A federated learning-based, adaptable, multifunctional control system for smart sensors to improve agricultural production

Abstract

An intelligent agricultural control system consisting of: a large number of distributed sensor units spatially arranged over an agricultural field, each sensor unit being configured to generate measurement data corresponding to at least soil moisture, temperature, humidity, plant condition and environmental parameters; a communication interface configured to transmit the data collected by the numerous sensor units via a wireless network; a processing unit that is operationally connected to the communication interface and includes at least one processor and one memory, wherein the processing unit is configured to receive the acquired data and perform decentralized learning using locally stored datasets corresponding to individual sensor units; a data aggregation unit configured to combine parameter updates from the decentralized learning process without transmitting raw sensor data; a productivity analysis unit configured to compare the collected data with stored reference data that correspond to historical agricultural conditions and crop-specific requirements; a sensor control validation unit configured to determine the operating states of the multitude of sensor units based on the outputs of the productivity analysis unit and the data aggregation unit; and An actuator control unit configured to generate control signals for agricultural equipment such as irrigation systems, fertilization systems, and environmental control mechanisms, with the actuator control unit changing the timing and intensity of operation based on validated control states to improve agricultural productivity.

Description

TECHNICAL FIELD

[0001] The present invention relates generally to precision agriculture systems and in particular to an intelligent, sensor-based agricultural control device and a system architecture for monitoring, analyzing, and adaptively controlling agricultural parameters. The invention specifically relates to a federated learning-based, adaptable, multifunctional control system for intelligent sensors to improve agricultural production. BACKGROUND OF THE INVENTION

[0002] Modern agricultural systems increasingly rely on distributed sensor technologies to monitor soil, plant, and environmental parameters. Conventional sensor-based agricultural systems are typically integrated into IoT infrastructures to centrally collect and analyze data. However, such systems suffer from high computational latency, inefficient adaptation to dynamic environmental conditions, and delayed response times due to centralized processing capacities.

[0003] Existing machine learning systems reach their limits when processing large, distributed datasets, especially when real-time capability and privacy-compliant learning are required. Furthermore, conventional sensor control architectures lack mechanisms for adaptively adjusting sensor and actuator functions based on historical and contextual agricultural data, leading to suboptimal productivity and increased resource consumption.

[0004] Therefore, there is a need for a device-based, machine-integrated intelligent sensor control system that enables decentralized learning, dynamic control modification, and efficient synchronization of sensor and actuator operations in agricultural environments.

[0005] The rapid development of precision agriculture has led to the widespread use of intelligent sensor technologies that monitor environmental conditions, soil properties, and plant health on farms of all sizes. Modern agricultural practices increasingly rely on networked sensor infrastructures, typically consisting of distributed sensor nodes integrated into IoT frameworks, cloud computing platforms, and data analytics engines. These systems collect real-time information on parameters such as soil moisture, temperature, humidity, nutrient content, wind speed, and plant growth conditions, enabling data-driven decision-making. The integration of such technologies has significantly improved monitoring capabilities and facilitated the automation of irrigation, fertilization, and pest control.Despite these advances, the operational efficiency and adaptability of existing smart farming systems remain limited by various technological constraints.

[0006] Conventional smart sensor systems in agriculture are primarily based on centralized data processing architectures. Sensor data collected by distributed nodes is transmitted to a central server or cloud platform for analysis and decision-making. While these architectures enable the aggregation and processing of large datasets, they result in significant system latency due to communication delays, data transmission overhead, and centralized computing bottlenecks. This latency impairs real-time capability, particularly in scenarios requiring immediate intervention, such as irrigation control or disease management. Furthermore, centralized systems are vulnerable to network outages and bandwidth limitations, which can interrupt continuous monitoring and lead to the loss of critical agricultural data.

[0007] Besides latency issues, centralized agricultural monitoring systems raise concerns about data privacy and security. Agricultural data, especially when collected from multiple farms or different regions, can contain sensitive information about crop yields, soil conditions, and cultivation methods. Transmitting such data to central servers increases the risk of unauthorized access, data breaches, and misuse. Therefore, there is a growing need for decentralized data processing approaches that ensure data privacy while maintaining high analytical accuracy. However, existing decentralized frameworks often lack efficient coordination mechanisms and do not offer consistent performance across heterogeneous sensor networks.

