Multi-operator collaboration involves the transfer of information and the completion of tasks at multiple levels. As the changes of task complexity and staff positions, the efficiency of collaboration changes accordingly. To analyze the cognitive level of the operators during collaborative tasks, a cognitive-based multi-operator evaluation model was established. To ensure the objectivity of the evaluation, entropy value method and mean square deviation method were selected for the assignment of indexes. The weight coefficients of the two methods were calculated to get the combined weights, and finally the quality value of collaborative communication was obtained. In addition, three sets of communication efficiency experiments were conducted to validate the assessment model. The EEG signal of one operator was tested during each experiment, and the metrics required for the evaluation model were recorded. The word number of the command, the average reaction time, the error rate of operation, the maximum value of the command hierarchy, total time of completion, and total number of operations were obtained. The data of the above indexes were substituted into the model. The quality scores of the three experiments were calculated as 0.983, 0.335 and 0.156, respectively. The relative power of θ wave was applicable to the cognitive interaction evaluation model (p = 0.034). The cognitive load during the experiments were consistent with the results calculated by the established model. In conclusion, the verbal communication evaluation model is consistent with the physiological signals during the cognitive interaction process. It can be assumed that the collaborative communication evaluation model is accurate during the cognitive interaction, and can be embedded into a comprehensive assessment system in the future research.