on multiple deviation transmission analysis models to generate a transmission correction dataset including multiple sets of transmission correction sample data. The correction pressing state sequence of each set of transmission correction sample data is determined, and the stage direction characteristics of each correction pressing state sequence in each pressing stage are determined based on the pressing reference state.

[0018] The stage deviation range of each pressing stage is determined based on the directional characteristics of multiple stages and the corrected pressing state sequence. Multiple state deviation intervals of the pressing stage are determined based on the stage deviation range. These multiple state deviation intervals are used as state deviation nodes of each pressing stage in the stage deviation evolution diagram.

[0019] Multiple stage deviation evolution paths are extracted from the transmission correction dataset. Based on these multiple stage deviation evolution paths, multiple deviation evolution edges for state deviation nodes are determined, and the evolution weight of each deviation evolution edge is calculated to construct a stage deviation evolution graph for multiple state deviation nodes.

[0020] Preferably, anomaly analysis is performed on the collected real-time cable crimping data using a stage deviation evolution map to generate production data anomaly analysis results, including:

[0021] Multiple real-time deviation evolution paths are extracted from the real-time cable crimping data. The global deviation evolution weight of each real-time deviation evolution path is calculated based on the stage deviation evolution map. Multiple real-time deviation evolution paths are marked as anomalies based on a preset deviation evolution threshold. The proportion parameter of the paths marked as abnormal evolution paths in the real-time cable crimping data is determined, and the production data anomaly analysis results of the real-time cable crimping data are generated.

[0022] Preferably, multiple sets of boundary sample data are corrected based on multiple deviation transmission analysis models to generate a transmission correction dataset including multiple sets of transmission correction sample data, including:

[0023] The multiple stage state deviation values ​​of each set of boundary sample data are input into the corresponding deviation transmission analysis model to generate the corrected state deviation value corresponding to the stage state deviation value. This yields the transmission corrected sample data corresponding to each set of boundary sample data, generating a transmission corrected dataset that includes multiple sets of transmission corrected sample data.

[0024] A second aspect of the present invention provides a production data anomaly analysis system for implementing the above-described production data anomaly analysis method, comprising:

[0025] The crimping data acquisition module is used to acquire historical cable crimping process record data about the target power equipment in production, including historical crimping process record data and historical crimping quality test data, and divides the historical cable crimping process record data into core sample datasets and boundary sample datasets;

[0026] The state offset analysis module is used to determine the crimping reference state based on the core sample dataset, extract the state deviation features of the boundary sample dataset based on the crimping reference state, and generate the state offset sequence corresponding to each group of boundary sample data in the boundary sample dataset at multiple crimping stages.

[0027] The deviation transmission analysis module is used to construct deviation transmission datasets corresponding to multiple pressing stages. Based on the deviation transmission datasets, multiple deviation transmission analysis models are trained. The trained deviation transmission analysis models are used to correct the deviation transmission of the boundary sample datasets to generate a transmission correction dataset.

[0028] The production data anomaly analysis module is used to construct a stage deviation evolution map based on the crimping reference state and conduction correction dataset. The stage deviation evolution map is used to perform anomaly analysis on the collected real-time cable crimping data and generate production data anomaly analysis results.

[0029] The present invention has the following beneficial effects:

[0030] This invention addresses the temporal continuity and multi-stage evolution characteristics of cable crimping processes in power equipment manufacturing. It proposes an anomaly detection mechanism that integrates deviation propagation modeling and deviation evolution spectrum analysis. By constructing a propagation model of inter-stage errors, it eliminates the interference of error compensation effects between different crimping stages on anomaly identification, extracts more representative true deviation features, and further combines historical representative sample data to construct a state offset evolution spectrum. This achieves accurate mapping of multi-stage deviation paths and anomaly weight analysis, effectively overcoming traditional anomaly judgment methods based on fixed thresholds or single-stage indicators. It supports trend identification and accurate early warning of potential anomalies in real-time crimping data, improving the stability and intelligence level of crimping process quality monitoring. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating a production data anomaly analysis method according to one embodiment of the present invention.

