The reuse of green sand in casting production is hindered by the accumulation of oolitic deposits,primarily composed of clay binder with surface degradation,which may adversely affect the the moulding sand performance...The reuse of green sand in casting production is hindered by the accumulation of oolitic deposits,primarily composed of clay binder with surface degradation,which may adversely affect the the moulding sand performance.Currently,there is a lack of standardized methods for quantifying the oolitic content.Accurate measurement of oolitic content is of great significance to the reuse of green sand.Attempts to determine oolitic content using potassium hydroxide(KOH)and phosphoric acid(H_(3)PO_(4))methods encounter challenges due to their excessive reactions with SiO_(2) in the sand.In this study,an improved method for measuring the oolitic content of green sand with repeated approximations was proposed.This method judges the chemical activity of the sample surface through the change of its mass to accurately obtain the mass of the reaction oolitic deposits.The test result of the used sand samples from the foundry shows that the oolitic deposits are completely removed after reacting with KOH solution three times at 300℃ for 20 min.SEM and EDS also show that after three times of reactions,the surface of green sand becomes smooth and the content of Al-containing oolitic deposits is very low.This indicates that the method can accurately control the extent of the reaction.Implementation of this method at Huangshi Dongbei Casting Co.,Ltd.has yielded consistent and reliable test results,effectively mirroring variations in green sand oolitic content on the production line.This new method is expected to be widely adopted to improve the efficiency and quality of reused green sand in casting operations.展开更多
现有的电离层总电子含量(Total Electron Content,TEC)时空预测模型主要以堆叠ConvLSTM单元及其变体为主.这种依赖于ConvLSTM的TEC时空预测模型在捕捉局部时空依赖性的时候比较有效.但由于缺乏存储长距离空间记忆的单元,致使长距离的TE...现有的电离层总电子含量(Total Electron Content,TEC)时空预测模型主要以堆叠ConvLSTM单元及其变体为主.这种依赖于ConvLSTM的TEC时空预测模型在捕捉局部时空依赖性的时候比较有效.但由于缺乏存储长距离空间记忆的单元,致使长距离的TEC空间特征依赖难以被ConvLSTM及其变体捕捉.为解决该问题,本文提出了一个基于自注意力记忆卷积长短期记忆网络的电离层TEC时空预测模型SA-ConvLSTM,该模型在具有短期记忆依赖的ConvLSTM基础上,增加了具有长距离记忆依赖的自注意力记忆(self-attention memory,SAM)模块,以便在TEC时空预测中同时兼顾短期记忆和长距离记忆.为了验证SA-ConvLSTM的性能,本文在12.5°S—87.5°N,25°E—180°E区域内选择3年太阳活动高年和3年太阳活动低年的TEC网格数据,在该数据上,将SA-ConvLSTM与目前主流的TEC时空预测模型ConvGRU、ConvLSTM、PredRNN、Residual Attention-BiConvLSTM及CODE提供的电离层预测产品C1PG进行了对比.结果表明,与C1PG、ConvGRU、ConvLSTM、PredRNN和Residual Attention-BiConvLSTM相比,SA-ConvLSTM的RMSE在太阳活动高年分别降低了6.58%、3.89%、5.79%、1.44%、1.21%;在太阳活动低年分别降低了13.42%、10.26%、11.40%、3.20%、4.37%.此外,本文还在不同月份和纬度区域情况下进行了对比,结果表明,在绝大多数月份和绝大多数纬度区域内,SA-ConvLSTM的预测性能更好.最后本文选取了两次磁暴事件来验证SA-ConvLSTM在极端情况下的预测能力.结果表明,SA-ConvLSTM在磁暴的大多数阶段均优于对比模型.展开更多
针对农业场景中草莓因枝叶遮挡、簇生分布及果面反光导致稀疏观测下难以三维重建的问题,本研究基于SA3D(Segment Anything in 3D)框架,在实验室条件下验证了自动化构建高保真草莓三维模型库的可行性。该方法融合DVGO与SAM,利用DVGO从14...针对农业场景中草莓因枝叶遮挡、簇生分布及果面反光导致稀疏观测下难以三维重建的问题,本研究基于SA3D(Segment Anything in 3D)框架,在实验室条件下验证了自动化构建高保真草莓三维模型库的可行性。该方法融合DVGO与SAM,利用DVGO从144张多视角图像中重建保留种子、果蒂等亚毫米细节的三维几何;结合SAM仅需1~2个提示点生成2D掩码,通过Mask逆渲染与跨视角自提示机制实现无标注的三维果实分割。为提升实用性,开发了基于Dash的交互式系统,集成图像上传、位姿估计、重建与分割全流程,支持非专业用户高效建模。实验表明,该方法平均PSNR达20.83 dB(较NeRF提升1.12 dB),IoU均值为0.803,显著增强遮挡与反光区域的重建鲁棒性。所构建的标准化点云库可为表型测量提供基准,并作为几何与语义先验支撑田间稀疏视角重建,服务于智能采摘系统的视觉感知。展开更多
State constraints in nonlinear systems are commonly pursued by resorting to barrier functions,which enforce constraints over the entire duration of system operation.We propose a universal intermittent state-constraine...State constraints in nonlinear systems are commonly pursued by resorting to barrier functions,which enforce constraints over the entire duration of system operation.We propose a universal intermittent state-constrained solution,which not only offers flexibility by activating constraints just during specific time periods of interest to the user,but also successfully accommodates different types of constraint boundaries.The innovative shifting functions are proposed to facilitate seamless transitions between constrained and unconstrained operational