【作者】 徐平；

【导师】 郑崇苏；

【作者基本信息】 福州大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 红外水分仪的水分检测技术是一门涉及多学科的综合性技术，三十多年来，已经取得了很大的进步。但由于被测信号十分微弱，测量过程干扰因素较多，尤其是温度，而现有的红外水分仪在信号处理方面往往只是进行简单的滤波与温度补偿，致使仪器在测量精度、适应环境、抗干扰性能、调试与维护等方面还不尽人意。针对以上存在的问题，本课题对红外水分仪的光路系统、硬件电路、系统软件进行了全面设计，还引入小波软阈值滤波与信息融合技术，在智能化信息处理方面进行了一些探索和尝试，为新型红外水分仪的数据处理提供了新的思路。本文的主要内容和主要成果可归纳如下：1. 介绍了红外水分仪的基本组成和工作原理。对系统的光路和硬件电路进行了详细的设计与装配，编写了系统软件程序，并进行调试。2. 制作神经网络训练所需的标准样本。3. 阐述了小波去噪的相关理论，探讨了小波软阈值去噪的几种方法。4. 阐述了信息融合的基本原理以及信息融合的几种方法，具体介绍了基于神经网络的信息融合技术。5. 针对红外水分仪出来的信噪比相对较低的微弱信号，利用MATLAB小波分析工具包进行了小波软阈值滤波的可行性仿真之后，编写C语言程序进行实际实现。6. 针对红外水分仪的水分传感器受温度影响很大的情况，利用BP神经网络进行了多传感器信息融合，用MATLAB进行仿真实验，融合后的效果较为理想。更多还原

【Abstract】 The moisture measurement technology of infrared moisture meter is a synthetic technology relating to many subjects. During the past 30 years, much prominent progress has been achieved. However, many influence factors especially the temperature have influence on the output signal of infrared moisture meter which belongs to the weak signal, and the existing infrared moisture meter have some defects such as just carrying simple filter and temperature compensation on it, so the measurement precision, the capability of adapting the environment and anti-noise are still not satisfying, as well as its performance and maintenance. In view of the above, the wavelet soft threshold value filter and the information fusion technology are introduced; some explorations and attempts are made on processing the information intelligently, which give a new data processing basis of the infrared moisture meter. The main content and achievements can be summed up as follows:1. The basic components and operation principle of infrared moisture meter, designing and assembling the light path and hardware circuit of the system are introduced in detail, the program is compiled and debugged.2. Standard samples that are needed for neural network training are made.3. The theory about the wavelet denoising method is introduced, as well as the methods of wavelet soft threshold selection.4. The basic concepts and several ways about information fusion are illustrated, especially the information fusion technology based on neural network.5. Considering the signal’s SNR is relatively low and it belongs to weak signal, some feasible simulations of the wavelet soft threshold filter with MATLAB wavelet analysis tool packet are done, and the C program is compiled and executed.6. The temperature’s influence on the moisture sensor is very obvious, so bi-sensors information fusion techniques based on BP neural network is<WP=4>proposed, and be simulated with MATLAB, the effect of it is ideal.更多还原