[0032] Figure 2 This is a schematic diagram of the structure of a production data anomaly analysis system provided in one embodiment of the present invention. Detailed Implementation

[0033] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention.

[0034] One embodiment of the present invention provides a method for analyzing production data anomalies; please refer to [link to relevant documentation]. Figure 1 The method includes the following steps:

[0035] Step S1: Obtain historical cable crimping process record data for the target power equipment, including historical crimping process record data and historical crimping quality test data, and divide the historical cable crimping process record data into core sample dataset and boundary sample dataset.

[0036] Specifically, the historical cable crimping process record data includes historical crimping process data, involving time-series data of various process parameters collected online by sensor devices during the cable crimping process. This includes real-time changes in applied pressure, real-time displacement data of the mold or crimping process, and instantaneous movement speed data of the crimping joint. For some products, data such as mold temperature changes in the crimping equipment can also be collected during the crimping process to avoid interference caused by thermal expansion and contraction. The historical crimping quality test data includes data obtained through automated or manual detection and recording after crimping, reflecting the performance quality of the crimped product. This includes data such as pull-out force characterizing the connection strength between the terminal and the cable, contact resistance characterizing conductivity, and crimping height characterizing whether the shaping meets standards. The collected data is standardized and aligned, for example, through normalization, before being used for subsequent modeling and analysis.

[0037] For the pre-processed historical cable crimping process records, the data is divided according to crimping quality to construct representative samples containing excellent crimping effects and small deviations. Multiple representative core sample data sets can be determined from a large dataset using technical personnel or pre-set quality thresholds to obtain a core sample dataset for constructing the reference state. The remaining data is marked as boundary sample data to obtain a boundary sample dataset, which contains normal samples with some deviation but still within an acceptable range, used to capture edge changes and deviation propagation patterns. In short, the collected historical cable crimping process records can specifically be data corresponding to products that meet quality requirements throughout the production process. Dividing the data based on crimping quality yields samples closer to the process boundary region, which helps model the dynamic evolution process of crimping states gradually deviating from target requirements.

[0038] Step S2: Determine the crimping reference state based on the core sample dataset, extract the state deviation features from the boundary sample dataset based on the crimping reference state, and generate the state offset sequence corresponding to each group of boundary sample data in the boundary sample dataset at multiple crimping stages.

[0039] Specifically, the crimping reference state is constructed by extracting the average state of multiple crimping stages from multiple representative samples in the core sample dataset as a central benchmark. This includes the reference crimping state sequence corresponding to multiple crimping stages in the cable crimping process of the target power equipment.

[0040] The crimping stage can be divided into multiple typical stages based on process characteristics, such as initial contact, plastic compression, and crimping shaping, or it can be divided into multiple stages based on crimping duration. Those skilled in the art can make reasonable divisions based on actual production processes. After determining the multiple crimping stages of the cable crimping process for the target power equipment, the core crimping state sequence corresponding to each set of core sample data in different crimping stages is extracted. First, the core crimping state sequence corresponding to each set of core sample data in the core sample dataset for each of the multiple crimping stages is constructed. This is represented as a time series vector group composed of the values ​​of multiple sampling points in different parameter dimensions within the crimping stage. Then, the multiple core crimping state sequences of each crimping stage are fused, for example, by calculating the mean, to generate a reference crimping state sequence corresponding to each crimping stage, thereby constructing a crimping reference state.

[0041] Furthermore, using multiple reference pressing state sequences as benchmarks, a comparative analysis of the pressing process is conducted on the boundary sample dataset. By comparing the differences between the process states and reference states of each group of boundary sample data, the state deviations at multiple pressing stages are extracted. For example, boundary pressing state sequences corresponding to multiple pressing stages of the boundary sample data are constructed, and deviation calculations are performed on the boundary pressing state sequences based on the reference pressing state sequences, such as obtaining the difference values, to form state offset sequences corresponding to multiple pressing stages of each group of boundary sample data, reflecting the degree and direction of its offset relative to the reference state at each pressing stage.

