Rapid method for determining precise fertilization of cigars

Abstract

The invention relates to a rapid method for determining precise fertilization of cigars. Comprising the following steps: 1) collecting spectral data of a cigar canopy in a cigar field by using a portable hyperspectrometer, 2) detecting soil nutrient indexes such as soil available nitrogen, available phosphorus, available potassium and pH in situ in the field by using an electrochemical soil nutrient sensor to obtain real-time detection data of soil nutrients; 3) acquiring meteorological data of the target plot and growth period information of the cigars, wherein the growth period information is divided based on the number of days after the cigars are transplanted; 4) preprocessing and standardizing the canopy spectral parameters, the soil nutrient real-time detection data, the meteorological data and the growth period information to form multi-source fusion input feature data; 5) inputting the multi-source fusion input feature data into a fertilization decision model based on fuzzy logic, and outputting a fertilization formula and a fertilization amount of the cigar in the current growth period; and (6) executing corresponding fertilization operation according to the fertilization formula and the fertilization dosage, regularly repeating the steps (1)-(5) after fertilization, and carrying out rolling correction on a fertilization strategy to realize precise fertilization management of cigars. The fertilizer utilization rate can be remarkably increased, the soil salinization risk caused by excessive fertilization is reduced, the proportion of first-class tobacco is increased, the oil content and the chromaticity of the eggplant coating are improved, economic benefits and ecological benefits are both considered, and the laboratory detection cost and the labor cost are reduced.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of tobacco cultivation, and particularly relates to a method for rapidly determining precise fertilization of cigar tobacco based on combination of canopy spectral analysis of tobacco plants and real-time detection of soil nutrients. BACKGROUND

[0002] Cigar tobacco is sensitive to nutrient demand, and the precision of fertilization directly affects the quality of the jacket and core. The existing technology mainly has the following deficiencies: Dependence on fixed fertilization formula: such as the pre-set fertilizer ratio (such as calcium-magnesium-phosphorus fertilizer, micro-fertilizer, etc.) in CN117296539A patent, but without considering the dynamic factors of real-time illumination, soil moisture, etc. in the field, which is easy to cause "over-fertilization" or "under-fertilization", resulting in mismatch between the amount of fertilization and the actual demand of tobacco plants.

[0003] Traditional soil detection needs long laboratory detection period, which takes several days to several weeks, and cannot guide the fertilization opportunity. The existing precise fertilization equipment (such as the infrared sensor fertilizer machine described in CN102799129B) is mostly based on mechanical control of quantitative or proportional control, which can only mechanically control the amount of fertilization, lacks the "diagnosis + decision" function based on the nutritional status of tobacco plants and the soil fertilization capacity, and cannot dynamically adjust the formula according to the nutritional status of tobacco plants.

[0004] The accumulation of dry matter and the absorption of nitrogen, phosphorus and potassium of cigar tobacco show an "S-shaped curve", and the demand for fertilizer in the rapid growth period of 50-60 days after transplanting accounts for 40% of the total amount, but the traditional method is difficult to capture the demand at this key period in time. SUMMARY

[0005] The purpose of the present application is to provide a rapid method for determining precise fertilization of cigar tobacco, which can complete "detection-diagnosis-decision" in a short time, solve the problems of static traditional cigar tobacco fertilization scheme, long soil detection period, lack of real-time feedback and dynamic decision-making, make the fertilization formula and amount reflect the actual nutritional status of tobacco plants, and take into account the soil fertilization potential and weather conditions, and realize dynamic, precise and efficient fertilization.