[0008] Machine learning and artificial intelligence are increasingly being used in agricultural systems to improve predictive analytics and automate decision-making. Various models, including support vector machines, random forests, neural networks, and deep learning architectures, are employed for tasks such as yield forecasting, crop classification, disease detection, and resource optimization. While these approaches have demonstrated higher accuracy in controlled environments, their practical application in real-world agricultural systems faces several challenges. One of the biggest limitations is the need for large amounts of annotated training data, which is not always readily available or obtainable in agriculture.Furthermore, machine learning models trained on centralized datasets often cannot be effectively transferred to different geographical regions, soil types, and climate conditions, which affects their reliability and robustness.

[0009] Another significant drawback of existing machine learning systems in agriculture is the high computational complexity involved in processing large volumes of sensor data. Agricultural environments continuously generate heterogeneous data streams containing temporal, spatial, and contextual information. Real-time processing of this data requires substantial computing resources, which are often unavailable on edge devices or resource-constrained sensor nodes. Consequently, many systems resort to cloud-based processing, which in turn introduces latency and reliance on network connectivity. Furthermore, the lack of adaptive learning mechanisms in conventional models limits their ability to dynamically respond to changing environmental conditions, leading to suboptimal control decisions.

[0010] Wireless sensor networks (WSNs) are widely used in agricultural monitoring systems due to their ability to collect and communicate data in a distributed manner. These networks consist of multiple sensor nodes distributed across farmland, each collecting data and transmitting it to a central coordinator. While WSNs offer advantages such as scalability and flexibility, they are often limited by restricted power resources, communication range, and network reliability. Battery-powered sensor nodes can experience power outages, which can lead to network instability and data loss. Furthermore, communication interference, node failures, and environmental factors can compromise the reliability of data transmission, thus reducing the overall performance of the monitoring system.

[0011] Existing sensor control mechanisms in agricultural systems are typically static or rule-based. They use predefined thresholds and conditions to trigger actions. For example, irrigation systems can be activated when soil moisture falls below a certain threshold. While such approaches are easy to implement, they cannot adapt to complex and dynamic agricultural conditions. Factors such as crop species, growth stage, weather fluctuations, and soil heterogeneity require more sophisticated control strategies that can consider multiple variables simultaneously. Static control mechanisms often lead to inefficient resource use, such as over-irrigation or under-fertilization, which can negatively impact crop yield and environmental sustainability.

[0012] Recent advances are exploring the use of optimization techniques such as particle swarm optimization and reinforcement learning to improve sensor placement, coverage, and control strategies in agricultural systems. While these methods can enhance performance under certain conditions, they often require extensive parameter adjustments and may not scale effectively in large and heterogeneous environments. Furthermore, optimization-based approaches typically rely on predefined models and assumptions that may not accurately reflect the real-world variability in agriculture. Consequently, their effectiveness is limited in dynamic and uncertain environments with frequently changing conditions.

[0013] Another challenge with existing agricultural monitoring systems is the lack of synchronization between sensors, analysis, and control. In many systems, data acquisition, processing, and control occur independently or are only inadequately coordinated, leading to inconsistencies and delays. For example, delayed data analysis can result in outdated information being used for control decisions, thus reducing the effectiveness of control measures. This lack of synchronization also contributes to delayed responses, which significantly impairs the responsiveness and efficiency of agricultural systems.

[0014] Furthermore, current systems often struggle to manage the variability and heterogeneity of agricultural data. Environmental conditions such as temperature, humidity, and soil composition can vary considerably depending on location and time frame. Existing models and control strategies may not adequately account for this variability, leading to inaccurate predictions and suboptimal decisions. The inability to effectively integrate historical data with real-time observations further limits these systems' ability to learn from past experiences and improve their future performance.

