Continuous and dynamic measurements of human respiratory parameters are very important for vital diseases of respiratory system during mechanical ventilation. This paper analyzed the structure and me-chanical properties of the human respiratory system, and designed an active intervening monitoring micro system for it. The mobile mechanism of the micro system is soft and earthworm-like movement actuated by pneumatic rubber actuator, the measurement and therapy unit of the system is an extensible mechanism with sensors in the front. The micro monitoring system can move in respiratory tract and measure the res-piratory parameters in bronchium continuously. Experiments had been done in swine's respiratory tract, the results proved that the micro robot system could measure the respiratory parameters in real-time suc-cessfully and its movement was smooth in swine's respiratory tract.

为提高脑机接口中脑电识别率，分析了特征提取方面时频特征组合法的缺点，探讨了一种改进的模式识别方法。该方法以样本类平均距离为判据，采用滑动窗优化技术，获取时域均值的最佳时间段和频域功率谱均值的最佳频率段。用经过优化的时域均值和功率谱均值组合作为特征，形成特征向量。基于该特征向量，用神经网络对脑电信号进行分类。以识别正确率为指标，将改进方法与原方法进行对比，实验结果表明改进方法能够提高脑电识别率，具有应用价值。

AIM: To propose a new, non-invasive method for monitoring 24-h pressure, temperature and pH value in gastrointestinal tract. METHODS: The authors developed a miniature, multifunctional gastrointestinal monitoring system, which comprises a set of indigestible biotelemetry capsules and a data recorder. The capsule, after ingested by patients, could measure pressure, temperature and pH value in the gastrointestinal tract and transmit the data to the data recorder outside the body through a 434 MHz radio frequency data link. After the capsule passed out from the body, the data saved in the recorder were downloaded to a workstation via a special software for further analysis and comparison. RESULTS: Clinical experiments showed that the biotelemetry capsules could be swallowed by volunteers without any difficulties. The data recorder could receive the radio frequency signals transmitted by the biotelemetry in the body. The biotelemetry capsule could pass out from the body without difficulties. No discomfort was reported by any volunteer during the experiment. In vivo pressure and temperature data were acquired. CONCLUSION: A non-invasive method for monitoring 24-h gastrointestinal parameters was proposed and tested by the authors. The feasibility and functionality of this method are verified by laboratory tests and clinical experiments.

阐述了电磁激励式漏磁通法检测煤气管道壁厚及缺陷的原理，介绍了检测装置的机械组成、电气组成以及装置的工作原理和技术特点，并对装置的控制流程和软件设计作了详细说明。该装置已研制成功，实测结果表明装置有较高的分辨率。

The computation algorithm of knot point planning for Cartesian trajectory generation of manjpulator is investigated. A novel inheritance bisection algorithm (IBA) based on conventional bisection algorithm (BA) is proposed. IBA has two steps. The first step is the 1st knot point planning under lower ste position accuracy; the sccond step is the 2nd knot point planning that inherits the results of the 1st planning under higher set position accuracy. The simulation results reveal that the number of inverse kinematical calcultaion (IKC) CAUSED BY IBA is decreased compared with BA. IBA is more efficient to plan knot points.

介绍了目前国际上常用的体内目标的定位方法即放射性同位素扫描和电磁场定位，分析了各自的优缺点，在此基础上提出利用目标体内密封的永久磁铁为磁标记物，在体表处检测多点磁场强度，并采用遗传算法解决非线性方程组的求逆问题从而获得胶囊在体内的位置和方向信息，实现目标在体内的连续定位。仿真实验表明，该方法定位精度在毫米级，完全满足实际应用需求。

脑机接口（brain-computer interface，BcI）利用脑电信号实现人脑与计算机或其他电子设备的通讯和控制，是一种新的人机接口方式。它在康复医学和控制工程等领域有应用前景。本文介绍了BCIs系统的工作原理，从系统设计、数据获取及处理方法两方面论述了BCIs系统设计中的关键技术，最后指出了BCIs存在的主要问题和发展趋势。这些探讨为BCIs的设计与研究提供了指导。

A new type of in-pipe mobile robot was designed and developed on the basis of wheel in-pipe robot and crawler in-pipe robot. The three sets of driving wheels circumferentilly 120 apart in the cross section, both front and rear driving wheels are distributed on the same parallelogram mechanism. The driving motor drives the three sets of driving wheels by worm couple, the regulating motor makes the three sets of drivin wheels push against the pipe inwall with stable and adequate pressig force by the ball screw pair and pressure sensor, so the in-pipe robot can provide adequate and stable traction force. The robot mechanism is simple and small in size and work reliably. It is particularly suitable to the pipe with diameter 400-650mm.

This paper described the structrue and control of a new kind of miniature hexapod bio-robot, analyzed the moving principle of the robot. The robot is based on the principle of bionics, its structure is simple, design novel, unique. It can move forwards and backwards. The external dimensions of bio-robto is: length 30 mm, width 40 mm, height 20 mm, weight 6.3g. Some tests about the model robot were made. The experimental results show that the robot bas mobility.

针对脑机接口（BCI）研究中存在脑电信号（EEG）识剐率低的问题，提出一种基于遗传算法（GA）和概率神经网络（PNN）的GA-PNN识别方法。用该方法对EEG提取时频特征，构成模式识别的初始特征。以训练样本识别正确率为适应度函数，采用GA对初始特征进行组合优化。基于优选后的特征，用PNN对测试样本进行分类。该方法使EEG识别正确率达到92.49%，与2003年BCI国际竞赛最好的处理结果（88.7%）相比，提高近4%，为BCI中EEG的识别提供了有效的手段。