Multiple-input Single-output Model Research Articles

Multiple-Input-Single-Output prediction models of crowd dynamics for Model Predictive Control (MPC) of crowd evacuations

Predicting crowd dynamics in real-time may allow the design of adaptive pedestrian flow control mechanisms that prioritize attendees’ safety and overall experience. Single-Input-Single-Output (SISO) AutoRegresive eXogenous (ARX) prediction models of crowd dynamics have been effectively used in Linear Model Predictive Controllers (MPC) that adaptively regulate the movement of people to avoid overcrowding. However, an open research question is whether Multiple-Input, State-space, and Nonlinear modeling approaches may improve MPC control performance through better prediction capabilities. This paper considers a simulated controlled evacuation scenario, where evacuees in a long corridor dynamically receive speed instructions to modulate congestion at the exits. We aim to investigate Multiple-Input-Single-Output (MISO) prediction models such that the inputs are the control action (speed recommendation) and pedestrian flow measurement, and the output is the local density of the pedestrian outflow. State-space and Input–output MISO models, linear and neural, are identified using a data-driven approach in which input–output datasets are generated from strategically designed microscopic evacuation simulations. Different estimation algorithms, including the subspace method, prediction error minimization, and regularized AutoRegressive eXogenous (ARX) model reduction, are evaluated and compared. Finally, to investigate the importance of measuring and modeling the pedestrian inflow, the case in which the models’ structure is defined as a Single-Input-Single-Output (SISO) system has been explored, where the pedestrian inflow is considered an unmeasured input disturbance. This study has important implications for the design of more effective MPC controllers for regulating pedestrian flows. We found that the prediction error minimization algorithm performs best and that nonlinear state-space modeling does not improve prediction performance. The study suggests that modeling the inner state of the evacuation process through a state-space model positively influences predicting system dynamics. Also, modeling pedestrian inflow improves prediction performance from a predefined prediction horizon value. Overall, linear state-space models have been deemed the most suitable option in corridor-type scenarios.

Prediction of flavor of Maillard reaction product of beef tallow residue based on artificial neural network.

The beef flavor of beef tallow residue was improved by enzymatic hydrolysis followed by the Maillard reaction, and the flavor could be predicted using an artificial neural network. Five beef tallow residue hydrolysates were prepared using different enzymes. The Flavourzyme and Papain (FP) hydrolysate had low molecular weight peptides and high degree of hydrolysis and free amino acid content. We identified 49 main compounds, including aldehydes, pyrazines, and furan. Furan and pyrazine were the dominant volatile compounds in the five beef tallow residue-derived Maillard reaction products (MRPs), and their profiles and levels in the FP MRPs were high. The FP MRPs had the best sensory characteristics. The artificial neural network analysis revealed that the multiple input single output model had a better performance than the single input single output model, and the prediction accuracy was>90%, indicating that the MRPs sensory evaluation scores could be accurately predicted.

Vision sensing and feedback control of weld penetration in helium arc welding process

Consistent joint penetration is essential to mechanical properties of weldments, yet difficult to realize in gas tungsten arc welding due to heat accumulation, varying welding conditions, and uncertain disturbance, especially for manufacturing large-scale structural components. The solution is to monitor welding penetration in real-time and implement closed-loop control. In this paper a directional light-assisted vision sensing scheme is proposed to measure the transient profile of the backside weld pool/bead for monitoring and further controlling the joint penetration state. The developing zone and the developed zone of the backside weld bead are recognized, helping determining its final width. And the relationship between the transient backside bead width and the corresponding final width is established. Furthermore, the backside bead profile is employed as a feedback signal and the closed-loop control of weld formation is implemented to achieve full penetration and a constant backside bead width. A multiple-input single-output model predictive controller is utilized in the control system, creatively combined with an anti-windup integrator eliminating the steady-state error caused by nonlinearity in welding. Experimental results have proved that the developed control system is effective in keeping a consistent backside bead width with a rapid response, a small overshoot and robustness against varying heat transfer conditions.

Leveraging long short-term memory (LSTM)-based neural networks for modeling structure\u2013property relationships of metamaterials from electromagnetic responses

We report a neural network model for predicting the electromagnetic response of mesoscale metamaterials as well as generate design parameters for a desired spectral behavior. Our approach entails treating spectral data as time-varying sequences and the inverse problem as a single-input multiple output model, thereby compelling the network architecture to learn the geometry of the metamaterial designs from the spectral data in lieu of abstract features.

