AI-Based Redundant Integrated Environmental Control System for Smart Farming

The AI-based smart farm control system addresses failure vulnerabilities by using predictive controllers and energy management to maintain stable environmental conditions, ensuring uninterrupted operation and improved efficiency.

KR103046428B1Active Publication Date: 2026-11-11유한회사한국기계설비기술
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
KR1020250135793
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2025-09-20
Publication Date
2026-11-11
Estimated Expiration
2045-09-20

AI Technical Summary

Technical Problem

Conventional smart farm environmental control systems are vulnerable to failure, leading to destabilization of critical conditions like temperature, humidity, and light intensity, and fail to account for unique factors such as weather changes and crop growth stages, compromising control efficiency and stability.

Method used

An AI-based complex environment control redundancy system with predictive controllers, energy utilization indicators, and distributed redundancy modules that ensure uninterrupted operation by anticipating failures and managing energy reserves, using AI algorithms to maintain optimal environmental conditions.

Benefits of technology

The system ensures stable and efficient operation by preventing incorrect control transitions and maintaining optimal environmental conditions, even in the event of failures, through predictive scheduling and energy management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112025107908770-PAT00054_ABST
    Figure 112025107908770-PAT00054_ABST
Patent Text Reader

Abstract

The present invention relates to a smart farm AI-based complex environmental control redundancy system. The system is composed of an environmental data collection unit (100), a predictive controller (110, 120), a stored energy calculation unit (130), a distributed / cooperative redundancy module (140), and an alarm / recovery module (150). The environmental data collection unit collects temperature, humidity, CO₂, light intensity, soil moisture, and external weather data. The predictive controller calculates an operation schedule using an AI-based algorithm, and in the event of a failure of the main controller (110), the auxiliary controller (120) automatically switches over according to the predicted schedule to perform uninterrupted operation. The stored energy calculation unit calculates a stored energy index (UET / UCT) through data from multi-point sensors and flow / calorimeters to support switching decisions and optimal operation. The distributed / cooperative redundancy module determines whether to switch based on consensus and voting among multiple controllers, and the alarm / recovery module notifies the administrator terminal of failure / switching events in real time and performs an automatic recovery procedure. The present invention can simultaneously achieve stable operation, improved energy efficiency, and continuity of the growth environment in a smart farm.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to complex environmental control technology for smart farms, specifically to a redundancy system for stably controlling temperature, humidity, CO₂, light intensity, ventilation, cooling and heating, and thermal and cooling storage facilities inside a greenhouse. In particular, it belongs to an AI-based complex environmental control redundancy system that utilizes AI-based predictive control and energy retention indicators (UET / UCT) to guarantee uninterrupted operation even in the event of failures or communication anomalies, and simultaneously secures energy efficiency and growth stability by reflecting weather forecasts and crop growth data. Background Technology

[0002] Smart farm integrated environmental control systems primarily utilize a centralized SCADA / BEMS approach to manage sensors and control devices in an integrated manner. However, this structure has limitations, as the entire control system immediately ceases if a server or network failure occurs. The existing Master / Slave switching method also relies on simple fault detection, posing a risk that a Slave may be promoted in the event of data inconsistencies, which can destabilize crop growth environments such as greenhouse temperature, humidity, CO₂, and light intensity. In particular, smart farms require the simultaneous consideration of weather changes, energy consumption, and crop growth stages; however, existing technologies fail to adequately reflect these factors, necessitating improvements. Prior art literature

[0003] Published Patent 10-2024-0139372 Published Patent 10-2024-0130894 The problem to be solved

[0004] This invention aims to solve the various limitations of existing technology in smart farm complex environment control systems.

[0005] Conventional centralized control systems have a vulnerability where the entire operation halts if a server or network fails, and simple Master / Slave failover structures allow promotion to occur even when data is not perfectly consistent, potentially leading to the transmission of incorrect control commands. These issues destabilize key environmental conditions inside the greenhouse, such as temperature, humidity, CO₂ concentration, and light intensity, which can adversely affect crop growth. Furthermore, control efficiency and stability are compromised because the systems fail to adequately account for factors unique to smart farms, such as weather changes, energy consumption of thermal and cooling storage equipment, and crop growth stages.

