NN-based Prediction Interval for Nonlinear Processes Controller

Neural networks (NNs) are extensively used in modelling, optimization, and control of nonlinear plants. NN-based inverse type point prediction models are commonly used for nonlinear process control. However, prediction errors (root mean square error (RMSE), mean absolute percentage error (MAPE) etc.) significantly increase in the presence of disturbances and uncertainties. In contrast to point forecast, prediction interval (PI)-based forecast bears extra information such as the prediction accuracy. The PI provides tighter upper and lower bounds with considering uncertainties due to the model mismatch and time dependent or time independent noises for a given confidence level. The use of PIs in the NN controller (NNC) as additional inputs can improve the controller performance. In the present work, the PIs are utilized in control applications, in particular PIs are integrated in the NN internal model-based control framework. A PI-based model that developed using lower upper bound estimation method (LUBE) is used as an online estimator of PIs for the proposed PI-based controller (PIC). PIs along with other inputs for a traditional NN are used to train the PIC to predict the control signal. The proposed controller is tested for two case studies. These include, a chemical reactor, which is a continuous stirred tank reactor (case 1) and a numerical nonlinear plant model (case 2). Simulation results reveal that the tracking performance of the proposed controller is superior to the traditional NNC in terms of setpoint tracking and disturbance rejections. More precisely, 36% and 15% improvements can be achieved using the proposed PIC over the NNC in terms of IAE for case 1 and case 2, respectively for setpoint tracking with step changes.

     

Modeling soft granular materials

Soft-grain materials such as clays and other colloidal pastes share the common feature of being composed of grains that can undergo large deformations without rupture. For the simulation of such materials, we present two alternative methods: (1) an implicit formulation of the material point method (MPM), in which each grain is discretized as a collection of material points, and (2) the bonded particle model (BPM), in which each soft grain is modeled as an aggregate of rigid particles using the contact dynamics method. In the MPM, a linear elastic behavior is used for the grains. In order to allow the aggregates in the BPM to deform without breaking, we use long-range center-to-center attraction forces between the primary particles belonging to each grain together with steric repulsion at their contact points. We show that these interactions lead to a plastic behavior of the grains. Using both methods, we analyze the uniaxial compaction of 2D soft granular packings. This process is nonlinear and involves both grain rearrangements and large deformations. High packing fractions beyond the jamming state are reached as a result of grain shape change for both methods. We discuss the stress-strain and volume change behavior as well as the evolution of the connectivity of the grains. Similar textures are observed at large deformations although the BPM requires higher stress than the MPM to reach the same level of packing fraction.     

Nonlocal buckling and vibration analysis of thick rectangular nanoplates using finite strip method based on refined plate theory

The buckling and vibration of thick rectangular nanoplates is analyzed in this article. A graphene sheet is theoretically assumed and modeled as a nanoplate in this study. The two-variable refined plate theory (RPT) is applied to obtain the differential equations of the nanoplate. The theory accounts for parabolic variation of transverse shear stress through the thickness of the plate without using a shear correction factor. Besides, the analysis is based on the nonlocal theory of elasticity to take the small-scale effects into account. For the first time, the finite strip method (FSM) based on RPT is employed to study the vibration and buckling behavior of nanoplates and graphene sheets. Hamilton’s principle is employed to obtain the differential equations of the nanoplate. The stiffness, stability and mass matrices of the nanoplate are formed using the FSM. The displacement functions of the strips are evaluated using continuous harmonic function series which satisfy the boundary conditions in one direction and a piecewise interpolation polynomial in the other direction. A matrix eigenvalue problem is solved to find the free vibration frequency and buckling load of the nanoplates subjected to different types of in-plane loadings including the uniform and nonuniform uni-axial and biaxial compression. Comparison studies are presented to verify the validity and accuracy of the proposed nonlocal refined finite strip method. Furthermore, a number of examples are presented to investigate the effects of various parameters (e.g., boundary conditions, nonlocal parameter, aspect ratio, type of loading) on the results.     

GSA: A Gravitational Search Algorithm

In recent years, various heuristic optimization methods have been developed. Many of these methods are inspired by swarm behaviors in nature. In this paper, a new optimization algorithm based on the law of gravity and mass interactions is introduced. In the proposed algorithm, the searcher agents are a collection of masses which interact with each other based on the Newtonian gravity and the laws of motion. The proposed method has been compared with some well-known heuristic search methods. The obtained results confirm the high performance of the proposed method in solving various nonlinear functions.     

