Converter junction temperature monitoring method and system based on neural network

The invention discloses a converter junction temperature monitoring method and system based on a neural network, and the method comprises the steps: carrying out the preprocessing of values of Tj, Rdson, RL and RC, and completing the collection of a training data set of a maximum junction temperature model; building a maximum junction temperature model based on a neural network; the input data set comprises Rdson, RL and RC; the output data set is Tj; the training data set is adopted to train the maximum junction temperature model, and a relational matrix containing independent variables Rdson, RL and RC and dependent variable junction temperature data Tj is obtained; parameters of inductive current, input voltage and capacitor voltage of the converter are collected; rdson, RL and RC are obtained through operation, numerical values of Rdson, RL and RC are input into the maximum junction temperature model, and junction temperature data Tj of the converter are obtained in real time based on the relational matri