Search

In this paper, we address the swing-down control of the Acrobot, a two-link planar robot operating in a vertical plane with only the second joint being actuated. The control objective is to rapidly stabilize the Acrobot around the downward equilibrium point, with both links in the downward position, from almost all initial states. Under the conditions of no friction and measurability of only the angle and angular velocity of the actuated joint, we present a sinusoidal-derivative (SD) controller. This controller consists of a linear feedback of the sinusoidal function of the angle of the actuated joint and a linear feedback of its angular velocity. We prove that the control objective is achieved if the sinusoidal gain is greater than a negative constant and the derivative gain is positive. We establish crucial relationships between the relative stability of the Acrobot under the SD controller and its physical parameters, presenting analytically all optimal control gains. These gains minimize the real parts of the dominant poles of the linearized model of the resulting closed-loop system around the downward equilibrium point. We demonstrate that the resulting dominant closed-loop poles can be double complex conjugate poles, or a quadruple real pole, or a triple real pole, depending on the Acrobot's physical parameters. Simulation studies indicate that the proposed SD controller outperforms the derivative (D) controller in rapidly stabilizing the Acrobot at the downward equilibrium point. • The control goal is to rapidly stabilize the Acrobot in a downward position. • A sinusoidal-derivative (SD) controller has been developed to meet this objective. • The challenge of minimizing the real parts of the dominant poles is addressed. • Comprehensive analytical solutions for the optimal design of both gains are provided. • The introduced SD controller demonstrates improved control performance compared to other approaches. [ABSTRACT FROM AUTHOR]

This paper presents an optimization design method for a two-dimensional (2D) modified repetitive control system (MRCS) with an anti-windup compensator. Using lifting technology, a 2D hybrid model of the MRCS considering actuator saturation is established to describe the control and learning of the repetitive control. A linear-matrix-inequality (LMI)-based sufficient condition is derived to ensure the stability of the MRCS. Two tuning parameters, the selection of which is critical to the system design, are used in the LMI to adjust the control and learning, and hence the reference-tracking performance. A new cost function, developed through time domain analysis, directly evaluates the control performance of the system without calculating control errors, thus reducing the optimization time. Based on this cost function, an adaptive multi-population particle swarm optimization algorithm is presented to select an optimal pair of tuning parameters in which multiple populations cooperatively search in non-intersecting search intervals. An anti-windup term is added between the low-pass filter and the time delay in the modified repetitive controller to mitigate the undesirable effect of actuator saturation on system performance and stability. Simulations and experiments on the speed control of a rotation control system demonstrate the validity of the approach. • Evaluate control performance only with the information of controller gains. • An optimization algorithm is used to optimize tuning parameters. • An anti-windup compensator mitigates the undesired influence of actuator saturation. • Separate design of repetitive control system and anti-windup compensator. [ABSTRACT FROM AUTHOR]

The analysis of convergence of extended state observers (ESOs) requires the total disturbance to be differentiable. However, this requirement is not satisfied in many control engineering practices. In this paper, we attempt to analyze the convergence of ESOs for nonlinear systems with non-differentiable uncertainties. A decomposition method is first presented to divide the non-differentiable total disturbance into a differentiable signal and a bounded but non-differentiable signal. Based on this decomposition, we give out the convergence of both nonlinear and linear ESOs (NLESO/LESO), low- and high-power ESOs (LPESO/HPESO), and fixed-time ESO (FxESO). We also derive the explicit formulas for the estimation errors of these ESOs. Simulations and experiments demonstrate the correctness of the analysis results. • The perturbation of ESO is established based on the proposed decomposition method. • The convergence of typical ESOs is analyzed when disturbance is non-differentiable. • Simulation and experimental testing of ESOs' convergence performance are carried out. [ABSTRACT FROM AUTHOR]

