Predictive System for Rapid Response Characteristics of Drone-Mounted Gas Analysis Algorithm

Abstract

The present invention relates to a prediction system for reducing the response time of a gas analysis system mounted on a drone. The initial concentration change pattern of an NDIR (Non-Dispersive Infrared) gas sensor is modeled as a characteristic function based on the Beer-Lambert law, and a system of equations is established from continuous sampling data to predict the final concentration value. The system characteristic value α is determined as a function of gas molecular weight, temperature, optical waveguide volume, and vent area. Experimental results show that the response time was reduced from the existing 240 seconds to 40 seconds, improving it to approximately one-sixth of the original level. Since the present invention can be implemented using only software algorithms without additional hardware, battery efficiency can be maximized without increasing the drone's payload. This allows for covering six times more measurement points during the same flight time, making it effective for large-scale pollution monitoring.

Description

Technology Field

[0001] An embodiment relates to a prediction system that shortens the response time by analyzing the initial change pattern of gas concentration in a gas analysis system mounted on a drone and predicting the final concentration value. Background Technology

[0002] The use of drones for monitoring air pollutants at inaccessible locations, such as chimneys, has recently been gaining attention. Equipping drones with gas detectors to approach emission points and take direct measurements is more efficient than traditional ground-based measurement methods.

[0003] U.S. Patent No. 8,097,856 (2012, Wong, “Super-miniaturized NDIR gas sensor”) discloses a miniature NDIR gas sensor using a “wave-guide” sample chamber concept. While this patent contributed to the miniaturization of NDIR sensors, it did not provide a specific solution for reducing response time.

[0004] In the study by Liu et al. (2022, "Improvements of a low-cost CO2 commercial nondispersive near-infrared (NDIR) sensor for unmanned aerial vehicle (UAV) atmospheric mapping applications", Atmospheric Measurement Techniques), an NDIR CO2 sensor system for drone mounting was developed, achieving an accuracy of ±2 ppm at 1 Hz. However, this study also pointed out the limitation that the sensor's response time is still slow and emphasized the importance of calibration before measurement.

[0005] The study by Szulczyρski et al. (2023, "The Gas Sensing Drone with the Lowered and Lifted Measurement Platform", Sensors) reported achieving a "very short response time" by optimizing gas flow within the gas chamber. While they improved response time through small cell size and gas flow path optimization, they still required a complex mechanical structure to address measurement errors caused by the drone's downwash effect.

[0006] A recent study by Kang et al. (2025, "Threshold trigger embedded gas sensor for low-power and real-time atmospheric NO2 monitoring using drones", Chemical Engineering Journal) reduced power consumption and extended drone flight time by 10% by activating the sensor only above a certain concentration. However, this has the limitation that continuous monitoring is not possible.

[0007] Conventional technologies have largely relied on hardware improvements (such as adding pumps, reducing sensor size, and modifying mechanical structures), which has led to new problems including increased drone payloads and battery consumption. Therefore, there is a need for a new approach that can shorten the response time of gas sensors without additional hardware under the constraints of the drone environment. Prior art literature

[0008] U.S. Patent No. 8,097,856 (January 17, 2012) U.S. Patent No. 5,444,249 (August 22, 1995) International Published Patent WO2012 / 027257 (March 1, 2012)

[0009] Y. Liu et al., “Improvements of a low-cost CO2 commercial nondispersive near-infrared (NDIR) sensor for unmanned aerial vehicle (UAV) atmospheric mapping applications”, Atmos. Meas. Tech., 15, 4431-4442, 2022B. Szulczyρski et al., “The Gas Sensing Drone with the Lowered and Lifted Measurement Platform”, Sensors, 23(3), 1253, 2023D. Kang et al., "Threshold trigger embedded gas sensor for low-power and real-time atmospheric NO2 monitoring using drones", Chemical Engineering Journal, 2025 The problem to be solved

[0010] The purpose of the embodiment is to provide a method to reduce response time using only software algorithms without additional hardware in a gas analysis system mounted on a drone.

[0011] Another objective of the embodiment is to provide an algorithm that can accurately predict the final concentration value by analyzing the initial concentration change pattern of the NDIR gas sensor.

