共查询到19条相似文献,搜索用时 656 毫秒
1.
模糊神经网络推理的实时故障诊断专家系统 总被引:5,自引:0,他引:5
将模糊神经网络推理引入专家系统,采用修正的RLS算法训练网络的权系数,以此开发了模糊神经网络实时故障诊断专家系统,并将其成功应用于某化工厂大型PTA装置。文章介绍了系统的总体结构和主要特点,并以溶剂脱水塔釜水浓度高事件为例阐述了系统的具体实现。现场运行表明:该系统预报准确,界面友好,能满足工厂的实际需求,具有良好的易维护性和可扩充性。 相似文献
2.
针对整流装置不同故障数据的特点,采用框架和规则混合知识表达方式设计专家系统,并结合C++语言和关系数据库实现了专家系统外壳;该系统具有框架式知识库对整流装置部件故障的离线诊断、规则式知识库对整流装置实时测点数据的在线维护两大功能;系统被用于某食盐水电解大型整流电源上,获得良好的应用效果。 相似文献
3.
4.
介绍了自动称重进料系统在天津石化PTA装置的成功应用。包括称重进料系统的设备组成,控制系统结构组成,PLC与现场仪表及DCS的信号连接,控制单元的组成,PLC功能及数据处理,操作控制原理和系统动作过程。该系统的成功应用,保证了PTA装置生产的优质和稳定,取得了良好的经济效益。 相似文献
5.
一个用于化工过程故障诊断的知识基系统 总被引:3,自引:0,他引:3
本文开发研制了一个用于化工过程故障诊断的实时专家系统 CPFD.该系统由动态数据库、知识库、推理机构和人一机界面等部分组成,知识库由带有确定性因子的产生式规则组成,推理机构(即故障诊断问题求解机构)采用不确定性推理、最佳深度优先控制策略与正反向混合搜索策略.该诊断系统已成功地应用于某大型工业催化裂化装置. 相似文献
6.
现代船用柴油机是机一电.液等各种子系统组成的大型机电设备,结构复杂决定了对其故障进行诊断的困难性。基于目前发展较快的船用柴油机故障诊断技术及较成熟的专家系统与人工神经网络理论,对专家系统和BP神经网络的融合方法在船用柴油机故障诊断方面的应用进行了有益的探讨,并提出了利用专家系统和BP神经网络的融合方法建立的船用柴油机神经网络故障诊断专家系统基本结构。 相似文献
7.
针对大型激光驱动器故障定位难,诊断、处置效率低的基本现状,提出利用专家系统进行故障诊断的基本思路,并以自动准直子系统为例,介绍了专家系统的体系架构、故障树模型、知识库设计和基于CLIPS工具的实现方法。初步应用结果表明,该专家系统有助于提高装置的故障诊断效率和智能化水平,增强系统可用性。 相似文献
8.
CENTUM-CS系统组态调试及维护 总被引:1,自引:0,他引:1
CENTUM -CS集散型控制系统是YOKOGAWA公司 2 0世纪 90年代中期推出的大型DCS ,集装置控制和生产管理于一体 ,采用标准的人机接口 ,开放性的网络结构 ,整个系统人机界面友好 ,稳定性高 ,系统组态简单 ,维护调试方便。自 1997年以来 ,在我厂 2套催化裂化装置、常减压装置、化纤芳烃抽提装置、PX ,PTA装置使用多套 ,提高了装置的控制水平 ,保证了安全生产。1 CS系统构成CS集散型控制系统主要由信息指令站ICS(操作站)、现场控制站FCS、通信门单元ACG、工程师站EWS、实时控制网V网以及信息局域网E网组成。典型的系统配置如图 1… 相似文献
9.
10.
运用ABC管理法控制精对苯二甲酸(PTA)装置的污水产生量,降低PTA污水单排,减少污水处理费用.ABC管理法实施后污水单排下降了0.14 t/tPTA,取得很好经济效益、环境效益和社会效益,实现了PTA装置的环保管理. 相似文献
11.
