ANN Based Inverse Dynamic Model of the 6-PGK Parallel Robot Manipulator

In a controller design process, it is interesting to find the inverse model, in which the desired input signal of the system is determined by using a desired output of the system. In many cases, the inverse model problem is very difficult, and sometimes impossible to determine and implement. There are some methods to make an inverse dynamic model, such as computed torque control, which had been studied in the previous Chapters. The capability of the NNs to learn the inverse model of the plant has been investigated for many years; the NN can be used to learn an approximate inverse system. In this approach, the desired output of the dynamic system should be known, and the NN can be trained by the desired output of the dynamic system to obtain the inverse dynamic model. In early studies of adaptive learning control using an NN model, Barto et al. [1], Jordan [2], Miller et al. [3] and Psaltis et al. [4] addressed the problem of how to obtain the error signal for training the NN controller. Generally, the cost function, consisting of the squared norm of the output reference errors, can not correctly train the NN controller as an inverse dynamic model. Therefore, Jordan [2] proposed a forward-inverse-modeling, and Albus [5], Atkeson and Reinkensmeyer [6], Psaltis et al. [4], and Kuperstein and Rubinstein [7] used a direct-inverse-modeling to obtain the command-error for forming the inverse dynamic model as a feedforward controller. Moreover, Watanabe et al. [8] proposed a linear NN using feedforward NNs with two layers and a linear unit function in the output-layer. Kawato [9] proposed a learning method to obtain a feedforward controller, which uses the output of a feedback controller as the error for training an NN model. The conventional feedback controller is configured in parallel to the NN controller, in the same manner as reported by Gomi and Kawato [10], and Miyamoto et al. [11], but with a different SF and learning algorithm. In another work, Wada and Kawato [12] used a feedforward NN as an inverse dynamic model.

The increase of the world population, especially in the global competition, together with the increasing use of fossil fuel resources to meet energy needs, leads to more greenhouse gases (more than one CO2, methane etc.) emissions and the global climate crisis. In this process, changes in meteorological events such as temperature, precipitation, and wind, attract attention moreover but when considered as a whole, we know that these negative changes in the ecosystem negatively affect many living groups. Sea Surface Temperature (SST) as measured meteorologically is the most important environmental parameter where these changes are monitored and observed. It draws attention to the fact that changes in SST are not limited to living organisms as habitats, but also catalyze many chain reactions, especially socio-economic impacts. Therefore, much of the work is devoted to forecasting studies to adapt to changing habitats and take the necessary precautions against potential risks. Feed-forward artificial neural networks have been commonly used to address these emerging needs. Artificial neural networks, which are a simple imitation of the human neurological system, have been used as an artificial intelligence method in forecasting problems due to their superior performance and not having the limitations of classical time series. In this study, the forecasting of the time series of monthly mean SST temperature obtained from Rize station between the years 2010 and 2020 is performed by using feed-forward artificial neural networks, and the forecasting performance of the corresponding time series is compared with many forecasting methods with different characteristics. The comparison of the methods used the mean square error and mean absolute percentage error criteria, which are commonly used in the forecasting literature. The analysis results showed that the analysis results obtained with the feed-forward artificial neural networks have the best prediction performance. As a result, it can be stated that the sea surface temperature can be forecasted with a very high accuracy using the feed-forward artificial neural networks.

This paper presents a study on the weld quality obtained by different values of the input parameters. The weld quality is characterized by two categories of parameters: geometrical parameters and mechanical parameters. They are dependent on the following process parameters: electric arc voltage, electric current intensity, welding speed, the feed wire velocity. Because the dependence between inputs and outputs is a nonlinear one was used an artificial feed forward neural network (ANN). The ANN was trained with the backpropagation algorithm, using as training patterns data measured from the mechanical process. This ANN can be used to estimate some parameters from future experiments of the mechanical process.

Local meteorological parameters are key in understanding the frequency of occurrence of extreme weather conditions such as floods, and droughts, among others. In this study, we present a method for simultaneous retrieval of two weather parameters. The method is based on already measured monthly average values of weather parameters from 2011 to 2016, which were used to train a Feed-forward radial basis function neural network (RBFNN) to obtain a fast and accurate method to compute global solar radiation for specified weather parameters pair. In inverse modelling, a multidimensional unconstrained non-linear optimization was employed to retrieve the weather parameters pair. The new approach was validated using weather parameter data measured at the Department of Physics, Makerere University (0.31° N, 32.58° E, 1200 m). Statistical tools were used to evaluate the method's performance. In the Feed-forward artificial neural network (ANN), the correlation coefficient (R), mean bias error (MnB), root mean square error (RMSE), and mean percentage error (MAPE) were in the ranges 0.80-0.95, -0.0011-0.0077, 0.55-1.04 and 2.49%-5.82%, respectively. The pairs (sunshine hours, relative humidity) and (sunshine hours, minimum temperature) had the highest correlation coefficient of 0.95. In the inverse artificial neural network (ANNi), the R, MnB, RMSE and MAPE were in the ranges 0.3-0.9, 0.01-0.08, 0.51-11.14 and 2.1%-14.5%, respectively. The pair (sunshine hours, relative humidity) had the highest correlation coefficients of 0.92 and 0.62, respectively. The method helps in obtaining weather parameter data sets in places where measuring equipment is lacking or during days when measuring equipment malfunctions.

