航天水介质轴承在水环境进行高速寿命试验的过程中，轴系在充满水的轴承座空腔中不断搅拌，造成轴系功率损耗较大，对系统的驱动设备、安全保障设计具有很高要求。研究航天水介质轴承需要一套能够模拟实际工况、集参数研究和产品抽检试验为一体的测控系统试验台。 论文首先利用EPLAN电气绘图软件设计了控制系统的电气接线图。利用以太网和PROFIBUS-DP总线建立通信网络，控制系统上位机通过WinCC组态软件编写监控界面，应用STEP7软件编写S7-300 PLC主站程序，控制电机转速并采集电机参数。同时，设计了WinCC Flexible触摸屏界面和电流串联采集电路，控制从站远程驱动气源，实现了高速水介质润滑加载试验台的设计。 然后，根据试验台结构和被试轴承试验原理，设计了具有传感器信息采集和记录功能的NI数据采集系统。根据工艺要求设计了试验器测点，利用LabVIEW图形化编程语言编写了上位机采集程序，通过OPC Sever建立了基于以太网的共享变量标签与控制系统PLC主站进行数据交换，并利用倒频谱分析了轴承振动信号。实现了轴承试验台工作参数的实时采集、处理、显示、记录等功能。 最后，介绍了粗糙集理论并分析了粗糙集在故障诊断中进行数据挖掘的可行性，应用Rosetta软件针对轴承性能试验台的故障试车数据进行属性约简，去除不相容的冗余因子，得到最小属性集合和约简规则。在MATLAB中对约简参数得到的样本数据进行分批训练、测试，建立BP神经网络模型，得到较高准确率的分类结果，证明了粗糙集属性约简有助于提高BP神经网络分类效果的结论，缩减了系统故障排查的时间，完善了轴承性能试验台测控系统。

When doing high-speed-life test for the aqueous medium aerospace bearings in the aqueous environment, the cavity of the bearing seat, in which installing the bearing system with a high rotating speed and leading bearing system losing large power, is almost full of water. Therefore a good quality of drive device and high safety factors play important roles in the bearing system. It is very important to study on the aerospace bearings needing a test rig with control and data- acquisition functions of simulating actual condition parameters while studying parameters and testing products. The paper, first of all, uses the EPLAN, a software which can design electrical control system, to design subsystems of control system including motor control and lubrication system, high and low pressure water-supply system, loading system and monitoring system. By utilizing the Ethernet and PROFIBUS-DP bus, the system has established a communication network for the monitoring interface, which is designed by WinCC configuration software in the control system, of the position machine. And then, control the speed of motor and acquire it’s data by the transporting control commands from S7-300 PLC master successfully using STEP7 software. At the same time, a current series acquisition circuit is devised to control the gas source distantly while the touch screen interface being designed with WinCC Flexible to achieve a test rig design of high-speed water lubricated environment and load. Secondly, the sensor information system of collecting and recording in NI data acquisition system is designed according to the structure of tester and bearing test principle. The tester measuring point has been planned according to the test process requirement and collected by the master making use of acquisition program in LabVIEW, a software in graphical programming language, while establishing OPC share variable labels based on the Ethernet to achieve the data exchange for the PLC master of control system. The data of the bearing vibration signal from acceleration sensor is gathered by the position computer of data acquisition system and analyzed by cepstrum to record the working parameters the same time of the bearing test rig. Finally, the paper describes the core of Rough Set theory, and analysis of the feasibility of using Rough Set in fault diagnosis for data mining. Aiming at the collected test data of bearing performance test bed, the software Rosetta is exploited to apply attribute reduction and removal of incompatible factors, resulting in a minimum set of attributes and reduction rules. To set up the BP neural network model, the sample test data after attribute reduction have been trained and tested in MATLAB. In this way a higher classification accuracy rate has been obtained proving a conclusion that attribute reduction of Rough Set can improve BP neural network classifier.