An Ensemble of Recurrent Neural Networks for Real Time Performance Modeling of Three-Spool Aero-Derivative Gas Turbine Engine

Abstract Gas turbine is a complex system operating in nonstationary operation conditions for which traditional model-based modeling approaches have poor generalization capabilities. To address this, an investigation of a novel data driven neural networks based model approach for a three-spool aero-derivative gas turbine engine (ADGTE) for power generation during its loading and unloading conditions is reported in this paper. For this purpose, a nonlinear autoregressive network with exogenous inputs (NARX) is used to develop this model in matlab environment using operational closed-loop data collected from Siemens (SGT-A65) ADGTE. Inspired by the way biological neural networks process information and by their structure which changes depending on their function, multiple-input single-output (MISO) NARX models with different configurations were used to represent each of the ADGTE output parameters with the same input parameters. First, data preprocessing and estimation of the order of these MISO models were performed. Next, a computer program code was developed to perform a comparative study and to select the best NARX model configuration, which can represent the system dynamics. Usage of a single neural network to represent each of the system output parameters may not be able to provide an accurate prediction for unseen data and as a consequence provides poor generalization. To overcome this problem, an ensemble of MISO NARX models is used to represent each output parameter. The major challenge of the ensemble generation is to decide how to combine results produced by the ensemble's components. In this paper, a novel hybrid dynamic weighting method (HDWM) is proposed. The verification of this method was performed by comparing its performance with three of the most popular basic methods for ensemble integration: basic ensemble method (BEM), median rule, and dynamic weighting method (DWM). Finally, the generated ensembles of MISO NARX models for each output parameter were evaluated using unseen data (testing data). The simulation results based on datasets consisting for experimental data as well as data provided by Siemens high fidelity thermodynamic transient simulation program show improvement in accuracy and robustness by using the proposed modeling approach.

Data‐Driven Modeling of Biodiesel Production Using Artificial Neural Networks

AbstractData‐driven modeling of biodiesel production was developed by simultaneous transesterification and esterification of rapeseed oil and myristic acid with methanol, without catalyst or with different amounts of sulfated zirconia catalyst. An artificial neural network (ANN)‐based model was created with experimental literature data. The input data, i.e., reaction time, catalyst, temperature, and methanol‐to‐oil ratio, and output data, i.e., total fatty acid methyl ester and oleic acid methyl ester, were considered to develop the model. Multiple input single output (MISO) ANN architecture was taken to predict the above targeted two output parameters. The proposed ANN model is computationally efficient and works reasonably well when tested on biodiesel production for solving the MISO model.

Data-Driven Modeling of Wireless Power Transfer Systems With Multiple Transmitters

This article develops a new method of data-driven modeling for a class of multiple-transmitter single-receiver wireless power transfer (WPT) systems. A continuous-time multiple-input single-output (MISO) model with pure time delays is used to characterize the input–output behavior of the system, where the transfer functions associated with each input channel are not constrained to have the same denominator. Moreover, the time delays are allowed to be a fraction of the sampling interval in order to account for the delay effects that stem from circuit components and wireless communication, which are, by nature, often a fraction of the sampling interval. An optimal refined instrumental variable method is proposed to estimate the parameters and time delays of the MISO model based on sampled input–output data. In contrast to the conventional circuit-theory-based modeling methods that rely on circuit parameters and result in models which are often complex, the proposed data-based method yields parsimonious models, whose parameters are directly estimated from input–output data. Due to the easy availability of sampled data in control engineering applications, the proposed method is clearly more user-friendly, having a broad prospect for efficient operation of WPT systems, such as prediction, optimization, and control. Numerical and experimental results are presented to validate the effectiveness and merit of the proposed method.

