CN116257764A - Nuclear power operation data correlation analysis method based on multi-scale time window - Google Patents
Nuclear power operation data correlation analysis method based on multi-scale time window Download PDFInfo
- Publication number
- CN116257764A CN116257764A CN202211572537.6A CN202211572537A CN116257764A CN 116257764 A CN116257764 A CN 116257764A CN 202211572537 A CN202211572537 A CN 202211572537A CN 116257764 A CN116257764 A CN 116257764A
- Authority
- CN
- China
- Prior art keywords
- data
- sensor
- state switching
- time
- time window
- 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.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000010219 correlation analysis Methods 0.000 title claims abstract description 15
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 238000001514 detection method Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 description 10
- 238000004458 analytical method Methods 0.000 description 4
- 239000003381 stabilizer Substances 0.000 description 4
- 230000001052 transient effect Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000005094 computer simulation Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000005428 wave function Effects 0.000 description 1
Images
Landscapes
- Monitoring And Testing Of Nuclear Reactors (AREA)
Abstract
Preprocessing original nuclear power operation data based on a nuclear power operation data correlation analysis method of a multi-scale time window; determining a state switching standard, dividing the sensors into different categories according to the characteristics of sensor data of different categories, determining corresponding time windows according to the sensor divisions of the different categories, and judging whether state switching occurs or not; selecting a target sensor, searching a time point when state switching occurs by utilizing a multi-scale time window; searching whether state switching occurs in the corresponding time neighborhood in each sensor data by utilizing the determined time window; and calculating the matching rate of each sensor and the target sensor, and determining the relevance of each sensor and the target sensor. According to the method, the sensors are classified according to the sensor data characteristics through a sliding time window, and the correlation between the found nuclear power data of the matching rate is calculated. The method can reduce the occupied memory in the searching process, does not need to convert data into matters, and improves the searching efficiency.
Description
Technical Field
The invention relates to the technical field of nuclear power operation analysis, in particular to a nuclear power operation data correlation analysis method based on a multi-scale time window.
Background
In the nuclear power operation analysis technology, searching for the sensor data with the relevance is a foundation for constructing an internal system model of a nuclear power plant, and screening the sensor data with the strong relevance from mass nuclear power operation data is an urgent technical requirement. The screening of the relevance data directly influences the accuracy of the subsequent system modeling, and accurate sensor data with strong relevance is very important for constructing a nuclear power plant system model.
For correlation analysis of nuclear power operation data, there are some data correlation analysis methods that can be used as a reference, for example: the Aproril algorithm and the FP-growth algorithm are methods for establishing association rules between data, which are commonly used at present. The april algorithm obtains a frequent item set by repeatedly scanning a data set continuously, and then generates association rules from the frequent item set. However, this method requires repeated scanning of the data set, which causes a problem of large memory occupation. The FP-growth algorithm only needs to scan the data set twice and establish the FP tree, thereby greatly improving the mining efficiency. However, the method still generates a large number of frequent item sets in the mining process, and the processing of high-dimensional data is relatively laborious.
Both the above two methods need to convert the data characteristics into matters, and establish association rules by searching frequent item sets, which also causes the problems of large occupied memory and low searching efficiency, and are not applicable to high-dimensional nuclear power operation data.
Disclosure of Invention
In order to solve the technical problems, the invention provides a nuclear power operation data relevance analysis method based on a multi-scale time window. The method can reduce the occupied memory in the searching process, does not need to convert data into matters, improves the searching efficiency, and has good searching effect on high-dimensional nuclear power operation data.
The technical scheme adopted by the invention is as follows:
the nuclear power operation data correlation analysis method based on the multi-scale time window comprises the following steps:
step 1: preprocessing original nuclear power operation data to eliminate noise influence;
step 2: determining a state switching standard, dividing the sensors into different categories according to the characteristics of sensor data of different categories, determining corresponding time windows according to the sensor divisions of the different categories, and judging whether state switching occurs or not;
step 3: selecting a target sensor, searching a time point when state switching occurs by utilizing a multi-scale time window;
step 4: searching whether state switching occurs in the corresponding time neighborhood in each sensor data by utilizing the time window determined in the step 2;
step 5: and calculating the matching rate of each sensor and the target sensor, and determining the relevance of each sensor and the target sensor.