[0006] The present application provides a rapid method for determining precise fertilization of cigar tobacco, which mainly includes the following core steps: 1) Collecting canopy spectral data of cigar tobacco in the field by using a portable hyperspectral instrument, the wavelength range of the hyperspectral instrument is 400-2500 nm, calculating spectral indices such as normalized difference vegetation index NDVI and red edge position REP according to the collected spectral data, to obtain canopy spectral parameters representing the nutritional status of tobacco plants; 2) Detecting soil available nitrogen, available phosphorus, available potassium and pH, etc. in the field in situ by using an electrochemical soil nutrient sensor, to obtain real-time detection data of soil nutrients; 3) Obtain meteorological data of the target land and growth period information of the cigar tobacco, the growth period information being divided based on days after the cigar tobacco is transplanted; 4) Preprocess and standardize the canopy spectral parameters, soil nutrient real-time detection data, meteorological data, and growth period information to form multi-source fusion input feature data; 5) Input the multi-source fusion input feature data into a fertilization decision model based on fuzzy logic, and output a fertilization formula and a fertilization amount for the current growth period of the cigar tobacco according to preset fuzzy membership functions and fuzzy rules; 6) Perform corresponding fertilization operation according to the fertilization formula and the fertilization amount, and periodically repeat steps 1) to 5) after fertilization to roll over the fertilization strategy to achieve precise fertilization management of the cigar tobacco.

[0007] Further, in step 1), the normalized vegetation index NDVI is calculated according to the following formula: wherein R NIR is the reflectivity of the near-infrared band, R Red is the reflectivity of the red light band, when NDVI>0.85, it is determined that the nitrogen of the tobacco plant is sufficient, and when NDVI<0.80, it is determined that there is a risk of nitrogen deficiency in the tobacco plant and fertilization adjustment is triggered.

[0008] Further, in step 2), the detection period of the electrochemical soil nutrient sensor is not more than 10 minutes, and the soil nutrient real-time detection data at least includes one of the following: soil available nitrogen content, soil available phosphorus content, soil available potassium content, soil pH value, and soil water content.

[0009] Further, in step 3), the growth period information includes three stages of the seedling recovery period, the rapid growth period, and the mature period, wherein the rapid growth period is 30 to 60 days after transplanting, and the weight of nitrogen and potassium nutrition in the fertilization decision model in the rapid growth period is higher than that in other stages.

[0010] Further, in step 5), the fertilization decision model adopts one or more of the following fuzzy rules: If NDVI is "low", soil available potassium is "low", and the growth period is "rapid growth period", the potassium fertilizer amount in this fertilization is "high"; If NDVI is "moderate", soil available nitrogen is "medium" level, and recent rainfall is "less", the nitrogen fertilizer amount in this fertilization is "medium"; If NDVI is "high" and soil available nitrogen is "high", reduce or delay the use of nitrogen fertilizer.

[0011] Further, in step 5), when the fuzzy inference result indicates that potassium fertilizer topdressing is needed, the potassium fertilizer application amount (Q_K) is calculated according to the following formula: Qκ =(D 需求 -D ±壤 ) ×0.2 Wherein, D 需求 ; is the theoretical potassium requirement in the current growth period, D 土壤 is the available potassium amount converted from the measured soil available potassium, and 0.2 is the conversion coefficient calibrated from field positioning test. The unit of QK is kg / acre.

[0012] Further, the fertilization formula comprises nitrogen fertilizer, phosphorus fertilizer, potassium fertilizer and at least one of zinc, boron and the like, and the fertilization decision model increases the recommended proportion of trace elements in the later growth stage of cigar tobacco to improve the aroma and oil of the tobacco.

[0013] Further, in the multi-source fusion input feature data, different types of features are respectively given different weights according to the growth period, and the weight of the canopy spectral feature and the soil available potassium index in the rapid growth period is higher than that of the meteorological factor.

[0014] Further, the fertilization decision model and the multi-source data processing module are integrated in a mobile terminal or a field control terminal, and the mobile terminal is configured with a display interface for real-time display of the fertilization formula and the fertilization amount suggestion.

[0015] Further, the method is applied in batches in multiple plots or blocks in the same tobacco field, and the differentiated fertilization scheme of each block is determined respectively according to the differences of canopy spectrum and soil nutrients at the block level, and the fertilization strategy of each block is adjusted according to the new monitoring results after fertilization.

[0016] The beneficial effects of the present application compared with the prior art are as follows: 1. Strong real-time performance: the spectrum and soil nutrient detection are completed on site in the field, combined with fuzzy logic decision, and the fertilization suggestion is given within 1 hour, which greatly shortens the decision cycle compared with laboratory detection.

