Power module Insulated Gate Bipolar Transistor switching losses include switch-on losses and switch-off losses, the switching losses can generate a great deal of heat which causes a larger temperature rise. It will seriously affect the safe and reliable operation of the device, but how to obtain accurate switching losses quickly and conveniently needs to be solved. The goal of this study is to accurately predict the switching losses of Insulated Gate Bipolar Transistor. Firstly a dynamic characteristics test was carried out, the dynamic voltage and current waveforms under different test conditions were obtained. Based on the data, the influence which the circuit operating environment parameters have on switching losses was analyzed. Secondly, for the defect of switching losses in small sample modeling, a novel switching losses prediction model based on the support vector machine optimized by the improved chicken swarm optimization algorithm was established. This novel model is modified on the basis of the chicken swarm optimization algorithm. The dynamic inertia weight and the part of learning from optimal individuals were introduced to the location updating formula of chicks, so that the chicks could forage flexibly and jump from the local optimal solutions. Finally, the novel model was used to predict the switching losses. To further verify the accuracy and effectiveness of the novel model, the true switching losses were compared respectively with the predicted results of the novel model and the other two models. The result showed that the novel model had a higher convergence precision and prediction precision. Compared with the other two models, the average relative error of switching losses prediction from the novel model is at least reduced by 1.2607%. The novel model contributes to the accurate prediction of the switching losses. It also has an important guiding significance for the improvement of renewable energy utilization and the system reliability.