基于神经网络的多状态网络设备故障预测的研究
发布时间：2018-08-11 16:54
【摘要】：随着网络规模的不断扩大,网络中运行的网络设备如路由器、交换机等设备日益增多,能够确保网络正常运行,维护网络设备不出现故障,在出现故障之后能够迅速、准确地定位问题并排除故障,对于网络维护和管理人员是个很大的挑战。 为了克服传统维修方式的不足,随着状态监测和故障诊断技术的不断进步,逐渐发展起来一种新的维修方式——基于状态的维修(CBM)。该维修方式综合运用各种技术手段来获取设备的运行状态数据,然后运用故障预测和诊断技术对设备的运行状态进行判别,并预测其发展趋势以及诊断发生何种故障,实现了通过状态监测预测即将发生的故障,制订合理的维修策略。故障预测技术是故障诊断技术的重要组成部分,是通过对历史和当前的故障特征值进行分析,预测出未来的故障特征值,从而预测出设备在未来一段时间内的运行状态,预测设备可能出现的故障,并且依据这些特征值,判断设备的故障级别,提前掌握设备故障的发展趋势,为提早预防和修复故障提供依据,具有重要的理论研究价值和工程实践意义。 本文提出了基于神经网络的故障预测方法,引入基于状态的维修技术,构建了基于多状态在网运行设备故障预测模型。该模型根据故障的严重性将预警等级划分为四层,对于不同的预警级别,分别构建神经网络,解决了设备故障预测精度不高的难题,提升了基于多状态的故障预测能力。通过收集网络设备运行特征信息,得到设备的特征信息样本集,应用设计完成的神经网络对样本集进行训练,进一步优化神经网络的设计结构,建立基于神经网络的故障预测模型,实现对设备故障的预测和诊断。 基于状态的维修获得主要是基于设备的状态信息来预测设备的剩余寿命,以设定的优化准则为目标对设备做出维修决策,即判断设备是否需要进行预防性维修,如果需要,何时进行维修最合适。这种维修方式的维修间隔期是不固定的,其最大的特点是根据每个设备具体的状态,在设备故障发生前提早进行维修。对于设备,基于状态的维修可以降低维护维修费用、提高设备的可用性和任务成功率；通过减少维修,尤其是计划外的维修次数,缩短维修时间,提高设备运行效率；通过减少备品备件、维修人员等日常维护保障开支,降低维护和维修成本；通过状态监测,降低任务失败的风险,进一步提高任务的成功率,极大的提升了设备维护和维修水平。
[Abstract]:With the continuous expansion of the network scale, the network equipment such as routers, switches and other devices running in the network is increasing day by day, which can ensure the normal operation of the network, maintain the network equipment without failure, and be able to quickly after the failure. It is a great challenge for network maintenance and management to locate and troubleshoot the problem accurately. In order to overcome the shortcomings of traditional maintenance methods, with the continuous progress of condition monitoring and fault diagnosis technology, a new maintenance mode, the condition based maintenance (CBM).), has been gradually developed. The maintenance method synthetically uses various technical means to obtain the running state data of the equipment, and then uses the fault prediction and diagnosis technology to distinguish the running state of the equipment, and predicts its development trend and what kind of fault to diagnose. Through the condition monitoring to predict the upcoming failure, a reasonable maintenance strategy is worked out. Fault prediction technology is an important part of fault diagnosis technology. By analyzing the history and current fault eigenvalues, it can predict the future fault eigenvalues, and then predict the running state of the equipment in a certain period of time. To predict the possible faults of the equipment, and to judge the fault level of the equipment according to these characteristic values, to grasp the development trend of the equipment faults in advance, and to provide the basis for the early prevention and repair of the faults. It has important theoretical research value and engineering practical significance. In this paper, a fault prediction method based on neural network is proposed, and the fault prediction model of equipment running in network based on multi-state is constructed by introducing the state-based maintenance technology. According to the severity of the fault, the model divides the warning level into four layers. For different early warning levels, neural networks are constructed, which solve the problem of low precision of equipment fault prediction and improve the ability of fault prediction based on multi-state. Through collecting the characteristic information of the network equipment, the characteristic information sample set of the equipment is obtained, and the designed neural network is used to train the sample set, and the design structure of the neural network is further optimized. The fault prediction model based on neural network is established to predict and diagnose the fault of equipment. Condition-based maintenance is mainly based on the state information of the equipment to predict the remaining life of the equipment, and make maintenance decisions on the equipment with the set optimization criteria as the goal, that is, to judge whether the equipment needs preventive maintenance, if so, When maintenance is most appropriate. The maintenance interval of this kind of maintenance method is not fixed, and its biggest characteristic is that according to the specific condition of each equipment, the maintenance should be carried out early before the failure of the equipment. For the equipment, the condition based maintenance can reduce the maintenance cost, improve the availability of the equipment and the success rate of the task, reduce the number of maintenance, especially the unplanned maintenance times, shorten the maintenance time, and improve the efficiency of the equipment operation. Reducing maintenance and maintenance costs by reducing daily maintenance support expenses such as spare parts and maintenance personnel, reducing the risk of mission failure through condition monitoring, and further improving the success rate of the task, Greatly improved the equipment maintenance and maintenance level.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.05;TP183
