Cheng-Tao Wu

Neural–Fuzzy Gap Control for a Current/Voltage-Controlled 1/4-Vehicle MagLev System

A magnetically levitated (MagLev) vehicle prototype has independent levitation (attraction) and propulsion dynamics. We focus on the levitation behavior to obtain precise gap control of a 1/4 vehicle. An electromagnetic levitation system is highly nonlinear and naturally unstable, and its equilibrium region is severely restricted. It is therefore a tough task to achieve high-performance vehicle-levitated control. In this paper, a MagLev system is modeled by two self-organizing neural-fuzzy techniques to achieve linear and affine Takagi-Sugeno (T-S) fuzzy systems. The corresponding linear-type optimal fuzzy controllers are then used to regulate both physical systems (voltage- and current-controlled systems). On the other hand, an affine-type fuzzy control design scheme is proposed for the affine-type systems. Control performance and robustness to an external disturbance are shown in simulation results. Affine T-S fuzzy representation provides one more adjustable parameter in the neural-fuzzy learning process. Therefore, an affine T-S-based controller possesses better performance for a current-controlled system since it is nonlinear not only to system states but also to system inputs. This phenomenon is shown in simulation results. Technical contributions include a nonlinear affine-type optimal fuzzy control design scheme, self-organizing neural-learning-based linear and affine T-S fuzzy modeling for both MagLev systems, and the achievement of an integrated neural-fuzzy technique to stabilize current- and voltage-controlled MagLev systems under minimal energy-consumption conditions.

Neural-network-based optimal fuzzy control design for half-car active suspension systems

Developing advanced design and synthesis of self-learning optimal intelligent active suspension systems. Artificial neural-based fuzzy modeling is applied to set up the neural-based fuzzy model based on the training data from the nonlinear half-car suspension system dynamics. Furthermore, a robust optimal fuzzy controller is designed based on the proposed fuzzy model to improve ride quality and support appropriate movement in suspension systems. Moreover, the development of self-learning optimal intelligent active suspension can not only absorb disturbance and shock, to adapt the model, the sensor and the actuator error but also cope with the parameter uncertainty with minimum power consumption. The simulation results also indicate the feasibility and the applicability of the designed controller.

Fuzzy-Based Self-Interactive Multiobjective Evolution Optimization for Reverse Engineering of Biological Networks

S-system modeling from time series datasets can provide us with an interactive network. However, system identification is difficult since an S-system is described as highly nonlinear differential equations. Much research adopts various evolution computation technologies to identify system parameters, and some further achieve skeletal-network structure identification. However, the truncated redundant kinetic orders are not small enough as compared with the preserved terms. In this paper, we integrate quantitative genetics, bacterium movement, and fuzzy set theory into evolution computation to develop a new genetic algorithm to achieve convergence enhancement and diversity preservation. The proposed exploration and exploitation genetic algorithm (EEGA) can improve the best-so-far individual and ensure global optimal search at the same time. The EEGA enhances evolution convergence by golden section seed selection, normal-distribution reproduction, mixed inbreeding and backcrossing, competition elitism, and acceleration operations. Search-then-conquer evolution direction operations, eugenics-based screen-sifting mutation, eugenic self-mutation, and fuzzy-based tumble migration preserve population diversity to avoid premature convergence. Furthermore, to ensure that a reasonable gene regulation network is inferred, fuzzy composition is introduced to derive a reconstruction index. This performance index let EEGA possess self-interactive multiobjective learning. The proposed fuzzy-reconstruction-based multiobjective genetic algorithm is examined by three dry-lab biological systems. Simulation results show that a safety pruning action is guaranteed (the truncation threshold is set to be 10-15), and only one- or two-step pruning action is taken.

Inference of Genetic Network of Xenopus Frog Egg : Improved Genetic Algorithm

An improved genetic algorithm (IGA) is proposed to achieve S-system gene network modeling of Xenopus frog egg. Via the time-courses training datasets from Michaelis-Menten model, the optimal parameters are learned. The S-system can clearly describe activative and inhibitory interaction between genes as generating and consuming process. We concern the mitotic control in cell-cycle of Xenopus frog egg to realize cyclin-Cdc2 and Cdc25 for MPF activity. The proposed IGA can achieve global search with migration and keep the best chromosome with elitism operation. The generated gene regulatory networks can provide biological researchers for further experiments in Xenopus frog egg cell cycle control

Evolution - Based Gene Regulatory Network of Yeast Cell Cycle

In the work, we try to construct the corresponding S-system and modified power-low model from a dataset. These two mathematical models are highly nonlinear. Though they can clearly describe reactions among genes in the biological system, the identification is a tough work, especially for huge genes. We adopt the evolution strategy to achieve 16-genes modeling with 544 or 288 parameters. The time-course data of the yeast cell cycle is concerned. The proposed two different gene regulatory networks and their corresponding pathways can provide biological researchers for further experiments in yeast cell cycle control.

