From the perspective of nonlinear dynamics and bifurcation theory, this paper analyzes the impact of time delay on small-signal stability region of doubly-fed induction generator (DFIG) grid-connected power system. Firstly, the differential-algebraic equation model of the system is established. It is theoretically demonstrated that time delay will affect the bifurcation behavior of the system, especially the degenerate Hopf bifurcation (DHB) under a specific time delay. Then, the time delay, the injected DFIG mechanical power, and the load reactive power are chosen as the bifurcation variables. The bifurcation diagram is obtained through bifurcation analysis, which can determine the multi-parameter small-signal stability boundary of the system. Finally, the impact of single and multiple time delays on the stability region is analyzed through the stability boundary. It is found that the DHB due to the time delay variation induces a hole effect in the system stability region. Moreover, increasing the time delay may also improve the system stability margin. The findings of this study have significant guiding implications for multi-time delay system parameter adjustment.

A power system is a strongly nonlinear dynamic system, and traditional small disturbance stability analysis cannot reflect the dynamic characteristics of the system under uncertain disturbances. This paper proposes a robust practical stability region (RPSR) partitioning method that considers uncertain disturbances to address the uncertainty of wind power injections and load volatility in wind power systems. First, a disturbance model of the wind turbine generator access system is constructed on the basis of perturbation theory to investigate the effects of uncertainty disturbances on the dynamic characteristics of the system. Second, through bifurcation analysis and limit cycle (L-cycle) tracking, a comprehensive bifurcation diagram that considers the variation trend of the L-cycle amplitude is drawn. Combined with the variation trend of the L-cycle, the RPSR of the system under uncertain disturbances is partitioned. Case studies of dual-machine system and multimachine system explore the new concept of the RPSR. Last, numerical simulations are used to verify the effectiveness of the analysis results and the proposed method.

Decentralized applications (Dapps), based on smart contract technology, have been increasingly applied in various fields such as healthcare, industrial IoT, agriculture, financial services, supply chain management, and insurance. In certain complex business scenarios, blockchain may require machine learning models to assist contract business. However, on-chain computations are often costly and slow, and there are limitations on contract size. Due to the transparency of on-chain data, there are also privacy concerns regarding user data during model training and inference. To address these challenges, we propose a trusted off-chain machine learning solution that integrates ZK-SNARK and Oracle technologies. Following the principle of “off-chain computation, on-chain verification”,our approach leverages ZK-SNARK to delegate the computation tasks of machine learning models to a trusted environment under the Oracle off-chain server. This solution significantly reduces the computational costs of the blockchain. User data and models are executed off-chain, effectively safeguarding user privacy. The execution results generate zero-knowledge proofs returned for on-chain verification. We have implemented TOMLS-ZKSO and conducted relevant experiments on insurance contract business on the Chainmaker. Experimental results demonstrate the effectiveness of our approach, with model proof generation taking approximately 0.4 seconds and Dapp response time around 0.52 seconds.

In consortium blockchain, full nodes face challenges such as increased storage space usage, higher storage costs, and longer query times due to the growing amount of data. In this paper, we propose a secure and flexible solution to address these issues by transferring old blockchain data to a trusted storage system. First, we introduce a full-node data transfer scheme that packages and stores block data off-chain to tackle insufficient node storage space. Second, we combine blockchain with Transparent Data Encryption (TDE), which stores the data in ciphertext to ensure the security of off-chain data. Additionally, in order to restore the initial state of the node, we add the data recovery function. Experimental results demonstrate that this scheme can save 60% to 70% of node storage space, improve query efficiency by 20% to 50%, and ensure data security.

Federated learning(FL) promotes collaborative learning among devices in the Internet of Things (IoT), achieving privacy-preserving data sharing. However, federated learning in IoT faces challenges of insufficient trust and low efficiency. Existing methods have not effectively addressed both problems simultaneously. In this paper, we proposed a blockchain-based trustworthy and efficient federated learning framework. Firstly, we introduced a proof-of-federated-work(PoFW) consensus algorithm, designed specific block and transaction structures to address trust problem. Secondly, to alleviate the low efficiency caused by the heterogeneous environment, we introduced a deep reinforcement learning-based client selection strategy to optimize training efficiency. Extensive experiments validated that our proposed solution established trust among federated learning participants within the IoT environment, simultaneously improving learning efficiency by 1.22× to 2.63× compared to existing solutions.

Aiming at the cross-chain interoperation difficult problem between isomorphic or heterogeneous blockchain systems, we propose a cross-chain mechanism based on the cross-chain smart contract execution framework. Firstly, a cross-chain smart contract execution framework is proposed to weaken the blockchain properties of relay chains. Secondly, a reliable cross-chain scheme based on the Pedersen VSS and VRF is designed for the framework. Lastly, we conduct multiple sets of experiments on the proposed cross-chain mechanism. The experimental results proved that the proposed cross-chain mechanism is able to execute cross-chain transactions efficiently and securely with realistic feasibility. Its cross-chain interoperation TPS is about 8 times higher than that of the traditional relay chain platform BitXHub on average, and about 82% higher than that of TCIP.

Cross-chain technology has emerged to address the challenges of achieving asset and data interoperability between different blockchains. However, existing cross-chain schemes face issues related to transaction privacy leakage. Therefore, we propose a solution to solve the issues. First, we design a cross-chain model based on a relay chain for storing the information generated during the entire cross-chain process on the chain. Second, under this model, a privacy protection scheme and a relay chain key sharing scheme are proposed, employing cryptographic techniques to safeguard the privacy and security of transaction information and achieving an auditable function. Based on the above proposed schemes, a system model is designed. Finally, the proposed schemes and models are implemented, performance tests are conducted, and comparisons are made with other relay chain solutions with privacy protection capabilities, along with an analysis of security aspects. The experimental results showed that the proposed schemes and the system model are feasible.

Federated Learning (FL) has become an ideal privacy-preserving learning technique that can train a global model in a collaborative manner while preserving the privacy of local data. Federated learning in the Industrial Internet of Things (IIoT) faces the threat of data poisoning attacks. In this paper, we propose a blockchain-based federated learning framework in IIOT, which can defend against poisoning attacks. Moreover, we propose a robust aggregation algorithm to eliminate poisoned local model from malicious participants during training. Experimental results demonstrate the efficacy of the blockchain-based federated learning framework. When tested on the CIFAR-10 and Fashion-MNIST datasets, the framework achieves higher accuracy compared to the Krum algorithm by 3.57% and 0.84%, respectively.