[0006] Therefore, the objective of the present invention is to enable uninterrupted operation even in the event of failure or disaster, and to simultaneously secure the stability and energy efficiency of the smart farm environment by comprehensively considering prediction-based operation, energy reserve indicators, growth data, etc. means of solving the problem

[0007] To achieve this objective, the present invention proposes a smart farm AI-based complex environment control redundancy system in which various functional modules are organically combined.

[0008] An environmental data collection unit is provided to comprehensively collect not only internal greenhouse temperature, humidity, CO₂, light intensity, and soil moisture but also external weather information. Based on this data, a predictive controller, consisting of a main controller and an auxiliary controller, calculates an operating schedule. The predictive controller calculates future operating conditions in advance using AI algorithms, and in the event of a failure in the main controller, the auxiliary controller takes over uninterrupted operation according to the predicted schedule.

[0009] Furthermore, it includes a module that calculates the Energy Utilization Token (UET / UCT) index by utilizing data from multi-point sensors, flow rates, and calorimeters in the thermal and cold storage tanks; this allows for the determination of whether to switch operations after comprehensively reviewing remaining energy levels and environmental conditions, rather than simply detecting failures. In addition, a distributed and cooperative redundancy module is configured where multiple controllers operate collaboratively, ensuring that the remaining controllers continue operation even if one controller fails. Along with this, the alarm and recovery module notifies the administrator of the situation in real time and executes automatic recovery procedures, further enhancing system stability. Effects of the invention

[0010] According to the present invention, the smart farm AI-based complex environment control redundancy system can overcome existing limitations and achieve the following effects.

[0011] First, even if the main controller fails, the auxiliary controller ensures stability by taking over uninterrupted operation through a predictive operation schedule.

[0012] Second, since the transition is carried out through an approval process utilizing energy holding indicators and prediction algorithms rather than a simple fault detection method, it is possible to prevent the problem of incorrect promotions occurring due to data inconsistencies.

[0013] Third, through control that reflects weather forecasts and crop growth data, heating, ventilation, shading, and CO₂ supply are maintained at optimal levels, stabilizing the crop growth environment.

[0014] Fourth, since control is carried out including thermal and cold storage energy management, energy efficiency is improved and operating costs can be reduced.

[0015] Therefore, the present invention is an AI-based complex environmental control redundancy system for smart farms capable of simultaneously realizing stable operation and energy efficiency in a smart farm environment, and can be extended and applied to various industrial fields in the future. Brief explanation of the drawing

[0016] FIG. 1 is a block diagram illustrating the overall configuration of a smart farm AI-based complex environment control redundancy system according to the present invention. FIG. 2 is a configuration diagram showing the structure of a distributed and cooperative predictive controller according to the present invention and the data sharing and cooperative relationship between each controller. Figure 3 is a flowchart illustrating the flow of calculating UET (Usable Energy Temperature) and UCT (Usable Cooling Temperature) in the energy calculation unit according to the present invention. FIG. 4 is a prediction-based fault switching procedure diagram showing the process of switching to an auxiliary controller when a predictive controller according to the present invention detects a fault. FIG. 5 is a mapping diagram illustrating a configuration in which heating, cooling, ventilation, shading, CO₂ supply loops, etc. are linked with a prediction controller and an energy management module as an example of smart farm application according to the present invention. Specific details for implementing the invention

[0017] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings. These embodiments are merely examples to illustrate the technical concept of the present invention, and various modifications are possible within the scope of the essence of the present invention as described in the claims.

[0018] FIG. 1 is a block diagram illustrating the overall configuration of a smart farm AI-based complex environment control redundancy system (100) according to the present invention.

