Gene regulation is the regulation of gene expression behavior by various related substances in cells, which controls almost all cell activities. Therefore, the study of gene regulation can not only explore the internal law of life activities, but also play a great role in the prediction, diagnosis, treatment and drug design of gene-related diseases. Using multi-source biological information such as gene expression profile data, transcription factor information and protein interaction information, a network model can be established to describe the regulatory relationship between genes, so as to facilitate the above studies. In order to solve the problem of low accuracy of traditional gene regulatory network construction methods, a new dynamic model of gene regulatory network was established by combining hybrid genetics and threshold restriction. The model is divided into two parts: solution space reduction and parameter fitting. In the phase of solution space reduction, singular value decomposition method is used to define the mathematically feasible gene regulatory network to reduce unnecessary calculation, and then the control gene of each gene is limited to a certain scale by threshold limitation method, which improves the computational efficiency and accords with the bioinformatics rules. In the parameter fitting part, the parallel genetic algorithm is used to optimize the whole solution space quickly, and then the mountain climbing method is used to solve the problem carefully in a small range to improve the calculation accuracy. In this study, we applied this method to the establishment of genetic regulatory systems for complex skin melanoma and type 2 diabetes. By comparing with the real network, the correctness of the method is proved. Compared with traditional genetic and PSO methods, the effectiveness of the proposed method was verified. In this paper, the deep mechanism of gene regulation is modeled, and the regulation process involving genes, proteins and small biological molecules is described in more detail, so as to be more detailed than other models and more consistent with the intracellular dynamics law.