Anticipatory NH3 injection control for SCR system based on the prediction of the inlet NOx concentration

The anticipatory NH3 injection control strategy with a prediction-assisted feed-forward was proposed to improve the accuracy of the outlet NOx control. The large lag of SCR system was ascribed to the lag duration by flue gas sampling for inlet NOx analysis. A 660 MW power plant equipped tangentially fired pulverized coal boiler was involved in the process of theoretical analysis and engineering application. 22 key running parameters of the studied power plant were determined to influence the inlet NOx concentration of SCR system substantially through the multiple linear regression (MLR) method. Multiple input single output (MISO) model was adopted to predict the non-lag inlet NOx concentration based on a multi layer perception (MLP) model with three hidden layers. An optimization strategy for MISO model application was proposed to improve the prediction accuracy of inlet NOx concentration. The root mean square error (RMSE) was decreased to be 5.52 mg·Nm−3 and the mean absolute percentage error (MAPE) was only 1.12%. The maximum absolute deviation was also decreased by 41.28% compared with the deviation in the direct application case. The average constant control deviation (CCD) of the outlet NOx concentration was calculated to be 2.27 mg·Nm−3 in the case analysis of engineering application. The technology can be developed as a potential strategy for improving the NH3 injection performance of SCR system.

Power Beacon-Based Wireless Power Transfer in MISO/SISO: An Application in Device-to-Device Networks

This paper considers device-to-device (D2D) together with single input single output and multiple input single output models in transmitting of nearby devices under help of wireless power transfer. To support more harvested energy, two modes are studied in which multiple-antenna/single antenna power beacons are proposed to robust D2D transmission network. Especially, enhanced successful communication is explored with short distance transmission. Accordingly, the alternative energy source can be used to maintain small devices which can operate at close position efficiently. In this paper, a model of radio frequency-assisted wireless energy transfer for D2D system with two realistic transmission schemes will be investigated, namely pure D2D and D2D with interference impact of conventional user equipment. As an important result, we derive analytical expressions for outage probability to achieve performance evaluation. This paper will analyze outage probability by matching Monte-Carlo and analytical simulations to corroborate the exactness of derived expressions.

Detection of reaching intention using EEG signals and nonlinear dynamic system identification

Background and objectivesLow frequency electroencephalography (EEG) signals are associated with preparation of movement and thus provide valuable information for brain-machine interface applications. The purpose of this study was to detect movement intention from EEG signals before execution of self-paced arm reaching movements. MethodsTen healthy individuals were recruited. Movement onset was determined from surface electromyography recordings time-locked with EEG signals. Unlike previous studies, which employed feature extraction and classification for decoding, a nonlinear dynamic multiple-input/single output (MISO) model was developed. The MISO model consisted of a cascade of Volterra structures and a threshold block, generating the binary output corresponding to intention/no-intention. The modeling process included input selection from a pool of different EEG channels. The predictive performance of the model was evaluated using the receiver operating characteristics curve, from which the optimum threshold was also selected. The Mann–Whitney statistics was employed to select the significant EEG channels for the output by examining the statistical significance of improvement in the predictive capability of the model when the respective channels were included. ResultsWith the proposed approach, movement intention was detected approximately 500 ms before the movement onset and on average, with an accuracy of 96.37 ± 0.94%, a sensitivity of 77.93 ± 4.40% and a specificity of 98.52 ± 1.19%. ConclusionsThe model output can be converted to motion commands for neuroprosthetic devices and exoskeletons in future applications.

Design of Eccentric Training System Based on Multiple-Input Single-Output Wiener Nonlinear Model.

The efficacy of eccentric cycling has been proved in research, clinical, and sport training activities. However, several constraints make it difficult to use commercially available eccentric cycle ergometers. This study will stimulate the application of eccentric cycling as an exercise modality. University research laboratory. In this study, the multiple-input single-output (MISO) Wiener nonlinear model was applied to the design of the eccentric training system. Particle Swarm Optimization (PSO) algorithm was then applied to identify the parameters of the MISO Wiener nonlinear model. The computer software embedded with MISO model was programmed for not only controlling and monitoring the running status of the eccentric training system, but also collecting data in a flexible and convenient manner. The correlation coefficient of 0.9985 and root mean square error of 3.3264 between the collected power value and the estimated power value from the model shows the good performance of the model. The proposed method for the design of the eccentric training system using MISO Wiener nonlinear model provides a new way for eccentric research and training.

