Development Of Pepper Precision Iirrigation System Based On Spectral Detection Technology

Precision irrigation is a control method with the goal of achieving precise and accurate management of agricultural irrigation water.Most of irrigation control used in agriculture in my country is based on the substrate moisture content.Traditional irrigation decision-making method has a certain water-saving effect,but control strategies combined with crop physiological needs can further improve the utilization rate of agricultural water resources.The degree of water stress directly affects photosynthetic rate of crops,and chlorophyll fluorescence parameter F_v/F_m of crops can reflect its photosynthetic potential and characterize the water demand.Spectral technology can realize the online measurement of chlorophyll fluorescence parameters.Therefore,the research on precise irrigation of peppers in greenhouse based on spectral technology is of great significance for the exploration of irrigation control methods that combine crop demand.In this paper,inadequacies of disadvantages of substrate moisture content as irrigation control target,on the basis of analyzing the variation law of pepper F_v/F_m under drought stress,characteristic bands highly correlated with F_v/F_m were screened by spectral analysis,and a detection device was developed.Irrigation decision-making model based on pepper growth needs,build a greenhouse pepper irrigation system,and explore precise irrigation control methods that combine crop growth needs.The main research contents and conclusions of this paper are as follows:(1)The characteristic wavelength extraction of F_v/F_m and selection of typical leaf regions were completed.Taking pepper as the test object,F_v/F_m images and hyperspectral images under water stress were obtained,and F_v/F_m image denoising was completed with median filter algorithm of 3×3 filter window.The matching of the hyperspectral image and F_v/F_m image is realized by finding edge feature points and coordinate transformation.The overall matching root mean square error is 0.51 pixels,and 73.5%of pixels on leaf edge are error-free.After preprocessing the spectral data with the SG algorithm,the SPA algorithm was used to extract characteristic bands and screen out 23 characteristic wavelengths.According to correlation coefficient of each characteristic wavelength,its contribution to the model was evaluated,and the changes in modeling accuracy of 4 to 10 different number of characteristic wavelength combinations with correlation coefficients from high to low under the PLSR algorithm were analyzed.Finally,412.0,670.2,692.9,737.1,762.7 and 993.0 nm were selected,a total of 6 characteristic wavelengths.By analyzing the F_v/F_m histogram,it is concluded that distribution of F_v/F_m values in the central area of the blade tends to be consistent,and this area can represent entire blade.(2)Developed a F_v/F_m detection device based on characteristic wavelengths.According to above research results,six narrow-band LEDs with central wavelengths of 412.5,670.0,690.0,735.0,760.0 and 975.0 nm were selected as the light source of the detection device.The hardware of detection device includes:a light source drive circuit,a reflected light intensity detection circuit,a wireless communication circuit,a microcontroller circuit and a housing structure,and a hardware driver program is written to control each characteristic wavelength light source to light up in turn and measure its reflected light intensity,and then pass the NB-Io T module uploads reflected light intensity to the server.The stability test of testing equipment was completed,test time was 42 h,and measurement error was less than0.1%.The dark current of detection equipment was measured,and dark current corresponding to the detection voltage was distributed in range of 0.00075～0.00081 V within the 20 h test time,indicating that detection equipment had good stability and low dark current noise,which had no effect on normal measurement.The BP-LM algorithm was used to establish the F_v/F_m prediction model of detection equipment,and model verification was completed.The R~2 was 0.878,RMSE was 0.019,and maximum error was 1.03%.(3)The irrigation decision-making method based on physiological needs is constructed.According to the characteristics and causes of water stress and high and low temperature stress,a method to discriminate drought stress in pepper by combining air temperature and substrate humidity was proposed.Based on experimental results of pepper recovery from drought stress,it is concluded that when F_v/F_m value of pepper is higher than 0.790,when F_v/F_m value is between 0.790～0.700,it can be recovered,and when F_v/F_m value is lower than0.700,the plant is close to death.To sum up,under premise of excluding the effects of low temperature and high temperature stress based on the lowest ambient temperature at night and highest temperature in afternoon,irrigation was carried out when water content of pepper substrate was lower than 10%and F_v/F_m test result was lower than 0.790.(4)Completed construction and verification of greenhouse pepper irrigation system based on spectral technology.According to needs of the irrigation system,a solenoid valve controller including a 4G communication circuit,a solenoid valve control circuit,a power supply circuit,a single-chip microcomputer and a sensor interface circuit was designed,a hardware driver was developed and its power consumption was tested,and low power consumption was completed real-time communication and solenoid valve control.Using the existing server platform,the data access,F_v/F_m prediction model call of detection device,and call of the irrigation decision model were realized,and a mobile APP was developed to realize human-computer interaction function of system.The water-saving performance of system was evaluated by total water consumption,plant height,stem diameter,fresh weight,dry weight and number of leaves in verification test.The test results show that irrigation method of pepper in greenhouse based on spectrum technology does not affect normal growth of pepper.Compared with irrigation method based on substrate moisture content,water saving performance is better,and water saving rate is 12.3%.