[0042] Step S3: Construct deviation transmission datasets corresponding to multiple pressing stages, train multiple deviation transmission analysis models based on the deviation transmission datasets, and use the trained deviation transmission analysis models to perform deviation transmission correction on the boundary sample datasets to generate a transmission correction dataset.

[0043] It is worth noting that in typical continuous manufacturing processes like cable crimping, the crimping process is usually completed by equipment under closed-loop control. Therefore, deviations in different parameters, such as displacement and pressure, may influence each other between different crimping stages. For example, if a deviation occurs at a certain stage, subsequent stages may automatically or passively compensate for the deviation due to factors such as the equipment's feedback control system (e.g., the crimping displacement reaches the set value but the pressure is too low), or the natural tendency of material deformation, resulting in an overall error that is ultimately insignificant. This relationship of deviation transmission and compensation is often overlooked in anomaly detection methods based on single-point threshold judgment, failing to identify states where anomalies have not yet manifested but have already begun to evolve. This invention improves the depth of analysis of potential anomalies by modeling the deviation transmission paths between stages.

[0044] Specifically, using the state offset sequences of each stage in the boundary sample dataset as input, a deviation propagation dataset between stages is constructed. A deviation propagation analysis model is trained for each stage to express the influence of the deviation of the previous stage on the deviation of that stage. Based on the trained model, the state offset sequences of the boundary sample dataset are corrected to remove some compensable deviations that may be caused by the previous stage, and the true residual offset of each stage is retained to generate a propagation correction dataset, which is used to characterize the actual independent deviation state of each sample after the mutual influence is eliminated.

[0045] As an optional implementation process, a deviation transmission dataset corresponding to multiple pressing stages is constructed, and multiple deviation transmission analysis models are trained based on the deviation transmission dataset, including:

[0046] Based on the state offset sequence, the state deviation values ​​of each group of boundary sample data at multiple stages are calculated. The deviation transmission path of each pressing stage except the first pressing stage is determined. Based on the state deviation values ​​of multiple stages, multiple sets of transmission sample data of each deviation transmission path are extracted to generate multiple deviation transmission datasets.

[0047] In this embodiment, for each state offset sequence, which involves the deviation values ​​of multiple parameter dimensions at different sampling points, in order to simplify the feature dimensions, the mean value of each parameter of the state offset sequence, such as pressure and displacement, is calculated to obtain the dimensionality-reduced state offset sequence. The modulus of the dimensionality-reduced sequence is calculated as the overall deviation intensity. While preserving the cross-stage transmission relationship, the actual deviation of the sample from the reference state in each pressing stage is quantified. Then, based on the concept of stage deviation propagation path, the modeling structure of deviation propagation is determined. Since the first pressing stage is not affected by the previous stage, it usually starts from the second pressing stage. Each pressing stage is affected by multiple upstream pressing stages. In this way, the deviation propagation path of the pressing stage is determined. For example, the fifth pressing stage will be affected by the deviation of the previous four pressing stages. After determining multiple deviation propagation paths, according to the pressing stages involved in the deviation propagation path, multiple stage state deviation values ​​belonging to the same set of boundary sample data corresponding to these stages are extracted to form one set of propagation sample data. Each set of propagation sample data includes the stage state deviation values ​​of multiple upstream pressing stages as input features and the stage state deviation value of the current pressing stage as the output target. In this way, the deviation propagation dataset corresponding to each pressing stage is constructed.

[0048] Based on the deviation propagation path, a deviation propagation analysis model is constructed for each deviation propagation dataset. The deviation propagation analysis model is trained using the deviation propagation dataset to obtain the deviation propagation analysis models corresponding to the remaining pressing stages, except for the first pressing stage.

