【作者】 孙云霞；

【导师】 陆奎；

【作者基本信息】 安徽理工大学 ， 计算机应用技术， 2011， 硕士

【摘要】 为了研究煤矿安全预测本质,正确有效地预测煤矿系统的安全状况,基于目前煤矿安全研究现状和手段,从煤矿系统安全预测的内在规律性、有效时间长度及有效性等方面对煤矿进行了研究,试图总结出煤矿系统安全预测的本质特征,并研究安全预测的建模过程。本论文依据以往的煤矿安全评价中得到的经验,结合我国以及国外煤矿安全评价与预测的发展情况,并利用从煤矿现场所采集到的数据建立煤矿安全动态评价以及预测模型。本论文是以神经网络和模糊神经网络两者的优点为基础,建立煤矿安全评价的模糊神经网络模型,此模型在并行计算、对复杂数据的处理以及自身的学习等方面具有更强的能力。本论文主要研究如何去建立一套完整评价指标体系,它能够客观的去反映评价对象的本质,该指标体系希望尽可能的包括能够影响到煤矿安全生产的所有因素(但实际上并不能把所有的影响因素都覆盖),同时建立的这个指标体系能够应用在煤矿安全的神经网络和模糊神经网络模型中,并能够进行应用研究；最后,对矿井安全状态的动态预测主要利用的是神经网络对时序性指标的预测功能。最后,本论文将结合具体实例,建立煤矿安全预测系统。因为非线性组合预测模型有较高的精度,并且非线性组合要好于线性组合,非线性组合预测模型能克服单一预测模型的缺点,从而能够解决由于系统状态安全指标数量随机变动而造成的预测困难,所以本论文将采用模糊神经网络与BP反馈神经网络相结合的非线性组合预测模型。图34表11参50更多还原

【Abstract】 The coal and mine system’s security status could be predicted accurately and efficiently by researching into nature of coal mine safety forecast. Based on actuality and means of present research on coal mine safety, research on internal regularity and effective time of coal mine safety forecast has been done in order to summarize substantive characteristics on coal mine safety forecast and study modeling process of safety forecast.According to experience gained from previous coal mine safety evaluation, combined with present status of development of coal mine safety evaluation and forecasting at home and abroad, coal mine safety evaluation and forecasted model is built by utilizing data collected on scene of coal mine.Based on neural network and fuzzy neural network, neural network model of coal mine safety evaluation is built. That model is better at parallel computing, complex data processing and ability of learning from itself.A suit of integrated targeted system used for evaluation which can reflect nature of evaluated object objectively will be mainly researched. That targeted system is wished to include all the factors which can affect coal mine safety production. In fact, not all the factors could be covered. This targeted system can be employed in model of neural network and fuzzy neural network of coal mine safety and be studied in applied aspect. The mining well security is dynamically forecasted by mainly utilizing the ability that neural network can forecast timing targets.Finally, coal mine safety forecasting system will be built combined with concrete examples. Nonlinear combination forecast model has very high accuracy and nonlinear combination is better than linear one. Nonlinear combination forecast model can overcome the shortcoming of single forecast model. Then difficulty in forecasting caused by stochastic alternation of targeted value for system safety can be solved. Forecast model with fuzzy neural network and BP feedback neural network will be used.Figure 34 table 11 references 50更多还原