基于里程信息融合的株间锄草刀定位数据优化方法

为提高株间锄草刀定位精度、降低机器视觉受外界因素的影响,该文提出里程信息融合机器视觉的方法对锄刀定位数据进行优化。通过分析定位数据校正和视觉滞后补偿的原理,设计了模糊逻辑校正器,通过模糊规则将模糊校正系统简化为单输入单输出形式,采用Mamdani模糊推理方法获得视觉数据可信度决策表,将可信度作为加权值生成校正锄刀定位数据,并提出采用实时里程信息作为视觉滞后补偿量的方法,给出补偿公式。田间刀苗距优化静态试验表明,视觉刀苗距误差为9.88 mm,优化后刀苗距误差为6.06 mm;动态试验表明,视觉数据出错率为4.8%～6.6%,刀苗距变化曲线显示,优化方法可有效过滤视觉坏点或不稳定的数据点,将视觉滞后纳入衡量标准,不同车速下动态优化后刀苗距平均误差为5.30～7.08 mm,较优化前降低了25%左右。研究结果表明,锄草刀定位数据优化方法可有效提高机器视觉静态和动态获取刀苗距的精度。该研究为提高株间锄草技术的锄刀定位精度提供了参考。     

Regulatory issues for genetically modified animals

Precision genetics and breeding have the potential to meet the agricultural needs and goals of the world in the 21st century. These needs include increasing the efficiency of production of animals and improving their products with minimal impact on the environment. The USA is the major innovator in genomic science and the acknowledged leader in formulating policies to regulate genetic applications in medicine and agriculture. However, governments worldwide have been exceedingly reluctant to support the introduction of genetically modified (GM) animals into agriculture. Regulatory policies have stagnated due to legal guidelines that could not anticipate the needs and solutions that are evident today. This must change if we are to maintain planetary integrity. I propose a new, market-based regulatory model for GM livestock that has both a strong scientific foundation and has worked for 10000 years. The model is similar to that for information technology in which specific algorithms drive computer and cell phone applications. Genome engineers write genetic algorithms that drive the traits in biological organisms. Accordingly, GM products should be viewed in terms of their use and public benefit rather than by limitations to the genetic programing coming from a few highly vocal groups. Genetic algorithms (Genapps) of the 21st century will include not only introduction of synthetic genes, but also complete natural and synthetic biochemical pathways to produce agricultural products that are maximally efficient, healthy to humans and animals, and sustainable in an era of changing climates while avoiding environmental degradation.     

倍半萜类抗癌抑制剂的定量构效关系研究

对具有抑制人早幼粒白血病细胞(HL-60)活性的倍半萜类化合物进行二维定量构效关系研究,利用遗传算法建立2D-QSAR模型,得到10个具有较好预测能力的QSAR模型。分析模型得出:脂水分配系数(ALogP)、分子量(Molecular_Weight)与化合物活性呈正相关,偶极距x分量(Dipole_X)、分子表面积(Molecular_SurfaceArea)与化合物活性呈负相关;倍半萜2D-QSAR模型中独特的2个分子参数分别是VSA_AlogP[2]和VSA_AlogP[10];VSA_AlogP[2]参数与化合物抗癌活性呈正相关,在此区域内增加羟基等影响分子脂水分配系数的官能团能增加化合物的抗癌活性,VSA_AlogP[10]参数与化合物活性呈负相关,此区域内减少羟基等影响分子酯水分配系数的官能团会增加化合物的抗癌活性。该组模型将为新型抗癌类药剂的筛选以及新型抗癌类药物设计提供指导。     

