CNN Noise Recognition Engine for Automated Construction Site Monitoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise monitoring systems are passive and require heavy human effort for real-time noise classification, making it difficult to identify and regulate noise pollution from construction equipment in urban environments.
Innovation Solution
A CNN+Shift+Pitch noise recognition engine is trained on cyclically time-shifted samples of repetitive noise sources, allowing for accurate identification of noise patterns even if detected out of sync with original recordings, and correlating pitch range with time intervals to enhance identification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If real-time noise monitoring is implemented using microphones and human review, then noise detection capability is provided, but heavy human effort and slow response time are required
Solution Approach 1:
The system implements automated noise classification using a trained CNN model that independently identifies and categorizes construction equipment noises without human intervention. The model processes audio data, detects noise patterns, and generates classifications autonomously, enabling the system to serve itself in the noise detection and classification task.
Solution Approach 2:
The patent replaces the mechanical human review process with an automated computational system. A convolutional neural network model processes audio signals and performs noise classification that previously required human experts to listen to and categorize recordings, thereby substituting mechanical human effort with automated computational analysis.
2Measurement precision
If simple microphone recording is used for noise monitoring, then basic noise detection is achieved, but accurate classification of similar construction equipment noises is difficult
Solution Approach 1:
The system transforms the raw audio signal into a spectrogram representation, changing the parameter space from time-domain waveforms to frequency-time visualizations. This parameter transformation enables the CNN model to more effectively distinguish between similar construction equipment noises by analyzing frequency patterns and temporal characteristics that are not apparent in raw audio signals.
Solution Approach 2:
The patent applies preliminary data processing and feature extraction steps before classification. Audio recordings are converted to spectrograms and preprocessed to highlight relevant acoustic features. This preliminary action prepares the data in an optimal format for the CNN model, improving classification accuracy without requiring complex hardware modifications.
3Speed
If traditional noise monitoring systems are used, then passive noise detection is provided, but rapid response to noise pollution issues cannot be achieved
Solution Approach 1:
The system implements continuous real-time noise monitoring and classification, processing audio streams without interruption. The CNN model continuously analyzes incoming audio data, providing ongoing noise classification and enabling rapid detection of regulatory violations as they occur, rather than relying on periodic human reviews of recorded data.
Solution Approach 2:
The system provides immediate feedback by classifying noises in real-time and generating alerts when regulated equipment is detected. This feedback mechanism enables rapid response by notifying authorities or stakeholders instantly when noise violations occur, allowing for immediate intervention rather than delayed response through manual review processes.
Data Source
AI summary
A method for analyzing environmental sounds to identify repeating noises due to power equipment for purposes of controlling noise pollution is disclosed. Ambient sounds are sampled with microphones and the resulting data analyzed by an edge computing device. The samples are compared to stored data with a convolutional neural network by the edge computing device which also performs pitch shifting and cycle shifting to reduce false negatives. Positive results are reported via LoRa and/or 4G to a cloud platform and/or backend server.