CN115496218B - Weld defect real-time detection method integrating evolutionary algorithm and fuzzy inference - Google Patents
Weld defect real-time detection method integrating evolutionary algorithm and fuzzy inference Download PDFInfo
- Publication number
- CN115496218B CN115496218B CN202211433127.3A CN202211433127A CN115496218B CN 115496218 B CN115496218 B CN 115496218B CN 202211433127 A CN202211433127 A CN 202211433127A CN 115496218 B CN115496218 B CN 115496218B
- Authority
- CN
- China
- Prior art keywords
- fuzzy inference
- fuzzy
- classification threshold
- rule
- evolutionary algorithm
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/048—Fuzzy inferencing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention discloses a welding defect real-time detection method fusing an evolutionary algorithm and fuzzy inference, which is characterized in that an evolutionary algorithm iterates a fuzzy inference rule, continuously optimizes the fuzzy inference rule, respectively obtains optimal fuzzy inference rule solutions under different classification threshold values, iterates the classification threshold values on the basis, obtains a classification threshold value corresponding to the highest accuracy rate, and further obtains a fuzzy inference system by reverse reasoning. The fuzzy inference rule and the classification threshold can be optimized, and during actual detection, the optimal fuzzy inference system and the optimal classification threshold can be found by selecting characteristics according to different welding defects; inputting the time sequence data into a fuzzy reasoning system, and comparing the output abnormal credibility with a corresponding classification threshold value to realize defect judgment; compared with the prior art, the method can realize the optimization effect of the fuzzy inference rule by iterating through two super parameters of the solution of the fuzzy inference rule and the classification threshold value.
Description
Technical Field
The invention belongs to the technical field of welding defect detection, and particularly relates to a welding defect real-time detection method fusing an evolutionary algorithm and fuzzy reasoning.
Background
Fuzzy reasoning is used as a branch of approximate reasoning and is the theoretical basis of fuzzy control. In practical application, the method is characterized by numerical calculation, replaces the traditional symbolic deduction, and calculates a conclusion from the premise of reasoning (rule antecedent) through a fuzzy reasoning algorithm. In the last 30 years, the successful use of fuzzy inference methods in industrial production control has led to the growing emphasis of fuzzy inference in the field of automatic control.
At present, a welding defect real-time detection method combining fuzzy inference and edge calculation exists in the prior art, and Chinese patent CN114700587A (published Japanese 2022.7.5) discloses a welding defect real-time detection method and a welding defect real-time detection system based on fuzzy inference and edge calculation. According to the scheme, only the classification threshold is iterated, whether the final fuzzy inference rule has the highest accuracy cannot be determined, the fuzzy inference rule cannot be iterated, and further the optimization of the rule cannot be better achieved.
Disclosure of Invention
The purpose of the invention is as follows: aiming at the problems in the background technology, the invention provides a welding defect real-time detection method fusing an evolutionary algorithm and fuzzy inference, which is characterized in that an evolutionary algorithm iterates a fuzzy inference rule and continuously optimizes the fuzzy inference rule to respectively obtain optimal fuzzy inference rule solutions under different classification threshold values, and the classification threshold values are further iterated on the basis to obtain the classification threshold values corresponding to the highest accuracy rate, so that a fuzzy inference system is obtained. The fuzzy inference rule and the classification threshold can be optimized, and during actual measurement, the optimal fuzzy inference system and the optimal classification threshold can be found by selecting characteristics according to different welding defects.