[0015] Furthermore, many existing solutions lack mechanisms for the continuous validation and adaptation of sensor functions. After installation, sensor configurations and control strategies are rarely updated dynamically, even under changing environmental conditions. This lack of adaptability impairs the system's ability to respond to new challenges such as emerging plant diseases, climate fluctuations, or altered soil properties. Consequently, the overall productivity and efficiency of agricultural operations are negatively impacted.

[0016] Therefore, there remains a significant need for an advanced agricultural control system that can overcome the limitations of existing solutions through decentralized, adaptive, and efficient management of sensor functions. Such a system should integrate real-time and historical data, enable collaborative learning between distributed sensor nodes, and dynamically adapt control strategies to optimize agricultural productivity. Developing a device-based architecture with federated learning and adaptive control mechanisms represents a promising approach to addressing these challenges and advancing the state of the art in precision agriculture. SUMMARY OF THE INVENTION

[0017] The invention relates in particular to a federated learning-based, modifiable multifunctional control system for intelligent sensors for improving agricultural production.

[0018] The present invention relates to a multifunctional intelligent agricultural control device, comprising a sensor arrangement, a federated learning processing unit, a control validation circuit and an adaptive actuation interface, which together are configured to enable continuous acquisition, analysis and control of agricultural parameters.

[0019] The invention presents an adaptable, multifunctional control mechanism in which the sensor functions are dynamically modified based on the results of federated learning from current and historical data. The system performs productivity analyses and validations of the sensor control sequentially and in parallel, thereby minimizing response delay and increasing adaptability.

[0020] The device enables decentralized learning across distributed sensor nodes, thus ensuring data protection-compliant data processing while maintaining high decision accuracy. The coordinated interaction of sensor units, learning modules, and actuator components leads to optimized agricultural productivity, reduced analysis time, and improved operational efficiency.

[0021] The main objective of the present invention is to provide an intelligent agricultural control device for monitoring, analyzing, and controlling environmental and plant-related parameters using an integrated multi-sensor architecture. The device enables real-time data acquisition and adaptive control to increase agricultural productivity. A further objective of the invention is to provide a decentralized data processing mechanism based on federated learning. In this mechanism, multiple distributed sensor nodes contribute to model training without transmitting raw data. This improves data privacy, reduces communication overhead, and increases computing efficiency in agricultural environments.

[0022] A further objective of the present invention is to provide an adaptable, multifunctional control system that can dynamically modify sensor functions, including acquisition intervals, control thresholds, and actuator responses. This is achieved by continuously validating current sensor data against previously stored optimal data for various crops and environmental conditions. A further objective of the invention is the synchronous coordination of acquisition, productivity analysis, and actuator processes, thereby minimizing delays, reducing response time, and ensuring the timely execution of control measures such as irrigation, fertilization, and environmental adjustments.

[0023] A further objective of the invention is to provide a device with a productivity analysis subsystem for evaluating fluctuations in agricultural parameters and identifying deviations that impair plant growth. This analysis supports intelligent decisions for optimizing resource utilization and maximizing yield. Furthermore, the invention incorporates a sensor control validation mechanism that determines the operating status and effectiveness of individual sensors, thus enabling the detection and correction of unsynchronized or faulty sensor nodes in the agricultural network.

[0024] A further objective of the present invention is to improve the adaptability of agricultural systems through continuous learning from historical and real-time data. The system adapts its control strategies to different climatic conditions, soil properties, and crop-specific requirements. Another objective is to reduce computational complexity and analysis time when processing large volumes of agricultural data by distributing the learning tasks across multiple sensor nodes and employing efficient aggregation techniques.

[0025] A further objective of the invention is to provide a robust and scalable device suitable for use in a wide variety of agricultural environments, from small farms to large fields. The system should deliver consistent performance despite fluctuations in sensor density, environmental conditions, and operational constraints. Another objective is to improve the reliability and efficiency of wireless sensor networks in agriculture through optimal coordination, reduced energy consumption, and increased data integrity.

[0026] A further objective of the invention is to provide a mechanism for minimizing errors in sensor control and actuation through continuous validation and updating of control decisions using federated learning, thereby avoiding failures and inefficiencies in agricultural operations. Furthermore, the invention aims to facilitate the intelligent management of agricultural resources by integrating multifunctional sensors with adaptive control strategies, thus enabling precise and efficient cultivation methods.

