Virtual reality research of the dynamic characteristics of soft soil under metro vibration loads based on BP neural networks
Autor: | Xiaotong Qin, Kai Cui |
---|---|
Rok vydání: | 2017 |
Předmět: |
0209 industrial biotechnology
Artificial neural network business.industry Computer science 02 engineering and technology Structural engineering Virtual reality Finite element method Vibration Dynamic simulation Pore water pressure 020901 industrial engineering & automation Artificial Intelligence Simple (abstract algebra) 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing business Software Simulation |
Zdroj: | Neural Computing and Applications. 29:1233-1242 |
ISSN: | 1433-3058 0941-0643 |
DOI: | 10.1007/s00521-017-2853-7 |
Popis: | In this paper, in order to research the dynamic characteristics of soft soil under metro vibration loads, the mathematical expression of metro vibration loads is obtained. According to the loading form, drainage requirement and vibration frequencies of the actual situation, the corresponding experiment is conducted through indoor dynamic triaxial equipment. Then, the dynamic characteristics of experimental results are analyzed. An empirical formula is proposed to compute the dynamic characteristics of soft soil. Then, the computational results obtained by empirical formula are compared with those of the experimental. They are consistent with each other, and the results show that empirical formula is reliable to compute the dynamic characteristics of soft soil. Then, based on the verified empirical formula, the dynamic characteristics such as the vertical strain and pore water pressure with different CSR are also computed and compared. With the increase in CSR, the dynamic characteristics will be larger when the other parameters are consistent. However, empirical formula can only predict the dynamic characteristics of the simple model. In order to realize virtual reality of the dynamic characteristics of the complex model more accurately, the BP neural network and finite element are adopted, respectively. Then, the computational results are also compared with those of the experimental to verify their reliabilities. In the future, the BP neural network and finite element method can be also used to realize virtual reality of the dynamic characteristics of the more complex model. |
Databáze: | OpenAIRE |
Externí odkaz: |