Study On Condition Monitoring And Fault Diagnosis Of Cnc Machine Tool Ball Screw

In modern industry, CNC machine tool is one of the most important large mechanical equipment and the ball screw as important components of CNC machine tool feed system, characterised by high transmission efficiency, precise positioning accuracy, strong stiffness, plays an irreplaceable role in torque transmission of the machine tool and positioning in the machining process. However, the ball screw is also one of the components prone to broken-down in feed system, Once failure occurs, it will affect the machining accuracy of machine tools, and even causes the halt of the whole machine.Therefore, the research and development of efficient and reliable condition monitoring and fault diagnosis technology, guiding for the maintenance of CNC machine tools, of great significance in improving the high speed operation and high precision of CNC machine tool.This paper takes the ball screw of CNC machine tools as the research object, on the basis of the comprehensive analysis of failure diagnosis technology of ball screw pairs in research domestic and aboard and the development of no sensor monitoring technology of numerical control machine.It puts forward a method that receiving internal information from sensorless monitoring method can be used for the fault diagnosis of CNC machine ballscrew,and compare analysis the obtaining external information with external sensors for traditional monitoring method.Experiments were designed aiming at the four state of the ball screw and that is operating well, raceway wear, ball damage and screw bending. Besides, mixed programming based on Lab VIEW and MATLAB was performed and eventually the model of ball screw intelligent diagnosis on the foundation of BP neural network and PNN neural network was established and identifying the state types of the ball screw pair of CNC machine tools was studied.Based on the analysis of the servo feed system of CNC machine tool components and internal sensor function,it renders the concept of the state information about ball screw in servo feed system of CNC machine tool and studies the methods of obtaining state information of the ball screw pair.Based on the structure of ball screw, it analyzes the fault mechanism and the vibration signal characteristicsand determines the approach for ball screw normal, raceway wear, ball damage, and screw bending, the four types of condition monitoring and fault diagnosis.The test scheme of the ball screw and the test system of overall architecture are designed and the test bed of state information monitoring is built.Based on the software platform Lab VIEW and SIEMENS STARTER, using the traditiona lmonitoring and sensorless monitoring methods the data acquisition system, of the ball screw pair state information is established, and more detailed design aiming at the hardware and software of data acquisition is promoted.The data is processed and analyzed and the ball screw pair state information collected by the two signal acquisition methods is analyzed by time domain, the frequency and time frequency domain, obtaining the initial feature.Through kernel principal component analysis method, the initial feature is processed by vector of dimensionality reduction, reduction treatment, and 33 group and 27 group the combination of effective characteristic quantity of each state type ball screw pair are finally determined.The BP neural network and PNN neural network to two network models of two kinds of methods of monitoring data obtained by the intelligent pattern recognition is established. Comparing the analysis of two kinds of network model and the recognition performance by the ball screw pair state information of the two kinds of monitoring methods,the results show that, the fault diagnosis method of ball screw based on without sensor information and advantagous PNN neural network can complete the fault diagnosis of ball screw pairs effectively.