Quanli Liu

Data-driven based model for flow prediction of steam system in steel industry

The steam system is one of the main energy systems in steel industry, and its operational scheduling plays a crucial role for energy utility and resources saving. For a reasonable resources operation, the accurate prediction of steam flow is required. Considering the large amount of production data in energy system, a data-driven based model is proposed to perform a time series prediction for steam flow, in which a Bayesian echo state network (ESN) is established. This method combines Bayesian theory with ESN to obtain optimal output weight via maximizing the posterior probability density of the weights to avoid over-fitting in the training process of sample data. To pursue optimized hyper-parameters in the proposed Bayesian ESN, the evidence framework based on sample data is further adopted in this work. Experimental results using the real production data from Shanghai Baosteel show the validity and practicality of the proposed data-driven based model in providing scientific decision guidance for the steam system.

A parallel immune algorithm for traveling salesman problem and its application on cold rolling scheduling

Parallel computing provides efficient solutions for combinatorial optimization problem. However, since the communications among computing processes are rather cost-consuming, the actual parallel or distributed algorithm comes with substantial expenditures, such as, hardware, management, and maintenance. In this study, a parallel immune algorithm based on graphic processing unit (GPU) that originally comes to process the computer graphics in display adapter is proposed. Genetic operators and a structure of vaccine taboo list are designed, and the internal memory utility of GPU structure is optimized. To verify the effectiveness and efficiency of the proposed algorithm, various middle-scale traveling salesman problems (TSP) are employed to demonstrate the potential of the proposed techniques. The simulation examples demonstrate that the developed method can greatly improve the computing efficiency for solving the TSP, and the results are more remarkable when the scale of TSP becomes higher. Furthermore, the derived algorithm is verified by a practical application in steel industry that arranges the cold rolling scheduling of a batch of steel coils.

Hybrid Neural Prediction and Optimized Adjustment for Coke Oven Gas System in Steel Industry

An energy system is the one of most important parts of the steel industry, and its reasonable operation exhibits a critical impact on manufacturing cost, energy security, and natural environment. With respect to the operation optimization problem for coke oven gas, a two-phase data-driven based forecasting and optimized adjusting method is proposed, where a Gaussian process-based echo states network is established to predict the gas real-time flow and the gasholder level in the prediction phase. Then, using the predicted gas flow and gasholder level, we develop a certain heuristic to quantify the users optimal gas adjustment. The proposed operation measure has been verified to be effective by experimenting with the real-world on-line energy data sets coming from Shanghai Baosteel Corporation, Ltd., China. At present, the scheduling software developed with the proposed model and ensuing algorithms have been applied to the production practice of Baosteel. The application effects indicate that the software system can largely improve the real-time prediction accuracy of the gas units and provide with the optimized gas balance direction for the energy optimization.

Effective Noise Estimation-Based Online Prediction for Byproduct Gas System in Steel Industry

A rapid and accurate prediction of byproduct gas flow in steel industry can help not only to become aware of the operational situations of gas system, but it also provides the energy scheduling workers with sound decision-making mechanisms. In this study, a least square support vector machine (LS-SVM) model based on online hyperparameters optimization is proposed, where the variance of effective noise of the sample is estimated, while a conjugate gradient algorithm is developed to optimize the width of Gaussian kernels and the regularization factor. To assess the quality of the proposed method, we experiment with a test function affected by additive noise and an industrial gas flow data from Shanghai Baosteel Company Ltd. A series of comparative experiments are reported as well. The results demonstrate that the proposed method shows the shortest computing time while ensuring the prediction accuracy. These two features make the approach applicable to real-time prediction of gas flow in steel industry.

Granular-computing based hybrid collaborative fuzzy clustering for long-term prediction of multiple gas holders levels

Linz-Donawitz converter Gas (LDG), regarded as an essential secondary energy resource, plays a significant role for the entire production process of steel industry. In a LDG system, the gas holders are crucial equipment for temporary energy storage and buffers connecting with the gas generation units and the gas users. The accurate long-term prediction for the holders levels of such a system would be very necessary for energy scheduling and its optimal decision making. Given the practical characteristics of the LDG system in a steel plant, a granular-computing (GrC)-based hybrid collaborative fuzzy clustering (HCFC) algorithm is proposed in this study for the long-term prediction of the multiple holders levels. The hybrid structure considers the features regarding to a gas holder, of which the horizontal part elaborates the mutual influences among different time spaces of a holder level, while the vertical one describes them among the influence factors (denoting the gas generation units or the users). Then, the modeling algorithm is also explicitly derived in this study. To verify the performance of the proposed approach, two groups of simulation are carried out by employing the real-world industrial data coming from this plant, in which the single-output method and the iterative computing-based one are comparatively analyzed. The results indicate that the proposed approach provides a remarkable accuracy for such an industrial application.

