作　　者： ; ; ;

机构地区： 湖南文理学院电气与信息工程学院

出　　处： 《计算机仿真》 2005年第6期135-138,186,共5页

摘　　要： 研究了运用神经网络在既不增加设备,也不对现有仪器做任何改动的情况下增加医院现有生化测量仪器化验项目的简便方法。以对胆固醇含量估计为例的仿真结果表明该方法是有效的。该方法不但简单易行,而且处理测量数据非常高效快捷,具有投资少,见效快,效益高的显著特点。在医疗生化测量中,利用物质的光谱特性对样本溶液某种或某几种特定物质含量进行测量是一种简便快速的方法,然而光谱测量数据往往是几种待测物质含量的线性或非线性多元函数,用一般方法很难,甚至无法辨别这种复杂的函数关系,利用研究设计的多层BP神经网络成功地解决了这一难题。 A simple approach using neutral network was introduced to develop more test projects in hospitals for present biochemistry test equipments neither adding apparatus nor improving the present equipments. An example has proved its validity which simulated the estimation for the content of Cholesterol. This approach is easy to operate, fast to process data and has outstanding advantages such as lower investment and high efficiency. In biochemistry survey for medical treatment, it is a quick, simple and convenient approach to survey the content of one certain or more kinds of specific substances in the sample solution by utilizing the spectrum of substances. However, the data measured by utilizing the spectrum is often a linear or non - linear multi - dimensional function of some kinds of substances to be surveyed, and it is very difficult to deal with it by ordinary approaches, even the sophisticated relation within the function is unable to distinguish. The multi - layer BP neural network in this paper solved this difficult problem successfully.

关 键 词： 神经网络 训练 仿真 生化测量 光谱测量

领　　域： [自动化与计算机技术] [自动化与计算机技术]