Water Sound Recognition Based on Support Vector Machine

Water flow monitoring is of importance for water dynamic controlling in river basins. Currently, most of the water monitoring in the basin is manual-based, there is often some disordered data, which does not conform to hydrological models input standards. In order to solve the above problems, we propose a sound signal processing method to identify the water sound and link it with stream-flow measurements in this paper. Firstly, the water sound features were extracted. On this basis, Support Vector Machine (SVM) and two other classifiers are used to build the classification models, and the model tested using trained and not trained data show the best agreement to identifying the water sound recognition. The best performance of the classification is given by further optimizing the kernel function and penalty factor. The experimental result shows that the model's recognition rate is 98.22% with SVM, and the result of recognition is superior to other classification models.