Real-Time Detection and Short-Term Prediction of Blast Furnace Burden Level Based on Space-Time Fusion Features

Abstract

Real-time, continuous and accurate blast furnace burden level information is of great significance for controlling the charging process, ensuring a smooth operation of a blast furnace, reducing energy consumption and emissions and improving blast furnace output. However, the burden level information measured by conventional mechanical stock rods and radar probes exhibit problems of weak anti-interference ability, large fluctuations in accuracy, poor stability and discontinuity. Therefore, a space-time fusion prediction and detection method of burden level based on a long-term focus memory network (LFMN) and an efficient structure self-tuning RBF neural network (ESST-RBFNN) is proposed. First, the space dimensional features are extracted by the space regression model based on radar data. Then, the LFMN is designed to predict the burden level and extract the time dimensional features. Finally, the ESST-RBFNN based on a proposed fast eigenvector space clustering algorithm (ESC) is constructed to obtain reliable and continuous burden level information with high accuracy. Both the simulation results and industrial verification indicate that the proposed method can provide real-time and continuous burden level information in real-time, which has great practical value for industrial production.

High-Precision Real-Time Detection of Blast Furnace Stockline Based on High-Dimensional Spatial Characteristics

“…The proposed SDSO-RBFNN method has little difference with the measured stockline data of the mechanical probe under normal and abnormal working conditions, and the measured values are closely distributed on both sides of the reference line, which is not only accurate, but also suitable for various abnormal working conditions of the blast furnace. Selecting the 12 h real operation data of the proposed method, the VS-RBFNN [20] method, and the ESST-RBFNN [21] method on a 2650 m 3 blast furnace, a long-term operation effect diagram, as shown in Figure 15, is drawn to compare the detection accuracy of the three detection methods. Obviously, the VS-RBFNN method is similar to the traditional RBFNN method, showing inhibition to the predicted stockline, and the predicted stockline fluctuates slightly at 1.55 m. Compared with the real stockline, the prediction of the stockline under normal working conditions is on the high side, and the prediction of stockline under abnormal working conditions has great deviation.…”

“…However, when predicting the low stockline, the predicted stockline frequently changes violently, which is caused by the failure of this method to extract the spatial features of radar data effectively, and the poor stability of the clustering algorithm. Under normal working conditions, the predicted stockline of the proposed method is basically equal to the actual material level, and under abnormal working conditions, the deviation between the predicted stockline and the actual stockline is small, and the predicted stockline data Selecting the 12 h real operation data of the proposed method, the VS-RBFNN [20] method, and the ESST-RBFNN [21] method on a 2650 m 3 blast furnace, a long-term operation effect diagram, as shown in Figure 15, is drawn to compare the detection accuracy of the three detection methods. Obviously, the VS-RBFNN method is similar to the traditional RBFNN method, showing inhibition to the predicted stockline, and the predicted stockline fluctuates slightly at 1.55 m. Compared with the real stockline, the prediction of the stockline under normal working conditions is on the high side, and the prediction of stockline under abnormal working conditions has great deviation.…”

“…For example, in [20], Chen et al proposed the dynamic self-growing RBFNN, based on the data distribution features of a radar probe, which combines the advantages of two probes online in real-time, improves the accuracy of stockline measurement on the premise of ensuring the continuity of stockline measurement, and has certain practical value. In [21], Chen et al extracted the spatiotemporal features of radar probe data, and based on the fast clustering algorithm of feature vector space, constructed the efficient structure self-tuning RBFNN, and realized the fusion measurement of two kinds of probe.…”

Adversarial Robustness Enhancement for Deep Learning-Based Soft Sensors: An Adversarial Training Strategy Using Historical Gradients and Domain Adaptation

“…Soft-sensor methods establish models by utilizing mathematical mapping relationships between easily measurable process variables (referred to as “auxiliary variables”) and key target variables that are difficult to measure (referred to as “primary variables”) [ 4 , 5 ]. Such soft-sensor models effectively emulate the functionality of hardware sensors used to predict primary variables, unaffected by the complex environments of industrial processes, and inherently possess advantages in terms of universality, flexibility, and cost-effectiveness [ 6 , 7 ].…”