The analysis of existing approaches to reducing human errors in the technological process of electric power facilities has been carried out, revealing the influence of psychophysiological and emotional state of the operator on the risk of accident occurrence. The influence of emotions on human decision making has been analyzed, and the necessity to consider emotions in decision support has been proved. As an approach to recognizing emotional states of operators, the use of artificial neural networks for recognizing by face on video is chosen. The task of technological process risk management by considering the psychophysiological and emotional states of the operator is set and formalized. An algorithm for assessing the risk of accident occurrence at the technological process is proposed, which combines the consideration of parameters caused by technological and human factors. A computational experiment was carried out on the example of a fire at an electrical installation. The VGG16 convolutional neural network architecture was used for the task of recognizing the emotional state of the operator from the image. The result shows the sufficient accuracy of the applied method. The architecture of the developed information system of technological process risk assessment is given.