phases,resulting in more user-friendly design and implementation.By blending an improved shifting transformation into intermittent constraint design,we construct a universal barrier function upon the constrained states,with which our control strategy removes the limitations on constraint functions and completely obviates the feasibility conditions.Furthermore,a modified fuzzy approximator driven by the prediction error rather than the tracking error achieves decoupling of the control and estimation loops,which not only ensures the estimation performance,but also facilitates proof of stability.Finally,the effectiveness of the proposed scheme is assessed by numerical simulation.展开更多
基金financially supported by the National Key Research and Development Program of China(Grant No.2022YFB3706800)the National Natural Science Foundation of China(Grant Nos.51905188 and 51775205).
文摘The reuse of green sand in casting production is hindered by the accumulation of oolitic deposits,primarily composed of clay binder with surface degradation,which may adversely affect the the moulding sand performance.Currently,there is a lack of standardized methods for quantifying the oolitic content.Accurate measurement of oolitic content is of great significance to the reuse of green sand.Attempts to determine oolitic content using potassium hydroxide(KOH)and phosphoric acid(H_(3)PO_(4))methods encounter challenges due to their excessive reactions with SiO_(2) in the sand.In this study,an improved method for measuring the oolitic content of green sand with repeated approximations was proposed.This method judges the chemical activity of the sample surface through the change of its mass to accurately obtain the mass of the reaction oolitic deposits.The test result of the used sand samples from the foundry shows that the oolitic deposits are completely removed after reacting with KOH solution three times at 300℃ for 20 min.SEM and EDS also show that after three times of reactions,the surface of green sand becomes smooth and the content of Al-containing oolitic deposits is very low.This indicates that the method can accurately control the extent of the reaction.Implementation of this method at Huangshi Dongbei Casting Co.,Ltd.has yielded consistent and reliable test results,effectively mirroring variations in green sand oolitic content on the production line.This new method is expected to be widely adopted to improve the efficiency and quality of reused green sand in casting operations.
基金supported by the Fundamental Research Funds for the Central Universities(N2404005)the National Key Research and Development Program of China(2018YFA0702200)+2 种基金Liaoning Revitalization Talents Program(XLYC1801005)the National Natural Science Foundation of China(U23B20118)the Nature Science Foundation of Liaoning Province of China(2022JH25/10100008)。
文摘State constraints in nonlinear systems are commonly pursued by resorting to barrier functions,which enforce constraints over the entire duration of system operation.We propose a universal intermittent state-constrained solution,which not only offers flexibility by activating constraints just during specific time periods of interest to the user,but also successfully accommodates different types of constraint boundaries.The innovative shifting functions are proposed to facilitate seamless transitions between constrained and unconstrained operational phases,resulting in more user-friendly design and implementation.By blending an improved shifting transformation into intermittent constraint design,we construct a universal barrier function upon the constrained states,with which our control strategy removes the limitations on constraint functions and completely obviates the feasibility conditions.Furthermore,a modified fuzzy approximator driven by the prediction error rather than the tracking error achieves decoupling of the control and estimation loops,which not only ensures the estimation performance,but also facilitates proof of stability.Finally,the effectiveness of the proposed scheme is assessed by numerical simulation.
