Three-phase cables in a motor driving system are the major ways of electromagnetic radiation, which result in potential interference to the nearby electronic equipment on electric vehicle. Numerical simulation of radiated emissions from cables is time-consuming, which becomes insufficient to compute large numbers of cables for actual applications. Hence, this article proposes a fast prediction approach that can highly improve the computational efficiency. Since cables are often close to each other and arbitrarily bent in the real application due to confined working space, the influence of proximity effects caused by closely-spaced cables on radiated emissions is also considered in the calculation model by multi-transmission line and Hertzian dipole theories. Besides, an equivalent model is presented to characterize the measured motor impedance. On the basis of the proposed cables model, a complete motor driving system model is originally developed by integrating a radiated electromagnetic interference (EMI) source model of inverter and a load equivalent model of motor impedance. The system model is of vital value to identify the mechanism concerning radiated EMI issues on electric vehicles and efficiently evaluated in the end. Both numerical simulations and experiments are conducted for the validation of proposed approach, which show that the proposed approach can save over 90% computational time with desired evaluation accuracy.

Radiated emission from electric motors interferes nearby electronic devices and, thus, degrades the electromagnetic compatibility performance of the systems. Establishing a full-wave model of electric motor is crucial to full understanding of the radiated emission. Nevertheless, challenges remain in modeling large quantities of winding wires that hinders the establishment of a full-wave model. In particular, the proximity effects of closely spaced winding wires can strongly affect their radiation characteristics by changing the charge distributions on wires, which should be critically considered in the reduction modeling. In this article, the geometrical and electrical characteristics considering proximity effects are first generated by deriving the per-unit-length capacitance matrix of wires, and a four-step reduction procedure is detailed to simplify the complete wires model influenced by proximity effects. Backpropagation (BP) neural network is used for the first time to compute the geometrical characteristics of reduced wires model. Our results show that the reduced model has identical radiation characteristics with the complete model, as such it can be applied to develop a full-wave electric motor model for the prediction of radiated emissions. The validation of our technology is testified both numerically and experimentally.

This paper proposes a modeling method for improvement of the SPICE Model of a diode based on measurement and a nonlinear fitting random optimization algorithm. First, the mechanism of electromagnetic interference generated by diodes at high frequencies was analyzed: dynamic characteristic parameters such as reverse recovery of diodes and junction capacitance. Second, the method of obtaining the I-V characteristic curve and junction capacitance characteristics of the diode through testing was introduced in detail, and a nonlinear fitting random optimization algorithm was proposed to calculate the static and dynamic characteristic parameters of the diode, which can be applied to improvement of the SPICE model of the diode. A simulation and test circuit were built, and the simulation and test results were compared in the time and frequency domains. The model in this study can not only truly reflect the peak voltage owing to the electromagnetic interference characteristics, but also ensure that the harmonic components and component amplitudes are within three times the fundamental frequency, verifying the accuracy of the electromagnetic compatibility model.