[0012] Another objective of the embodiment is to provide a gas monitoring system capable of covering more measurement points by maximizing the battery usage efficiency of the drone. means of solving the problem

[0013] To achieve the above objective, a prediction system for rapid response characteristics of a gas analysis system for a drone according to one embodiment of the embodiment includes the following configuration.

[0014] First, an NDIR sensor unit is provided to measure gas concentration using a non-dispersive infrared method. The NDIR sensor unit may include a pulse-driven infrared light source, an optical waveguide acting as a gas chamber, a photodetector that detects infrared rays passing through the optical waveguide, and a signal processing circuit that processes the signal from the photodetector.

[0015] Next, a data acquisition unit is included that samples gas concentration data from the NDIR sensor unit at regular time intervals (βt). The data acquisition unit can sample concentration data by synchronizing with the pulse driving period of the infrared light source.

[0016] In addition, a prediction algorithm processing unit is provided that predicts a final concentration value by applying the sampled initial concentration data to a characteristic function based on the Beer-Lambert law. The prediction algorithm processing unit [describes] the change in gas concentration It is modeled as follows. Here, X(t) is the gas concentration at time t, X f is the final concentration to be predicted, X i is the initial concentration, and t represents the elapsed time since the start of measurement.

[0017] In the above formula, α is a characteristic value of the gas detection system and is determined as a function of the gas molecular weight (M), temperature (T), the volume of the optical waveguide (V) of the NDIR sensor unit, and the ventilation area (A). In particular, the system characteristic value α has a value equal to the time constant 1 / α, which is the time constant for the gas concentration to reach the (1-1 / e) level of the final value.

[0018] The above prediction algorithm processing unit uses two consecutive sampling points t k and t k+1 Concentration value X at k and X k+1 Establish a system of linear equations using this. Through this The final concentration is calculated using the formula. Here, X k is at time t kConcentration measurement at, X k+1 is at time t k+1 The concentration measurement at , Δt represents the sampling time interval.

[0019] Furthermore, it may further include a drone control unit that receives a measurement completion signal from the prediction algorithm processing unit and moves the drone to the next measurement point.

[0020] Meanwhile, a prediction method for the rapid response characteristics of a gas analysis system for a drone according to another embodiment of the embodiment includes the following steps.

[0021] First, a step of measuring gas concentration using an NDIR sensor is performed. Subsequently, a step of sampling concentration data at regular time intervals (Δt) during the initial transient period is carried out.

[0022] Next, the characteristic function based on the Beer-Lambert law A step is performed to analyze the concentration change pattern by applying . Here, X(t) is the gas concentration at time t, X f is the final concentration to be predicted, X i is the initial concentration, and α is a system characteristic value that is a function of gas molecular weight (M), temperature (T), optical waveguide volume (V), and vent area (A). The system characteristic value α can be determined through prior calibration for each type of gas to be measured.

[0023] Subsequently, the concentration value X at consecutive sampling points k and X k+1 Using The final concentration value (X) using the formula f A step to predict ) is performed. Here, X k is at time t k Concentration measurement at, X k+1 is at time t k+1 It refers to the concentration measurement value at.

[0024] Finally, the method may further include the step of terminating the measurement and moving the drone to the next measurement point once the predicted final concentration value stabilizes. Effects of the invention

[0025] The prediction system of the embodiment can reduce the response time for gas concentration measurement to about one-sixth of the existing level. As a result of the experiment, the measurement, which previously took 240 seconds, can be completed in about 40 seconds, thereby significantly improving the battery efficiency of the drone.

[0026] Since the implementation example can be achieved using only software algorithms without additional hardware, measurement efficiency can be improved without increasing the drone's payload.

[0027] The implementation can cover six times more measurement points during the same flight time, enabling efficient pollution mapping of large-scale industrial complexes or entire cities. Brief explanation of the drawing

[0028] FIG. 1 is an overall configuration diagram of a gas analysis system for drones according to one embodiment of the present invention. FIG. 2 is a conceptual diagram showing the configuration of an NDIR gas sensor according to one embodiment of the present invention. FIG. 3 is a system configuration diagram for modeling changes in gas concentration according to one embodiment of the present invention. Figure 4 is a graph showing a gas concentration increase characteristic curve according to one embodiment of the present invention. FIG. 5 is a flowchart of the operation of a prediction algorithm according to one embodiment of the present invention. Figure 6 is a graph comparing the predicted value and the actual value when the concentration changes to 1000 ppm in an experimental example of the present invention. Figure 7 is a graph comparing the predicted value and the actual value when the concentration changes to 2000 ppm in an experimental example of the present invention. Figure 8 is a comparison graph of the predicted value and the actual value when the concentration changes to 3000 ppm in the experimental example of the present invention. Specific details for implementing the invention

[0029] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings so that those skilled in the art can easily implement the present invention. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein. Throughout the specification, similar parts are denoted by the same reference numerals.