[0027] Finally, an objective of the present invention is to provide an advanced, intelligent sensor-based agricultural device that achieves improved performance indicators, including an increased analysis rate, an improved control rate, higher adaptability, reduced analysis time and minimized actuation delay, thereby contributing to sustainable and efficient agricultural production systems. BRIEF DESCRIPTION OF THE IMAGE

[0028] These and other features, aspects and advantages of the present invention will be better understood if the following detailed description is read with reference to the accompanying drawing, in which the same symbols represent the same parts: Fig. Figure 1 shows a block diagram of an intelligent agricultural control system.

[0029] Furthermore, those skilled in the art will recognize that the elements in the drawing are simplified and not necessarily drawn to scale. For example, the flowcharts illustrate the process by highlighting the main steps to facilitate understanding of the present disclosure. With regard to the construction of the device, one or more components may be represented in the drawing by conventional symbols. The drawing may show only those specific details relevant to understanding the embodiments of the present disclosure, so as not to clutter the drawing with details that are already apparent to those skilled in the art from the description contained herein. Detailed description of the invention

[0030] To facilitate understanding of the principles of the invention, reference is made below to the embodiment shown in the drawing, which is described using specific terms. It is understood, however, that this does not limit the scope of protection of the invention. Rather, modifications and further developments of the depicted system, as well as further applications of the inventive principles shown therein, are conceivable, insofar as they would normally occur to a person skilled in the art in the field of the invention.

[0031] It will be clear to those skilled in the art that the foregoing general description and the following detailed description are exemplary and explanatory of the invention and are not to be understood as a limitation of it.

[0032] References to “an aspect”, “another aspect”, or similar phrases in this description mean that a particular feature, structure, or property described in connection with the embodiment is included in at least one embodiment of the present disclosure. Therefore, phrases such as “in one embodiment”, “in another embodiment”, and similar expressions in this description may, but do not necessarily, all refer to the same embodiment.

[0033] The terms "includes," "comprehensive," or similar formulations denote non-exclusive inclusion. Likewise, the mention of one or more devices, subsystems, elements, structures, or components by "includes..." without further limitations does not exclude the existence of other devices, subsystems, elements, structures, or components.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as is generally known to those skilled in the art in the field to which this invention belongs. The system and the examples contained herein serve only for illustration and are not to be construed as a limitation.

[0035] Embodiments of the present disclosure are described in detail below with reference to the attached drawing.

[0036] Fig.Figure 1 shows a block diagram of an intelligent agricultural control system. The system 100 comprises: several distributed sensor units (102) spatially arranged across an agricultural field, each recording measurement data on soil moisture, temperature, humidity, plant condition, and environmental parameters; a communication interface (104) for transmitting the measurement data from the sensor units via a wireless network; a processing unit (106) connected to the communication interface, comprising at least one processor and memory. The processing unit receives the measurement data and performs decentralized learning based on locally stored datasets from the individual sensor units; and a data aggregation unit (108) for combining the parameter updates derived from decentralized learning without transmitting the raw data.a productivity analysis unit (110) for comparing the measurement data with stored reference data on historical agricultural conditions and plant-specific requirements; a sensor control validation unit (112) for determining the operating states of the sensor units based on the outputs of the productivity analysis unit and the data aggregation unit; and an actuation control unit (114) configured to generate control signals for agricultural equipment such as irrigation systems, fertilization systems, and environmental control mechanisms, wherein the actuation control unit modifies the timing and intensity of operation based on validated control states to improve agricultural productivity.

[0037] In one embodiment, each of the numerous sensor units (102) comprises a multi-layered sensor structure with a subsurface probe part positioned at different depths within the soil layers and a surface-mounted sensor part. The subsurface probe part is configured to detect moisture gradients and conductivity changes, and the surface-mounted sensor part is configured to detect atmospheric parameters. This enables multidimensional data acquisition along vertical and horizontal profiles of the agricultural field.

[0038] In one embodiment, the processing unit (106) is configured to iteratively update a global learning representation by aggregating locally calculated parameter updates received from individual sensor units at predefined time intervals. This aggregation eliminates the need to transmit the raw measurement data and instead uses parameter-level synchronization to protect data privacy and reduce communication overhead.

