LETI: LETI Scientists Сreate Smart Metal Detectors
Researchers at ETU “LETI” have created a prototype of a unique metal detector based on a chaotic circuit.
Today, walk-through metal detectors can be found in airports and places of mass gatherings all over the world. Devices that detect metal objects without contact are widely used for security, archaeology, mineral exploration, construction, and industry.
“Artificial intelligence and neuromorphic systems are an in-demand and topical area of research. In our opinion, classical neural networks are not the future, structures based on models of biological neurons are. Our development uses the action principle of a sensor neuron to solve the problem of metal detection. There are many applications for such detectors.”
Olga Druzhina, a postgraduate student of the Department of Computer Aided Design, engineer of the Youth Research Institute at LETI
The main problem of modern metal detector designs is insufficient selectivity, that is, they cannot always determine what kind of metal the detected object is made of. “Such qualities as low energy consumption, high sensitivity, and resistance to interference are also important,” commented Timur Karimov, Associate Professor at the Department of Computer Aided Design, Head of the Advanced Circuits Engineering Group of the Youth Research Institute at LETI.
Given the existing shortcomings of modern metal detectors, scientists have created a new type of sensor, which can be used to detect metal objects, determine what metal they are made of, and at what distance they are located. In this case, the detection range is increased compared with classic designs. The research results by the research group were published in the Nonlinear Dynamics journal.
The developed sensor consists of two parts: a sensitive element in the form of an inductive coil and an electronic circuit in which chaotic oscillations mimicking impulses of a biological neuron occur. The signals from the circuit go to the computer, which processes them with special algorithms and shows the type of target and the distance to it. According to the researchers, the combination of properties of chaotic and neuromorphic systems in this development gives it favorable qualities of both high sensitivity to changes in parameters and comparative ease of interpretation of behavior.
“Our electrical circuit simulates a biological neuron, with a sensitive coil being part of the model. When a metal object approaches the coil, eddy currents are induced in the object, changing the oscillations in the electric circuit. The distribution of the distances between the pulses in the circuit changes. Using a computer, we calculate this distribution and decode it, because it contains information about the type of target and the range to it.”
Timur Karimov, Associate Professor of the Department of Computer Aided Design at LETI
The developed device based on a chaotic oscillator has no direct counterparts. Scientists are confident that international colleagues engaged in nonlinear dynamics and chaotic sensors will be interested in this paper.
Researchers have plans to solve the problem of the device’s unwanted sensitivity to temperature and interference. “Chaotic systems, unfortunately, are not very resistant to changes in operating conditions. We plan to solve this problem by switching from analog circuitry to implementing most of the circuitry on a microcontroller. The processing algorithms will also be implemented on the controller without the need for a separate calculator,” comments Timur Karimov.
The prospects are to further improve the detector design and apply chaotic neuron-like systems in other tasks, for example, for diagnosing faults of electric machines, such as electric motors in factories or wind turbines.