The interpretation of the cone penetration test(CPT)still relies largely on empirical correlations that have been predominantly developed in resource-intensive and time-consuming calibration chambers.This paper presen...The interpretation of the cone penetration test(CPT)still relies largely on empirical correlations that have been predominantly developed in resource-intensive and time-consuming calibration chambers.This paper presents a CPT virtual calibration chamber using deep learning(DL)approaches,which allow for the consideration of depth-dependent cone resistance profiles through the implementation of two proposed strategies:(1)depth-resistance mapping using a multilayer perceptron(MLP)and(2)sequence-to-sequence training using a long short-term memory(LSTM)neural network.Two DL models are developed to predict cone resistance profiles(qc)under various soil states and testing conditions,where Bayesian optimization(BO)is adopted to identify the optimal hyperparameters.Subsequently,the BO-MLP and BO-LSTM networks are trained using the available data from published datasets.The results show that the models with BO can effectively improve the prediction accuracy and efficiency of neural networks compared to those without BO.The two training strategies yielded comparable results in the testing set,and both can be used to reproduce the whole cone resistance profile.An extended comparison and validation of the prediction results are carried out against numerical results obtained from a coupled Eulerian-Lagrangian(CEL)model,demonstrating a high degree of agreement between the DL and CEL models.Ultimately,to demonstrate the usability of this new virtual calibration chamber,the predicted qc is used to enhance the preceding correlations with the relative density(Dr)of the sand.The improved correlation with superior generalization has an R^(2) of 82%when considering all data,and 89.6%when examining the pure experimental data.展开更多
The virtual in-situ calibration method has been effective in calibrating multiple sensors in HVAC systems when the fault types are known.However,due to the high cost of physical calibration and the strict data accurac...The virtual in-situ calibration method has been effective in calibrating multiple sensors in HVAC systems when the fault types are known.However,due to the high cost of physical calibration and the strict data accuracy requirements of data-driven methods,obtaining benchmarks for sensors during actual operation is challenging,making it difficult to diagnose the specific fault types of the sensors.To address this issue,an enhanced multi-sensor calibration(EMC)method has been developed to operate without prior knowledge of fault types.The primary soft faults encountered—bias and drift deviations—differ in whether they vary over time.The proposed method employs an interval sliding approach to identify and calibrate these faults within each interval effectively.Furthermore,the influence of interval size on calibration accuracy has been systematically analyzed to optimize performance.The proposed method has been validated on a chiller plant in a large public building in Hong Kong.The experimental results indicate that,under conditions involving eight sensor faults,including even three drift deviations,the EMC method achieved average calibration accurate rates of 100%for bias faults and 95%for drift faults.Notably,in calibrating drift faults,the enhanced method outperformed the high-dimensional sensor calibration method and the Improved simulated annealing method by 87%and 34%,respectively.展开更多
基金support from the National Natural Science Foundation of China(Grant No.52408356)the China Scholarship Council(CSC).
文摘The interpretation of the cone penetration test(CPT)still relies largely on empirical correlations that have been predominantly developed in resource-intensive and time-consuming calibration chambers.This paper presents a CPT virtual calibration chamber using deep learning(DL)approaches,which allow for the consideration of depth-dependent cone resistance profiles through the implementation of two proposed strategies:(1)depth-resistance mapping using a multilayer perceptron(MLP)and(2)sequence-to-sequence training using a long short-term memory(LSTM)neural network.Two DL models are developed to predict cone resistance profiles(qc)under various soil states and testing conditions,where Bayesian optimization(BO)is adopted to identify the optimal hyperparameters.Subsequently,the BO-MLP and BO-LSTM networks are trained using the available data from published datasets.The results show that the models with BO can effectively improve the prediction accuracy and efficiency of neural networks compared to those without BO.The two training strategies yielded comparable results in the testing set,and both can be used to reproduce the whole cone resistance profile.An extended comparison and validation of the prediction results are carried out against numerical results obtained from a coupled Eulerian-Lagrangian(CEL)model,demonstrating a high degree of agreement between the DL and CEL models.Ultimately,to demonstrate the usability of this new virtual calibration chamber,the predicted qc is used to enhance the preceding correlations with the relative density(Dr)of the sand.The improved correlation with superior generalization has an R^(2) of 82%when considering all data,and 89.6%when examining the pure experimental data.
基金the Natural Science Foundation of Jiangsu Province(BK20240560)the China Postdoctoral Science Foundation(2024M764219)+1 种基金the Science and Technology Project of Jiangsu Provincial Department of Housing and Urban Rural Development(2023ZD026)the Science and Technology Project of Nanjing Municipal Commission of Urban and Rural Development(Ks2415).
文摘The virtual in-situ calibration method has been effective in calibrating multiple sensors in HVAC systems when the fault types are known.However,due to the high cost of physical calibration and the strict data accuracy requirements of data-driven methods,obtaining benchmarks for sensors during actual operation is challenging,making it difficult to diagnose the specific fault types of the sensors.To address this issue,an enhanced multi-sensor calibration(EMC)method has been developed to operate without prior knowledge of fault types.The primary soft faults encountered—bias and drift deviations—differ in whether they vary over time.The proposed method employs an interval sliding approach to identify and calibrate these faults within each interval effectively.Furthermore,the influence of interval size on calibration accuracy has been systematically analyzed to optimize performance.The proposed method has been validated on a chiller plant in a large public building in Hong Kong.The experimental results indicate that,under conditions involving eight sensor faults,including even three drift deviations,the EMC method achieved average calibration accurate rates of 100%for bias faults and 95%for drift faults.Notably,in calibrating drift faults,the enhanced method outperformed the high-dimensional sensor calibration method and the Improved simulated annealing method by 87%and 34%,respectively.
