目 錄

中文摘要 Ⅰ
ABSTRACT Ⅱ
引言 1
1緒論 3
1.1 設(shè)備狀態(tài)預(yù)測的意義 3
1.2 神經(jīng)網(wǎng)絡(luò)理論國內(nèi)外發(fā)展現(xiàn)狀 3
1.3 人工神經(jīng)網(wǎng)絡(luò)在預(yù)測領(lǐng)域的發(fā)展 5
2 人工神經(jīng)網(wǎng)絡(luò)及其應(yīng)用于狀態(tài)預(yù)測原理 7
2.1 人工神經(jīng)網(wǎng)絡(luò) 7
2.1.1人工神經(jīng)網(wǎng)絡(luò)的特點…………………………………………………………………8
2.1.2 人工神經(jīng)網(wǎng)絡(luò)的特性…………………………………………………………………8
2.1.3人工神經(jīng)網(wǎng)絡(luò)的模型…………………………………………………………………9
2.1.4人工神經(jīng)網(wǎng)絡(luò)的轉(zhuǎn)移函數(shù)………………………………………………………… .13
2.1.5人工網(wǎng)絡(luò)的學(xué)習(xí)……………………………………………………………………..14
2.2 基于人工神經(jīng)網(wǎng)絡(luò)的預(yù)測問題 15
2.2.1設(shè)備狀態(tài)預(yù)測問題概述…………………………………………………………… .15
2.2.2設(shè)備狀態(tài)預(yù)測常用方法…………………………………………………………….17
2.2.3應(yīng)用神經(jīng)網(wǎng)絡(luò)對設(shè)備狀態(tài)預(yù)測問題進行預(yù)測建模………………………………..19
3 設(shè)備狀態(tài)預(yù)測的實現(xiàn) 23
3.1 人工神經(jīng)網(wǎng)絡(luò)及BP網(wǎng)絡(luò)學(xué)習(xí)規(guī)則介紹 23
3.1.1 人工神經(jīng)網(wǎng)絡(luò)的一般學(xué)習(xí)規(guī)則 23
3.1.2 BP算法概述 24
3.1.3 BP網(wǎng)絡(luò)的學(xué)習(xí)算法 25
3.2 建立于BP網(wǎng)絡(luò)的預(yù)測模型策略 26
3.3 用MATLAB7.0實現(xiàn)BP網(wǎng)絡(luò) 28
3.3.1 MATLAB簡介 28
3.3.2 MATLAB神經(jīng)網(wǎng)絡(luò)工具箱 28
3.3.3 BP算法程序流程 29
4 基于BP網(wǎng)絡(luò)的狀態(tài)預(yù)測算例訓(xùn)練 33
4.1 狀態(tài)分類器 33
4.2 BP網(wǎng)絡(luò)設(shè)計 34
4.3 小結(jié) 37
5 基于BP網(wǎng)絡(luò)的狀態(tài)預(yù)測算例驗證 38
5.1 大型機械系統(tǒng)的劣化模型 38
5.2 算例介紹及BP網(wǎng)絡(luò)預(yù)測壓縮機綜合劣化度……………………………………………39
5.3 BP神經(jīng)網(wǎng)絡(luò)的算法……………………………………………………………………. 40
5.4 本章小結(jié)………………………………………………………………………………… 42
6 總結(jié) 43
致謝…………………………………………………………………. 44
參考文獻……………………………………………………………….45



摘要

隨著現(xiàn)代工業(yè)技術(shù)的快速發(fā)展，機械設(shè)備在工作中的運行狀態(tài)越來越受到關(guān)注。人工神經(jīng)網(wǎng)絡(luò)作為一種新興的交叉技術(shù)，被應(yīng)用于機械設(shè)備狀態(tài)預(yù)測問題之上并取得了長足的進步?；谌斯ど窠?jīng)網(wǎng)絡(luò)的設(shè)備狀態(tài)預(yù)測技術(shù)能在降低維修成本，制定生產(chǎn)規(guī)劃上發(fā)揮重要作用。
本文介紹了設(shè)備狀態(tài)和趨勢預(yù)測的研究現(xiàn)狀，提出了計算原理及其改進和實現(xiàn)的過程。研究目的是針對設(shè)備狀態(tài)問題建立基于人工神經(jīng)網(wǎng)絡(luò)的在線實時觀測模型。本文首先闡述了人工神經(jīng)網(wǎng)絡(luò)的基本工作原理，如學(xué)習(xí)規(guī)則、權(quán)值、網(wǎng)絡(luò)結(jié)構(gòu)等等，然后引進具體的BP網(wǎng)絡(luò)模型，介紹了其學(xué)習(xí)規(guī)則和基于此網(wǎng)絡(luò)的預(yù)測原理和模型，并介紹了MATLAB7.0及其對BP網(wǎng)絡(luò)的實現(xiàn)過程，最后是全文工作的總結(jié)。
在評述和分析了幾種常見的剩余壽命的方法基礎(chǔ)上，提出了相對劣化度的概念，并以此建立基于人工神經(jīng)網(wǎng)絡(luò)的大型機械系統(tǒng)的剩余壽命的預(yù)報模型。
由本文可得出的結(jié)論：基于人工神經(jīng)網(wǎng)絡(luò)的設(shè)備狀態(tài)預(yù)測模型考慮了影響機械設(shè)備狀態(tài)的多重因素，能夠作出多步預(yù)測，它的學(xué)習(xí)特性能夠以較好的穩(wěn)定性和較高的精度模擬機械狀態(tài)中輸入輸出間的非線性關(guān)系，人工神經(jīng)網(wǎng)絡(luò)在設(shè)備狀態(tài)預(yù)測中的應(yīng)用和發(fā)展前景是非常廣闊的。
關(guān)鍵詞：設(shè)備狀態(tài)預(yù)測，人工神經(jīng)網(wǎng)絡(luò)，BP網(wǎng)絡(luò)，MATLAB7.0，壽命預(yù)報，非線形




ABSTRCT

With the development of modern industry, the working state of mechanical equipment now is attracting more and more attention. Artificial neural network, as a rising interdisciplinary technology, is applied universally in equipment state prediction field. Prediction of equipment state based on artificial neural network now plays an important role in reducing maintaining cost and establishing product plan.
This paper introduces the research actuality of equipment state prediction and trend estimation. Moreover, it puts forward computing method and its improvement and realization process in mechanical field. The aim of this project is to establish the state model of equipment. First, the essential working principles of ANN are introduced, such as training rules is investigated and architecture, then the concrete BP model is introduced, whose training rules is investigated, and the predictive model based on BP algorithm is described and then its realization in Matlab7.0 is studied, finally the summary of this paper is given
A residual life prediction model of large machinery system based on artificial neuralnet was developed.
We draw the conclusion that the predictive model based on artificial neural network can take many factors affecting the machinery condition into account and make multi-step prediction. Its intelligent feature is with good stability and high sensitivity for simulating nonlinear I/O relationship in mechanical behavior. Its application foreground and developing future are quite broad.

Keywords: Prediction of equipment state, Artificial neural network, BP network, MATLAB7.0, Residual life prediction, Nonlinear.