Neural Network Self Research Articles

Multi-Source Fusion Deformation-Monitoring Accuracy Calibration Method Based on a Normal Distribution Transform–Convolutional Neural Network–Self Attention Network

Multi-Source Fusion Deformation-Monitoring Accuracy Calibration Method Based on a Normal Distribution Transform–Convolutional Neural Network–Self Attention Network

Character Recognition Using Neural Network Self- Organizing Map (NN-SOM)

Character Recognition Using Neural Network Self- Organizing Map (NN-SOM)

Continuous nutrient observations capture fine-scale estuarine variability simulated by a 3D biogeochemical model

The fine-scale (<100m; <1day) skill of a 3D biogeochemical estuarine model cannot be assessed by traditional discrete monitoring programmes. Whilst continuous temperature, salinity, chlorophyll, turbidity and oxygen sensors based on electrical and optical properties have been deployed at individual sites, in underway and in profiling configurations, continuous estuarine nutrient sensors are in their infancy. In this paper we present data from 3 continuous nutrient observing systems deployed in a temperate estuary: a Systea Wet Chemistry analyser deployed at the head of the estuary to characterise river water entering the estuary; an in situ ultra-violet optical nitrate analyser deployed on a benthic lander at the mouth of the estuary to ascertain the marine nitrate concentrations entering the estuary; and a rapid reverse flow injection analysis system to achieve a high resolution spatial survey of conditions throughout the estuary. Our observations capture seasonal and synoptic timescale variability in marine and river nutrient concentrations and characterise plumes of nutrient enrichment associated with point source discharge against a background estuarine gradient. At the head of the estuary nutrient concentrations were greater and varied in ratio compared to previous observations. Results from a fine-scale 3D biogeochemical model (Wild-Allen et al., 2013) captured 66% of the observed synoptic timescale variability in nitrate concentration at the mouth of the estuary, but did not propagate the observed synoptic timescale fluctuations in river nutrients into the domain when forced with an upstream boundary condition of interpolated monthly observations. The absence of proxies for observing river nutrients dictates further investment in automated nutrient analysis to better inform the model and characterise evolving river conditions. Modelled surface nutrient fields matched observed spatial snapshots of estuarine conditions in contrasting years with surprising accuracy and suggest persistent spatial patterns that might usefully be characterised with neural network self organising maps.

A humid electronic nose based on pulse voltammetry: A proof-of-concept design

We report herein the design, manufacture and use of a “humid electronic nose” prototype based on voltammetric techniques. It consists of an array of four working electrodes (i.e., Au, Pt, Ir and Rh) housed inside a homemade stainless steel cylinder, in contact with a fabric mesh made of nylon damped with a NaCl aqueous solution, used as the supporting humid membrane. The “humid electronic nose” was tested for the discrimination of different samples displaying different volatile compounds. The samples chosen involve aqueous solutions of different simple volatile products (i.e., ammonia, acetone, acetic acid and 6-amino-1-hexanol) and different food samples (i.e., onion, coffee and Roquefort cheese). Under working conditions, the volatile compounds from the corresponding sample were generated in the measurement chamber and were partially dissolved in the damped nylon fabric, which was in contact with the set of electrodes. It was envisioned that provided different samples offer different vapours, the application of a suitable set of pulses to the electrodes will differentiate the samples. This proof-of-concept study employed a Large Amplitude Pulse Voltammetry (LAPV) waveform. The increment for the potential steps was of 200mV and potentials ranged from +1 to −1V with each pulse applied for 50ms. PCA studies from the response obtained by the “humid electronic nose” discriminated the different samples studied. The neural network Self Organized Map (SOM) was also used to analyze the electrochemical data obtained from the “humid electronic nose”.

A neural network self tuner based on input shapers behavior for anti sway system of gantry cranes

The rapid and sway-less conveyance of crane loads has been an essential task and a field of research; considering the broad application of transportation of goods and materials by these systems. In this paper, a neural network self tuner (NNST) controller, which aims to move the load precisely as well as to eliminate its sway, is presented. This controller is capable of being trained flexibly for either increasing the speed of load transportation or decreasing its sway. The developed training algorithm, in this paper, is inspired by a sway elimination method of input shapers. The proposed training method provides the controller with the capability to remove the majority of load sway, at the outset of transportation, and transport it with a slight sway. The results show that the neural network controller with the proposed training method, in the same time, is capable of removing the sway of the load in a better way and be more appropriate for longer distances compared to a proportional-integral-derivative controller with constant gain. Besides, compared to standard input shapers, the NNST controller is more robust to cable length variations as well as length uncertainty by utilizing the feedback. Using a lab scale gantry crane model, the feasibility of empirical implementation is proved and results are verified.

Design of Vehicle Weigh-in-Motion System Based on Neural Network

In order to improve the dynamical respond of the weighing sensor and to meet the demand of rapid weigh-in-motion(WIM), a new adaptive weighing system is introduced. The system consists of sensors, data processing module, control module and interface circuit. For restraining the influence of vehicle speed and vehicle liberation, the system adopts self-adaptive variable step size LMS algorithm based on neural network to filter out the noise of WIM signal. Based on this method of signal processing, a vehicle WIM system based on neural network self adaptive filtering was designed. The experimental results proved that the neural network based method can be used to reduce the weighing time and increase the accuracy simultaneously.

