Applied Computer Sciences in Engineering

Effective resource allocation during natural disasters and response strategies are crucial to mitigate impacts and save lives. This paper presents a comprehensive mathematical model, which is designed to optimize disaster response operations. The proposed model integrates various components, including the demand for different types of relief services, facilities distributed by zones, as well as the assessment of resource requirements, and optimization of the location capacities per zone. We propose the use of mixed-integer linear programming (MILP), where we propose three different objective functions. The model was tested over a real case in Bogotá-Colombia and we evaluate different scenarios to analyze the performance of the model and obtained results. The proposed framework offers a strategic tool for policymakers and emergency managers to enable data-driven decision-making in disaster preparedness and response planning.

The Index Tracking Problem involves the creation of an investment portfolio that accurately replicates the behavior of a market index. Being an NP-Hard optimization problem, it is well-suited for metaheuristic approaches. Drawing inspiration from the Harmony Search algorithm, the Harmony Search algorithm for Portfolio Optimization is presented. This algorithm addresses the problem’s complex constraints by utilizing two search operators that effectively handle both the problem itself and the most difficult constraints, and it incorporates two population initialization strategies that aid in achieving convergence and can deal with problems of big size.

In this paper, we binarize a novel algorithm called the Fox Optimizer using a two-step technique and test its performance against the Set Covering Problem. Additionally, we explore the incorporation of chaotic maps into the binarization process. To benchmark the binary Fox Optimizer, we compare it with two well-known and documented metaheuristics: Particle Swarm Optimization and Grey Wolf Optimizer. Each algorithm is tested with standard, sine chaotic, elitist, and elitist sine chaotic binarization rules. Our findings demonstrate that elitist configurations, especially when combined with sine chaotic binarization, consistently yield superior results, providing robust and reliable performance in obtaining high-quality solutions. Conversely, standard binarization configurations exhibit enhanced convergence capabilities, proving effective for problems with rapid convergence requirements or lower complexity. This study highlights the importance of aligning algorithm configurations with specific problem characteristics to optimize performance in practical applications.

In modern power distribution systems, the integration of distributed generation (DG) projects is crucial for enhancing system efficiency, safety, and sustainability. However, the optimal selection of DG projects amid various technical, economic, and operational constraints remains challenging. Previous research has focused on optimal placement and sizing of DG, but the realities faced by distribution system operators (DSOs) often involve evaluating predetermined project proposals. This paper introduces a novel approach for the optimal selection of DG projects in power distribution systems using a genetic algorithm (GA) framework integrated with DIgSILENT PowerFactory. The proposed methodology employs a GA to maximize system performance by considering factors such as voltage regulation, line loadability, power losses, and the direction of power flow while adhering to system constraints. The integration with DIgSILENT PowerFactory enables realistic simulation through unbalanced power flow analysis and evaluation of candidate DG projects within the distribution system context. The approach is implemented using DIgSILENT’s built-in programming language, facilitating direct utilization of utility databases and intricate modeling of power system elements. A case study on a real Colombian utility’s distribution feeder demonstrates the effectiveness of the proposed approach. The method achieved a 70.79% reduction in power losses, a 66.57% decrease in maximum line loadability, and voltage profile improvements of up to 7.5% at critical buses, while ensuring no reverse power flow towards the substation. This research contributes to advancing power system optimization by providing DSOs with a practical tool for assessing and selecting DG projects that enhance system performance while mitigating potential negative impacts. The integration with DIgSILENT PowerFactory ensures applicability in real-world scenarios. Future work may explore incorporating additional objectives such as reliability indices and economic factors, and revisiting the constraint on reverse power flow as distribution systems evolve.

Agri-food supply chains (ASC) focus on stakeholders with the objective of providing access to essential inputs, processing primary products, marketing and distributing the products to end consumers. In this context, it is proposed to design a model for locating rural centers that serves as logistical support for the yam agri-food supply chain. This location model for hubs aims to plan the distribution in the agri-food supply chain of yams in the Sabanas Subregion of the Department of Sucre, Colombia. Additionally, the use of the Hub Location Problem is considered, which is addressed as a mixed integer linear programming model. To this end, assumptions have been adopted that will allow obtaining results that maximize the profits of the warehouses (hubs). These results have been obtained through the use of GAMS software, solved with the CPLEX Solver. According to the results obtained, the optimal location for the opening of two strategic hubs has been identified. These distribution centers not only help reduce logistics costs, but also optimize delivery times, thus strengthening the supply chain.