This paper presents a high-order low-pass filter for the equivalent-input-disturbance (EID) approach to improving the disturbance-rejection performance. The configuration characteristic of the presented filter clearly explains the reason why the disturbance-rejection performance is improved and provides a guideline to design it. Using the presented filter to replace the conventional filter derives a high-order EID (HEID) approach. It is easy to apply the small-gain theorem in deriving stability conditions of the HEID-based control system. Moreover, the presented filter is proved to be better than the conventional one. Finally, a comparison shows the validity and superiority of the presented method. And a simulation result shows that the HEID approach is easily extended in a multiple-input, multiple-output system even with effects of a white noise and parameter uncertainties. • The presented method decreases the estimation error caused by the phase lag, and thus improves the disturbance-rejection performance. • It is easy to derive the frequency response of the presented filter. And analyzing the frequency response, we prove that the phase lag of the presented filter is smaller than that of the first-order low-pass filter, which has not been analyzed in previous studies. • Compared with the previous method of designing a high-order low-pass filter, the disturbance-rejection performance of our method is excellent and the designed high-order low-pass filter is simple. • The configuration of the HEID-based control system is simple. And it is easy to extend the small-gain theorem to simply derive the stability conditions of the HEID-based control system. • The disturbance-rejection mechanism of the HEID approach is clear and is explained via transforming the configuration of the HEID estimator into an equivalent configuration that contains two low-pass filters, which provides a design guideline for the high-order low-pass filter. [ABSTRACT FROM AUTHOR]

Exogenous disturbances largely affect the control performance of systems with time delays. This study considers a control problem of rejecting a disturbance in a PI control system for a time-varying state-delay plant. The equivalent-input-disturbance (EID) approach is integrated in a PI control system. An EID estimator estimates the overall effects of a time-varying delay and a disturbance. An EID estimate is combined into a PI control law to improve control performance. A less-conservative stability condition of the control system is derived using a Lyapunov–Krasovskii functional together with the Jensen's integral inequality and the reciprocally convex combination lemma. Parameters of the controllers in the system are calculated using the condition. Engine idle speed control is used to verify the effectiveness of this approach. Compared with the generalized extended-state observer and the sliding-mode control methods, our method reduced the tracking error to about one third and one sixth, respectively. This demonstrates the validity and superiority of our method. • The equivalent-input-disturbance (EID) approach is incorporated into a PI control system to improve disturbance-rejection performance. • The EID approach is also extended to deal with a time-varying delay. • A less-conservative stability condition is derived to design the gains of the PI controller and state observer. • A comparison of our method with the GESO and SMC methods for engine idle speed control show the validity of this method and its superiority over others. [ABSTRACT FROM AUTHOR]

This paper presents a new method that uses the equivalent-input-disturbance (EID) approach to estimate the wind load of a seismically isolated building. To verify the control performance of wind-resistant design methods and devices, wind-load data are important for structural design of skyscrapers. However, wind loads cannot be measured directly and is usually estimated based on structural responses. Previous studies focused on a building model in which the damping coefficient is proportional to a linear combination of mass and stiffness (referred to as Rayleigh damping). These existing methods cannot be directly applied to a seismically isolated building because its damping is not Rayleigh damping. Moreover, a structural model usually contains uncertainties caused by modeling errors that cause the degradation of estimation performance. To solve these problems, this paper combines the EID approach with a robust control strategy to design a wind-load estimator for a seismically isolated building with uncertainties. First, the configuration of an EID-based estimation system is shown. Next, a robust wind-load estimator is presented. Then, simulations are carried out to demonstrate the validity of the presented method using wind-load data obtained from a wind-tunnel test. An 11-degree-of-freedom model of a seismically isolated building is used for numerical verification. The results show that, even if a model contains 20% uncertainties, the presented method accurately estimates the wind load of a seismically isolated building. [ABSTRACT FROM AUTHOR]

Highlights • A new method is developed to design LQR controller for active structural control. • Both the acceleration and the drift are considered based on Bayesian optimization. • Weights of LQR for 11 DoF building model are automatically optimized by this method. Abstract This paper presents a method for the automatic selection of weighting matrices for a linear-quadratic regulator (LQR) in order to design an optimal active structural control system. The weighting matrices of a control performance index, which are used to design optimal state-feedback gains, are usually determined by rule of thumb or exhaustive search approaches. To explore an easy way to select optimal parameters, this paper presents a method based on Bayesian optimization (BO). A 10-degree-of-freedom (DOF) shear building model that has passive-base isolation (PBI) under the building is used as an example to explain the method. A control performance index that contains the absolute acceleration, along with the inter-story drift and velocity of each story, is chosen for the design of the controller. An objective function that contains the maximum absolute acceleration of the building is chosen for BO to produce optimal weighting matrices. In the numerical example, a restriction on the displacement of the PBI is used as a constraint for the selection of weighting matrices. First, the BO method is compared to the exhaustive search method using two parameters in the weighting matrices to illustrate the validity of the BO method. Then, thirty-three parameters (which are automatically optimized by the BO method) in the weighting matrices are used to elaborately tune the controller. The control results are compared to those for the exhaustive search method and conventional optimal control, in terms of the control performance of the relative displacement, absolute acceleration, inter-story-drift angle, and the story-shear coefficient of each story. The damping ratio for each mode, and the control energy and power are also compared. The comparison demonstrates the validity of the method. [ABSTRACT FROM AUTHOR]