[0030] In this specification, the term “connection” should be interpreted to include not only direct connections but also indirect connections through other components. Furthermore, terms such as “comprising” or “having” are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0031] The computation means mentioned in the present invention can be implemented in various forms such as CPU, GPU, microprocessor, ASIC, FPGA, etc., and the memory can be implemented in various storage means such as RAM, ROM, flash memory, SSD, etc.

[0032] The following describes an example of implementation in more detail.

[0033] Referring to FIG. 1, a gas analysis system (100) for drone mounting according to one embodiment of the present invention includes a drone platform (110), an NDIR sensor unit (120), a data acquisition unit (130), a prediction algorithm processing unit (140), and a drone control unit (150).

[0034] The drone platform (110) is a multicopter-type unmanned aircraft and includes a battery (111), a flight controller (112), and a communication module (113). The battery (111) generally provides a flight time of 20-30 minutes, which is the main reason why the fast response algorithm of the present invention is required.

[0035] As illustrated in detail in FIG. 2, the NDIR sensor unit (120) includes an infrared light source (121), an optical waveguide (122), an optical detector (123), and a signal processing circuit (124). The optical waveguide (122) acts as a gas chamber, and the volume (V) and the ventilation area (A) are the main parameters that determine the system characteristic value α.

[0036] The light source (121) operates in a pulse-driven manner and emits infrared light at intervals of, for example, 3 seconds. This pulse-driven method enables periodic data sampling while reducing power consumption. Various forms of light sources can be used, such as MEMS heaters, tungsten filaments, and LEDs.

[0037] Theoretical Background and Mathematical Modeling

[0038] The prediction algorithm of the present invention applies the Beer-Lambert law to changes in concentration within a gas chamber. First, the following physical laws are applied to understand the gas movement mechanism.

[0039] Law of Partial Pressure Accordingly, when chamber 1 and chamber 2 are connected to allow gas movement, even if the total pressure of each chamber is the same, if the partial pressure of the type of gas constituting it is different, the gas moves from the chamber with the higher partial pressure to the chamber with the lower partial pressure.

[0040] Gas State Equation If a single gas present in the chamber is in equilibrium according to [this], the following relationship holds:

[0041] [Mathematical Formula 1]

[0042]

[0043] Here, P is the pressure inside the chamber, V is the volume of the chamber, N is the number of gas molecules, k B ε is the Boltzmann constant, and T is the chamber temperature at equilibrium.

[0044] Since pressure P is proportional to the kinetic energy of the molecules, the following relationship holds in equilibrium:

[0045] [Mathematical Formula 2]

[0046]

[0047] [Mathematical Formula 3]

[0048]

[0049] Here, M is the total mass of the gas molecules, ε is the average velocity of the gas molecules, and ΔP is the pressure difference of the partial gas in chamber 1 and chamber 2.

[0050] Referring to Fig. 3, a system configuration for modeling changes in gas concentration is illustrated. Gas exchange between the inside of the chamber (optical waveguide) and the outside (atmosphere) occurs due to a difference in partial pressure. The change in the partial pressure of gas molecules within the chamber over time is expressed as follows:

[0051] [Mathematical Formula 4]

[0052]

[0053] Here, α is a system characteristic value, which is a function of gas molecular weight (M), temperature (T), optical waveguide volume (V), and vent area (A), as in α = α(M, T, V, A).

[0054] The solution to mathematical equation 4 can be found as follows:

[0055] [Mathematical Formula 5]

[0056]

[0057] Here, P0 = P i - P f is, P iis the initial pressure, P f is the final pressure.

[0058] Under the condition that the chamber volume is constant, using the gas state equation, the number of molecules and density can be expressed as follows:

[0059] [Mathematical Formula 6]

[0060]

[0061] [Mathematical Formula 7]

[0062]

[0063] Here, n(t) is the gas density of the chamber with respect to time, and n0 is the initial (t=0) gas density. Equation 7 is the characteristic function of the pattern of decreasing gas density.