The technical scheme is as follows:
a welding defect real-time detection method integrating an evolutionary algorithm and fuzzy reasoning comprises the following steps:
s1, acquiring high-frequency welding time sequence data, constructing a sample according to a preset sliding window length, and marking the sample according to actual welding defects;
s2, extracting specific sample characteristics; the sample characteristics comprise time domain characteristics, frequency domain characteristics and time-frequency domain characteristics; selecting n specific sample characteristics from the characteristics, and constructing a fuzzy inference characteristic library;
s3, taking the specific sample characteristics extracted in the step S2 as input, and taking the abnormal credibility as output, and constructing a fuzzy inference system, wherein the abnormal credibility domain is [0,1];
s4, setting rule antecedents in the fuzzy inference system according to the selected specific sample characteristics, determining the fuzzy inference rule according to the output abnormal credibility, and further acquiring a solution space of the fuzzy inference rule;
s5, setting an evaluation function, and evaluating whether the final determination solution is the optimal solution; the evaluation function comprises a classification threshold value and 2 hyper-parameters of a fuzzy inference rule solution; specifically, iteration classification threshold values are adopted, a plurality of rounds of fuzzy inference systems are iterated through an evolutionary algorithm in each iteration process, solutions output in each round of iteration are transmitted into an evaluation function, the accuracy is calculated according to the current classification threshold values, and solutions corresponding to the highest accuracy under the current classification threshold values, namely the optimal fuzzy inference rule, are obtained; iterating all the classification threshold values, and finding out the classification threshold value with the highest accuracy and the corresponding optimal fuzzy inference rule;
s6, reversely resolving the solution of the obtained optimal fuzzy inference rule into a fuzzy inference rule to obtain a final fuzzy inference system;
and S7, deploying a fuzzy inference system on the edge side, extracting corresponding specific sample characteristics by acquiring high-frequency time sequence data transmitted by the multi-dimensional sensor, inputting the specific sample characteristics into the fuzzy inference system, and judging whether the welding defects exist or not by comparing the output abnormal credibility with the classification threshold with the highest accuracy.
Further, the high-frequency time sequence data collected in the step S1 includes welding current, welding voltage, wire feeding speed and shielding gas flow rate.
Further, n specific sample features are selected from the sample features based on a Relie-F algorithm in the step S2, and a fuzzy inference feature library is constructed; wherein the n specific sample features are the first n features of all the features with the feature weights arranged from high to low.
Further, the specific method for constructing the fuzzy inference system in step S3 includes:
s3.1, recording the input specific sample characteristics as TF1, TF2, \ 8230, TFn; the discourse domain of the specific sample characteristics is [ Min (Train [ Fea ]), max (Train [ Fea ]) ]; wherein Min (Train [ Fea ]), max (Train [ Fea ]) sequentially represent the minimum value and the maximum value in the sample characteristic set;
s3.2, fuzzy processing is carried out on the input sample characteristics by adopting a trapezoidal or triangular membership function, and an input fuzzy set is obtained as follows:
TF1→{TF1_low,TF1_middle,TF1_high}
TF2→{TF2_low,TF2_middle,TF2_high}
…
TFn→{TFn_low,TFn_middle,TFn_high}
and step 3.3, performing fuzzification processing on the output similarly, wherein the fuzzy set of the output abnormal credibility A is as follows: { A _ low, A _ middle, and A _ high }, which in turn represent low, medium, and high confidence levels of the anomaly.
Further, a rule antecedent is determined based on the input fuzzy set obtained in step S3.2, the rule antecedent comprisingThe bar, i.e. the correspondingly generated fuzzy inference rule, includes>A strip; the output abnormal confidence fuzzy set is serialized as follows: a _ low =0, a _middle =1, a _high =2, the solution space length of the final fuzzy inference rule is £ h =>Value range of。
Further, in step S5, a classification threshold is set to [0,1], the step length is 0.1, and in each iterative classification threshold process, an evolutionary algorithm is used to iterate 5000 rounds of fuzzy inference systems.
Further, a SimpleGA algorithm is adopted in the step S6 to iterate a fuzzy inference system.
Further, an openAI ES algorithm is adopted in the step S6 to iterate a fuzzy inference system.
Further, the time domain features in step S2 include a mean, a median, a maximum, a minimum, a variance, a standard deviation, a quantile, a square root amplitude, a root mean square, a peak-to-peak value, a skewness, a kurtosis, a peak factor, a margin factor, a form factor, and a pulse index; the frequency domain features include a frequency mean with an amplitude greater than 60, a frequency minimum with an amplitude greater than 60, a 1/4 quantile with an amplitude greater than 60 for frequencies, a 3/4 quantile with an amplitude greater than 60 for frequencies, a frequency with a second largest amplitude, a frequency at a maximum amplitude, a frequency with a third largest amplitude; VMD conversion is carried out on the sample data, and the time-frequency domain characteristics of the sample are extracted; the time-frequency domain features comprise zero crossing rate of the modal component IMF1 and five crossing rate of the modal component IMF 2.
Compared with the prior art, the technical scheme adopted by the invention has the following beneficial effects:
(1) According to the invention, appropriate sample characteristics can be selected according to specific welding defect conditions, the number of the characteristics does not need to be fixed, and the more the selected characteristics are, the closer the final fuzzy inference system is to displaying the welding conditions. The most important characteristic of a certain defect is artificially selected according to different actual welding environments, so that a fuzzy inference system suitable for a specific defect can be constructed, and the fuzzy inference system has good adaptation performance.
