CN117315408A - Fault criterion model construction method and system based on AI target detection model - Google Patents
Fault criterion model construction method and system based on AI target detection model Download PDFInfo
- Publication number
- CN117315408A CN117315408A CN202311193948.9A CN202311193948A CN117315408A CN 117315408 A CN117315408 A CN 117315408A CN 202311193948 A CN202311193948 A CN 202311193948A CN 117315408 A CN117315408 A CN 117315408A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- fault
- functional
- target detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 117
- 238000010276 construction Methods 0.000 title claims abstract description 28
- 238000000034 method Methods 0.000 claims abstract description 47
- 238000004140 cleaning Methods 0.000 claims abstract description 42
- 238000012545 processing Methods 0.000 claims abstract description 41
- 238000002474 experimental method Methods 0.000 claims abstract description 23
- XOFYZVNMUHMLCC-ZPOLXVRWSA-N prednisone Chemical compound O=C1C=C[C@]2(C)[C@H]3C(=O)C[C@](C)([C@@](CC4)(O)C(=O)CO)[C@@H]4[C@@H]3CCC2=C1 XOFYZVNMUHMLCC-ZPOLXVRWSA-N 0.000 claims description 28
- 230000002159 abnormal effect Effects 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 11
- 238000012417 linear regression Methods 0.000 claims description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 10
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000012216 screening Methods 0.000 abstract description 2
- 238000004458 analytical method Methods 0.000 description 10
- 230000006399 behavior Effects 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 230000006872 improvement Effects 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 238000012806 monitoring device Methods 0.000 description 3
- 239000012141 concentrate Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/776—Validation; Performance evaluation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
- G06F18/15—Statistical pre-processing, e.g. techniques for normalisation or restoring missing data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Probability & Statistics with Applications (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention relates to the technical field of fault criterion model construction of AI (advanced technology attachment) target detection models, and discloses a fault criterion model construction method and system based on an AI target detection model, wherein the method comprises the following steps: by acquiring the historical operation data of the AI target detection model, including the historical fault data, basic information is provided for the establishment of the fault criterion model. And secondly, data cleaning is carried out on the historical fault data of the AI target detection model, and the data fault data are integrated into a data fault set, so that the quality and usability of the data are ensured. And then, extracting operation parameters from the data fault set, further screening out key functional parameters, and then, establishing a functional model and carrying out a fitting experiment to improve the performance of the model. In this process, the functional data set is subjected to data processing to determine optimal functional data, thereby improving accuracy of the model. And finally, determining a fault criterion model of the AI target detection model through the functional model, and providing support for the reliability of the establishment of the fault criterion model.
Description
Technical Field
The invention relates to the technical field of fault criterion model construction of an AI target detection model, in particular to a fault criterion model construction method and system based on an AI target detection model.
Background
With the rapid development of the field of computer vision, especially the wide application of deep learning technology, the AI target detection model has made a great breakthrough in image and video processing. These models are capable of detecting and locating various target objects in a highly accurate manner, both in road identification of an autonomous car and in quality inspection of products in intelligent manufacturing.
Currently, the modern industry and technology fields are becoming more widespread and sophisticated for the acquisition and recording of data. A large amount of operating parameters and performance data are collected in real time and used to monitor the status of the device or AI system. However, these data inevitably include errors, disturbances, or erroneous judgment, which results in inaccurate judgment of the state of the monitoring device or the AI system.
Therefore, it is urgently needed to invent a fault criterion model technology for constructing the state of the monitoring device or the AI system, so as to solve the problem of inaccurate state judgment of the monitoring device or the AI system caused by errors, interference or erroneous judgment in the judgment of data.
Disclosure of Invention
The purpose of the invention is that: the fault criterion model construction method and system based on the AI target detection model are provided, and are used for solving the problem of inaccurate state judgment of monitoring equipment or an AI system.
On the one hand, the embodiment of the invention provides a fault criterion model construction method based on an AI target detection model, which comprises the following steps:
acquiring historical operation data of an AI target detection model of a model to be constructed;
acquiring historical fault data in the historical operation data of the AI target detection model of the model to be constructed, and cleaning the historical fault data;
acquiring the historical fault data after data cleaning, and establishing a data fault set;
acquiring operation parameters in the data fault set, extracting key functional parameters of the operation parameters, and establishing a functional model for fitting experiments;
acquiring a functional data set of the functional model in the fitting experiment, and establishing a fault criterion model of the AI target detection model, wherein,
performing data processing on the functional data set of the functional model, and acquiring optimal functional data according to the relation between the functional data in the functional data set of the functional model after the data processing;
And acquiring a functional model of the optimal functional data, and determining a fault criterion model of the AI target detection model according to the functional model of the optimal functional data.
Further, obtaining historical fault data in the historical operation data of the AI target detection model to be modeled, and when the historical fault data is subjected to data cleaning, the method comprises the following steps:
acquiring historical fault data in the historical operation data of the AI target detection model;
removing and cleaning repeated data in the historical fault data of the AI target detection model;
and acquiring historical fault data of the AI target detection model for eliminating and cleaning repeated data, and eliminating error judgment data in the historical fault data of the AI target detection model for eliminating and cleaning repeated data based on a Z-Score method.
