Artificial Neural Network Structure Research Articles

Position control of a digital electrohydraulic system with limited sensory data using double deep q-network controller

In this study, we propose a reinforcement learning controller for position control of a digital electrohydraulic system with independent metering valve structure. Our study addresses the inherent challenges of traditional control methods, such as the need for continuous pressure measurement and complex control structures. As a solution, a double deep q-network (DDQN) controller which relies solely on position feedback from D-EHS is suggested. A simulation environment was developed to train the DDQN controller, wherein the artificial neural network structure, training hyperparameters, and reward function were optimized through preliminary studies. The DDQN controller was trained using a sinusoidal position reference signal of a single frequency and evaluated experimentally for its position control performance. Tracking performance was assessed against various position reference signals, including ramp and sinusoidal signal with different frequencies. Robustness against external disturbances was also tested by increasing the nominal supply pressure by 20% and 40%. The results demonstrate that the DDQN controller successfully controls the digital electrohydraulic system and is achieving good tracking performance for the training position reference signal. Furthermore, the controller exhibits satisfactory tracking performance for non-training position reference signals, highlighting its ability to learn the digital electrohydraulic system dynamics with limited training data. Even under varying supply pressure conditions, the DDQN controller maintains an acceptable level of control performance. Our findings emphasize the generalizability and reliability of the DDQN reinforcement learning controller, making it a suitable candidate for application in digital electrohydraulic systems and contributing to the advancement of control theory research.

Artificial Neural Networks and Experimental Analysis of the Resistance Spot Welding Parameters Effect on the Welded Joint Quality of AISI 304.

The automobile industry relies primarily on spot welding operations, particularly resistance spot welding (RSW). The performance and durability of the resistance spot-welded joints are significantly impacted by the welding quality outputs, such as the shear force, nugget diameter, failure mode, and the hardness of the welded joints. In light of this, the present study sought to determine how the aforementioned welding quality outputs of 0.5 and 1 mm thick austenitic stainless steel AISI 304 were affected by RSW parameters, such as welding current, welding time, pressure, holding time, squeezing time, and pulse welding. In order to guarantee precise evaluation and experimental analysis, it is essential that they are supported by a numerical model using an intelligent model. The primary objective of this research is to develop and enhance an intelligent model employing artificial neural network (ANN) models. This model aims to provide deeper knowledge of how the RSW parameters affect the quality of optimum joint behavior. The proposed neural network (NN) models were executed using different ANN structures with various training and transfer functions based on the feedforward backpropagation approach to find the optimal model. The performance of the ANN models was evaluated in accordance with validation metrics, like the mean squared error (MSE) and correlation coefficient (R2). Assessing the experimental findings revealed the maximum shear force and nugget diameter emerged to be 8.6 kN and 5.4 mm for the case of 1-1 mm, 3.298 kN and 4.1 mm for the case of 0.5-0.5 mm, and 4.031 kN and 4.9 mm for the case of 0.5-1 mm. Based on the results of the Pareto charts generated by the Minitab program, the most important parameter for the 1-1 mm case was the welding current; for the 0.5-0.5 mm case, it was pulse welding; and for the 0.5-1 mm case, it was holding time. When looking at the hardness results, it is clear that the nugget zone is much higher than the heat-affected zone (HZ) and base metal (BM) in all three cases. The ANN models showed that the one-output shear force model gave the best prediction, relating to the highest R and the lowest MSE compared to the one-output nugget diameter model and two-output structure. However, the Levenberg-Marquardt backpropagation (Trainlm) training function with the log sigmoid transfer function recorded the best prediction results of both ANN structures.

Artificial intelligence based succored power prediction in stand alone PV system

ABSTRACT Accurate forecasting of solar power generation is crucial for effective management of renewable energy resources. This literature review emphasises the effectiveness of Artificial Neural Networks (ANNs) in predicting solar energy potential. While traditional ANNs with single hidden layers and predefined neuron configurations have shown success in various applications, they are not suitable for predicting nonlinear time-series databases in the context of this research. The limitations of traditional ANNs in predicting solar power potential necessitate the exploration of alternative approaches. Configuring the structure of traditional ANNs for solar power prediction is a time-consuming process that often requires manual or trial-and-error methods. To address this issue, optimisation techniques are incorporated into the ANN model. Various traditional methods have been explored in this research, but none has achieved satisfactory performance. Drawing from experiences with traditional techniques, a novel Evolution of Cub to Predator (ECP) technique is proposed. The investigation results demonstrate that the incorporation of optimisation techniques yields superior performance compared to traditional ANNs. The proposed ECP technique achieves an impressive prediction accuracy of 97.2%, surpassing existing optimisation techniques. These findings highlight the potential of the ECP technique for accurate solar power prediction and its contribution to optimising renewable energy resource management.