(2) The method designed by the invention integrates an evolutionary algorithm, continuously iterates fuzzy inference rules in the process of iterating classification threshold values, finally obtains the accuracy under different classification threshold values and the corresponding fuzzy inference rules, finally compares the highest accuracy under each classification threshold value and the corresponding fuzzy inference rules, namely, reversely deduces a fuzzy inference system, inputs real-time data into the fuzzy inference system, and compares the output abnormal credibility with the corresponding classification threshold value, namely, realizes defect judgment. Compared with the prior art, the fuzzy inference rule optimization effect can be realized by iterating the solution of the fuzzy inference rule and the classification threshold value.
Drawings
FIG. 1 is a diagram illustrating a result of permutation and combination of rule antecedents provided in an embodiment of the present invention;
FIG. 2 is a diagram illustrating a part of a procedure of an iterative process of fuzzy rules according to an embodiment of the present invention;
fig. 3 is a diagram of the result of the optimal fuzzy rule provided in the embodiment of the present invention.
Detailed Description
The invention will be further explained by the following description of an embodiment thereof, which is provided in conjunction with the accompanying drawings.
The embodiment provides a welding defect real-time detection method integrating an evolutionary algorithm and fuzzy inference, which comprises the following steps of:
s1, acquiring high-frequency welding time sequence data, constructing a sample according to a preset sliding window length, and carrying out data marking on the sample according to actual welding defects. It should be noted here that, when the fuzzy inference system is trained in the early stage, the high-frequency time series data to be acquired includes the welding current, the welding voltage, the wire feeding speed and the shielding gas flow rate. And (3) carrying out a sliding Window construction sample according to a preset sliding Window length Window _ Size =10s (the step is 1 s), namely, taking long-time sequence data of each Window as a sample, and carrying out data annotation on the sample according to the real welding defect condition, wherein the normal label is 0, and the abnormal label is 1.
S2, extracting specific sample characteristics; the sample characteristics comprise time domain characteristics, frequency domain characteristics and time-frequency domain characteristics; and selecting n specific sample characteristics from the sample characteristics to construct a fuzzy inference characteristic library. The fuzzy inference characteristic library comprises the following characteristics:
(1) Time domain characterization
And extracting time domain characteristics of each sample data, further extracting square root amplitude, root mean square, peak-to-peak value, skewness, kurtosis, peak value factor, margin factor, form factor, pulse index and other time domain characteristics on the basis of the traditional time domain characteristics such as mean value, median, maximum value, minimum value, variance, standard deviation and quantile, and forming a time domain characteristic library of the sample.
(2) Extracting related characteristics of a frequency spectrum obtained by Fourier transform of each sample data, further extracting related spectrum characteristics on the basis of frequency-related time domain characteristics (including a mean value, a median, a maximum value, a minimum value, a variance, a standard deviation and a quantile), wherein the related spectrum characteristics include but are not limited to a mean value of frequencies with amplitudes larger than 60, a minimum value of frequencies with amplitudes larger than 60, a 1/4 quantile value of frequencies with amplitudes larger than 60, a 3/4 quantile value of frequencies with amplitudes larger than 60, a frequency with the second largest amplitude, a frequency at the maximum amplitude and a frequency with the third largest amplitude, and constructing a frequency domain characteristic library of the samples.
(3) Time-frequency domain characteristics
And extracting relevant features of a time-frequency spectrum obtained by VMD conversion of each sample data, such as zero crossing rate of modal component IMF1, five crossing rate of modal component IMF2 and the like, and constructing a time-frequency domain feature library of the sample. Compared with the EMD, the VMD can specify the number of the split modal components, and greatly improve the calculation speed.
And selecting n specific sample features from the feature library to construct a fuzzy inference feature library. For fuzzy inference, the more features mean that an extremely large fuzzy inference rule base needs to be constructed. In order to solve the problem, a sample feature of the feature weight top n is extracted from the features by utilizing a Relie-F algorithm, and finally a fuzzy inference feature library is formed. The feature data extracted in the early stage is continuity data, the target is classification data, and a Relie-F algorithm is preferentially selected for calculating the correlation of the problems. In this example, n =2 is taken, and the selected sample characteristics are as follows:
TF1: a spectral peak; TF2: an IMF2 feature;
and S3, taking the specific sample characteristics extracted in the step S2 as input, and taking the abnormal credibility as output, and constructing a fuzzy inference system.