Further, when the error judgment data in the historical fault data of the AI target detection model for eliminating and cleaning the repeated data is eliminated based on the Z-Score method, the method comprises the following steps:
acquiring each characteristic data point E in the historical fault data of the AI target detection model eliminating and cleaning repeated data, setting the characteristic data points E as E=E1, E2 and E3 … En, and acquiring a Z-Score value delta E of each characteristic data point E according to a formula Z= \frac { X\mu } { \sigma }, wherein the delta E is (E, zi) and i=1, 2,3 and … i;
Wherein Z is Z-Score of each characteristic data point E, X is actual value of each characteristic data point E in the historical fault data of the AI target detection model, mu is average value of each characteristic data point E in the historical fault data of the AI target detection model, sigma is standard deviation value of each characteristic data point E in the historical fault data of the AI target detection model;
presetting a first preset Z-Score value delta E1 and a second preset Z-Score value delta E2,
judging whether the Z-Score value delta E of the characteristic data point E is an abnormal value according to the magnitude relation between the Z-Score value delta E of the characteristic data point E and the first preset Z-Score value delta E1 and the second preset Z-Score value delta E2;
when delta E1 is less than or equal to delta E < [ delta ] E2, judging the Z-Score value delta E of the characteristic data point E as a normal value;
when delta E < [ delta ] E1, judging the Z-Score value delta E of the characteristic data point E as an abnormal value, and eliminating the characteristic data point E;
when delta E > -delta E2, judging the Z-Score value delta E of the characteristic data point E as an abnormal value, and eliminating the characteristic data point E.
Further, when the historical fault data after data cleaning is obtained and the data fault set is established, the method includes:
Acquiring each fault data of the historical fault data of the AI target detection model after error judgment data are removed and the data characteristics in each fault data;
clustering is carried out according to the data characteristics in each fault data, and a fault data set is established according to each fault data under each cluster.
Further, obtaining the operation parameters in the data fault set, extracting the key functional parameters of the operation parameters, and establishing a functional model for fitting experiments, wherein the method comprises the following steps:
acquiring non-fault data in historical operation data of the AI target detection model of a model to be constructed, and acquiring each operation parameter in the non-fault data;
acquiring the lowest value and the highest value of each operation parameter in the non-fault data, and establishing an operation parameter threshold of the AI target detection model according to the lowest value and the highest value of each operation parameter in the non-fault data;
acquiring each operation parameter in the data fault set, comparing each operation parameter in the data fault set with the operation parameter threshold of the AI target detection model, and acquiring the operation parameters in the data fault set which are not in the operation parameter threshold of the AI target detection model;
And determining each operation parameter in the data fault set, which is not in the operation parameter threshold of the AI target detection model, as an operation high correlation parameter in the data fault set.
Further, when determining each operation parameter in the data failure set that is not in the operation parameter threshold of the AI target detection model as an operation high correlation parameter in the data failure set, the method includes:
acquiring weight coefficients of all operation high-correlation parameters in the data fault set based on a linear regression calculation method;
and determining the high correlation parameter with the maximum absolute value of the weight coefficient of each operation high correlation parameter as the key functional parameter of the data fault set according to the magnitude relation among the weight coefficients of each operation high correlation parameter.
Further, when obtaining the weight coefficient of each operation high correlation parameter in the data fault set based on the linear regression calculation method, the method includes:
acquiring a data variable Y of each fault data in the data fault set;
obtaining a weight coefficient of each operation high correlation parameter in the data fault set according to a linear regression calculation equation Y=beta < 0 > +beta < 1 > X < 1 > +beta < 2 > X < 2 > + … +beta < pXp + >;
Wherein X is the operation high correlation parameter in the data failure set, x=x1, X2, … Xp, β is the weight coefficient of the operation high correlation parameter in the data failure set, and β=β0, β1, β2 … βp is set.
Further, when the data processing is performed on the functional data set of the functional model, and the relationship between the functional data in the functional data set of the functional model after the data processing obtains the optimal functional data, the method includes:
acquiring each piece of functional data in a functional data set of a functional model in the fitting experiment, and eliminating repeated data in each piece of functional data in the functional data set;
and removing abnormal data in each functional data in the functional data set after the repeated data are removed by using a clustering method, and solving the value of k in the k-means cluster by adopting a secondary clustering idea.
Further, the secondary clustering includes:
obtaining each function data in the function data set and a change interval between the function data, and adopting a direct clustering method to obtain a clustering result { C1, C2, ck }, wherein the clustering center of each cluster is { W1, W2, wk };
obtaining a cluster center Wmax with the largest median value of { W1, W2, …, wk }, and a cluster Cmax corresponding to the largest cluster center Wmax;
Obtaining a distance di of each cluster center except Wmax in Wmax and { W1, W2, …, wk }, and setting di=Wmax-Wi;
acquiring optimal functional data in the functional data set according to the acquired relation between the distance di of each clustering center and a preset threshold T; wherein,
when di is more than or equal to T, judging that the distance di between the clustering center Wi and the maximum clustering center Wmax is more than or equal to a preset threshold T, wherein the clustering center Wi and the maximum clustering center Wmax have larger difference in characteristic space, and eliminating the clustering Ci corresponding to the clustering center Wi;
when di is smaller than T, judging that the distance di between the clustering center Wi and the maximum clustering center Wmax is smaller than a preset threshold T, wherein the difference between the clustering center Wi and the maximum clustering center Wmax in a characteristic space is smaller, and combining the clustering Ci corresponding to the clustering center Wi and the maximum clustering Cmax into a new cluster;
and obtaining the number of finally reserved clusters, namely the value of k in a k-means algorithm, wherein the reserved clusters are the optimal functional data in the functional data set.
