Search

To improve the control performance of calcination zone temperature in a lime rotary kiln, a predictive control method based on a support vector machine (SVM) and improved particle swarm optimization (PSO) algorithm is proposed. As high-temperature thermal equipment, the lime rotary kiln requires accurate modelling because of its complex non-linearity and long delay characteristics. SVM has strong normalization and good learning ability compared with other modelling models such as neural network, partial least squares model and other non-linear regression models, which can avoid overfitting and local minimization problems. At the same time, it is sometimes difficult to obtain a large number of production sample data of lime rotary kiln. The modelling process based on SVM requires only a small amount of sample data. SVM is appropriate for the modelling of calcination zone temperature of the lime rotary kiln. The predictive control method in this paper utilizes SVM to establish a non-linear prediction model of calcination zone temperature of the lime rotary kiln. The calcination zone temperature can be achieved by output feedback of input control variables, the error and the error correction. The performance index function is established by the control deviations and control variables. An improved PSO algorithm with better convergence speed and accuracy is employed to obtain optimal control laws by rolling optimization. The stability of the control method has also been demonstrated. The proof process shows that the control method of this paper is asymptotically stable. The simulation results show that the prediction error of calcination zone temperature based on SVM is within ±20°C and the prediction accuracy is better. The model of calcination zone temperature in the lime rotary kiln based on SVM has good performance. The proposed predictive control method can make the output value of the calcination zone temperature of the lime rotary kiln fast and stable to track the change of the reference value. At the same time, in the presence of interference, the system can also track the reference value. The average single step rolling optimization time of the control variables needs to be 0.29 s, which can be used for the practical applications. The simulation results show that the proposed control method is effective. [ABSTRACT FROM AUTHOR]

A scheduling method based on fuzzy feedback priority and variable sampling period for a networked control system with resource constraints is proposed in this paper. Firstly, each control loop weight is adjusted dynamically according to a control loop data transmission error through a quadratic square root mapping function. A different weight is given for each loop. A fuzzy feedback priority scheduling strategy is designed combined with dynamic weight, system output and system output error rate. The priority of each control loop is adjusted dynamically. Then, considering the current network status, the new network utilization is predicted through proportional control. A least-squares support vector machine (LSSVM) algorithm is used for controlled loop packet transmission time prediction. A variable sampling period for each control loop is determined by using a controlled loop dynamic weight, network utilization and data transmission time. Finally, a fuzzy feedback priority scheduler and variable sampling period scheduler are combined. The merits of the two schedulers are utilized to achieve the optimization goal. A simulation experiment is carried out using the True Time toolbox. Simulation results show that the scheduling method can make the network utilization rate converge to the desired value under resource constraints, improve the output response performance of the control loop, reduce the integral absolute error value of the control loop and decrease the data transmission time of the control loop. The networked control system performance is improved and the scheduling method of this paper is effective. [ABSTRACT FROM AUTHOR]