Results of Experimental Research on Computerized Intellectual Monitoring Means of Effective Greenhouse Illumination

The approaches to the study of computerized means of metrical monitoring of the effective greenhouse illumination in the visible optical range, which are used in this research, are based on modern achievements in the theory of experiment planning, the theory of errors and the concept of uncertainty, the theory of physical modeling, and experimental methods of laboratory testing of a prototype measuring system.

To implement the main stages of aggregating and processing the results of observations of the effective greenhouse illumination in the visible optical range, the following modern software have been used:

the software for the microprocessor subsystem has been developed and tested by means of Arduino IDE;

The method of standard devices has been chosen as the basic methodology for carrying out the research on evaluation of the metrological characteristics of measuring channels of effective illumination. This method is one of the most common and thoroughly studied methods of metrological verification and certification of metrical instruments of various physical and chemical quantities. The generalized block diagram of the setup for evaluating the metrological characteristics of the meters under study is shown in Figure 3. The algorithm for conducting laboratory tests is shown in Figure 4. The number of iterations during the research is equal to the number of control points for metrological certification of measuring instruments, namely 14 (Laktionov et al., 2017). The inquiry period of the measuring channels is 1 second, which meets the requirements. The distance from the light source to the surface of sensitive elements of the system is 0.1 m. The measurement conditions are as follows: the temperature is 20 ± 0.5°С, the relative humidity is 70%.

The laboratory setup of an automated greenhouse has been designed and constructed in the laboratory of information-measuring systems of the Department of Electronic Engineering in the SHEE “Donetsk National Technical University”.

Taking into account the basic principles of the theory of physical modeling, the designed laboratory setup meets the conditions of geometric similarity to real objects. The constant of the model-nature linear similarity is equal to: (1) k l = h r e a l h m o d e l = a r e a l a mod e l = b r e a l b m o d e l = 65 , where k_l – linear similarity constant; h_real, a_real, b_real – height, length, and width of the standard all-year greenhouse, correspondingly; h_model, a_model, b_model – height, length, and width of the designed model of the automated greenhouse, correspondingly.

The physical configuration of the laboratory sample of the greenhouse is shown in Figure 5. The physical configuration of the setup for recording the conversion characteristics of the illumination sensors under testing is presented in Figure 6. The setup is equipped with the following technological systems: drip irrigation; ventilation; artificial lighting; air heating and moistening.

The laboratory tests of the greenhouse model allowed us to establish that it meets the mandatory condition of referring the processes in the model and real greenhouses to the same class of phenomena.

To determine the conversion characteristics and the value of the main measurement error, experimental studies of the measuring channel of effective illumination at ambient temperature (20 ± 0.5)°С have been carried out. To exclude transients in the measuring channel of effective illumination (E_v, lx), measurements have been carried out during 120 sec. The illumination value has been measured by the standard Benetech GM1020 luxmeter (Benetech Digital Lux Meter GM1020, 2018). The value change during 100 sec is shown in Figure 7. When analyzing the experimental data, the transient stage is not shown; measurements are taken during the first 20 sec. The E_v measurement has been performed in the range from 4,470 to 96,850 lx, which corresponds to E_e illumination in terms of the energy system (Vovna et al., 2018) from 15 to 325 W·m⁻². Figure 7 shows the graphs of the illumination for the three measurement series: 4,470, 44,700, and 89,400 lx, which correspond to the values E_e: 15, 150, and 300 W·m⁻², respectively. The values of illumination (see Figure 7) are used to evaluate the metrological characteristics of the measuring channel of effective illumination, based on the GY-302 module of the BH1750FVI photodiode (Light intensity Sensor Module GY-30 BH1750FVI, 2018).

Figure 7

Figures 8, 9 and 10 show the results of the illumination changes: 4,470, 44,700, and 89,400 lx, respectively. These values are measured by GY-302 BH1750FVI illumination sensor and Benetech GM1020 luxmeter.

Figure 8

Figure 9

Figure 10

When analyzing the obtained dependences, it has been found out that the results of illumination measuring for the GY-302 BH1750FVI sensor contain both random and systematic components of the error. To reduce the value of the random component of the illumination measurement error, the results of Е ¯ v ref and Е ¯ v meas observations (Matula et al., 2016; Weisstein, 2018) have been averaged: (2) Е ¯ v ref = 1 n ⋅ ∑ i = 1 n Е v ref i ; Е ¯ v meas = 1 n ⋅ ∑ i = 1 n Е v meas i , where E_v ref_i, lx and E_v meas_i, lx – observation results obtained by means of the standard Benetech GM1020 luxmeter and the GY-302 BH1750FVI illumination sensor under investigation.