A Novel Single-Beacon Navigation Method for Group AUVs Based on SIMO Model

Single-beacon navigation for group autonomous underwater vehicles (AUVs) suffers from long navigation period and asynchronization. This paper addresses the problem by introducing the single-input-multiple-output (SIMO) model, where, a single AUV transmits the signal and other AUVs receive the signal. Utilizing the time-delay of signals from the beacon and the sender, the navigation for all AUVs is achieved simultaneously, and the navigation period is dramatically decreased. Moreover, based on the new SIMO model, a tracking method utilizing extended Kalman filter is also put forward to increase the navigation precision and reduce the computational burden. Finally, a coordinate fusion algorithm based on the location, error, and time-delay is presented, which fuses the independent AUV coordinates to be a higher precision holistic coordinate system for group AUVs. Simulation results show the effectiveness of the proposed method.

Developing and Testing Dynamic Models for HVAC Systems Using System Identification Approach

This paper proposes integrated modeling and optimization methods for a chilled water HVAC system using a system identification approach. Two multiple input-single output models are developed to find the supply air temperature and fan power of the investigated chilled water air handling unit AHU. To test the proposed models, actual data are collected from existing HVAC systems. Different fan and cooling coil model structures with various time delays and orders are tested to find the optimal model structure in term of normalized root-mean-square deviation or Coefficient Of Variances (COV). The paper also proposes an optimization procedure integrated into system identification model to automate the process of finding optimal fan and cooling coil model structures yielding the best accurate predictions. The testing results show that the proposed methods can provide accurate predictions that can used for several applications such as control optimization, energy assessment and fault detection and diagnosis.

System identification of the upper part of Murray River

Data based modelling is an important tool for obtaining models of rivers. In this paper we consider several identification methods, namely Prediction Error Method, Maximum Likelihood, continuous time system identification, Refined Instrumental Variable method (used within the context of Data-Based Mechanistic modelling) and Subspace Identification Method, and apply them to real data from the Murray River in Australia. Both Multiple-Input Single-Output models where the output is the water level in a lake and Multiple-Input, Multiple-Output models where in addition a flow is also modelled, are considered. The models are compared in terms of their accuracy on validation data and on how easily the methods can incorporate prior knowledge.

Estimation of a large-scale generalized Volterra model for neural ensembles with group lasso and local coordinate descent.

Estimation of neural models based on observed spike timing faces challenges as the amount of recorded units increases, especially when identifying detailed model features. Given that neural regions are generally sparsely connected, input selection is a critical step in model estimation but oftentimes computationally and theoretically challenging. In this paper, we detail an efficient methodology for estimating a sparse, nonlinear dynamical multiple-input, single-output model (MISO) applicable to large-scale (n > 50) single-unit recorded activity. The main contribution of this paper is the complete implementation of a principled group-lasso and local coordinate descent (LCD) algorithm into a generalized Volterra model (GVM) framework to achieve efficient sparse model estimation. Input selection is achieved with group-lasso by simultaneously selecting groups of parameters that are associated with each input. LCD yields efficient computation as the amount of inputs and parameters increase. We investigate and validate the performance of this estimation procedure with the application to a 64 input simulated model.

Fuzzy dynamic modelling and predictive control of a coagulation chemical dosing unit for water treatment plants

In this paper, a fuzzy model predictive control (FMPC) strategy is proposed to regulate the output variables of a coagulation chemical dosing unit. A multiple-input, multiple-output (MIMO) process model in form of a linearised Takagi–Sugeno (T–S) fuzzy model is derived. The process model is obtained through subtractive clustering from the plant's data set. The MIMO model is described by a set of coupled multiple-input, single-output models (MISO). In the controller design, the T–S fuzzy model is applied in combination with the nonlinear model predictive control (MPC) algorithm. The results show that the proposed controller has good set-point tracking when compared with nonlinear MPC and adequate disturbance rejection ability required for efficient coagulation control and process optimisation in water treatment operations.

Application of FRF with SISO and MISO model for accelerometer-based in-cylinder pressure reconstruction on a 9-L diesel engine