Rural development faces major challenges related to infrastructure, youth migration, employment shortages, and food security. Agriculture is a dynamic sector that is crucial in addressing these challenges. In this context, this work conducted a theoretical review of the optimization techniques used and the information available in a specific case study. From the above, two techniques were selected, linear programming and stochastic linear programming, to efficiently solve this problem. Four models were proposed to define the number of hectares to be planted for the crops \(i\) in the municipalities \(j\), considering food security conditions and some limiting factors in each territory. The results indicate that the amount of food and income produced in the municipalities is significantly lower than that obtained with the proposed models; the tons of food produced represent 15% of the amount projected with the linear programming model and do not account for 1% of the income generated by said model. Likewise, it is observed that the third model, which has two stochastic parameters, yields better results regarding the number of hectares planted, the tons of food produced, and the income generated in comparison with the second model, which only defines one parameter. In conclusion, these optimization models allow analyzing municipalities’ participation and diversification in relation to food production in the region, and their results are better that those of current empirical decisions.

This paper presents the implementation of two methods for sparse index tracking of the US Nasdaq 100 index. The proposed methods employ a regularization approach based on a non-convex cardinality constraint approximation of the ℓ_p-norm to identify the optimal asset weights of the tracking portfolio. The results demonstrate that the ℓ₀-norm-constrained sparse tracking portfolio is computationally efficient and exhibits a notable reduction in tracking errors during the out-of-sample testing periods. Furthermore, we present an empirical comparison of the results of performance measures with those of traditional constrained strategies using norm constraints for index tracking.

This article describes the details of the configuration of a physical benchmark that is useful for assessing performance degradation of networked control systems (NCS) when exposed to cyber-attacks (for example, denial-of-service (DoS) attacks, deception attacks, man-in-the-middle (MiTM) attacks, and others, as stated in an NCS vulnerability taxonomy performed here), and explore countermeasures. The benchmark is a temperature control system controlled remotely through WIFI communication established in a private network under the UDP protocol. This is controlled here by an offset-free model predictive control. Two scenarios were visualized and tested, allowing input constraints, disturbances, mismatches, and DoS attacks. Experimental results are discussed, illustrating the potential use of this benchmark.

Oncolytic virotherapy with measles represents a promising approach to combat cancer, often combined with other therapies to eradicate tumors. However, it faces many challenges, for instance, the toxic side effects that must be carefully managed during treatment, the optimal dose to apply, the determination time to extirpate the tumor, the monitoring system, and others. In this study, predictive control methodologies are employed based on the nonlinear system representing oncolytic virotherapy with measles. The objective is to present strategies for cancer treatment by either limiting the maximum dose per patient or bringing the tumor to a point where it can be surgically removed. This method utilizes system approximations at discrete time intervals for the model subsystems. To address the limitation of the maximum allowed injected virus dose, an additional state describing the total injected dose is incorporated. Simulations demonstrate dose optimization for tumor reduction, with successful therapy in up to 18 days. Moreover, careful dosing strategies maximize therapeutic efficacy while minimizing potential toxicity, underscoring the promising potential in achieving optimal treatment outcomes. These findings support the idea that feedback control has the potential to enhance robustness and toxicity reduction in oncolytic virotherapy.

This work presents an application of a linear Kalman filter to control the effects of thermal expansion in a recirculating ball screw of a computer numerical control (CNC) machine, impacting the final position accuracy of the machining tool. Through the Kalman filter, the system’s state ball screw position is estimated iteratively, allowing for real-time compensation of thermal variations, dilation, and positional inaccuracies on machining precision. The effectiveness of the proposed approach is demonstrated through experimental validation using a dataset comprising positional and temperatures measurements. The performance of the Kalman filter was evaluated through statistical analysis and for measurements taken during the heating cycle, the Y-axis exhibited R² of 0.913 and \({\rm R}^{2}_{\rm adj}\) of 0.906, with an RMSE of 9.32 μm, indicating a good fit of the model to the data and acceptable precision of the estimates. For the X-axis under similar thermal conditions, higher values of R² and R²_adj were obtained with a lower RMSE confirming the high accuracy of the estimates and the method.