基于遗传算法的液肥变量施肥控制系统

为解决大田牵引式液肥施肥机的变量施肥作业精度不高、施肥流量不均匀以及肥料浪费问题,该研究针对液肥变量施肥控制系统,基于遗传算法的模糊PID(Proportion Integral Derivative)对电动比例阀的控制过程进行优化。首先对牵引式液肥变量施肥机的控制过程进行分析,建立液肥变量施肥控制系统的负反馈控制模型。根据控制系统要求,将模糊控制规则进行染色体编码,通过选择、交叉、变异等遗传算子对模糊控制规则进行仿真寻优,得到最优模糊控制规则表。依据得到的最优模糊控制规则对模糊PID控制器进行设置,并通过MATLAB软件进行仿真分析,结果表明,基于遗传算法的模糊PID控制的响应时间为4.86 s,小于传统PID控制的8.4 s和模糊PID控制的7.32 s。搭建试验平台进行液肥变量施肥控制系统流量控制的稳定性试验和变量控制试验,得到传统PID、模糊PID以及基于遗传算法的模糊PID在系统稳定运行时流量控制的相对误差分别为5.19%、3.40%、1.14%,响应时间分别为5.19、4.12、3.21 s,基于遗传算法的模糊PID较传统PID的相对误差减少了4.05个百分点,响应时间减少了1.98 s;基于遗传算法的模糊PID较模糊PID的相对误差减少了2.26个百分点,响应时间减少了0.91 s。基于遗传算法的模糊PID对液肥流量的控制效果优于传统PID和模糊PID,本文控制方法为变量施肥的研究提供了一种可行方案。     

遗传算法在MDF热压模糊控制规则优化中的应用

对于给出的中密度纤维板(MDF)热压模糊控制器的建模,采用二进制编码策略,基于分层遗传算法通过改变控制基因和参数基因,应用再生、交叉、变异等遗传操作在模糊控制规则中寻优.仿真结果表明,优化得到的中密度纤维板热压机模糊控制器与传统的模糊控制器相比具有响应快、超调小的特点,动态性能和稳态误差都得到提高.     

Bi-level Programming Model and Genetic Algorithms for the Selection of Vendors

The traditional vendor selection model often pay attention to the profit of the buyers and the selection behavior of vendor is ignored,but the vendors only trade with the buyers according with their rational selection,so the vendors selection bi-level programming model is presented with constraint of minimal batches,the ability of supply,product matching etc.A solution of the model based on genetic algorithm is proposed.The buyers' vendor selection of minimum cost is realized in the upper programming and the allocation of requirements is proposed in the lower programming.The application of the model and its algorithm are illustrated with a practical example.     

Bilevel Programming Model and Solution Algorithm for the Supplier Selection in Automobile Components

Focusing on the characteristics of automobile components industry, such as long-term supplier selection, large number of suppliers, various assessment index, complicated processes, and extremely strict assessment requirements, an improved supplier selection bilevel programming model for automobile components industry is established. With this model, the authors analyzed selected objectives from viewpoints of both suppliers and purchasers, assessed several selectied factors with upper and lower layer objective functions, and produced a optimal supplier selection method to achieve minimum purchasing costs. This model takes several constraints, such as technology, quality, quoted price, supply capacity and services of suppliers, into consideration, which makes the model very practical. A genetic-algorithms-based calculation method is designed, and the model is validated through a case study. Both the model and the method provide valuable support to supplier selection in the automobile components industry.     

Analysis of the Algorithm of Incremental Mining of Sequential Patterns

The author analyzes an algorithm for Incremental Mining of Sequential Patterns (IMSP), which is presented in a paper to speed up the current mining process by using the previous results when the underlying database has minor changes. Some improper results of in the paper are pointed out, and counterexamples are presented respectively.     

基于二分法的鸡蛋图像边缘快速检测方法

机器视觉技术广泛应用于鸡蛋感官品质无损检测研究中,快速得到鸡蛋图像边缘有助于高效识别鸡蛋的几何特征参数。借鉴计算机二分法快速解方程的算法,提出用二分法快速检测图像边缘,把对图像的整体研究转换到具体的行图像来研究,利用行图像信息构建了满足二分法求解条件的方程,并把对边缘的检测构造成对方程的求解问题,从而建立了二分法求解鸡蛋图像边缘的理论基础,最后在应用部分给出了详细的程序实现步骤。试验结果表明：该方法可以检测鸡蛋图像边缘,检测效率提高了约20倍,大大加快了检测速度,可为鸡蛋自动化检测提供更高的效率。