[0019] The system is composed of an environmental data collection unit (100), a prediction controller (110, 120), a holding energy calculation unit (130), a distributed and cooperative redundancy module (140), an alarm / recovery module (150), and a control target device (200), and each component is interconnected through a wired and wireless communication network.

[0020] The environmental data collection unit (100) collects internal data such as temperature, humidity, CO₂, light intensity, and soil moisture inside the greenhouse, and combines it with external environmental data such as external weather forecast, solar radiation, and wind speed. These data are exchanged in formats such as JSON, XML, and OPC-UA through a standard interface (210).

[0021] The predictive controller (110, 120) consists of a primary controller and a secondary controller, and generates a future operation schedule by applying an AI-based prediction algorithm. In the event of a failure, the secondary controller is immediately switched over to ensure uninterrupted operation.

[0022] The energy calculation unit (130) analyzes the multi-point sensor and flow rate / heat quantity data of the thermal storage / cooling tank to calculate the Usable Energy Temperature (UET) and Usable Cooling Temperature (UCT), and calculates the heating / cooling capacity.

[0023] The distributed and cooperative redundancy module (140) performs consensus and voting between multiple controllers, plan replication, and transition hysteresis, and implements a cooperative structure rather than a simple Master-Slave switching.

[0024] The alarm / recovery module (150) notifies the administrator terminal (smartphone, PC) of failure occurrence, transition, and recovery events and performs an automatic recovery procedure.

[0025] The control target device (200) consists of a heating and cooling pump, a ventilation fan, a shading curtain, a CO₂ supplier, etc., and each device is controlled according to the operation schedule of the predictive controller.

[0026] The standard interface (210) according to the present invention may adopt standard specifications for data exchange. For example, it may apply internationally widely used data formats such as JSON, XML, or OPC-UA as a basis.

[0027] Furthermore, depending on operating conditions such as farm size, network bandwidth, and device processing power, additional support may be provided for lightweight data formats such as CSV, YAML, MessagePack, Protobuf, and CBOR. These lightweight formats reduce packet size, minimize processing latency, and enable stable operation even on resource-constrained edge devices.

[0028] By supporting such diverse formats, the system ensures interoperability between heterogeneous environments, including sensors, actuators, gateways, servers, and cloud platforms, and enables flexible integration with various IoT devices and external cloud servers. This enhances data processing efficiency at smart farm sites and facilitates integration with cloud-based artificial intelligence services or big data analysis platforms.

[0029] Figure 2 illustrates the distributed and cooperative predictive controller structure and the data sharing and cooperative relationship between the controllers according to the present invention.

[0030] Controller A (110) (primary candidate), Controller B (120) (secondary candidate), and Controller C (130) (backup / edge) each independently include a health score module, a priority management unit, and a lease token issuance unit. Each controller internally evaluates its own status information and exchanges it in real time with other controllers through a heartbeat channel (240) and a QoS monitoring module (241). Through this, each controller periodically checks quality information such as the normality, delay, and loss rate of the other controller.

[0031] The shared model / data module (250) periodically shares and synchronizes the AI ​​model version, Feature Store, learning checkpoint, etc. used by each controller, so that operation can continue on the same data basis even if the controller is switched.

[0032] The consensus and voting module (260) adjusts control authority using a Quorum method. For example, if two or more of the three controllers detect a failure of a specific controller, a new leader is elected by majority consensus. In this process, a priority table, a timeout manager, and a leader election algorithm are used.

[0033] The state synchronization module (270) exchanges control plans, logs, and event sequences, and prevents data collisions using time synchronization techniques for distributed systems, such as Lamport timestamps or Vector Clocks. Additionally, it includes a plan replication unit and a collision resolution unit to ensure that the same schedule is consistently reflected in all controllers.

[0034] With this distributed cooperative structure, even if controller A (110) fails, controllers B (120) and C (130) can elect a new leader through the consensus / voting module (260) and reflect the latest operation plan through the state synchronization module (270), thereby ensuring control continuity. Therefore, uninterrupted operation is possible despite specific controller failures, and control stability and reliability are greatly improved.