Engine control with feedback from engine combustion process diagnostics can help improve fuel efficiency and assist in meeting stricter emission regulations. The standard is to use in-cylinder pressure measurements with analysis including rate of heat release. The measurement is usually obtained with intrusive sensors that require a special mounting process and engine structure modification. The potential of the low-cost non-intrusive accelerometer as an alternative means to reconstruct the in-cylinder pressure has been demonstrated by previous investigations. In this work, start of injection (SOI) sweep test conditions at varied speed spanning both low load and high load were conducted on an inline 6-cylinder, 9 L diesel engine. The relationship between the in-cylinder pressure and the accelerometer signal was quantified with frequency response function (FRF). The robustness of the obtained FRF was evaluated by applying the single-test-based FRF to reconstruct the in-cylinder pressures for other test conditions. Two models, single-input single-output (SISO) and multiple-input single-output (MISO), were investigated and compared where the accelerometer signal was taken as the input and in-cylinder pressure as the output. The optimal channel used to acquire the input signal in the SISO model was selected on the basis of coherence analysis. Results show that the MISO model assisted by principal component analysis (PCA) and offset-compensation processes can result in better in-cylinder pressure estimation than the SISO model for conditions with 2200 rpm engine speed. With the purpose of minimizing the cost for accelerometer employment, the minimum number of inputs used to reconstruct the in-cylinder pressure in the MISO model was pursued. Thresholds were set based on three estimated in-cylinder pressure parameters to select the qualified input channels and two input channels were finally determined. Results showed that the two-input single-output FRF model coupled with the PCA and offset-compensation processes improves the FRF’s robustness for the in-cylinder pressure estimation in comparison to the SISO FRF model based on all the tests conducted in this paper.

Sparse-representation algorithms for blind estimation of acoustic-multipath channels

Acoustic channel estimation is an important problem in various applications. Unlike many existing channel estimation techniques that need known probe or training signals, this paper develops a blind multipath channel identification algorithm. The proposed approach is based on the single-input multiple-output model and exploits the sparse multichannel structure. Three sparse representation algorithms, namely, matching pursuit, orthogonal matching pursuit, and basis pursuit, are applied to the blind sparse identification problem. Compared with the classical least squares approach to blind multichannel estimation, the proposed scheme does not require that the channel order be exactly determined and it is robust to channel order selection. Moreover, the ill-conditioning induced by the large delay spread is overcome by the sparse constraint. Simulation results for deconvolution of both underwater and room acoustic channels confirm the effectiveness of the proposed approach.

Practical experience with unified discrete and continuous–time, multi–input identification for control system design

The paper is concerned with the practical aspects of a unified approach to the identification and estimation of multiple–input, single–output (MISO) transfer function models for both continuous and discrete–time systems. The estimation algorithms considered in the paper are based on the Refined Instrumental Variable (RIV) approach to identification and estimation, where the MISO model denominator polynomials are normally constrained to be equal. Unconstrained RIV estimation presents a more difficult problem and it is necessary to exploit an iterative, back–fitting routine to handle this more general situation. The paper focuses on the practical realization of this back–fitting algorithm, including its initiation from either common denominator MISO or repeated SISO estimation. The rivcdd algorithm for continuous–time model estimation, as implemented in the CAPTAIN Toolbox for Matlab™, is then used in three practical examples: first, the modelling of solute transport and dispersion in a water body; secondly, modelling for two control problems, namely a pair of connected laboratory DC motors and a nonlinear wind turbine simulation.

A dynamic data-based model describing nephropathia epidemica in Belgium

Nephropathia epidemica (NE) is a human infection caused by Puumala virus (PUUV), which is naturally carried and shed by bank voles (Myodes glareolus). Population dynamics and infectious diseases in general, such as NE, have often been modelled with mechanistic SIR (Susceptible, Infective and Remove with immunity) models. Precipitation and temperature have been found to be indicators of NE, however most SIR models do not take them into account. The objective of this paper was to develop a dynamic model of incidences of NE in Belgium by taking into account climatological data. A multiple–input, single-output (MISO) transfer function was used to model the incidence of NE. In a first step, the NE cases were modelled based on data from 1996 until 2003 with an R T 2 of 0.68. In the next step the MISO model was validated for incidences of NE in Belgium from 2003 to 2008 ( R T 2 of 0.54). The output of the MISO models was the number of NE cases in Belgium over the time period 1996 until 2008 and the inputs were average measured monthly precipitation (mm), and temperature (°C) as well as estimated carrying capacity (vole ha−1). The monthly values of carrying capacity (K) were estimated for the whole period by using an existing mechanistic SIR model. K is related to the variation in seed production in Northern Europe, which has an effect on the population of bank voles. In the future, such modelling approaches may be used to predict and monitor forthcoming NE cases based on climate and vegetation data as a tool for prevention of NE cases.