[0017] 2. Dynamic adaptation: automatically adjust the fertilization strategy according to different growth stages of cigar tobacco, focus on nitrogen and potassium supply in the rapid growth period, and strengthen the regulation of trace elements in the later period to improve the aroma and oil.

[0018] 3. Multi-source data fusion: comprehensive actual nutrient state of tobacco plants and soil fertility supply capacity, avoid misjudgment caused by single data source, and improve the reliability of fertilization decision.

[0019] 4. Quality improvement and efficiency enhancement: It can significantly improve the utilization rate of fertilizers (increase the utilization rate of fertilizers by 28% compared with traditional methods), reduce the risk of soil salinization caused by excessive fertilization; test results show that the proportion of superior tobacco in the cigar tobacco using this method increases by 15%, the proportion of superior tobacco increases, the oil content and color of the tobacco jacket improve, the economic and ecological benefits are balanced, the laboratory testing cost and labor cost are reduced, and it is suitable for large-scale tobacco field management. DETAILED DESCRIPTION

[0020] The following examples are used to illustrate the technical solutions of the present application and do not constitute a limitation on the protection scope defined by the claims. In actual applications, equivalent substitutions or modifications can be made within the spirit of the present application.

[0021] 1. Test area and equipment configuration 1) Select a typical test field located in a main cigar tobacco producing area, the soil type is red soil or yellow soil, and the previous crop is tobacco or non-legume crop.

[0022] 2) Configure the following in each test plot: A set of portable hyperspectral instruments with a wavelength range of 400-2500 nm and an optimal spectral resolution of 1-5 nm; collect cigar canopy spectral data, extract red edge position (REP), normalized vegetation index NDVI, and other spectral indices; establish a relationship model between spectral indices and chlorophyll content, nitrogen nutrition status, for example: NDVI>0.85 is determined as sufficient nitrogen, and NDVI<0.8 triggers a nitrogen or potassium deficiency warning Several soil electrochemical nutrient sensors are used to measure soil available nitrogen, available phosphorus, available potassium, and pH, available nitrogen, available phosphorus, and available potassium content; the single detection time is controlled within 10 minutes, which replaces laboratory testing, and soil moisture and temperature sensors can be selected to assist in evaluating fertilizer effectiveness.

[0023] An integrated collection terminal is used to receive sensor output signals and interact with fertilizer decision-making software, collect local weather data including temperature, rainfall, and light duration; determine the growth period of cigar tobacco (such as the slow recovery period, rapid growth period, and mature period) according to the days after transplanting or stage investigation records; standardize canopy spectral indicators, soil nutrient data, weather data, and growth period information to construct a multi-source data feature set.

[0024] 2. Canopy spectral collection and processing 1) Collect canopy spectral data at 10:00-14:00 on a representative sunny day during different growth periods (such as 10 d, 20 d, 30 d, and 45 d) after cigar tobacco transplanting.

[0025] 2) During the collection, the hyperspectral probe is aimed at the top of the tobacco canopy or at a certain angle of inclination, keeping the same observation height and angle to avoid direct sunlight and shadow interference.

[0026] 3) In each plot, no less than 10 tobacco plants are selected for repeated measurement, and the average spectrum is taken as the canopy spectrum of the plot.

[0027] 4) The red and near-infrared reflectance is extracted from the original spectrum using spectral analysis software, NDVI is calculated, and the red edge position REP is determined using the first derivative spectrum method.

[0028] 5) According to the relationship model between NDVI and chlorophyll content, nitrogen nutrition index established in the early stage of calibration, the NDVI value is converted into nitrogen nutrition status grade, such as "low", "medium", "high".

[0029] Input: canopy spectral indicators (NDVI, REP, etc.), soil available N / P / K content, meteorological factors, growth period; Language variables such as "nitrogen deficiency / moderate / excessive" and "soil available potassium low / medium / high" are described by membership functions; Pre-set "if-then" type fuzzy rules, such as: "if NDVI is low and soil available potassium is low and growth period is rapid growth period, then increase potassium fertilizer application rate"; Model output: recommended fertilization formula (N, P, K and trace element ratio) and fertilization amount.

[0030] 3. Real-time detection of soil nutrients 1) On the day of canopy spectrum collection, 3-5 points are taken inside each test plot in a cluster pattern, and an electrochemical soil nutrient sensor is inserted into the 0-20 cm soil layer for in-situ measurement.