To eliminate the systematic error, we have measured the illumination of the GY-302 BH1750FVI sensor under study and the standard Benetech GM1020 luxmeter in the range from 4,470 to 96,850 lx, which corresponds to E_e illumination from 15 to 325 W·m^–2. The results of the research are presented in Figure 11, where the illumination measurement results obtained by GY-302 BH1750FVI sensor and standard values obtained by Benetech GM1020 luxmeter have been compared in one coordinate system. In Figure 11, a and b are the nominal and real characteristics when comparing the measurement results of GY-302 BH1750FVI sensor with those of the standard Benetech GM1020 luxmeter.

Figure 11

When analyzing the results of measuring the illumination with GY-302 BH1750FVI sensor and a standard Benetech GM1020 luxmeter, it has been found that there is a systematic error component in the GY-302 BH1750FVI sensor results. This component is multiplicative in nature. To estimate its value, the relative error in illumination measuring has been calculated (Matula et al., 2016; Weisstein, 2018): (3) δ E v = Е ¯ v ref − Е ¯ v meas Е ¯ v ref . 100 , where Е ¯ v ref , lx and Е ¯ v meas , lx – averaged illumination values measured by the Benetech GM1020 luxmeter and the GY-302 BH1750FVI sensor.

The change in the value of the relative error in illumination measuring of the GY-302 BH1750FVI sensor due to its measured value is shown in Figure 12.

Figure 12

When analyzing the dependence of the change in the relative error in illumination measuring on its measured value (see Figure 12), it has been found that in the initial part of the GY-302 BH1750FVI sensor conversion characteristics in the range of illumination measurement from 4,470 to 29,800 lx, the error is from δ_{Ev min} = −7.8% to δ_{Ev max} = −18.8%. When increasing the value of the measured illumination from 29,800 to 96,850 lx, the value of the relative measurement error ranges from –11.9 to –13.8%. Moreover, in the specified range, the relative error is oscillating in vicinity of δ_{Ev system} = −12.7%. This is the value of the systematic error of illumination measuring by the GY-302 BH1750FVI sensor. To eliminate the systematic component of the illumination measurement error, taking into account its multiplicative nature, a linear calibration equation has been used: (4) Е v calibr = k ⋅ Е v real , where E_{v real}, lx – the measured illumination value of the GY-302 BH1750FVI sensor; E_{v calibr}, lx – the illumination value after calibration; k – the calibration coefficient, the value of which is 0.889; this value has been established experimentally while studying the GY-302 BH1750FVI sensor.

To assess the effectiveness of applying the linear calibration Equation (4) to the GY-302 BH1750FVI sensor, the illumination measurement results (see Figure 13) obtained by the GY-302 BH1750FVI sensor during its calibration and those of the Benetech GM1020 luxmeter have been compared graphically.

Figure 13

Having analyzed the graphical dependence (see Figure 13), we have found that the linear calibration Equation (4) almost completely eliminates the systematic component of the measurement error of the GY-302 BH1750FVI illumination sensor. For the quantitative evaluation of the reduction in the value of the systematic error during illumination measuring by means of the calibration Equation (4), the change in the relative error of illumination measuring calculated by the Equation (3) due to its measured value has been established (see Figure 14).

Figure 14

A comparative analysis of the dependence of the change in the relative illumination measurement error of the GY-302 BH1750FVI sensor on its measured value (see Figure 14) has shown that by using the calibration Equation (4), it has become possible to reduce the value of the systematic illumination measurement error by 127 times from δ_{Ev system} = 12.7% to δ_{Rv system} = 0.1% in the range from 29,800 to 96,850 lx. When using the calibration equation in the initial part of the GY-302 BH1750FVI sensor conversion characteristics, the value of the error is reduced by 3.5 times from δ_{Ev max} = 18.8% to δ_{Ev max calibr} = 5.4% in the measurement range from 4,470 to 29,800 lx. Having conducted studies involving the developed recommendations, we will be able to improve the metrological characteristics of the effective illumination meter by using the GY-302 BH1750FVI sensor and its calibration equation, which will allow measuring in the range from 4,470 to 29,800 lx with the relative error of not more than 5.4%. In the range from 29,800 to 96,850 lx it will not be more than 1.0%.