Design and implementation of MVB protocol analyzer

The Train Communication Network (TCN) is a high-performance local area network for information transfer of modern train control, state detection, fault diagnosis and passenger service and so on. The Multifunction Vehicle Bus (MVB) is a critical part of the TCN and the design of the MVBs protocol analysis tool plays a very important function. Supported by the friendly Windows graphics user interface, a powerful MVBs protocol analysis tool is developed in this paper and this tool fills in a vacancy of the technical application presently. Starting from the MVB protocol, the frame construction of the MVB protocol analyzer and the function design of each module of up level computer software are introduced. Finally this protocol analyzer is applied to the developing and debugging process of state detection, fault diagnosis and vehicle equipment of the TCN.

Prediction intervals for industrial data with incomplete input using kernel-based dynamic Bayesian networks

Reliable prediction intervals (PIs) construction for industrial time series is substantially significant for decision-making in production practice. Given the industrial data feature of high level noises and incomplete input, a high order dynamic Bayesian network (DBN)-based PIs construction method for industrial time series is proposed in this study. For avoiding to designate the amount and type of the basis functions in advance, a linear combination of kernel functions is designed to describe the relationships between the nodes in the network, and a learning method based on the scoring criterion—the sparse Bayesian score, is then reported to acquire suitable model parameters such as the weights and the variances. To verify the performance of the proposed method, two types of time series which are the classical Mackey-Glass data mixed by additive noises and a real-world industrial data are employed. The results indicate the effectiveness of our proposed method for the PIs construction of the industrial data with incomplete input.

Data-Based Predictive Optimization for Byproduct Gas System in Steel Industry

In light of significant complexity of the byproduct gas system in steel industry (which limits an ability to establish its physics-based model), this study proposes a data-based predictive optimization (DPO) method to carry out real-time adjusting for the gas system. Two stages of the method, namely the prediction modeling and real-time optimization, are involved. At the prediction stage, the states of the optimized objectives, the consumption of the outsourcing natural gas and oil, the power generation and the tank levels, are forecasted based on a proposed mixed Gaussian kernel-based prediction intervals (PIs) construction model. The Jacobian matrix of this model is represented by a kernel matrix through derivation, which greatly facilitates the subsequent calculation. At the second stage, a rolling optimization based on a mathematical programming technique involving continuous and integer decision-making variables is developed via the prediction intervals.

Granular Computing Concept Based Long-Term Prediction of Gas Tank Levels in Steel Industry

Abstract The converter gas, especially Linz Donawitz converter gas (LDG), is one of the most significant secondary energy resources in a large scale steel plant. For such a thing, the accurate prediction for the gas tank levels largely contributes to the energy optimization operations. Taking the LDG system of a steel plant in China into consideration, a regression model based on the Granular Computing (GrC) is proposed in this study to provide a long-term prediction for the LDG tank levels, in which the data segments are entirely considered for the prediction horizon extension rather than the generic data point-oriented modeling. For being more practical, this study specially granulates the initial data with regard to industrial semantic meaning. And, different from ordinary time series analysis, this method considers the factor related to the gas tank levels. Bearing this in mind, the fuzzy rules by adopting a fuzzy C-means based clustering is established. To verify the effectiveness of the proposed method, a series of practical experiments by using the industrial data coming from the energy data center of this plant are conducted, and the results demonstrate the practicability of the proposed approach.

A multiple surrogates based PSO algorithm

Particle swarm optimization (PSO) usually requires a large number of fitness evaluations to obtain a sufficiently good solution, which poses an obstacle for applying PSO to computationally expensive problems. In this paper, a multiple surrogates based PSO (MSPSO) framework is proposed, which consists of an inner loop optimization and an outer one. In the outer loop optimization, a PSO algorithm is used in both the optimization mode and the sampling one. In the inner loop optimization, a multiple surrogate based parallel optimization strategy is designed. Furthermore, the search history and the possible solutions from the outer loop optimization are provided for the inner one, and the result of the inner loop optimization is employed to guide the search of the outer one. To verify the performance of the proposed approach, a number of numerical experiments are conducted by using ten benchmark test functions and three time series regression modeling problems. The results indicate that the proposed framework is capable of converging to a good solution for the low-dimensional, non-convex and multimodal problems.