[0031] 2) The sensor outputs soil available nitrogen, available phosphorus, available potassium concentration and pH value, and the single-point detection time is controlled within 5-10 min.

[0032] 3) The average of each measurement point data is taken as the real-time detection result of soil nutrients in the plot, and it is divided into "low", "medium", "high" three grades according to the local soil fertility classification standard.

[0033] 4. Acquisition of meteorological data and growth period information 1) Obtain the meteorological data such as air temperature, rainfall, sunshine hours on the day of measurement and the previous period through the nearest automatic weather station or online platform.

[0034] 2) Determine the transplanting date according to the cultivation record of cigar tobacco, calculate the number of days from the determination date to the transplanting date, and determine the current growth period type according to the pre-divided growth stages (such as 0-10 d for the slow recovery period, 10-45 d for the rapid growth period, and 45 d or later for the mature period).

[0035] 5. Multi-source data preprocessing and fusion 1) Normalize the characteristics of NDVI, REP, soil available N / P / K content, pH, air temperature, rainfall, and growth period to unify the numerical range of each characteristic. 2) Use principal component analysis or weighted standardization method to reduce the redundancy between characteristics and improve the calculation efficiency of the model. 3) Set different feature weights according to the growth period, for example, increase the weight of NDVI and soil available potassium in the rapid growth period, and increase the weight of soil available phosphorus and trace elements in the mature period.

[0036] The normalized vegetation index NDVI is calculated according to the following formula: where R NIR is the reflectivity of the near-infrared band, R Red is the reflectivity of the red light band, and when NDVI>0.85, it is determined that the nitrogen of the tobacco plant is sufficient, and when NDVI<0.80, it is determined that there is a risk of nitrogen deficiency in the tobacco plant and the fertilizer adjustment is triggered.

[0037] 6. Fertilizer decision based on fuzzy logic 1) Design fuzzy membership functions for NDVI, soil available nitrogen, available phosphorus, available potassium, and rainfall variables, respectively, and map them to "low", "medium", "high" and other language variables.

[0038] 2) According to the nutrient diagnosis theory of cigar tobacco and the results of field tests, a number of "if-then" type fuzzy rules are constructed, for example: Rule 1: If NDVI is "low" and soil available nitrogen is "low", then the nitrogen fertilizer application rate is "high"; Rule 2: If NDVI is "medium" and soil available phosphorus is "low", then the phosphorus fertilizer application rate is "medium-high"; Rule 3: If NDVI is "low" and soil available potassium is "low" and the growth period is "rapid growth period", then the potassium fertilizer application rate is "high"; Rule 4: If NDVI is "high" and soil available nitrogen is "high", then the nitrogen fertilizer application rate is "low" or suspended.

[0039] 3) Use conventional fuzzy reasoning and defuzzification methods to convert fuzzy output into specific nitrogen, phosphorus, and potassium usage recommendations, in kg / acre.

[0040] 7. Fertilizer amount calculation and fertilizer implementation 1) For the plot determined to need potassium fertilizer topdressing, the potassium fertilizer application amount (Q_K) is calculated according to the following formula: Q_K = (D_K - D_S) x 0.2 需求 ±壤 wherein D_K is the theoretical potassium requirement of the current growth period, D_S is the soil measured available potassium, and 0.2 is the conversion coefficient calibrated by field positioning test. The unit of Q_K is kg / acre. 需求 土壤

[0041] 2) The application amounts of nitrogen fertilizer and phosphorus fertilizer can be calculated according to the difference between the respective theoretical fertilizer requirement and the soil measured content by using similar formulae, and matched with the fuzzy decision result.3) The fertilizer formula and amount are displayed to the field management personnel through an integrated terminal or transmitted to an intelligent fertilizer machine through a wireless network to realize strip application, hole application or drip irrigation fertilization of the corresponding plot.4) The whole process from on-site detection of spectrum and soil data to completion of fertilizer recommendation calculation is controlled within 1 hour to ensure the timeliness of the decision.

