本文選題：報警管理　切入點：小波分析　出處：《青島科技大學》2015年碩士論文　論文類型：學位論文

【摘要】：報警系統構成了大型現代化工裝置操作界面的一個重要組成部分,在預防、控制和減輕異常情況方面具有重要的作用。因此,報警的優(yōu)化以及報警處理策略對化工裝置的正常運行具有十分重要的意義。精餾塔是化工企業(yè)生產過程中應用廣泛的傳質設備,在化工行業(yè)有著很重要的地位。精餾塔的安全、平穩(wěn)運行己經成為了化工生產的重要環(huán)節(jié),也直接關系到化工企業(yè)的經濟效益。本文針對精餾塔DCS報警信息冗余問題,進行了報警優(yōu)化以及報警根原因分析研究。本文首先對報警系統關鍵性能指標進行了統計分析,并對報警系統的性能進行了評估,為后續(xù)報警系統優(yōu)化做好了準備。針對重復報警,在考慮死區(qū)的基礎上,引入了基于數據過濾的報警限優(yōu)化設計。數據過濾的引入有效地解決了過程數據受噪聲干擾的不足,增強了魯棒性。死區(qū)和數據過濾的結合,有效地解決了重復報警的問題。但是,僅根據測量信號進行報警優(yōu)化是十分有限的,只有找到報警根源才能徹底解決報警的問題。故障診斷策略也是處理報警的一種有效方法,只要將故障消除,報警也會隨之消失。但是,直接將化工機理模型用于故障診斷還存在計算量大、收斂性不好的問題。因此,本文研究了基于雙層法模型的故障檢測和診斷方法。使用非線性模型監(jiān)控精餾過程和識別大的測量值偏差時的異常情況,通過基于線性模型的線性最小二乘法估計精餾塔內部的故障參數。為了進一步縮小模型計算范圍,還運用希梅爾布勞算法對精餾系統進行分解,將高維的、不易求解的問題分解成若干低維的子問題。然后對分解后的小系統采用基于中值濾波和提升小波分析去噪的方法對數據進行去噪處理,增強了報警優(yōu)化的魯棒性。最后,將上述報警優(yōu)化策略應用到TEP仿真流程的汽提塔中。案例研究結果表明,雙層診斷結構有效地獲得了故障參數的變化情況,比單純的非線性模型結構更有效。還給出了因物流C的壓差推動力發(fā)生階躍下降導致的故障7時的塔底進料損失參數的變化情況。結果表明,該方法不僅可以準確地給出故障原因還可以大大縮短報警診斷的時間。
[Abstract]:The alarm system constitutes an important part of the operating interface of large-scale modern chemical plants and plays an important role in preventing, controlling and mitigating anomalies. The optimization of alarm and the strategy of alarm processing are very important to the normal operation of chemical plant. Distillation tower is a mass transfer equipment widely used in the production process of chemical enterprises and plays an important role in the chemical industry. Smooth operation has become an important part of chemical production, and also directly related to the economic benefits of chemical enterprises. This paper aims at the redundancy of DCS alarm information in distillation tower. The alarm optimization and the cause analysis of the alarm root are carried out. Firstly, the key performance indexes of the alarm system are statistically analyzed, and the performance of the alarm system is evaluated. For the repeated alarm, considering the dead zone, the optimal design of alarm limit based on data filtering is introduced. The introduction of data filtering can effectively solve the problem of process data being disturbed by noise. The combination of dead zone and data filtering can effectively solve the problem of repeated alarm. However, it is very limited to optimize the alarm only according to the measured signal. Only by finding the root cause of the alarm can the problem of alarm be solved completely. The fault diagnosis strategy is also an effective way to deal with the alarm. As long as the fault is eliminated, the alarm will disappear. However, The direct application of chemical mechanism model to fault diagnosis still has the problems of large computation and poor convergence. In this paper, the fault detection and diagnosis method based on the two-layer model is studied. The nonlinear model is used to monitor the distillation process and identify the abnormal situation of the large deviation of the measured value. The linear least square method based on the linear model is used to estimate the internal fault parameters of the distillation tower. In order to further reduce the calculation range of the model, the Simmel Braugh algorithm is also used to decompose the distillation system. The problem that is difficult to solve is decomposed into some sub-problems of low dimension. Then, the decomposed small system is de-noised based on median filter and lifting wavelet analysis, which enhances the robustness of alarm optimization. The above alarm optimization strategy is applied to the stripper of TEP simulation flow. The case study shows that the two-layer diagnosis structure can effectively obtain the variation of fault parameters. It is more effective than the simple nonlinear model structure. The variation of the parameters of the bottom feed loss caused by the step drop of the pressure differential driving force of the logistics C at 7:00 is also given. The results show that, This method can not only give the cause of fault accurately, but also shorten the time of alarm diagnosis.
【學位授予單位】：青島科技大學
【學位級別】：碩士
【學位授予年份】：2015
【分類號】：TQ053.5

【參考文獻】

相關期刊論文 前4條

1 潘全文,艾弘飛,房振旭;專家系統的基本原理和基于CLIPS的專家系統設計與實現[J];飛機設計;2004年04期

2 耿志強,朱群雄,顧祥柏;現代化工流程報警系統優(yōu)化算法及應用[J];計算機與應用化學;2005年03期

3 安維中,胡仰棟,袁希鋼;最優(yōu)化技術在精餾過程綜合中的應用及研究進展[J];計算機與應用化學;2005年05期

4 高東;張貝克;吳重光;;基于提升格式的過程數據在線去噪方法及其應用[J];計算機應用研究;2008年10期

，

本文編號：1567984