On the other hand, the embodiment of the invention also provides a fault criterion model construction system based on the AI target detection model, which is applicable to the fault criterion model construction method based on the AI target detection model in the embodiments of the invention, and comprises the following steps:
The data acquisition module is electrically connected with the operation database of the AI target detection model and is used for acquiring historical operation data of the AI target detection model of the model to be constructed;
the first data processing module is electrically connected with the data acquisition module and is used for acquiring historical fault data in the historical operation data of the AI target detection model to be modeled, carrying out data cleaning on the historical fault data, acquiring the historical fault data after data cleaning and establishing the data fault set;
the second data processing module is electrically connected with the first data processing module and is used for acquiring the operation parameters in the data fault set, extracting the key functional parameters of the operation parameters and establishing a functional model to carry out fitting experiments, the second data processing module is also used for acquiring the functional data set of the functional model in the fitting experiments and establishing a fault criterion model of the AI target detection model,
the second data processing module is further configured to perform data processing on the functional data set of the functional model, and obtain optimal functional data according to a relationship between the functional data in the functional data set of the functional model after the data processing, and the second data processing module is further configured to obtain a functional model of the optimal functional data, and determine, according to the functional model of the optimal functional data, a fault criterion model of the AI target detection model.
Compared with the prior art, the fault criterion model construction method and system based on the AI target detection model have the beneficial effects that: by collecting and analyzing historical operational data of the model, a comprehensive understanding of the performance and behavior of the model can be obtained. This helps to identify the behavior of the model under different conditions, providing baseline data for further improvement. Secondly, historical fault data is extracted from the historical operation data and is subjected to data cleaning, which is a key step. By cleaning up the data, potential errors or anomalies can be eliminated, ensuring that models are built based on accurate and reliable information. Subsequently, a data failure set is created, which helps to concentrate the failure data in a manageable data set, facilitating subsequent analysis and modeling. After the data fault set is obtained, the operation parameters are further analyzed, the key functional parameters are extracted, and a functional model is built for fitting experiments. This process helps to understand the relationship between parameters and model performance and provides guidance for subsequent optimization. Through fitting experiments, a functional data set of the functional model is obtained, and a foundation is laid for constructing a fault criterion model. This model will help to better understand the behavior of the model, especially in the face of faults. Finally, by data processing and analysis of the functional data set of the functional model, optimal functional data can be determined, which will be used as a fault criterion model of the AI target detection model. This will help to improve the stability and robustness of the model, reduce the risk of failure, and thus enhance the reliability and practicality of the model.
Drawings
FIG. 1 is a flow chart of a fault criterion model construction method based on an AI target detection model in an embodiment of the invention.
FIG. 2 is a block diagram of a fault criteria model building system based on an AI target detection model in accordance with an embodiment of the invention.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. It should be noted that, without conflict, the embodiments of the present invention and features of the embodiments may be combined with each other. The invention will be described in detail below with reference to the drawings in connection with embodiments.
As shown in fig. 1, the fault criterion model construction method based on the AI target detection model in the embodiment of the invention includes:
step S100, acquiring historical operation data of an AI target detection model of a model to be constructed.
Step 200, acquiring historical fault data in the historical operation data of the AI target detection model of the model to be constructed, and cleaning the historical fault data.
Step S300, historical fault data after data cleaning is obtained, and a data fault set is established.
And step 400, acquiring operation parameters in the data fault set, extracting key functional parameters of the operation parameters, and establishing a functional model to perform a fitting experiment.
Step S500, acquiring a functional data set of a functional model in a fitting experiment, and establishing a fault criterion model of an AI target detection model, wherein the functional data set of the functional model is subjected to data processing, and the relation between the functional data in the functional data set of the functional model after the data processing acquires optimal functional data; and acquiring a functional model of the optimal functional data, and determining a fault criterion model of the AI target detection model according to the functional model of the optimal functional data.
It will be appreciated that, in step S100, by acquiring historical operation data, a reference for the model may be established, and the performance of the model under various circumstances is known, so as to provide a basis for subsequent improvement. Step S200 extracts the historical fault data from the historical operating data and performs data cleaning, which is important for accurately capturing problems and abnormal conditions of the model, and contributes to stability and reliability of the model. Step S300 creates a specialized data fault set that makes the fault data easier to manage and analyze, providing a clear source of data for fault analysis and modeling. Through step S400, the operational parameters in the data failure set are analyzed, the critical functional parameters are extracted, and a functional model is built, which is helpful for understanding the relationship between the parameters and the model performance, and the factors that may lead to failure. Step S500 further builds a fault criterion model of the AI target detection model, and determines optimal functional data by processing a functional data set of the functional model, thereby improving performance and robustness of the model. The process can better predict and diagnose potential faults, reduce instability of the model in practical application, and improve usability and maintainability of the model.
It can be seen that the collection of historical data, the cleaning of fault data, the parameter analysis, the construction of a functional model and the establishment of a fault criterion model are helpful to create a fault criterion model of an AI target detection model with more reliability and high performance. The process can reduce faults and problems, and improve the stability and reliability of the model in practical application, so that a solid foundation is provided for judging the state of the AI target detection model.
