A multimodal large model-based agricultural sentiment perception auxiliary system and method

Abstract

The present invention discloses an agricultural sentiment perception assistance system and method based on a multimodal large model, which relates to the field of agricultural sentiment perception assistance technology. The present invention includes the following methods: S10: screening out abnormal areas in a target area, and determining the processing order of each abnormal area based on the real-time agricultural risk index of each abnormal area; S20: analyzing the correlation between each abnormal area based on the distribution characteristic coefficient corresponding to each abnormal cause in the abnormal area; S30: determining the agricultural situation management strategy for each abnormal area; S40: predicting the agricultural situation management time for each abnormal area. The present invention predicts the distribution characteristic coefficient corresponding to each abnormal cause in each abnormal area, realizes accurate analysis of the cause of abnormal crop growth, and realizes reasonable distribution of the management time of crops planted in the abnormal area, which is conducive to effective management of the abnormal area in a short time.

Description

Technical Field

[0001] The present invention relates to the field of agricultural sentiment perception assistance technology, and in particular to an agricultural sentiment perception assistance system and method based on a multimodal large model. Background Art

[0002] Agricultural performance monitoring utilizes field monitoring technology to provide real-time insights into crop growth, helping agricultural producers make timely decisions and improve crop yields and quality. With technological advancements, these monitoring methods are evolving from traditional manual sampling and measurement to automated, intelligent, and rapid monitoring, improving monitoring efficiency.

[0003] When sensing agricultural conditions, existing technologies usually determine whether crops are growing normally based on monitoring data output by monitoring equipment. However, they are unable to accurately analyze the causes of abnormal crop growth and require manual analysis of crop growth conditions, which reduces work efficiency. At the same time, existing technologies are unable to quickly determine crop management strategies or reasonably distribute crop management time. Summary of the Invention

[0004] The purpose of the present invention is to provide an agricultural sentiment perception assistance system and method based on a multimodal large model to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for assisting agricultural sentiment perception based on a multimodal large model, the method comprising:

[0006] S10: Several sets of agricultural monitoring equipment are evenly arranged in the target area. Abnormal areas in the target area are screened out based on the monitoring data output by each set of agricultural monitoring equipment arranged in the target area. The processing order of each abnormal area is determined based on the real-time agricultural risk index of each abnormal area.

[0007] S20: analyzing the correlation between the abnormal areas according to the distribution characteristic coefficients corresponding to the abnormal causes in the abnormal areas;

[0008] S30: adjusting the processing order of each abnormal area according to the correlation between the abnormal areas, and determining the agricultural situation management strategy for each abnormal area based on the adjustment result;

[0009] S40: Predicting the agricultural condition management time of each abnormal area, and performing agricultural condition management processing on each abnormal area according to the predicted management time.

[0010] Furthermore, the S10 includes:

[0011] S101: Divide the target area by a distance d to obtain a plurality of square areas with a side length of d, wherein a set of agricultural monitoring equipment is arranged in each square area, wherein the set of agricultural monitoring equipment includes a soil sensor, a meteorological monitoring sensor, a seedling condition detector, and an insect pest detection and reporting instrument. Monitoring data output by each set of agricultural monitoring equipment arranged in the target area is collected. Based on the collected monitoring data, the monitoring data includes the moisture content of the soil surface layer and the soil deep layer at a distance of X meters from the soil surface monitored by the soil sensor, the temperature and humidity monitored by the meteorological monitoring sensor, the plant height and leaf area of ​​the crops monitored by the seedling condition detector, and the number of pests monitored by the insect pest detection and reporting instrument. A three-dimensional simulation of the growth conditions and growth environment of the crops planted in the target area is performed in real time to obtain a real-time agricultural condition perception three-dimensional model of the target area, where X represents a constant and 0<X<the average burial depth of the crops planted in the target area.

[0012] S102: Calculate the abnormal index of each divided area based on the real-time agricultural sentiment perception three-dimensional model of the target area. The specific calculation formula is:

[0013] F i =a*(g i / G i )+b*(s i / S i );

[0014] Where i = 1, 2, ..., m, represents the number of each square area in the target area, m represents the total number of square areas in the target area, a and b represent the proportional coefficient and a + b = 1, s i represents the total area of ​​abnormal leaves of crops monitored by the seedling condition detectors arranged in the i-th square area, S i G represents the total leaf area of ​​crops monitored by the seedling condition detectors arranged in the i-th square area, i represents the total number of crops planted in the i-th square area, g i F represents the total number of abnormal crops in the i-th square area, i represents the abnormal index of the ith square area;

[0015] If 0.4<F i ≤1, the ith square area is considered to be an abnormal area. If 0≤F i ≤0.4, the i-th square area is considered not to be an abnormal area;