[0035] Figure 3 illustrates the procedure of the retained energy calculation unit (130) according to the present invention.

[0036] First, the layer temperature values ​​from the multi-point temperature sensor (101) and the flow rate / heat meter (102) installed in the heat storage / cooling storage tank , flow rate , temperature difference Δ??, power ??, etc. are collected. This data is recorded together with the time log (137), and accuracy is ensured through the sensor correction coefficient and the time synchronization module.

[0037] The collected data undergoes processes such as outlier removal, drift correction, and missing interpolation in the preprocessing module (103), and the state estimation module generates highly reliable temperature and flow rate estimates through techniques such as Kalman filters.

[0038] In the next step, the temperature profile module (131) reproduces a layered profile. Dividing the tank into layers and the average temperature of each layer and volume Calculate to reflect the layered structure.

[0039] Afterwards, the energy integration module (132, 133) is at the reference temperature Calculate the stored energy based on .

[0040] - Heat capacity (heating energy available) :

[0041]

[0042] - Retained cooling capacity (cooling energy available):

[0043]

[0044] Here, is the density of the medium, is specific heat and can reflect temperature dependence if necessary.

[0045] The calculated value is adjusted to effective energy by taking into account the heat loss on the outer wall of the tank and the heat exchange efficiency in the loss correction module (134).

[0046] - Effective heat retention

[0047]

[0048] - Effective cooling capacity

[0049]

[0050] Here, is heat exchange efficiency, is the amount of loss due to insulation performance.

[0051] Next, the load prediction module (135) predicts the heating / cooling demand of the h-hour horizon based on weather forecasts, solar radiation, and internal environment data. Calculates.

[0052] - Continuous type :

[0053]

[0054] - Dispersed type :

[0055]

[0056] Finally, the threshold judgment module (136) compares the effective retained energy and the load prediction value to calculate the capacity indicator (UET / UCT) and the margin value (margin).

[0057] - Heating margin :

[0058]

[0059] - Cooling margin :

[0060]

[0061] - UET (Heating Supply Temperature Prediction) :

[0062]

[0063] - UCT (Cooling Supply Temperature Prediction) :

[0064]

[0065] Here Silver water change temperature, is the expected supply temperature.

[0066] Based on the judgment result, if the expected supply temperature is above the required value (heating) or below (cooling), it is judged as “sufficient,” otherwise it is considered “insufficient,” and the auxiliary controller switching or load reduction mode is activated. All calculation results and judgments are recorded in a log and notified to the administrator terminal (151) via the alarm / recovery module (150).

[0067] FIG. 4 is a prediction-based fault switching procedure diagram illustrating the process of switching to an auxiliary controller (120) when a prediction controller (110) according to the present invention detects a fault.

[0068] First, the system performs a self-diagnosis during the initialization phase and checks the sensor, network, and time synchronization status. When a normal state is confirmed, the main controller (110) starts operation. During operation, the status of the main controller is continuously monitored through the heartbeat channel (240) and the QoS monitoring module (241), and quality indicators such as heartbeat loss rate, latency, and packet loss rate are collected.

[0069] At the same time, the prediction engine (111) predicts the load by reflecting weather forecasts, external environment, and internal sensor data, and updates the operation schedule for a certain time horizon h. The prediction-to-actual deviation Δx is calculated through the difference between the actual data and the predicted data of the main controller, and is defined as follows.

[0070]

[0071] Here, is the actual state vector, is the predicted state vector.

[0072] deviation When the heartbeat and QoS indicators are within the allowable range and are at normal levels, the main controller (110) continues to operate. However, If it is above the threshold, or the risk index This predefined threshold If it exceeds, it is determined to be a “high-risk state.”

[0073] Risk Index is defined as follows.

[0074]

[0075] Here,

[0076] - is the heartbeat loss rate,

[0077] - is a QoS degradation indicator,

[0078] - is the predicted-actual deviation,

[0079] - is an energy reserve capacity indicator,

[0080] - is a weighting coefficient.