Specifically, in some embodiments of the present invention, obtaining historical fault data in historical operation data of an AI target detection model to be modeled, and performing data cleaning on the historical fault data includes: acquiring historical fault data in the historical operation data of the AI target detection model; removing and cleaning repeated data in the historical fault data of the AI target detection model; and acquiring historical fault data of the AI target detection model for eliminating and cleaning the repeated data, and eliminating error judgment data in the historical fault data of the AI target detection model for eliminating and cleaning the repeated data based on a Z-Score method.
Specifically, when removing erroneous judgment data in the historical fault data of the AI target detection model for removing and cleaning duplicate data based on the Z-Score method in some embodiments of the present invention, the method includes: acquiring each characteristic data point E in the historical fault data of the AI target detection model for eliminating and cleaning the repeated data, setting the characteristic data points E as E=E1, E2 and E3 … En, and acquiring a Z-Score value delta E of each characteristic data point E according to a formula Z= \frac { X- \mu } { \sigma }, wherein the delta E is (E, zi), and i=1, 2,3 and … i; wherein Z is Z-Score of each characteristic data point E, X is actual value of each characteristic data point E in the historical fault data of the AI target detection model, mu is average value of each characteristic data point E in the historical fault data of the AI target detection model, sigma is standard deviation value of each characteristic data point E in the historical fault data of the AI target detection model; presetting a first preset Z-Score value delta E1 and a second preset Z-Score value delta E2, and judging whether the Z-Score value delta E of the characteristic data point E is an abnormal value according to the magnitude relation between the Z-Score value delta E of the characteristic data point E and the first preset Z-Score value delta E1 and the second preset Z-Score value delta E2; when delta E1 is less than or equal to delta E < [ delta ] E2, judging the Z-Score value delta E of the characteristic data point E as a normal value; when delta E < [ delta ] E1, judging the Z-Score value delta E of the characteristic data point E as an abnormal value, and eliminating the characteristic data point E; when DeltaE >. DeltaE 2, the Z-Score value DeltaE of the characteristic data point E is judged to be an abnormal value, and the characteristic data point E is eliminated.
It can be appreciated that by eliminating the duplicate data in the historical fault data during the data cleaning stage, it helps to reduce redundant information in the data set, ensuring data quality and efficiency of subsequent analysis and modeling processes. Second, by further screening the data in the historical fault data using the Z-Score method, outlier data points can be effectively identified and culled. Z-Score is a statistical method commonly used to detect outliers by calculating the deviation between the data points and the mean and standard deviation of the data set to determine if the data is outliers. This helps to clear invalid data that may be caused by errors or other anomalies, thereby improving the accuracy and reliability of the data. Finally, by integrating the steps into the data cleaning flow, the used historical fault data set can be ensured to have higher reliability, and more reliable input data is provided for the subsequent establishment of an AI target detection model, so that the performance and reliability of the model are improved. The data cleaning method is beneficial to reducing the erroneous judgment of the model caused by noise data, so that the accuracy and the stability of the model in practical application are improved.
Specifically, in some embodiments of the present invention, acquiring historical fault data after data cleaning, and establishing a data fault set includes: acquiring each fault data of historical fault data of an AI target detection model after error judgment data are removed and data characteristics in each fault data; clustering is carried out according to the data characteristics in each fault data, and a fault data set is established according to each fault data under each cluster.
It will be appreciated that by obtaining cleaned historical fault data and data signatures, a more realistic data set can be provided for analysis, including detailed information for different fault conditions. This helps to more fully understand the performance issues of the model and accurately identify different types of failure modes. Second, by clustering using data features, grouping similar fault data points together helps to sort the fault data into a more organized form. Such clustering can help identify failure modes, find commonalities, and better understand the weaknesses and problems of the model. Finally, further analysis and modeling work will be facilitated by building a fault dataset. It allows data for different failure modes to be more easily used for model improvement, optimization and prediction, thereby improving the performance and reliability of the model. The method is helpful for extracting valuable information from the historical fault data so as to improve and maintain the AI target detection model, reduce the occurrence of potential faults and enhance the stability and usability of the model in practical application.
Specifically, in some embodiments of the present invention, the steps of taking the operation parameters in the data fault set, extracting the key functional parameters of the operation parameters, and establishing a functional model for fitting experiments include: acquiring non-fault data in historical operation data of an AI target detection model of a model to be constructed, and acquiring each operation parameter in the non-fault data; acquiring the lowest value and the highest value of each operation parameter in the non-fault data, and establishing an operation parameter threshold of the AI target detection model according to the lowest value and the highest value of each operation parameter in the non-fault data; acquiring each operation parameter in the data fault set, comparing each operation parameter in the data fault set with an operation parameter threshold of the AI target detection model, and acquiring the operation parameters in the data fault set which are not in the operation parameter threshold of the AI target detection model; and determining each operation parameter in the data fault set which is not in the operation parameter threshold of the AI target detection model as an operation high correlation parameter in the data fault set.