This paper presents a method to estimate a transformer health condition based on diagnostic tests. A feed forward artificial neural network (FFANN) is used to find the health index of the transformer. The health index is used to find the health condition of the transformer. The training of the FFANN is done using real measurements of 59 working transformers. The testing of the trained neural network performance is done using real data for 29 working transformers. The performance evaluation of the trained FFANN shows that the trained neural network is reliable in finding the health condition of any working transformer.

The authors examine the usefulness of the feedforward neural network as a controller. In particular, the authors consider as an example the problem of design of a power system stabilizer that has the structure of a classical feedforward artificial neural network. In this case, the controller is developed for a class of problems associated with linear plants. Certain observations are made regarding the generic behavior of systems with this type of controller structure. The authors incorporate a robustness criterion into the design as well as a conventional dynamic performance criterion.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

The aim of the study is verification of a stochastic optimization approach for redesign of heat-integrated two-stage Autothermal Thermophilic Aerobic Digestion (ATAD) bioreactors system for municipal wastewater treatment operating under uncertainties. It was implemented through simulation of the ATAD system operation using feed-forward Artificial Neural Networks (ANN) for bioreactors modelling of the both stages and heat integration model of ATAD system with one heat storage tank. For the simulation purpose the design parameters values of the heat integration equipment associated with the obtained optimal solution of the stochastic approach were used. The ANN models were applied for prediction of the thermal shock occurred in the first bioreactor stage at incoming of each new portion of raw sludge, the expected temperature of the sludge at the end of the process and the volatile solids reduction at constant parameters of the inlet flows. They were included in two sequentially linked modules for simulation of the bioreactors operation. An appropriate data transfer between the modules, simulating the bioreactors and the Heat Integration module was provided. The simulation was carried out for two 15 d winter and summer periods. The simulation results have shown that applying the heat-integration of the process can lead to an increase in the temperatures of the inlet raw sludge about 6-8 °? and a decrease in the depth of thermal shock in both bioreactors about 5-7 °?. It also can provide higher and sustainable temperatures of the hot treated sludge at the end of the process which get much closer to the optimal of 55 °? for the first bioreactor and 65 °? for the second bioreactor and higher volatile solids reduction in both bioreactors about 1.5 wt.%.

A feedforward artificial neural network has been realized and trained to individualize subharmonics frequencies in an electric network. The adopted sampling window is fixed in 20 ms correspondent to a period of the fundamental frequency of 50 Hz, nevertheless the net is able to identify the subharmonics frequencies even if their period is not completed. The particular binary successive approximation structure realized for the ANN forecasts the scalability of the same net so allowing to recognize the subharmonics with a prefixed resolution. The net has been widely tested on synthesized signals.

Of late, traffic sign detection and recognition are becoming very prevalent topic as it enhances drivers towards safety and alert them with precaution information. This study reports about processing time of the individual color detection and recognition of the partial occlusion traffic sign that have been previously implemented using HSV and RGB color ratio and ANN and PCA method respectively for detection and recognition. The data set for detection and classification process has been successfully created in various places in Malaysia that involved with degradation and out of planes rotated of the signs. There are three standard types of colored images have been used in the study namely Red, Blue and Yellow signs. In this study, we analyze the system processing speed of individual color detection and classification respectively using red, green and blue (RGB) and hue, saturation and value (HSV) color segmentation techniques, supervised feed forward artificial neural network (ANN) and principal component analysis (PCA). The experimental result shown that processing time of individual color detection during daytime and at night using HSV method is slightly faster than RGB technique. On the other hand, supervised feed forward neural network has reached almost 1s in recognizing traffic sign images rather than PCA with only 0.0238s.

On the Circuit Complexity of Sigmoid Feedforward Neural Networks

The bandsawing as a multi-point cutting operation is the preferred method for cutting off raw materials in industry. Although cutting off with bandsaw is very old process, research efforts are very limited compared to the other cutting process. Appropriate online tool condition monitoring system is essential for sophisticated and automated machine tools to achieve better tool management. Tool wear monitoring models using artificial neural network are developed to predict the tool wear during cutting off the raw materials (American Iron and Steel Institute 1020, 1040 and 4140) by bandsaw. Based on a continuous data acquisition of cutting force signals, it is possible to estimate or to classify certain wear parameters by means of neural networks thanks to reasonably quick data-processing capability. The multi-layered feed forward artificial neural network (ANN) system of a 6 × 9 × 1 structure based on cutting forces was trained using error back-propagation training algorithm to estimate tool wear in bandsawing. The data used for the training and checking of the network were derived from the experiments according to the principles of Taguchi design of experiments planned as L27. The factors considered as input in the experiment were the feed rate, the cutting speed, the engagement length and material hardness. 3D surface plots are generated using ANN model to study the interaction effects of cutting conditions on sawblade. The analysis shows that cutting length, hardness and cutting speed have significant effect on tooth wear, respectively, while feed rate has less effect. In this study, the details of experimentation and ANN application to predict tooth wear have been presented. The system shows that there is close match between the flank wear estimated and measured directly.

