Method for detecting abnormalities of smart farm actuators based on actuator response data according to control commands

Abstract

A method for detecting an abnormality in a smart farm actuator includes: a step in which a smart farm management device transmits a control command to an actuator according to a diagnostic schedule; a step in which the smart farm management device collects actuator response data generated according to the control command; a step in which the smart farm management device extracts a plurality of features corresponding to different time scales from the actuator response data; and a step in which the smart farm management device detects an abnormality in the actuator based on a plurality of features corresponding to different time scales.

Description

Technology Field

[0001] The technology described below pertains to the field of smart farm technology. Background Technology

[0002] A smart farm refers to a system that applies Information and Communication Technology (ICT) to agricultural production facilities to manage the crop growth environment automatically or semi-automatically. For example, a smart farm aims to maintain stable crop growth by using sensors to measure environmental information such as temperature, humidity, CO₂ concentration, solar radiation, soil moisture, and wind speed in greenhouses or cultivation facilities, and by controlling the cultivation environment based on the measured data.

[0003] Various actuators may be installed in a smart farm to control the cultivation environment. For example, actuators may include devices such as motors for opening and closing side or roof windows, ventilation fans, irrigation pumps, nutrient valves, heaters, or coolers. These actuators operate according to control commands transmitted from a control unit to regulate the environment of the smart farm.

[0004] Meanwhile, actuators in smart farms may experience performance degradation or failure due to factors such as repetitive operation, changes in the external environment, the ingress of foreign substances, and wear or aging of parts. Actuator malfunctions can lead to a failure in environmental control, adversely affecting crop growth, and in severe cases, resulting in facility shutdowns or increased maintenance costs. Therefore, stable operation of smart farms requires technology capable of continuously monitoring the status of actuators and detecting abnormalities at an early stage.

[0005] Traditionally, actuator malfunctions were often managed through operator inspections or regular maintenance; however, with the expansion of smart farms and the increasing demand for unmanned and automated operations, there is a growing need for methods to automatically diagnose and manage actuator malfunctions. Prior art literature

[0006] Korean Published Patent Application: 10-2025-0065018 The problem to be solved

[0007] Smart farms are equipped with various actuators, such as side window motors, ventilation fans, irrigation pumps, and nutrient valves, to control the cultivation environment. While these actuators may experience performance degradation or failure due to factors such as repetitive operation, disturbances (wind, temperature, pressure differences, etc.), foreign substances, wear, and aging, field operations often still rely on operator inspections or simple threshold-based monitoring. Consequently, this can lead to reduced diagnostic reliability, such as the failure to detect actuator abnormalities early or misidentifying normal response changes caused by environmental shifts as abnormalities.

[0008] Furthermore, the response characteristics of smart farm actuators can appear across various time scales, ranging from short-term transient responses (e.g., inrush current, ripple change) to performance indicators during the operating period (e.g., cycle time, average current / power) and long-term accumulated degradation trends (e.g., changes in health index). Therefore, if only a single observation period or a single indicator is used, early signs of anomalies may be missed or the progression of degradation may not be sufficiently reflected, which may reduce the accuracy and earlyness of anomaly detection.

[0009] The technology described below aims to provide a method for reliably detecting actuator abnormalities by systematically executing control commands (or diagnostic sequences) for diagnostic purposes on actuators in a smart farm, collecting corresponding actuator response data, and then extracting and combining multiple features corresponding to different time scales from the response data. Furthermore, by considering response deviations caused by changes in environmental conditions, this invention aims to reduce false positives and detect actuator abnormalities early, thereby improving the stable operation and maintenance efficiency of the smart farm.

[0010] The problem to be solved is not limited to those described as examples, and a person skilled in the art can derive various modified problems within the technical concept of the present invention. means of solving the problem

[0011] In one embodiment, a smart farm actuator fault detection method comprises: a step in which a smart farm management device transmits a control command to an actuator according to a diagnostic schedule; a step in which the smart farm management device collects actuator response data generated according to the control command; a step in which the smart farm management device extracts a plurality of features corresponding to different time scales from the actuator response data; and a step in which the smart farm management device detects an actuator fault based on a plurality of features corresponding to different time scales.

[0012] In one embodiment, the feature may be a mathematical indicator calculated to detect whether there is an abnormality in the actuator from the actuator response data.

[0013] In one embodiment, the diagnostic schedule may include at least one execution condition, execution time, and command sequence.

[0014] In one embodiment, the diagnostic schedule may be constructed based on at least one of smart farm environment information, smart farm crop information, smart farm actuator information, and smart farm aging information.

[0015] In one embodiment, establishing a diagnostic schedule based on the smart farm environment information may include determining execution conditions, execution time, and command sequence so that changes in the environment within the smart farm or the impact on the actuator within the smart farm are minimized by the operation of the actuator according to the control command.

[0016] In one embodiment, establishing the diagnostic schedule based on the smart farm crop information may include determining execution conditions, execution time, and command sequence so as to minimize the impact on crops within the smart farm resulting from the operation of an actuator according to the control command.

[0017] In one embodiment, establishing a diagnostic schedule based on the smart farm actuator information may include determining execution conditions, execution time, and command sequence to minimize the possibility of an accident within the smart farm due to the operation of the actuator according to the control command.

[0018] In one embodiment, establishing a diagnostic schedule based on the smart farm aging information may include determining execution conditions, execution time, and command sequences so as to minimize the impact on the actuator caused by the operation of the actuator according to the control command, taking into account the degree of aging of the smart farm actuator.

[0019] In one embodiment, the smart farm actuator anomaly detection method may further include: a step in which the smart farm management device collects environmental data at the time when the actuator response data is measured; and a step in which the smart farm management device selects only actuator response data within a preset environmental tolerance range by referring to the environmental data at the time when the actuator response data is measured.

[0020] In one embodiment, the actuator response data used to extract the plurality of features may be the selected actuator response data.

[0021] In one embodiment, the preset environmental tolerance range may include a range for a single environmental physical quantity included in the environmental data or a range for a combination of multiple environmental physical quantities.

[0022] In one embodiment, selecting only actuator response data within the preset environment tolerance range may include selecting only actuator response data in which the similarity between environment vectors, each having a plurality of environment physical quantities included in the environment data as a component, is within a preset range.

[0023] In one embodiment, selecting the response data of the actuator may further include selecting only the response data of the actuator according to the same control command.

[0024] In one embodiment, the smart farm actuator anomaly detection method may further include: a step in which the smart farm management device collects environmental data at the time when the actuator response data is measured; and a step in which the smart farm management device calculates an environmental impact component from the actuator response data based on the environmental data and the actuator response data, and removes the calculated environmental impact component from the actuator response data.

[0025] In one embodiment, the actuator response data used to extract the plurality of features may be actuator response data from which the environmental influence component has been removed.

[0026] In one embodiment, removing the environmental influence component from the actuator response data may include utilizing a correction model.

[0027] In one embodiment, the correction model may be a model that estimates the environmental influence component included in the actuator response data using environmental data and actuator response data as inputs, and outputs actuator response data from which the environmental influence component has been removed based on the estimation result.