Evolutionary neural networks for learning turbulence closure models with explicit expressions

Developing physical closure models with explicit expressions based on a given dataset is essential to science and engineering. For such symbolic regression tasks, biology-inspired evolutionary algorithms are most widely used. However, typical evolutionary algorithms do not utilize any structural information inherent in training data, which limits their performance in finding accurate model structures and coefficients. By combining one evolutionary algorithm, gene expression programing (GEP), with an artificial neural network (ANN) for symbolic regression, we propose a novel evolutionary neural network method, in which candidate expressions are specifically designed so that they can be transformed between the GEP and ANN structures during training iterations. By combining the GEP's global searching and the ANN's gradient optimization capabilities, efficient and robust convergence to accurate models can be achieved. In addition, sparsity-enhancing strategies have been introduced to improve the interpretability of the trained models. The present method has been tested for finding different physical laws and then applied to turbulence modeling problems with different configurations, showing advantages compared to the existing GEP and ANN methods.

Velocity Estimations in Blood Microflows via Machine Learning Symmetries

Improving velocity forecasts of blood microflows could be useful in biomedical applications. We focus on estimating the velocity of the blood in capillaries. Modeling blood microflow in capillaries is a complex process. In this paper, we use artificial intelligence techniques for this modeling: more precisely, artificial neural networks (ANNs). The selected model is able to accurately forecast the velocity, with an R2 of 0.8992 comparing the forecast with the actual velocity. A key part of ANN model creation is selecting the appropriate parameters for the ANN, such as the number of neurons, the number of layers and the type of training algorithm used. A grid approach with 327,600 simulations was used. It is shown that there are substantial, statistically significant differences when different types of ANN structures are used. It is also shown that the proposed model is robust regarding the initial random initialization of weights in the ANN. Additionally, the sensitivity of the selected models to additional noise was also tested.

Adaptive Online Neural Predictive Control of Hydraulic Actuator Using FPGA Acceleration

Hydraulic actuators have recently been widely used in control systems as efficient output devices. However, high-precision control of hydraulic actuators in industrial applications remains challenging due to the conflict between controllers' complexity, especially in theoretical modeling, and their generalization abilities, including adaptability to parametric uncertainties and versatility for various working conditions. To address this issue, we propose an adaptive online neural predictive control (AONPC) scheme, which is capable of handling unmodeled nonlinearities and uncertainties with a simple structure and low demand for system modeling. Specifically, an artificial neural network (ANN) is responsible for data-driven system identification and predictive control, while upper confidence bound (UCB) algorithm is introduced for hyperparameter self-adaptation. The ANN structure and the iteration number for solving the optimization problem are determined through a simulation study. Then, to complete the whole control loop within a short and fixed tracking interval, the proposed controller is implemented on a field programmable gate array (FPGA), which can effectively improve the tracking precision. Finally, extensive experimental results on a realistic hydraulic actuator device demonstrate the superiority of the proposed controller.

A memetic dynamic coral reef optimisation algorithm for simultaneous training, design, and optimisation of artificial neural networks

Artificial Neural Networks (ANNs) have been used in a multitude of real-world applications given their predictive capabilities, and algorithms based on gradient descent, such as Backpropagation (BP) and variants, are usually considered for their optimisation. However, these algorithms have been shown to get stuck at local optima, and they require a cautious design of the architecture of the model. This paper proposes a novel memetic training method for simultaneously learning the ANNs structure and weights based on the Coral Reef Optimisation algorithms (CROs), a global-search metaheuristic based on corals’ biology and coral reef formation. Three versions based on the original CRO combined with a Local Search procedure are developed: (1) the basic one, called Memetic CRO; (2) a statistically guided version called Memetic SCRO (M-SCRO) that adjusts the algorithm parameters based on the population fitness; (3) and, finally, an improved Dynamic Statistically-driven version called Memetic Dynamic SCRO (M-DSCRO). M-DSCRO is designed with the idea of improving the M-SCRO version in the evolutionary process, evaluating whether the fitness distribution of the population of ANNs is normal to automatically decide the statistic to be used for assigning the algorithm parameters. Furthermore, all algorithms are adapted to the design of ANNs by means of the most suitable operators. The performance of the different algorithms is evaluated with 40 classification datasets, showing that the proposed M-DSCRO algorithm outperforms the other two versions on most of the datasets. In the final analysis, M-DSCRO is compared against four state-of-the-art methods, demonstrating its superior efficacy in terms of overall accuracy and minority class performance.

