Electromagnetic pulses （EMPs） primarily couple into an engine electronic control unit （ECU） through harness conduction. Bulk current injection （BCI） is the standard method for assessing controller sensitivity. To optimize EMP protection design for the ECU during the design phase， a virtual injection method based on harness conduction characteristics is proposed. This method uses a harness conduction prediction model based on a 1D convolutional neural network （CNN） and a simulation circuit built in Multisim. Injection current signals from BCI experiments serve as input to the model， generating predicted port signals， which are then injected into the simulation circuit. By monitoring the circuit’s output， the electromagnetic susceptibility of the engine electronic control system is analyzed. The predicted signals have an error margin of less than 5.8% compared to actual measurements. The virtual injection results match the BCI test outcomes and observed sensitivities. The method can provide a rapid analysis method for ECU module susceptibility during the design phase and aid EMP protection design.