| Autor: |
Wenhao CUI, Sheng ZHENG, Xiongjie QIN, Shuguang ZENG |
| Jazyk: |
English<br />Chinese |
| Rok vydání: |
2023 |
| Předmět: |
|
| Zdroj: |
南方能源建设, Vol 10, Iss 2, Pp 143-150 (2023) |
| Druh dokumentu: |
article |
| ISSN: |
2095-8676 |
| DOI: |
10.16516/j.gedi.issn2095-8676.2023.02.019 |
| Popis: |
[Introduction] Nuclear power operation data is characterized by high dimension and large volume, and the complexity of the internal system of nuclear power plant makes it difficult to build a corresponding mechanism model. Therefore, it is very difficult to manually screen out relevant parameters from nuclear power data, and the introduction of non-relevant parameters will greatly affect the accuracy of the model. By means of improving the model accuracy, the purpose of accurate modeling can be reached. [Method] This paper proposed a correlation analysis method based on multi-scale time window. This method extracted state switch points for target parameters, classifies each sensor according to the characteristics of the data recorded by different sensors, and then designs detection windows for different kinds of sensors that meet their characteristics. The state switch detection was carried out in the corresponding time neighborhood of each sensor, and the correlation matching rate between each sensor and the target sensor was calculated to judge the correlation. [Result] Based on the actual historical operation data of nuclear power plant, the sensor parameters associated with the target sensor are selected successfully by the established correlation matching rate rule. [Conclusion] The experimental results show that the proposed method can screen out the correlation parameters more accurately. Compared with the commonly used Pearson correlation coefficient, the proposed method is more accurate. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
