滚筒式采煤机故障诊断研究

发布时间：2019-02-15 21:08

【摘要】：煤炭生产的关键设备是采煤机,其构造复杂,且在环境恶劣的井下工作,因此其电气、液压等系统故障率较高,经过调研陕西神木市数十家煤炭企业得知现在采煤设备故障诊断非常落后,主要凭借传统经验去判断出现的问题,其准确性和效率无法保证,自动化程度较低,已经成为煤炭产业发展的重要瓶颈,因此研究采煤机故障诊断方法很有现实意义和实用价值。本文主要研究采煤机故障诊断神经网络模型、总体结构,训练算法与步骤,提出了最优学习算法(ELM算法),建立采煤机滚动轴承BP神经子网络故障诊断模型,并进行MATLAB诊断误差研究;研究基于专家系统、模糊神经网络的混合智能算法采煤机故障诊断方法,建立采煤机液压牵引装置系统过热故障诊断样本及模型,采用单自适应BP网络算法和模糊BP网络算法进行故障诊断,并对比分析诊断误差、训练速度。本文的主要研究结果如下:1.基于神经网络的采煤机故障诊断方法采用最优学习算法(ELM算法)避免前馈神经网络算法误差和权值范数较大等缺点,提高故障诊断网络的泛化性。2.采煤机滚动轴承BP神经子网络故障诊断模型经6次循环训练后达到目标误差,诊断误差小于0.01,说明基于BP神经网络采煤机故障诊断方法是可行的、高效的。3.采煤机液压牵引装置过热故障诊断模型误差达到0.001,模糊模块自适应BP网络算法训练样本仅仅迭代1500次,而自适应BP网络算法必须要迭代3500次,前者故障诊断效率更高。4.相同样本训练情况下,基于混合智能算法的采煤机故障诊断方法诊断误差小,适应能力更好。5.基于混合智能算法的采煤机故障诊断方法使整个故障诊断系统的诊断效果得到优化。
[Abstract]:The key equipment in coal production is shearer, whose structure is complex, and it works in the downhole where the environment is bad, so its electrical, hydraulic and other systems have a high failure rate. After investigating dozens of coal enterprises in Shenmu City, Shaanxi Province, they know that the fault diagnosis of coal mining equipment is very backward now. It mainly relies on traditional experience to judge the problems that appear. Its accuracy and efficiency cannot be guaranteed, and the degree of automation is low. It has become an important bottleneck in the development of coal industry, so it is of practical significance and practical value to study the fault diagnosis method of shearer. In this paper, the neural network model, overall structure, training algorithm and steps for fault diagnosis of shearer are studied. The optimal learning algorithm (ELM algorithm) is put forward, and the fault diagnosis model of BP neural network for shearer rolling bearing is established. The diagnostic error of MATLAB was studied. Based on expert system and fuzzy neural network, the fault diagnosis method of shearer based on hybrid intelligent algorithm is studied, and the diagnosis sample and model of overheating fault of hydraulic traction device system of shearer are established. The single adaptive BP network algorithm and fuzzy BP network algorithm are used for fault diagnosis, and the diagnosis error and training speed are compared and analyzed. The main results of this paper are as follows: 1. The fault diagnosis method of shearer based on neural network adopts the optimal learning algorithm (ELM algorithm) to avoid the shortcomings of the feedforward neural network algorithm such as large error and weight norm, and to improve the generalization of fault diagnosis network. 2. After six cycles training, the fault diagnosis model of BP neural network of shearer rolling bearing reaches the target error, and the diagnostic error is less than 0.01, which shows that the fault diagnosis method based on BP neural network is feasible and efficient. The model error of overheating fault diagnosis of hydraulic traction device of shearer is 0.001. The training sample of fuzzy module adaptive BP network algorithm is only 1500 iterations, and the adaptive BP network algorithm must be iterated 3500 times. The former fault diagnosis efficiency is higher. 4. In the case of the same sample training, the fault diagnosis method of shearer based on hybrid intelligent algorithm has less error and better adaptability. The fault diagnosis method of shearer based on hybrid intelligent algorithm optimizes the diagnosis effect of the whole fault diagnosis system.
【学位授予单位】：西北农林科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD421.61;TP183

【参考文献】