[0028] In one embodiment, the feature may include at least one of average current, peak current, RMS (Root Mean Square), operation completion time, harmonic component, specific frequency band energy, position of spectrum peak, magnitude of spectrum peak, target arrival time, whether the target is achieved, whether an error code is generated, whether an interlock operation is performed, and whether an overcurrent is generated.

[0029] In one embodiment, the time scale includes a first time scale, a second time scale, and a third time scale, wherein the second time scale has a time interval larger than the first time scale, and the third time scale may have a time interval larger than the second time scale.

[0030] In one embodiment, detecting an abnormality of an actuator based on a plurality of features corresponding to different time scales may include detecting an abnormality of an actuator based on the number of features among the plurality of features corresponding to different time scales that are determined to be outside a normal threshold value.

[0031] In one embodiment, detecting an abnormality of an actuator based on a plurality of features corresponding to different time scales may include calculating an abnormality score corresponding to each of the plurality of features corresponding to different time scales, assigning different weights to the abnormality scores, and determining whether the sum of the abnormality scores assigned different weights exceeds a preset abnormality score threshold.

[0032] The different weights assigned to the above abnormal scores can be set differently depending on the type of actuator and the degree of aging.

[0033] In one embodiment, detecting an abnormality of an actuator based on a plurality of features corresponding to different time scales may include hierarchically analyzing the plurality of features corresponding to different time scales to detect whether an abnormality exists.

[0034] In one embodiment, the smart farm actuator fault detection method may further include: a step in which the smart farm management device classifies the type of fault based on whether the fault is detected and infers the cause of the fault; and a step in which the smart farm management device proposes a maintenance plan based on the inferred cause of the fault.

[0035] In one embodiment, the smart farm management device may include a computing device and a storage device including 'commands' that cause the smart farm management device to perform operations when executed by the computing device.

[0036] In one embodiment, the operations may include: an operation in which the smart farm management device transmits a control command to an actuator according to a diagnostic schedule; an operation in which the smart farm management device collects actuator response data generated according to the control command; an operation in which the smart farm management device extracts a plurality of features corresponding to different time scales from the actuator response data; and an operation in which the smart farm management device detects an abnormality in the actuator based on a plurality of features corresponding to different time scales.

[0037] In one embodiment, the computer program stored on the recording medium may be a computer program stored on the recording medium for executing a smart farm actuator abnormality detection method. Effects of the invention

[0038] By utilizing the technology described below, actuator response data can be systematically collected using control commands (or diagnostic sequences) according to a diagnostic schedule, thereby securing response data with higher consistency and reproducibility compared to data obtained under random operation or non-standard operating conditions. Consequently, the data quality for evaluating the actuator status is improved, and the reliability of the diagnostic results can be increased.

[0039] By using the technology described below, multiple features corresponding to different time scales are extracted from actuator response data, and anomalies are determined either individually or in combination. This allows for the complementary reflection of early signs of anomalies (short-term transient response), operational performance degradation (operational interval characteristics), and long-term degradation trends (cumulative change), compared to diagnosis based on a single indicator or a single time interval. As a result, the accuracy and early detection of anomalies can be improved.

[0040] By utilizing the technology described below, it is possible to exclude response deviations caused by environmental changes or determine abnormalities based on corrected response data (or residuals) by using environmental data together. This reduces false positives caused by misidentifying normal response changes resulting from disturbances, such as wind speed, temperature, and pressure differences, as abnormalities. Consequently, the maintenance efficiency of smart farm actuators is improved, and the effect of preventing environmental control failures and crop damage caused by malfunctions can be expected.

[0041] The effects of the invention are intended for ease of explanation, and the description of effects in the specification is not to be interpreted as grounds for limiting the scope of the claims. Brief explanation of the drawing

[0042] FIG. 1 shows one embodiment in which a smart farm management device (100) performs a smart farm actuator abnormality detection method. FIG. 2 shows one of the embodiments (200) of a smart farm actuator abnormality detection method. FIG. 3 shows one example (300) of a control command according to the type of actuator according to an embodiment. FIG. 4 shows one example (400) of a diagnostic schedule according to an embodiment. Figure 5 is one example (500) of using more environmental data. FIG. 6 is one of the embodiments (600) for analyzing multiple time scales. FIG. 7 shows one of the embodiments of a smart farm management device (700). Specific details for implementing the invention

[0043] The technology described below is capable of various modifications and may include various embodiments. Specific embodiments of the technology described below may be described in the drawings of the specification. However, this is for the purpose of explaining the technology described below and is not intended to limit the technology described below to specific embodiments. Accordingly, it should be understood that all modifications, equivalents, and substitutions that fall within the spirit and scope of the technology described below are included in the technology described below.

[0044] Terms such as first, second, A, B, etc., may be used to describe various components. However, the above terms are used merely to distinguish one component from other components and are not intended to limit the components to the said terms. For example, without departing from the scope of rights of the technology described below, the first component may be named the second component, and similarly, the second component may be named the first component. The term “and / or” includes a combination of multiple related described items or any of the multiple related described items.

[0045] In the terms used below, singular expressions should be understood to include plural expressions unless the context clearly indicates otherwise, and terms such as "includes" should be understood to mean that the described features, number, steps, actions, components, parts, or combinations thereof exist, and not to exclude the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0046] Before providing a detailed description of the drawings, it is to clarify that the classification of components in this specification is merely based on the primary function each component is responsible for. That is, two or more components described below may be combined into a single component, or a single component may be divided into two or more components based on more subdivided functions. Furthermore, each component described below may additionally perform some or all of the functions of other components in addition to its primary function, and it is obvious that some of the primary functions of each component may be exclusively performed by other components.

[0047] Furthermore, in performing the method or operation method, each process constituting the method may occur differently from the specified order unless a specific order is clearly indicated in the context. That is, each process may occur in the same order as specified, may be performed substantially simultaneously, or may be performed in the reverse order.

[0049] FIG. 1 shows one embodiment in which a smart farm management device (100) performs a smart farm actuator abnormality detection method.

[0050] A smart farm (110) may be a farm in which the crop cultivation environment is managed automatically or semi-automatically using information and communication technology (ICT), sensor technology, and automatic control technology. A smart farm (110) may be implemented in various forms of cultivation facilities such as greenhouses, vinyl greenhouses, and plant factories, and the structure and components of the cultivation space may differ depending on the form of the facility.

[0051] A smart farm (110) can grow one or more crops. For example, a smart farm (110) can grow various crops such as leafy vegetables, fruit vegetables, flowers and seedlings, but is not limited thereto.

[0052] A smart farm (110) may be equipped with a plurality of sensors for measuring the internal environment and a plurality of actuators for controlling the internal environment. In one embodiment, the sensors may include sensors for measuring temperature, humidity, carbon dioxide concentration, solar radiation (or light intensity), wind direction, wind speed, soil (or growing medium) moisture, EC, pH, etc. Additionally, the actuators may include at least one of a ventilation fan, a heater, a cooling device, a nutrient solution supply device, a spraying device, a shading screen, a lighting device, and a window opening / closing actuator, but are not limited thereto.