Having analyzed the technical characteristics of VT83N1 photoresistor of KY-018 module, we have found that by using this photoresistor the developed measuring channel of effective illumination will allow measurements in the working range from 10 to 10⁴ lx with a relative error of no more than ±10%. Experimental studies of the measuring channel of effective illumination based on VT83N1 photoresistor of KY-018 module (KY-018 Photoresistor Module, 2018) have been performed at an ambient temperature of (20 ± 0.5)°С. To check the metrological characteristics of the measuring channel, experimental studies have been carried out in the range of illumination measuring from 4,470 to 96,850 lx. The change in VT83N1 photoresistor resistance due to illumination in the indicated range is shown in Figure 15. Benetech GM1020 luxmeter has been used as a standard illumination measuring tool.

Figure 15

Having analyzed the results of experimental studies (see Figure 15), it has been established that it is potentially possible to make measurements of the illumination with VT83N1 photoresistor of the KY-018 module up to 30,000 lx. When illumination increases by more than 30,000 lx, the change in resistance of VT83N1 photoresistor of KY-018 module becomes almost indistinguishable and is oscillating in vicinity of 330 Ohm. For a qualitative evaluation of the error value in the range from 10 to 30,000 lx, the change in resistance of VT83N1 photoresistor of KY-018 module due to the illumination has been approximated by the equation: (5) R = R 0 ⋅ ( exp ( − α R ⋅ E v ) + β R ) , where R – Ohm is the change in resistance of VT83N1 photoresistor of KY-018 module; R₀ – Ohm is the resistance of VT83N1 photoresistor taking into account the conversion characteristic form (β_R), the values of which are determined experimentally with the minimum illumination of the photoresistor and make up R₀ = 528 Ohm and β_R = 0.682; α_R, lx⁻¹ is the damping decrement of the conversion characteristic, the value of which is determined upon approximation and is α_R = 22¹0⁻⁵ lx⁻¹.

Approximated conversion characteristics of VT83N1 photoresistor of KY-018 module due to illumination changes up to 30,000 lx and the results of experimental studies are shown in Figure 16. In the indicated illumination range, the relative error of approximation of the conversion characteristics by the Equation (5) due to illumination changes to 30,000 lx does not exceed 1.5%, which fully meets the requirements for the development of the measuring channel of effective illumination for industrial applications.

Figure 16

To quantify the real measurement range of VT83N1 photoresistor of KY-018 module, the illumination measurement results (see Figure 17) obtained by means of VT83N1 photoresistor of KY-018 module during its calibration and those of the standard Benetech GM1020 luxmeter have been graphically compared.

Figure 17

Having analyzed the characteristics (see Figure 17), it has been found that the results of illumination measurements by means of VT83N1 photoresistor and those of the standard Benetech GM1020 luxmeter correlate with each other up to 20,000 lx. With an increase of illumination from 20,000 to 30,000 lx, there is a significant increase in measurement error. For a quantitative evaluation of the metrological characteristics of the measuring channel of effective illumination, built on the basis of VT83N1 photoresistor (KY-018 Photoresistor Module, 2018), the relative error of illumination measuring is calculated by means of the Equation (3) due to its measured value. This dependence is shown in Figure 18.

Figure 18

When analyzing the dependence of the change in the relative illumination measurement error by VT83N1 photoresistor of KY-018 module on its measured value (see Figure 18), it was found that in the measurement range from 10 to 10,000 lx, which is regulated by the technical characteristics of VT83N1 photoresistor of KY-018 module (KY-018 Photoresistor Module, 2018), the value of the basic error does not exceed ±1.5%. This result is 6.7 times less than the value of the relative measurement error specified in the technical characteristics of VT83N1 photoresistor of KY-018 module, which is not more than ±10%. When extending the range of illumination measuring of VT83N1 photoresistor to 22,000 lx, the value of the relative error increases to ±10%, and with an increase to 30,000 lx it exceeds ±30%. Experimental studies of VT83N1 photoresistor of KY-018 module allow us to establish the real measurement range from 10 to 22,000 lx, in which the relative measurement error does not exceed ±10%.