In the step 2, according to the change characteristics of the sensor data, the time required for the state switching of the sensors of different types is different, and the sensors are classified into three main categories: instantaneous, gradual change and slow change, and making time windows respectively corresponding to different scales, and judging whether the state switching standard occurs.
In the step 2, the state switching criteria are defined as follows: when the amplitude of the data in a certain window is larger than or equal to a certain proportion of the whole amplitude of the data, the state switching is confirmed to occur in the time window, and the basis for judging whether the state switching occurs is as follows:
Data * ＝Data max -Data min (1)；
data * ＝data max -data min (2)；
data * ＞＝K*Data * (3)；
wherein: data max For the maximum value of a certain sensor in the whole time range, data min Data, which is the minimum value over the entire sensor time range * The maximum variation in data amplitude over time. data max Data is the maximum value in the time window of a certain sensor min Data, which is the minimum value in the time window of a certain sensor * K is a scaling factor for the maximum amplitude variation within the time window.
The step 2 comprises the following steps:
s2.1, counting the amplitude variation of each sensor data, and counting the time required by each sensor for one-time state switching;
s2.2, dividing all the sensors into three types of instantaneous variable, gradual variable and slow variable according to the time length required by the data recorded by different sensors to generate one-time state switching.
S2.3, counting the amplitude change belonging to state switching in the data of each sensor, and calculating the proportion of the amplitude of each sensor to the difference between the maximum value and the minimum value in the whole data.
The step needs to have certain knowledge on the data characteristics of each sensor, and the change states of each sensor in the nuclear power historical operation data can be manually extracted. The statistics were made here using three months of historical nuclear power data. The proportion of the state switching amplitude refers to the proportion of the amplitude change of each state switching to the whole data amplitude change, and the calculation mode is shown in the formula (4):
t＝mag/Data * (4)
wherein: mag is the magnitude of the amplitude change during state switching, data * The maximum variation in data amplitude over time. And calculating the proportion of the amplitude change of each state switching according to the formula, recording and counting each proportion coefficient, and finding out the change interval of the proportion. The results are shown in Table 1.
Table 1 statistics of various sensor changing magnitudes
When the state switching of various sensors is obtained, the proportion of the amplitude change to the integral change can be known, and when the amplitude change occurs once in the data, the state switching can be considered once if the proportion of the amplitude change to the integral amplitude change is within the statistical interval. Because the size of the time window can not completely comprise the whole change interval, and the state switching with smaller amplitude change possibly exists in the unknown data, the K value can be smaller than the minimum statistical proportionality coefficient, and the K value is the proportionality coefficient for judging whether the state switching occurs in the formula (3).
S2.4, aiming at sensors of different types, formulating a time window with corresponding length so as to facilitate the subsequent detection of state switching.
The step 3 comprises the following steps:
s3.1, selecting an object sensor: the most visual change is needed to be selected as the target sensor, such as voltage and current, corresponding to the dependent variable.
S3.2, performing state switching detection on the target sensor data by utilizing the sliding window, reserving the window with detected state switching, and discarding the window which is not detected.
S3.3, further detecting the area with the detected state switching by utilizing a sliding window with smaller scale, still reserving the area with the detected state switching, and discarding the area without the detected state switching;
s3.4, repeating the step S3.3 until an accurate state switching time point can be obtained, and reserving one time point for the repeated time point.
The step 4 comprises the following steps:
and S4.1, marking the time points searched in the step 3 in each sensor data.
S4.2, detecting the switching state of the neighborhood of each time point by taking each time point as the center and utilizing the time window of the corresponding category formulated in the step 2.
The method comprises the steps of firstly restoring each searched time point to the time point of each sensor, then judging whether state switching occurs or not by taking the time point as the center according to a classified window belonging to the time point, judging according to formulas (1), (2) and (3), firstly obtaining the total maximum change amplitude Data of Data, then obtaining the maximum fluctuation amplitude Data in the window, finally substituting each value into the formula (3) for judging, and if the formula (3) is met, indicating that the state switching occurs in the time window, otherwise, judging nothing.
And S4.3, recording the occurrence times of state switching detected by each sensor data.
Said step 5 comprises the steps of:
s5.1, calculating the matching rate of each sensor and the target sensor, wherein the calculation formula is as follows:
cor＝(m/n)*100％ (4)
wherein: n is the number of times of state switching of the target sensor, and m is the number of times of state switching of the detected sensor.