It will be appreciated that by obtaining non-faulty data and various operating parameters from the historical operating data, a benchmark can be created that represents the performance and parameter range of the model under normal operating conditions. This provides an important understanding of the expected behaviour of the model. Secondly, by obtaining the lowest value and the highest value of each operating parameter in the non-fault data and establishing operating parameter thresholds based on these values, a clear boundary can be set for the normal operating range of the model. This helps identify any anomalies that deviate from the normal range. Finally, by comparing the operation parameters in the data fault set with the operation parameter threshold values, the parameters which do not accord with the normal operation range can be identified, so that the operation parameters in the data fault set with higher correlation with the model performance are determined. The method can accurately diagnose the problem, find out parameters related to the fault, and contribute to the improvement and optimization of the model so as to improve the performance and stability of the model.
Specifically, in some embodiments of the present invention, determining each operating parameter in the data failure set that is not in the operating parameter threshold of the AI target detection model as an operating high correlation parameter in the data failure set includes: acquiring weight coefficients of all running high-correlation parameters in a data fault set based on a linear regression calculation method; and determining the high correlation parameter with the maximum absolute value of the weight coefficient of each operation high correlation parameter as the key functional parameter of the data fault set according to the magnitude relation between the weight coefficients of each operation high correlation parameter.
Specifically, in some embodiments of the present invention, obtaining weight coefficients for each operational high correlation parameter in a data fault set based on a linear regression calculation method includes: acquiring a data variable Y of each fault data in the data fault set; obtaining weight coefficients of all operation high-correlation parameters in the data fault set according to a linear regression calculation equation Y=β0+β1x1+β2x2+ … +β pXp +E; wherein X is a high correlation parameter of each operation in the data failure set, and x=x1, X2, … Xp, β is a weight coefficient of each high correlation parameter of each operation in the data failure set, and β=β0, β1, β2 … βp is set.
It can be appreciated that by obtaining the weight coefficients of each run-height correlation parameter based on a linear regression calculation method, the degree of performance impact of these parameters on the target detection model can be measured quantitatively. This helps to determine which parameters play a critical role in the model performance, so that the working principle of the model is better understood. Secondly, according to the magnitude relation among the weight coefficients, which parameter has the greatest influence on the model performance can be determined and regarded as a key functional parameter. This helps to improve and optimize the model more specifically, focusing on solving the most performance critical issues, improving the effectiveness and maintainability of the model. Finally, by determining critical functional parameters, changes in these parameters can be more easily focused and tracked during model monitoring and maintenance, identifying potential problems in time, and taking action. This helps to reduce the risk of failure and improves the reliability and practicality of the model.
It can be seen that the most critical functional parameters in the data fault set are helped to be determined through data analysis and weight coefficient calculation, so that a targeted guide is provided for optimizing and maintaining a criterion model of the AI target detection model, and the performance and reliability of the model are improved.
Specifically, in some embodiments of the present invention, when performing data processing on a functional data set of a functional model, and obtaining optimal functional data according to a relationship between functional data in the functional data set of the functional model after the data processing, the method includes: acquiring each piece of functional data in a functional data set of a functional model in a fitting experiment, and eliminating repeated data in each piece of functional data in the functional data set; and removing abnormal data in each functional data in the functional data set after the repeated data are removed by using a clustering method, and solving the value of k in the k-means cluster by adopting a secondary clustering idea.
Specifically, in some embodiments of the invention the secondary clustering comprises: obtaining each function data in the function data set and a change interval between the function data, and adopting a direct clustering method to obtain a clustering result { C1, C2, ck }, wherein the clustering center of each cluster is { W1, W2, wk }; obtaining a cluster center Wmax with the largest median value of { W1, W2, …, wk }, and a cluster Cmax corresponding to the largest cluster center Wmax; obtaining a distance di of each cluster center except Wmax in Wmax and { W1, W2, …, wk }, and setting di=Wmax-Wi; acquiring optimal functional data in the functional data set according to the acquired relation between the distance di of each clustering center and a preset threshold T; when di is more than or equal to T, judging that the distance di between the clustering center Wi and the maximum clustering center Wmax is more than or equal to a preset threshold T, wherein the clustering center Wi and the maximum clustering center Wmax have larger difference in characteristic space, and eliminating the clustering Ci corresponding to the clustering center Wi; when di is smaller than T, judging that the distance di between the clustering center Wi and the maximum clustering center Wmax is smaller than a preset threshold T, wherein the difference between the clustering center Wi and the maximum clustering center Wmax in a characteristic space is smaller, and combining the clustering Ci corresponding to the clustering center Wi and the maximum clustering Cmax into a new cluster; and obtaining the number of the finally reserved clusters, namely the value of k in a k-means algorithm, wherein the reserved clusters are the optimal functional data in the functional data set.
It will be appreciated that by eliminating duplicate data in the functional data set, a clean and high quality data set is ensured. This helps to reduce interference of redundant information on analysis and modeling, improving usability of the data. And secondly, removing abnormal data by adopting a clustering method, and dividing the data into different clusters by adopting the clustering method, so that the distribution and the characteristics of the data are better understood. The secondary clustering thought is adopted, the problem of selecting the k value in the k-means clustering is solved, the data clusters are divided more accurately, and abnormal data which do not meet the model requirements are eliminated. At the same time, most importantly, by determining the optimal functional data, the data points with the greatest influence on the model performance can be accurately selected. The method is beneficial to optimizing the functional model, improving the prediction performance, reducing the error of the model and enhancing the usability and reliability of the model in practical application.