For inputs TF1, TF2, their universe of discourse is [ Min (Train [ Fea ]), max (Train [ Fea ]) ]; wherein Min (Train [ Fea ]), max (Train [ Fea ]) sequentially represent the minimum value and the maximum value in the sample feature set. Fuzzification processing is carried out by utilizing a trapezoidal or triangular membership function, and an input fuzzy set is obtained as follows:
TF1→{TF1_low,TF1_middle,TF1_high}
TF2→{TF2_low,TF2_middle,TF2_high}
for the output of the fuzzy system: and the abnormality reliability A sets the domain of discourse as [0,1], and outputs a fuzzy set as { A _ low, A _ middle and A _ high }, wherein the output fuzzy set respectively represents low abnormality reliability, medium abnormality reliability and high abnormality reliability.
Fuzzy inference rule antecedents are determined below. In this embodiment, there are two features, each feature has a length of the smear set of three, and after the arrangement and combination, there are 9 pieces of the regular front piece, as shown in fig. 1. The fuzzy rule formed by the corresponding output anomaly confidence level a has 9 pieces. And (3) carrying out serialization representation on fuzzy rule output: a _ low =0, a middle =1, a high =2, the solution space length of the final fuzzy inference rule is 9, and the range is。
And S5, setting an evaluation function, and evaluating whether the final determination solution is the optimal solution. The merit function includes a classification threshold and a solution 2 hyper-parameters of the fuzzy inference rule. As shown in fig. 2, the classification threshold is iterated, the classification threshold is set to [0,1], the step length is 0.1, and in the process of each iterative classification threshold, an evolutionary algorithm is adopted to iterate 5000 rounds of fuzzy inference systems. Transmitting the solution output by each iteration into an evaluation function, and calculating the accuracy according to the current classification threshold to obtain the solution corresponding to the highest accuracy under the current classification threshold, namely the optimal fuzzy inference rule; and iterating all the classification threshold values, and finding out the classification threshold value with the highest accuracy and the corresponding optimal fuzzy inference rule. Wherein the function topic structure in FIG. 2 is: and loading the marked data, operating a fuzzy reasoning system, iterating the characteristic data, outputting a fuzzy reasoning result, and calculating the accuracy according to a threshold value.
In the actual operation process, the fuzzy inference rule can be iterated by adopting an evolutionary algorithm including a SimpleGA algorithm and an openAI ES algorithm according to different conditions.
The classification threshold and fuzzy inference rule solution corresponding to the highest accuracy obtained are shown in fig. 3. And reversely resolving the obtained optimal fuzzy inference rule solution into a fuzzy inference rule to obtain a final fuzzy inference system.
In formal implementation, a fuzzy reasoning system is deployed on the edge side, corresponding specific sample characteristics are extracted by collecting high-frequency time sequence data transmitted by a multi-dimensional sensor and input into the fuzzy reasoning system, and whether welding defects exist can be judged by comparing the output abnormal credibility with a classification threshold with the highest accuracy.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.
Claims (9)
1. A welding defect real-time detection method integrating an evolutionary algorithm and fuzzy reasoning is characterized by comprising the following steps:
s1, acquiring high-frequency welding time sequence data, constructing a sample according to a preset sliding window length, and marking the sample according to actual welding defects;
s2, extracting specific sample characteristics; the sample characteristics comprise time domain characteristics, frequency domain characteristics and time-frequency domain characteristics; selecting n specific sample characteristics from the sample characteristics, and constructing a fuzzy inference characteristic library;
s3, taking the specific sample characteristics extracted in the step S2 as input, and taking the abnormal credibility as output, and constructing a fuzzy inference system, wherein the abnormal credibility domain is [0,1];
s4, setting rule antecedents in the fuzzy inference system according to the selected specific sample characteristics, determining the fuzzy inference rule according to the output abnormal credibility, and further acquiring a solution space of the fuzzy inference rule;
s5, setting an evaluation function, and evaluating whether the final determination solution is the optimal solution or not; the evaluation function comprises a classification threshold value and 2 hyper-parameters of a fuzzy inference rule solution; specifically, iteration classification threshold values are adopted, a plurality of rounds of fuzzy inference systems are iterated through an evolutionary algorithm in each iteration process, solutions output in each round of iteration are transmitted into an evaluation function, the accuracy is calculated according to the current classification threshold values, and solutions corresponding to the highest accuracy under the current classification threshold values, namely the optimal fuzzy inference rule, are obtained; iterating all the classification threshold values, and finding out the classification threshold value with the highest accuracy and the corresponding optimal fuzzy inference rule;
s6, reversely resolving the solution of the obtained optimal fuzzy inference rule into a fuzzy inference rule to obtain a final fuzzy inference system;
and S7, deploying a fuzzy inference system on the edge side, extracting corresponding specific sample characteristics by acquiring high-frequency time sequence data transmitted by the multi-dimensional sensor, inputting the specific sample characteristics into the fuzzy inference system, and judging whether the welding defects exist or not by comparing the output abnormal credibility with the classification threshold with the highest accuracy.
