Resumo:
Power transformers (TP) connect the entire electrical energy network, which goes from the
generation and production of energy to the final consumer. Therefore, it is necessary to monitor
their operating conditions. The operating condition of the TP is related to its insulation system,
which includes the refrigerant oil and insulating paper, which must not operate at temperatures
above technical specifications, as these conditions cause failures in the TP that can lead to
interruption of the power supply. In this research, poly(glycerol dendrimer) microspheres
encapsulating cobalt-doped hopeita (CoHoM) was synthesized for its potential use as optical
thermometer with reversible thermochromic property coupled to a multilayer feed-forward
artificial neural network as a proposal for temperature monitoring in power transformers. The
purpose of this sensor is to show color change when the power transformer temperature reaches
alert levels. The structures of the microspheres were studied using scanning electron
microscopy (SEM), infrared spectroscopy (ATR-FTIR), X-ray diffraction (XRD) and
thermogravimetric analysis (TGA). Through TGA it was possible to obtain information about
the thermal tolerance of the material, which supports temperatures above 300°C. From the
micrographs obtained from the powder material, it is possible to observe microspheres with
smooth surfaces and with an average diameter of approximately 0.82μm. Through the analysis
of X-ray diffractometry (XRD) it was observed the formation of crystalline phase in the planes
(110) and (200) in the structure of Hopeite. After these analyzes an ink was produced as a
thermochromic coating, which had its optical properties analyzed by UV-Vis techniques and
automated color recognition in CIEL*a*b* space coordinates. Finally, it was shown that the
CoHoM reversible thermochromic sensors allow detecting temperatures around 60°C, changing
the color from dark blue to light pink. CoHoM sensors have potential for monitoring power
transformers and show promise for use in real-time analysis.
