A Microwave imaging system for in-line packaged food contamination monitoring

A Microwave imaging system for in-line packaged food contamination monitoring The Master Thesis focus was the investigation of contamination monitoring of packaged food using a microwave imaging system. Microwave imaging is an engineering approach derived as an evolution of older detecting/locating techniques (e.g., radar) for hidden or embedded objects detection in a tested system, exploiting electromagnetic waves in the microwave regime. The scope of this technique is to detect a target and localize it within the system under analysis, exploiting the dielectric properties and contrast of media interacting with electromagnetic fields. Several techniques have been employed for customer safety and conformity of the product, but each of them presents some limitations: infrared detectors (work only small items), X-rays (ionizing radiation risks), metal detectors (for metallic type intrusion only). The principle of microwave imaging is to derive information about detection and localization of an object/target in a system exploiting the dielectric properties of the parts composing the latter, and their interaction with electromagnetic waves from incoming sources (antennas). The sensors/antennas work both as transmitters and as receivers in the frequency range 9-11 GHz during the imaging process according to a preassigned sequence, irradiating and capturing the irradiated electromagnetic (EM) fields respectively. Ideally, the system should be shielded from EM disturbances (outcoming EM fields from electrical and electronical devices, power supply networks or natural sources), to avoid compromising the measurements. The retrieval of the dynamic operator requires the static configurations (provided as CAD projects) to be discretized, and the finite element method is used to approach the problem by computing the electric fields on the discrete mesh elements. At the end, the 6 static operators are compacted by means of geometrical translation and interpolation techniques. In particular, 3 interpolation formulations have been investigated and tested to select the most accurate choice. Measured scattering matrices have been employed for the analysis. The original scattering parameters have been furtherly filtered according to a Truncated Singular Value Decomposition (TSVD) to solve the ill-conditioning mathematical issues. Moreover, a statistical analysis of the noise in the measurements collection has been performed to discard the noisiest parameters according to a tolerance criterion. Also, the illumination balancing problem has been investigated to allow a uniform resolution detection within the sample under test and avoid blind spots, where contaminants might be omitted due to the unfavourable position with respect to the sensors. This innovative technology presents several advantages, since: -??non-invasive; -??non-destructive; -??low-power; -??portable; -??real-time; -??easy to operate. The microwave imaging technology is very promising for industrial applications on large scale production chains, requiring therefore a scientific effort to improve quality detection and limit noise effects due to external factors. The main purpose is achieving an optimal distinct binary classification reducing both false-positive and false-negative cases, as well as refining the accuracy of the intrusion quest.