Deep learning (DL) methods have brought world-shattering breakthroughs, especially in computer vision and classification problems. Yet, the design and deployment of DL methods in time series prediction and nonlinear system identification applications still need more progress. In this paper, we present DL frameworks that are developed to provide novel approaches as solutions to the aforementioned engineering problems. The proposed DL frameworks leverage the advantages of autoencoders and long-short term memory network, which are known being data compression and recurrent structures, respectively, to design Deep Neural Networks (DNN) for modeling time series and nonlinear systems with high performance. We provide recommendations on how deep AEs and LSTMs should be utilized to end up with efficient Prediction-focused (Pf) and Simulation-focused (Sf) DNNs for time series and system identification problems. We present systematic learning methods for the DL frameworks that allow straightforward learning of Pf-DNN and Sf-DNN models in detail. To demonstrate the efficiency of the developed DNNs, we present various comparative results conducted on the benchmark and real-world datasets in comparison with their conventional, shallow, and deep neural network counterparts. The results clearly show that the deployment of the proposed DL frameworks results with DNNs that have high accuracy, even with a low dimensional feature vector.
In this study, an inverse controller based on a type-2 fuzzy model control design strategy is introduced and this main controller is embedded within an internal model control structure. Then, the overall proposed control structure is implemented in a pH neutralization experimental setup. The inverse fuzzy control signal generation is handled as an optimization problem and solved at each sampling time in an online manner. Although, inverse fuzzy model controllers may produce perfect control in perfect model match case and/or non-existence of disturbances, this open loop control would not be sufficient in the case of modeling mismatches or disturbances. Therefore, an internal model control structure is proposed to compensate these errors in order to overcome this deficiency where the basic controller is an inverse type-2 fuzzy model. This feature improves the closed-loop performance to disturbance rejection as shown through the real-time control of the pH neutralization process. Experimental results demonstrate the superiority of the inverse type-2 fuzzy model controller structure compared to the inverse type-1 fuzzy model controller and conventional control structures. 相似文献
In situ synthesis of conductive polymers, poly(Aniline) (p(An)), poly(Pyrrole) (p(Py)), and poly(Thiophene) (p(Th)) within network of superporous cryogels with tunable functionalities as neutral poly(acrylamide) (p(AAm), anionic poly(acrylic acid) (p(AAc)), and cationic poly(4-vinylpyridine) (p(4-VP)) were carried out via oxidation polymerization technique. The highest conductivity values were measured for p(AAm)/p(An) semi-IPN cryogel with 1.4 × 10?2 S.cm?1 and for p(AAc)/p(Py) cryogel with 3.2 × 10?4 S.cm?1. In addition, to increase the amounts of conductive polymers within cryogel networks, reloading/polymerization cycle was carried out thrice, and found that there is no significant increase in the amounts of conductive polymers and the measured conductivity values. The prepared p(AAm), p(AAc), and p(4-VP) cryogels and their corresponding p(An), p(Py), and p(Th) composites were tested potential sensor materials against HCl and NH3 vapor. The changes on conductivities for bare p(4-VP) cryogel were observed as 70 and 52-fold increase upon HCl and NH3 gas treatment, respectively. The p(4-VP)/p(An) p(An) composites showed 7-fold conductivity decrease upon the treatments of HCl and NH3 vapors. The p(AAm)/p(Py) composite responded 2-fold increase upon HCl vapor exposure and 50-fold decrease upon NH3 vapor exposure. Furthermore, p(AAm)/p(Th) cryogel composite responded 7-fold decrease and 300-fold increase in their conductivities upon HCl and NH3 vapor exposure, respectively.
Heparan sulfate proteoglycans (HSPGs) play diverse roles in cell recognition, growth, and adhesion. In vitro studies suggest that cell-surface HSPGs act as coreceptors for heparin-binding mitogenic growth factors. Here we show that the glycosylphosphatidylinositol- (GPI-) anchored HSPG glypican-1 is strongly expressed in human pancreatic cancer, both by the cancer cells and the adjacent fibroblasts, whereas expression of glypican-1 is low in the normal pancreas and in chronic pancreatitis. Treatment of two pancreatic cancer cell lines, which express glypican-1, with the enzyme phosphoinositide-specific phospholipase-C (PI-PLC) abrogated their mitogenic responses to two heparin-binding growth factors that are commonly overexpressed in pancreatic cancer: fibroblast growth factor 2 (FGF2) and heparin-binding EGF-like growth factor (HB-EGF). PI-PLC did not alter the response to the non-heparin-binding growth factors EGF and IGF-1. Stable expression of a form of glypican-1 engineered to possess a transmembrane domain instead of a GPI anchor conferred resistance to the inhibitory effects of PI-PLC on growth factor responsiveness. Furthermore, transfection of a glypican-1 antisense construct attenuated glypican-1 protein levels and the mitogenic response to FGF2 and HB-EGF. We propose that glypican-1 plays an essential role in the responses of pancreatic cancer cells to certain mitogenic stimuli, that it is relatively unique in relation to other HSPGs, and that its expression by pancreatic cancer cells may be of importance in the pathobiology of this disorder. 相似文献
As the extension of the linear inverted pendulum (LIP) and planar inverted pendulum (PIP), this paper proposes a novel spatial inverted pendulum (SIP). The SIP is the most general inverted pendulum (IP) than any existing IP. The model of the SIP is presented for the first time. The SIP inherits all the characteristics of the LIP and the PIP, which is a nonlinear, unstable and underactuated system. The SIP has five degrees of motion freedom and three control forces. Thus, it is a multiple-input and multiple-output (MIMO) system with nonlinear dynamics. To realize the spatial trajectory tracking of the SIP, the control structure with five PID controllers will be designed. The parameter tuning of the multiple PIDs is a challenging work for the proposed SIP model. To alleviate the difficulties of the parameter tuning for the multiple PID controllers, optimal PIDs can be achieved with the help of Big Bang-Big Crunch (BBBC) optimization. The BBBC algorithm can successfully optimize the parameters of the multiple PID controllers with high convergence speed. The optimization performance index of the BBBC algorithm is compared with that of the particle swarm optimization (PSO). Simulation results certify the rightness and effectiveness of the proposed control and optimization methods. 相似文献