[0039] In one embodiment, the productivity analysis unit (110) is configured to normalize the acquired data by correlating the current environmental conditions with stored optimal conditions corresponding to specific plant species, growth stages, and seasonal variations. The normalization also takes into account deviations in the maximum and minimum thresholds of agricultural production observed over multiple acquisition intervals.

[0040] In one embodiment, the sensor control validation unit (112) is configured to evaluate the synchronization of the sensor operations by identifying discrepancies between the expected sensor behavior derived from historical data and the current operating outputs, and selectively activating, deactivating, or modifying individual sensor units depending on the detected inconsistencies.

[0041] In one embodiment, the actuation control unit (114) is configured to dynamically adjust the actuation intervals by decreasing or increasing the frequency of the control signals based on the stability or variability of the detected environmental parameters. This minimizes unnecessary actuation operations and conserves energy resources.

[0042] In one embodiment, the processing unit (106) is further configured to divide the acquired data into sequential and successive instances corresponding to time-based intervals. A first instance is used for an initial productivity assessment, and a subsequent instance serves to validate control decisions. This enables a stepwise evaluation of agricultural conditions.

[0043] In one embodiment, the data aggregation unit (108) is configured to detect unsynchronized sensor behavior by comparing parameter updates across multiple sensor units and to trigger corrective actions to restore synchronization between the numerous sensor units.

[0044] In one embodiment, the productivity analysis unit (110) is configured to determine changed control requirements by identifying deviations between currently acquired data and optimal stored data. These deviations include agricultural constraints such as soil dryness, excessive moisture, indicators of plant infestation, and climatic anomalies.

[0045] In one embodiment, the actuator control unit (114) is configured to generate differentiated control signals for various agricultural operations such as irrigation, fertilization and harvesting assistance, with each control signal being adapted based on the crop-specific requirements and the environmental conditions determined by the productivity analysis unit.

[0046] The present invention relates to an intelligent agricultural control system implemented as an integrated device structure with distributed sensor units, a processing unit with memory, a data aggregation unit, a productivity analysis unit, a sensor control validation unit, and an actuator control unit. All components are interconnected via a communication interface. System operation is based on a sequential and iterative process that enables continuous data acquisition, decentralized learning, validation, and adaptive actuator control to increase agricultural productivity.

[0047] In operation, sensor units distributed across an agricultural field continuously record environmental and plant-related parameters, including soil moisture at various depths, air temperature, humidity, wind conditions, and indicators of plant health. Each sensor unit generates timestamped measurement data corresponding to its location and installation depth. The measurement data is temporarily stored locally in each sensor unit and periodically transmitted to the processing unit via the communication interface. Communication takes place using energy-efficient radio protocols, and the transmission intervals are dynamically adjusted to environmental changes to avoid unnecessary data exchange.

[0048] After receiving the measurement data, the processing unit performs a decentralized learning process in which each sensor unit contributes locally calculated parameter updates instead of transmitting raw data. To this end, the processing unit distributes a common initial learning representation to the sensor units, which then calculate local updates based on their respective measurement datasets and historical data stored in their local memory. These local updates are sent back to the processing unit, where the data aggregation unit combines them to generate an updated global representation. The aggregation process is iterative and is performed at predefined time intervals. This enables continuous refinement of the learning representation without exposing the raw sensor data.

[0049] The productivity analysis unit receives both the measurement data and the aggregated learning results and performs a correlation-based evaluation using stored reference datasets that correspond to optimal growing conditions. These reference datasets contain crop-specific requirements such as preferred soil moisture ranges, temperature thresholds, nutrient conditions, and growth stages. The productivity analysis unit normalizes the measurement data by comparing current observations with historical optimal ranges and calculating deviations that indicate potential inefficiencies or unfavorable conditions. The normalization process also considers fluctuations in maximum and minimum production levels from previous measurement intervals, thus enabling a contextual assessment of current conditions.