Design of Vehicle Weigh-in-Motion System Based on Neural Network

In order to improve the dynamical respond of the weighing sensor and to meet the demand of rapid weigh-in-motion(WIM), a new adaptive weighing system is introduced. The system consists of sensors, data processing module, control module and interface circuit. For restraining the influence of vehicle speed and vehicle liberation, the system adopts self-adaptive variable step size LMS algorithm based on neural network to filter out the noise of WIM signal. Based on this method of signal processing, a vehicle WIM system based on neural network self adaptive filtering was designed. The experimental results proved that the neural network based method can be used to reduce the weighing time and increase the accuracy simultaneously.

Effect of SRTM resolution on morphometric feature identification using neural network—self organizing map

In this study, we present a semi-automatic procedure using Neural Networks--Self Organizing Map--and Shuttle Radar Topography Mission DEMs to characterize morphometric features of the landscape in the Man and Biosphere Reserve "Eastern Carpathians". We investigate specially the effect of two resolutions, SIR-C with 3 arc seconds and X-SAR with 1 arc second for morphometric feature identification. Specifically we investigate how the SRTM/C band data with 30 m interpolated grid, corresponding to SRTM/X band 30 m, affect the morphometric characterization and topography derivatives. To reduce misregistration between the DEMs, spatial co-registration was performed and a RMSE of 0.48 pixel was achieved. Morphometric parameters such as slope, maximum curvature, minimum curvature and cross-sectional curvature are derived using a bivariate quadratic approximation on 90 m, 30 m and interpolated 30 m DEMs. Self Organizing Map (SOM) is used for the classification of morphometric parameters into ten exclusive and exhaustive classes. These classes were analyzed as morphometric features such as ridge, channel, crest line and planar for all data sets based on feature space (scatter plot), morphometric signatures and 3D inspection of the area. The map quality is analyzed by oblique views with contour lines overlaid. Using the X band DEM with 30 m grid as benchmark, a change detection technique was used to quantify differences in morphometric features and to assess the scale effect going from a 90 m (C-band) DEM to an interpolated 30 m DEM. The same procedure is used to study the effect of different resolutions on morphometric features. Morphometric parameters were computed by a moving window size 5?×?5 (corresponding to 450 m on the ground) over SRTM- 90 m. To cover the same ground area, a moving window size of 15?×?15 is used for the 30 m DEM. The change analysis showed the amount of resolution dependency of morphometric features. Overall, the results showed that the introduced method is very useful for identification of morphometric features based on SRTM resolution. Decreasing the grid size from 90 m to 30 m reveals considerably more detailed information emphasizing local conditions. Comparison between results from DEM-30 m as reference data set and interpolated 30 m, showed a rate of change of 31.5% which is negligible. About 17% of this rate correspond to classes with mean slope?>?10°. Of the morphometric parameters, the cross sectional curvature is most sensitive to DEM resolution. Increasing spatial resolution reduces the main constrains for morphometric analysis with SRTM 90 m data, such as unrealistic features and isolated single elements in the output map. So in case of lack of high resolution data, the SRTM 90 m data could be interpolated and used for further geomorphic analysis.

신경망 자율 적응제어를 이용한 발전기의 전압제어

PI controller is widely used as voltage control system of generator. However when a generator system has various characters of continuance, a new PI parameter decision for accurate control is a hard task as method of solving this problem, in this paper, the method to generator voltage control using Neural Network self adaptive control is presented. A property continuous feedback control gain of voltage control system is decided by a rule of delta learning. The function of proposed control method is verified by voltage control experiment results of DC generator.

Parameterization and reconstruction from 3D scattered points based on neural network and PDE techniques

Reverse engineering ordinarily uses laser scanners since they can sample 3D data quickly and accurately relative to other systems. These laser scanner systems, however, yield an enormous amount of irregular and scattered digitized point data that requires intensive reconstruction processing. Reconstruction of freeform objects consists of two main stages: parameterization and surface fitting. Selection of an appropriate parameterization is essential for topology reconstruction as well as surface fitness. Current parameterization methods have topological problems that lead to undesired surface fitting results, such as noisy self-intersecting surfaces. Such problems are particularly common with concave shapes whose parametric grid is self-intersecting, resulting in a fitted surface that considerably twists and changes its original shape. In such cases, other parameterization approaches should be used in order to guarantee non-self-intersecting behavior. The parameterization method described in this paper is based on two stages: 2D initial parameterization; and 3D adaptive parameterization. Two methods were developed for the first stage: partial differential equation (PDE) parameterization and neural network self organizing maps (SOM) parameterization. The Gradient Descent Algorithm (GDA) and Random Surface Error Correction (RSEC), both of which are iterative surface fitting methods, were developed and implemented.