[0053] The smart farm management device (120) can be implemented with various hardware and software configurations. For example, the smart farm management device (120) can be implemented in the form of a computer, server, edge computing device, PLC, microcontroller-based controller, smart device, or chipset dedicated to data processing, and can communicate with the sensors and actuators of the smart farm (110) via wired or wireless communication (e.g., RS-485, Ethernet, Wi-Fi, BLE, LPWAN, etc.). However, the implementation form of the smart farm management device (120) is not limited to the examples described above.

[0054] The smart farm management device (120) may be configured to perform the smart farm actuator abnormality detection method described below. The smart farm management device (120) may store a diagnostic schedule. The smart farm management device (120) may transmit a control command according to the diagnostic schedule to the actuator. The actuator within the smart farm performs an operation according to the control command, and actuator response data may be generated during this process. The smart farm management device (120) may receive the actuator response data according to the transmitted control command. The smart farm management device (120) may detect an abnormality based on the actuator response data.

[0056] FIG. 2 shows one of the embodiments (200) of a smart farm actuator abnormality detection method.

[0057] The smart farm management device can transmit control commands to the actuator according to the diagnostic schedule (210).

[0058] Control commands based on a diagnostic schedule may be commands specialized for diagnosing actuator abnormalities. These control commands may be special commands intended to diagnose smart farm actuators, rather than general control commands. Control commands may refer to electrical, logical, or digital signals transmitted from a smart farm management device to an actuator to instruct its operation. Control commands may include the start of actuator operation, stop of operation, driving speed, position, torque, opening / closing rate, or patterns of change thereof. Control commands may be provided in the form of a command sequence consisting of a single command or multiple commands arranged in temporal sequence.

[0059] Alternatively, control commands can be set differently depending on the actuator type (e.g., side window motor, ventilation fan, irrigation pump, nutrient valve) or the diagnostic purpose (e.g., diagnosis of starting characteristics, diagnosis of friction changes, diagnosis of brake characteristics).

[0060] FIG. 3 shows one example (300) of a control command according to the type of actuator according to an embodiment.

[0061] In one embodiment, when the actuator is a side window motor, the control command may be [open → close → open] to analyze current spikes during reversal, thereby enabling the detection of gear wear and rail foreign matter.

[0062] In one embodiment, when the actuator is a ventilation fan, the control command may be [OFF→→] to analyze the start / stop transient characteristics, thereby allowing the detection of bearing wear or blade imbalance.

[0063] In one embodiment, when the actuator is a water pump, the control command to analyze the repeated start response may be [ON 5 seconds → OFF → 5 seconds], thereby detecting priming problems, leakage, or cavitation.

[0064] In one embodiment, when the actuator is a nutrient valve, the control command to evaluate flow rate stability after opening may be [open → hold for 30 seconds → close], thereby detecting blockage, leakage, or operation delay.

[0065] A diagnostic schedule may refer to a set of predefined execution rules specifying when, under what conditions, and in what manner control commands are executed to diagnose abnormalities in an actuator.

[0066] The diagnostic schedule may include at least one execution condition, execution time, and command sequence.

[0067] The execution condition may be a condition for executing a control command. In one embodiment, the execution condition may include a time condition (e.g., 04:00 every day, 03:00 every Sunday), an environmental condition (e.g., wind speed is below a threshold, within an external temperature range), an equipment condition (e.g., no manual operation, safety interlock satisfied, absence of a worker), or a combination thereof.

[0068] The smart farm management device can transmit control commands to the actuator when execution conditions are met. This allows the device to collect actuator response data only when specific execution conditions are satisfied, rather than collecting data when the actuator operates by chance. This ensures the consistency, reliability, reproducibility, and quality of the collected actuator response data. Alternatively, data quality can be guaranteed by collecting actuator response data during time periods free from physical noise, which facilitates diagnosis. Furthermore, by acquiring actuator response data under identical execution conditions, it is possible to collect data with clear cause-and-effect relationships and specificity.

[0069] The execution time may include at least one of the time when the operation according to the control command starts, the time when the operation according to the control command ends, and the period during which the operation according to the control command is in operation.

[0070] A command sequence may refer to the sequence, combination, and temporal arrangement of control commands executed to diagnose an abnormality of an actuator according to a diagnostic schedule. A command sequence may be composed of multiple control commands arranged in a temporally continuous or repetitive manner. For example, a command sequence may include a target value of the control command (e.g., opening / closing rate, speed, torque, position), a holding time, a switching condition (e.g., upon reaching the target value, after a certain period of time), a number of repetitions, and / or acceleration / deceleration conditions. For example, a command sequence may include at least one of a step sequence that gradually changes the target value within the actuator's operating range, a micro-reciprocating sequence that reciprocates the target value, a hold sequence that maintains a stationary state at a specific target value, or a stress sequence that performs high-speed repetitive operation.

[0071] FIG. 4 shows one example (400) of a diagnostic schedule according to an embodiment.

[0072] In one embodiment, the diagnostic schedule may include a Daily Auto-Diagnosis. The Daily Auto-Diagnosis may collect actuator response data according to a control command at a predetermined time during the day. For example, the diagnostic schedule may include [Execution condition: 4:00 AM], [Execution time: 30 seconds], and [Command sequence: Fine back-and-forth of the smart farm opening / closing rate from 10% → 20% → 10%].

[0073] In one embodiment, the diagnostic schedule may include Weekly Precision Diagnosis. Weekly Precision Diagnosis may collect actuator response data according to control commands on a predetermined day of the week. For example, the diagnostic schedule may include [Execution condition: 3:00 AM on Sunday], [Execution time: 3 minutes], [Command sequence: Repeat opening / closing rate of smart farm at 0%→→→→→].

[0074] In one embodiment, the diagnostic schedule may include a Monthly Load Diagnosis. The Monthly Load Diagnosis may collect driver response data according to a control command on a fixed date each month. For example, the diagnostic schedule may include [Execution condition: 1st day of every month], [Execution time: 1 minute], [Command sequence: 3 pauses for 10 seconds at 50%].

[0075] In one embodiment, the diagnostic schedule may include an Annual Stress Diagnosis. The Annual Stress Diagnosis may collect actuator response data according to a control command at a set time each year. For example, the diagnostic schedule may include [Execution condition: Spring / Autumn off-season], [Execution time: 10 minutes], [Command sequence: 10 high-speed repetitions of 0% → 100%].

[0076] The diagnostic schedule can be constructed based on at least one of smart farm environment information, smart farm crop information, smart farm actuator information, and smart farm aging information.

[0077] In one embodiment, environmental information may include temperature, humidity, solar radiation, wind speed / direction, rainfall, pressure difference, ventilation status, etc. In one embodiment, crop information may include crop type, growth stage, crop sensitivity (e.g., light / temperature sensitivity), cultivation schedule, work schedule, etc. In one embodiment, actuator information may include actuator type (e.g., window actuator, ventilation fan, pump, valve, etc.), rating, operating range, installation location, safety interlock conditions, recent operation history, maintenance history, etc. In one embodiment, aging information may include cumulative operation count or time, trend of change in current / power time series characteristics, anomaly score, probability of failure, predicted remaining lifespan (RUL) value, or alarm history, etc.