[0042] 8. Rolling correction and closed-loop management The decision result is displayed through a terminal or transmitted wirelessly to a fertilization device or a human operator; the process from "on-site detection - data analysis - fertilization decision - implementation" is completed within 1 hour; the fertilization strategy is rolling corrected according to the subsequent monitoring results (new spectrum and soil data), forming a closed-loop management.

[0043] Specifically, after completing a fertilization, crown layer spectrum and soil nutrient on-site detection are performed again after 7-15 days, and the new monitoring data are input into the fertilization decision model to compare with the previous decision effect; if the NDVI and nutrient status reach the expected target, the topdressing intensity is maintained or slowed down, otherwise the next fertilization formula and amount are adjusted according to the new model output, thereby forming a closed-loop precision fertilization management based on multi-source data.

[0044] Example 2: Fertilization decision example under potassium deficiency condition in rapid growth period This example illustrates how the method of the present application quickly gives a targeted fertilization scheme when potassium deficiency occurs in the rapid growth period of cigar tobacco.

[0045] 1) About 45 days after transplanting, crown layer spectrum and soil detection are performed on a certain test plot, obtaining NDVI = 0.75 and red edge position REP moving obviously to the shortwave direction; the soil available potassium content is 140 mg / kg, and the available nitrogen and available phosphorus are at the "medium" level.

[0046] ​​2) According to the NDVI and nitrogen nutrition model, the nitrogen of the tobacco plant is in a "medium-low" state, and the soil potassium content is "low"; according to the growth period information, the plot is in the mature period.

[0047] 3) The fertilization decision model calls rule 3: "if NDVI is 'low' and soil available potassium is 'low' and growth period is 'fast mature period', then potassium fertilizer application amount is'medium'", and combines the theoretical potassium requirement curve and soil available potassium to calculate the potassium fertilizer application amount: QK = (D 需求 -D ±壤 ) x 0.2 Assuming that the current theoretical potassium requirement D 需求 is 80 kg / acre, the soil available potassium D 土壤 is converted to 40 kg / acre, then Q k = (80-40) x 0.2 = 8 kg / acre 4) At the same time, according to the fact that NDVI is only slightly lower than the ideal level, the model only recommends a small amount of nitrogen fertilizer, and the nitrogen fertilizer application amount is set to the "medium" level range.

[0048] 5) Through the terminal, the manager is output with the suggestion of "applying 8 kg / acre of potassium fertilizer and X kg / acre of nitrogen fertilizer to the plot at one time, and re-measuring NDVI and soil nutrients after 10-14 days", realizing rapid response.

[0049] Example 3: Popularization and application in large-area cigar tobacco planting base This embodiment illustrates the integration and popularization method of the method of the application in a large-area cigar tobacco planting base.

[0050] 1) Fixed or semi-fixed soil sensor arrays are arranged in several typical plots, and mobile hyperspectral measurement equipment is configured; 2) The canopy spectral data, soil nutrient data, and meteorological data of each plot are uploaded in real time to the base management platform through a wireless network; 3) The multi-source data fusion and fuzzy logic fertilization decision software of the application is deployed on the management platform, and batch calculations are performed on different plots to give differentiated fertilization schemes; 4) The fertilization prescription map of each plot is issued to intelligent fertilization machinery to realize prescription fertilization; 5) Through annual comparative tests, it is shown that the plot applying the method of the application has higher fertilizer utilization rate, significantly reduced excessive fertilization, significantly improved proportion of high-quality tobacco, and increased economic benefits compared with the traditional fertilization system.

Claims

1. A rapid method for determining precise fertilization of cigars, characterized in that, Includes the following steps: 1) A portable hyperspectral instrument was used to collect spectral data of the cigar tobacco canopy in the cigar tobacco field. The wavelength range of the hyperspectral instrument was 400-2500 nm. The normalized vegetation index (NDVI) and red edge position (REP) and other spectral indices were calculated based on the collected spectral data to obtain canopy spectral parameters that characterize the nutritional status of tobacco plants. 2) An electrochemical soil nutrient sensor was used to detect soil nutrient indicators such as available nitrogen, available phosphorus, available potassium and pH in situ in the field to obtain real-time soil nutrient data. 3) Obtain meteorological data and cigar growing season information for the target plot, wherein the growing season information is divided based on the number of days after the cigars are transplanted; 4) The canopy spectral parameters, real-time soil nutrient monitoring data, meteorological data, and growth period information are preprocessed and standardized to form multi-source fusion input feature data; 5) Input the multi-source fusion input feature data into the fertilization decision model based on fuzzy logic. The fertilization decision model outputs the fertilization formula and fertilization amount for the current growth stage of cigars according to the preset fuzzy membership function and fuzzy rules. 6) Perform the corresponding fertilization operation according to the fertilization formula and dosage, and repeat steps 1) to 5) periodically after fertilization to make rolling corrections to the fertilization strategy in order to achieve precise fertilization management of cigars.