Comparative study of different training algorithms in backpropagation neural networks for generalized biomass higher heating value prediction

When selecting biomass feedstock for sustainable heat and electricity generation, higher heating value (HHV) is an important consideration. Meanwhile, the laboratory procedures of using an adiabatic oxygen bomb calorimeter to determine the HHV are strenuous, costly, and time-consuming. As a result, researchers have turned to artificial intelligence techniques such as artificial neural networks (ANN) to predict HHV using data from proximate analysis. Notwithstanding, this approach has been hampered by different case-specific techniques and methodologies given the heterogeneous nature of biomass materials and intricate ANN structures. This study, therefore, examined and compared the efficacy of six training algorithms comprising thirteen distinct training functions of feedforward backpropagation neural networks to predict the HHV of a variety of biomass materials as a function of the proximate analysis. In creating the networks, the neurons of the hidden layer were iterated from 1 to 20 leading to 260 investigated scenarios. Compared to other training algorithms, the Bayesian Regularization and Levenberg-Marquardt with 15 and 12 hidden neurons respectively, demonstrated superior prediction performances based on the Nash-Sutcliff's efficiencies of 0.9044 and 0.8877, and mean squared errors of 0.002271 and 0.00267. It is envisaged that this study will create an insightful paradigm for a rapid selection of best-performing ANN algorithms for biomass HHV prediction.

Optimal design of heat pipes for city gate station heaters by applying genetic and Bayesian optimization algorithms to an artificial neural network model

This study contains a novel idea to improve the performance of city gate station heaters, using pulsating heat pipes with optimized structure. In this research, a hybrid method including Bayesian and genetic algorithms is implemented to optimize the architecture of an artificial neural network (ANN) model and find the optimal operating point for heat pipes. In the previous study, the authors have proposed a neural network-based model to predict the thermal performance of heat pipes, based on a wide range of their design parameters, including dimensions of heat pipe, number of turns, working fluid, inclination angle, filling ratio, and heat input. In the present study, the prediction model architecture is optimized using Bayesian optimization algorithm. The best performance is achieved when iterations of the employed Bayesian optimization is set to 300 iterations or samples, which 150 samples are produced to train the Bayesian model and other 150 iterations are utilized in order to find the optimal ANN structure and hyper-parameter tuning of the model. The optimal ANN model with structure of 9-97-21-68-100-1 has evaluation parameters of 0.980 and 0.014 for R2 and MSE of total dataset, respectively. The performance metrics indicate an improvement in accuracy of the ANN model with a 53.33% reduction in MSE value and a 3.15% increase in R2 score. Then, the genetic algorithm is applied to the optimal model to find the optimum operating parameters of the heat pipe. In this regard, the design parameters of the best heat pipe in order to use in the city gate station heaters, are obtained by setting the genetic algorithm population equal to 300. Finally, as a practical case study, a real heater with capacity of 10,000 SCMH which is located in Iran is selected. Considering constraints and limitations of this heater, the proposed model indicates that a vertical pulsating heat pipe with four turns which is filled 55% with water as the working fluid, has the optimal thermal resistance of 0.104 K/W. Utilizing the optimal heat pipe in the mentioned heater is a promising solution for efficiency improvement and fuel consumption and CO2 emission reduction, which decreases the operating costs of the heater as well.

Analysis of Compressive Strength of Sustainable Fibre Reinforced Foamed Concrete Using Machine Learning Techniques

Abstract This study emphasizes the usage of Silica Fume (SF) and Marble Sludge Powder (MSP) as a partial replacement for fly ash in Fibre Reinforced Foamed Concrete (FRFC). The compressive strength for various samples was analyzed using Artificial Neural Network (ANN) methods. In this research work, the utilization of silica fume, fly ash, marble sludge powder, polypropylene fiber, and foaming agent in fiber-reinforced foamed concrete is presented and a sincere attempt has been made to use silica fume and marble sludge powder for the replacement of fly ash with various percentages. In addition to that polypropylene fiber (PPF) was used in various proportions of 0%, 0.1%, 0.2%, 0.3%, 0.4%, and 0.5%. The Feed Forward Propagation (FFP) network of the machine learning method with one hidden layer was taken as the ANN structure of FRFC. In this ANN work, cement, silica fume, fly ash, marble sludge powder, foaming agent, water, and polypropylene fiber were used as input parameters and compressive strength is the output parameter. The correlation coefficient with the ANN methods was found as 0.940 for compressive strength. In machine learning techniques, the ANN method was found to be accurate in estimating and analyzing strength prediction responses with effective parameters.