[0050] The system divides the analysis into successive instances. A first instance performs a primary productivity assessment based on the currently collected data, while a subsequent instance performs validation using historical and aggregated learning results. This staggered evaluation ensures that temporary fluctuations in sensor readings do not lead to premature or erroneous control measures. The system guarantees temporal continuity by comparing successive measurement intervals and identifying patterns of change, thus enabling more reliable decision-making.

[0051] The sensor control validation unit receives output from the productivity analysis unit and evaluates the operating status of each sensor unit. This evaluation involves comparing the expected sensor behavior, derived from historical data and aggregated learning results, with the actual measured values. If deviations such as abnormal readings, inconsistent data patterns, or communication delays are detected, the sensor control validation unit determines the need for corrective action. This may include recalibrating sensor units, adjusting measurement intervals, or temporarily disabling faulty sensors. The validation process ensures that only reliable and synchronized sensor data is used for subsequent control decisions.

[0052] Based on validated sensor data and productivity analyses, the actuator control system generates control signals for agricultural equipment. It determines the type, timing, and intensity of necessary control measures to optimize growing conditions. For example, the system generates irrigation signals when the soil moisture level for a specific crop falls below optimal thresholds. These parameters correspond to the required amount and duration of water. If nutrient deficiencies or environmental stress are detected, appropriate fertilization or environmental protection measures are initiated. Furthermore, the actuator control system prioritizes control measures according to the severity of the detected conditions, thus ensuring the prompt resolution of critical problems.

[0053] A key aspect of the system is the dynamic adjustment of measurement and response intervals. The processing unit continuously assesses the stability of the environment by analyzing changes in measurement data over time. Under stable conditions, the system lengthens the interval between measurement and response processes to conserve energy and computing resources. Conversely, under rapidly changing conditions, the system shortens these intervals to enable more frequent monitoring and timely intervention. This adaptive interval adjustment is essential to ensure an optimal balance between system efficiency and responsiveness.

[0054] The system also includes a delay minimization mechanism that analyzes the time difference between detection events and the corresponding actuator responses. By comparing successive detection intervals and actuator timestamps, the processing unit identifies delays in control execution and adjusts the scheduling parameters to reduce these delays. This mechanism ensures that control actions are closely timed with the detection of relevant states, thus improving the effectiveness of interventions.

[0055] The data aggregation unit also performs a cross-node comparison of parameter updates to detect unsynchronized sensor behavior. If inconsistencies are detected between the sensor units, the system makes corrections to restore synchronization and thus ensure smooth operation across the entire agricultural field. This synchronization is particularly important in large-scale installations where the sensor nodes may be exposed to varying environmental conditions and communication delays.

[0056] The processing unit additionally calculates key performance indicators (KPIs) for each sensor unit based on aggregated learning and validation results. These KPIs determine whether the individual sensor units are operating within acceptable limits. Sensor units with low KPIs can be recalibrated or replaced to ensure the overall reliability of the system.

[0057] The system also supports the continuous updating of stored reference datasets based on validated results from control measures. If a particular control measure leads to improved agricultural conditions, the corresponding data are incorporated into the reference dataset, allowing the system to learn from past experiences. This adaptive learning capability improves the system's ability to respond more precisely to future conditions.

[0058] The system can also manage multiple agricultural plots, each with its own sensors and local data. The processing unit aggregates the learning results from these plots to improve the global learning representation while preserving local adaptability. This ability to manage multiple plots enables scalability and facilitates deployment in diverse agricultural environments.

[0059] The system according to the invention enables the coordinated, continuous, and adaptive execution of sensor, learning, validation, and actuator processes. By integrating decentralized learning with dynamic control adjustment and synchronized operation of the sensor units, the system achieves higher analysis accuracy, lower computational latency, improved adaptability, and minimized actuator delays. This significantly improves agricultural productivity and resource utilization.

[0060] The present invention discloses an intelligent agricultural control device configured as a machine-integrated structure within an agricultural field and consisting of a plurality of interconnected hardware and computing components.

[0061] The device comprises a multimodal sensor array distributed across an agricultural area. This array consists of optical sensors, electrochemical sensors, dielectric soil moisture sensors, airflow sensors, temperature sensors, humidity sensors, and sensors for monitoring plant health. The sensors are mounted on distributed nodes or embedded probes positioned at varying depths and locations within the field to capture environmental and plant-related parameters in real time.

