This article addresses the problem of increasing the energy efficiency of electromechanical systems driven by asynchronous electric drives. In this context, one of the promising areas is the application of a predictive control strategy that allows for reducing energy losses in dynamic modes of electric drives. This paper proposes a computerised method for the multiparameter optimisation of predictive control algorithms for asynchronous electric drives. A computer model was designed in MATLAB and Simulink R2024a based on the gradient-based model predictive control strategy. A series of simulation experiments were carried out by varying the sampling step, number of iterations, prediction horizon, loss function parameters, and maximum linear search step to identify their impact on the control quality indicators. A taxonomic approach was used for multi-criteria optimisation. The study results show that the optimal setting of the algorithmic parameters improves the accuracy of task processing, reduces energy consumption, and reduces computation time. The results obtained can be used to design and operate energy-efficient control systems for asynchronous electric drives in industrial and transport applications. Prospects for further research will focus on hybrid intelligent architectures to enhance adaptability and integration into automated systems.

This article is devoted to solving the scientific and technical problem of increasing the energy efficiency of electromechanical systems driven by asynchronous electric drives, which is an important task in industry, transport and energy in the context of sustainable development and rational use of resources. Due to the large number of drives, energy efficient operation is increasingly important in drive control. In this context, one of the promising areas is the application of a predictive control strategy that allows for reducing energy losses in dynamic modes of electric drives. For a systematic parametrisation of such algorithms, this paper proposes a computerised method for multiparameter optimisation of predictive control algorithms for asynchronous electric drives. In the course of the study of this article, a computer model was built in Matlab & Simulink based on the gradient-based model predictive control strategy. A series of simulation experiments were carried out with variation of the sampling step, number of iterations, prediction horizon, loss function parameters and maximum linear search step to identify their impact on the control quality indicators. For the multicriteria optimisation a taxonomic approach is proposed. The study results show that the optimal setting of the algorithmic parameters improves the accuracy of task pro-cessing, reduces energy consumption and reduces computation time. The results obtained can be used to design and operate energy-efficient control systems for asynchronous electric drives in industrial and transport applications. Prospects for further research include the development of hybrid intelligent architectures to increase the adaptability of predictive control and its integration into automated control systems.

This paper addresses the problem of improving the robustness of image steganographic methods under lossy compression while preserving high perceptual quality and low computational complexity. The paper proposes an adaptive code-controlled steganographic method that enables spectrally selective embedding in the spatial domain through structured codewords. The proposed approach introduces block-level adaptivity in which the energy of the embedding codeword is dynamically selected according to the robustness characteristics of each image block. Instead of applying uniform embedding strength, the method determines the minimal codeword energy required to guarantee reliable message extraction under a predefined worst-case JPEG compression level. Experimental evaluation demonstrates that the proposed adaptive strategy significantly improves robustness to compression attacks while preserving high perceptual reliability and strong resistance to statistical steganalysis techniques. In particular, for JPEG quality factor (QF) = 50, the bit error rate is reduced to 1.25% while a high perceptual quality of 52.07 dB peak signal-to-noise ratio (PSNR) is achieved. For stronger attack conditions, QF = 20, the method achieves 6.6% bit errors with a PSNR of 47.7 dB. Overall, the proposed adaptive energy selection provides up to 22.68% fewer errors or up to 6.05 dB higher PSNR compared to the classical code-controlled steganographic method, confirming its effectiveness for practical steganographic applications.