The paper presents a method for neural network approximation of helicopter turboshift engine (TE) parameters to optimize engine efficiency, reliability, and flight safety. A three-layer feedforward neural network with linear neurons in the output layer is used, and the scale conjugate gradient algorithm is modified by introducing a moment coefficient to avoid local minima. The modified algorithm helps in calculating new model parameters, ensuring accurate predictions of the relationship between energy release during compressor rotation and gas-generator rotor r.p.m. The optimal ratio of energy consumption and compressor operating efficiency is achieved, leading to improved performance and reliability of the helicopter TE. Experimental data show that the proposed method outperforms existing analogues, with an efficiency coefficient of 0.994 compared to 0.914. The study also includes a detailed analysis of the neural network's training process, including the selection of optimal network structure and the impact of the moment coefficient on avoiding local minima. The results provide valuable insights for further research and development in improving helicopter TE performance and efficiency.The paper presents a method for neural network approximation of helicopter turboshift engine (TE) parameters to optimize engine efficiency, reliability, and flight safety. A three-layer feedforward neural network with linear neurons in the output layer is used, and the scale conjugate gradient algorithm is modified by introducing a moment coefficient to avoid local minima. The modified algorithm helps in calculating new model parameters, ensuring accurate predictions of the relationship between energy release during compressor rotation and gas-generator rotor r.p.m. The optimal ratio of energy consumption and compressor operating efficiency is achieved, leading to improved performance and reliability of the helicopter TE. Experimental data show that the proposed method outperforms existing analogues, with an efficiency coefficient of 0.994 compared to 0.914. The study also includes a detailed analysis of the neural network's training process, including the selection of optimal network structure and the impact of the moment coefficient on avoiding local minima. The results provide valuable insights for further research and development in improving helicopter TE performance and efficiency.