[0078] In one embodiment, a smart farm management device can determine the execution conditions, execution time, and command sequence of a diagnostic schedule based on smart farm environment information, smart farm crop information, smart farm actuator information, and smart farm aging information.

[0079] In one embodiment, establishing a diagnostic schedule based on smart farm environment information may include determining execution conditions, execution time, and command sequences so that changes in the environment within the smart farm or the impact on the actuator within the smart farm are minimized through the operation of the actuator according to control commands. For example, the smart farm management device may select a time period that satisfies environmental conditions where disturbances are relatively small, such as when the wind speed is below a threshold or the pressure difference is within an allowable range, as the diagnostic execution time period.

[0080] In one embodiment, establishing a diagnostic schedule based on smart farm crop information may include determining execution conditions, execution times, and command sequences so that the impact on crops within the smart farm resulting from the operation of actuators according to control commands is minimized. For example, the smart farm management device may restrict the performance of the diagnosis to only during time periods when the impact of opening, closing, or ventilation operations on the crops can be minimized (e.g., times when workers are absent, nighttime, or times permitted by the cultivation schedule), depending on the growth stage or characteristics of the crops.

[0081] In one embodiment, establishing a diagnostic schedule based on smart farm actuator information may include determining execution conditions, execution times, and command sequences to minimize the possibility of accidents within the smart farm caused by the operation of actuators according to control commands. For example, the smart farm management device may include equipment status conditions such as “no manual operation,” “safety interlock satisfied,” and “emergency stop released” in the execution conditions to exclude time periods where manual operation is highly likely or where there is a safety risk.

[0082] In one embodiment, establishing a diagnostic schedule based on smart farm aging information may include determining execution conditions, execution time, and command sequences so as to minimize the impact on the actuator caused by the operation of the actuator according to control commands, while considering the degree of aging of the smart farm actuator.

[0083] In one embodiment, a smart farm management device may determine or select a command sequence according to the type of actuator and the purpose of diagnosis. For example, the smart farm management device may store a library of predefined diagnostic command sequences for each actuator type and select a command sequence corresponding to the actuator type or suspected failure mode. In this case, the command sequence may include a target position (or opening / closing rate), driving speed, torque (or current / power limit), holding time, number of repetitions, acceleration / deceleration conditions, or a temporal change pattern thereof, and at least one of the parameters may be variably adjusted according to aging information.

[0084] In one embodiment, the diagnostic schedule may be set fixedly, but it may also be constructed as an adaptive schedule that is dynamically updated based on aging information or anomaly detection results. For example, if the health index falls below a reference value, the anomaly score exceeds a reference value, or the remaining lifespan is predicted to be below a threshold value, the smart farm management device may increase the diagnostic frequency (e.g., daily / weekly / monthly), allocate additional time for diagnostic execution, or increase the diagnostic intensity (e.g., number of command sequence repetitions, range of change, duration). Conversely, if the actuator status is determined to be stable, the diagnostic frequency or intensity may be decreased.

[0085] The smart farm management device can collect actuator response data generated according to control commands (220).

[0086] Actuator response data may refer to data representing physical, electrical, temporal, and / or state characteristics that occur during the actual operation of the actuator according to a control command. Actuator response data may be collected continuously or discretely during the period from when the control command is transmitted until the operation of the actuator is completed.

[0087] Actuator response data may be measured or calculated from the actuator itself, the actuator drive circuit, a controller (e.g., PLC), or an external sensor (e.g., CT sensor). For example, the actuator response data may include current, voltage, power, frequency components, torque (or estimated torque), actuation time, operation completion time, speed and speed change, position and position change, whether the target is reached, target error, overshoot / undershoot, number of repeated operations, alarm / error code, whether an interlock operation is performed, or at least one of these.

[0088] Actuator response data can be recorded in association with specific control commands. Actuator response data can be collected and categorized according to identical or similar control commands. By comparing changes in the actuator response data, it can be used to determine changes in the actuator's state, progression of deterioration, or abnormalities. Actuator response data can be stored aligned with the timestamps of the control commands.

[0089] Actuator response data can be recorded in conjunction with environmental data. In other words, it may be collected separately based on identical or similar environments and identical or similar control commands. In smart farm environments, environmental conditions such as temperature, wind speed, humidity, and pressure difference can change over time; therefore, actuator responses to the same control command may vary depending on these conditions. Consequently, if the characteristics of actuator response data collected under varying environmental conditions are simply compared or analyzed using the same criteria, there is a possibility that normal response deviations caused by environmental changes may be misinterpreted as actuator malfunctions. Furthermore, even if data is collected under similar execution conditions via a diagnostic schedule, it is difficult to ensure that environmental conditions at the time of measurement remain completely identical. To address this, it is necessary to utilize actuator response data obtained from identical environments as well as identical control commands.

[0090] FIG. 5 is one example (500) of further utilizing environmental data. For example, a smart farm management device can select only the actuator response data collected under the same control command and under conditions where the internal temperature is 20°C to 22°C and the external wind speed is 1 m / s to 2 m / s. The smart farm management device can extract time-scale characteristics (e.g., current characteristics, cycle time, response delay, target arrival time, or trends of change thereof) from the selected actuator response data and detect whether there is an actuator abnormality while excluding normal response deviations caused by environmental changes based on the above characteristics.

[0091] In one embodiment, the smart farm management device may further utilize smart farm environment data to detect an abnormality in an actuator. The smart farm management device may detect an abnormality in an actuator by further utilizing smart farm environment data along with multiple features corresponding to different time scales. Specifically, the smart farm management device may use environment data representing the smart farm environment to select only the features extracted from actuator response data measured under the same environmental conditions, and detect an abnormality in the actuator based on the selected features.

[0092] Environmental data may be data regarding internal / external environmental conditions or environmental changes within the smart farm. Environmental data may include physical quantities that can affect the load or response characteristics of the actuator. In one embodiment, environmental data may include at least one of temperature, humidity, wind speed, pressure difference, solar radiation, CO₂ concentration, soil moisture, or a combination thereof.

[0093] In one embodiment, a smart farm management device can select only actuator response data within a preset environmental tolerance range by referring to environmental data at the time when the actuator response data was measured. By using only actuator response data within the preset environmental range, response deviations due to environmental differences can be minimized, and response characteristics caused by changes in the state of the actuator itself or abnormalities can be analyzed more clearly.

[0094] The preset environmental tolerance range used to select environmental data may refer to the acceptable range of environmental conditions that must be satisfied for actuator response data to be used for analysis or comparison.

[0095] In one embodiment, the preset environmental tolerance range may be a range for a single environmental physical quantity included in the environmental data or a range for a combination of multiple environmental physical quantities. For example, the environmental tolerance range may be defined as an upper limit and / or lower limit (or center value ± tolerance) for at least one of the environmental data, such as temperature, humidity, wind speed, pressure difference, solar radiation, CO₂ concentration, and soil moisture.

[0096] In one embodiment, a preset environmental tolerance range may be determined by whether the similarity between environmental vectors, each having multiple environmental physical quantities included in the environmental data as components, falls within a preset range. For example, a smart farm management device may calculate a distance or similarity scale between a reference environmental vector and an environmental vector at a measurement point in time, and select actuator response data at that point in time only if the distance (or similarity) falls within a preset threshold range. Here, the distance or similarity scale may be at least one of Euclidean distance, Manhattan distance, Mahalanobis distance, cosine similarity, or a combination thereof, and normalization or weighting may be applied to correct for differences in the unit and scale of each environmental physical quantity.