本文编号：2177632
[Abstract]:With the continuous expansion of the network scale, the network equipment such as routers, switches and other devices running in the network is increasing day by day, which can ensure the normal operation of the network, maintain the network equipment without failure, and be able to quickly after the failure. It is a great challenge for network maintenance and management to locate and troubleshoot the problem accurately. In order to overcome the shortcomings of traditional maintenance methods, with the continuous progress of condition monitoring and fault diagnosis technology, a new maintenance mode, the condition based maintenance (CBM).), has been gradually developed. The maintenance method synthetically uses various technical means to obtain the running state data of the equipment, and then uses the fault prediction and diagnosis technology to distinguish the running state of the equipment, and predicts its development trend and what kind of fault to diagnose. Through the condition monitoring to predict the upcoming failure, a reasonable maintenance strategy is worked out. Fault prediction technology is an important part of fault diagnosis technology. By analyzing the history and current fault eigenvalues, it can predict the future fault eigenvalues, and then predict the running state of the equipment in a certain period of time. To predict the possible faults of the equipment, and to judge the fault level of the equipment according to these characteristic values, to grasp the development trend of the equipment faults in advance, and to provide the basis for the early prevention and repair of the faults. It has important theoretical research value and engineering practical significance. In this paper, a fault prediction method based on neural network is proposed, and the fault prediction model of equipment running in network based on multi-state is constructed by introducing the state-based maintenance technology. According to the severity of the fault, the model divides the warning level into four layers. For different early warning levels, neural networks are constructed, which solve the problem of low precision of equipment fault prediction and improve the ability of fault prediction based on multi-state. Through collecting the characteristic information of the network equipment, the characteristic information sample set of the equipment is obtained, and the designed neural network is used to train the sample set, and the design structure of the neural network is further optimized. The fault prediction model based on neural network is established to predict and diagnose the fault of equipment. Condition-based maintenance is mainly based on the state information of the equipment to predict the remaining life of the equipment, and make maintenance decisions on the equipment with the set optimization criteria as the goal, that is, to judge whether the equipment needs preventive maintenance, if so, When maintenance is most appropriate. The maintenance interval of this kind of maintenance method is not fixed, and its biggest characteristic is that according to the specific condition of each equipment, the maintenance should be carried out early before the failure of the equipment. For the equipment, the condition based maintenance can reduce the maintenance cost, improve the availability of the equipment and the success rate of the task, reduce the number of maintenance, especially the unplanned maintenance times, shorten the maintenance time, and improve the efficiency of the equipment operation. Reducing maintenance and maintenance costs by reducing daily maintenance support expenses such as spare parts and maintenance personnel, reducing the risk of mission failure through condition monitoring, and further improving the success rate of the task, Greatly improved the equipment maintenance and maintenance level.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.05;TP183
【引证文献】
相关期刊论文 前2条
1 杨健华;;网络环境大背景下的设备动态故障诊断与预测维修[J];西部广播电视;2016年22期
2 姚仲敏;沈玉会;;基于GA-BP的移动通信设备故障诊断[J];计算机测量与控制;2015年10期
相关硕士学位论文 前3条
1 张钱龙;基于信息融合的设备故障预测研究[D];郑州大学;2016年
2 贾永青;多变天气环境下消防给水设备智能巡检系统研究[D];湘潭大学;2015年
3 王振华;基于日志分析的网络设备故障预测研究[D];重庆大学;2015年
，本文编号：2177632
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2177632.html
