Jinshou Yu’s research while affiliated with East China University of Science and Technology and other places

294 fault states of 12-pulse waveform controlled rectifier circuit are studied. According to the fault voltage waveforms of rectifier, a special fault classification method is proposed. To enhance the performance of particle swarm optimization (PSO), a novel PSO with disturbance (DPSO) is put forward by introducing an evolution speed factor in standard PSO. Simulation results and comparisons with standard PSO show that DPSO enhances the searching efficiency and improves the searching quality effectively. Finally, DPSO is successfully applied in neural network fault diagnosis modeling.

An expert system has been developed for monitoring the abnormal process and avoiding the dangers of purifying units in ammonia plant in this work. The regular knowledge expression method of ldquorule frame + rule bodyrdquo is used to building the knowledge base. It uses hybrid strategy for inference engine. The system is stable and reliable when put into operation, and it can monitor the abnormal process real-time, keep away the dangers of purifying units in ammonia plant, and realize fault-tolerant control under certain operation state.

support vector machine (SVM) is a kind of machine learning method based on the statistical learning theory, it has been applied in the fault diagnosis field. After analyzing SVM pattern classification theory, a hierarchical structure fault detection and identification (FDI) system is presented in this paper, and simulation results show that this method can effectively handle the complex process characteristic and improve FDI model performance.

By studying cellular automata, a new optimization method based on cellular automata is proposed by this paper. The new optimization method assumes that "life game" are applied in operator of genetic algorithm (GA). Experiment results show that the new method has good optimization performance. Then, a hybrid neural network algorithm based on life game, GA and back-propagation algorithm is presented to train soft-sensing model of acrylonitrile yield. Experiment results show that the hybrid soft sensing model proposed in this paper has good performance and high measuring precision.

For aggregation to be effective the component artificial neural networks (ANNs) must be as accurate and diverse as possible, an improved Bagging neural network ensemble algorithm is proposed to cope with this problem. The Euclidean distances between two arbitrary samples of the original training set are analyzed, the training subsets of component ANNs are distilled from this set then. The subsets elements have good properties of ergodicity and representativeness in sample space. The outputs of component ANNs are combined via weighted averaging and the optimal weights are determined by particle swarm optimization. Experimental studies on four typical regression datasets show that this approach has improved the quality of training subsets. Thus, the ensemble generalization ability is improved. Finally the improved algorithm is applied to construct an ANN-based soft sensor model for real-time measuring the ethylene yield. Application results show that this model has high measuring precision as well as good generalization ability.

Considering fault diagnosis is a small sample problem in real chemical process industry, Support Vector Machines (SVM) is adopted as classifier to discriminate chemical process steady faults. To improve fault diagnosis performance, it is essential to reduce the dimensionality of collected data. This paper presents a modified discrete binary ant colony optimization (MDBACO) to optimize discrete combinational problems, and then further combines it with SVM to accomplishing fault feature selection. The tests of optimizing benchmark functions show the developed MDBACO is valid and effective. The fault diagnosis results and comparisons of simulations based on Tennessee Eastman Process (TEP) prove the feature selection method based on MDBACO and SVM can find the essential fault variables quickly and exactly, and greatly increases the fault diagnosis correct rates as irrelevant variables are eliminated properly.

Fault diagnosis is a small sample problem as fault data are absent in the real production process. To tackle it, Support Vector Machines (SVM) is adopted to diagnose the chemical process steady faults in this paper. Considering the high data dimensionality in the large-scaled chemical industry seriously spoil classification capability of SVM, a modified adaptive chaotic binary ant system (ACBAS) is proposed and combined with SVM for fault feature selection to remove the irrelevant variables and ensure SVM classifying correctly. Simulation results and comparisons of Tennessee Eastman Process show the developed ACBAS can find the essential fault feature variables effectively and exactly, and the SVM fault diagnosis method combined with ACBAS-based feature selection greatly improve the diagnosing performance as unnecessary variables are eliminated properly.

Three sub-swarm discrete particle swarm optimization algorithm (THSDPSO) is proposed. The new algorithm assumes that all particles are divided into three sub- swarms. One sub-swarm flies toward the global best position. The second sub-swarm flies in the opposite direction. The last sub-swarm flies randomly around the global best position. In THSDPSO algorithm, two ways are used to handle the position of particles. One way is using the corresponding velocity as a probability measure by the transfer function and THSDPSO with this way is called BTHSDPSO. Another is directly using the hard limit function and THSDPSO with this way is called HTHSDPSO. The two THSDPSOs and basic discrete particle swarm optimization algorithm (DPSO) are all used to solve two well-known test functions' optimization problems. Simulation results show that the two THSDPSOs are both able to find the best fitness more quickly and more precisely than DPSO. Especially the HTHSDPSO has more wonderful optimization performance.

This paper presents a decoupled path planning based on ant colony algorithm and distributed navigation with collision avoidance for multi-robot systems. An improved ant colony algorithm is proposed to plan a reasonable collision-free path for each mobile robot of multi-robot system in the decoupled path planning scheme in complicated static environment. The special functions are added into the regular ant colony algorithm to improve the selective strategy. When an ant explores a dead-corner in path searching, a dead-corner table is established and a penalty function is used for the trail intensity updated in order to avoid the path deadlock of mobile robot. A behavior strategy on "first come and first service" is adopted to solve the conflict between moving robots in distributed local navigation. Simulation results show that the proposed method can effectively improve the performance of the planned path, and the individual robots with collision-free can achieve to reach their goal locations by the simple local navigation strategies

An improved radial basis function neural network (IRBFNN) with unsymmetrical Gaussian function is presented to simplify the structure of RBFNN. The improved resource allocating network (IRAN) is developed to design IRBFNN online for nonlinear dynamic system modeling, integrating the typical resource allocating network (RAN) with merging method for similar hidden units, deleting strategy for redundant hidden units, and LMS learning algorithm with moving data window for output link weights of network. The proposed approach can effectively improve the precision and generalization of IRBFNN. The combination of IRBFNN and IRAN is competent for the online modeling of nonlinear dynamic systems. The feasibility and effectiveness of the modeling method have been demonstrated by simulations.