水电机组故障诊断技术是水电厂开展状态检修的基础。水电机组作为大型旋转机械,其振动可以反映机组大部分故障。专家系统是人工智能领域应用比较成熟的技术,现在故障诊断专家系统已在国内水电厂逐步推广。本文在水电机组状态监测的基础上,针对水电机组的振动故障,提出一种基于产生式规则的专家系统。它主要由知识库、推理机、数据库、人机接口等部分组成。最后,利用实际数据进行诊断和分析,并根据机组检修情况对诊断结果进行验证,证明专家诊断系统能够给检修人员提供机组故障产生的可能原因和部位,对机组检修起到一定的指导作用。 相似文献
12.
An Experimental Study on Three General Interface Layout Designs for Chemical Process Plants 下载免费PDF全文
The human‐machine interface design is not a new problem; however, existing knowledge appears to be focused either on interface conceptual design, which determines what information at what time needs to be displayed, or on interface detailed design, which determines the form of display elements. This article presents a study on interface layout design. Three interface layout designs were proposed based on the proximity compatibility priniciple, which were on the same interface conceptual design. In particular, the first interface layout design is with the highest degree of proximity cognition and the second and third with a reduced degree of proximity cognition. They are called sophisticated, semisophisticated, and nonsophisticated interfaces, respectively. An experiment with a simple process plant was conducted to understand user behaviors on the three interfaces. It is noted that in order to provide an unbiased comparative evaluation of these interfaces the same application problem and similar look‐and‐feel forms of interfaces were designed. In the experiment, three general classes of tasks were considered, namely, normal control operation, fault detection (or monitoring), and fault diagnosis. Two categories of measures were used: the performance measure and the subjective measure. The major results obtained from the experiment are 1) the nonsophisticated interface is the best for fault detections in terms of performance measure; besides, this interface has the lowest mental workload for fault detection; 2) the sophisticated interface is the best for normal operation in terms of performance measure; 3) there appears no significant difference for fault diagnosis in terms of performance measures for all these three interfaces; 4) for normal operation and fault diagnosis, the mental workloads for the three interfaces have no significant difference. Overall, the experiment suggests that the nonsophisticated interface be used in practice with additional two reasons: 1) most of time in the chemical process plant today is the fault detection or operation monitoring task, and 2) it is relatively easier for the nonsophisticated interface to adapt to changes in the chemical process plant design, which are common in today's manufacturing environments. 相似文献
13.
14.
15.
《Computers & Electrical Engineering》2014,40(7):2216-2226
In this paper, an effective strategy for fault detection of sludge volume index (SVI) sensor is proposed and tested on an experimental hardware setup in waste water treatment process (WWTP). The main objective of this fault detection strategy is to design a system which consists of the online sensors, the SVI predicting plant and fault diagnosis method. The SVI predicting plant is designed utilizing a fuzzy neural network (FNN), which is trained by a historical set of data collected during fault-free operation of WWTP. The fault diagnosis method, based on the difference between the measured concentration values and FNN predictions, allows a quick revealing of the faults. Then this proposed fault detection method is applied to a real WWTP and compared with other approaches. Experimental results show that the proposed fault detection strategy can obtain the fault signals of the SVI sensor online. 相似文献
16.
Wei-Hua Gui Chun-Hua Yang Jing Teng 《国际自动化与计算杂志》2007,4(2):135-140
According to the fault characteristic of the imperial smelting process (ISP), a novel intelligent integrated fault diagnostic system is developed. In the system fuzzy neural networks are utilized to extract fault symptom and expert system is employed for effective fault diagnosis of the process. Furthermore, fuzzy abductive inference is introduced to diagnose multiple faults. Feasibility of the proposed system is demonstrated through a pilot plant case study. 相似文献
17.
为了及时、方便地根据火电厂的报警信息诊断出故障原因,使用Visual C 和专家系统工具CLIPS开发了火电厂报警诊断专家系统.给出了该系统的总体结构并介绍了其功能.同时详细说明了汽包水位高报警诊断的实现过程.实践证明该系统界面友好,功能完善,知识库易于维护和更新. 相似文献
18.