Industrial fault diagnosis is crucial for ensuring the safety and efficiency of modern production systems. Industrial big data, particularly large-scale tabular data capturing multivariate time-series processes, offer valuable operational insights. Existing methods face significant challenges due to extreme label scarcity and massive unlabeled data volumes. Large Language Models (LLMs) hold great potential to address these issues due to their strong heterogeneous and few-shot learning capabilities. However, the application of LLMs to fault diagnosis with industrial big data, especially for tabular data, remains unexplored. In view of this, we propose a novel semi-supervised prefix tuning of LLMs for fault diagnosis with industrial big data. We first generate auxiliary prediction tasks based on the unlabeled data as the semi-supervised training materials for LLMs. Then we design a prefix-based soft embedding layer to fine-tune the LLMs, so that the model is able to learn the task-specific information in a parameter-efficient way. To make the model applicable to industrial big data, we also implement the Sparse Gaussian Processes (SGP) to filter the most informative samples to relieve the computational cost. Finally, we design a hybrid prompt template to effectively combine the hard and soft prompts and formulate the final prediction prompt for the industrial diagnosis tasks. The experiments have proven the superiority of the proposed method.

With the large scale adoption of Internet of Things (IoT) applications in people’s lives and industrial manufacturing processes, IoT security has become an important problem today. IoT security significantly relies on the security of the underlying hardware chip, which often contains critical information, such as encryption key. To understand existing IoT chip security, this study analyzes the security of an IoT security chip that has obtained an Arm Platform Security Architecture (PSA) Level 2 certification. Our analysis shows that the chip leaks part of the encryption key and presents a considerable security risk. Specifically, we use commodity equipment to collect electromagnetic traces of the chip. Using a statistical T-test, we find that the target chip has physical leakage during the AES encryption process. We further use correlation analysis to locate the detailed encryption interval in the collected electromagnetic trace for the Advanced Encryption Standard (AES) encryption operation. On the basis of the intermediate value correlation analysis, we recover half of the 16-byte AES encryption key. We repeat the process for three different tests; in all the tests, we obtain the same result, and we recover around 8 bytes of the 16-byte AES encryption key. Therefore, experimental results indicate that despite the Arm PSA Level 2 certification, the target security chip still suffers from physical leakage. Upper layer application developers should impose strong security mechanisms in addition to those of the chip itself to ensure IoT application security.