公开(公告)号：WO2019118460A1

公开(公告)日：2019-06-20

申请号：PCT/US2018/064949

申请日：2018-12-11

申请人： THE TEXAS A&M UNIVERSITY SYSTEM

发明人： SUN, Lijia ,  HU, Jiang ,  PORTER, Dana, O. ,  MAREK, Thomas, H. ,  HILLYER, Charles, C. ,  YANG, Yanxiang

IPC分类号： E02B13/00 ,  G01W1/10 ,  G05B13/02 ,  G06G7/66 ,  G06N3/06 ,  G06N5/04 ,  G06Q50/02

摘要： Disclosed are various embodiments for reinforcement learning-based irrigation control to maintain or increase a crop yield or reduce water use. A computing device may be configured to determine an optimal irrigation schedule for a crop planted in a field by applying reinforcement learning (RL), where, for a given state of a total soil moisture, the computing device performs an action, the action comprising waiting or irrigating crop. An immediate reward may be assigned to a state-action pair, the state-action pair comprising the given state of the total soil moisture and the action performed. The computing device may instruct an irrigation system to apply irrigation to at least one crop in accordance with the optimal irrigation schedule determined, where the optimal irrigation schedule includes an amount of water to be applied at a predetermined time.

公开(公告)号：WO2021007363A1

公开(公告)日：2021-01-14

申请号：PCT/US2020/041270

申请日：2020-07-08

申请人： THE TEXAS A&M UNIVERSITY SYSTEM

发明人： YANG, Yanxiang ,  KONG, Hongxin ,  HU, Jiang ,  PORTER, Dana, O. ,  MAREK, Thomas, H. ,  HEFLIN, Kevin, R.

IPC分类号： A01G25/16 ,  A01G1/00 ,  A01G25/00 ,  G05D7/00 ,  G06N3/02 ,  G06N3/08

摘要： Disclosed are various embodiments for deep reinforcement learning-based irrigation control to maintain or increase crop yield and/or other desired crop status, and/or reduce water use. One or more computing devices can be configured to determine an amount of water to be applied to at least one crop in at least one of a plurality of irrigation management zones through execution of a deep reinforcement learning routine. Further, the computing devices can determine a start time and an end time to be applied to the at least one of the plurality of irrigation management zones based at least in part on the amount of water determined by the deep reinforcement learning module. Finally, the computing devices can instruct an irrigation system to apply irrigation to the at least one of the plurality of irrigation management zones in accordance with the start time and the end time.

公开(公告)号：WO2019118453A1

公开(公告)日：2019-06-20

申请号：PCT/US2018/064942

申请日：2018-12-11

申请人： THE TEXAS A&M UNIVERSITY SYSTEM

发明人： YANG, Yanxiang ,  HU, Jiang ,  PORTER, Dana O. ,  MAREK, Thomas H. ,  HILLYER, Charles C. ,  SUN, Lijia

IPC分类号： G06N3/12 ,  H04W64/00 ,  H04W84/18

摘要： Disclosed are various embodiments for optimized sensor deployment and fault detection in the context of agricultural irrigation and similar applications. For instance, a computing device may execute a genetic algorithm (GA) routine to determine an optimal sensor deployment scheme such that a mean-time-to-failure (MTTF) for the system is maximized, thereby improving communication of sensor measurements. Moreover, in various embodiments, a centralized fault detection scheme may be employed and a soil moisture of a field can be determined by statistically inferring soil moistures at locations of faulty nodes using spatial and temporal correlations.