This article presents a method to evaluate batch production scheduling in a cassava starch production company. The company uses specialized machinery to process cassava. The method is based on a study of the practical operations of the company and the production technologies used, including the construction of a simulation model. The simulation model reflects the technical aspects of the production process and the company’s requirements. The simulation program FlexSim (from FlexSim Software Products, Inc.) was used in the experiments. The proposed method was evaluated using 12 possible scenarios. When comparing the simulated scenarios with the real data, it was concluded that there is an average percentage variation of 0.018%. These results highlight the effectiveness of advanced tools to optimize processes and improve efficiency in cassava starch production. The method was applied directly in a company to improve its performance; Furthermore, it is scalable and can be applied to problems of varying complexity and to production systems of different types and sizes, especially in small and medium-sized companies in the agri-food industries of the region.

This paper presents a cascade control strategy for attitude tracking in unmanned aerial vehicles (UAVs), addressing external actuator disturbances. UAVs have become increasingly relevant in various fields, necessitating precise control strategies for enhanced performance. The study focuses on a quadcopter model and employs a PID-P cascade controller, incorporating an additional loop to account for battery dynamics and disturbances in the electronic speed controllers (ESC) and brushless DC motors (BLDC). The proposed control strategy aims to improve trajectory tracking, disturbance rejection, and robustness. The research includes the development of a specialized test platform that allows for safe and accurate implementation and telemetry of the control algorithms. The results demonstrate the efficacy of the cascade control with an auxiliary loop in achieving desired angular positions and velocities, highlighting significant performance improvements compared to traditional control methods.

This study aims to estimate the battery life of a solar-electric boat using Python-based algorithms, with a primary focus on predicting battery degradation during the boat’s operation. Real-life operational data, including power consumption along a predetermined route and charging process information, were collected from the solar-electric boat. Laboratory experiments provided essential parameters of the battery cells. Leveraging this dataset, several Python-based algorithms were employed to simulate the expected battery life of the electric boat. The integration of real-world operational data and laboratory-derived cell parameters enhances the accuracy of the predictions, contributing valuable insights into the sustainable and efficient use of electric propulsion in waterway applications.

Conventional PID controllers have been a practical solution when controlling linear processes, but their time response is degraded in many non-linear processes. Fuzzy control has non-linear capabilities that may improve the performance against industrial processes. This article presents a comparative analysis for tuning fuzzy architectures: Direct, Supervisor, and Parallel in order to provide guidance in design decision-making. A non-linear mathematical model and continuous stirred tank reactor process have been considered to assess reference tracking and disturbance rejection. Results show that the Parallel fuzzy logic controller presents a more consistent response in the temporal performance indexes.

This study presents a dynamic orthosis designed to support the therapies of patients with muscle stiffness in the hand through gripping and relaxation movements. The orthosis features a modular design with sections for the phalanges, palm, and forearm, enabling adjustments to fit different hand and arm sizes. It uses motors to generate hand movements and feedback sensors for precise position adjustments during therapy. Testing demonstrated its effectiveness and adaptability in different users. Future improvements include validating the device with professionals for clinical use.

The maintenance and diagnosis of failures in mechanical machines, especially in induction motors, represent critical and costly challenges. Rotor failures, such as breaks or fractures, are especially problematic and can affect operational safety and efficiency. To address these challenges, intelligent fault diagnosis has become vital, employing machine learning techniques such as KNN, SVM and decision trees to prevent failures in real time. Motor current signature analysis (MCSA) stands out as a non-intrusive diagnostic technique, complemented by signal processing such as FFT and wavelet. Alternatives to MCSA include vibration signal analysis, with accelerometers capturing data and classification techniques identifying rotor and bearing faults. In this paper, the CRISP-DM model was applied, including data preprocessing, Fourier analysis and Hamming window for current and vibration signals. Machine learning models such as Random Forest and SVM were trained and evaluated, reaching an average accuracy of 90% when combining current and vibration data.

Ecoacoustics is a widely used wildlife passive monitoring discipline. Professional audio recorders offer robust casings able to resist adverse environmental conditions. However, the need of using clouds of recorders (for e.g. landscape level analyses) is obligating most researchers to migrate to low-cost devices, which unfortunately do not offer robust low-cost casings. In this work, we propose a home-made casing design and define a protocol of experiments for testing the recorder performance under real-life conditions. Specifically, we set-up an Audiomoth recorder for long-term wildlife monitoring in a Colombian rainforest and as a result, we find that it behaves similar to a professional SM4 recorder in terms of gain and dynamic range, and better in terms of soundscape information provided.