2. The weld defect real-time detection method by fusion of evolutionary algorithm and fuzzy inference as claimed in claim 1, wherein the high frequency time series data collected in step S1 comprises welding current, welding voltage, wire feeding speed and shielding gas flow rate.
3. The method for detecting the welding defects through the fusion of the evolutionary algorithm and the fuzzy inference in real time as claimed in claim 1, wherein n specific sample characteristics are selected from the sample characteristics based on a Relie-F algorithm in the step S2, and a fuzzy inference characteristic library is constructed; wherein the n specific sample features are the first n features of all the features with the feature weights arranged from high to low.
4. The method for detecting the welding defects by fusing the evolutionary algorithm and the fuzzy inference in real time as claimed in claim 1, wherein the specific method for constructing the fuzzy inference system in the step S3 comprises:
s3.1, recording the input specific sample characteristics as TF1, TF2, \ 8230, TFn; the discourse domain of the specific sample characteristics is [ Min (Train [ Fea ]), max (Train [ Fea ]) ]; wherein Min (Train [ Fea ]), max (Train [ Fea ]) sequentially represent the minimum value and the maximum value in the sample characteristic set;
s3.2, fuzzy processing is carried out on the input sample characteristics by adopting a trapezoidal or triangular membership function, and an input fuzzy set is obtained as follows:
TF1→{TF1_low,TF1_middle,TF1_high}
TF2→{TF2_low,TF2_middle,TF2_high}
…
TFn→{TFn_low,TFn_middle,TFn_high}
and step 3.3, performing fuzzification processing on the output similarly, wherein the fuzzy set of the output abnormal credibility A is as follows: { a _ low, a _ middle, and a _ high }, which in turn represent low, medium, and high confidence in the anomaly.
5. The weld defect real-time detection method by fusion of evolutionary algorithm and fuzzy inference as claimed in claim 4, characterized in that the rule antecedent is determined based on the input fuzzy set obtained in step S3.2, and comprisesThe bar, i.e. the correspondingly generated fuzzy inference rule, includes>A strip; the output abnormal confidence fuzzy set is serialized as follows:a _ low =0, a _middle =1, a _high =2, the solution space length of the final fuzzy inference rule is £ h =>Value field is->。
6. The method for detecting the welding defects through the fusion evolutionary algorithm and the fuzzy inference in real time as claimed in claim 5, wherein in the step S5, a classification threshold value is set to [0,1], the step length is 0.1, and in the process of each iterative classification threshold value, the evolutionary algorithm is adopted to iterate 5000 rounds of fuzzy inference systems.
7. The method for detecting the welding defects through the fusion of the evolutionary algorithm and the fuzzy inference in real time as claimed in claim 6, wherein a SimpleGA algorithm iterative fuzzy inference system is adopted in the step S6.
8. The method for detecting the welding defects through the fusion of the evolutionary algorithm and the fuzzy inference in real time as claimed in claim 7, wherein an openAI ES algorithm is adopted in the step S6 to iterate the fuzzy inference system.
