Wyniki wyszukiwania - BazTech

Ograniczanie wyników

Znaleziono wyników: 2

Liczba wyników na stronie

Wyniki wyszukiwania

Sortuj według:

Ogranicz wyniki do:

Kernel extreme learning machine combined with gray wolf optimization for temperature compensation in pressure sensors

Wang Huan, Wu Ting, Liu Pan, Zou Yijun, Zeng Qinghua

Metrology and Measurement Systems

2025

Vol. 32, nr 1

art. no. 4, 1--14

In order to improve the measurement accuracy of pressure sensors, a method based on gray wolf optimization (GWO) to optimize kernel extreme learning machine (KELM) is proposed to address the problem of nonlinear drift that can be easily affected by temperature in the working environment. Firstly, the fast search capability of the GWO algorithm is used to find optimal regularization coefficients and kernel function parameters of the KELM algorithm; secondly, the random mapping of the traditional ELM algorithm is replaced by the kernel mapping of the KELM algorithm to improve the generalization and stability degradation brought by the random assignments. Finally, the voltage signal values under different temperature and pressure environments are obtained through calibration experiments and compensated by the GWO-KELM algorithm. The results show that the GWO-KELM method has a better compensation effect compared with the traditional BP neural network with a full-scale error of 0.13% (FS), the ELM algorithm with a full-scale error of 0.12%FS, and the KELM algorithm with a full-scale error of 0.12% in the range of 0 to 700 kPa absolute pressure and -40˚ to 70˚. The full-scale error is only 0.07% and the maximum absolute error is as low as 0.5446 kPa, which improves the accuracy index by one order of magnitude.

Study on groundwater table distribution of loess tableland in the South Jingyang tableland, Northwest China

Wu Jiaqi, Dai Fuchu, Liu Pan, Huang Zhiquan, Wu Daixuan

Acta Geophysica

2024

Vol. 72, no. 1

257--272

The South Jingyang tableland, located in Shaanxi Province, China, is composed of loess deposits. Agricultural irrigation in this area has caused the continuous rise of the groundwater table, which triggered a series of loess landslides. Although extensive research has been conducted on the effect of rising groundwater tables on induced loess landslides, less is known about the groundwater table distribution. This study jointly investigated the distribution of the groundwater table in the South Jingyang tableland using boreholes, electrical resistivity tomography (ERT), and laboratory tests. A qualitative analysis between the apparent resistivity and the water content was estimated at different depths, using the borehole data and ERT inversion data. A quantitative relationship between the resistivity and the water content was established in the laboratory tests. A combination of the qualitative analysis of the inversion results and the quantitative relationship was used to determine the distribution of the groundwater table. The groundwater table is present at the depth where the water content of the loess is 30% (liquid limit of loess) and the resistivity is 28.35 Q m. The groundwater table elevation in the southern study area is higher than that of the northern study area and decreases in a stepwise gradient from south to north. This study not only identified the distribution of the groundwater table in the South Jingyang tableland, but also makes up for the limitation and inaccuracy in determining groundwater tables by relying solely on the borehole information or inversion results of ERT. The ascertained distribution characteristics of the groundwater table provides a basis for the analysis of the formation mechanism of landslides.