基于人工神经网络烟度计有效光路的算法  

Algorithm for the Effective Optical Path of Smoke Meter Based on Artificial Neural Network

作　　者：杨礼河[1] 孙玉德[1] 解世文[1] 李志军[2] 

机构地区：[1]天津大学内燃机研究所,天津300072 [2]天津大学机械工程学院,天津300072

出　　处：《天津大学学报》2006年第9期1072-1076,共5页Journal of Tianjin University(Science and Technology)

基　　金：国家自然科学基金(50276042).

摘　　要：为准确计算烟度计的有效光路,基于人工神经网络提出了对QN-1型全流消光式烟度计有效光路的计算方法．该方法利用烟度计有效光路的关键影响因素即空气扫气压力和发动机的转速为输入,构造了一种反向传播神经网络,并利用试验数据进行训练,然后仿真烟度计的有效光路．试验表明,仿真数据与试验数据偏差在-0．0399～0．0486之间．由此可见,利用人工神经网络计算烟度计的有效光路准确实用．To calculate the effective optical path of smoke meter accurately, an algorithm for the effective optical path of QN-1 total flow extinction type smoke meter was proposed based on artificial neural network in this paper. Taking scavenging pressure of air and the engine rotation speed as inputs which were the main factors affecting the effective optical path of the smoke meter, the back-propagation network was constructed. By training the testing data, the network was used for simulating the effective optical path of the smoke meter. The results indicate that the deviations between the simulating and testing data range from - 0. 039 9 to 0. 048 6. In conclusion, the effective optical path of the smoke meter can be calculated by the artificial neural network exactly and practically.

关 键 词：人工神经网络 烟度计 有效光路 扫气 压力 算法 

分 类 号：TK16[动力工程及工程热物理—热能工程]