Study On Intelligent Prediction Of Frozen Temperature Field And Back Analysis Of Thermal Conductivity Of Peat Soil

The northeast and northwest regions of China belong to seasonal frozen soil regions,and there are great engineering hidden dangers in frozen soil during engineering construction,so it is extremely important to study the characteristics of frozen soil.The freezing method is widely used in engineering construction and large-scale mine mining.The key to the freezing method construction is to ensure the stability and safety of the frozen wall.Therefore,it is of great significance to study the development and change of the temperature field during the freezing process.We selected peat soil from a certain area in Xinjiang to conduct a bottom-up indoor unidirectional freezing test.During the test,the temperature field,moisture field and displacement field during the freezing process are monitored by controlling the cold end temperature,initial moisture content and compactness.The variation laws of water content field,temperature field and displacement field in the freezing process are analyzed by three different factors.Based on the theoretical basis of BP neural network,the weights and thresholds of the neural network are constantly adjusted through artificial fish school algorithm in training.The model of temperature field change considering time factor is established.The temperature data based on time series are obtained through indoor freezing test,and the obtained data are simulated and predicted by BP neural network.Compared with the test value and the model prediction value,the overall error is small,and the prediction model can more accurately predict the temperature field change law.The freezing temperature field model is established through the ANSYS software.The size of other thermal parameters of the frozen soil and the variation range of the thermal conductivity of the frozen soil are obtained in advance through laboratory tests and engineering experience.On the basis of the back analysis of the thermal conductivity,the freezing model under different thermal conductivity is numerically simulated,and the simulated temperature field value is compared with the temperature data measured in the test,An optimal thermal conductivity is determined by using the approximation method.The obtained equivalent thermal conductivity is used for numerical simulation,and it is verified that the equivalent thermal conductivity is1.625 k J/(m?h?℃)by comparing with the measured value in the test.In this paper,the indoor freezing test is carried out to simulate the changes of temperature field during the freezing process,and the temperature field is predicted and analyzed through the neural network.By comparing with the actual value,the prediction model can predict the temperature field more accurately.The finite element software is used to back analyze the thermal conductivity to determine the optimal equivalent thermal conductivity.The simulated temperature field value is compared with the actual data to verify that the simulation effect is good,which provides a reference value for the research of temperature field in actual projects.Figure [39] Table [13] Reference [79]...