A5080453263- Fault Detection and Control Systems
- Neural Networks and Applications
- Water Quality Monitoring and Analysis
- Machine Learning and ELM
- Image and Signal Denoising Methods
- Structural Health Monitoring Techniques
- Advanced Algorithms and Applications
- Probabilistic and Robust Engineering Design
- Energy Load and Power Forecasting
- Mineral Processing and Grinding
- Advanced Chemical Sensor Technologies
- Air Quality Monitoring and Forecasting
- Advanced Control Systems Optimization
- Fuzzy Logic and Control Systems
- Spectroscopy and Chemometric Analyses
- Stock Market Forecasting Methods
Shandong University of Science and Technology
2021-2025
Shandong Academy of Sciences
2019-2024
Qilu University of Technology
2019-2024
The development of accurate data-driven models is great significance for online monitoring and control industrial processes. In this study, a nonlinear dynamic soft-sensing algorithm measuring predicting the nitrogen oxides (NOx) emissions denitration system power plant was developed. First, processes principles were studied. Second, based on long short-term memory (LSTM) least absolute shrinkage selection operator (LASSO) proposed. dynamics process captured by LSTM, influential input...
The overflow slurry concentration (OSC) of a hydrocyclone is key performance indicator semi-autogenous ball mill crusher (SABC) system. Accurate modeling and prediction the can improve grinding efficiency product quality process. However, mechanism this process complex, gradual wear equipment leads to data drifts in measured results. To address these problems, an online updating soft sensor that combines stochastic configuration network (SCN) with dynamic forgetting factor sliding window...
The data collected from complex process industries are usually time series with considerable nonlinearities and dynamics, as well excessive redundancy. Moreover, there temporal spatial correlations between input variables key performance variables. These characteristics bring great difficulties to data-driven modeling of the To overcome problems, a new regularized spatiotemporal attention (STA)-based long short-term memory (LSTM) was developed. First, standard LSTM network an STA module...
In the paper, a new hybrid variable selection algorithm for nonlinear regression multi-layer perceptron (MLP) is proposed. The proposed applies nonnegative garrote (NNG) to compress input weights of MLP. zero dependent variables will be removed from initial dataset. Next, further carried out by extremal optimization (EO) algorithm. integrates powerful global ability NNG and accurate local search EO. Finally, two examples artificial data sets an industrial application debutanizer column are...
In this study, sensitivity analysis and nonnegative garrote (NNG) are combined to realize adaptive variable selection for gated recurrent unit (GRU). Firstly, the based on variance decomposition is used quantify correlation between each input output variable, total index of calculated. Secondly, added NNG as an weight vector achieve selection. Finally, artificial dataset with characteristic time series verify effectiveness proposed algorithm. Simulation results show that algorithm can...
In the paper, a nonlinear regression with long short-term memory (LSTM) and least absolute shrinkage selection operator (LASSO) is developed. The LSTM used to handle strong nonlinearity, dynamic time-series, LASSO applied perform input variable for LSTM. Firstly, deep neural network of constructed trained from initial data set. After that, introduced shrink weights well-trained LSTM, in which Monte Carlo method (MCM) moving block cross-validation (MBCV) are optimization. performance proposed...