[0049] In this embodiment, the deviation transmission analysis model is preferably a linear regression model. By minimizing the sum of squared residuals, the model parameters are trained based on the deviation transmission dataset, enabling the deviation transmission analysis model to reflect the real cross-stage deviation compensation, amplification, or transmission trends in historical samples. Ultimately, deviation transmission analysis models corresponding to all pressing stages except the first pressing stage can be obtained. These models can be used to correct the state deviation sequence of boundary sample data, eliminate the masking effect caused by deviation transmission and other factors between stages, and thus obtain a more essential and independent true deviation structure.

[0050] Step S4: Construct a stage deviation evolution map based on the crimping reference state and conduction correction dataset. Perform anomaly analysis on the collected real-time cable crimping data through the stage deviation evolution map to generate production data anomaly analysis results.

[0051] Specifically, for each sample group in the conduction correction dataset, a symbolic representation of the deviation value is constructed for each stage, and a stage deviation evolution map is generated to show the deviation evolution relationship across multiple crimping stages. In practical applications, the collected cable crimping data is calculated and corrected for deviation in the same way, mapped onto the map structure, and its path is observed to see if it falls into a high-risk area, deviates from the high-frequency main path, or exhibits rare jumps. This generates anomaly analysis results for production data, enabling early warning of potential abnormal trends and quality assurance.

[0052] As an optional implementation process, a stage deviation evolution map is constructed based on the press-fit reference state and the conduction correction dataset, including:

[0053] Multiple boundary sample data are corrected based on multiple deviation transmission analysis models to generate a transmission correction dataset including multiple sets of transmission correction sample data. The correction pressing state sequence of each set of transmission correction sample data is determined, and the stage direction characteristics of each correction pressing state sequence in each pressing stage are determined based on the pressing reference state.

[0054] The process of correcting multiple sets of boundary sample data involves inputting the multiple stage state deviation values ​​of each set of boundary sample data into the deviation transmission analysis model of the corresponding pressing stage, thereby obtaining a more fundamental corrected state deviation value for each set of boundary sample data at each pressing stage. For example, for the stage state deviation value of the third pressing stage, according to the corresponding deviation transmission analysis model, the stage state deviation values ​​of the first and second pressing stages are input into the trained linear regression model to generate the corrected state deviation value for the third pressing stage. In this way, multiple sets of boundary sample data are corrected to obtain multiple sets of transmission corrected sample data. Each set of transmission corrected sample data includes a corrected pressing state sequence based on the corrected state deviation values ​​of multiple pressing stages.

[0055] Considering that the corrected state deviation value is a non-negative scalar value, and if all deviations were non-negative, this would result in the fusion deviation feature only having "deviation magnitude" but no "deviation tendency." Therefore, a sign correction mechanism based on the center reference direction offset is introduced. Specifically, the dot product between the state offset sequence of the boundary sample data and the corresponding reference pressing state sequence of the pressing stage is calculated, and the sign information of the dot product, i.e., positive or negative state, is extracted as the stage direction feature of the pressing stage, used to describe whether the deviation represented by the corrected state deviation value is towards or away from the reference direction.

[0056] The stage deviation range of each pressing stage is determined based on the directional characteristics of multiple stages and the corrected pressing state sequence. Multiple state deviation intervals of the pressing stage are determined based on the stage deviation range, and these multiple state deviation intervals are used as state deviation nodes of each pressing stage in the stage deviation evolution diagram.

[0057] Specifically, based on the stage direction characteristics, the corrected state deviation value in each corrected pressing state sequence is assigned a direction sign, i.e., a positive or negative sign is added. For each pressing stage, multiple corrected state deviation values ​​containing positive and negative signs are summarized to obtain the stage deviation range of the pressing stage. The stage deviation range of each pressing stage can be divided by a fixed step size or a fixed number of steps, for example, into multiple state deviation intervals with equal step sizes, and the state deviation intervals are used as state deviation nodes in the stage deviation evolution map.