[0097] The smart farm management device can calculate environmental impact components from actuator response data by utilizing environmental data together, and remove the calculated environmental impact components from the actuator response data. The smart farm management device can detect actuator abnormalities based on the actuator response data from which environmental impact components have been removed.

[0098] The environmental influence component may refer to the variation caused by the environment included in the actuator response data as the actuator load or response characteristics fluctuate due to environmental conditions such as wind speed, temperature, pressure difference, humidity, and solar radiation. For example, the environmental influence component may include at least one of an increase in current, an increase in power, an increase in operating time, or an increase in response delay that may occur solely due to changes in environmental conditions, regardless of deterioration or failure of the actuator itself.

[0099] To remove environmental impact components from actuator response data, the smart farm management device may utilize a correction model. The correction model may be a model that uses environmental data and actuator response data as inputs to estimate environmental impact components included in the actuator response data, and outputs actuator response data from which environmental impact components have been removed based on the estimation results.

[0100] In one embodiment, the correction model may include a physics-based model, a statistical model (e.g., a regression model), and a machine learning-based model. For example, the correction model may be a model based on a physics formula that approximates the effect of environmental factors on the actuator load, such as wind pressure load according to wind speed, changes in lubrication characteristics according to temperature, and opening / closing resistance according to pressure difference.

[0101] For example, the correction model can generate a correction current with environmental influences removed by calculating the aerodynamic wind pressure load torque applied at the current window opening / closing rate based on wind speed information included in environmental data, and estimating the current component corresponding to the wind pressure load torque and subtracting it from the actual driving current.

[0102] Furthermore, the above correction model is not limited to window actuators but can be applied to various actuators such as ventilation fans, irrigation pumps, and nutrient valves. For example, the above correction model estimates load fluctuations of the actuator using environmental data and / or operating condition data (e.g., temperature, wind speed, pressure difference (differential pressure), flow rate, head, fluid viscosity, etc.) and estimates environmental influence components included in response data (e.g., driving current, power consumption, response delay time, etc.) due to said load fluctuations. Then, by removing (subtracting or correcting) the estimated environmental influence components from the measured response data, corrected response data from which environmental influences have been removed can be generated.

[0103] The smart farm management device can extract multiple features corresponding to different time scales from actuator response data (230).

[0104] The time scale may be the temporal observation range for interpreting actuator response data. The time scale may refer to the length of the analysis interval (observation window) or the temporal resolution. Different time scales can be used to extract and highlight different physical / operational characteristics from actuator response data corresponding to the same control command.

[0105] Extracting multiple features corresponding to different time scales may mean setting different temporal observation ranges (observation windows) or temporal resolutions for actuator response data, and calculating feature values ​​that quantify the physical / operational characteristics emphasized at each time scale. The purpose of extracting multiple features corresponding to different time scales is to interpret the actuator response data differently depending on the length of the temporal observation interval.

[0106] Features may refer to mathematical indicators or expressions calculated or derived from actuator response data (e.g., current, voltage, power, speed, position, operating time, etc.) to determine the state, operating characteristics, or abnormalities of the actuator. Features may not be the raw response data itself, but rather values ​​obtained through preprocessing, transformation, statistical summarization, frequency analysis, or model-based estimation of the raw data.

[0107] Features can be a single value (scalar) or a vector composed of multiple values ​​(feature vector). Features may also be in the form of a time series arranged according to time, calculated for each specific time interval (observation window). Alternatively, features can be calculated corresponding to different time scales (e.g., micro / macro / long-term) or different processing methods (e.g., filtering band, downsampling rate).

[0108] In one embodiment, the features may include average current (or average power), peak current (or inrush current), RMS and / or operation completion time (cycle time) calculated in the time domain of the driver response data. Alternatively, the features may include harmonic components, specific frequency band energy, and / or the position or magnitude of spectrum peaks calculated in the frequency domain of the driver response data. Alternatively, the features may include target arrival time, target error, and / or whether the target was not reached calculated in terms of the operation performance (control response) of the driver response data. Alternatively, the features may include event / state information acquired from the driver or controller, such as whether an alarm or error code occurred, whether an interlock operation occurred, and / or whether an overcurrent / overload occurred.

[0109] FIG. 6 is one of the embodiments (600) for analyzing multiple time scales.

[0110] In one embodiment, the time scale may include a first time scale, a second time scale, and a third time scale. The second time scale may have a time interval (observation interval) larger than the first time scale, and the third time scale may have a time interval larger than the second time scale. Alternatively, the third time scale may have a time interval larger than the first time scale. For example, a smart farm management device may set a reference interval from the point in time when a control command is applied to the actuator response data until the completion of operation (or stable state), and for the reference interval, calculate (i) a first time scale feature based on a short time scale, (ii) a second time scale feature based on the entire operation or a medium-length time scale, and (iii) a third time scale feature based on a long time scale aggregated from multiple operation histories.

[0111] The first time scale may be a micro-scale. The first time scale may be a time scale for capturing electrical or dynamic characteristics (dynamic properties) occurring in very short time intervals (e.g., at the millisecond (ms) level) that are difficult to distinguish with the human eye.

[0112] In one embodiment, the first time scale may include current, voltage, power, or change components among the driver response data, and may be a time scale for analyzing minute variations of the electrical response (e.g., ripple, peak, harmonic components, etc.). For example, the first time scale may enable the acquisition of driver response data with a resolution capable of high-speed sampling of 1 kHz or more (or the extraction of a corresponding section from already acquired data), and may enable the analysis of periodicity, irregularity, or frequency components of the ripple pattern within the section. In this case, the irregularity of the ripple may be related to specific failure factors, such as commutator damage or initial cracking of the bearing.

[0113] In one embodiment, the first time scale may be a time scale for analyzing the transient response during the starting phase. For example, by observing actuator response data with the first time scale to calculate the magnitude, duration, or I²t energy integral value of the inrush current, and comparing the calculated value with a reference value or previous history, the possibility of insufficient starting torque can be diagnosed.

[0114] The second time scale may be a macro-scale. The second time scale may be a time scale for evaluating the mechanical behavior or operation completion characteristics of the actuator, and can be analyzed by setting an operation interval in units of seconds to minutes as an observation interval.

[0115] In one embodiment, the second time scale can calculate the cycle time, average current per section, average power, or load estimation indicator during the operation period from the time the drive command is transmitted until the drive is completed, and can analyze whether there is a change in said indicator. For example, if the cycle time increases or the average current increases compared to the previous day, it can be determined that there is a possibility that an abnormality has occurred in the actuator due to a change in mechanical load factors, such as foreign matter jamming in the rail, lack of lubricant, or increased friction in the mechanism.

[0116] The third time scale may be a long-term scale. The third time scale may be a time scale for tracking the cumulative trend of aging or performance degradation occurring in the actuator over time, and can be interpreted based on long-term observation periods of days, weeks, months, or longer.

