13 July 2018 in IEEE Transactions on Fuzzy Systems
A novel adaptive energy management strategy is proposed for real time power split between fuel cells and supercapacitors in a hybrid electric vehicle in view of the fact that driving patterns greatly affect fuel economy. The driving pattern recognition (DPR) is achieved based on the features extracted from the historical velocity window with a multi-layer perceptron neural network. After the DPR has been obtained, an adaptive fuzzy energy management controller is utilized for power split according to the required power for vehicle running. In order to prolong the fuel cell lifetime whilst decreasing the hydrogen consumption, a genetic algorithm is applied to optimize critical factors such as adaptive gains and fuzzy membership function parameters for several standard driving cycles. In the proposed method, the future driving cycles are not required and the current driving pattern can be successfully recognized, demonstrating that less current fluctuations and fuel consumption can be achieved under various driving conditions. Compared with conventional energy management systems, the proposed framework can ensure the state of charge of supercapacitors within the desired limit.
24 November 2017 in IEEE Transactions on Industrial Electronics
Fuzzy neural networks (FNNs) are quite useful for nonlinear system identification when only the input/output information is available. A new FNN framework is first proposed by combining an AutoRegressive with exogenous input (ARX) with the nonlinear Tanh function in the Takagi-Sugeno (T-S) type fuzzy consequent part. An improved genetic algorithm is then designed to optimize the structure and parameters of the FNN simultaneously under unknown plant dynamics. The hybrid encoding/decoding, neighborhood search operator, and maintain operator are presented to optimize the input structure of the ARX plus the nonlinear function submodel, the number of the fuzzy rules, and the parameters of the membership function. Three benchmarks and a liquid level modeling problem in an industrial coke furnace are utilized to compare the performance of several typical methods. Simulation results show that the proposed method is superior in structure simplification, modeling precision, and generalization capability.
1 November 2017 in ISA Transactions
30 December 2020 in IEEE Transactions on Cybernetics
Deep learning technology has been widely used in fault diagnosis for chemical processes. However, most deep learning technologies currently adopted only use a single network stack or a certain network stack with multilayer perceptron (MLP) behind it. Compared with traditional fault diagnosis technologies, this method has made progress in both the diagnosis accuracy and speed, but due to the limited performance of a single network, the accuracy or speed cannot meet the requirements to the greatest extent. In order to overcome such problems, this article proposes a fault diagnosis method using deep learning multimodel fusion. Different from previous deep learning diagnosis methods, this method uses long short-term memory (LSTM) and convolutional neural network (CNN) to extract features separately. The extracted features are then fused and MLP is taken as the input for further feature compression and extraction, and finally the diagnosis results will be obtained. LSTM has long-term memory capabilities, the extracted features have temporal characteristics, and CNNs have a good effect on the extraction of spatial features. The proposed method integrates these two aspects for diagnosis such that the features finally extracted by the network have both spatial and temporal characteristics, thereby improving the network's diagnostic performance. Finally, a TE chemical process and an industrial coking furnace process are taken for simulation testing. It is proved that the performance of this method is superior to existing deep learning fault diagnosis methods with simple sequential stacking for unilateral feature extraction.
11 December 2017 in IEEE Transactions on Industrial Electronics
Iterative learning control (ILC) has been successfully applied to numerous batch processes over the past decades. Monotonic convergence of tracking error is a desired characteristic that attracts much attention in academia. Many factors arising in industrial practice, such as strong nonlinearity and parameter uncertainty, have challenged the most existing monotonically convergent ILC approaches. This motivates the development of the nonlinear monotonically convergent ILC (NMC-ILC) in this paper. The proposed NMC-ILC is an optimization-based control strategy, in which the original nonlinear process is linearly approximated to reduce the complexity of the optimization problem accompanied. An upper bound of the tracking error is derived by considering the effects of model mismatches that consist of two components: structural mismatch induced by the linearization and parametric mismatch arising from the indetermination of process parameters. The NMC-ILC is then devised by minimizing this upper bound. Numerical experiments are conducted on an ILC-testing benchmark and the velocity control in an injection molding process.
4 December 2017 in IEEE Transactions on Fuzzy Systems
The combination of the fuel cell (FC) and Super-capacitor (SC) for Hybrid Electric Vehicle (HEV) has the benefit of compensating for the slow dynamic response and avoiding reactant starvation of FC. Energy management system (EMS) is critical to the HEV and a fuzzy controller plus low pass filter is proposed to prolong the FC lifetime and decrease the hydrogen consumption. The constrained bi-objective optimization problem for fuzzy EMS is then solved by an improved genetic algorithm (GA), where the decimal and rule base encoding, constraint handling, the pruning and maintain operator are designed to optimize both the fuzzy rule base and the parameters of the membership functions. Simulation results of Highway fuel economy certification test (HWFET), Urban Dynamometer Driving Schedule (UDDS) and New European Drive Cycle (NEDC) illustrate that the proposed approach can smooth the output of FC with robustness and be implemented in real time, which decreases 19% current variation with about 10% increase of H2 consumption.