A new method of active structural control, which suppresses vibrations in civil structures due to seismic shocks, has been developed. It is based on the equivalent-input-disturbance (EID) approach, which estimates the effect of a seismic shock and produces an equivalent control signal on the control input channel to compensate for it. A system designed by this method can be viewed as a conventional state-feedback control system with an EID estimator plugged in. Unlike conventional control systems, this one has two degrees of freedom, which yields better control performance. Simulations on a model of a ten-degree-of-freedom building demonstrated the validity of the method. In addition, the effect of the parameters of the low-pass filter in the EID estimator on the vibration suppression performance was examined. A comparison revealed that this method is superior to a linear-quadratic regulator and sliding-mode control. [ABSTRACT FROM AUTHOR]

This paper presents a new system configuration and a design method to improve control performance for a system with an input time delay and disturbances. The equivalent-input-disturbance approach is extended to handle a time-delay system. It is combined with the Smith predictor to reject disturbances. A delay-dependent stability condition is devised in terms of a matrix inequality by using the free-weighting matrix approach. The gain of the observer is designed by applying the cone complementary linearization method to the matrix inequality. A numerical example demonstrates the validity of the method. [ABSTRACT FROM AUTHOR]

Abstract: This paper concerns the design of a robust observer-based modified repetitive-control system for a class of strictly proper plants with periodic time-varying uncertainties. It exploits the inherent structure characteristics of repetitive control to convert the design problem to a robust stabilization problem for a continuous-discrete two-dimensional system. The singular-value decomposition and Lyapunov stability theory are used to derive a linear-matrix-inequality-based design algorithm for the parameters of control and state-observer gains. Two tuning parameters are introduced to perform the preferential adjustment of control and learning. A numerical example illustrates the adjusting procedure and demonstrates the validity of the method. [Copyright &y& Elsevier]

Abstract: This paper describes an autoselection method for the pedal load of an electric cart, and the control system. The method checks the driver’s physical condition just before the start of driving, calculates the driver’s rating of perceived exertion, and uses that to automatically select a suitable pedal load. To ensure that the level of physical exertion is safe, the largest pedal load is determined on the basis of Borg’s CR10, the Karvonen formula, a pedaling experiment, and statistical data. A driver’s degree of fatigue is estimated from his heart rate, and a pedal load (between the largest and almost zero) is automatically selected. In addition, a dynamic output-feedback cart control system was designed that guarantees the stability of the cart system for any allowable pedal load. A stability condition is derived by employing the concept of dynamic parallel distributed compensation. Finally, experiments demonstrate the validity of the method. [Copyright &y& Elsevier]

Abstract: This paper concerns motion planning and tracking control for an acrobot. We first introduce an artificial friction torque in order to construct a downward trajectory, and rewind it to make an upward trajectory. Then, we combine the upward trajectory with a stabilizing trajectory to make a complete trajectory. Finally, we use the pole assignment method to design a tracking controller that makes the acrobot exponentially track the whole trajectory. This enables the acrobot to be swung up from the straight-down position and stabilized at the straight-up position. Unlike the most commonly used switching stabilization control methods, the strategy presented here features a single controller for motion control in the whole motion space. It is simple and efficient. Simulation results demonstrate the validity of the method. [Copyright &y& Elsevier]

Abstract: This paper concerns the problem of designing a modified repetitive-control system for a class of strictly proper plants. Repetitive control involves two types of actions: control and learning; but the insertion of a low-pass filter in a modified repetitive controller, which is employed to guarantee the stability of the system, mixes the two actions together. In this paper, a continuous–discrete two-dimensional model is first constructed. Next, the continuity of repetitive control and Lyapunov stability theory are applied to the model to establish two linear-matrix-inequality (LMI) based sufficient stability conditions, one for the design of the cutoff angular frequency and one for the design of the feedback gains. The features of these conditions are exploited to develop an iterative algorithm that searches for the best combination of the maximum cutoff angular frequency of the low-pass filter and the feedback gains. A numerical example illustrates the design procedure and demonstrates the validity of the method. [Copyright &y& Elsevier]

