Data Processing in Digital Twin

Data Processing in Digital Twin

Data processing means extracting useful information from a large volume of incomplete, unstructured, noisy, fuzzy, and random raw data.
Stage:

• Data preprocessed: removing redundant, irrelevant, misleading, duplicate, and inconsistent data. The relevant technologies include data cleaning, data compression, data smoothing, data reduction, data transformation, etc.
• Data analysis :
  • statistical methods: descriptive statistics (e.g., frequency, central tendency, discrete tendency, and distribution analysis), hypothesis testing (e.g., u-test, t-test, χ2 test, and F-test), correlation analysis (e.g., linear correlation, partial correlation, and distance analysis), regression analysis (e.g., linear regression, curve regression, binary regression, and multiple regression), clustering analysis (e.g., partition clustering, hierarchical clustering, density-based clustering, and grid-based clustering), discriminant analysis (e.g., maximum likelihood, distance discriminant, bayesian discriminant, and fisher discriminant), dimension reduction (e.g., principal component analysis, and factor analysis), time series analysis, etc.
  • neural network methods: forward neural network (i.e., neural network based on gradient algorithm such as BP network, optimal regularization method such as SVM, radial basis neural network, and extreme learning machine neural network), feedback network (e.g., Hopfield neural network, Hamming network, wavelet neural network, bidirectional contact storage network, and Boltzmann machine), and self-organizing neural network (e.g., self-organizing feature mapping and competitive learning).
  • deep learning
  • database methods: multidimensional data analysis and OLAP methods.