Application Of Wavelet Principle On Acoustic Leak Signal Detection

Fast Reactors are that fast neutrons cause nucleus fission chain reaction. Their main characteristics are that they are enriching Uranium during the fission chain reaction which allows an operational fast reactor to generate more fissile material than it consumes. So they are also called fast breeder reactor. In order to relieve the contradiction of energy shortage and irregular distribution, and achieve on harmonious development between human and the nature, China has established a strategic target that we should realize the commercialization of fast reactor nuclear power system around the year of 2035. We must strive to achieve this goal in thirty or forty years. We should realize the technical breakthrough in each link and form China's nuclear power industry of fast reactor gradually, thus solving the issues for China's sustainable development on energy.In development of the fast reactors, the steam generator is one of the most important equipments. As a result of the problems of the material, the manufacture and the processing, it's difficult to avoid that the water in steam generator and the steam in pipe may leak. Water and sodium are two kinds of substances that have high chemical affinity. Once the leak happens, sodium and water (water steam) will break out a violent chemical reaction, it will release a large amount of heat, which results in the breadth and extension. So it's one of the chief dangers which seriously affect the motion of the power station. Consequently, detecting the leak in time and taking relevant and safe measures to control the extension are the significant conditions which should guarantee the fast reactors to operate safely. In fact, developing the leak detection technology has become on of the main measures for the fast reactor power station moving safely.The paper theoretically analyzes the mechanism of water leak signal, background noises and the transmission characteristics in the steam generator (SG) of a Liquid Metal Fast Breeder Reactor (LMFBR). The improvement of measurement section in the acoustic leak detection system and the experiments on sodium loop are introduced. Through the massive data analysis and processing, the paper has studied the signal characteristic of the water leak and the leak signal influence of every kind of factor, as well as every kind of background noise time domain and frequency range. In this foundation, the paper introduces the wavelet analysis to the water/steam acoustic leak detection system signal processing of the fast reactor and proposes the leak extracting method which based on wavelet/wavelet packet transformation energy and rectangular characteristic. It analyzes the processing results and gives some proposals to further research work.The following are the specific contribution and research results:1. Acoustic leak detection technology in the steam generator (SG) of a Liquid Metal Fast Breeder Reactor (LMFBR) is reviewed comprehensively. Study on tendencies of domestic research, the problems and tendency of development are discussed. This paper introduces the experimental equipments and methods of experimental system in sodium loops, and analyzes the systemic noise generator and the communication mechanism. Through the analysis of leak signal, and time domain and frequency domain in background noise, it carries out an effective test on stability, periodicity and normality for random signals. Effective research on acoustic leak system at the noise characteristics of sensitivity, response time of system, the influence of sodium temperature to noise, current, flow, pump and blower, which supports the foundation for which uses correct methods to deal with analysis and the data.2. It uses singularity detection of wavelet transform, theory of wavelet transform maximum modulus and principles of wavelet noise elimination to make fault detection for acoustic leak of fast reactor steam generator. So it presents these leak detection methods, which based on singularity detection of wavelet transform, features of maximum modulus and theory of wavelet noise elimination. The findings indicate that the signal singularity detection of wavelet transform can carry on the accurate time localization at the high temperature (350℃) when the leak signal happens the sudden change. The method can also judge whether the leak occurs at the low temperature (250℃), but the time localization at the low temperature is not accurate as the high temperature. When the theory of wavelet transform maximum modulus is introduced to the leak signal detection, it discovers that maximum modulus contains background noise and maximum modulus of leak signal when original signals are transformed by wavelet. Through the study of the difference of their maximum modulus and the position of maximum modulus, the leak can be examined effectively, by using the location and numerical value of maximum modulus to judge the sudden change degree of the signals. Wavelet noise elimination can enable the signal noise ratio to improve extent, is also a very effective method to assess leak of steam generator promptly and accurately.3.The leak detection method is put forth and based on of wavelet transforming energy characteristic in this research. The experiment discovers that the structure of signal spectrum has changed when the leak happens. Using multi-resolution of wavelet transform to put the signal onto different criterion (frequency band), which will decompose the signal energy, it forms characteristic vector which can distinguish occurrence of the leak. Energy characteristic vector may distinguish the leak information and distribution from wavelet transform on different frequency channel.In different sodium temperature, the system possessed different normal characteristic vectorγnormal. Different leak rate has different leak characteristic vectorγleak, but the change rule ofγleak along with the leak rate is not obvious. We can use the average ofγleak andγnormal to distinguish the leak, but the size recognition needs to be carried on further analysis.4. The research proposes the leak detection method which is based on wavelet packet transformation. After the wavelet packet analysis is introduced to the leak signal processing, energy characteristic vector of wavelet packet can directly show the change of energy proportion of normal signal and leak signal in various frequency bands. It not only can distinguish the characteristic frequency band, but also can provide the reference with the leak degree. From the energy characteristic vector of wavelet packet normalization, we can gain the characteristic frequency band of leak signal is below 6 kHz when sodium reacts with water. On this characteristic frequency band, the coefficient rectangular characteristic of wavelet packet had good effect on leak distinction.The paper proposes the leak detection and extracting methods are based on wavelet transform, which can successfully detect the leak rate from 5 ml/h to 400 ml/h at high temperature and over 100 ml/h at low temperature, and can orient accurately for leak time, surpassing the traditional signal processing method. In order to enhance the detection reliability at low temperature and carry out the research on leak rate recognition, more repetitions experiment should be done at the same operating mode to gain the more accurate data. About setting the threshold for leak distinction, the available experimental data may satisfy the need of sodium return. But for actual Liquid Metal Fast Breeder Reactor, we should gather the actual background noise and the experimental background noise to compare, and at last determine the reasonable threshold to distinct leak.