[0058] Multiple stage deviation evolution paths are extracted from the transmission correction dataset. Based on these multiple stage deviation evolution paths, multiple deviation evolution edges for state deviation nodes are determined, and the evolution weight of each deviation evolution edge is calculated to construct a stage deviation evolution graph for multiple state deviation nodes.

[0059] Specifically, for the stage deviation evolution path, the stage direction features of the boundary sample data can be fused into the corrected pressing state sequence. The corrected pressing state sequence after fusion of direction features is then determined, and the corresponding state deviation interval for each pressing stage is identified, which is the state deviation node in the stage deviation evolution map. This method allows for the determination of the stage deviation evolution path for different boundary sample data. Then, based on multiple stage deviation evolution paths, the path transition frequency of all samples can be statistically analyzed, i.e., the evolution weight between any two adjacent state node pairs in any two consecutive pressing stages can be calculated. Finally, based on all state deviation nodes and the deviation evolution edges between them, a complete stage deviation evolution map is constructed, serving as the joint evolution structure of multi-stage state deviations in the sample population.

[0060] As an optional implementation process, anomaly analysis is performed on the collected real-time cable crimping data using a stage deviation evolution graph to generate production data anomaly analysis results, including:

[0061] Multiple real-time deviation evolution paths are extracted from the real-time cable crimping data. The global deviation evolution weight of each real-time deviation evolution path is calculated based on the stage deviation evolution map. Anomalies are marked on multiple real-time deviation evolution paths based on a preset deviation evolution threshold.

[0062] Specifically, similar to extracting the stage deviation evolution path of multiple sets of boundary sample data in the boundary sample dataset, the real-time deviation evolution path of the real-time cable crimping data corresponding to multiple sets of real-time sample data is constructed. Each real-time deviation evolution path is mapped to the stage deviation evolution map, and multiple evolution weights involved in the real-time deviation evolution path are determined. The mean of multiple evolution weights is calculated as the global deviation evolution weight of the real-time deviation evolution path, so as to comprehensively reflect the rarity or deviation of the path as a whole in the historical evolution map.

[0063] Then, based on a preset deviation evolution threshold, each real-time deviation evolution path is marked as abnormal. For example, if the global deviation evolution weight of a path is lower than the preset deviation evolution threshold, it is determined to be an abnormal evolution path. The preset deviation evolution threshold can be reasonably set based on data from the actual production process. For example, if 100 consecutive products meet the standard, several consecutive products can be selected as representatives. The global deviation evolution weight of each product is calculated, and the average of these global deviation evolution weights is used as the preset deviation evolution threshold. That is, if no abnormalities occur for a long period during continuous production, it indicates that the overall production status is relatively stable and can be considered representative of normal production. Those skilled in the art can reasonably set the preset deviation evolution threshold according to actual needs.

[0064] Furthermore, the system identifies anomalous evolution paths marked in real-time cable crimping data, calculates the proportion of these marked anomalous paths, and generates a path proportion parameter to reflect the overall volatility and anomalous trend of the current crimping process. Finally, based on the anomalous analysis of individual paths and the overall path anomalous proportion parameter, the system outputs the production data anomalous analysis results for the current real-time cable crimping data. In other words, if numerous anomalous evolutionary processes occur in consecutive products, even if there are no obvious overall anomalies, there may be potential anomalous factors causing the product to gradually deviate from normal production conditions. Such potential risks may require early warning mechanisms. Ultimately, this method provides real-time monitoring and auxiliary decision-making support for the quality control of the crimping process.

[0065] Based on the same concept as the above-described method for analyzing production data anomalies, one embodiment of the present invention also provides a system for analyzing production data anomalies. Please refer to [link to relevant documentation]. Figure 2 The system includes:

[0066] The crimping data acquisition module is used to acquire historical cable crimping process record data about the target power equipment in production, including historical crimping process record data and historical crimping quality test data, and divides the historical cable crimping process record data into core sample datasets and boundary sample datasets;

[0067] The state offset analysis module is used to determine the crimping reference state based on the core sample dataset, extract the state deviation features of the boundary sample dataset based on the crimping reference state, and generate the state offset sequence corresponding to each group of boundary sample data in the boundary sample dataset at multiple crimping stages.

