Yuhan Ding

Measurement of process variables in solid-state fermentation of wheat straw using FT-NIR spectroscopy and synergy interval PLS algorithm.

The feasibility of rapid determination of the process variables (i.e. pH and moisture content) in solid-state fermentation (SSF) of wheat straw using Fourier transform near infrared (FT-NIR) spectroscopy was studied. Synergy interval partial least squares (siPLS) algorithm was implemented to calibrate regression model. The number of PLS factors and the number of subintervals were optimized simultaneously by cross-validation. The performance of the prediction model was evaluated according to the root mean square error of cross-validation (RMSECV), the root mean square error of prediction (RMSEP) and the correlation coefficient (R). The measurement results of the optimal model were obtained as follows: RMSECV=0.0776, R(c)=0.9777, RMSEP=0.0963, and R(p)=0.9686 for pH model; RMSECV=1.3544% w/w, R(c)=0.8871, RMSEP=1.4946% w/w, and R(p)=0.8684 for moisture content model. Finally, compared with classic PLS and iPLS models, the siPLS model revealed its superior performance. The overall results demonstrate that FT-NIR spectroscopy combined with siPLS algorithm can be used to measure process variables in solid-state fermentation of wheat straw, and NIR spectroscopy technique has a potential to be utilized in SSF industry.

Rapid determination of pH in solid-state fermentation of wheat straw by FT-NIR spectroscopy and efficient wavelengths selection

AbstractIn the work discussed in this paper we investigated the feasibility of determination of the pH of a fermented substrate in solid-state fermentation (SSF) of wheat straw. Fourier-transform near-infrared (FT-NIR) spectroscopy was combined with an appropriate multivariate method of analysis. A genetic algorithm and synergy interval partial least-squares (GA–siPLS) were used to select the efficient spectral subintervals and wavelengths by k-fold cross-validation during development of the model. The performance of the final model was evaluated by use of the root mean square error of cross-validation (RMSECV) and correlation coefficient (Rc) for the calibration set, and verified by use of the root mean square error of prediction (RMSEP) and correlation coefficient (Rp) for the validation set. The experimental results showed that the optimum GA-siPLS model was achieved by use of seven PLS factors, when four spectral subintervals were selected by siPLS and then 45 wavelength variables were chosen by use of the GA. The predicted precision of the best model obtained was: RMSECV = 0.0583, Rc = 0.9878, RMSEP = 0.0779, and Rp = 0.9779. Finally, the superior performance of the GA-siPLS model was demonstrated by comparison with four other PLS models. The overall results indicated that FT-NIR spectroscopy can be successfully used for measurement of pH in solid-state fermentation, and use of the GA-siPLS algorithm is the best means of calibration of the model. FigureThe above figure shows the algorithm implementation process in this study. siPLS combined with a genetic algorithm (GA), called GA-siPLS, was used for model calibration. The specific objectives of the study were: to preprocess the raw spectra by use of the first derivative; to split the full-spectrum region into equidistant spectral subintervals and to select several efficient subintervals by use of siPLS; to select useful wavelength variables by use of a GA from the efficient subintervals selected; and to develop a PLS model based on the useful variables selected by GA-siPLS

A review of learning algorithm for radius basis function neural network

In this paper, Radius Basis Function Neural Networks basic learning algorithms are reviewed from the aspects of convergence, training speed, network structure, generalization, etc. Advantages and disadvantages of each learning algorithm are pointed out. And the prospect of dynamic neural network is considered.

A review of decoupling control based on multiple models

In this article, we review and discuss methods for decoupling control which include conventional decoupling and adaptive decoupling. Decoupling methods which is used to solve the multivariable coupled in nonlinear control are always a hot issue. The major algorithms such as neural network and fuzzy control are described in details and critically reviewed in this work. The advantages and disadvantages of these algorithms are analyzed. The review reveals the tremendous prospect of decoupling algorithms in nonlinear control. It is important to seek effective and simple decoupling methods.

Development of improved inverse compensator for two-dimensional sensor

In this paper, we developed an improved inverse compensator to ameliorate the dynamic performance of the two-dimensional sensor. The main feature of the improved compensator is the use of a special state observer, which is designed on the state-space model the two-dimensional sensor. As the state-space model can fully describe the complex sensor, the obtained compensator will be more effective in compensating the output of the sensor. Besides, the compensator uses an ANN (artificial neural network) to approximate the imprecisely known nonlinear function, which makes it more applicable for compensating real sensors.

A novel soft sensor model based on artificial neural network in the fermentation process

Some crucial process variables in fermentation process could not be measured directly. Soft sensor technology provided an effective way to solve the problem. There has been considerable interest in modeling a soft sensor by using artificial neural network (ANN) in bioprocess. To generate a more efficient soft sensor model, we proposed a novel soft sensor model based on artificial neural network (SS-ANN). By analyzing a grey-box model of fermentation process, the secondary variables were selected. In modeling, on-line measurable variables could be taken as the input of ANN and the output is the derivatives of immeasurable variables. The estimated values of immeasurable variables were calculated by integrating the outputs of the well-trained ANN. The novel SS-ANN is different from the general SS-ANN. Experimental results of erythromycin fermentation process showed the novel soft sensor model could estimate mycelia concentration, sugar concentration and chemical potency with higher accuracy and generalization ability than the general soft sensor based on ANN. The novel soft sensor modeling method provides the theory basis for selecting the secondary variables. The dynamic characteristic of the process is considered, the novel model improves the estimation accuracy and generation ability. It can be concluded that the soft sensor model mentioned in this paper is reasonable and effective. Key words : Soft sensor model, artificial neural network, fermentation process, dynamic characteristics.

Soft sensor model of marine enzyme fermentation process based on NN-MIV variable selection

To solve online estimation of the key variables that cannot be used in the online real time measurement in the process of marine enzyme fermentation, a variable selection method based on the mean impact value (MIV) of neural network (NN) is proposed. The principle of NN-MIV soft sensing is that firstly through the analysis of the mechanism of marine enzyme fermentation process, the external contribution rate of auxiliary variable to key variables is computed by the MIV method and the internal contribution rate of auxiliary variables to key variables is computed by the NN method, then considering the two contribution as a basis for the choices of auxiliary variables. NN-MIV variable selection method is applied to soft sensor model of marine enzyme which can screen input variables purposefully, so the optimized soft sensor model of NN-MIV can realize online soft sensing of enzyme activity. The simulation results show that this kind of soft sensing model has higher prediction precision and stronger generalization ability.

Flux observer of induction motor based on “assumed inherent sensor” inversion

In induction motor vector control system, the obtaining of accurate rotor flux is of essential significance. Based on the “assumed inherent sensor” reverse principle, a new method for observing flux is proposed. It firstly assumes that there exists a subsystem inside the induction motor, whose inputs are just the to-be-estimated variables while outputs the measurable variables and establishes the mathematical model of the subsystem. Then it analyzes the reversibility of the subsystem according to inverse system theory, and establishes the flux observer inverse model. Compared with traditional observation method, the proposed method provides a precise mathematical model with high measurement accuracy for observing flux. Simulation results show that: the inverse method of observation is extremely robust to changes of the rotor resistance and load and can achieve an accurate observation of the rotor flux.