Traceless Kalman filter threshold estimation for distributed power loads
Load forecasting is an important part of smart grid construction, energy management, and sustainable design of power systems, and has a great impact on the reliable operation of power grids, facility planning and other decisions. In this paper, we utilize the traceless transform of the UKF algorithm...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Sciendo
2024-01-01
|
| Series: | Applied Mathematics and Nonlinear Sciences |
| Subjects: | |
| Online Access: | https://doi.org/10.2478/amns.2023.2.01681 |
| _version_ | 1797340493539966976 |
|---|---|
| author | Bao Chengjia Zhang Tianyi Hu Zhixi |
| author_facet | Bao Chengjia Zhang Tianyi Hu Zhixi |
| author_sort | Bao Chengjia |
| collection | DOAJ |
| description | Load forecasting is an important part of smart grid construction, energy management, and sustainable design of power systems, and has a great impact on the reliable operation of power grids, facility planning and other decisions. In this paper, we utilize the traceless transform of the UKF algorithm to obtain the sigma feature points generated by the mean value of the high-frequency wavelet components of the power signals and combine them with the EKF algorithm. We put forward a novel neural network hybrid Kalman TUKF algorithm, which will be used to carry out simulation experiments on the distributed electric loads and to estimate the threshold value of the loads in the numerical experiments. The results show that in comparison with the actual measurements, the TUKF algorithm improves by 34.7% in the RMSE metrics, 38.7% in the MAE metrics, and 40.6% in the MAPE metrics compared to the PFWNN. The TUKF algorithm is closer to the real curves and has the best prediction performance for all the time intervals of the prediction. The change of the threshold value has a more obvious effect on the prediction accuracy, and the best effect is in the scale δ = 0.5, i.e., the threshold frequency is selected as the middle value of the intermediate frequency. |
| first_indexed | 2024-03-08T10:04:16Z |
| format | Article |
| id | doaj.art-8e34237091a74394bdc2e6b84de207ab |
| institution | Directory Open Access Journal |
| issn | 2444-8656 |
| language | English |
| last_indexed | 2024-03-08T10:04:16Z |
| publishDate | 2024-01-01 |
| publisher | Sciendo |
| record_format | Article |
| series | Applied Mathematics and Nonlinear Sciences |
| spelling | doaj.art-8e34237091a74394bdc2e6b84de207ab2024-01-29T08:52:45ZengSciendoApplied Mathematics and Nonlinear Sciences2444-86562024-01-019110.2478/amns.2023.2.01681Traceless Kalman filter threshold estimation for distributed power loadsBao Chengjia0Zhang Tianyi1Hu Zhixi21State Grid Gansu Electric Power Company, Lanzhou, Gansu, 730030, China.2State Grid Lanzhou Electric Power Supply Company, Lanzhou, Gansu, 730070, China.2State Grid Lanzhou Electric Power Supply Company, Lanzhou, Gansu, 730070, China.Load forecasting is an important part of smart grid construction, energy management, and sustainable design of power systems, and has a great impact on the reliable operation of power grids, facility planning and other decisions. In this paper, we utilize the traceless transform of the UKF algorithm to obtain the sigma feature points generated by the mean value of the high-frequency wavelet components of the power signals and combine them with the EKF algorithm. We put forward a novel neural network hybrid Kalman TUKF algorithm, which will be used to carry out simulation experiments on the distributed electric loads and to estimate the threshold value of the loads in the numerical experiments. The results show that in comparison with the actual measurements, the TUKF algorithm improves by 34.7% in the RMSE metrics, 38.7% in the MAE metrics, and 40.6% in the MAPE metrics compared to the PFWNN. The TUKF algorithm is closer to the real curves and has the best prediction performance for all the time intervals of the prediction. The change of the threshold value has a more obvious effect on the prediction accuracy, and the best effect is in the scale δ = 0.5, i.e., the threshold frequency is selected as the middle value of the intermediate frequency.https://doi.org/10.2478/amns.2023.2.01681power loadukf algorithmwavelet neural networkthresholdtukf algorithm11l07 |
| spellingShingle | Bao Chengjia Zhang Tianyi Hu Zhixi Traceless Kalman filter threshold estimation for distributed power loads Applied Mathematics and Nonlinear Sciences power load ukf algorithm wavelet neural network threshold tukf algorithm 11l07 |
| title | Traceless Kalman filter threshold estimation for distributed power loads |
| title_full | Traceless Kalman filter threshold estimation for distributed power loads |
| title_fullStr | Traceless Kalman filter threshold estimation for distributed power loads |
| title_full_unstemmed | Traceless Kalman filter threshold estimation for distributed power loads |
| title_short | Traceless Kalman filter threshold estimation for distributed power loads |
| title_sort | traceless kalman filter threshold estimation for distributed power loads |
| topic | power load ukf algorithm wavelet neural network threshold tukf algorithm 11l07 |
| url | https://doi.org/10.2478/amns.2023.2.01681 |
| work_keys_str_mv | AT baochengjia tracelesskalmanfilterthresholdestimationfordistributedpowerloads AT zhangtianyi tracelesskalmanfilterthresholdestimationfordistributedpowerloads AT huzhixi tracelesskalmanfilterthresholdestimationfordistributedpowerloads |