[0068] The deviation transmission analysis module is used to construct deviation transmission datasets corresponding to multiple pressing stages. Based on the deviation transmission datasets, multiple deviation transmission analysis models are trained. The trained deviation transmission analysis models are used to correct the deviation transmission of the boundary sample datasets to generate a transmission correction dataset.

[0069] The production data anomaly analysis module is used to construct a stage deviation evolution map based on the crimping reference state and conduction correction dataset. The stage deviation evolution map is used to perform anomaly analysis on the collected real-time cable crimping data and generate production data anomaly analysis results.

[0070] The above are merely specific embodiments of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A method for analyzing anomalies in production data, characterized in that, include: Acquire historical cable crimping process records for the target power equipment, including historical crimping process records and historical crimping quality test data. Divide the historical cable crimping process records into core sample datasets and boundary sample datasets. This includes determining multiple sets of core sample data from the historical cable crimping process records based on pre-set quality thresholds and constructing a core sample dataset. Mark the remaining data in the historical cable crimping process records as boundary sample data to obtain a boundary sample dataset. The crimping reference state is determined based on the core sample dataset. The crimping reference state includes the reference crimping state sequence corresponding to multiple crimping stages in the cable crimping process of the target power equipment. This includes constructing the core crimping state sequence corresponding to each group of core sample data in the core sample dataset for multiple crimping stages, fusing multiple core crimping state sequences of crimping stages to generate the reference crimping state sequence for the crimping stage, and extracting state deviation features from the boundary sample dataset based on the crimping reference state to generate the state offset sequence corresponding to each group of boundary sample data in the boundary sample dataset for multiple crimping stages. Construct deviation transmission datasets corresponding to multiple pressing stages, including calculating multiple stage state deviation values ​​for each set of boundary sample data based on the state offset sequence, determining the deviation transmission path for each pressing stage except the first pressing stage, extracting multiple sets of transmission sample data for each deviation transmission path based on the multiple stage state deviation values, generating multiple deviation transmission datasets, training multiple deviation transmission analysis models based on the deviation transmission datasets, and using the trained deviation transmission analysis models to perform deviation transmission correction on the boundary sample datasets to generate a transmission correction dataset; Based on the crimping reference state and conduction correction dataset, a stage deviation evolution map is constructed. Anomaly analysis is performed on the collected real-time cable crimping data using the stage deviation evolution map to generate production data anomaly analysis results.

2. The method for analyzing production data anomalies according to claim 1, characterized in that, Based on the crimping reference state, state deviation features are extracted from the boundary sample dataset, including: Construct boundary pressing state sequences for each group of boundary sample data in the boundary sample dataset at multiple pressing stages. Calculate the deviation of the boundary pressing state sequences based on the reference pressing state sequences to generate multiple state offset sequences for the boundary sample data.

3. The method for analyzing production data anomalies according to claim 2, characterized in that, Several deviation transmission analysis models were trained based on the deviation transmission dataset, including: Based on the deviation transmission path, a deviation transmission analysis model is constructed for each deviation transmission dataset. The deviation transmission analysis model is a linear regression model. The deviation transmission analysis model is trained using the deviation transmission dataset to obtain the deviation transmission analysis models corresponding to the remaining pressing stages, except for the first pressing stage.

