As we all know safe and reliable operation of electrical equipment is the first line of defense to avoid major accidents in power systems. Electrical equipment will produce some aging characteristics due to electrical and thermal aging in the operation process, such as furfural, acetone, methanol, formic acid, acetic acid, etc. The study shows that furfural can be used to evaluate the aging life of transformers accurately. At present, the detection of furfural dissolved in oil mainly adopts high performance liquid chromatography and ultraviolet spectrophotometry. However, oil-paper insulation degradation products are diverse and complex, and these traditional methods are only suitable for laboratory operations. Raman spectroscopy enables in situ detection of furfural dissolved in oil without the need for extraction. In 2015, the Osaka Laser Technology Research Center in Japan realised the first in-situ Raman spectroscopic detection of furfural standard samples in transformer oil, with a minimum concentration of 104 mg/L, which is far from the standard for oil-paper insulation aging. Surface Enhanced Raman Spectroscopy (SERS) can effectively improve the detection sensitivity of the features to be tested, and it is very effective to introduce this technology to evaluate the aging state of electrical equipment. To the aging of the transformer oil in many features high sensitivity tests of furfural, the researchers developed the gold and silver core-shell structure used as SERS substrate. They found that the furfural characteristic peak intensity corresponding to the relative standard deviation of only 5.8%, the stability of the core-shell structure of basement is higher. Still, the gold nanoparticles were almost were silver shell coated leading to the sensitivity of detection can't play to the largest, the detection limit could only reach 20 mg/L, and the antioxidant capacity was weak. To solve the problems of SERS substrate such as low sensitivity, easy oxidation and poor homogeneity, some scientific researchers introduced carbon nanomaterials into the preparation of SERS substrate. Based on their large specific surface area, strong adsorption capacity and oxidation resistance, when combined with precious metal materials with strong electromagnetic enhancement characteristics, they were used to prepare of SERS substrate. It is expected that more metal nanoparticles will be adsorbed on the surface of the carbon material, forming more "hot spots" and generating stronger Raman signals. SERS has the potential to improve the in situ detection limit of dissolved furfural in oil. In this paper, the chemical reduction method was adopted, AgNO₃ was used as the precursor, ascorbic acid was used as the reducing agent, the crystal morphology was controlled by PVP, and the rough flower-like silver nanoparticles were generated by controlling the stirring rate. The functionalised CNTs were mixed with flower-like silver nanoparticles to prepare carbon nanotube-modified flower-like silver nanoparticles (CNT-FAgNPs). The key point of this technology is that after acidification treatment, carbon nanotubes adsorb more hydroxyl and carboxyl groups on their surface, which is convenient for better combination with FAgNPs. The CNT-FAgNPs were adsorbed on the surface of gold-coated (Indium Tin Oxide, ITO) glass as the SERS substrate to achieve high-sensitivity in situ detection of furfural dissolved in oil with different concentrations. Studies have shown that the structural gap of the rough three-dimensional nanostructures formed by the combination of nano silver flowers with electromagnetic enhancement effect and CNTs can effectively enhance the Raman signal, while the large specific surface area and strong adsorption capacity of CNTs promote the formation of more "hot spots". The lower limit of detection for furfural in transformer oil is up to 7 mg/L, and the relative standard deviation of characteristic peaks at different substrate positions is 3.01%, which proves that the prepared substrate has good sensitivity and consistency, showing great potential for in-situ detection of trace aging characteristics in transformer oil.