【作者】 王瑞梅；

【导师】 傅泽田；

【作者基本信息】 中国农业大学 ， 管理科学与工程， 2003， 博士

【摘要】 池塘水质是制约池塘养殖产品产量和质量的关键因素，由于缺乏科学合理的评价、预测、预警方法和手段，当前的水质管理还处于经验和半经验管理状态，制约了渔业的高产、优质、高效和可持续发展。为了解决上述问题，本文运用环境评价、人工智能、生态系统与模糊数学等理论，研究了池塘水质中各因子的变化规律，建立了评价、预测及预警模型，开发了池塘水质管理智能决策支持系统，为提高池塘水质管理提供了科学的理论、方法和手段。 池塘水体是一个多因子相互作用的复杂巨系统，基于实验和专家访谈，运用系统动力学的分析方法对池塘水质因子及其影响因素进行系统分析，建立了池塘水温、PH值、溶解氧、氮、磷、碱度、光照和透明度等水质因子的系统动力学模型，阐明了池塘水质因子间相互作用、相互影响的关系。 在综合分析池塘水质各影响因子的基础上，通过专家问卷调查法和德尔裴法对水质因子的重要程度进行了排序，建立了池塘水质评价指标体系和评价标准。用重要程度法和实测值相对隶属度法确定了指标权重，建立了池塘水质评价模型，对池塘水质进行了评价。 在分析池塘水质各因子之间的关系的基础上，建立了系统动力学模型，对气温、水温、日照和浮游植物生物量进行模拟；在大量监测数据的基础上，对四个关键且难以监测的水质因子，建立了基于神经网络模糊系统预测模型，应用改进的神经网络模型对上述四个水质因子进行预测，该模型预测值与实测数据拟合良好，解决了生产实际中水质因子状况无法及时预报和动态监测难等问题。 在池塘水质评价和预测的基础上，建立了池塘水质单因子状态预警模型、多因子状态预警模型、趋势预警模型和鱼类生存指数预警模型，确定了池塘水质预警的警级标准。并提出了池塘水质预警管理措施。 最后，本文遵循软件工程设计思想，运用ASP、VC++等开发工具，完成了池塘水质管理智能决策支持系统的设计与实现。更多还原

【Abstract】 Pond water quality is the key factor of limiting the output and quality of breed product. Because of lacking the scientific and reasonable method and instrument for evaluation, forecast and early-warning. The current water quality management is in the state of experience and semi-experience management. This method restricts fishery high output, high quality, high efficiency and sustainable development. In order to solve above problem, the paper using the theory of environment evaluation, of artificial intelligence, of biogeocenose and of fuzzy mathematics investigates the transformation rule of the every factor in the pond water quality, establishes the estimation model, forecast model and early-warning model, achieves the intelligent decision support system of pond water quality management which provides scientific theory, method and instrument for improving the pond water quality management.Pond water is a complicated and great system composed of many factors interactional. Based upon experiment and expert visit, after analyzing the factors of the pond water quality and other factors which influencing them through the method of system dynamics, the paper founds the system dynamics model for water temperature-, PH value, dissolved oxygen, nitrogen, phosphor-, alkalinity, illumination and transparence, and also clarifies the relation of interaction of the factors of pond water quality.On the basis of the general analysis of every factor influencing the pond water quality, the paper founds the system of target estimation and evaluation standard after sorting by importance of the factors by means of Delphi and expert investigation. In the paper, Target weight was confirmed using the method of important program and relative membership of grade of the measured density, evaluation model of pond water quality was founded, and water quality was evaluated.Based on the analysis of the relation among all factors of the pond water quality, the paper founds system dynamics model to simulate air temperature, water temperature, sunlight and quantity of the alga biology, On the basis of much testing data, the paper establishes a forecast model for four key factors difficult to watching and measuring based on neural network fuzzy system, the model using the improved neural network forecasts the above four factors, and the result accords with the testing data approximately. So it solves the problem that the status of the water quality couldn’t be predicted in time and it’s hard to watching and measuring the factors dynamic.On the basis of the evaluation and forecast of pond water quality, the paper establishes forecast model of status of single factor and multi-factor of pond water quality, of trend, of survival index of fish, confirms the warning grade of pond water quality forecast, and puts forward the measures of forecast management of pond water quality.In the end, the paper follows the idea of software engineering, uses the tools ASP, VC++ and etc, achieves intelligent decision support system of pond water quality.更多还原