The Study On Weak Signal Detection And Its Application In Accurate Measurement Of Torque

As an important parameter to reflect the assembly quality of thread fastener, the torque is of great significance to guarantee the safe operation and improve the service life of the equipment. In engineering practice, torque signal is difficult to be processed by traditional filtering methods because it is very weak, vulnerable to external noise and low SNR. As a sophisticated technique which is used to detect noise drowned weak signal, weak signal detection is widely used in geological exploration, industrial control, aviation and space, machine design and manufacture and other fields. The relevant instrument has become indispensable equipment in modern scientific research and industrial production. By the study on weak signal detection theory and method and by combining with the characteristics of the signal, this thesis effectively improved traditional method for de-noising and then applied it to actual detection of weak torque signals. Finally, a new high precision measurement is developed to improve torque measurement accuracy.This thesis first briefly discusses the features of torque measurement and research significance. On this basis, the current development and research status of weak signal detection and the torque sensor are analyzed.Secondly, the principle of strain torque measurement is summarized. Aiming at the resistance strain torque sensor which is the widely used in torque measurement, the features of output signal of the sensor and the source and regularity of noise are studied, and the torque signal model that accords with the actual is given. This thesis focuses on the principles and methods of wavelet threshold de-noising algorithm. On this basis, an improved weak signal detection algorithm, which is verified its feasibility and effectiveness to the torque weak signal detection by a simulation with MATLAB, is put forward.Finally, a set of torque measurement system is designed. It adopts the resistance strain torque sensor to collect the weak signal when the torque changes. This signal is amplified and filtered to eliminate the high frequency interference by signal conditioning circuit and then converted into digital signal through 24 bit high precision Sigma-Delta analog to digital converter ADS7192. Then, this digital signal is delivered to STM32F4 digital signal controller based on ARM Cortex-M4 core to process. This system uses ARM-MDK programming to realize wavelet threshold de-noising and calculate the torque value, and then sends the value to LCD display. Furthermore, this torque value could be transmitted wirelessly for remote data record and analysis. For the convenience of users and improve the reliability, this system allows setting torque measurement conditions. If the torque value reaches or exceeds the limit value, this system can send out audible and visual alarm.Measured results show that the accuracy of this torque measurement system designed in this thesis up to 0.6% F.S., which realizes the accurate measurement and real-time display of the torque. In addition, the system takes on low power consumption, stable and reliable. The research results of this thesis not only can be used in the tool calibration, screw assemblies, security detection and other actual torque detection field, but also have a certain theoretical and practical significance to the weak signal detection and hardware implementation in the future.