Controlled-environment agriculture may be entering a new phase in which greenhouses do more than measure their surroundings: they continuously interpret noisy data, anticipate changing conditions and coordinate several pieces of equipment at once. A new study describes an intelligent greenhouse control system that combines distributed sensors, Kalman filtering, LabVIEW software and a genetic-algorithm-optimized fuzzy controller. The architecture is designed to regulate temperature, relative humidity, carbon dioxide and light as a coupled system rather than as four independent problems. That distinction matters because greenhouse variables interact constantly. Heating can reduce relative humidity, ventilation can remove carbon dioxide, lighting can add heat, and attempts to correct one condition can unintentionally destabilize another. The researchers present their platform as a practical route toward more reliable climate control and lower energy use without requiring the highly complex mathematical models associated with some advanced predictive-control systems.
The system is built around four principal components: a host computer, an edge processing unit, a distributed sensor network and an actuator output module. Six sensor nodes are deployed across the greenhouse, with one node assigned to every 50 square metres according to the layout standards described by the researchers. Each node measures air temperature, air humidity, carbon dioxide concentration and photosynthetic photon flux density, or PPFD—the amount of photosynthetically active light reaching a surface. The measurements are transmitted at a frequency of one sample per second through Wi-Fi using the 802.11n standard. An edge acquisition controller communicates with the host computer and also supports local data handling, while the system stores measurements in a MySQL 8.0 database. The hardware is intended to remain operational across temperatures from −20 to 60 degrees Celsius and relative humidity from 0 to 100 percent, conditions that reflect the demanding environment in which greenhouse electronics must function.
Raw sensor readings are rarely clean enough to drive a control system directly. Electronic noise, calibration drift, communication interruptions and local disturbances can produce values that jump sharply or disagree with measurements from nearby sensors. In a greenhouse, even a genuine reading may represent only a small microclimate rather than conditions across the entire growing area. The researchers therefore place Kalman filtering between data acquisition and control. A Kalman filter is a mathematical estimation method that combines a system’s previous state with new observations. Instead of accepting every measurement at face value, it weighs the expected reliability of the sensor and the uncertainty of the underlying process. In simplified form, the filter predicts the next environmental state using a state-transition model and then corrects that prediction using the incoming observation. The balance is governed by process-noise and observation-noise covariance matrices, commonly represented as Q and R. When sensor readings become especially noisy, the filter can rely more heavily on its prediction; when observations are considered reliable, the new data exert greater influence.
This preprocessing stage is crucial because control decisions can amplify measurement errors. If a noisy humidity sensor suddenly reports an implausible increase, a conventional controller may activate ventilation or heating, potentially wasting energy and creating a second environmental disturbance. A filtered estimate provides a more stable representation of the greenhouse state before the controller decides how strongly to operate an actuator. The system also uses multiple sensing locations rather than depending on a single instrument. Data fusion allows the controller to compare measurements spatially and identify broader trends, while filtering reduces the effect of short-lived fluctuations. Together, these techniques aim to make the greenhouse’s digital representation closer to its physical condition. The approach reflects a broader shift in agricultural automation, in which sensors are not treated merely as instruments that report numbers but as uncertain sources of evidence that must be combined, evaluated and continuously interpreted.
At the centre of the control architecture is a fuzzy controller whose parameters and rule structure are optimized with a genetic algorithm. Fuzzy logic is well suited to environments that are difficult to describe with precise equations. Instead of requiring an exact mathematical relationship between every input and output, it uses linguistic categories such as “low,” “moderate” and “high,” together with rules that approximate the reasoning of an experienced operator. In this system, the controller considers errors in temperature, humidity, carbon dioxide and PPFD, then generates coordinated commands for devices including heaters, fans, carbon-dioxide generators and supplementary lights. A temperature error, for example, does not automatically trigger a fixed response in isolation. Its effect can be interpreted alongside humidity and light conditions, reducing the likelihood that separate controllers will issue contradictory commands. This is particularly important because greenhouse climate is a multi-input, multi-output system: several actuators influence several variables simultaneously, and the same actuator may affect more than one target.
Manually designing fuzzy membership functions and control rules, however, can be laborious and highly dependent on expert judgement. The genetic algorithm is intended to reduce that dependence by searching for better controller settings. Genetic algorithms imitate evolutionary processes: candidate solutions are encoded, evaluated according to a fitness function, recombined through crossover and altered through mutation. In the greenhouse application, a candidate may represent the shapes of fuzzy membership functions, the weighting of rules or other controller parameters. The fitness function can balance environmental tracking against energy consumption, comparing actual system energy use with the energy required by a conventional proportional-integral-derivative, or PID, controller. Through repeated generations, the algorithm retains configurations that produce more desirable combinations of stability, accuracy and efficiency. The result is not a controller that simply reacts faster, but one that is computationally tuned to the nonlinear and coupled behaviour of the greenhouse.