[0081] if ≥ Then, the system begins preparation for switching to the auxiliary controller (120).

[0082] At this time, the Catch-up synchronization module (144) checks whether the operation plan, log, and time information of the main controller and the auxiliary controller are aligned. Once synchronization is complete, the auxiliary controller (120) is granted Master authority through approval by the consensus / voting module (260). Subsequently, control authority is transferred via a non-stop handover method.

[0083] When the transition is completed, the alarm / recovery module (150) sends a notification to the administrator terminal (151), and the reason for the transition, the time, and the data status are recorded in the event log (154). When the main controller (110) is restored to normal, it is incorporated as a Slave through a rejoining procedure, and if necessary, it is returned to being a Master.

[0084] Meanwhile, in the event of a serious failure in both the main controller and the auxiliary controller, the system enters safety mode to protect the crop growth environment by maintaining only essential functions such as ventilation, shading, and minimal heating and cooling.

[0085] FIG. 5 is a mapping diagram illustrating an example of smart farm application according to the present invention, showing the process in which the heating, cooling, ventilation, shading, and CO2 supply loops inside the greenhouse are operated in conjunction with a predictive controller and a storage energy management module.

[0086] First, the heating and cooling loop (201) includes a pump, valve, heat pump, etc., and the retained heat amount calculated by the retained energy calculation unit (130). and retained cold energy Determine whether to maintain the target temperature based on.

[0087] The stored energy is calculated as follows.

[0088] ,

[0089] Based on this, the predictive controller calculates the predicted supply temperature value.

[0090]

[0091] Here Silver water change temperature, is mass flow rate, is the forecast horizon. This value is the target supply temperature. If satisfied, normal operation is maintained.

[0092] The ventilation loop (202) drives fans and dampers based on the temperature, humidity, and CO2 concentration inside the greenhouse and is also linked to external weather conditions. It can usually be expressed as a PID algorithm.

[0093]

[0094] Here is the deviation between the target ventilation rate and the actual ventilation rate. When the ventilation loop is activated, the CO2 loop is blocked by a safety interlock.

[0095] The shading / curtain roof (203) receives solar radiation and internal temperature data as input and adjusts the opening and closing rate in steps. It maintains the target photosynthetic effective radiation (PAR) in conjunction with the solar radiation prediction of the prediction controller.

[0096] The shading control rate can be defined as follows.

[0097]

[0098] Here is the target solar radiation, is the measured concentration.

[0099] In this way, each control loop operates by receiving the predicted operation schedule and UET / UCT indicators provided by the predictive controller (110, 120) and the energy reserve calculation unit (130). Additionally, the alarm / recovery module (150) monitors the status of each loop, notifies the administrator terminal if an abnormality is detected, and performs a recovery procedure.

[0100] Therefore, Figure 5 shows that the control loops inside the smart farm are operated by comprehensively reflecting prediction-based scheduling, determination of available energy capacity, and safety interlocks, rather than simple feedback control. As a result, the smart farm can simultaneously achieve uninterrupted operation, improved energy efficiency, and stability of the growing environment. Explanation of the symbols

