基于WSN的智能养殖控制系统研究-检测技术与自动化装置专业论文.docxVIP

万方数据 万方数据 摘要摘要 摘要 摘要 水体是高复杂度、非线性、大时滞的系统，水体水质特别是溶解氧一直以来 都是人工手动调控。人类专家通过观察水体水质状况，结合养殖经验、气象要素 即可做出决策。本研究使用人工神经网络完成水质分类，使用专家系统的知识和 规则表示气象要素以及控制经验，通过将二者相结合的方式解决水体溶解氧调控 和泛塘预测问题。 (1)采用神经网络对水质进行评价本质是多分类问题，建立综合水质评价 神经网络对水质进行分类，并自定义评价指标以指导网络训练过程，并阐明综合 水质评价网络的有效性、合理性。 (2)本研究根据文献以及养殖专家提供的知识，利用Jess实现溶解氧的“专 家调控策略”，采用模糊专家控制和“专家调控策略”相结合的方法完成对水体 溶解氧的控制，对比实验表明可达到专家调控效果。由于养殖过程中泛塘出现频 率高且对水产品影响严重，本研究使用综合水质评价网络的评价结果结合气象因 素以及专家经验对养殖水体是否发生泛塘进行预测，使用Jess完成实验，实验验 证了此方法的可行性。 (3)为解决数据采集以及远程控制问题，设计并实现基于WSN的实时监 控系统。为解决系统实时性问题，将非实时和实时数据请求分别处理，并采用 WebSocket进行实时数据传输。对比实验结果表明可显著地提高实时性。 关键字：无线传感网人工神经网络水质评价模糊专家系统智能决策云平台 监控系统 摘要 11 AbstractABSTRACT Abstract ABSTRACT The aquaculture water is highly complex，nonlinear,large delay system，the regulation of water quality,especially the dissolved oxygen has always been manually controlled．Human experts can make decisions just by observing the quality of aquaculture water,the experience of fish farming and meteorological condition． Artificial neural networks are used for water quality classification，meteorological conditions and control experience represented by the knowledge and rules of expert system．This study solve the regulation problem ofthe dissolved oxygen and prediction problem of fish floating head or asphyxia by combining decision outcomes by artificial neural networks with expert knowledge． (1)The nature of water quality classification by the artificial neural networks is a multi-classification problem．A Comprehensive Water Quality Evaluation Neural Network(CWQNN)is constructed to make classification and this study develops an indicator to guide the trainning and demonstrate the effectiveness，reasonability of CWQNN according to the evaluation results． (2)This study implements‘Expert Regulation Strategy of Dissolved Oxygen’ according to the related publications and the knowledge offered by the experts by Jess and achieves the control ofthe dissolved oxygen by combining the fuzzy expert control wim‘Expert Regulation Strategy ofDissolved Oxygen’．Experi