Inverse modeling of semiconductor manufacturing processes by neural networks

An inverse recall neural network model and a feedback pattern recognition method based on the model are proposed. The inverse recall neural network model is trained by the same method as that used for a typical multilayer feedforward model. The model can produce an inverse mapping of the trained feedforward mappings to show the parts of an input pattern. The model is applied to the feedback recognition method which can extract features from input patterns and discriminates between them by the inverse recall neural network model. The feedback recognition method adjusts feature extraction parameters so as to detect the important parts shown by the neural network model in order to present them to the network, and to produce more certain recognition results. This method is examined on handwritten alpha-numerics. It is found that rejection ratio can be reduced by half at the same error ratio. >

Development of Process Inverse Neural Network Model to Determine the Required Alum Dosage at Segama Water Treatment Plant Sabah, Malaysia

In order to identify and model and effectively control to the complicated nonlinear systems, principle of identification and control of the ANN's (artificial neural network) inverse model was analysed for nonlinear system in the paper. Simulation researches of the composite control structure scheme based on the feedforward counterbalance of the ANN's inverse model was done for nonlinear system. Simulation researches were done for the composite control scheme with conventional PID controller + counterbalance of the MFN (multilayer feedforward network) ANN's inverse model. The simulation results show, the composite control structure scheme designed is successful, and the relatively simple network architecture can be improved generalization ability of the ANN's inverse model and the control precision for nonlinear system.

In this paper, based on vector control system of double-fed motor, the artificial neural network (ANN) inverse control strategy of double-fed motor control is discussed. The double-fed motor mathematical model with stator flux oriented in the synchronous MT reference frame is given, and the reversibility of the system through Interact or Algorithm is confirmed, thus the artificial neural network inverse system model of motor is contributed, then the model is applied to double-fed speed-regulating system. At last, the double-fed motor control strategy based on artificial neural network inverse model is simulated on MATLAB. The simulation results shows that applying the neural network inverse system control strategy to double-fed motor is feasible.

Nowadays model based techniques play very important role in solving measurement and control problems. Recently for representing nonlinear systems fuzzy and neural network (NN) models became very popular. For evaluating measurement data and for controller design also the inverse models are of considerable interest. In this paper, different observer based techniques to perform fuzzy and neural network model inversion are presented. The methods are based on solving a nonlinear equation derived from the multiple-input single-output (MISO) forward fuzzy model simple by interchanging the role of the output and one of the inputs. The utilization of the inverse model can be either a direct compensation of some measurement nonlinearities or a controller mechanism for nonlinear plants. For discrete-time inputs the technique provides good performance if the iterative inversion is fast enough compared to system variations, i.e., the iteration is convergent within the sampling period applied. The proposed method can be considered also as a simple nonlinear state observer which reconstructs the selected input of the forward (fuzzy or NN) model from its output using an appropriate strategy and a copy of the fuzzy or neural network model itself. Improved performance can be obtained by introducing genetic algorithms in the prediction-correction mechanism. Although, the overall performance of the suggested technique is highly influenced by the nature of the non-linearity and the actual prediction-correction mechanism applied, it can also be shown that using this observer concept completely inverted models can be derived. The inversion can be extended towards anytime modes of operation, as well, providing short response time and flexibility during temporal loss of computational power and/or time.Keywordsinverse modelinginverse controliterative methodsfuzzy modelingneural network modelsanytime techniquesgenetic algorithmsobserverssingular value decomposition

: Manufacturing process models are developed for a data-driven, reconfiguarable manufacturing facility. A human assembly process model is developed to characterize assembly processes. Machine and tool resources are characterized for manufacturing processes to provide a broad characterization of manufacturing process coverage in a hypothetical final assembly factory for infantry fighting vehicles. A sample population of manufacturing process parameters and resources is collected and incorporated into a manufacturing capability and process model library. Requirements for a virtual manufacturing environment are developed, and an initial, prototype version is implemented to facilitate process model validation, configuration, and experimentation.

According to error and error rate, the phase plane divisional control (PPDC) applies different actions in the phase plane to make the control system up to the requirement. This controlling structure makes the system possess the features of closed-open loop. Furthermore, just as the PID control, itpsilas unnecessary to build up a mathematical model for the controlled object. This paper combines neural network (NN) inverse model with PPDC to generate a new kind of intelligent controller - NN-PPDC. This controller gives real time relationship between the basic performance of the system response and the action logical divisional in the phase plane. Thus the control parameters in each divisional can make automatic on-line regulation in accordance with changing in the controlling process, possessing adaptive ability. The simulation and its application in the feed-supporting system for the square kilometer array (SKA) show that this controller has better synthesized performance and robustness.

