12 September 2024 in IEEE Transactions on Industrial Informatics
To accurately predict the furnace temperature and flue gas oxygen content of the municipal solid waste incineration (MSWI) process under variable operating conditions, a data-driven multitarget online predictive modeling method based on an improved stochastic configuration network (SCN) is proposed in this article. This method configures the input weights and biases of the new hidden layer nodes of the SCN through a competitive guidance strategy. The model output weights are sparsely constrained using a matrix elastic net, the modeling accuracy is improved by using the correlation between the furnace temperature and flue gas oxygen content. On this basis, the output weights of the model are recursively updated using a sparse constrained direction forgetting algorithm to improve the online prediction accuracy. Finally, the performance of the proposed method is validated using historical data from the MSWI process. The experimental results show that the proposed multitarget prediction model is adaptive and can accurately predict the trends of furnace temperature and flue gas oxygen content under variable operating conditions.
19 September 2024 in Information Sciences
17 February 2025 in IEEE Sensors Journal
Accurately monitoring key parameters of the production process is one of the prerequisites for ensuring efficient and stable production. However, some key parameters are difficult to measure online in real-time, and their change mechanisms are poorly understood. This paper proposes a data-driven optimization stochastic configuration network (DO-SCN) soft sensor modeling method to build high-performance monitoring models. The DO-SCN is incrementally constructed within a newly designed configuration-evaluation-learning-modification framework. The parameters and connection ways of the model are determined via parallel construction and an adaptive supervisory evaluation mechanism. A parameter modification strategy is proposed to reduce the redundancy of the hidden layer nodes. The performance of the DO-SCN model is evaluated on six benchmark regression datasets and a furnace temperature dataset derived from a municipal solid waste incineration (MSWI) power plant. The experimental results show that the DO-SCN model has advantages in model accuracy and structural compactness, achieving the lowest RMSE and MAPE values of 3.486 and 5.811 on the MSWI dataset, respectively. It has good potential for application in production process monitoring modeling tasks.
16 June 2025 in IEEE Transactions on Automation Science and Engineering
A multi-target model can simultaneously monitor the changing trends of multiple related parameters, which plays a crucial role in achieving intelligent perception and decision-making in the production process. However, efficiently learning the non-linearity between input and output variables and exploring the correlations between multiple targets to construct high-performance monitoring models have always been challenging. This paper proposes a multi-target stochastic configuration network (SCN) model from the perspectives of network structure design and parameter group sparse constraints. This model assigns hidden layer input weights and biases using a weighted supervisory mechanism and adds a latent layer to the network structure. It facilitates multi-target balanced learning while exploiting inter-target correlations to improve modeling accuracy. The output weights of the hidden and latent layers are collaboratively constrained by constructing a generalized sparse matrix elastic net to induce their group sparsification. An alternating optimization algorithm based on matrix decomposition is designed to iteratively solve the model parameters. The performance of the proposed model is experimentally validated on benchmark multi-target datasets and historical data of municipal solid waste incineration process. The results demonstrate that the proposed model achieves superior accuracy and exhibits potential for multi-target modeling applications.
2 February 2026 in IEEE Transactions on Industrial Informatics
Industrial production data often exhibits concept drift, resulting in the performance degradation of the models constructed offline. This article proposes an adaptive dynamic multitarget regression model based on stochastic configuration networks to achieve precise monitoring of multiple correlated targets in industrial processes. First, a slack supervisory mechanism is developed to adaptively optimize the model hidden layer parameters. The correlation and differentiation between multiple targets are jointly learned through parameter adaptive sharing. Second, an online collaborative updating strategy of model structure and parameters is designed based on a regularized projection algorithm and node effectiveness analysis. The hidden layer nodes and output weights of the model are dynamically updated to avoid model mismatch. Finally, the performance of the proposed model is validated using synthetic, benchmark, and real industrial datasets. The experimental results demonstrate that the model can accurately monitor multiple targets and effectively adapt to variable operating conditions.
20 December 2025 in Expert Systems with Applications