[0117] In one embodiment, the third time scale may construct time series data by aggregating feature values ​​or indicators calculated from the first time scale and / or the second time scale on a daily / weekly / monthly basis, and may estimate the state of deterioration progression using the trend, gradient, or health index of the time series data. Alternatively, the remaining useful life (RUL) or maintenance time of the actuator may be predicted using the point at which the health index reaches a threshold or the result of trend extrapolation.

[0118] The smart farm management device can detect an abnormality in the actuator based on multiple features corresponding to different time scales (240).

[0119] Detecting an anomaly may mean determining whether actuator response data collected during the process of the actuator operating according to a control command deviates from the range or reference characteristics expected in a normal state. Here, an anomaly may include an abnormal state caused by the failure or deterioration of the actuator. For example, an anomaly may include conditions such as initial signs of an anomaly, performance degradation, intermittent malfunction, failure to reach a target, or overload.

[0120] In one embodiment, a smart farm management device establishes a normal threshold based on normal response data for the same control command and can detect an anomaly based on whether a value according to a plurality of features extracted from collected actuator response data exceeds the normal threshold. For example, if at least one of the features, such as average current, peak current, cycle time, target arrival time, and current ripple magnitude, exceeds a reference range, or if the deviation (distance) relative to the reference distribution is greater than or equal to the threshold, the smart farm management device can determine that the actuator is in an abnormal state.

[0121] Alternatively, anomaly detection may be performed through binary judgment (normal / abnormal), or by comparing thresholds after calculating continuous values ​​such as anomaly scores or risk levels. Furthermore, the smart farm management device may detect anomalies by utilizing environment-corrected response data (or corrected features, residuals) to exclude normal deviations caused by environmental changes.

[0122] Detecting an abnormality in an actuator based on multiple features corresponding to different time scales may mean calculating features corresponding to multiple time scales from actuator response data collected according to the same control command, and determining whether the features deviate from the expected range or reference characteristics in a normal state, either individually or in combination.

[0123] In one embodiment, a smart farm management device can detect whether an actuator is abnormal based on the number of features determined to be outside a normal threshold among a plurality of features corresponding to different time scales.

[0124] For example, a smart farm management device may determine that an actuator is abnormal if at least two of the first, second, and third time-scale features each deviate from their corresponding normal threshold values. Alternatively, the smart farm management device may classify the actuator as being in an observed state if only one of the first, second, and third time-scale features deviates from its corresponding normal threshold value. Specifically, even if the first time-scale feature exceeds the threshold value, if the second and / or third time-scale features are within the normal range, the actuator may be classified as being in an observed state rather than being confirmed as abnormal. Alternatively, if the first and second time-scale features simultaneously exceed the threshold values, it may be determined that there is an abnormality. By utilizing a combination of multiple conditions for time-scale features in this manner, false positives can be reduced compared to judgments based on a single time-scale feature.

[0125] In one embodiment, a smart farm management device calculates an anomaly score corresponding to each of a plurality of features corresponding to different time scales, assigns different weights to the anomaly scores, and can detect whether an actuator is abnormal based on whether the sum of the anomaly scores with different weights exceeds a preset anomaly score threshold. At this time, the different weights assigned to the anomaly scores may vary depending on the type of actuator and the degree of aging.

[0126] For example, in the case of an actuator (e.g., a motor drive) where electrical transient response or ripple components are effective for abnormal diagnosis, a relatively large weight may be assigned to the first time-scale feature. Or, in the case of an actuator (e.g., a valve, a pump) where operation completion time or load change is effective for abnormal diagnosis, a relatively large weight may be assigned to the second time-scale feature. Or, in the case of an actuator where long-term degradation trends are important, a relatively large weight may be assigned to the third time-scale feature. Alternatively, the weights may be predefined as user-set values, or may be learned or updated based on past operation data or diagnostic history.

[0127] In one embodiment, a smart farm management device can detect anomalies by hierarchically analyzing multiple features corresponding to different time scales. For example, the smart farm management device can detect anomaly candidates using a first time scale feature, verify whether the operational performance of the anomaly candidates has deteriorated using a second time scale feature, and finally determine whether an anomaly exists by confirming the persistence of anomaly signs or the cumulative trend of deterioration using a third time scale feature. By applying such stepwise judgment, early detection and anomaly confirmation can be performed simultaneously.

[0128] In one embodiment, a smart farm management device can generate an integrated feature vector by combining multiple features calculated over a time scale and detect whether an actuator is abnormal based on the integrated feature vector. For example, the smart farm management device can calculate the distance (deviation) between the integrated feature vector and a reference feature vector (or reference distribution) and compare it with a threshold value, or determine whether an abnormality exists based on the output of a rule-based discriminator, a statistics-based discriminator, or a machine learning-based anomaly detection model that takes the integrated feature vector as input.

[0129] For example, a smart farm management device can detect electrical abnormal signs appearing in a short period of time using features calculated in a first time scale (e.g., peak of inrush current, harmonic or ripple components, etc.), detect degradation of the actuator's operating performance or changes in mechanical load using features calculated in a second time scale (e.g., cycle time, average current / power, target arrival time or target error, etc.), and detect the cumulative trend of deterioration using features calculated in a third time scale (e.g., moving average of the first / second time scale features, rate of change, long-term trend or health index, etc.).

[0130] The smart farm management device can classify the type of abnormality based on whether an abnormality is detected and infer the cause of the abnormality (250).

[0131] In addition to detecting abnormalities in actuators, the smart farm management device can classify the types of abnormalities detected and infer the root cause of the abnormalities based on the classification results.

[0132] In one embodiment, a smart farm management device can classify fault types based on the temporal change pattern of actuator response data or features extracted therefrom (e.g., current, power, cycle time, current ripple, inrush current, target arrival time, etc.). For example, the fault type can be classified into at least one of (i) abrupt fault, (ii) gradual fault, and (iii) intermittent fault.

[0133] For example, sudden failures may manifest as response characteristics changing rapidly over a short period or exceeding thresholds, and may be associated with sensor open circuits, fuse short circuits, or mechanical damage. Gradual failures may appear as long-term trends in characteristic values ​​changing gradually, and may be associated with bearing wear, reduced lubrication, sensor drift, or insulation degradation. Intermittent failures may appear as abnormal patterns occurring and disappearing irregularly, and may be associated with poor contact, temperature-dependent failures, or electromagnetic interference (EMI).

[0134] In one embodiment, the smart farm management device can infer the root cause by using at least one of rule-based inference or learning-based inference based on a plurality of combination patterns based on the above type classification results.

[0135] For example, if “increase in current” and “increase in cycle time” occur simultaneously, the smart farm management device may infer the possibility of increased mechanical friction or increased load as the root cause. Alternatively, if “increase in current ripple” occurs along with “increase in current,” the smart farm management device may infer the possibility of bearing damage or a malfunction in the rotating part. Alternatively, if “there is almost no current fluctuation but no operation is performed,” the smart farm management device may infer the possibility of mechanical connection failure or power transmission disconnection. Alternatively, if a combination of “decrease in inrush current and maintenance of normal current” is observed, the smart farm management device may infer the possibility of an electrical fault, such as a winding short circuit. Alternatively, if “increase in inrush current and overheating” are observed together, the smart farm management device may infer signs of rotor seizure or overload. Alternatively, if “intermittent current fluctuation” is repeatedly observed, the smart farm management device may infer the possibility of poor contact or electromagnetic interference. Furthermore, if a pattern is observed in which the current fluctuates significantly with temperature changes, the smart farm management device can infer the possibility of mechanical interference due to poor lubrication or thermal expansion.

[0136] The smart farm management device may also suggest maintenance measures based on the inferred cause of the abnormality (260).

[0137] For example, if increased mechanical friction is suspected, removing foreign matter from the rails, checking the lubrication status, and verifying mechanical interference may be suggested. Alternatively, if bearing damage is suspected, bearing inspection or replacement may be suggested. Or, if poor contact is suspected, checking the connector connection status and wiring may be suggested, and if electromagnetic interference is suspected, checking the grounding / shielding status may be suggested. Or, if the possibility of an electrical fault (e.g., winding short circuit) is suspected, checking the winding resistance or insulation status may be suggested.

[0139] FIG. 7 shows one of the embodiments of a smart farm management device (700).

[0140] The smart farm management device (700) may correspond to the smart farm management device (100) described above in FIG. 1. That is, the smart farm management device (700) may be a device that performs the aforementioned smart farm actuator abnormality detection method.

[0141] The smart farm management device (700) may include at least one input device (710), a storage device (720), a computing device (730), an output device (740), an interface device (750), and a communication device (760).

[0142] The input device (710) may receive data, information, or models necessary for performing the aforementioned smart farm actuator anomaly detection method. The input device (710) may receive at least one of a diagnostic schedule, control command, actuator response data, features, actuator anomaly detection results, and time scale information. The input device (710) may receive an analysis model and a correction model. The input device (710) may receive training data necessary for training the analysis model and the correction model. The input device (710) may include a device for inputting certain commands or data (keyboard, mouse and touchscreen, joystick, trackball, touchpad, scanner, webcam, etc.). The input device (710) may include a configuration for receiving data through a separate storage device (USB, CD, hard disk, etc.). The input device (710) may receive data through a separate measuring device or a separate database. The input device (710) may receive data via a wired or wireless connection through a communication device (760). The input device (710) may receive a control signal for controlling the smart farm management device (700).

[0143] The storage device (720) can store data, information, or models necessary for performing the aforementioned smart farm actuator anomaly detection method. The storage device (720) can store at least one of a diagnostic schedule, control commands, actuator response data, features, actuator anomaly detection results, and time scale information. The storage device (720) can store an analysis model and a correction model. The storage device (720) can store training data necessary for training the analysis model and the correction model. The storage device (720) may be a device that stores certain data, information, or models. The storage device (720) can store data, information, and models received through the input device (710). The storage device (720) can store commands that cause the computing device (730) to perform operations necessary for the smart farm actuator anomaly detection method. The storage device (720) can store information generated during the computing process of the computing device (730). That is, the storage device (720) may include memory. For example, storage devices may include HDD (Hard Disk Drive), SSD (Solid State Drive), ROM, RAM, CD-ROM, magnetic tape, or floppy disk.

[0144] The computing device (730) can perform calculations necessary to perform the aforementioned smart farm actuator abnormality detection method. The computing device (730) can perform calculations necessary to transmit control commands to the actuator according to the diagnosis schedule. The computing device (730) can perform calculations necessary to collect actuator response data generated according to the control commands. The computing device (730) can perform calculations necessary to extract multiple features corresponding to different time scales from the actuator response data. The computing device (730) can perform calculations necessary to detect actuator abnormalities based on multiple features corresponding to different time scales. The computing device (730) may be a device such as a processor, an AP (Application Processor), or a chip with an embedded program that processes data and performs certain calculations. For example, the computing device (730) may include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an NPU (Neural Processing Unit). The computing device (730) can generate a control signal to control the smart farm management device (700). The computing device (730) can generate a control signal to control the input device (710), storage device (720), output device (740), interface device (750), and communication device (760) included in the smart farm management device (700).

[0145] The output device (740) may be a device that outputs certain data, information, and models. The output device (740) may be a device that outputs certain data, information, and models outside the smart farm management device (700). The output device (740) may output interfaces, input data, analysis results, etc., necessary for the data processing process. The output device (740) may include a device that outputs data, etc., through tactile, visual, auditory, gustatory, and olfactory methods. The output device (740) may be implemented in various physical forms, such as a display, speaker, vibration motor, or document output device. The output device (740) may output data, information, or models, etc., stored in the storage device (720). The output device (740) may output data, information, and models, etc., generated during the process of calculation by the computing device (730). The output device (740) may output the results calculated by the computing device (730).

[0146] The interface device (750) may be a device that receives certain commands and data from the outside. The interface device (750) may receive control signals for controlling the smart farm management device (700). The interface device (750) may output results analyzed by the smart farm management device (700). The interface device (750) may receive information necessary to perform the aforementioned smart farm actuator abnormality detection method from a physically connected input device or an external storage device.

[0147] The communication device (760) can receive information necessary for performing the aforementioned smart farm actuator anomaly detection method. The communication device (760) can receive a model necessary for performing the aforementioned smart farm actuator anomaly detection method. The communication device (760) can transmit and receive at least one of a diagnostic schedule, a control command, actuator response data, features, actuator anomaly detection results, and time scale information. The communication device (760) can transmit and receive an analysis model and a correction model.

[0148] The communication device (760) can receive control signals necessary to control the smart farm management device (700). The communication device (760) can transmit the results analyzed by the smart farm management device (700). The communication device (760) may refer to a configuration that receives and transmits certain data, information, and models, etc., through a wired or wireless network. The communication device (760) can perform network communication such as Wi-Fi (Wireless Fidelity), Wi-Fi Direct, Bluetooth, UWB (Ultra-Wide Band), NFC (Near Field Communication), USB (Universal Serial Bus), HDMI (High Definition Multimedia Interface), LAN (Local Area Network), etc.

[0150] The aforementioned smart farm actuator anomaly detection method can be implemented as a program (or application) including an executable algorithm that can be executed on a computer.

[0151] The above program may be provided by storing it on a transitory or non-transitory computer-readable medium.

[0152] The above-mentioned temporary readable medium refers to various types of RAM such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synclink DRAM (SLDRAM), and Direct Rambus DRAM (RDRAM).

[0153] The above-mentioned non-transient readable medium refers to a medium that stores data semi-permanently and can be read by a device, rather than a medium that stores data for a short moment, such as a register, cache, or memory. Specifically, the various applications or programs described above may be stored and provided on a non-transient readable medium such as a CD, DVD, hard disk, Blu-ray disc, USB, memory card, ROM (read-only memory), PROM (programmable read-only memory), EPROM (Erasable PROM, EPROM), EEPROM (Electrically EPROM), or flash memory.

[0155] The embodiments and drawings attached to this specification merely clearly illustrate a part of the technical ideas included in the aforementioned technology, and it is self-evident that variations and specific embodiments that can be easily inferred by a person skilled in the art within the scope of the technical ideas included in the specification and drawings of the aforementioned technology are all included within the scope of the rights of the aforementioned technology.

Claims

Claim 1 A step in which a smart farm management device transmits a control command to an actuator according to a diagnostic schedule; a step in which the smart farm management device collects actuator response data generated according to the control command; a step in which the smart farm management device extracts a plurality of features corresponding to different time scales from the actuator response data; and a step in which the smart farm management device detects an abnormality in the actuator based on the plurality of features corresponding to different time scales. ...including, wherein the above feature is a mathematical indicator calculated to detect whether there is an abnormality in the actuator from the actuator response data, and the diagnostic schedule is constructed based on at least one of smart farm environment information, smart farm crop information, smart farm actuator information, and smart farm aging information; constructing the diagnostic schedule based on the smart farm environment information includes determining execution conditions, execution time, and command sequences so that changes in the environment within the smart farm or the impact on the actuator within the smart farm are minimized by the operation of the actuator according to the control command; constructing the diagnostic schedule based on the smart farm crop information includes determining execution conditions, execution time, and command sequences so that the impact on crops within the smart farm is minimized by the operation of the actuator according to the control command; constructing the diagnostic schedule based on the smart farm actuator information includes determining execution conditions, execution time, and command sequences so that the possibility of an accident within the smart farm is minimized by the operation of the actuator according to the control command; and constructing the diagnostic schedule based on the smart farm aging information includes determining execution conditions, execution time, and command sequences so that the impact on the actuator caused by the operation of the actuator according to the control command is minimized by considering the degree of aging of the smart farm actuator. A smart farm actuator anomaly detection method including determining a sequence. Claim 2 A smart farm actuator anomaly detection method according to claim 1, wherein the diagnostic schedule includes at least one execution condition, execution time, and command sequence. Claim 3 delete Claim 4 delete Claim 5 The smart farm actuator anomaly detection method according to claim 1 further comprises: a step in which the smart farm management device collects environmental data at the time when the actuator response data is measured; and a step in which the smart farm management device selects only actuator response data within a preset environmental tolerance range by referring to the environmental data at the time when the actuator response data is measured; wherein the actuator response data used to extract the plurality of features is the selected actuator response data. Claim 6 A smart farm actuator anomaly detection method according to claim 5, wherein the preset environmental tolerance range includes a range for a single environmental physical quantity included in the environmental data or a range for a combination of multiple environmental physical quantities. Claim 7 A smart farm actuator anomaly detection method according to claim 5, wherein selecting only actuator response data within the preset environment allowable range includes selecting only actuator response data in which the similarity between environment vectors, each having multiple environmental physical quantities included in the environment data as components, is within a preset range. Claim 8 A smart farm actuator anomaly detection method according to claim 5, wherein selecting the response data of the actuator further includes selecting only the response data of the actuator according to the same control command. Claim 9 The smart farm actuator anomaly detection method according to claim 1 further comprises: a step in which the smart farm management device collects environmental data at the time when the actuator response data is measured; and a step in which the smart farm management device calculates an environmental influence component from the actuator response data based on the environmental data and the actuator response data, and removes the calculated environmental influence component from the actuator response data; wherein the actuator response data used to extract the plurality of features is the actuator response data from which the environmental influence component has been removed. Claim 10 A smart farm actuator anomaly detection method according to claim 9, wherein removing the above-mentioned environmental impact component from actuator response data includes utilizing a correction model, wherein the correction model estimates the environmental impact component included in the actuator response data using environmental data and actuator response data as inputs, and outputs actuator response data from which the environmental impact component has been removed based on the estimation result. Claim 11 A method for detecting abnormalities in a smart farm actuator according to claim 1, wherein the above features include at least one of average current, peak current, RMS (Root Mean Square), operation completion time, harmonic component, specific frequency band energy, position of spectrum peak, magnitude of spectrum peak, time to reach target, whether target is achieved, whether an error code is generated, whether an interlock is operated, and whether overcurrent is generated. Claim 12 A method for detecting abnormalities in a smart farm actuator according to claim 1, wherein the time scale includes a first time scale, a second time scale, and a third time scale, wherein the second time scale has a time interval larger than the first time scale, and the third time scale has a time interval larger than the second time scale. Claim 13 A smart farm actuator fault detection method according to claim 1, wherein detecting an actuator fault based on a plurality of features corresponding to different time scales includes detecting an actuator fault based on the number of features among the plurality of features corresponding to different time scales that are judged to be outside a normal threshold value. Claim 14 A smart farm actuator fault detection method according to claim 1, wherein detecting an fault in an actuator based on multiple features corresponding to different time scales comprises calculating an fault score corresponding to each of the multiple features corresponding to different time scales, assigning different weights to the fault scores, and determining whether the sum of the fault scores assigned different weights exceeds a preset fault score threshold. Claim 15 A smart farm actuator anomaly detection method according to claim 14, wherein different weights assigned to the above anomaly scores are set differently depending on the type and degree of aging of the actuator. Claim 16 A smart farm actuator fault detection method according to claim 1, wherein detecting an fault in an actuator based on multiple features corresponding to different time scales includes hierarchically analyzing multiple features corresponding to different time scales to detect whether there is a fault. Claim 17 A smart farm actuator fault detection method according to claim 1, further comprising: a step in which the smart farm management device classifies the type of fault based on whether the fault is detected and infers the cause of the fault; and a step in which the smart farm management device proposes a maintenance plan based on the inferred cause of the fault. Claim 18 A storage device comprising a computing device and instructions that cause a smart farm management device to perform operations when executed by said computing device; wherein the operations include: an operation in which the smart farm management device transmits a control command according to a diagnostic schedule to an actuator; an operation in which the smart farm management device collects actuator response data generated according to the control command; an operation in which the smart farm management device extracts a plurality of features corresponding to different time scales from the actuator response data; and an operation in which the smart farm management device detects an abnormality in the actuator based on a plurality of features corresponding to different time scales....including, wherein the above feature is a mathematical indicator calculated to detect whether there is an abnormality in the actuator from the actuator response data, and the diagnostic schedule is constructed based on at least one of smart farm environment information, smart farm crop information, smart farm actuator information, and smart farm aging information; constructing the diagnostic schedule based on the smart farm environment information includes determining execution conditions, execution time, and command sequences so that changes in the environment within the smart farm or the impact on the actuator within the smart farm are minimized by the operation of the actuator according to the control command; constructing the diagnostic schedule based on the smart farm crop information includes determining execution conditions, execution time, and command sequences so that the impact on crops within the smart farm is minimized by the operation of the actuator according to the control command; constructing the diagnostic schedule based on the smart farm actuator information includes determining execution conditions, execution time, and command sequences so that the possibility of accidents within the smart farm is minimized by the operation of the actuator according to the control command; and constructing the diagnostic schedule based on the smart farm aging information includes determining execution conditions, execution time, and command sequences so that the impact on the actuator caused by the operation of the actuator according to the control command is minimized by considering the degree of aging of the smart farm actuator. A smart farm management device including determining a sequence.; Claim 19 A computer program stored on a recording medium to execute the smart farm actuator abnormality detection method described in paragraph 1.

Images