[0062] Each sensor node is operationally connected to a wireless communication interface that enables the transmission of the acquired data to a central or on-device processing unit. The processing unit comprises a federated learning engine implemented in hardware circuitry and memory architecture, configured to aggregate locally acquired data while maintaining data isolation at the node level.

[0063] The device also includes a productivity analysis subsystem that compares the collected data with stored optimal growing conditions for various crops and environmental factors. This subsystem normalizes and correlates the current sensor data with historical datasets, thus enabling the identification of deviations and inefficiencies in growing conditions.

[0064] A sensor control validation unit is structurally integrated into the processing architecture and configured to determine the validity of current sensor operations based on federated learning results. The validation unit calculates control parameters that determine whether sensor operations need to be adjusted, continued, or aborted at specific time intervals.

[0065] The invention further comprises an adaptive actuator interface that is operationally connected to irrigation systems, fertilization mechanisms, and environmental control units. Based on validated control decisions, the actuator interface generates control signals for carrying out agricultural processes such as water dosing, nutrient application, and environmental adjustments.

[0066] The device is configured to perform a two-stage operating process, namely a first stage with data acquisition and productivity analysis, and a second stage with control validation and actuator operation.

[0067] The system also includes a modification engine that dynamically adjusts the sensors' operating parameters, including measurement intervals, activation times, and operating thresholds. The modification engine analyzes deviations between actual and target conditions and corrects the sensor control strategies. This mechanism reduces errors due to environmental influences and ensures consistent agricultural performance.

[0068] The federated learning process also identifies unsynchronized sensor behavior and operational anomalies, enabling corrective actions to be applied across distributed nodes. This ensures the synchronization of multiple sensors and smooth operation throughout the entire agricultural field.

[0069] The system is further configured to calculate probabilistic control metrics that determine the probability of successful sensor operation and successful actuator activation. These metrics are used to dynamically allocate resources and prioritize critical agricultural interventions.

[0070] In one embodiment, the device supports time-variable detection intervals, whereby the frequency of detection and actuation is adaptively adjusted based on the stability or variability of the environment.

[0071] Furthermore, the device features a delay minimization mechanism that analyzes successive sensing intervals to detect and suppress delays in the actuators. This mechanism ensures the prompt execution of control actions, thereby improving responsiveness and agricultural efficiency.

[0072] The system is particularly effective in managing various crop species, such as chickpeas and winter wheat, by adapting the sensor functions to crop-specific requirements, including temperature, soil type and moisture content.

[0073] Through the integrated interaction of sensors, federated learning, validation and actuators, the device can achieve improved performance indicators, including a higher analysis rate, an improved control rate, higher adaptability, reduced analysis time and minimized actuator delay, thereby significantly increasing agricultural productivity.

[0074] The present invention relates to the field of precision agriculture and intelligent control systems, in particular an intelligent, sensor-based agricultural control system. This system is implemented as a device structure with distributed sensors, decentralized data processing, and adaptive actuators. The invention specifically relates to a system that utilizes multiple sensor units in combination with a processing unit. This processing unit is configured for data-based aggregation using federated learning, productivity analysis, sensor control validation, and dynamic actuator control. The described system enables real-time monitoring and adaptive management of agricultural parameters such as soil quality, environmental factors, and plant health, thereby improving the efficiency, responsiveness, and productivity of farms.

[0075] The drawing and the preceding description illustrate embodiments. Those skilled in the art will recognize that one or more of the described elements can be combined to form a single functional element. Alternatively, certain elements can be divided into several functional elements. Elements of one embodiment can be added to another. For example, the process flows described here can be modified and are not limited to the manner described herein. Furthermore, the actions of a flowchart need not be performed in the sequence shown; nor do all actions necessarily need to be carried out. Actions that do not depend on other actions can be performed in parallel with the other actions. The scope of protection of the embodiments is in no way limited by these specific examples. Numerous variations, whether explicitly stated in the description or not, such as...Differences in structure, dimensions, and materials are possible. The scope of protection of the embodiments is at least as comprehensive as described by the following claims.