9. The method for detecting the welding defects through the fusion of the evolutionary algorithm and the fuzzy inference in the real time mode as claimed in claim 1, wherein the time domain characteristics in the step S2 comprise a mean value, a median value, a maximum value, a minimum value, a variance, a standard deviation, a quantile, a square root amplitude, a root mean square, a peak-to-peak value, a skewness, a kurtosis, a peak factor, a margin factor, a form factor and a pulse index; the frequency domain features include a frequency mean with an amplitude greater than 60, a frequency minimum with an amplitude greater than 60, a 1/4 quantile with an amplitude greater than 60 for frequencies, a 3/4 quantile with an amplitude greater than 60 for frequencies, a frequency with a second largest amplitude, a frequency at a maximum amplitude, a frequency with a third largest amplitude; VMD transform is carried out on the sample data, and time-frequency domain characteristics of the sample are extracted; the time-frequency domain features comprise zero crossing rate of the modal component IMF1 and five crossing rate of the modal component IMF 2.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211433127.3A CN115496218B (en) | 2022-11-16 | 2022-11-16 | Weld defect real-time detection method integrating evolutionary algorithm and fuzzy inference |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211433127.3A CN115496218B (en) | 2022-11-16 | 2022-11-16 | Weld defect real-time detection method integrating evolutionary algorithm and fuzzy inference |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115496218A CN115496218A (en) | 2022-12-20 |
CN115496218B true CN115496218B (en) | 2023-04-07 |
Family
ID=85115880
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211433127.3A Active CN115496218B (en) | 2022-11-16 | 2022-11-16 | Weld defect real-time detection method integrating evolutionary algorithm and fuzzy inference |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115496218B (en) |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105787563B (en) * | 2014-12-18 | 2018-12-21 | 中国科学院沈阳自动化研究所 | A kind of Rapid matching fuzzy reasoning method based on self-study mechanism |
CN104936148B (en) * | 2015-07-03 | 2018-04-27 | 中南大学 | A kind of WIFI indoor orientation methods based on fuzzy KNN |
CN106228232A (en) * | 2016-07-07 | 2016-12-14 | 淮北师范大学 | A kind of dynamic multi-objective based on fuzzy reasoning Population forecast strategy teaching optimization method |
CN106300977B (en) * | 2016-10-17 | 2018-10-23 | 东南大学 | A kind of single input fuzzy PID control method of Buck types DC-DC converter |
CN110533102A (en) * | 2019-08-30 | 2019-12-03 | 东北林业大学 | Single class classification method and classifier based on fuzzy reasoning |
CN114492527A (en) * | 2022-01-27 | 2022-05-13 | 上海理工大学 | Fuzzy neural network and principal component analysis based surface roughness online prediction method |
-
2022
- 2022-11-16 CN CN202211433127.3A patent/CN115496218B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN115496218A (en) | 2022-12-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112910859B (en) | Internet of things equipment monitoring and early warning method based on C5.0 decision tree and time sequence analysis | |
CN110309771B (en) | GBDT-INSGAII-based EAS (Acoustic magnetic System) label identification algorithm | |
CN112603334B (en) | Spike detection method based on time sequence characteristics and stacked Bi-LSTM network | |
CN113935440A (en) | Iterative evaluation method and system for error state of voltage transformer | |
CN113949549B (en) | Real-time traffic anomaly detection method for intrusion and attack defense | |
CN112600800A (en) | Network risk assessment method based on map | |
CN117421684B (en) | Abnormal data monitoring and analyzing method based on data mining and neural network | |
CN115130499A (en) | Vibration signal-based online measurement and prediction method for slip ratio of bearing retainer under variable working conditions | |
CN116452542A (en) | GIS partial discharge defect diagnosis method based on nerve supervision decision tree | |
CN115496218B (en) | Weld defect real-time detection method integrating evolutionary algorithm and fuzzy inference | |
CN108665001B (en) | Cross-tested idle state detection method based on deep belief network | |
CN116911201B (en) | Analog motor energy efficiency detection system based on deep generation countermeasure network algorithm | |
Pourabdollah et al. | Changes under the hood-a new type of non-singleton fuzzy logic system | |
CN116758922A (en) | Voiceprint monitoring and diagnosing method for transformer | |
CN114896228B (en) | Industrial data stream cleaning model and method based on filtering rule multistage combination optimization | |
CN115659323A (en) | Intrusion detection method based on information entropy theory and convolution neural network | |
Revathi et al. | Rainfall prediction using machine learning classification algorithms | |
CN113610167A (en) | Equipment risk detection method based on metric learning and visual perception | |
Wang et al. | Two-layer generalization boosting model for anomaly detection of e-commerce orders | |
CN112232329A (en) | Multi-core SVM training and alarming method, device and system for intrusion signal recognition | |
CN117147966B (en) | Electromagnetic spectrum signal energy anomaly detection method | |
Li | Research on the application of DT and K-NN-based data mining algorithms in network intrusion monitoring | |
CN117648588B (en) | Meteorological radar parameter anomaly identification method based on correlation network graph cluster analysis | |
CN115659162B (en) | Method, system and equipment for extracting intra-pulse characteristics of radar radiation source signals | |
CN113919237A (en) | Method for online working condition segmentation and fault diagnosis of fan equipment |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |