The electrification of offshore oil and gas platforms (OOGPs), using offshore wind turbines, is now a reality, aimed at minimizing carbon emissions. In this scenario, the utilization of battery energy storage systems ...
详细信息
Humanoid robot is able to mimic repetitive activities done by human. By having them, humans can focus on tasks that require extra focus, while leaving the simple one to be done by humanoid robot. To extend the functio...
详细信息
Steelmaking slag has been extensively studied as aggregate for cement-based composites. Because of the distinct properties of this residue, traditional mix design methods are not suitable to determine its target compr...
详细信息
Steelmaking slag has been extensively studied as aggregate for cement-based composites. Because of the distinct properties of this residue, traditional mix design methods are not suitable to determine its target compressive strength, which hinders research studies and compromises its use on a large scale. In this context, the present work aims to develop machine learning (ML)-based models to predict the compressive strength of steel slag concretes from their mix proportion. For this purpose, a global data survey on steel slag concretes was carried out to create a benchmark dataset. Then, four ML-based models were trained and cross-validated using this dataset: Support Vector Regression (SVR), Artificial Neural Networks (ANN), Extreme Gradient Boost (XGBoost), and Gaussian Process Regression (GPR). Finally, new steel slag concrete specimens were built and tested to experi-mentally validate the adjusted models. The model that achieved the best performance using the literature dataset was the ANN, with a R2 of 0.79. However, the experimental validation was not satisfactory - the GPR, XGBoost and SVR models presented negative R2 values. These results brought light to some pivotal aspects that must be considered when using ML techniques: i) the size and homogeneity of the dataset;ii) the proper choice of input parameters;and iii) the use of cross-validation to adjust the models. Hence, although such techniques are promising and powerful, care must be taken on the generalization of their predictions, especially when the available data is limited.
In this work, a sliding detection algorithm was proposed for a previously developed robotic hand through the utilization of force sensors. Previous works have designed several hardware architectures to filter sensor d...
In this work, a sliding detection algorithm was proposed for a previously developed robotic hand through the utilization of force sensors. Previous works have designed several hardware architectures to filter sensor data, implement dynamic control, and machine learning models for controlling the robotic fingers using a single Field programmable Gate Arrays (FPGA) chip. In this regard, the slip detection algorithm was developed to comprise three stages, each with a low computational cost, including a moving average filter, a first-order derivative, and a peak detection algorithm. A reference model was constructed and validated through an experimental protocol employing daily use objects to create a database. Subsequently, the slip detection algorithm was mapped onto hardware utilizing previously developed floating-point arithmetic IP cores and implemented using a Zynq 7020 device. The FPGA implementation was characterized in terms of resource occupation, power consumption, execution time, and numerical error with the reference model.
The development of interoperability is more and more an essential task for all kinds of organizations. It needs to be measured, verified, and continuously improved. With the advent of the Internet of Things, Industry ...
详细信息
Several coffee producers rely on manual inspection for quality control, which is prone to inconsistencies and time consumption. Recent studies have proposed Comparison performance on State-of-The-Art (SOTA). Deep lear...
详细信息
ISBN:
(数字)9789887581598
ISBN:
(纸本)9798331540845
Several coffee producers rely on manual inspection for quality control, which is prone to inconsistencies and time consumption. Recent studies have proposed Comparison performance on State-of-The-Art (SOTA). Deep learning models were investigated as part of the proposed intelligent computer vision for faster and more reliable coffee bean grading. Performance Deep learning models, i.e., MobileNetV3Small, ResNet50, EfficientNetV2L, and DenseNet201, were evaluated to grade the quality of coffee beans that followed a grading procedure based on Indonesian National Standard (SNI) 01-2907-2008. A specific tray was designed to position coffee beans evenly without overlap for sample preparation, which consist of about 500–600 coffee beans. The process of capturing images is conducted systematically, focusing on each type of coffee bean defect individually. For comparison of SOTA models, five classes of coffee bean defects, With 1053 beans per class, the total dataset comprises 5265 beans. This dataset was divided into 10% for testing and 90% for training. Within the training data, 12% was reserved for validation. The result from this paper shows that MobileNetV3Small achieves the highest accuracy (99.43%) with minimal memory and parameters. Its efficient design allows for robust feature extraction, crucial for distinguishing subtle defect features. While larger models like ResNet50 and EfficientNetV2L also perform well, they exhibit higher memory and parameters.
Robotic manipulators are multi-input multi-output (MIMO) systems with nonlinear points affected by numerous uncertainties and disturbances. PID controllers are widely used in industry for kinematic and dynamic control...
详细信息
ISBN:
(数字)9781665462808
ISBN:
(纸本)9781665462815
Robotic manipulators are multi-input multi-output (MIMO) systems with nonlinear points affected by numerous uncertainties and disturbances. PID controllers are widely used in industry for kinematic and dynamic control. However, when applied to MIMO systems, they are not easy to tune and require performance improvements. In this work, a PID controller is proposed with a fuzzy precompensator (FP-PID), both tuned by the bioinspired particle swarm optimization (PSO) algorithm to a two-degree of freedom (2-DOF) robotic manipulator representing a human leg. To validate the system, two real datasets of human gait were used: normal walking and stair climbing to estimate the error trajectory of the manipulator. The statistical analysis of the PSO algorithm with 16 experiments was satisfactory, and the addition of the fuzzy precompensator to the conventional PID resulted in a reduction of the mean square error of one of the manipulator links by up to 73 percent.
The electrification of offshore oil and gas platforms (OOGPs), using offshore wind turbines, is now a reality, aimed at minimizing carbon emissions. In this scenario, the utilization of battery energy storage systems ...
详细信息
ISBN:
(数字)9798350368154
ISBN:
(纸本)9798350368161
The electrification of offshore oil and gas platforms (OOGPs), using offshore wind turbines, is now a reality, aimed at minimizing carbon emissions. In this scenario, the utilization of battery energy storage systems (BESS) improves both efficiency and power quality. As this integration comes with a high CAPEX, industries demand exploring BESS for multiple ancillary services on OOGPs. Thus, this paper proposes an alternative approach to energizing a power transformer (PT) using the BESS power conversion system (PCS) on the platform, aiming to minimize inrush currents. This avoids the necessity for additional bulky devices, converters, or dc voltage sources specifically designated for this purpose. The strategy consists of manipulating residual core flux using BESS PCS followed by a controlled switching, which reduces inrush currents during PT energization. The results show effective reduction of the inrush current from 2.31 pu to 0.16 pu for the presented case study.
With the advancement of technology, systems in many different areas of application have become more complex, and obtaining reliable mathematical models is a task that is increasingly becoming more difficult as well. T...
详细信息
ISBN:
(数字)9781665462808
ISBN:
(纸本)9781665462815
With the advancement of technology, systems in many different areas of application have become more complex, and obtaining reliable mathematical models is a task that is increasingly becoming more difficult as well. To overcome these issues, new methodologies have emerged, such as the one known as Data-Driven-control (DDC), which uses the input and output data of the system to design a controller, and Behavior-Based-control (BBC), which provides the controller based on the desired behavior of a system. In this work, these two methodologies were applied to obtain the transfer function of a Single-Input Single-Output mobile robotic platform that moves parallel to a wall. A Proportional Integral Derivative controller was used to control the left wheel, while the right wheel was set to a constant speed, allowing the mobile robot to maintain a constant distance from the wall. The controller was tuned utilizing root locus methodology with an overshot of 16.7%, settling time of 6.45 seconds, and a static error of 2.1%. Additionally, disturbances were added to test the robustness of the mobile robot model and the PID controller.
Automated Guided Vehicles (AGVs) are usual in industrial settings, with an increasing economic impact on processes. They move through numerous environments inside factories, commonly navigating long distances, while p...
详细信息
ISBN:
(数字)9781665462808
ISBN:
(纸本)9781665462815
Automated Guided Vehicles (AGVs) are usual in industrial settings, with an increasing economic impact on processes. They move through numerous environments inside factories, commonly navigating long distances, while performing several activities. The constant detection of such vehicles, especially in cases of maintenance and safety, is a main issue in an industry setting. Usually, this information could be provided by a supervisory system, but many applications are not so large as to make such a system viable. Thus, a solution via machine learning and computer vision is developed in this work, by using simple cameras, such as the factories' security cameras. As the industrial environment is a scenario with a lot of variation and noise, the Transfer Learning technique is used to improve the training step of the developed AGV detection system. Finally, a database with 1067 images is used to build and validate the model, achieving a result greater than 80% of F1-score for various confidence values.
暂无评论