[0076] The advantages, other benefits, and problem solutions have been described above with reference to specific embodiments. However, the advantages, benefits, problem solutions, and any components that can effect or enhance an advantage, benefit, or solution are not to be construed as critical, necessary, or essential features or components of the claims. REFERENCES 100 An intelligent control system for agriculture is presented. 102 multiple distributed sensor units 104 Communication interface 106 processing units 108 Data aggregation unit 110 Productivity Analysis Unit 112 Sensor control validation unit 114 Actuation control unit

Claims

An intelligent agricultural control system consisting of: a multitude of distributed sensor units spatially arranged across an agricultural field, each sensor unit configured to generate measurement data corresponding to at least soil moisture, temperature, humidity, plant condition, and environmental parameters; a communication interface configured to transmit the data acquired by the numerous sensor units via a wireless network; a processing unit operationally connected to the communication interface and comprising at least one processor and memory, the processing unit being configured to receive the acquired data and perform decentralized learning using locally stored datasets corresponding to individual sensor units;a data aggregation unit configured to combine parameter updates from the decentralized learning process without transmitting raw sensor data; a productivity analysis unit configured to compare the acquired data with stored reference data corresponding to historical agricultural conditions and crop-specific requirements; a sensor control validation unit configured to determine the operating states of the multitude of sensor units based on the outputs of the productivity analysis unit and the data aggregation unit;and an actuator control unit configured to generate control signals for agricultural equipment such as irrigation systems, fertilization systems, and environmental control mechanisms, wherein the actuator control unit changes the timing and intensity of operation based on validated control states to improve agricultural productivity. System according to claim 1, wherein each of the numerous sensor units comprises a multi-layered sensor structure comprising a subsurface probe section positioned at several depths within the soil layers and a surface-mounted sensor section, wherein the subsurface probe section is configured to detect moisture gradients and conductivity changes and the surface-mounted sensor section is configured to detect atmospheric parameters. System according to claim 1, wherein the processing unit is configured to iteratively update a global learning representation by aggregating locally computed parameter updates received from individual sensor units at predefined time intervals, wherein the aggregation excludes the transmission of raw measurement data and instead uses parameter-level synchronization to protect data privacy and reduce communication overhead. System according to claim 1, wherein the productivity analysis unit is configured to perform the normalization of the recorded data by correlating the current environmental conditions with stored optimal conditions for specific plant species, growth stages and seasonal variations, wherein the normalization further takes into account deviations in the maximum and minimum agricultural production thresholds observed over several recording intervals. System according to claim 1, wherein the sensor control validation unit is configured to evaluate the synchronization of the sensor operations by detecting discrepancies between the expected sensor behavior derived from historical data and the current operating outputs, and wherein the sensor control validation unit selectively activates, deactivates, or modifies individual sensor units based on detected inconsistencies. System according to claim 1, wherein the actuation control unit is configured to dynamically adjust the actuation intervals by decreasing or increasing the frequency of the control signals based on the stability or variability of the detected environmental parameters, thereby minimizing unnecessary actuation operations and conserving energy resources. System according to claim 1, wherein the processing unit is further configured to divide the acquired data into sequential and successive instances corresponding to time-based intervals, wherein a first instance is used for the initial productivity assessment and a subsequent instance is used for validating control decisions. System according to claim 1, wherein the data aggregation unit is configured to detect unsynchronized sensor behavior by comparing parameter updates across multiple sensor units, and wherein the system triggers corrective actions to restore synchronization between the numerous sensor units. System according to claim 1, wherein the productivity analysis unit is configured to determine changed control requirements by identifying deviations between currently recorded data and optimal stored data, wherein the deviations correspond to agricultural constraints such as soil dryness, excessive moisture, indicators of plant infestation and climatic anomalies. System according to claim 1, wherein the actuation control unit is configured to generate differentiated control signals for various agricultural operations such as irrigation, fertilization and harvesting assistance, wherein each control signal is adapted on the basis of crop-specific requirements and environmental conditions determined by the productivity analysis unit.

Images