Algorithms Don’t Have A Future: On the Relation of Judgement and Calculation

This paper is about the opposite of judgement and calculation. This opposition has been a traditional anchor of critiques concerned with the rise of AI decision making over human judgement. Contrary to these approaches, it is argued that human judgement is not and cannot be replaced by calculation, but that it is human judgement that contextualises computational structures and gives them meaning and purpose. The article focuses on the epistemic structure of algorithms and artificial neural networks to find that they always depend on human judgement to be related to real life objects or purposes. By introducing the philosophical concept of judgement, it becomes clear that the property of judgement to provide meaning and purposiveness is based on the temporality of human life and the ambiguity of language, which quantitative processes lack. A juxtaposition shows that calculations and clustering can be used and referred to in more or less prejudiced and reflecting as well as opaque and transparent ways, but thereby always depend on human judgement. The paper clearly asserts that the transparency of AI is necessary for their autonomous use. This transparency requires the explicitness of the judgements that constitute these computational structures, thereby creating an awareness of the conditionality of such epistemic entities.

Feasibility of Six Metaheuristic Solutions for Estimating Induction Motor Reactance

Industry is the primary application for induction machines. As such, it is essential to calculate the induction devices’ electrical properties accurately. With DC testing, no-load rotor tests, and locked rotor tests, one may empirically evaluate the electrical variables of induction motors. These tests are expensive and difficult to conduct, however. The information supplied by machine makers can also be used to accurately approximate the equivalent variables of the circuits in induction machines. This article has successfully predicted motor reactance (Xm) for both double- and single-cage models using artificial neural networks (ANN). Although ANNs have been investigated in the literature, the ANN structures were trained to use unmemorized training. Besides ANN, six other approaches have been suggested to address this issue: heap-based optimization (HBO), leagues championship algorithm (LCA), multi-verse optimization (MVO), osprey optimization algorithm (OOA), cuckoo optimization algorithm (COA), and sooty tern optimization algorithm (STOA). The efficaciousness of the suggested approaches was compared with each another. Regarding the obtained outcomes, the suggested MVO- multi-layer perceptron (MLP) technique performed better than the other five methods regarding reactance prediction, with R2 of 0.99598 and 0.9962, and RMSE of 20.31492 and 20.80626 in the testing and training phases, respectively. For the projected model, the suggested ANNs have produced great results. The novelty lies in the mentioned methods’ ability to tackle the complexities and challenges associated with induction motor reactance optimization, providing innovative approaches to finding optimal or near-optimal solutions. As researchers continue to explore and refine these techniques, their impact on motor design and efficiency will likely grow, driving advancements in electrical engineering.

Optimization of Neurons Number in Artificial Neural Network Model for Predicting the Power Production of PV Module

In this work, an Artificial Neural Network (ANN) with a backward-propagation technique was used to predict the power generation of the Photovoltaic (PV) module in weather conditions of Baghdad city-Iraq. Experiment tests were investigated in the summer of 2022. Three weather parameters, including: (solar radiation, ambient temperature, and wind speed), the output electrical characteristics of the PV module (voltage, current, power), and module temperature (were measured. Therefore, the dataset of the ANN system consists of four input and one output parameter. Furthermore, the structure of ANN includes a single hidden layer with a backward propagation technique. The main goal of this study was to optimize the number of neurons in the training process. The evaluation of the ANN model depended on the determination coefficient (R) and Root Mean Squared Error (RMSE). The obtained results show that the architecture of ANN is appropriate for predicting the power generated from the PV module. The developed ANN model has good accuracy. Where the MSE is 0.002747 at epoch 9 in the model. Furthermore, the R is recorded as 0.99078, 0.98254, 0.99125, and 0.99005 for training, testing, validation, and all respectively in the proposed model. In addition, the optimization number of neurons in the hidden layer gave sufficient accuracy without referring to the choice of the number of neurons by using the trial-and-error method that most researchers relied.