[0064] When gas molecules enter the chamber from the outside, the gas molecule density in the chamber increases in the following pattern:

[0065] [Mathematical Formula 8]

[0066]

[0067] The data acquisition unit (130) samples the concentration data output from the NDIR sensor unit (120) at regular time intervals (Δt). In this embodiment, sampling is performed at 3-second intervals, but this is exemplary and various intervals between 1 second and 10 seconds may be used depending on the sensor characteristics and measurement environment.

[0068] Concentration prediction algorithm

[0069] The prediction algorithm processing unit (140) applies a characteristic function based on the Beer-Lambert law to the sampled data. As shown in FIG. 4, the increase in gas concentration follows an exponential function form.

[0070] If there is no change in chamber volume and no change in pressure, gas density is equivalent to expressing it in terms of concentration, which is the volume ratio of the constituent gases. Therefore, if we transform density n(t) into concentration X(t) and apply it:

[0071] [Mathematical Formula 9]

[0072]

[0073] Here X f is the final concentration, X i is the initial concentration.

[0074] Non-dispersive infrared gas measurement pulses the light source, so it outputs data at each pulse-driven time interval. Therefore, Equation 9 can be expressed in a discrete-time form as follows:

[0075] [Mathematical Formula 10]

[0076]

[0077] , k = 0, 1, 2, ...

[0078] If we let the time interval be Δt, at the next sampling point:

[0079] [Mathematical Formula 11]

[0080]

[0081] The prediction algorithm processing unit (140) above processes two consecutive sampling times t k wa t k+1 Using the concentration value at, establish the system of equations 10 and 11. Solving this system of equations yields the final concentration X f can be obtained as follows:

[0082] [Mathematical Formula 12]

[0083]

[0084] FIG. 5 shows the operation flowchart of the prediction algorithm. First, gas measurement is started (S510), and an initial concentration value is obtained (S520). Then, concentration data is sampled at regular time intervals (S530), and when at least two data points are collected (S540), the final concentration is predicted using Equation 12 (S550). When the predicted concentration value stabilizes (S560), the measurement is terminated and the process moves to the next measurement point (S570).

[0085] Generally, the response time of a gas detector is the time taken to reach 90% of the final concentration, starting from the initial concentration (t 90 ) or the time t to reach the 1-1 / e level (approx. 63.2%) 1 / e It is defined as. In particular, t 1 / e is t in mathematical equation 9 1 / e = equal to the value of 1 / α. This is an intrinsic characteristic value of the system, and in the present invention, this characteristic is used to predict the final value using only the data of the initial transient period.

[0086] When the drone control unit (150) receives a measurement completion signal from the prediction algorithm processing unit (140), it moves the drone to the next measurement point. This maximizes battery usage efficiency and allows more measurement points to be covered.

[0087] The following provides a more detailed explanation through specific embodiments. The following embodiments are merely examples to aid in understanding the present invention and do not limit the scope of the invention.

[0088] [Example 1] CO 2 Gas concentration prediction experiment

[0089] The prediction algorithm of the present invention was verified using the S-300 NDIR CO2 sensor from ELTsensor. The system characteristic value α of the sensor was set to 1 / 80 (average value), and the data sampling interval was set to 3 seconds.

[0090] Figure 6 is a graph comparing the predicted and actual values ​​as the concentration changes from 0 ppm to 1000 ppm. It can be confirmed that the final concentration value predicted at 38 seconds using Equation 12 is almost identical to the actual final value after 240 seconds.

[0091] Figure 7 shows the results when the concentration changes from 0 ppm to 2000 ppm. The predicted value at the 40-second mark showed a high degree of agreement with the actual value after 250 seconds.

[0092] Figure 8 shows the results when the concentration changes from 0 ppm to 3000 ppm. The predicted value at the 42-second mark matched the actual value after 240 seconds exactly.