• This paper uses an SMC controller to suppress the estimation error neglected by the conventional EID approach. The application of the SMC improves the disturbance-rejection performance of the EID approach. • A variable switching gain is designed to reduce the chattering. The switching gain is dependent on the estimation error and is very small. Thus, the chattering is smaller than the conventional SMC method. • A comparison shows that the disturbance-rejection performance of our method is better than either the conventional EID approach or the conventional SMC method while the chattering is almost free. This paper uses the sliding-mode control (SMC) to compensate for an estimation error caused by the filter in an equivalent-input-disturbance (EID) compensator. The application of the SMC brings another shortcoming: The chattering problem. To solve such a problem, this paper makes use the features of the EID approach and presents a method of designing a small switching gain based on the estimation error. Combining the SMC, the EID approach, and the switching-gain design law, this paper presents a new system configuration. The presented SMC-based-EID control system not only improves the disturbance-suppression performance of the EID approach but also solves the chattering problem in the SMC. A comparison shows the validity and superiority of our method. Furthermore, the robustness of the disturbance-rejection performance with respect to measurement noise is also shown by the example. [ABSTRACT FROM AUTHOR]

This paper addresses the weighted sensitivity design of multivariable proportional–integral–derivative (PID) controllers for multi-input multi-output (MIMO) processes. First, the PID reference tracking problem is formulated into an H ∞ PID control problem by using the weighted sensitivity synthesis, and the problem is further casted into an H ∞ SOF control problem. Second, based on the idea of T–K iteration, a new iterative LMI algorithm has been developed to locally solve the optimal H ∞ SOF control problem. Then, a design procedure is presented for weighted sensitivity design of multivariable PID Controllers for MIMO processes. Finally, three application examples are provided to show the effectiveness and good performance of our method for practical industrial processes. • The weighted sensitivity design of multivariable PID controllers for MIMO processes is formulated into an H ∞ SOF control problem. • A new iterative LMI algorithm has been developed to solve the optimal H ∞ SOF control problem. • Our method provides a locally optimal solution for the weighted sensitivity design of multivariable PID controllers. [ABSTRACT FROM AUTHOR]

Highlights • A model with sensor faults and disturbance is transformed by using a first-order filter. • Current sensor faults and disturbance are decoupled by coordinate transformation. • Sliding-mode observers are designed to estimate faults and disturbance. Abstract Current sensor faults leads to system degradation of a permanent magnet synchronous motor (PMSM) drive. Accurate estimation of the faults for fault-tolerant control has become an important issue. However, unknown disturbance in measurement circuits greatly aggravates the difficulty of the estimation and seriously influences the performance of the estimation. This paper presents a new estimation method of sensor faults and unknown disturbance in current measurement circuits for an interior PMSM drive system. The system model, which contains faults of virtual sensors for α β axis currents and the unknown disturbance, is first constructed. Then, the model is transformed into a model with actuator faults by a first-order low-pass filter. Moreover, the faults and disturbance in the model are decoupled by coordinate transformation. And sliding-mode observers are designed based on the new decoupled system. Single and simultaneous faults of current sensors and an unknown disturbance are accurately estimated based on sliding-mode equivalent control methodology. An experiment is carried out to demonstrate the validity and effectiveness of the method. [ABSTRACT FROM AUTHOR]

This paper concerns the problem of tracking a position-related periodic signal for an uncertain nonlinear rotational system. Applying the additive-state-decomposition (ASD) technique decomposes the original uncertain nonlinear system into two subsystems: A primary linear time-invariant system that performs spatial repetitive control, and a secondary nonlinear system that ensures robust stability of the whole system with respect to unknown nonlinear dynamics and exogenous disturbances. This method has four features: (i) The ASD technique provides an effective way to deal with the coupling between the spatial periodicity and other factors (temporal and state-dependent uncertainties and disturbances) in a rotational system. This technique ensures satisfactory tracking and disturbance rejection performance; (ii) A real-time digital implementation of a spatial repetitive controller enables repetitive control with an exact period in the position. This ensures the synchronous execution of repetitive control in the position domain and robust stability in the time domain; (iii) The method of accurate phase compensation in a spatial repetitive controller leads to a significant improvement in steady-state tracking performance; and (iv) An extended-state observer precisely estimates the overall effect of disturbances. Integrating a disturbance estimate into a control input effectively compensates for disturbances. Stability criteria and design algorithms are presented in detail. Finally, experiments of a rotational speed-control system demonstrate the effectiveness of this method. [ABSTRACT FROM AUTHOR]

