生物氧化提金预处理中氧化槽温度控制系统故障诊断研究

发布时间：2018-02-08 16:25

本文关键词： 生物氧化预处理氧化槽温度控制系统学习型和声算法(LHS) BP神经网络故障诊断模型　出处：《新疆大学》2017年硕士论文　论文类型：学位论文

【摘要】：在生物氧化的过程中,温度是影响细菌生存的重要条件之一。细菌繁殖的快慢及活性的高低,直接影响矿浆在预处理过程中氧化的程度,从而影响提金率,所以氧化槽内温度能否维持恒定就显得尤为重要。一旦系统中执行器或传感器发生故障造成氧化槽内温度突变,将会导致细菌失活甚至死亡,造成停产等难以挽回的损失。因此,对氧化槽内温控系统做出故障诊断在实际生产中具有重大意义。通过结合现场生产过程,以氧化槽实验设备为对象,开展氧化槽温控系统故障诊断方法研究。本文主要的研究工作如下:针对氧化槽故障发现以人工排查方式为主的现状,提出一种基于BP神经网络的智能故障诊断模型并应用于氧化槽温控系统。通过对氧化槽内热量变化做理论分析,得到氧化槽热量机理模型,进而对模型分析、查找导致温度变化的故障形成原因和相应的故障检测点,确定模型的输入输出量和基本结构,为后期模型完善奠定了基础。对于故障诊断模型建立时的参数最优值选择问题,将学习策略与和声搜索算法相结合用于BP神经网络参数优化。利用引入学习策略的和声算法(LHS)较好的全局优化性弥补BP神经网络易陷入局部最优的劣势,建立最优参数LHS-BP神经网络故障诊断模型。与基本和声算法(HS)相比,该算法一方面将参数设置变化成为动态的,用于增加解的多样性和提高算法的收敛速度,另一方面将教与学优化算法与和声算法两种启发式算法结合,用于位置更新来提升算法跳出局部最优的能力。通过测试验证该算法具有更强的全局寻优性能。为了验证本文方法的可行性和有效性,利用实验设备搭建氧化槽模拟仿真平台,获得样本数据和测试数据,通过对样本数据进行网络训练确定模型参数,将测试样本带入模型得到测试结果。比对结果表明,本文方法提高了故障诊断的准确率,在理论和实践中都起到了一定的指导意义。
[Abstract]:In the process of biological oxidation, temperature is one of the important conditions affecting the survival of bacteria. So whether or not the temperature in the oxidation tank can be kept constant is very important. Once the actuator or sensor in the system fails to cause the temperature in the oxidation tank to change, it will lead to bacteria inactivation and even death. The failure diagnosis of the temperature control system in the oxidation tank is of great significance in actual production. By combining with the field production process, the experimental equipment of the oxidation tank is taken as the object. The main research work of this paper is as follows: in view of the current situation of the oxidation tank fault detection by manual inspection, the main work of this paper is to study the fault diagnosis method of oxidation tank temperature control system. An intelligent fault diagnosis model based on BP neural network is proposed and applied to the oxidation tank temperature control system. Finding out the cause of fault formation and the corresponding fault detection points, determining the input and output quantity and the basic structure of the model, which lays the foundation for the later model perfection, for the problem of selecting the optimal value of the parameters when the fault diagnosis model is established. The learning strategy and harmony search algorithm are combined to optimize the parameters of BP neural network. The better global optimization of LHS-based learning strategy is used to make up for the disadvantage that BP neural network is prone to fall into local optimum. The optimal parameter LHS-BP neural network fault diagnosis model is established. Compared with the basic harmonic algorithm, the algorithm changes the parameter setting into a dynamic one, which is used to increase the diversity of the solution and improve the convergence speed of the algorithm. On the other hand, it combines two heuristic algorithms: teaching and learning optimization algorithm and harmony algorithm. It is used to update the position to improve the ability of the algorithm to jump out of the local optimum. The test results show that the algorithm has stronger global optimization performance. In order to verify the feasibility and effectiveness of the proposed method, the oxidation tank simulation platform is built with experimental equipment. Sample data and test data are obtained, and model parameters are determined by network training of sample data, and test samples are brought into the model to obtain test results. The comparison results show that this method improves the accuracy of fault diagnosis. In theory and practice has played a certain guiding significance.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP273;TP277

【参考文献】