With the rapid development of the semiconductor industry, the demand for high-speed testing in the large-scale production of semiconductor devices and integrated circuit production lines continues to grow. As one of the key tools in semiconductor device performance testing and integrated circuit testing, source measure unit (SMU) plays a crucial role in high-precision transient response testing scenarios. In high-precision measurement scenarios, multiple measurements are often required and averaged to improve measurement accuracy, but this can slow down the measurement speed. This article proposes a measurement acceleration algorithm based on BA-Informer time series prediction to solve the problem of decreased measurement speed in high-precision measurement. On the one hand, this algorithm improves the encoder structure. Traditional time series prediction models may have limitations in handling long-term dependencies and trend extraction. BiRNN is an extended version of recurrent neural network (RNN), which consists of two directional RNN. One forward RNN processes data from the beginning to the end of the sequence, while the other reverse RNN processes data from the end to the beginning of the sequence. In the end, the outputs from both directions are merged at each time step. Compared to traditional one-way RNN, BiRNN can more effectively handle data with before and after dependencies. Based on its characteristics, this article integrates BiRNN into the encoder structure. This algorithm can simultaneously process input sequences from both positive and negative directions, effectively limiting the bidirectional contextual information of data and significantly enhancing the model’s ability to capture time series trends. In this paper, BiRNN is integrated into the encoder structure, and the algorithm can simultaneously process input sequences from both positive and negative directions, more effectively capturing the bidirectional contextual information of data and significantly enhancing the model’s ability to capture time series trends. This improvement enables the model to more accurately grasp the overall trend of data changes during prediction, thereby improving prediction accuracy. On the other hand, an attention discrete cosine transform (ADCT) module is introduced between the encoder and decoder to convert time-domain signals into frequency-domain representations. This not only reveals the spectral characteristics of the signal but also reduces data redundancy and improves the efficiency of subsequent processing by combining attention mechanisms. Finally, the algorithm performance is analyzed by analyzing the output characteristic curves of loads with different properties. The experiment shows that the prediction algorithm and the combination of measurement and prediction method proposed in this article save half of the measurement time by combining measurement and prediction while ensuring the same amount of data obtained, verifying the effectiveness of the proposed method.

The widespread deployment of unmanned aerial vehicles has raised growing privacy and security concerns, creating an urgent need for effective drone detection and identification. To address these challenges, this paper proposes a semantic learning framework termed Drone Time–Frequency-Semantic Network (DroneTFS-SemNet) for drone recognition. The main contributions are as follows. A Temporal Enhancement module is designed to improve model adaptability and robustness through radio frequency (RF) signal noise simulation, feature pre-extraction, learnable frequency-band partitioning, and an attention mechanism. A Cross-Channel Recalibration Module is introduced to promote cross-channel interaction of RF features via dilated convolutions and Channel Shuffle, thereby enhancing the model’s learning capability. Third, Feature Fusion Attention is employed to weight, align, and fuse RF features from different levels, thereby improving the model’s ability to represent differences among drone categories and operation modes. A semantic construction and decision module is further developed to extract high-level semantic features and accomplish drone detection, classification, and recognition. Experimental results on the DroneRF dataset show that DroneTFS-SemNet achieves a drone classification accuracy of 99.73% and an operation mode recognition accuracy of 99.65%, outperforming multiple comparison methods. These results demonstrate that the proposed model has strong recognition capability on this public dataset and under controlled experimental conditions, providing an effective method for RF-based drone recognition research.