[0101] 100 Environmental Data Collection Department 101 Temperature / Humidity / CO₂ / Light Intensity / Soil Sensor 102 RTU / Gateway 103 Preprocessing / Outlier Removal Module 110 Predictive Controller (Primary Controller) 111 Prediction Engine 120 Predictive Controller (Secondary Controller) 121 Predictive Controller (Backup / Edge Controller) 130 Energy Capacity Calculation Section (UET / UCT Calculation) 131 Temperature Profile Processing Module 132 Energy Integration Module 133 Cold and Thermal Integration Module 134 Loss Compensation Module (Tank Loss, Heat Exchange Efficiency) 135 Load Prediction Module (Qh Calculation) 136 Threshold determination unit (margin calculation) 137-hour log 140 Distributed / Cooperative Redundancy Modules 144 Catch-up Synchronization Module 150 Alarm / Recovery Modules 151 Notification Module (App / SMS / Gateway) 154 Event Log / Historian 160 Data Storage / Historian (Raw / Processed Data, Alarm / Setting History) 200 Controlled device 201 Heating and Cooling Loop (Pump, Valve, Heat Pump) 202 Ventilation Loop (VFD Fan, Heat Recovery Unit) 203 Shade / Curtain Roof 204 CO₂ supply loop (including injector, safety interlock) 210 Standard Interfaces (OPC-UA, REST, MQTT, JSON / XML, CSV, Protobuf, etc.) 240 Heartbeat Channels 241 QoS Monitoring Module 250 Shared Model / Data Modules (AI Model Versions, Feature Store) 260 Consensus and Voting Module 270 State Synchronization Module

Claims

Claim 1 A smart farm AI-based complex environment control redundancy system is characterized by comprising: an environment data collection unit (100) that collects environmental data; a prediction controller (110, 120) that calculates an operation schedule through an AI prediction algorithm based on the environmental data and external weather data, and automatically switches to an auxiliary controller (120) according to the predicted operation schedule to perform uninterrupted operation when a failure occurs in the main controller (110); an energy calculation unit (130) that calculates an energy index (UET, UCT) using multi-point sensors and flow rate / heat meter data of a thermal storage / cooling storage facility; a distributed / cooperative redundancy module (140) in which multiple controllers operate in a cooperative structure to agree on and decide whether to switch and prevent unnecessary repeated switching; an alarm / recovery module (150) that notifies a manager terminal of controller failure, switching, and recovery events and performs an automatic recovery procedure; and a control target device (200) that is controlled according to the operation schedule of the prediction controller (110, 120). Claim 2 A system according to claim 1, wherein the retained energy calculation unit (130) calculates the remaining energy of the heat storage / cooling storage facility using multi-point temperature sensors and flow rate / heat quantity data, and calculates UET and UCT based thereon. Claim 3 A system according to claim 1, wherein the prediction controller (110, 120) generates a future operation schedule by reflecting weather forecasts, external solar radiation, and a crop growth database (DB). Claim 4 A system according to claim 1, wherein the distributed and cooperative redundancy module (140) is characterized by a plurality of controllers deciding whether to switch through consensus and voting, rather than a Master-Slave single switching method. Claim 5 A system according to claim 1, characterized in that when a failure occurs in the main controller (110), the deviation and risk indicator between the predicted state vector calculated by the predictive controller (110, 120) and the actual state vector of the main controller (110) are evaluated, and a switch to the auxiliary controller (120) is performed only when the allowable range is exceeded. Claim 6 A system according to claim 1, wherein the alarm / recovery module (150) notifies a failure, transition, or recovery event to an administrator terminal in real time and includes a retransmission and acknowledgment (ACK) function. Claim 7 A system according to claim 1, wherein the auxiliary controller (120) performs data synchronization upon normal return of the main controller and then switches to a standby state through a rejoining procedure. Claim 8 A system according to claim 1, characterized in that the system ensures interoperability between controllers by using standard data formats such as JSON, XML, and OPC-UA. Claim 9 A system according to claim 1, wherein the control target device (200) comprises a heating and cooling pump, a ventilation fan, a shading curtain, a CO₂ supply device, and an irrigation and fertilization device. Claim 10 A system according to claim 1, wherein the prediction controller (110, 120) updates target environment values ​​based on a crop-specific growth database and reflects them in a control schedule.

Citation Information

Patent Citations

  • Distributed Control Smart Farm

    KR1020200027252A

  • Intelligent multiplex controller system for smart farm integrated management

    KR1020240130894A

  • System for ensuring continuity of environmental control in preparation for power outages in smart farm facility cultivation method thereof

    KR1020240139372A

  • Selective Control Apparatus of Smart Greenhouse and Selective Control Method Thereof and Smart Greenhouse System

    KR1020250090717A