S5.2, sorting the matching rate of each sensor from large to small.
S5.3, the high-matching-rate sensor is identified with strong relevance, the medium-matching-rate sensor is checked manually, and the low-matching-rate identification is identified with low relevance or no relevance.
Compared with the existing manual experience searching relevance data, and the Aproril algorithm and the FP-growth algorithm, the nuclear power operation data relevance analysis method based on the multi-scale time window has the following technical effects:
1) The invention can automatically analyze the high-dimensional nuclear power operation data to find out the data with strong relevance.
2) Compared with the Aproril algorithm and the FP-growth algorithm, the method occupies less memory in the searching process and has higher searching efficiency.
3) The invention does not need to convert data into matters, and the result is more visual.
Drawings
FIG. 1 is a flowchart of the search algorithm according to the present invention.
FIG. 2 (a) is a state switching label diagram of transient (current);
fig. 2 (b) is a partially enlarged view of fig. 2 (a).
Fig. 3 (a) is a state switching map of the gradation amount (electric power);
fig. 3 (b) is a partially enlarged view of fig. 3 (a).
FIG. 4 (a) is a state switching label diagram of a slow variable (temperature);
fig. 4 (b) is a partially enlarged view of fig. 4 (a).
Detailed Description
As shown in fig. 1, the nuclear power operation data correlation analysis method based on a multi-scale time window comprises the following steps:
step 1: preprocessing original nuclear power operation data to eliminate noise influence;
and selecting a method for decomposing and reconstructing the wavelet packet to realize data noise reduction. And determining a proper maximum decomposition layer number for an input signal according to the data length and the wave function, decomposing the signal into a plurality of components according to a wavelet tree, then reordering the components of each layer according to frequency information, selecting proper components according to a set rule, and reconstructing the data to achieve the noise reduction effect.
Step 2: determining a state switching standard, classifying sensors according to the characteristics of sensor data, determining corresponding time windows according to different sensor divisions, and judging whether state switching occurs or not; the method comprises the following steps:
s2.1, counting the amplitude change of each sensor data, and counting the time required by each sensor for one-time state switching.
S2.2, according to the time length required by the data recorded by different sensors to generate one-time state switching, all the sensors can be divided into three types of instantaneous variable, gradual variable and slow variable. For example: a state switch of the current generally occurs instantaneously and can be categorized as a transient. The pressure and power will typically increase gradually, but the magnitude of the change will typically be large, which may be categorized as a gradual change. The long duration of the temperature change often lasts several hours and can therefore be classified as a slow variable.
S2.3, counting the amplitude change belonging to state switching in the data of each sensor, and calculating the proportion of the amplitude of each amplitude to the difference between the maximum value and the minimum value in the whole data, so as to know whether the amplitude change of a certain amplitude can reach the category regarded as state switching.
S2.4, aiming at sensors of different types, formulating a time window with corresponding length so as to facilitate the subsequent detection of state switching.
Step 3: selecting a target sensor, searching a time point when state switching occurs by utilizing a multi-scale time window; the method comprises the following steps:
s3.1, selecting the target sensor is equivalent to selecting a dependent variable, and the most visual and obvious variable is required to be selected as the target sensor, such as voltage, current and the like.
S3.2, performing state switching detection on the target sensor data by utilizing the sliding window, reserving the window with detected state switching, and discarding the window which is not detected.
And S3.3, further detecting the area with the detected state switching by utilizing a sliding window with a smaller scale, still reserving the area with the detected state switching, and discarding the area with the undetected state switching.
S3.4, repeating the third step until an accurate state switching time point (the time window is smaller than or equal to 2) can be obtained, and reserving one time point for the repeated time point.
Step 4: searching whether state switching occurs in the corresponding time neighborhood in each sensor data by utilizing the time window formulated in the step 2; the method comprises the following steps:
and S4.1, marking the time points searched in the step 3 in each sensor data.
S4.2, detecting the switching state of the neighborhood of each time point by taking each time point as the center and utilizing the time window of the corresponding category formulated in the step 2.
And S4.3, recording the occurrence times of state switching detected by each sensor data.
