光储微网中光储单元接口变换器的输出功率及散热条件变化下，现有的IGBT结温计算方法难以评估其对IGBT模块热扩散角的影响，导致结温计算精度受限，为系统热管理带来巨大挑战。为解决上述问题，本文提出一种考虑三维横向热传导的光储微网系统接口变换器IGBT结温计算方法。首先，在光储微网系统下，考虑多芯片间热耦合，建立功率器件物理热模型；然后，通过所建物理模型，进一步提出一种考虑三维横向热传导的结温计算方法，建立考虑三维横向热传导的热网络模型，有效提升了当前状态热参数及功率模块热扩散角的计算精度；最后，在散热柱型(PinFin)散热器结构下，利用有限元分析验证所提模型的精确性。仿真结果表明：与多种结温计算方法相比，本文所提方法在稳态、功率突变工况下结温计算误差均最小，分别约为3.11%，3.65%，较不考虑热扩散角(α=0)计算方法精度分别提升11.53%，61.93%；不同散热条件下，本文所提方法同样结温精度最高，误差最小。

Under the changing output power and heat dissipation conditions of the optical storage unit interface converter in the optical storage microgrid, existing junction temperature algorithms of insulated gate bipolar transistor (IGBT) are difficult to evaluate its impact on the thermal diffusion angle of the IGBT module, resulting in limited junction temperature algorithm accuracy and posing a huge challenge to system thermal management. To address the above issues, this article proposes a junction temperature algorithm of IGBT in the interface converter in optical storage microgrid systems that considers three-dimensional transverse heat conduction (3-D THC). Firstly, in the optical storage microgrid system, considering the thermal coupling between multiple chips, a physical thermal model of power devices is established; Then, based on the established physical model, a junction temperature algorithm considering 3-D THC is further proposed, and a thermal network model considering 3-D THC is established, effectively improving the calculation accuracy of current state thermal parameters and power module thermal diffusion angle; Finally, the accuracy of the proposed model was verified using finite element analysis under the PinFin heat sink structure. The simulation results show that compared with various junction temperature algorithms, the proposed algorithm in this article has the smallest error in junction temperature calculation under steady-state and sudden power change conditions, with approximately 3.11% and 3.65% respectively, which is more accurate than not considering thermal diffusion angle (α=0) algorithm improved accuracy by 11.53% and 61.93% respectively; Under different heat dissipation conditions, the algorithm proposed in this article also has the highest junction temperature accuracy and the smallest error.