Research On The Method Of Processing The Signal From The Vehicle-mounted Sensor Used To Measure Of The Amount Of Oil

With the rapid development of Chinese automobile industry, cars have entered ordinary family gradually. It brought great convenience to people in the trip and has a tremendous impact on human life. And because of limited natural petroleum resources and oil still being the main energy, oil price rises. It has brought great pressure to chemical production and the transportation industry. And the behavior of stealing oil and the oil spill happen often. In order to regulate the behavior of using oil, people designed a oil level monitoring system to control the amount of oil.The workflow of the oil level monitoring system is recording the height of oil surface within the drum and converting into the amount of oil by calculating. However, due to the process of the car moving having to face complex road conditions, the vibration is inevitable, as well as sudden acceleration and deceleration. All of these will generate oscillation of the oil level, which result in a large deviation between data collected by fuel sensor and the actual data. The deviation of these measurements and other interference are all regarded as noise. This paper focuses on how to reduce the noise component in the data by handling the signal collected by fuel sensor. It is expected the processed data can reflect the true amount of oil more intuitively with it becoming smoother than original.The purpose of this paper is trying to filter the oil date with using conventional filtering algorithms in the case of date having large noises component which is difficult to distinguish. Generally noise components of signal generate in the process of signal acquisition and transmission affected by internal or external reason. But in this paper, the noises are not mainly generated in this process. They mainly generate in the object whose date is collected. Since the acquisition target is susceptible and not easily measured, the date change strongly. The noise in the signal is different from the noise is known traditionally. This case happen frequently in practice. This research paper is written in this context. The paper studies to find a more practical algorithm to deal with this situation.The process of collecting data is always accompanied by a variety of interference. The noises caused by interference contains the noise which come from the external environment and the errors generated at the time of measurement. Noises generated by a variety of reasons which have different characteristics. But a variety of filtering methods developed with varying noises.First, this paper describes the classification of noises and the characteristics of various types of noises, meanwhile also introduces several common types of filtering methods and what noises they against. Then, the paper analyzes what factors influence the date collected by the fuel sensor and classify the signal noises by analysis. Finally, it uses different filtering algorithms to eliminate noises, such as FFT transform, Kalman filter and wavelet transform methods. After a comparative analysis of the filtering effect, it proposed methods of combinations and optimization of different filtering methods. It is found that signal filtering effect is better after double filtered of FFT and wavelet transform.