Step 5: calculating the matching rate of each sensor and the target sensor, and determining the relevance of the sensors, wherein the matching rate comprises the following steps:
s5.1, calculating the matching rate of each sensor and the target sensor.
S5.2, sorting the matching rate of each sensor from large to small.
S5.3, the high-matching-rate (50% and above) sensors are identified with strong relevance, the medium-matching-rate (25% to 50%) sensors can be manually checked, and the low-matching-rate sensors can be identified as low relevance or no relevance identification.
Examples:
1: the experimental data of the method comes from the actual operation data of a nuclear power plant in China, and the data set contains 29 sensor parameters, namely MI (current sensor), ZV (fan switch), MY (generator electric power), VE (reactor thermal power), P0 (pump switch), MT (temperature sensor), MV (voltage sensor), KM (loop average temperature and voltage stabilizer pressure). Each sensor recorded from 2019-07-01t00:00 to 2019-12-30t23:59:59, with a sampling frequency of once a second.
2: the data from 2019-07-01T00:00 to 2019-9-30T23:59:59 are used for statistics, the characteristics of switching of each sensor are counted, 29 sensors can be divided into three types of transient (current, voltage, pump and fan), gradual change (electric power, reactor thermal power and voltage stabilizer pressure) and slow change (temperature), the corresponding three types of detection window sizes are transient (the first 3 minutes and the last 3 minutes of a time point, the proportionality coefficient is 0.1), gradual change (the first 3 minutes and the last 30 minutes of the time point are adopted, the proportionality coefficient is 0.08), slow change (the first 3 minutes and the last 1 hour are adopted, the proportionality coefficient is 0.03), and the data are used as initial setting values of an algorithm.
The basis for determining whether the state switching occurs is as follows:
Data * ＝Data max -Data min (1)
data * ＝data max -data min (2)
data * ＞＝K*Data * (3)
wherein: data max Data is the maximum value of a certain sensor in the whole time range min For the whole sensorMinimum value in inter-range, data * The maximum variation in data amplitude over time. data max Data is the maximum value in the time window of a certain sensor min Data, which is the minimum value in the time window of a certain sensor * Is the maximum amplitude change within the time window. K is a proportionality coefficient and corresponds to three different types of sensors.
Fig. 2 (a), 2 (b), 3 (a), 3 (b), 4 (a), and 4 (b) are state switching labels of three different types of sensors, i.e., electric current, electric power, and temperature, respectively, and an enlarged view of the part.
It can be seen from fig. 2 (a) and 2 (b) that the current reacts very rapidly when a state switch occurs, and the entire process of the state switch is usually completed only within 1 time point. The sensors with this characteristic are four types of sensors, namely current, voltage, pump and fan.
It can be seen from fig. 3 (a) and 3 (b) that the electric power is in a gradual state when the state switching occurs, and the entire switching process is usually completed at around 7000 time points. The sensors with this characteristic are three types of sensors, namely electric power, reactor thermal power and pressure stabilizer pressure.
As can be seen from fig. 4 (a) and 4 (b), the electric power is in a slowly changing state when the state switching occurs, and one state switching lasts for a very long time, typically about 13000 time points. The sensors with this characteristic are three types of sensors, namely electric power, reactor thermal power and pressure stabilizer pressure.
3: the RCV234MV sensor, which was targeted at 2019-10-01T00:00 to 2019-12-30T23:59:59, was selected as the target sensor, and a multiscale time window was used to begin the search for the state switching time point. As shown in table 2 below, the state switching time points are found.
Table 2 status switch time point
| Sequence number | Time point | Sequence | Time point | |
| 1 | 2019-10-03T11:31:39.00 | 7 | 2019-11-23T10:37:51.00 | |
| 2 | 2019-10-29T11:22:42.00 | 8 | 2019-11-23T10:42:56.00 | |
| 3 | 2019-11-02T13:45:35.00 | 9 | 2019-12-23T10:50:18.00 | |
| 4 | 2019-11-02T13:54:07.00 | 10 | 2019-12-27T21:30:28.00 | |
| 5 | 2019-11-03T05:11:15.00 | 11 | 2019-12-28T06:03:17.00 | |
| 6 | 2019-11-03T05:14:37.00 |
4: in the other 28 sensor data, the searched time point is taken as an original point, detection is started with the corresponding sensor time window, and the state switching times searched by each sensor are recorded.