In summary, the embodiment of the invention provides a fault criterion model construction method based on an AI target detection model, which can obtain comprehensive understanding of model performance and behavior by collecting and analyzing historical operation data of the model. This helps to identify the behavior of the model under different conditions, providing baseline data for further improvement. Secondly, historical fault data is extracted from the historical operation data and is subjected to data cleaning, which is a key step. By cleaning up the data, potential errors or anomalies can be eliminated, ensuring that models are built based on accurate and reliable information. Subsequently, a data failure set is created, which helps to concentrate the failure data in a manageable data set, facilitating subsequent analysis and modeling. After the data fault set is obtained, the operation parameters are further analyzed, the key functional parameters are extracted, and a functional model is built for fitting experiments. This process helps to understand the relationship between parameters and model performance and provides guidance for subsequent optimization. Through fitting experiments, a functional data set of the functional model is obtained, and a foundation is laid for constructing a fault criterion model. This model will help to better understand the behavior of the model, especially in the face of faults. Finally, by data processing and analysis of the functional data set of the functional model, optimal functional data can be determined, which will be used as a fault criterion model of the AI target detection model. This will help to improve the stability and robustness of the model, reduce the risk of failure, and thus enhance the reliability and practicality of the model.
As shown in fig. 2, some embodiments of the present invention further provide a fault criterion model building system based on an AI target detection model, which is applicable to the fault criterion model building method based on the AI target detection model in the foregoing embodiments of the present invention, and includes: the system comprises a data acquisition module, a first data processing module and a second data processing module. The data acquisition module is electrically connected with an operation database of the AI target detection model and is used for acquiring historical operation data of the AI target detection model of the model to be constructed; the first data processing module is electrically connected with the data acquisition module and is used for acquiring historical fault data in the historical operation data of the AI target detection model of the model to be constructed, carrying out data cleaning on the historical fault data, acquiring the historical fault data after the data cleaning, and establishing a data fault set; the second data processing module is electrically connected with the first data processing module, and is used for acquiring operation parameters in a data fault set, extracting key functional parameters of the operation parameters, establishing a functional model to carry out fitting experiments, acquiring a functional data set of the functional model in the fitting experiments, and establishing a fault criterion model of an AI target detection model, wherein the second data processing module is also used for carrying out data processing on the functional data set of the functional model, acquiring optimal functional data according to the relation between the functional data in the functional data set of the functional model after the data processing, acquiring a functional model of the optimal functional data, and determining the functional model of the optimal functional data as the fault criterion model of the AI target detection model.
It can be understood that the fault criterion model construction system based on the AI target detection model in the embodiments of the present invention is applicable to the fault criterion model construction method based on the AI target detection model in the embodiments of the present invention, so that the fault criterion model construction system based on the AI target detection model in the embodiments of the present invention and the fault criterion model construction method based on the AI target detection model have the same beneficial effects, and therefore are not described in detail.
The foregoing is merely an example of the present invention, and the scope of the present invention is not limited thereto, and all changes made in the structure according to the present invention should be considered as falling within the scope of the present invention without departing from the gist of the present invention.
It will be clear to those skilled in the art that, for convenience and brevity of description, the specific working process of the system described above and the related description may refer to the corresponding process in the foregoing method embodiment, which is not repeated here.
It should be noted that, in the system provided in the foregoing embodiment, only the division of the foregoing functional modules is illustrated, in practical application, the foregoing functional allocation may be performed by different functional modules, that is, the modules or steps in the embodiment of the present invention are further decomposed or combined, for example, the modules in the foregoing embodiment may be combined into one module, or may be further split into multiple sub-modules, so as to complete all or part of the functions described above. The names of the modules and steps related to the embodiments of the present invention are merely for distinguishing the respective modules or steps, and are not to be construed as unduly limiting the present invention.
Those of skill in the art will appreciate that the various illustrative modules, method steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the program(s) corresponding to the software modules, method steps, may be embodied in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. To clearly illustrate this interchangeability of electronic hardware and software, various illustrative components and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as electronic hardware or software depends upon the particular application and design constraints imposed on the solution. Those skilled in the art may implement the described functionality using different approaches for each particular application, but such implementation is not intended to be limiting.
The terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus/apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus/apparatus.
Thus far, the technical solution of the present invention has been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of protection of the present invention is not limited to these specific embodiments. Equivalent modifications and substitutions for related technical features may be made by those skilled in the art without departing from the principles of the present invention, and such modifications and substitutions will fall within the scope of the present invention.
The above is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention.
Claims (10)
1. A fault criterion model construction method based on an AI target detection model is characterized by comprising the following steps:
acquiring historical operation data of an AI target detection model of a model to be constructed;
acquiring historical fault data in the historical operation data of the AI target detection model of the model to be constructed, and cleaning the historical fault data;
acquiring the historical fault data after data cleaning, and establishing a data fault set;
acquiring operation parameters in the data fault set, extracting key functional parameters of the operation parameters, and establishing a functional model for fitting experiments;
Acquiring a functional data set of the functional model in the fitting experiment, and establishing a fault criterion model of the AI target detection model, wherein,
performing data processing on the functional data set of the functional model, and acquiring optimal functional data according to the relation between the functional data in the functional data set of the functional model after the data processing;
and acquiring a functional model of the optimal functional data, and determining a fault criterion model of the AI target detection model according to the functional model of the optimal functional data.