Weight initialization of sigmoidal feed forward artificial neural network (SFFANN) and the Convolutional neural networks (CNN) has been a known factor which affects the learning abilities of the neural network. The uniform random weight initialization approach has been quite often used as the conventional network weight initial technique, due to its simplicity. However, various researches have shown that the random technique may not be the ideal choice of weight initialization for these neural networks. In this work, we analyze two separate chaotic functions and explore the possibilities of these being used as the weight initialization methods against the conventional random initialization technique for SFANNs as well as for the CNNs. For the SFFANNs, this analysis were done over 8 function approximation problems chosen for experimentation. The mean test error values along with a two sample t-test results strongly suggest that the Chebyshev chaotic map based weight initialization technique outperforms the conventional random initialization technique for most of the problems under consideration and hence may be used as an alternative weight initialization technique for the SFFANNs. For the CNN experiment, the MNIST dataset was used for analyzing the performance of the random and the Chebyshev based initialization scheme. Results strongly support the use of the Chebyshev chaotic map based initialization scheme as an alternate to the conventional random initialization.

The global increase in greenhouse gas emissions from automobiles has brought about the manufacture and usage of large quantities of electric vehicles (EVs). However, to ensure proper integration of EVs into the grid, there is a need to forecast the charging demand of EVs accurately. This paper presents a short-term electric vehicle charging demand forecast using a feedforward artificial neural network optimized with a modified local leader phase spider monkey optimization (MLLP-SMO) algorithm, a proposed variant of spider monkey optimization. A proportionate fitness selection is employed to improve the update process of the local leader phase of the spider monkey optimization. The proposed algorithm trains a feedforward neural network to forecast electric vehicle charging demand. The effectiveness of the proposed forecasting model was tested and validated with electric vehicle public charging data from the United Kingdom Power Networks Low Carbon London Project. The model's performance was compared to a feedforward neural network trained with particle swarm optimization, genetic algorithm, classical spider monkey optimization, and two conventional forecasting models, multi-linear regression and Monte Carlo simulation. The performance of the proposed forecasting model was assessed using the mean absolute percentage error of forecast and forecasting accuracy. The model produced a forecast accuracy and mean absolute percentage error of 99.88% and 3.384%, respectively. The results show that MLLP-SMO as a trainer predicted better than the other forecasting models and met industry standard forecast accuracy.

One of many applications of artificial neural networks is discovering non-linear patterns in time series data. In this paper, analysis of the efficacy of applying an artificial neural network to the time series data produced by fluctuating stock prices is discussed in more detail. There are few current models that are capable of analyzing stocks but they lack in predicting effectively. Results of this neural network are examined through the generated return on investment. The network used for stock analysis is a four layer, feed-forward artificial neural network. The results of this network reveal that artificial neural networks are capable of performing technical analysis on stock prices. The return on investment ranged anywhere from 0.8 to 5.28 % per month or as extrapolated over a year, as high as 17 %.

To investigate the feasibility of artificial neural networks (ANN) to reconstruct dose maps for intensity modulated radiation treatment (IMRT) fields compared with those of the treatment planning system (TPS). An artificial feed forward neural network and the back-propagation learning algorithm have been used to replicate dose calculations of IMRT fields obtained from PINNACLE(3) v9.0. The ANN was trained with fluence and dose maps of IMRT fields for 6 MV x-rays, which were obtained from the amorphous silicon (a-Si) electronic portal imaging device of Novalis TX. Those fluence distributions were imported to the TPS and the dose maps were calculated on the horizontal midpoint plane of a water equivalent homogeneous cylindrical virtual phantom. Each exported 2D dose distribution from the TPS was classified into two clusters of high and low dose regions, respectively, based on the K-means algorithm and the Euclidian metric in the fluence-dose domain. The data of each cluster were divided into two sets for the training and validation phase of the ANN, respectively. After the completion of the ANN training phase, 2D dose maps were reconstructed by the ANN and isodose distributions were created. The dose maps reconstructed by ANN were evaluated and compared with the TPS, where the mean absolute deviation of the dose and the γ-index were used. A good agreement between the doses calculated from the TPS and the trained ANN was achieved. In particular, an average relative dosimetric difference of 4.6% and an average γ-index passing rate of 93% were obtained for low dose regions, and a dosimetric difference of 2.3% and an average γ-index passing rate of 97% for high dose region. An artificial neural network has been developed to convert fluence maps to corresponding dose maps. The feasibility and potential of an artificial neural network to replicate complex convolution kernels in the TPS for IMRT dose calculations have been demonstrated.