马斯京根洪水演算模型的改进--兼论其参数的蚁群算法率定

针对马斯京根洪水演算模型的近似性问题，用曲线积分法改进了原模型中时段流量的简单算术平均值法，并利用连续性空间优化问题的蚁群算法对改进的马斯京根洪水演算模型进行参数最优估计，使演算流量更加接近实际。     

基于遗传神经网络的入侵检测研究

入侵检测技术是计算机网络信息安全检测的重要手段之一,入侵检测作为一种动态的安全防护技术,提供了对内部攻击、外部攻击和误操作的实时保护,在网络系统受到危害之前拦截和响应入侵。对计算机网络数据进行特征提取,提出了采用遗传算法和神经网络相结合入侵检测技术。遗传算法具有计算简单、优化效果好的特点。利用遗传算法来避免BP算法的局部极小点,从而达到均方根误差全局最小点,也解决了BP算法的收敛慢的问题;同时也解决了单独利用GA往往不能在短时间内寻找到接近最优解的这一问题。通过计算机实验验证了入侵检测的效果,提高了识别率,使得误报率和漏报率降低。     

基于层次分析法—模糊综合评价(AHP-FCE)模型优化矿井通风系统的研究

矿井通风系统是一个复杂的非线性系统,具有随机性、模糊性和不确定性特征。应用层次分析法(AHP)和模糊综合评价(FCE)的基本理论建立的AHPFCE模型,可以全面考虑矿井通风系统的各种因素,既能体现评价过程的模糊性,又能尽量减少个人主观臆断带来的弊端,使优选的方案更符合实际,评价结果更可靠。应用加速遗传算法检验和修正判断矩阵的一致性,具有搜索效率高、计算时间短、求解精度高、计算结果稳定、适用性强等特点,是一种全局优化方法。采用AHPFCE方法确定权重,具有较强的逻辑性、实用性和系统性,并能准确地得出各评价指标的权系数,使计算结果更客观和稳定。实例计算结果表明:AHPFCE方法简便和通用,不仅能指导生产实际,而且为类似的方案优化工作提供了一种新的思路和方法。     

快速路交通流异常数据判断算法研究及实证

对快速路交通流数据进行异常数据判断,有利于避免使用异常数据带来的损失,提高信息利用的有效性。笔者分别根据逻辑推理、交通流的重复性和连续性以及交通流变量之间的机理分析提出了3种判断快速路交通流异常数据的算法,并讨论了这3种算法之间的集成。利用北京快速路实测数据对算法进行了验证,验证结果表明该算法基本是有效的。     

A FAST RECURSIVE GIS ALGORITHM FOR COMPUTING STRAHLER STREAM ORDER IN BRAIDED AND NONBRAIDED NETWORKS1

ABSTRACT: Stream ordering is a useful property of every river network, having a wide range of applications. A method for determining stream orders that quickly and easily addresses various network topologies and magnitudes is therefore needed. This paper introduces a general recursive stream ordering framework for vector hydrography. It also presents a linear, O(n), stream ordering procedure for braided river networks, which is a major improvement to the existing quadratic, O(n²), procedure. The discussion includes results and interpretations, and the appendices present procedure pseudocodes and thorough line by line explanations.     