5.: and calculating the matching rate of each sensor and the target sensor. The calculation formula is as follows:
cor＝m/n*100％ (4)
wherein: n is the number of times of state switching of the target sensor, and m is the number of times of state switching of the detected sensor.
Table 3 degree of matching of each sensor
The method can directly judge 16 sensors with strong relevance, the RCV002MI is judged to be the strong relevance sensor through manual rechecking, the number of the searched strong relevance sensors is 17, and the total number of the sensors with strong relevance in the data set is 18 including the RCV234MV through nuclear power operation and maintenance personnel inspection, which is just the sensor searched by the algorithm of the invention, and the accuracy is 100%.
The method is proved to be effective and reasonable by testing the historical operation data of the nuclear power plant and combining the working experience of operation and maintenance personnel.
Claims (7)
1. The nuclear power operation data correlation analysis method based on the multi-scale time window is characterized by comprising the following steps of:
step 1: preprocessing original nuclear power operation data;
step 2: determining a state switching standard, dividing the sensors into different categories according to the characteristics of sensor data of different categories, determining corresponding time windows according to the sensor divisions of the different categories, and judging whether state switching occurs or not;
step 3: selecting a target sensor, searching a time point when state switching occurs by utilizing a multi-scale time window;
step 4: searching whether state switching occurs in the corresponding time neighborhood in each sensor data by utilizing the time window determined in the step 2;
step 5: and calculating the matching rate of each sensor and the target sensor, and determining the relevance of each sensor and the target sensor.
2. The nuclear power operation data correlation analysis method based on the multi-scale time window according to claim 1, wherein the method comprises the following steps of: in the step 2, according to the change characteristics of the sensor data, the time required for the state switching of the sensors of different types is different, and the sensors are classified into three main categories: instantaneous, gradual change and slow change, and making time windows respectively corresponding to different scales, and judging whether the state switching standard occurs.
3. The nuclear power operation data correlation analysis method based on the multi-scale time window according to claim 1, wherein the method comprises the following steps of: in the step 2, the state switching criteria are defined as follows: when the amplitude of the data in a certain window is larger than or equal to a certain proportion of the whole amplitude of the data, the state switching is confirmed to occur in the time window, and the basis for judging whether the state switching occurs is as follows:
Data * ＝Data max -Data min (1)；
data * ＝data max -data min (2)；
data * ＞＝K*Data * (3)；
wherein: data max Data is the maximum value of a certain sensor in the whole time range min Data, which is the minimum value over the entire sensor time range * Maximum variation of data amplitude over time; data max Data is the maximum value in the time window of a certain sensor min Data, which is the minimum value in the time window of a certain sensor * K is a scaling factor for the maximum amplitude variation within the time window.
4. The nuclear power operation data correlation analysis method based on a multi-scale time window according to claim 3, wherein the method comprises the following steps of: the step 2 comprises the following steps:
s2.1, counting the amplitude variation of each sensor data, and counting the time required by each sensor for one-time state switching;
s2.2, dividing all sensors into three types of instantaneous variable, gradual variable and slow variable according to the time length required by the data recorded by different sensors for one-time state switching;
s2.3, counting the amplitude change belonging to state switching in the data of each sensor, and calculating the proportion of the amplitude of each sensor to the difference between the maximum value and the minimum value in the whole data;
the proportion of the state switching amplitude refers to the proportion of the amplitude change of each state switching to the whole data amplitude change, and the calculation mode is shown in the formula (4):
t＝mag/Data* (4)
wherein: mag is the amplitude variation during state switching, and Data is the maximum variation of the Data amplitude in the whole time range; the proportion of the amplitude change of each state switch is calculated according to the formula,
s2.4, aiming at sensors of different types, formulating a time window with corresponding length so as to facilitate the subsequent detection of state switching.
5. The nuclear power operation data correlation analysis method based on the multi-scale time window according to claim 1, wherein the method comprises the following steps of: the step 3 comprises the following steps:
s3.1, selecting an object sensor: selecting the most intuitive and most obvious quantity of change as a target sensor, such as voltage, current and the like;
s3.2, performing state switching detection on the target sensor data by utilizing a sliding window, and reserving the window with detected state switching, wherein undetected window can be discarded;
s3.3, further detecting the area with the detected state switching by utilizing a sliding window with smaller scale, still reserving the area with the detected state switching, and discarding the area without the detected state switching;
s3.4, repeating the step S3.3 until an accurate state switching time point can be obtained, and reserving one time point for the repeated time point.