This paper considers the problem of designing a state-feedback controller with both passive base isolation (PBI) and active structural control (ASC). In order to improve control performance, state-feedback gains are designed based on the linear quadratic regulator (LQR) method that optimizes a new performance index containing absolute acceleration, and inter-story drifts and velocity. Simulations on a model of an eleven degree-of-freedom shear building for four earthquake accelerograms are used to verify this method. Comparison studies show that, compared with PBI, the combination of PBI and ASC improves control performance; and this method yields better control results than the conventional ASC, which considers relative displacement and relative velocity of each story. The results are also discussed from the viewpoint of control system structure regarding the location of system zeros. In addition, the effect of weights in the LQR on control performance is discussed. A method for selecting the weights is presented by using the infinity norm of a system as a criterion to visualize their effect. [ABSTRACT FROM AUTHOR]

Unknown disturbances such as parameter perturbations, PM demagnetization fault, and external disturbances degrade the control performance of a permanent magnet synchronous motor (PMSM) drive system. This paper presents a novel model-free super-twisting sliding-mode control (MFSTSMC) method based on an extended sliding-mode disturbance observer (ESMDO) for a PMSM drive system. First, a novel model-free ultra-local model (ULM) is presented for PMSM with unknown disturbances. Next, the speed loop and d − q axis current loops controllers are designed by using the MFSTSMC method. This control strategy drives the system state quickly converges to the reference value in finite time. Then, an ESMDO is constructed to estimate the unknown parts of the novel model-free ULM. The estimated unknown part is fed back to the MFSTSMC controller to perform feedforward compensation, which effectively eliminates the influence caused by unknown disturbances. Moreover, the stability of the designed observer-based controller is discussed using a quasi-quadratic Lyapunov approach. Finally, comparisons with other methods demonstrate that the presented method improves the dynamic response, anti-disturbance, and robustness of the PMSM drive system. • The novel ultra-local model (ULM) is derived for PMSM. • The model-free super-twisting sliding-mode control method is presented. • The extended sliding-mode disturbance observer (ESMDO) is designed. • The unknown part of ULM is estimated by ESMDO. • The presented method performs fault-tolerant control of PMSM drive system. [ABSTRACT FROM AUTHOR]

Iron ore sintering is the second-most energy-consuming process in steelmaking. The main source of energy for it is the combustion of carbon. In order to reduce energy consumptions and improve industrial competitiveness, it is important to improve carbon efficiency. Reliable online prediction of the carbon efficiency would be extremely beneficial for making timely adjustments to the process to improve it. In this study, the comprehensive carbon ratio (CCR) was taken to be a measure of the carbon efficiency; and a soft sensing system was built to make an online estimation of the CCR. First, the sintering process was analyzed, and the key characteristics of the process parameters were extracted. Then, the configuration of the soft sensing system was devised based on the characteristics of the process. The system consists of three parts: an image selection, an image segmentation, and a hybrid just-in-time learning soft sensor (HJITL-SS). First, an image selection method was devised to automatically select the key frames (KFs) from the video taken at the discharge end of the sintering machine. Then, a genetic-algorithm-based fuzzy c-means clustering method was devised to extract feature parameters from the KFs. Finally, an HJITL-SS, which consists of online and offline submodels, was devised to estimate the CCR using the extracted feature parameters as inputs. Actual run data were used to verify the validity of our system. Accuracy, overfitness, and error distribution of the HJITL-SS, offline, and JITL-based soft sensing methods were compared, which show the validity of the HJITL-SS. The actual run results also show the validity of the soft sensing system with 97% of the actual runs are in an acceptable range. [ABSTRACT FROM AUTHOR]

This paper presents a new system configuration and a design method that improves disturbance rejection performance for a nonlinear system. The equivalent-input-disturbance (EID) approach is used to construct an EID estimator that estimates the influence of exogenous disturbances and nonlinearities on the output of the system. Sufficient stability conditions for state- and output-feedback control are derived in terms of linear matrix inequalities. New EID-based control laws that combines an EID estimate with a state- or an output-feedback control laws ensure good control performance. A numerical example illustrates the design method. A comparison between the EID-based control, the conventional disturbance observer, the disturbance-observer-based-control, and the sliding mode control methods demonstrates the validity and superiority of the EID-based control method. [ABSTRACT FROM AUTHOR]

