In this paper,we summarise the outcome of a set of experiments aimed at classifying cattle behaviour based on sensor data.Each animal carried sensors generating time series accelerometer data placed on a collar on the...In this paper,we summarise the outcome of a set of experiments aimed at classifying cattle behaviour based on sensor data.Each animal carried sensors generating time series accelerometer data placed on a collar on the neck at the back of the head,on a halter positioned at the side of the head behind the mouth,or on the ear using a tag.The purpose of the study was to determine how sensor data from different placement can classify a range of typical cattle behaviours.Data were collected and animal behaviours(grazing,standing or ruminating)were observed over a common time frame.Statistical features were computed from the sensor data and machine learning algorithms were trained to classify each behaviour.Classification accuracies were computed on separate independent test sets.The analysis based on behaviour classification experiments revealed that different sensor placement can achieve good classification accuracy if the feature space(representing motion patterns)between the training and test animal is similar.The paper will discuss these analyses in detail and can act as a guide for future studies.展开更多
Fine-grained silt is widely distributed in the Huanghe River Delta(HRD)in China,and the sedimentary structure is complex,meaning that the clay content in the silt is variable.The piezocone penetration test(CPTu)is the...Fine-grained silt is widely distributed in the Huanghe River Delta(HRD)in China,and the sedimentary structure is complex,meaning that the clay content in the silt is variable.The piezocone penetration test(CPTu)is the most widely approved in situ test method.It can be used to invert soil properties and interpret soil behavior.To analyse the strength properties of surface sediments in the HRD,this paper evaluated the friction angle and its inversion formula through the CPTu penetration test and monotonic simple shear test and other soil unit experiments.The evaluation showed that the empirical formula proposed by Kulhawy and Mayne had better prediction and inversion effect.The HRD silts with clay contents of 9.2%,21.4%and 30.3%were selected as samples for the CPTu variable rate penetration test.The results show as follows.(1)The effects of the clay content on the tip resistance and the pore pressure of silt under different penetration rates were summarized.The tip resistance Q_t is strongly dependent on the clay content of the silt,the B_(q)value of the silt tends to 0 and is not significantly affected by the change of the CPTu penetration rate.(2)Five soil behavior type classification charts and three soil behavior type indexes based on CPTu data were evaluated.The results show that the soil behavior type classification chart based on soil behavior type index ISBT,the Robertson 2010 behavior type classification chart are more suitable for the silty soil in the HRD.展开更多
Classification of sheep behaviour from a sequence of tri-axial accelerometer data has the potential to enhance sheep management.Sheep behaviour is inherently imbalanced(e.g.,more ruminating than walking)resulting in u...Classification of sheep behaviour from a sequence of tri-axial accelerometer data has the potential to enhance sheep management.Sheep behaviour is inherently imbalanced(e.g.,more ruminating than walking)resulting in underperforming classification for the minority activities which hold importance.Existing works have not addressed class imbalance and use traditional machine learning techniques,e.g.,Random Forest(RF).We investigated Deep Learning(DL)models,namely,Long Short Term Memory(LSTM)and Bidirectional LSTM(BLSTM),appropriate for sequential data,from imbalanced data.Two data sets were collected in normal grazing conditions using jaw-mounted and earmounted sensors.Novel to this study,alongside typical single classes,e.g.,walking,depending on the behaviours,data samples were labelled with compound classes,e.g.,walking_-grazing.The number of steps a sheep performed in the observed 10 s time window was also recorded and incorporated in the models.We designed several multi-class classification studies with imbalance being addressed using synthetic data.DL models achieved superior performance to traditional ML models,especially with augmented data(e.g.,4-Class+Steps:LSTM 88.0%,RF 82.5%).DL methods showed superior generalisability on unseen sheep(i.e.,F1-score:BLSTM 0.84,LSTM 0.83,RF 0.65).LSTM,BLSTM and RF achieved sub-millisecond average inference time,making them suitable for real-time applications.The results demonstrate the effectiveness of DL models for sheep behaviour classification in grazing conditions.The results also demonstrate the DL techniques can generalise across different sheep.The study presents a strong foundation of the development of such models for real-time animal monitoring.展开更多
文摘In this paper,we summarise the outcome of a set of experiments aimed at classifying cattle behaviour based on sensor data.Each animal carried sensors generating time series accelerometer data placed on a collar on the neck at the back of the head,on a halter positioned at the side of the head behind the mouth,or on the ear using a tag.The purpose of the study was to determine how sensor data from different placement can classify a range of typical cattle behaviours.Data were collected and animal behaviours(grazing,standing or ruminating)were observed over a common time frame.Statistical features were computed from the sensor data and machine learning algorithms were trained to classify each behaviour.Classification accuracies were computed on separate independent test sets.The analysis based on behaviour classification experiments revealed that different sensor placement can achieve good classification accuracy if the feature space(representing motion patterns)between the training and test animal is similar.The paper will discuss these analyses in detail and can act as a guide for future studies.