6. The nuclear power operation data correlation analysis method based on the multi-scale time window according to claim 1, wherein the method comprises the following steps of: the step 4 comprises the following steps:
s4.1, marking the time points searched in the step 3 in the data of each sensor;
s4.2, detecting the switching state of the neighborhood of each time point by taking each time point as the center and utilizing the time window of the corresponding category formulated in the step 2;
and S4.3, recording the occurrence times of state switching detected by each sensor data.
7. The nuclear power operation data correlation analysis method based on the multi-scale time window according to claim 1, wherein the method comprises the following steps of: said step 5 comprises the steps of:
s5.1, calculating the matching rate of each sensor and the target sensor, wherein the calculation formula is as follows:
cor＝m/n*100％(4)
wherein: n is the number of times of state switching of the target sensor, and m is the number of times of state switching of the detected sensor;
s5.2, sorting the matching rate of each sensor from large to small;
s5.3, the high-matching-rate sensor is identified with strong relevance, the medium-matching-rate sensor is checked manually, and the low-matching-rate identification is identified with low relevance or no relevance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211572537.6A CN116257764A (en) | 2022-12-08 | 2022-12-08 | Nuclear power operation data correlation analysis method based on multi-scale time window |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211572537.6A CN116257764A (en) | 2022-12-08 | 2022-12-08 | Nuclear power operation data correlation analysis method based on multi-scale time window |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116257764A true CN116257764A (en) | 2023-06-13 |
Family
ID=86681669
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211572537.6A Withdrawn CN116257764A (en) | 2022-12-08 | 2022-12-08 | Nuclear power operation data correlation analysis method based on multi-scale time window |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116257764A (en) |
-
2022
- 2022-12-08 CN CN202211572537.6A patent/CN116257764A/en not_active Withdrawn
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11586913B2 (en) | Power equipment fault detecting and positioning method of artificial intelligence inference fusion | |
| CN108447057B (en) | SAR image change detection method based on significance and depth convolution network | |
| CN113838054B (en) | Mechanical part surface damage detection method based on artificial intelligence | |
| CN101788378A (en) | Mechanical failure diagnostic method and device | |
| CN111275570A (en) | Wind turbine generator set power abnormal value detection method based on iterative statistics and hypothesis test | |
| CN116167010B (en) | Rapid identification method for abnormal events of power system with intelligent transfer learning capability | |
| CN112380992B (en) | Method and device for evaluating and optimizing accuracy of monitoring data in machining process | |
| CN108956111A (en) | A kind of the abnormal state detection method and detection system of mechanical part | |
| CN114490156A (en) | Time series data abnormity marking method | |
| CN113342610A (en) | Time sequence data anomaly detection method and device, electronic equipment and storage medium | |
| CN111898637A (en) | Feature selection algorithm based on Relieff-DDC | |
| CN113515450A (en) | Environment anomaly detection method and system | |
| CN113112067A (en) | Method for establishing TFRI weight calculation model | |
| CN111091194B (en) | Operation system identification method based on CAVWBB _ KL algorithm | |
| CN116257764A (en) | Nuclear power operation data correlation analysis method based on multi-scale time window | |
| CN117330906A (en) | Equipment arc fault detection method, device, equipment and storage medium | |
| CN116953435A (en) | Cable instantaneous discharge anomaly identification method based on wave recording file | |
| JP5640796B2 (en) | Name identification support processing apparatus, method and program | |
| CN114819260A (en) | Dynamic generation method of hydrologic time series prediction model | |
| CN112526558B (en) | System operation condition identification and cutting method under partial data loss condition | |
| Krämer et al. | Dimension-scalable recurrence threshold estimation | |
| CN114598627A (en) | Abnormal network information detection method based on knowledge graph | |
| CN110990383A (en) | Similarity calculation method based on industrial big data set | |
| JPH0743352A (en) | Method and device for automatic construction of learning type abnormality diagnosis algorithm | |
| CN111460005B (en) | JSD-based outlier detection method for time sequence data |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20230613 |