[0093] [Example 2] Analysis of Response Time Reduction Effect

[0094] The following table compares the response times for each concentration change:

[0095] Concentration change: 0 to 1000 ppm, Actual time to reach: 240 seconds, Predicted time to reach: 38 seconds

[0096] Concentration change: 0 to 2000 ppm, Actual time to reach: 250 seconds, Predicted time to reach: 40 seconds

[0097] Concentration change: 0 to 3000 ppm, Actual time to reach: 240 seconds, Predicted time to reach: 42 seconds

[0098] Average time reduction: Approximately 1 / 6

[0099] As a result, by applying the prediction system of the present invention, the final concentration can be predicted within an average of 40 seconds, thereby achieving a measurement speed improvement of approximately 6 times compared to the existing system.

[0100] [Example 3] Application to various types of gases

[0101] The system of the present invention is applicable to various gases, including not only CO2 but also CH4 (methane), NO2 (nitrogen dioxide), and SO2 (sulfur dioxide). Since the average velocity varies according to Equation 3 based on the molecular weight (M) of each gas, and the system characteristic value α varies accordingly, the α value is calibrated in advance for each type of gas.

[0102] If a gas detector intended for drone mounting is manufactured by enhancing the actual ventilation characteristics of the optical waveguide, which is the gas chamber, it is estimated that the prediction according to Equation 12 can be achieved within a few seconds.

[0103] Although preferred embodiments of the present invention have been described in detail above, the scope of the present invention is not limited thereto, and various modifications and improvements by those skilled in the art using the basic concept of the present invention as defined in the following claims also fall within the scope of the present invention. Explanation of the symbols

[0104] 100: Gas Analysis System 110: Drone Platform 111: Battery 112: Flight controller 113: Communication module 120: NDIR sensor section 121: Infrared light source 122: Optical waveguide 123: Photodetector 124: Signal processing circuit 130: Data Acquisition Unit 140: Prediction Algorithm Processing Unit 150: Drone control unit

Claims

Claim 1 A prediction system for rapid response characteristics of a gas analysis system mounted on a drone comprises: an NDIR sensor unit that measures gas concentration using a non-dispersive infrared method, including an optical waveguide functioning as a gas chamber; a data acquisition unit that samples gas concentration data from the NDIR sensor unit at a constant time interval (△t) during an initial transient period before the output of the NDIR sensor unit reaches a steady state; and a prediction algorithm processing unit that predicts a final concentration value before reaching a steady state by applying the sampled initial concentration data to a first-order transient response characteristic function based on the Beer-Lambert law, wherein the prediction algorithm processing unit [describes] a change in gas concentration Modeled as follows, where X(t) is the gas concentration at time t, Xf is the final concentration to be predicted, Xi is the initial concentration, t is the elapsed time since the start of measurement, α is a system characteristic value that is a function of the gas molecular weight (M), temperature (T), the optical waveguide volume (V) and the vent area (A) of the NDIR sensor unit, and 1 / α has a value equal to the time constant at which the gas concentration reaches the level of the final value (1-1 / e); and two consecutive sampling times t k and t k+1 Concentration value X at k and X k+1 Using The final concentration value (X) using the formula f Calculate ), where X k is at time t k Concentration measurement at, X k+1 is at time t k+1 A system comprising a prediction algorithm characterized by further including a drone control unit that receives a measurement completion signal from the prediction algorithm processing unit and moves the drone to the next measurement point, wherein △t is a sampling time interval and a concentration measurement value at the above-mentioned point. Claim 2 delete Claim 3 delete Claim 4 A prediction method for rapid response characteristics of a gas analysis system for a drone, comprising: a step of measuring gas concentration using an NDIR sensor including an optical waveguide functioning as a gas chamber; a step of sampling concentration data at a fixed time interval (△t) during an initial transient period; and a first-order transient response characteristic function based on the Beer-Lambert law. A step of analyzing the concentration change pattern by applying, where X(t) is the gas concentration at time t, X f is the final concentration to be predicted, X i is the initial concentration, α is a system characteristic value that is a function of gas molecular weight (M), temperature (T), optical waveguide volume (V), and vent area (A), and 1 / α has a value equal to the time constant at which the gas concentration reaches the (1-1 / e) level of the final value; and the concentration value X at consecutive sampling points k and X k+1 Using The final concentration value (X) before reaching steady state using the formula f Includes a step of predicting ), where X k is at time t k Concentration measurement at, X k+1 is at time t k+1 A prediction method characterized by being a concentration measurement value at; further comprising the step of terminating the measurement and moving the drone to the next measurement point when the predicted final concentration value stabilizes. Claim 5 delete Claim 6 delete

Images