基金The National Natural Science Foundation of China under contract No.U2006213。
文摘Fine-grained silt is widely distributed in the Huanghe River Delta(HRD)in China,and the sedimentary structure is complex,meaning that the clay content in the silt is variable.The piezocone penetration test(CPTu)is the most widely approved in situ test method.It can be used to invert soil properties and interpret soil behavior.To analyse the strength properties of surface sediments in the HRD,this paper evaluated the friction angle and its inversion formula through the CPTu penetration test and monotonic simple shear test and other soil unit experiments.The evaluation showed that the empirical formula proposed by Kulhawy and Mayne had better prediction and inversion effect.The HRD silts with clay contents of 9.2%,21.4%and 30.3%were selected as samples for the CPTu variable rate penetration test.The results show as follows.(1)The effects of the clay content on the tip resistance and the pore pressure of silt under different penetration rates were summarized.The tip resistance Q_t is strongly dependent on the clay content of the silt,the B_(q)value of the silt tends to 0 and is not significantly affected by the change of the CPTu penetration rate.(2)Five soil behavior type classification charts and three soil behavior type indexes based on CPTu data were evaluated.The results show that the soil behavior type classification chart based on soil behavior type index ISBT,the Robertson 2010 behavior type classification chart are more suitable for the silty soil in the HRD.
文摘Classification of sheep behaviour from a sequence of tri-axial accelerometer data has the potential to enhance sheep management.Sheep behaviour is inherently imbalanced(e.g.,more ruminating than walking)resulting in underperforming classification for the minority activities which hold importance.Existing works have not addressed class imbalance and use traditional machine learning techniques,e.g.,Random Forest(RF).We investigated Deep Learning(DL)models,namely,Long Short Term Memory(LSTM)and Bidirectional LSTM(BLSTM),appropriate for sequential data,from imbalanced data.Two data sets were collected in normal grazing conditions using jaw-mounted and earmounted sensors.Novel to this study,alongside typical single classes,e.g.,walking,depending on the behaviours,data samples were labelled with compound classes,e.g.,walking_-grazing.The number of steps a sheep performed in the observed 10 s time window was also recorded and incorporated in the models.We designed several multi-class classification studies with imbalance being addressed using synthetic data.DL models achieved superior performance to traditional ML models,especially with augmented data(e.g.,4-Class+Steps:LSTM 88.0%,RF 82.5%).DL methods showed superior generalisability on unseen sheep(i.e.,F1-score:BLSTM 0.84,LSTM 0.83,RF 0.65).LSTM,BLSTM and RF achieved sub-millisecond average inference time,making them suitable for real-time applications.The results demonstrate the effectiveness of DL models for sheep behaviour classification in grazing conditions.The results also demonstrate the DL techniques can generalise across different sheep.The study presents a strong foundation of the development of such models for real-time animal monitoring.
