Paulo Estevão Cruvinel

Accurate pressure prediction is essential for predictive control in variable-rate agricultural sprayers, but changes in pump command, valve opening, and intermittent nozzle activation modify the hydraulic resistance and reduce the validity of fixed AutoRegressive with eXogenous input (ARX) models. This work evaluates hydraulic-regime-dependent ARX pressure models for a Smart Weed Sprayer and proposes a fixed-denominator adaptive-numerator structure for control-oriented prediction. An experimental dataset with 84 tests was collected under different pump duty cycles, valve openings, and active-nozzle configurations. The hydraulic regime was characterized by the equivalent resistance index ρ = log10(KTeq). Local ARX models showed good accuracy within their identification regimes, but poor portability across operating conditions, with the mean Root Mean Squared Error (RMSE) increasing from 0.4492 in diagonal validation cases to 2.2377 in off-diagonal transfers. The analysis showed that the denominator polynomial was not robustly related to ρ, while the numerator coefficients were more sensitive to hydraulic changes but not sufficiently explained by direct scheduling alone. Therefore, the proposed model keeps a nominal denominator A0(z⁻¹) fixed and adapts the numerator B(z⁻¹, k) online using Recursive Least Squares (RLS). Compared with fixed, scheduled, lookup-table, and oracle benchmarks, the RLS model reduced the free-run mean RMSE from 6.6580 to 1.4762, corresponding to a 77.8% reduction. These results show that online error-driven adaptation provides more robust pressure prediction than fixed or directly scheduled ARX models across changing hydraulic regimes.

Precise flow rate control in agricultural sprayers is critical for variable-rate application, but is challenged by nonlinear dynamics, dead zones, and model variations due to operational changes and nozzle switching. This work aims to develop an efficient control strategy capable of handling these challenges while ensuring compliance with operational constraints. To this end, a generalized predictive controller (GPC) integrated with a dynamic output-linearization scheme is proposed. The dynamic linearization plays a key role to enable adaptive control under practical operating conditions, without requiring a detailed characterization of the hydraulic system. To support this framework, a simplified Kalman filter is employed, to estimate, in real time, the total equivalent fluidic resistance of the system. Experimental validation was conducted using an agricultural sprayer development system, demonstrating superior performance compared to traditional methods, with improved accuracy and adaptability in the presence of disturbances. The proposed output linearizing fuzzy GPC controller (L-FGPC) effectively handles nonlinearities and operational constraints, offering a practical solution for advanced agricultural spraying systems.

This study presents an intelligent method for evaluating the risk of Asian rust (Phakopsora pachyrhizi) based on its development stage in soybean crops (Glycine max (L.) Merrill). It has been designed using smart computer systems supported by image processing, environmental sensor data, and an embedded model for evaluating favorable conditions for disease progression within crop areas. The approach also includes the use of machine learning techniques and a Markov chain algorithm for data fusion, aimed at supporting decision-making in agricultural management. Rules derived from time-series data are employed to enable scenario prediction for risk evaluation related to disease development. Measured data are stored in a customized system designed to support virtual monitoring, facilitating the evaluation of disease severity stages by farmers and enabling timely management actions.

This paper presents a method for dynamic pattern recognition and classification of one dangerous caterpillar species to allow for its control in maize crops. The use of dynamic pattern recognition supports the identification of patterns in digital image data that change over time. In fact, identifying fall armyworms (Spodoptera frugiperda) is critical in maize production, i.e., in all of its growth stages. For such pest control, traditional agricultural practices are still dependent on human visual effort, resulting in significant losses and negative impacts on maize production, food security, and the economy. Such a developed method is based on the integration of digital image processing, multivariate statistics, and machine learning techniques. We used a supervised machine learning algorithm that classifies data by finding an optimal hyperplane that maximizes the distance between each class of caterpillar with different lengths in N-dimensional spaces. Results show the method’s efficiency, effectiveness, and suitability to support decision making for this customized control context.

Among the different stages involved in agricultural production processes, agricultural spraying used for phytosanitary control encompasses a significant portion of the total production cost, and its incorrect execution can increase human and biological risks. Therefore, research is necessary to ensure the proper use of agricultural pesticides, ensuring the quality of their application. Among the quality indicators provided by the droplet spectrum, the volumetric median diameter (VMD) is one of the most commonly used. However, its analysis is based on post-application tests, thus preventing action during pesticide spraying. This research aims to implement an artificial neural network (ANN) architecture in an agricultural sprayer with direct injection technology, in order to perform real-time classification of droplet sizes based on measurements of flow rate, pressure, and fluidic resistance from the instruments on the sprayer, as well as historical operational characteristics of the machinery. Therefore, training and testing are performed based on experimental results and tables from manufacturers’ technical manuals of the nozzles. The obtained results allow us to characterize that the utilized methodology provides accurate estimations of droplet sizes, thereby enhancing the quality of the application.