Prediction of ball-on-plate friction and wear by ANN with data-driven optimization

For training artificial neural network (ANN), big data either generated by machine or measured from experiments are used as input to “learn” the unspecified functions defining the ANN. The experimental data are fed directly into the optimizer allowing training to be performed according to a predefined loss function. To predict sliding friction and wear at mixed lubrication conditions, in this study a specific ANN structure was so designed that deep learning algorithms and data-driven optimization models can be used. Experimental ball-on-plate friction and wear data were analyzed using the specific training procedure to optimize the weights and biases incorporated into the neural layers of the ANN, and only two independent experimental data sets were used during the ANN optimization procedure. After the training procedure, the ANN is capable to predict the contact and hydrodynamic pressure by adapting the output data according to the tribological condition implemented in the optimization algorithm.

Повышение точности идентификации и настройки линейных систем с регуляторами состояния при помощи искусственной нейронной сети

High potential capabilities of control systems with state controllers can be realized only if automatic tuning tools are available. Since the tuning is carried out in real-time mode, which places increased demands on performance, it is proposed to use an artificial neural network to reduce its duration. However, under the conditions of noise in the measurement channels, the quality of identification of the parameters of the control object is significantly reduced. In this regard, the aim of the study is to find the optimal composition of measurement channels at the network input, which allows minimizing the influence of noise on the estimates of object parameters to improve the quality of tuning. During the study, state space methods are used to design a vector-matrix model of an object and synthesize a state controller. A radial artificial neural network is used to solve the problem of identifying the parameters of a vector-matrix model. The training of networks, the study of the effectiveness of their work, as well as the development of models is carried out using the tools of the MatLab software package. The authors have developed the method to select the optimal composition of measurement channels which gives the maximum signal-to-noise ratio and forming the corresponding structure of a radial artificial neural network to solve the problems of object parameters identification and control system tuning with state controller. It is proposed to use the sensitivity functions of the state coordinates of control object parameters variation to estimate power of information signals at the inputs of neural network. The results of the conducted computational experiments have confirmed the effectiveness of the developed method, which makes it possible to increase the accuracy of identification and tuning of systems with state regulators under noise conditions. The obtained results can be used to ensure a given quality of control with parametric uncertainty of the object.

A decreasing failure rate model with a novel approach to enhance the artificial neural network's structure for engineering and disease data analysis

The study focuses on key metrics used to examine the characteristics of a lifetime random variable distribution in reliability and survival theory research. In this analysis, metrics including the probability density function time, mean residual lifespan, mean time between failures, hazard rate, and reliability function are essential. The focus of the inquiry is these important parameters in relation to the Burr-Hatke exponential model specifically. The study focuses on key metrics used to examine the characteristics of a lifetime random variable distribution in reliability and survival theory research. In this analysis, metrics including the probability density function time, mean residual lifespan, mean time between failures, hazard rate, and reliability function are essential The focus of the inquiry is these important parameters in relation to the Burr-Hatke exponential model specifically. A key component of the research is a comparison of the outcomes from the artificial intelligence approach and those from conventional literature-based methodologies. This comparison study sheds light on how well the artificial neural network framework performs while evaluating the Burr-Hatke exponential model's technical features. The study allows a comprehensive analysis of the training and prediction capabilities of the growing neural network by calculating multiple performance measures. This comprehensive strategy improves our comprehension of the model's survival traits and reliability, offering significant contributions to the larger field of study. The network structure's mean square error was estimated to be 5.19E-04, and its coefficient of determination value was 0.99987 for the first neural network model. For the second neural network model, the coefficient of determination value was 0.99999 and the mean square error value was 4.58E-06. The outcomes amply revealed the neural network structure's extraordinarily high prediction accuracy and the degree to which the prediction outputs agree with those of the Maximum Likelihood Estimation technique.

