Abstarct: Abstract Insulators are one of the vital components in electrical energy transmission and distribution systems that play an important role in the stability of the power grid. In addition to mechanical strength, insulators must also have appropriate electrical strength. Polymer insulators are used today at various voltage levels due to their low weight and proper insulation. However, environmental factors such as pollution, humidity, and ultraviolet radiation can affect the performance of these insulators. In the meantime, ultraviolet radiation, due to its effect on the silicone rubber material used in the insulator body, causes them to deteriorate and worsens the performance conditions of the insulator in the presence of pollution and humidity. In addition, the location of the insulator, the severity of pollution, and the way it is installed on the insulator body affect its performance. Under these conditions, the occurrence of an electric arc on the insulator is considered a threatening factor that can cause the destruction of the insulator. In this context, various parameters have been proposed to provide a solution to predict the occurrence of electric arcs, all of which are baxsed on the occurrence of uniform contamination on a healthy insulator. However, the performance of the insulator under different contamination intensities and humidity is different due to the establishment of non-uniform contamination on an aged insulator with uniform contamination. In this thesis, a suitable criterion for predicting the condition of 20 kV polymer insulators baxsed on the analysis of the leakage current passing through the insulator is presented. In order to collect the required information, tests have been conducted on two different insulators in two healthy and aged states with ultraviolet radiation, as well as under non-uniform sector contamination. Considering the effect of humidity and contamination intensity at different levels has led to the results obtained from the tests being collected more comprehensively. In order to analyze and evaluate the obtained leakage currents, harmonic analysis was performed on the samples and examined in terms of amplitude and phase. In general, 24 features were extracted from the analyses performed, which were used to predict the occurrence of electrical discharges in two scenarios. The scenarios were defined baxsed on the effect of the leakage current amplitude or environmental factors on data classification. Due to the high overlap of the data, artificial intelligence was used. According to the defined scenarios, their training and testing stages were performed. Four methods including decision tree, random forest, support vector machine and multilxayer perceptron were used to evaluate the results. All of these methods were able to provide acceptable performance in determining insulator conditions in normal, precautionary and critical states. However, the random forest (RF) method had the best performance compared to other methods. Using the RF method with three, five and seven features has increased the data classification accuracy to 77, 85 and 85% for the first scenario and 94, 97 and 97% for the second scenario, respectively. In addition, the accuracy, importance of features and error diagram are presented for each of the methods. The results obtained show that the indicators baxsed on phase angle and harmonic ratios have played an important role in achieving the objectives of this research. Also, the presented artificial intelligence model in predicting the occurrence of electric arcs in the scenario baxsed on the leakage current amplitude had a better performance than the scenario baxsed on environmental conditions.