Development Of Ultrasonic Wind Velocity Measurement System

Wind is generally known as one of the most active meteorological elements, so its measurement is significant in fields like military, meteorology, scientific experiment, manufacture, navigation and aviation. However, the current common anemometers cannot meet the growing measuring requirements, which therefore lead to ultrasonic anemometer with simple construction, anti-vibration, rapid measurement, high accuracy, wide measuring range, and extreme-environment adaption. For oversea ultrasonic anemometer research, products with high accuracy and stability have been made within more than a hundred years’ history. On the other hand, for domestic research, only few products less than the oversea ones are produced, and oversea products are exceedingly expensive because of technological lock-in. Therefore, domestic products with good performance and proprietary intellectual property rights are urgent needs for enterprises to cut down the measuring costs and Improve competitiveness.To solve the problem and entrusted by Changchun Meteorological Instrument Research Institute, this paper focuses on the key technologies of ultrasonic anemometer and the application on actual systems. The work can be described as follows.1)Analog improved time-difference ultrasonic anemometer cannot solve the noise and disturbance problem because of the limitation of circuits, which leads to inaccurate measurement of delay in high wind speed environment. However, improved analog circuit design and production can improve measurement performance at some extent, but technologies to improve analog circuits are beyond domestic institutes’ abilities. So digital signal processing methods are chosen as the solution of restraining noise and disturbance.2)Cyclic correlation method based on cyclic statistics, aimed at ultrasonic echo signal time delay estimation,which containing gaussian noise and periodic interference, has smaller root mean square error and higher estimation precision; But when the ultrasonic echo signal contains not only Gaussian noise, but also impulsive noise, which the so-called alpha stable noise, cross-correlation and cyclic correlation method based on second-order statistics are degraded, or even failure. Therefore, the ultrasonic anemometer must request the time delay estimation algorithm are valid for alpha stable noise,gaussian noise and periodic disturbances.3)Two delay estimation methods fitting fractional-order alpha stable noise,gaussian noise and periodic interference are proposed, named ultrasonic echo signal delay estimation algorithm based on cycle covariant and normalized cycle correlation. Then they are analyzed and compared on root mean square error, accuracy, and computational complexity in low and high signal-to-noise ratio respectively. We find the algorithm based on Cycle covariant algorithm has a smaller RMSE(Root Mean Square Error) and higher accuracy then covariant algorithm, especially when the signal has lower SNR(Signal Noise Ratio) or less characteristic index of noise. But there is still some problem, such as root mean square error and the computation still large. So this algorithm can not meet the requirement of the engineering application; But the algorithm based on normalized cycle correlation perfect solution this problem, Finally considering the calculation ability of the processor is limited, and the computation of the normalized cycle correlation algorithm is still great, So fast normalized cycle correlation algorithm based on the frequency shift and fast Fourier transform is proposed. the algorithm is testified by experiments and the results show that the computational complexity is reduced by 50 times.4)First the hardware construction of the digital ultrasonic anemometer is designed and the circuit is debugged. Then the software ultrasonic echo signal delay estimation algorithm based on normalized cycle correlation is programmed. At last, the system is tested in windless environment, self-made small wind tunnel, and low-speed wind tunnel, respectively. The results testify that the algorithm we proposed meets the design requirements in this paper, and that the system’s main characteristics have reached global advanced level.