PipeMonitor is a hardware/software prototype for real-time detection and recognition of objects and object states using elastic wave propagation analysis.

This is a part of the research development for detection of illegal operations with oil, gas and water pipelines, but system can detect changes in various types of structures and objects on surfaces.

Based on my PhD work and patented invention with added neural networks using NeuralLibrary:

PhD Work with Articles from Russian Science Journals and Other Projects

• Based on a principle of generation and analysis of elastic waveforms, propagated through controlled object
• Working Modes:
• Autonomous detection of abnormal object condition without PC connection
• Real-time detection and recognition of current object state with C# software
• Detection and recognition of objects on a sensitive surface with C# software
• Software-only mode: loading and analyzing of experimental recorded WAV-files for object classification
• Hardware:
• Arduino Due for impulse generation, coherent accumulation, correlation, plotting on display, sending signals to C# software via serial connection
• A periodic impulse generator (solenoid - long range, relay - working with demo mini prototype)
• A vibration sensor, powered from DC or batteries
• Software:
• Сreating a list of states (objects) and setting the corresponding images
• Receiving of accumulated impulses from a microcontroller via serial connection
• Working with recorded WAV-files
• Increasing signal-to-noise ratio by coherent impulse accumulation (from loaded files)
• Fast Fourier Transform for signals
• Training a neural network on a spectrum or waveforms [NeuralLibrary]
• FFT classifiers training with feature extraction (patented invention)
• Recognition of object states based on a signal classification
• A Sound indication of recognized state by speech synthesizer (not shown in demos)