Research On Nuclear Pipeline Fault Diagnosis Technology Based On Quantum Computing

In nuclear power plants,nuclear pipelines are very important transmission devices,but because they often work in harsh environments and are prone to failures that can lead to safety problems,it is especially important to diagnose the types of nuclear pipeline failures quickly and accurately.Quantum computing theory integrates quantum mechanics and classical computing science,and can effectively solve some uncertain problems by using quantum properties.Therefore,this thesis combines quantum computing with machine learning in traditional fault diagnosis techniques to perform fault diagnosis of vibration signals of nuclear pipelines.The specific research is as follows:(1)In thesis,we first introduce some common faults of nuclear pipelines,as well as their causes and consequences,and show the state diagrams in both time and frequency domains,and then use Fourier transform to convert the original vibration signals into frequency domain signals and extract their features from both time and frequency domains.In order to reduce the diagnostic workload and improve the diagnostic accuracy,the extreme random tree algorithm is used to filter the features.(2)For the situation that the fault samples of some nuclear pipelines are small and uneven,this thesis uses generative adversarial network and quantum generative adversarial network for sample generation,and after the comparison of the loss function convergence,it is found that quantum generative adversarial network has more advantages.Then the generated data is mixed with the original data for fault diagnosis and compared with the diagnosis results before the generated data is not mixed.After the experiment,it is shown that the quantum generative adversarial network learns the feature distribution of the original data set better and can solve the problem of unbalanced sample distribution more effectively.(3)For the nuclear pipeline fault diagnosis based on quantum computing,this thesis combines quantum computing theory with classical BP neural network model and classical Support Vector Machine model respectively,and designs quantum line and quantumization scheme to fuse into two new model algorithms.And the classical genetic algorithm and quantum genetic algorithm are used to optimize the quantum BP neural network model,and the optimization results are used as the initial parameters of quantum BP neural network quantum circuit to form two new models.Then the fault diagnosis of the vibration characteristic signal of nuclear pipeline using the above models,and finally the accuracy of the above models are summarized and compared,and it is concluded that after combining with quantum computing theory,the accuracy of all kinds of classical machine learning algorithms are improved,and the optimization efficiency and effect of quantum genetic algorithm are better than classical genetic algorithm.