OPTIMIZATION PROCEDURE FOR COST EFFECTIVE BMP PLACEMENT AT A WATERSHED SCALE1

ABSTRACT: A combinatorial optimization procedure for best management practice (BMP) placement at the watershed level facilitates selection of cost effective BMP scenarios to control non point source (NFS) pollution. A genetic algorithm (GA) was selected from among several optimization heuristics. The GA combines an optimization component written in the C++ language with spatially variable NFS pollution prediction and economic analysis components written within the Arc View geographic information system. The procedure is modular in design, allowing for component modifications while maintaining the basic conceptual framework. An objective function was developed to lexicographically optimize pollution reduction followed by cost increase. Scenario cost effectiveness is then calculated for scenario comparisons. The NPS pollutant fitness score allows for evaluation of multiple pollutants, based on prioritization of each pollutant. The economic component considers farm level public and private costs, cost distribution, and land area requirements. Development of a sediment transport function, used with the Universal Soil Loss Equation, allows the optimization procedure to run within a reasonable timeframe. The procedure identifies multiple near optimal solutions, providing an indication of which fields have a more critical impact on overall cost effectiveness and flexibility in the final solution selected for implementation. The procedure was demonstrated for a 1,014‐ha watershed in the Ridge and Valley physiographic region of Virginia.     

预测低温冷害的门限回归模型

提出用门限回归模型（TR）来描述低温灾害指标变量在灾害状况与非灾害状况的突变性。研制了TR建模的一套简便通用的方案，用笔者改进的遗传算法可同时优化门限值和回归系数，从而解决了TR建模过程所涉及的大量复杂寻优工作这一问题，为TR模型的广泛应用提供了有效工具。实例的计算结果说明了其可行性和有效性。该方案具有通用性，在各种自然灾害时序预测中具有广泛的实用价值。     

基于遗传算法与人工神经网络相结合的玉米估产研究

在遗传算法 (GeneticAlgorithm )与误差反传 (BackPropagation)网络结构模型相结合的基础上 ,设计了用遗传算法训练神经网络权重的新方法 ,并对吉林省梨树县的玉米进行了估产研究 ,同时与BP算法和灰色系统理论模型进行了比较。经检验 ,计算值与实际值接近 ,并优于灰色理论模型 ,具有良好的预测效果 ,从而为农作物估产提供了新方法     

Predicting the Range of Chinese Mitten Crabs in Europe

Abstract:  Ecological niche modeling provides a means for predicting the potential future distribution of a nonindigenous species based on environmental characteristics of the species' native range. We applied this method to the Chinese mitten crab (Eriocheir sinensis) , a catadromous crustacean with a long history of invasion in Europe. We used genetic algorithm for rule-set prediction to predict the potential European distribution of mitten crab based on its distribution in 42 locations in its native Asia. The climatic variables, air temperature, number of days, amount of precipitation, and wetness index, contributed significantly to predictions of native distribution limits. Although the genetic algorithm for rule-set prediction model was developed for the native range, the species' extensive distribution in Europe ( n = 434) allowed independent validation of the predictions. Application of the model to Europe was successful, with 84% of occurrences in regions predicted to be suitable by >80% of the models and <4% of occurrences in areas predicted suitable by <50% of the models (mainly along the northern range). At the watershed scale, areas with established mitten crab populations had significantly higher habitat matching than sites that were not invaded. The independent validation of the Asian-based model by the European distribution revealed that predictions were highly accurate. The model also identified large areas of Europe, particularly along the Mediterranean coast, as vulnerable to future invasion. These predictions can be used to develop strategies to control the spread of mitten crab by preventing introductions into vulnerable areas.     

Area-Based Refinement for Selection of Reserve Sites with the Benefit-Function Approach

Abstract:  Optimization of resource use is necessary for efficient conservation planning. Many reserve-selection algorithms aim to identify representative but inexpensive networks, which may lead to selecting small sites due to their lower costs and collectively higher species richness. Nevertheless, larger sites would be preferable regarding species' long-term persistence. An area-based refinement can be used to overcome this problem. We used a reserve-planning framework in which continuous benefit functions valued representation (numbers of populations), and differential species weights were based on a species' local rarity and threatened status. We introduced a refinement based on the species-area relationship that provides relatively higher values for larger sites. We applied the proposed method to rich fen vegetation in southern Finland. The species-area refinement resulted in a network of significantly larger sites with minor trade-offs with representation (numbers of populations). Giving endangered species higher weights ensured that the trade-off occurred mostly between site size and representation of low-priority species. We recommend using a species-area refinement for practical, maximum-coverage conservation planning.