Urban noise identification method of multilayer random neural network based on quantization error entropy

Abstract

The invention discloses an urban noise identification method of a multilayer random neural network based on quantization error entropy. The method comprises the following steps of 1, processing a collected sound signal, performing feature extraction, and establishing a sound database; 2, introducing the minimum quantization error entropy into an encoder of the multi-layer random neural network toserve as a training criterion of encoder output weight, and learning features in data; 3, determining a network structure and parameters of the multilayer random neural network, and taking data in theestablished sound database as training data of the multilayer random neural network; 4, introducing a non-Gaussian noise model; and 5, generating non-Gaussian noise data, adding the non-Gaussian noise data into the original data, and then putting the original data into the model for training. The random neural network framework which is simple in algorithm, free of reverse iteration, high in calculation speed and high in generalization ability is adopted, a QMEE algorithm is added to the coding layer of the model to serve as the training criterion of the coding layer, and noise recognition ismore accurate.

Description

technical field [0001] The invention relates to an urban noise recognition method based on quantization error entropy, and relates to various technical fields such as signal processing, speech recognition, pattern recognition, information entropy, error entropy and random neural network. Background technique [0002] With the rapid development of China's social economy, the process of urbanization continues to accelerate, and a large amount of noise is generated during the construction of urbanization, which has a greater impact on residents' daily life, study, and physical health. Complaints are becoming more and more frequent, directly or indirectly affecting social stability and order, so the identification and management of noise is becoming more and more important. The development of modern technology enables the collection of acoustic signals related to the environment and mechanical engineering construction. It is hoped that technologies in the fields of information s...

AI Technical Summary

Problems solved by technology

In the process of urban noise identification, the influence of interfering sounds is usually considered, but the existence of noise in the acquisition equipment is rarely considered

In the process of data collection, electronic components will be affected by electromagnetic pulse radiation generated by TV broadcast transmitters, radar and wireless communication equipment, and at the same time, electronic components such as sensors will generate short-term pulses when switching states , coupled with factors such as environmental noise, power frequency interference, and zero-point drift caused by on-site temperature changes, the collected data will contain a lot of non-Gaussian noise, making it difficult to achieve the ideal classification results

Images