The use of LabVIEW gives the platform a real-time supervisory layer. LabVIEW, or Laboratory Virtual Instrument Engineering Workbench, is a graphical programming environment widely used for measurement, instrumentation and industrial control. In the reported architecture, it links sensor acquisition, filtering, database storage, fuzzy inference and actuator commands in a single operating framework. The distributed arrangement also separates functions: sensor nodes collect local conditions, the edge unit handles networked acquisition and the host computer performs higher-level processing and visualization. Such modularity could make the system easier to expand, since additional sensing points or actuators can be integrated without redesigning every component. It may also help operators diagnose failures by preserving time-stamped records of environmental conditions and control actions. For commercial growers, this type of traceability is valuable because it can connect climate events with crop performance, equipment operation and energy demand over long periods.
The researchers describe a long-term greenhouse deployment experiment to evaluate the system’s reliability and practical applicability. According to the study, the integrated approach improves environmental regulation while supporting energy-efficient operation. The significance of the claim lies in the combination of methods rather than in any single component. Multi-sensor fusion addresses spatial variability; Kalman filtering addresses uncertainty and noise; fuzzy logic handles nonlinear interactions; and genetic optimization reduces the need to tune the controller entirely by hand. The architecture also aims to avoid the computational burden of some model-predictive or reinforcement-learning systems, which may require detailed plant and climate models, substantial training data or powerful hardware. At the same time, the system is not a simple plug-and-play device. Its performance depends on sensor placement, calibration, communication reliability, the quality of the fuzzy rules and the way the genetic algorithm’s fitness function prioritizes climate accuracy against energy use.
The work arrives as controlled-environment agriculture expands from specialized production into a broader strategy for food security and resource management. Greenhouses can protect crops from external weather and extend growing seasons, but they can also consume large amounts of electricity and heat when their control systems operate inefficiently. A system that recognizes the greenhouse as a dynamic, interconnected environment could help reduce unnecessary actuator operation while maintaining conditions required for plant growth. The authors’ framework therefore represents a step toward greenhouses that behave less like collections of independent machines and more like coordinated cyberphysical systems. Its eventual impact will depend on validation across crop types, climates, greenhouse designs and production scales, as well as on transparent measurements of energy savings and crop outcomes. Even so, the combination of sensing, estimation and evolutionary control offers a compelling blueprint for agricultural automation in which cleaner data and smarter decisions work together to make high-precision cultivation more resilient.
Subject of Research: Multi-sensor fusion and intelligent environmental control for greenhouses
Subject of Research: Agriculture
Article Title: Multi-sensor fusion based intelligent greenhouse environmental control system with LabVIEW and GA-optimized fuzzy control
Article References: Wang, N., Zhu, G., Zhang, R., Zhen, H., Guo, B., & Papadakis, G. (2026). Multi-sensor fusion based intelligent greenhouse environmental control system with LabVIEW and GA-optimized fuzzy control. Smart Agricultural Technology, 15, Article 102508. https://doi.org/10.1016/j.atech.2026.102508
Image Credits: AI Generated
DOI: 10.1016/j.atech.2026.102508
Keywords: controlled-environment agriculture, smart greenhouse, multi-sensor fusion, Kalman filtering, fuzzy control, genetic algorithm, LabVIEW, energy-efficient agriculture, greenhouse automation, IoT agriculture
Cite this news
APA MLA Chicago
SCIENMAG. (August 28, 2026). Smart Greenhouse Uses Multi-Sensor Fusion and Genetic Algorithm-Optimized Fuzzy Control. https://scienmag.com/smart-greenhouse-uses-multi-sensor-fusion-and-genetic-algorithm-optimized-fuzzy-control/
Copy citation Download RIS
Tags: climate regulation in agricultureclimate regulation in greenhousescoupled environmental parameter regulationcoupled environmental variable controldistributed sensor architectureenergy-efficient greenhouse automationenergy-efficient greenhouse systemsgenetic algorithm fuzzy controlgenetic algorithm-optimized fuzzy controlintegrated temperature and humidity managementintelligent environment managementKalman filtering for noisy sensor dataKalman filtering in environmental monitoringLabVIEW-based automationLabVIEW-based greenhouse managementmulti-sensor data fusionmulti-sensor data fusion in agriculturemulti-variable environmental control systempredictive climate control in agriculturesensor network for environmental monitoringsensor network for greenhousesSmart greenhouse controlSmart greenhouse control system