Thermal performance estimation of a v-corrugated solar air heater using ann techniques

A solar air heater (SAH) is a unique form of solar thermal collector that utilizes solar energy emitted from the sun to produce heated air. Various experimental and theoretical investigations have been undertaken to improve the poor thermal performance of SAHs. The difficulties related to these studies drew attention toward a reliable soft computing technique exemplified by the Artificial Neural Network (ANN) technique. The current work applied actual meteorological data from Miskolc City, Hungary, to an ANN model with the structure of a Multi-layer Perceptron (MLP) to forecast the energy performance of a V-corrugated solar-powered air heater. Seven input parameters and one output parameter make up the ANN structure, with a single hidden layer. For the purpose of selecting the most effective network for predicting output parameters, ten neurons have been assessed. The suggested ANN model was trained with 336 data sets using the Levenberg-Marquardt (LM) learning technique. The comparison of anticipated and real thermal performance values shows a very good agreement. The statistical error analysis showed that the optimal ANN model structure of 7-8-1 can reliably and accurately predict SAH’s thermal performance and thus it can save both time and cost.

Combining Digital Imaging and Artificial Neural Networks for Individual Volumetric Estimation of Non-Straight Trees

Tree volume models have become an important decision-making tool in forest resources management. The objective of this work was to create Artificial Neural Networks (ANN) to fit individual tree models from the Brazilian Savanna (the Cerrado biome), characterized by non-straight and very bifurcated tree stems. Data were collected in two different areas and through different methods: i) individual tree volume by digital photographs, and also by tree climbing measurements (both obtained in a non-destructive way); ii) tree volumes measured destructively by the traditional rigorous cubing procedure after thinning. Data obtained from climbing measurements were compared with the corresponding tree obtained from digital photographs, resulting in a reasonable operational linear relation of 0.80. Ten ANN structures were created with different combinations of forestry variables. Five ANN structures depicted predictive ability results above 0.80. Then, the results obtained by ANN were compared with those obtained by classic tree volumetric models, once again showing the predictive superiority of ANN. The present study indicates that tree volume obtained from digital photographs, achieved by popular mobiles, can be safely used for the estimation of very-crooked trees from the Brazilian Savanna and demonstrated the greater predictive ability of ANN in relation to the classic regression-based volumetric models in terms of lower modeling bias.

Modelling of chemical processes using artificial neural network

ABSTRACT In this paper, a methodical strategy is put forth for the development of models for chemical processes by using an Artificial Neural Network (ANN). To show the effectiveness of the proposed methodology, three industrial chemical processes were considered. The intended work is to introduce the guideline for designing any ANN model by identifying the optimum number of neurons and hidden layers and coming up with the most suitable architecture for the neural network. The effect of viable parameters on predicting the specified parameter was evaluated using an ANN model. A systematic procedure adopted, ANN networks were designed to predict the boiler efficiency, frost thermal conductivity, and gas holdup under different operating conditions. At last, a comparative study on the experimental output and ANN approximated output for the three chemical processes was done. For developing the ANN model for boiler and frost conductivity, it was found that employing two layers provided a superior fit. The R² values for these fits were found to be 0.98877 and 0.9891, respectively. On the other hand, when configuring the column flotation model, a single layer with nine nodes was identified as the optimal ANN structure, yielding an excellent R² fit value of 0.99337.

Adaptive Neuroevolution With Genetic Operator Control and Two-Way Complexity Variation

Topology and weight evolving artificial neural network (TWEANN) algorithms optimize the structure and weights of artificial neural networks (ANNs) simultaneously. The resulting networks are typically used as policy models for solving control and reinforcement learning (RL) type problems. This paper presents a neuroevolution algorithm that aims to address the typical stagnation and sluggish convergence issues present in other neuroevolution algorithms. These issues are often caused by inadequacies in population diversity preservation, exploration/exploitation balance, and search flexibility. This new algorithm, called the Adaptive Genomic Evolution of Neural-Network Topologies (AGENT), builds on the neuroevolution of augmenting topologies (NEAT) concept. Novel mechanisms for adapting the selection and mutation operations are proposed to favorably control population diversity and exploration/exploitation balance. The former is founded on a fundamentally new way of quantifying diversity by taking a graph-theoretic perspective of the population of genomes and inter-genomic differences. Further advancements to the NEAT paradigm occur through the incorporation of variable neuronal properties and new mutation operations that uniquely allow both the growth and pruning of ANN topologies during evolution. Numerical experiments with benchmark control problems adopted from the OpenAI Gym illustrate the competitive performance of AGENT against standard RL methods and adaptive HyperNEAT, and superiority over the original NEAT algorithm. Further parametric analysis provides key insights into the impact of the new features in AGENT. This is followed by evaluation on an unmanned aerial vehicle collision avoidance problem where maneuver planning models are learnt by AGENT with 33% reward improvement over 15 generations.