1. In this study, neural network (NN) and response surface (RS) models were developed to investigate the response [average daily gain (ADG) and feed efficiency (FE)] of young broiler chickens to dietary protein and lysine. For this purpose, data on their responses to dietary protein and lysine were extracted from the literature and separate NN and RS models were constructed. 2. Comparison between the NN and RS models revealed higher accuracy of prediction with the NN models compared to the RS models. In terms of R 2 values, the NN models developed for both ADG (R 2 = 0.923) and FE (R 2 = 0.904) were far superior to the RS models (R 2 for ADG = 0.511; R 2 for FE = 0.67). This suggests that the NN models can serve as an alternative option to conventional regression approaches including use of RS models. 3. Optimisation of the NN models developed for response to protein and lysine showed that diets containing 220.7 (g/kg of diet) protein and 12.85 (g/kg of diet) lysine maximise ADG, whereas maximum FE is achieved with diets containing 241.3 and 13.12 (g/kg) protein and lysine, respectively. Based on the optimisation results, optimal dietary protein and lysine concentrations for maximum FE in broiler chickens during the starting period are higher than for ADG.

Background: Data mining (DM) is an approach used in extracting valuable information from environmental processes. This research depicts a DM approach used in extracting some information from influent and effluent wastewater characteristic data of a waste stabilization pond (WSP) in Birjand, a city in Eastern Iran. Methods: Multiple regression (MR) and neural network (NN) models were examined using influent characteristics (pH, Biochemical oxygen demand [BOD 5 ], temperature, chemical oxygen demand [COD], total suspended solids [TSS], total dissolved solid [TDS], electrical conductivity [EC] and turbidity) as the regression input vectors. Models were adjusted to input attributes, effluent BOD 5 (BODout) and COD (CODout). The models performances were estimated by 10-fold external cross-validation. An internal 5-fold cross-validation was also used for the training data set in NN model. The models were compared using regression error characteristic (REC) plot and other statistical measures such as relative absolute error (RAE). Sensitivity analysis was also applied to extract useful knowledge from NN model. Results: NN models (with RAE = 78.71 ± 1.16 for BODout and 83.67 ± 1.35 for CODout) and MR models (with RAE = 84.40% ± 1.07 for BODout and 88.07 ± 0.80 for CODout) indicate different performances and the former was better (P < 0.05) for the prediction of both effluent BOD 5 and COD parameters. For the prediction of CODout the NN model with hidden layer size (H) = 4 and decay factor = 0.75 ± 0.03 presented the best predictive results. For BODout the H and decay factor were found to be 4 and 0.73 ± 0.03, respectively. TDS was found as the most descriptive influent wastewater characteristics for the prediction of the WSP performance. The REC plots confirmed the NN model performance superiority for both BOD and COD effluent prediction. Conclusion: Modeling the performance of WSP systems using NN models along with sensitivity analysis can offer better understanding on exploring the most significant parameters for the prediction of system performance. The findings of this study could build the foundation for prospective work on the characterization of WSP operations and optimization of their performances with a view to conducting statistical approaches.

Although the learning‐based surrogate model could achieve fast prediction, the model uncertainty in data and predictions remains a significant challenge. To alleviate the uncertainty problems for the surrogate model, a learning‐based uncertainty analysis method for improving the stability of the inverse model of RF devices is proposed. The inherent uncertainty in the inverse model is analyzed by leveraging ensemble adversarial learning, enabling the prediction of confidence intervals for the inverse surrogate model output without changing the original neural network structure. Subsequently, the differential evolution (DE) algorithm is employed to optimize the geometric parameters of the inverse model output within the predicted interval. To validate feasibility and effectiveness of the method, RF hairpin filter is used for the design example via the neural networks. RF filters with multiple center frequencies and bandwidths are simulated, fabricated and measured. Both simulated and measured results demonstrate a notable enhancement, with a 37.22% predicted accuracy improvement in the degree of fitting between the optimized response and the label of the input network, compared to the circuit response (S‐parameter) pre‐optimization. The proposed method exhibits much better performance in terms of the predicted accuracy and computational time compared to direct optimization without inverse model. The learning‐based method could also be applicable to the other RF devices, affirming the practical applicability and robustness of the approach.