2. The rapid method according to claim 1, characterized in that, In step 1), the Normalized Difference Vegetation Index (NDVI) is calculated according to the following formula: Among them, R NIR R represents the reflectance in the near-infrared band. Red The reflectance is measured in the red light band. When NDVI > 0.85, the tobacco plants are considered to have sufficient nitrogen. When NDVI < 0.80, the tobacco plants are considered to be at risk of nitrogen deficiency, triggering fertilizer adjustments.

3. The rapid method according to claim 1 or 2, characterized in that, In step 2), the electrochemical soil nutrient sensor shall be used for a single detection cycle of no more than 10 minutes. The real-time soil nutrient detection data shall include at least one of the following: available nitrogen content, available phosphorus content, available potassium content, soil pH value, and soil moisture content.

4. The rapid method according to any one of claims 1 to 3, characterized in that, In step 3), the growth period information includes three stages: the seedling establishment period, the rapid growth period, and the maturity period. The rapid growth period is 30 to 60 days after transplanting. During the rapid growth period, the fertilization decision model gives higher weights to nitrogen and potassium nutrition than to other stages.

5. The rapid method according to any one of claims 1 to 4, characterized in that, In step 5), the fertilization decision model employs one or more of the following fuzzy rules: If NDVI is "low" and available potassium in the soil is "low" and the growth period is "rapid growth period", then the amount of potassium fertilizer used in this fertilization should be "high". If the NDVI is "moderate", the available nitrogen in the soil is "medium", and the recent rainfall is "slightly low", then the amount of nitrogen fertilizer used in this fertilization should be "medium". If NDVI is "high" and available nitrogen in the soil is "high", then reduce or postpone the application of nitrogen fertilizer.

6. The rapid method according to any one of claims 1 to 5, characterized in that, In step 5), when the fuzzy inference result indicates that potassium fertilizer topdressing is required, the potassium fertilizer application rate (Q_K) is calculated according to the following formula: Qκ =(D 需求 -D ±壤 ) ×0.2 Among them, D 需求 ; represents the theoretical potassium requirement for the current reproductive period, D 土壤 QK represents the available potassium quantity calculated from the measured available potassium in the soil. 0.2 is the conversion factor determined by field location experiments. The unit of QK is kg / mu.

7. The rapid method according to any one of claims 1 to 6, characterized in that, The fertilizer formula includes nitrogen fertilizer, phosphorus fertilizer, potassium fertilizer, and at least one trace element from zinc and boron. The fertilization decision model increases the recommended proportion of trace elements during the later growth stage of cigars to improve the aroma and oil content of the wrapper.

8. The rapid method according to any one of claims 1 to 7, characterized in that, In the multi-source fusion input features, different weights are assigned to various features according to the growth period. During the rapid growth period, the weights of canopy spectral features and soil available potassium index are higher than those of meteorological factors.

9. The rapid method according to any one of claims 1 to 8, characterized in that, The fertilization decision model and multi-source data processing module are integrated into a mobile terminal or field control terminal. The mobile terminal is equipped with a display interface for real-time display of fertilization formula and fertilization amount recommendations.

10. The rapid method according to any one of claims 1 to 9, characterized in that, This method is applied in batches across multiple plots or plots within the same tobacco field. By analyzing the differences in canopy spectroscopy and soil nutrients at the plot level, differentiated fertilization schemes are determined for each plot. After fertilization, the fertilization strategies for each plot are adjusted based on the results of a new round of monitoring.

Images