2. The AI-target-detection-model-based fault criterion model construction method according to claim 1, wherein acquiring historical fault data in historical operation data of the AI target detection model to be modeled and performing data cleaning on the historical fault data comprises:
acquiring historical fault data in the historical operation data of the AI target detection model;
removing and cleaning repeated data in the historical fault data of the AI target detection model;
and acquiring historical fault data of the AI target detection model for eliminating and cleaning repeated data, and eliminating error judgment data in the historical fault data of the AI target detection model for eliminating and cleaning repeated data based on a Z-Score method.
3. The method for constructing a fault criteria model based on an AI target detection model according to claim 2, wherein when removing erroneous judgment data in historical fault data of the AI target detection model from which duplicate data is removed based on a Z-Score method, comprising:
acquiring each characteristic data point E in the historical fault data of the AI target detection model eliminating and cleaning repeated data, setting the characteristic data points E as E=E1, E2 and E3 … En, and acquiring a Z-Score value delta E of each characteristic data point E according to a formula Z= \frac { X\mu } { \sigma }, wherein the delta E is (E, zi) and i=1, 2,3 and … i;
wherein Z is Z-Score of each characteristic data point E, X is actual value of each characteristic data point E in the historical fault data of the AI target detection model, mu is average value of each characteristic data point E in the historical fault data of the AI target detection model, sigma is standard deviation value of each characteristic data point E in the historical fault data of the AI target detection model;
presetting a first preset Z-Score value delta E1 and a second preset Z-Score value delta E2,
judging whether the Z-Score value delta E of the characteristic data point E is an abnormal value according to the magnitude relation between the Z-Score value delta E of the characteristic data point E and the first preset Z-Score value delta E1 and the second preset Z-Score value delta E2;
When delta E1 is less than or equal to delta E < [ delta ] E2, judging the Z-Score value delta E of the characteristic data point E as a normal value;
when delta E < [ delta ] E1, judging the Z-Score value delta E of the characteristic data point E as an abnormal value, and eliminating the characteristic data point E;
when delta E > -delta E2, judging the Z-Score value delta E of the characteristic data point E as an abnormal value, and eliminating the characteristic data point E.
4. The AI-target-detection-model-based fault criteria model construction method of claim 3, wherein, when acquiring the historical fault data after data cleaning and establishing the data fault set, comprising:
acquiring each fault data of the historical fault data of the AI target detection model after error judgment data are removed and the data characteristics in each fault data;
clustering is carried out according to the data characteristics in each fault data, and a fault data set is established according to each fault data under each cluster.
5. The method for constructing a fault criteria model based on an AI objective detection model of claim 4, wherein obtaining the operating parameters in the data fault set, extracting the critical functional parameters of the operating parameters, and building a functional model for fitting experiments comprises:
Acquiring non-fault data in historical operation data of the AI target detection model of a model to be constructed, and acquiring each operation parameter in the non-fault data;
acquiring the lowest value and the highest value of each operation parameter in the non-fault data, and establishing an operation parameter threshold of the AI target detection model according to the lowest value and the highest value of each operation parameter in the non-fault data;
acquiring each operation parameter in the data fault set, comparing each operation parameter in the data fault set with the operation parameter threshold of the AI target detection model, and acquiring the operation parameters in the data fault set which are not in the operation parameter threshold of the AI target detection model;
and determining each operation parameter in the data fault set, which is not in the operation parameter threshold of the AI target detection model, as an operation high correlation parameter in the data fault set.
6. The AI-target-detection-model-based fault criteria model construction method of claim 5, wherein determining each operating parameter in the data fault set that is not in the AI target detection model's operating parameter threshold as an operating high-correlation parameter in the data fault set comprises:
Acquiring weight coefficients of all operation high-correlation parameters in the data fault set based on a linear regression calculation method;
and determining the high correlation parameter with the maximum absolute value of the weight coefficient of each operation high correlation parameter as the key functional parameter of the data fault set according to the magnitude relation among the weight coefficients of each operation high correlation parameter.
7. The AI-target-detection-model-based fault criteria model construction method of claim 6, wherein obtaining the weight coefficients of each operational high-correlation parameter in the data fault set based on a linear regression calculation method comprises:
acquiring a data variable Y of each fault data in the data fault set;
obtaining a weight coefficient of each operation high correlation parameter in the data fault set according to a linear regression calculation equation Y=beta < 0 > +beta < 1 > X < 1 > +beta < 2 > X < 2 > + … +beta < pXp + >;
wherein X is the operation high correlation parameter in the data failure set, x=x1, X2, … Xp, β is the weight coefficient of the operation high correlation parameter in the data failure set, and β=β0, β1, β2 … βp is set.
8. The AI-object-detection-model-based fault criteria model construction method of claim 6, wherein when data processing is performed on the functional data set of the functional model and the relationship between the functional data in the functional data set of the functional model after the data processing obtains optimal functional data, comprising:
Acquiring each piece of functional data in a functional data set of a functional model in the fitting experiment, and eliminating repeated data in each piece of functional data in the functional data set;
and removing abnormal data in each functional data in the functional data set after the repeated data are removed by using a clustering method, and solving the value of k in the k-means cluster by adopting a secondary clustering idea.