Computation Intelligent for Eukaryotic Cell-Cycle Gene Network

Computational intelligent approaches is adopted to construct the S-system of eukaryotic cell cycle for further analysis of genetic regulatory networks. A highly nonlinear power-law differential equation is constructed to describe the transcriptional regulation of gene network from the time-courses dataset. Global artificial algorithm, based on hybrid differential evolution, can achieve global optimization for the highly nonlinear differential gene network modeling. The constructed gene regulatory networks will be a reference for researchers to realize the inhibitory and activatory operator for genes synthesis and decomposition in Eukaryotic cell cycle

Seeding-inspired chemotaxis genetic algorithm for the inference of biological systems

A large challenge in the post-genomic era is to obtain the quantitatively dynamic interactive information of the important constitutes of underlying systems. The S-system is a dynamic and structurally rich model that determines the net strength of interactions between genes and/or proteins. Good generation characteristics without the need for prior information have allowed S-systems to become one of the most promising canonical models. Various evolutionary computation technologies have recently been developed for the identification of system parameters and skeletal-network structures. However, the gaps between the truncated and preserved terms remain too small. Additionally, current research methods fail to identify the structures of high dimensional systems (e.g., 30 genes with 1800 connections). Optimization technologies should converge fast and have the ability to adaptively adjust the search. In this study, we propose a seeding-inspired chemotaxis genetic algorithm (SCGA) that can force evolution to adjust the population movement to identify a favorable location. The seeding-inspired training strategy is a method to achieve optimal results with limited resources. SCGA introduces seeding-inspired genetic operations to allow a population to possess competitive power (exploitation and exploration) and a winner-chemotaxis-induced population migration to force a population to repeatedly tumble away from an attractor and swim toward another attractor. SCGA was tested on several canonical biological systems. SCGA not only learned the correct structure within only one to three pruning steps but also ensures pruning safety. The values of the truncated terms were all smaller than 10-14, even for a thirty-gene system.

Prototype of integrated pseudo-dynamic crosstalk network for cancer molecular mechanism.

In this study, we attempted to solve two important challenges in systems biology. First, although the Michaelis-Menten (MM) model provides local kinetic information, it is hard to generalize MM models to model a large system because increasingly large amounts of experimental data are necessary for the parameter identification. In addition, it is not possible to develop an MM model that provides information about the strength of the interactions in the system. Second, although the dynamic simulation of various signal transduction pathways is important in cancer research, it is impossible to theoretically derive a mathematical model to describe the cancer molecular mechanism. Predictive computational approaches can be used to analyze the dynamics of a system and to determine the dysfunction of a regulatory process. In this report, we first propose a pseudo-dynamic pathway to describe protein interactions in an MM system. We then discuss the dynamic behavior of two large-scale systems (antigrowth-signal-induced cell cycle and apoptotic-signal-transduction mechanism). These two systems were constructed through the in-series and organic integration, respectively, of MM modules with Petri net modules; moreover, more than 30% additional reactions were added during this integration step. We then described an extremely large multi-stream system (growth signal transduction); however, the analysis of this system to obtain dynamic predictions is critical but appears impossible. Thus, we introduced a fuzzy concept that can be used to develop a physically realizable model prototype. In the future, through step-by-step in vivo modifications, researchers will be able to develop a complete model of cancer metabolism to achieve accurate predictions.

Winner-Take-All Memetic Differential Evolution for Genetic Interaction: Parameter Identification

The emerging large-scale biological tools (e.g., micro array) challenge biologists to realize the connectivity of genes and/or proteins at the system level (global view). Having advantages in good generalization and showing the direct interaction of genes and/or proteins, the S-system becomes one of the popular models, which is able to capture the dynamic behavior of the biological system. Differential evolution (DE) and its variants have recently applied to solve various optimization problems in engineering fields. However, the exploitative and explorative abilities are insufficient. In this study, we propose a winner-take-all memetic differential evolution scheme to infer the parameters of the S-type gene regulatory networks. This method was tested with a genetic-branch pathway and a twenty-gene network. The learning was implemented in a wide search space ([0, 100] for rate constants and [-100, 100] for kinetic orders) with a bad initial start (All parameters were randomly initialized at the neighborhood of 80). Simulation results show high-accuracy solutions are obtained.

Affined-type Neural-Fuzzy Gap Control of Electromagnetic Suspension Systems

Electromagnetic suspension (EMS) systems are highly nonlinear, especially for current-controlled EMS, which is not only nonlinear to system states abut also to system inputs. And hence, theoretically approach to derive T-S fuzzy system is unsuitable. In this work, we use self-organizing neural-fuzzy technique to obtain affine-TS fuzzy models for both voltage-and current-controlled EMS systems. Based on these two models, affine-type optimal fuzzy controls for these two highly nonlinear systems are derived. The control performance and robustness to external disturbance are demonstrated and are compared with linear type. Since affine T-S fuzzy model can provide one more adjustable parameter for neural-fuzzy modelling, the derived affine TS-based controller could have better performance than linear type as system complexity increases. This phenomena obviously exists in simulation for current-controlled EMS systems