In recent years, the development of web applications for teaching and learning has become a focus of research, leading professionals from different fields to form interdisciplinary teams to work together on the integration and development of such tools. Requirements engineering has enabled researchers to identify and prioritize the needs and specifications that these technological tools must meet in order to have the greatest possible impact on the training process. From the implementation of semi-structured interviews, the evaluation of the needs of the environment, and the analysis of the state of the art, the methodological phases for the requirements engineering of web applications developed for the learning of topics related to the field of the health sciences are approached and prioritized. The methodological and strategic aspects applied in the elicitation of five groups of requirements are presented: (i) content, (ii) pedagogical, (iii) functional, (iv) non-functional, and (v) technical. All groups consider the synergy with the actors involved in the process of developing the application to support the learning of students belonging to health sciences or related programmes. The interaction and prioritization of the strategies proposed for the collection of requirements have allowed the generation of a guide for the fluid, clear and organized development of a web application for teaching in the health sciences. In which the involvement and high synergy of the actors involved promote the development of technological tools that have a greater impact on the acquisition and reinforcement of knowledge and skills.

About 65% of students who start engineering programs do not graduate. This situation causes significant economic losses and social problems for families and society and risks achieving social sustainability worldwide. Sometimes, engineering programs focus on enrolling students rather than providing strategies to secure their academic success, which often leads to student dropout. This study proposes a model to explain and predict engineering dropout through pedagogical, sociodemographic, and institutional factors. Using data from 4127 engineering students (cohorts 2005 – 2019), a structural equation model (SEM) and logistic regression demonstrate that institutional, demographic, and pedagogical variables explain and predict dropout in computer, electronic, environmental, and industrial engineering. According to SEM, institutional, sociodemographic, and pedagogical factors confirm the theoretical model. With Logistic regression as a predictable model, we could identify variables that predict almost 74.8% of student dropouts and 72.4% of student success. Our results provide novel insights to engineering programs and Higher Education institutions to implement curricula and pedagogical strategies leading to decreased engineering dropouts and increased student success.

Globally, the need for territorial entities to have a territorial information system in which the existing real estate and its physical and legal characteristics are registered is identified. This system would facilitate the administration and management of the resources of the same and thus allow the decision-making of different administrative entities according to the various purposes required. For this reason, free software plays a fundamental role in land administration due to its versatility in being studied, modified, distributed, and improved. It is for this reason that this article exposes a vision of the concept of free software and provides a description of the tools that intervene and facilitate land administration from a flow that starts from the management of databases with the creation of a model that fits the process of cadastral formation in Colombian territory, then describes the customization of forms for field acquisition by mobile device using the model described above. It also details the automation processes for the edition and validation of the information from desktop tools. Finally, it concludes with free tools’ role in land administration and how they contribute to the definition of an agile and efficient multipurpose cadastre.

The need to understand the factors influencing financial indicators, especially those associated with business management, is essential for making sound decisions, maintaining good financial health, strategically planning, and effectively managing risks. Purpose: In this research, the impact of management system maturity on the revenue levels of various types of companies in the city of Barranquilla (Colombia) is analyzed. Methods: To do this, the maturity levels of 201 companies were measured for two consecutive years prior to the Covid-19 pandemic. During the research process, inferential statistics and Machine Learning tools were applied. This was done with the purpose of identifying which established financial indicators directly affect the maturity of organizations’ management systems, highlighting revenue and asset levels as those exhibiting behavior that, when more extensive, leads to better economic benefits. Finally, a Bayesian neural network classifier was used to establish the forecasting capacity of financial indicators and provide key information for the continuous improvement of companies. Results: As main findings, it was evidenced that out of the 20 items used to measure the maturity of management systems, 15 showed a statistically significant relationship with the annual income level of the companies. Furthermore, these 15 items achieved an 89.15% of well-classified companies in their income level. Additionally, during the variable selection process with the help of inferential statistics, it was established that the higher the level of implementation of the 15 identified items from the maturity instrument, the greater the economic income of the company.