9. The AI-target-detection-model-based fault criteria model construction method of claim 8, wherein the secondary clustering comprises:
obtaining each function data in the function data set and a change interval between the function data, and adopting a direct clustering method to obtain a clustering result { C1, C2, ck }, wherein the clustering center of each cluster is { W1, W2, wk };
obtaining a cluster center Wmax with the largest median value of { W1, W2, …, wk }, and a cluster Cmax corresponding to the largest cluster center Wmax;
obtaining a distance di of each cluster center except Wmax in Wmax and { W1, W2, …, wk }, and setting di=Wmax-Wi;
acquiring optimal functional data in the functional data set according to the acquired relation between the distance di of each clustering center and a preset threshold T; wherein,
When di is more than or equal to T, judging that the distance di between the clustering center Wi and the maximum clustering center Wmax is more than or equal to a preset threshold T, wherein the clustering center Wi and the maximum clustering center Wmax have larger difference in characteristic space, and eliminating the clustering Ci corresponding to the clustering center Wi;
when di is smaller than T, judging that the distance di between the clustering center Wi and the maximum clustering center Wmax is smaller than a preset threshold T, wherein the difference between the clustering center Wi and the maximum clustering center Wmax in a characteristic space is smaller, and combining the clustering Ci corresponding to the clustering center Wi and the maximum clustering Cmax into a new cluster;
and obtaining the number of finally reserved clusters, namely the value of k in a k-means algorithm, wherein the reserved clusters are the optimal functional data in the functional data set.
10. The fault criterion model construction system based on the AI target detection model, which is applicable to the fault criterion model construction method based on the AI target detection model as claimed in any one of claims 1 to 9, and is characterized by comprising:
the data acquisition module is electrically connected with the operation database of the AI target detection model and is used for acquiring historical operation data of the AI target detection model of the model to be constructed;
The first data processing module is electrically connected with the data acquisition module and is used for acquiring historical fault data in the historical operation data of the AI target detection model to be modeled, carrying out data cleaning on the historical fault data, acquiring the historical fault data after data cleaning and establishing the data fault set;
the second data processing module is electrically connected with the first data processing module and is used for acquiring the operation parameters in the data fault set, extracting the key functional parameters of the operation parameters and establishing a functional model to carry out fitting experiments, the second data processing module is also used for acquiring the functional data set of the functional model in the fitting experiments and establishing a fault criterion model of the AI target detection model,
the second data processing module is further configured to perform data processing on the functional data set of the functional model, and obtain optimal functional data according to a relationship between the functional data in the functional data set of the functional model after the data processing, and the second data processing module is further configured to obtain a functional model of the optimal functional data, and determine, according to the functional model of the optimal functional data, a fault criterion model of the AI target detection model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311193948.9A CN117315408A (en) | 2023-09-15 | 2023-09-15 | Fault criterion model construction method and system based on AI target detection model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311193948.9A CN117315408A (en) | 2023-09-15 | 2023-09-15 | Fault criterion model construction method and system based on AI target detection model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117315408A true CN117315408A (en) | 2023-12-29 |
Family
ID=89259469
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311193948.9A Pending CN117315408A (en) | 2023-09-15 | 2023-09-15 | Fault criterion model construction method and system based on AI target detection model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117315408A (en) |
-
2023
- 2023-09-15 CN CN202311193948.9A patent/CN117315408A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111459700B (en) | Equipment fault diagnosis method, diagnosis device, diagnosis equipment and storage medium | |
| KR102141391B1 (en) | Failure data management method based on cluster estimation | |
| CN116559598B (en) | Smart distribution network fault positioning method and system | |
| CN115643159B (en) | Equipment abnormity early warning method and system based on edge calculation | |
| CN108304348B (en) | Bearing residual life prediction method based on binary wiener process | |
| CN115828170B (en) | Fault detection method based on electronic control data of tunneling machine | |
| CN106708738B (en) | Software test defect prediction method and system | |
| CN108763048B (en) | Hard disk fault early warning and reliability evaluation method based on particle filtering | |
| CN112000081B (en) | Fault monitoring method and system based on multi-block information extraction and Mahalanobis distance | |
| CN116028887B (en) | Analysis method of continuous industrial production data | |
| CN111077876B (en) | Power station equipment state intelligent evaluation and early warning method, device and system | |
| JP2009070834A (en) | Method and device of classifying substrate, method and device of estimating abnormal facility, program for executing the method of classifying substrate or the method of estimating abnormal facility by computer, and computer readable recording medium storing the program | |
| CN117315408A (en) | Fault criterion model construction method and system based on AI target detection model | |
| CN112346893A (en) | Fault prediction method, device, terminal and storage medium | |
| Jobi-Taiwo et al. | Mahalanobis-Taguchi system for multiclass classification of steel plates fault | |
| CN112016193A (en) | Online prediction method and system for lubrication failure of shield tunneling machine system | |
| CN1920863A (en) | Real time management system for production management and yield analytical integration and processing method thereof | |
| CN115659271A (en) | Sensor abnormality detection method, model training method, system, device, and medium | |
| WO2020230436A1 (en) | Diagnosis device, diagnosis method, and machining device | |
| US20100010763A1 (en) | Method for detecting variance in semiconductor processes | |
| CN114637793B (en) | Equipment fault frequent region positioning method based on big data analysis | |
| CN117665221B (en) | Performance detection method and system for composite material product | |
| CN117647725B (en) | Aging test method and system for PCBA | |
| CN117932520B (en) | Solid biological waste treatment equipment monitoring method based on data identification | |
| CN115488696B (en) | Cutter abrasion prediction method based on variation modal decomposition and neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |