基于注意力机制的时空卷积数控机床热误差模型研究

为了提高数控机床热误差模型的精度与泛化性,提出了基于注意力机制的长短时记忆卷积神经网络(Long short term memory convolutional neural network based on attention mechanism, AM-CNN-LSTM)热误差模型。利用卷积神经网络提取高维数据空间状态特征的能力和长短时记忆网络提取长时间序列状态特征的能力,构建具有2个支路的热误差模型,分别提取特征后输入到注意力机制中进行特征重要性重构,建立原始数据与热误差的特征映射,最后通过全连接层进行热误差预测。采用G460L型数控机床进行实验数据采集,将不同季节采集到的温度数据和热误差作为模型输入,采用循环学习率与正则化优化方法对模型进行训练。与LSTM、ConvLSTM和CNN-LSTM热误差模型对比,结果表明,AM-CNN-LSTM模型对特征还原能力最强,残差波动范围最小,其残差范围较最大值下降62.09%,模型预测精度在2.4μm以内。     

马铃薯清选机气力悬浮薯石分离装置设计与试验

针对以联合收获为主的北方马铃薯主产区,马铃薯收获后薯石分离人工捡拾工作量大、清选效率低,且清选洁净率较低等问题,利用薯块和石块密度不同的特点,采用气力悬浮输送技术设计了马铃薯清选机气力悬浮薯石分离装置,并基于该装置研究了不同参数调整条件下的清选特性。利用高速气流悬浮与振动筛的摆动作用,发挥气力悬浮和振动筛分离的双重优势,使薯块与石块在运动过程中实现自动分离。试验表明：当气流速度为35m/s、筛面倾角为18°和曲柄角速度为30rad/s时,马铃薯选出率均值为96.71%,清选洁净率均值为98.34%,各项性能指标均满足马铃薯清选作业要求。     

刚柔耦合串联机械臂末端位置误差分析与补偿

机械臂连杆柔性、关节柔性等非线性变形的综合影响,导致其末端位置发生偏离而产生误差。本文以IRB1410型串联机械臂为研究对象,采用理论分析、仿真分析与实验验证相结合的方式,对机械臂末端位置误差进行分析与补偿研究。首先,建立机械臂刚柔耦合理论误差模型,并运用Newmarkβ法进行数值仿真分析;联合ANSYS和ADAMS进行刚柔耦合机械臂末端位置运动误差仿真;为了实现快速补偿,提出基于BP神经网络的伪目标点法对位置误差进行补偿,补偿后其位置误差均方根减小了68.3%,说明该方法具有较好的补偿效果;最后,自主设计并搭建了测量实验平台,采用所提算法进行了误差补偿实验,对比补偿前后距离误差,补偿后误差均方根减小了77.01%,验证了伪目标点法对柔性误差补偿的有效性。     

大空间运动3-RRRU并联机器人运动学标定与误差分析

并联机器人具有高速、高刚度和大负载等明显优势,被广泛应用到农业和工业领域,但多关节导致该类机器人控制精度不高。针对大空间运动3-RRRU并联机器人的运动学建模和误差标定方法展开了系统、深入研究。综合应用DH法和空间矢量法建立了机器人的运动学模型,在此基础上,借助偏微分理论推导并建立机器人的误差模型;应用激光跟踪仪进行不同轨迹下机器人的空间位置数据采集,对一般遗传算法进行改进,以等步距搜索策略实现主要遗传算子的优化,并通过全局数值寻优获取机器人的误差补偿数据,完成标定和补偿工作。实验表明：基于直线标定方式,补偿后直线轨迹跟踪误差控制在0.14～1.34mm,但不适用于曲线轨迹补偿,其实测补偿后的最大误差高达5.08mm。曲线轨迹标定精度高于直线轨迹标定,补偿后将直线和曲线两种路径下的最大误差分别降低至1.18mm和1.56mm。该标定方法自动化程度高,适用于含有大量关节并联机器人的误差标定工作。     

Screening of Feather Degrading Bacteria and Optimization of Fer-mentation Conditions from Poultry

The purpose of this experiment was to isolate and screen the microorganisms from the soil where chicken feathers were piled for a long time and identify them biologically. Single-factor test and response surface methodology (RSM) analyses were used to explore the optimum conditions for the growth and fermentation of a strain. Screening of bacterial strains from the soil where feathers were piled for a long time was performed, and 12 keratinolytic bacterial strains were isolated. One of these isolates, CH-7, was found to be the most effective feather-degrading strain, which was identified as Lactococcus lactis KU568489.1. The growth situation of feather keratin degrading bacterium analysis results showed that the best degradation effect of CH-7 was found in oxygen, inoculation 5％, initial pH=6.5, fermentation temperature 30℃, speed 220 r ? min-1 and fermentation time 36 h, and CH-7 had the highest degradation rate of feather keratin. The optimization analysis of RSM showed that there was more significances of the three factors, 32℃, pH 6.3 and amount of inoculum at 5.7％ on the degradation of feather keratin. Based on the above results, the feather degrading bacterium, Lactococcus lactis CH-7, was obtained by screening, 30℃ and pH 5.5-6.5 were the best growth conditions. The oxygen, the amount of inoculum 5.7％, the fermentation temperature, 32℃, pH 6.3 and the rotational speed 220 r ? min-1 were the best fermentation conditions. Under these conditions, 38.48％ degradation rate was obtained after 36 h fermentation, which demonstrated the strain CH-7 was potential to the fermented feather meal for feed.     

基于注意力机制的小样本啤酒花病虫害识别

为提高啤酒花产业的品质和产量,针对于传统人工识别鉴定难度大,效率低和客观性不够,以及啤酒花病虫害识别无大型公开数据集等问题。本文提出一种基于软注意力机制的小样本啤酒花病虫害识别方法,对传统深度残差网络ResNet模型进行改进,并使用图像增强技术-直方图均衡化处理图片得到新的数据集。实验结果表明,在小样本情况下,相比于传统的模型,改进过后的模型A-ResNet50和A-ResNet101都能准确识别不同类型的病虫害图像,在测试集上的准确率为93.27%和93.11%,Kappa指数达到了0.9027和0.8996,证实了A-ResNet50和A-ResNet101模型在啤酒花病虫害识别上的可行性以及可靠性。本文提出的方法识别精度高,实现了啤酒花病虫害的智能识别,同时也对小样本数据集的高精度识别提供了一种途径。     

红毛藻中多组分藻胆蛋白及其亚基的分离

采用DEAE-52离子交换柱对红毛藻(Bangia fusco-purpurea Lyngb.)中的藻胆蛋白进行分离，依次洗脱出蓝色、紫色、深红色、红色和浅红色等几种组分的藻胆蛋白；经光谱分析发现蓝色组分为R-藻蓝蛋白，紫色组分为1种三峰型的藻胆蛋白，深红色组分和红色组分为2种不同但又非常相似的藻红蛋白，其中深红色组分可能与某些色素物质结合较为紧密，而最后洗脱出的浅红色组分则为变性的藻红蛋白。尿素洗脱结果表明，蓝色组分的R-藻蓝蛋白是由2种双峰型的亚基构成。本研究旨为揭示藻胆蛋白的能量传递机制提供理论和技术依据。     

Evaluation of different approaches to individual tree growth and survival modelling using data collected at irregular intervals - a case study for Pinus patula in Kenya

Background： The minimum set of sub-models for simulating stand dynamics on an individual-tree basis consists of tree-level models for diameter increment and survival. Ingrowth model is a necessary third component in uneven-aged management. The development of this type of model set needs data from permanent plots, in which all trees have been numbered and measured at regular intervals for diameter and survival. New trees passing the ingrowth limit should also be numbered and measured. Unfortunately, few datasets meet all these requirements. The trees may not have numbers or the length of the measurement interval varies. Ingrowth trees may not have been measured, or the number tags may have disappeared causing errors in tree identification. Methods： This article discussed and demonstrated the use of an optimization-based approach to individual-tree growth modelling, which makes it possible to utilize data sets having one or several of the above deficiencies. The idea is to estimate all parameters of the sub-models of a growth simulator simultaneously in such a way that, when simulation begins from the diameter distribution at the first measurement occasion, it yields a similar ending diameter distribution as measured in the second measurement occasion. The method was applied to Pinus patula permanent sample plot data from Kenya. In this dataset, trees were correctly numbered and identified but measurement interval varied from 1 to 13 years. Two simple regression approaches were used and compared to the optimization-based model recovery approach. Results： The optimization-based approach resulted in far more accurate simulations of stand basal area and number of surviving trees than the equations fitted through regression analysis. Conclusions： The optimization-based modelling approach can be recommended for growth modelling when the modelling data have been collected at irregular measurement intervals.