In order to get an accurate and reliable model of the rise of temperature of connection fitting in value hall of UHVDC,a method of fuzzy system combined ant colony algorithm was used.After analyzing the characteristics of basic ant colony algorithm and gradient descent algorithm,an improved ant colony algorithm was put forward by combining basic ant colony algorithm and gradient descent algorithm,in which gradient descent algorithm was processed after basic ant colony algorithm. Through training data obtained from experiment,the fuzzy system was trained by basic ant colony algorithm,gradient descent algorithm and improved ant colony algorithm respectively.The convergence effect of improved ant colony algorithm was better than that of other two algorithms.All models were tested by testing data obtained from experiment,and the prediction effect of the fuzzy system trained by improved ant colony algorithm was best of all models.The prediction results showed that if training data obtained from experiment was enough,the fuzzy system trained by improved ant colony algorithm was reliable to predict the rise of temperature of connection fitting.