Comparison of control strategies for hysteresis attenuation in electromechanical actuators subject to dispersion

Tracking control of shape memory alloy (SMA) actuators is essential in many applications such as vibration controls. Due to the hysteresis, an inherent nonlinear phenomenon associated with SMAs, open-loop control design has proven inadequate for tracking control of these actuators. Aimed at eliminating the position sensor to reduce cost of an SMA actuator system, in this paper, a neural network open loop controller is proposed for tracking control of an SMA actuator. A test stand, including a titanium-nickel (TiNi, or Nitinol) SMA wire actuator, a position sensor, bias springs, and a programmable current amplifier, is used to generate training data and to verify the neural networks open loop controller. A digital data acquisition and real-time control system was used to record experimental data and to implement the control strategy. Based on the training data obtained from the test stand, two neural networks are used to respectively model the forward and inverse hysteresis relations between the applied voltage and the displacement of the SMA wire actuator. To control the SMA actuator without using a position sensor, the neural network inverse model is used as a feedforward controller. The experimental results demonstrate the effectiveness of the neural network open loop controller for tracking control of the SMA wire actuator.

To advance the state of the art of physical-principle-enhanced hybrid artificial neural network (ANN) modeling, network configurations with parallel modules (PMNN) reflecting the structural information of the physical principles have been developed (Cao, 2001). In this paper, the PMNN configuration is applied to develop a scalable and invertible dynamic magneto-rheological (MR) fluid damper model. To advance the state of the art and address issues of the current ANN-based MR damper models found in the open literature, two ANN-based MR damper models are developed in this study. The first one is a conventional first-principle-enhanced hybrid neural network model (defined as the baseline model in the current study) that improves upon the previous ANN-based MR damper models by introducing feedback loops to represent the dynamic behaviors of the MR damper. A PMNN-based MR damper model is then derived to further improve the control-input-output scalability and realize the invertible model concept. “Input-output scalability” refers to the model's capability to accurately estimate the system response with input profiles significantly different from the training data. “Invertible model” means that the resultant forward model can be directly transformed into an inverse model through a simple algebraic operation. The network training/testing results indicate that while both models provide satisfactory performance, the PMNN model outperforms the baseline model by showing superior control-input-output scalability. The candidacy of PMNN as a control-oriented actuator modeling tool is further strengthened by the fact that it is invertible, in other words, the inverse model with desired force as input and the control signal—voltage as output can be easily established by algebraically manipulating the forward model. This study indicates that PMNN, as a scalable and invertible dynamic modeling tool, is feasible for developing system-design-oriented models of vibration control purposes.

Inverse modeling is important for microwave design, diagnosis, and tuning. Neural network modeling techniques have been presented for microwave inverse modeling in the existing literature. This paper provides an overview of advances in neural network techniques for microwave inverse modeling, including the basic formulation and a comprehensive neural network inverse modeling methodology, the multivalued neural network inverse modeling, and recent advances towards deep neural network for microwave inverse modeling.

This paper presents an application of the neural network (NN) model for forecasting and filtering problems. Instead of using the gradient descent method as used by the standard back-propagation network, the present study adopted the Levenberg-Marquardt algorithm. It was shown that the Levenberg-Marquardt algorithm tends to reach termination criteria more quickly making the searches employing the full model less time consuming. In the first part of forecasting problem, multilayer feedforward (MLFF) network was constructed by trials to forecast the tidal-level variations at the Chao Phraya river mouth in Thailand. Unlike the well-known conventional harmonic analysis, the neural network model uses a set of previous data for learning and then forecasting directly the time series of tidal levels. It was found that lead time of 1 to 24 hourly tidal levels can be successfully predicted using only a short-time hourly learning data.In the second part of filtering problem, the MLFF network was also used to establish a stage-discharge relationship for tidal river. Comparisons were made among conventional methods, i.e. by using multivariate functions (Linear and power models) and the NN model. The NN model was found to be able to model the transfer function too much higher accuracy than multiple regression analysis. The root mean square errors (RMSE) and the mean relative error (MRE) for the NN model are about 8 cm and 6.4% whereas the same errors for multiple regression analysis are 16 cm and 12%, respectively. The efficiency index for the linear, power, and NN models is 0.4, 0.7, and 0.9, respectively. In addition, the obtained stage-discharge relationship by the NN model can indicate reasonably important behavior of the tidal influences, i.e. there are stronger influences of tide during the dry period than the flood period.