4. The production data anomaly analysis method according to claim 3, characterized in that, A stage deviation evolution map was constructed based on the press-fit reference state and conduction correction dataset, including: Multiple boundary sample data are corrected based on multiple deviation transmission analysis models to generate a transmission correction dataset including multiple sets of transmission correction sample data. The correction pressing state sequence of each set of transmission correction sample data is determined, and the stage direction characteristics of each correction pressing state sequence in each pressing stage are determined based on the pressing reference state. The stage deviation range of each pressing stage is determined based on the directional characteristics of multiple stages and the corrected pressing state sequence. Multiple state deviation intervals of the pressing stage are determined based on the stage deviation range, and the multiple state deviation intervals are used as the state deviation nodes of each pressing stage in the stage deviation evolution diagram. Multiple stage deviation evolution paths are extracted from the transmission correction dataset. Based on these multiple stage deviation evolution paths, multiple deviation evolution edges for state deviation nodes are determined, and the evolution weight of each deviation evolution edge is calculated to construct a stage deviation evolution graph for multiple state deviation nodes.

5. The production data anomaly analysis method according to claim 4, characterized in that, Anomaly analysis was performed on the collected real-time cable crimping data using a stage deviation evolution graph, generating production data anomaly analysis results, including: Multiple real-time deviation evolution paths are extracted from the real-time cable crimping data. The global deviation evolution weight of each real-time deviation evolution path is calculated based on the stage deviation evolution map. Multiple real-time deviation evolution paths are marked as anomalies based on a preset deviation evolution threshold. The proportion parameter of the paths marked as abnormal evolution paths in the real-time cable crimping data is determined, and the production data anomaly analysis results of the real-time cable crimping data are generated.

6. The production data anomaly analysis method according to claim 4, characterized in that, Based on multiple deviation propagation analysis models, multiple sets of boundary sample data are corrected to generate a propagation correction dataset that includes multiple sets of propagation correction sample data, including: Multiple stage state deviation values ​​of each set of boundary sample data are input into the corresponding deviation transmission analysis model to generate the corrected state deviation value corresponding to the stage state deviation value. This yields the transmission corrected sample data corresponding to each set of boundary sample data, generating a transmission corrected dataset that includes multiple sets of transmission corrected sample data.

7. A production data anomaly analysis system, characterized in that, The system is used to implement the production data anomaly analysis method according to any one of claims 1-6, comprising: The crimping data acquisition module is used to acquire historical cable crimping process record data for the target power equipment, including historical crimping process record data and historical crimping quality test data. The historical cable crimping process record data is divided into core sample datasets and boundary sample datasets. This includes determining multiple sets of core sample data from the historical cable crimping process record data according to a pre-set quality threshold and constructing a core sample dataset. The remaining data in the historical cable crimping process record data is marked as boundary sample data to obtain a boundary sample dataset. The state offset analysis module is used to determine the crimping reference state based on the core sample dataset. The crimping reference state includes the reference crimping state sequence corresponding to multiple crimping stages in the cable crimping process of the target power equipment. This includes constructing the core crimping state sequence corresponding to each group of core sample data in the core sample dataset for multiple crimping stages, fusing multiple core crimping state sequences of crimping stages to generate the reference crimping state sequence for the crimping stage, and extracting state deviation features from the boundary sample dataset based on the crimping reference state to generate the state offset sequence corresponding to each group of boundary sample data in the boundary sample dataset for multiple crimping stages. The deviation transmission analysis module is used to construct deviation transmission datasets corresponding to multiple pressing stages. This includes calculating multiple stage state deviation values ​​for each set of boundary sample data based on the state offset sequence, determining the deviation transmission path for each pressing stage except the first pressing stage, extracting multiple sets of transmission sample data for each deviation transmission path based on the multiple stage state deviation values, generating multiple deviation transmission datasets, training multiple deviation transmission analysis models based on the deviation transmission datasets, and using the trained deviation transmission analysis models to perform deviation transmission correction on the boundary sample datasets to generate a transmission correction dataset. The production data anomaly analysis module is used to construct a stage deviation evolution map based on the crimping reference state and conduction correction dataset. The stage deviation evolution map is used to perform anomaly analysis on the collected real-time cable crimping data and generate production data anomaly analysis results.

Images