Grinding in mineral processing is used to control the ore at the technically feasible and economically optimum particle size to achieve mineral liberation for separation. A circulating-load ratio (CLR) during a semi-autogenous grinding (SAG) milling process is critical for controlling particle size and energy consumption. This paper presents a CLR-prediction model based on clustering SAG milling operating conditions. First, operating parameters affecting the CLR are identified by comprehensively analyzing the complex mechanism and characteristics of a typical industrial SAG milling process. Next, a method is developed to cluster operating conditions of the SAG milling process based on the power consumption and CLR of the process. The method reveals the actual physical significance of each operating condition. Then, support vector regression (SVR) is used to model the CLR in each operating condition. After that, a distance-based model integration strategy is designed to determine the weights of each SVR model to predict the CLR. Finally, integrating the SVR submodels yields a CLR prediction model. Actual run data demonstrated the accuracy and effectiveness of the model in predicting CLR. This method has significant practical value for improving SAG milling efficiency via its utilization in control system design. • A clustering-regression-integration strategy for model prediction is proposed. • SAG milling operating conditions are classified by experience and operating data. • The integrated model predicts CLR accurately under multiple operating conditions. • K-means effectively classify operating conditions and SVR accurately predicts CLR. • Multi-model integration performs better than a single model for predicting CLR. [ABSTRACT FROM AUTHOR]

Abstract: This paper concerns the existence and exponential stability of periodic solution for the high-order discrete-time bidirectional associative memory (BAM) neural networks with time-varying delays. First, we present the criteria for the existence of periodic solution based on the continuation theorem of coincidence degree theory and the Young’s inequality, and then we give the criteria for the global exponential stability of periodic solution by using a non-Lyapunov method. After that, we give a numerical example that demonstrates the effectiveness of the theoretical results. The criteria presented in this paper are easy to verify. In addition, the proposed analysis method is easy to extend to other high-order neural networks. [Copyright &y& Elsevier]

Abstract: In this paper, we revisit the energy-based swing-up control solutions for the Pendubot, a two-link underactuated planar robot with a single actuator at the base joint. The control objective is to swing the Pendubot up to its unstable equilibrium point (at which two links are in the upright position). We improve the previous energy-based control solutions by analyzing the motion of the Pendubot further. Our main contributions are threefold. First, we provide a bigger control parameter region for achieving the control objective. Specifically, we present a necessary and sufficient condition for avoiding the singular points in the control law. We obtain a necessary and sufficient condition on the control parameter such that the up–down equilibrium point (at which links 1 and 2 are in the upright and downward positions, respectively) is the only undesired closed-loop equilibrium point. Second, we prove that the up–down equilibrium point is a saddle via an elementary proof by using the Routh–Hurwitz criterion to show that the Jacobian matrix valued at the point has two and two eigenvalues in the open left- and right-half planes, respectively. We show that the Pendubot will eventually enter the basin of attraction of any stabilizing controller for all initial conditions with the exception of a set of Lebesgue measure zero provided that these improved conditions on the control parameters are satisfied. Third, we clarify the relationship between the swing-up controller designed via the partial feedback linearization and that designed by the energy-based approach. We present the simulation results for validation of these results. [Copyright &y& Elsevier]

Abstract: This paper presents an integrated neural-network-based model for predicting the burn-through point (BTP) of a lead–zinc sintering process. This process features strong nonlinearity and time-varying parameters. First, experiments were carried out to establish a model of the gas temperature distribution (GTD) in the sintering machine; and based on the GTD model, a surface temperature model of the material (STMM) was established. Second, based on the STMM, a method of estimating the BTP that uses a soft-sensing technique was devised. In order to improve the estimation precision, a time-sequence-based model for predicting the BTP was built using grey system theory. Since the BTP is also affected by process parameters, a technological-parameter-based model for predicting the BTP was then built using a neural network. Finally, an integrated model for predicting the BTP was constructed by combining the time-sequence-based and the technological-parameter-based models using a fuzzy classifier. The result of actual runs shows that, compared to the manual control, the integrated prediction model reduced the variation in BTP by about 50%. This guarantees the improvement of the quality and quantity of the sinter. [Copyright &y& Elsevier]