The groundbreaking advancements in deep learning technology establish a new paradigm for interdisciplinary research in “AI + meteorology”， and severe convective weather prediction emerges as a cutting-edge research focus due to its complex dynamical characteristics and significant socioeconomic impacts. Therefore， the theoretical advancements and methodological innovations of Deep Neural Networks （DNN） in severe convective weather prediction were reviewed systemically， and their specific applications were explored deeply. Firstly， based on the spatio-temporal sequence prediction paradigm， the mechanisms of high-frequency feature extraction by Recurrent Neural Networks （RNNs） and non-RNNs in meteorological sequence modeling were dissected. Secondly， from a generative modeling perspective， the advantages of Generative Adversarial Network （GAN） and diffusion model in probabilistic prediction of extreme weather events were demonstrated. Thirdly， the theoretical breakthroughs of meteorological large-scale models realizing multimodal data fusion and cross-scale feature learning via pre-training and fine-tuning paradigms were revealed， along with their generalization enhancement mechanisms in global numerical weather prediction. Fourthly， aiming at model evaluation systems， the limitations of traditional statistical metrics in extreme weather prediction were analyzed， and pathways for constructing novel evaluation frameworks such as physical consistency constraints were discussed. Finally， the key scientific challenges faced currently and future research directions were distilled， aiming to provide theoretical support and methodological references for constructing the next-generation intelligent system for severe convective weather prediction.

McCabe's basic path testing method (McCABE T J. A complexity measure. IEEE Transactions on Software Engineering, 1976, SE-2(4): 308-320) is a more rigorous software testing one in dynamic white-box testing techniques, but the efficiency of McCabe method is lower. To solve this problem, this paper proposed an algorithm to design basic path test case according to the basic program structure. The algorithm first created a basic unit chart based on the Z-path coverage. Next, the rules that combined a control flow graph from basic unit were established. On this basis, the combination algorithm of the basic path was constructed. The algorithm collected the path set of basic program structure by scanning a program only once, and then generated the basic path set using those paths by combination ways. This method is more concise than the method proposed by McCabe, and it can improve the efficiency of the basic path test case design.

In order to solve the shortage of the method based on fragile watermarking, a new chaotic system and fragile watermarking based image tamper detection algorithm was proposed. Firstly, the method used Arnold cat map k times to scramble the original image, and then chose the Least Significant Bit (LSB) plane of the scrambled image as the embedding position. The actual watermark information was formed by using exclusive-or operation between a random binary sequence produced by Logistic map and the original watermark, and the LSB replacement method was utilized to embed watermark. Finally, the watermarked image was obtained by utilizing Arnold cat map T-k times. The experimental results show that, the introduction of chaotic system to a great extend improves the security of the proposed scheme. In addition, the proposed method achieves superior tamper detection and localization accuracy under different common attacks.

To decrease the modification of original data during the process of database watermarking, and to make watermark information more secretly, an algorithm based on parity modulation in two-dimensional space was proposed. One dimension was made up of primary keys' Hash values, the other was made up of numeric attributes' redundant bits. Watermark information was embedded by modulating the parity of Hash values and redundant bits. After watermarking, the modification of data average was 0.5296×10^(-2)%, and the modification of data variance was 0.6509×10^(-4)%. The experimental results indicate that the modification amount of original data is reduced and this algorithm is efficient and robust.

Concerning the lack of research on User Datagram Protocol (UDP) flow detection, and its low accuracy and efficiency, an adaptive UDP flow detection based on multi-granularity was proposed. Two different dynamic timeout strategies based on different granularity were designed by analyzing the characteristics of UDP flow. Compared with other timeout strategies, the accuracy of the proposed method was similar to the fixed timeout strategy, and the memory usage was only about 75% of the others. The experimental results show that the proposed method is suitable to the UDP flow detection.

The error concealment is an important technique in the video transmission, and it can ensure the reconstruction video quality and efficiently recover the data loss and the data errors in the transmission process caused by severe transmission environments. In order to enhance the error resilience of AVS-P2, the paper proposed a new adaptive temporal-spatial error concealment method based on the redundancy motion vectors. To conceal a lost block, the paper used the spatial error concealment for the I-frame macroblocks, and used the temporal error concealment for the non-I-frame macroblocks. At the same time, according to the motion intensity of the macroblocks, it used the default error concealment of AVS-P2 and error concealment method based on redundancy motion vectors, respectively. Lastly, the proposed algorithm was realized based on the platform of the AVS-P2 RM52_20080721. The simulation results show that the proposed method is significantly better than the existing techniques in terms of both objective and subjective quality of reconstruction video.

Concerning the specific application environment of embedded microprocessor, the branch prediction technology was researched in this paper, and a new scheme of branch prediction was proposed. Compatible with cache design, jump direction and destination address of branch prediction happened on extended instruction bus. The unexecuted instruction and address pointer were saved for possible recovery after misprediction, which reduced misprediction penalty, simultaneously guaranteed the instruction flow to execute correctly. The study shows this scheme is of little hardware spending, high prediction efficiency and low misprediction penalty.

Under the extremely sparse data environment, the traditional collaborative filtering algorithms only depenging on users rating data cannot achieve satisfactory recommended quality. A recommendation algorithm based on user characteristics and item attributes was provided. First, the timerelated interest degree was introduced in the process of user similarity calculation，which made a more accurate nearest neighbor set. While predicting the rating for the target user, the trust measure was used to reflect the neighbors contribution level for the ultimate recommendation. In addition, the users preference on item attribute instead of rating score was used to recommend the new items. The experimental results based on MovieLens data set show that the improved algorithm can solve the problem of coldstart and improve the accuracy of system recommendation significantly.

According to the urgent characteristics of casualty evacuation, this paper established a casualty evacuation mathematical model of multidisaster places, multihospital and multipatients with the goal of the shortest time. Considering the hospital resource competition of patients, Operation Table Method (OTM) was used to solve the model and realize the goal of the shortest time. The method has been realized on the provincial level information demonstration platform of emergency management, and can deal with different injury types and the patient priorities. With this platform and Geographic Information System (GIS) technology, the method is simulated and the effect is good.

Aiming at the need of very large spatial database applications and taking notice of the disadvantagesf existing spatial indexes, the thesis introduces an improved R*-tree spatial indexing structure after addressing the related concepts, data structures, algorithms, and performannalyses of R-tree and R*-tree. This paper gave a more efficient algorithm which analyses the shortcoming of R*-tree spatial index algorithm. The experiment results show that improved R*-tree is better than the original R*-tree in index efficiency.