Classification of non-alcoholic beer based on aftertaste sensory evaluation by chemometric tools

Sensory evaluation is the application of knowledge and skills derived from several different scientific and technical disciplines, physiology, chemistry, mathematics and statistics, human behavior, and knowledge about product preparation practices. This research was aimed to evaluate aftertaste sensory attributes of commercial non-alcoholic beer brands (P1, P2, P3, P4, P5, P6, P7) by several chemometric tools. These attributes were bitter, sour, sweet, fruity, liquorice, artificial, body, intensity and duration. The results showed that the data are in a good consistency. Therefore, the brands were statistically classified in several categories. Linear techniques as Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) were performed over the data that revealed all types of beer are well separated except a partial overlapping between zones corresponding to P4, P6 and P7. In this research, for the confirmation of the groups observed in PCA and in order to calculate the errors in calibration and in validation, PLS-DA technique was used. Based on the quantitative data of PLS-DA, the classification accuracy values were ranked within 49-86%. Moreover, it was found that the classification accuracy of LDA was much better than PCA. It shows that this trained sensory panel can discriminate among the samples except an overlapping between two types of beer. Also, two types of artificial networks were used: Probabilistic Neural Networks (PNN) with Radial Basis Functions (RBF) and FeedForward Networks with Back Propagation (BP) learning method. The highest classification success rate (correct predicted number over total number of measurements) of about 97% was obtained for RBF followed by 94% for BP. The results obtained in this study could be used as a reference for electronic nose and electronic tongue in beer quality control.     

A new second-order method for blind signal separation from dynamic mixtures

This paper presents a new approach to separate colored stationary signals mixed by convolutive channels. A cost function is proposed by employing linear constraint to the demixing vectors. The linear constraint is shown to be sufficient for avoiding trivial solution. The minimization of the cost function is performed using the Lagrangian method. Simulation results demonstrate the performance of the algorithm.     

Design, manufacture and evaluation of a new and simple mechanism for transmission of power between intersecting shafts up to 135 degrees (Persian Joint)

This paper introduces a new mechanism which is designed for the transmission of power between two intersecting shafts. The mechanism consists of one drive shaft and one driven shaft, six guide arms, and three connecting arms. The intersecting angle between the input shaft and the output shaft can be varied up to 135° while the velocity ratio between the two shafts remains constant. The research also includes a kinematic analysis and a simulation using Visual NASTRAN, Autodesk Inventor Dynamic and COSMOS Motion. The softwares showed that this mechanism can transmit constant velocity ratios at all angles between two shafts. By comparing the graphs of analytical analysis and simulation analysis, validity of equations was proved. Finally, by fabrication and evaluation of the mechanism it was shown that this mechanism can transmit constant velocity practically.     

Identification of concurrent control chart patterns with singular spectrum analysis and learning vector quantization

Identification of unnatural control chart patterns (CCPs) from manufacturing process measurements is a critical task in quality control as these patterns indicate that the manufacturing process is out-of-control. Recently, there have been numerous efforts in developing pattern recognition and classification methods based on artificial neural network to automatically recognize unnatural patterns. Most of them assume that a single type of unnatural pattern exists in process data. Due to this restrictive assumption, severe performance degradations are observed in these methods when unnatural concurrent CCPs present in process data. To address this problem, this paper proposes a novel approach based on singular spectrum analysis (SSA) and learning vector quantization network to identify concurrent CCPs. The main advantage of the proposed method is that it can be applied to the identification of concurrent CCPs in univariate manufacturing processes. Moreover, there are no permutation and scaling ambiguities in the CCPs recovered by the SSA. These desirable features make the proposed algorithm an attractive alternative for the identification of concurrent CCPs. Computer simulations and a real application for aluminium smelting processes confirm the superior performance of proposed algorithm for sets of typical concurrent CCPs.     

A 54.2?μW 5?MSps 9-bit ultra-low energy analog-to-digital converter in 180?nm technology

This paper aims to address the growing need for ultra-low power analog-to-digital converters (ADC). For this purpose, we pushed the limitations of conventional successive approximation register ADCs through the use of deep voltage scaling, a novel iterative precharging scheme, and topological improvements over recent works. From the simulations results we achieve a figure of merit of 31?fJ per conversion step, with an 8.45 effective number of bits, working at 5?MSps.