[0016] S103: Predicting the real-time agricultural risk index of each abnormal area screened out. The specific prediction formula is: , where j = 1, 2, ..., n, represents the number corresponding to each abnormal area, n represents the total number of abnormal areas screened out, t represents the time value, y jt It represents the number of pests detected by the insect monitoring device in the abnormal monitoring area numbered j at time t, e represents a constant and e>1, v=1,2,3,4, when v=1, w vjt It represents the moisture content of the soil surface detected by the soil sensor in the abnormal monitoring area numbered j at time t. When v=2, w vjt It represents the moisture content of the deep soil layer X meters away from the soil surface detected by the soil sensor arranged in the abnormal monitoring area numbered j at time t. When v=3, 4, w vjt They represent the temperature and humidity monitored by the meteorological monitoring sensor in the abnormal monitoring area numbered j at time t. If w vjt If it can meet the growth needs of crops, , if w vjt If it cannot meet the growth needs of crops, , H jt represents the agricultural risk index of the abnormal area numbered j at time t;

[0017] The processing order of each abnormal area is determined according to the order of the real-time agricultural risk index of each abnormal area from large to small.

[0018] Furthermore, the S20 includes:

[0019] S201: collecting monitoring data output by agricultural monitoring equipment arranged in each abnormal area within the time period [t0, t];

[0020] At time t, according to Calculate the distribution characteristic coefficient corresponding to the abnormal cause of pest infestation in the abnormal area numbered j, where F jt represents the abnormal index of the abnormal area numbered j at time t, t0 represents the most recent start-up time of the agricultural monitoring equipment from time t, and k1 represents the relationship coefficient between the number of pests and the degree of damage to crops planted in the square area obtained based on experimental data;

[0021] At time t, according to Calculate the distribution characteristic coefficient corresponding to the abnormal cause of humidity abnormality in the abnormal area numbered j;

[0022] Among them, k2 represents the relationship coefficient between humidity and the degree of damage to crops planted in the square area obtained based on experimental data. If 0.5*(w´ 1jt +w´ 2jt )≤0.5*(w1jt +w 2jt )≤0.5*(w´´ 1jt +w´´ 2jt ),but , if 0.5*(w 1jt +w 2jt )>0.5*(w´´ 1jt +w´´ 2jt ) or 0.5*(w 1jt +w 2jt )<0.5*(w´ 1jt +w´ 2jt ),but or ,w´ 1jt 、w´´ 1jt They represent the lower limit of soil surface moisture content and the upper limit of soil surface moisture content temperature corresponding to the normal growth of crops planted in the abnormal area numbered j at time t, w´ 2jt 、w´´ 2jt They represent the lower limit of soil moisture content at a distance of X meters from the soil surface and the upper limit of soil moisture content at a distance of X meters from the soil surface when the crops planted in the abnormal area numbered j grow normally at time t, respectively. p1=1 or p1=-1. When 0.5*(w 1jt +w 2jt )>0.5*(w´´ 1jt +w´´ 2jt ), p1=-1, when 0.5*(w 1jt +w 2jt )<0.5*(w´ 1jt +w´ 2jt ), p1=1;

[0023] At time t, according to The distribution characteristic coefficient corresponding to the abnormal cause of temperature abnormality in the abnormal area numbered j is calculated, where k3 represents the relationship coefficient between the temperature and the degree of damage to crops planted in the square area obtained based on experimental data, T jt It represents the temperature value monitored by the meteorological monitoring sensor in the abnormal area numbered j at time t. If T´ jt ≤T jt ≤T´´ jt ,but , if T jt >T´´ jt or T jt <T´ jt ,but or , T´ jt 、T´´ jtThey represent the lower and upper temperature limits of the crops grown in the abnormal area numbered j at time t when they grow normally, p2=1 or p2=-1. When T jt >T´´ jt When p2=-1, when T jt <T´ jt When p2=1;

[0024] S202: Classify the abnormal area according to the distribution characteristic coefficient corresponding to each abnormal cause in the abnormal area. The specific classification processing method is as follows:

[0025] The abnormal area corresponding to the maximum value of the real-time agricultural risk index is taken as the target abnormal area, and the number of the selected abnormal area is c, c = 1, 2, ..., n and c ≠ j, where Q 1ct , Q 2ct , Q 3ct All are not equal to 0;

[0026] When Q 1jt , Q 2jt , Q 3jt When both are not equal to 0, , L jct The correlation between the abnormal area numbered j and the abnormal area numbered c at time t is calculated, q = 1, 2, 3. At this time, number j is put into set A to obtain the first abnormal area subset;

[0027] When Q 1jt , Q 2jt , Q 3jt There is a 0, , now put number j into set A;

[0028] When Q 1jt , Q 2jt , Q 3jt When there are two 0s, , now put number j into set A;

[0029] When Q 1jt , Q 2jt , Q 3jt When both are equal to 0, At this time, it is considered that the abnormal cause of the crops planted in the abnormal area numbered j is abnormal fertilization, and the number j is put into set A.

[0030] Furthermore, the S30 includes:

[0031] S301: For an abnormal region stored in the first abnormal region subset, if the condition 0.7<L is satisfied jct ≤1, then 0.7<L jct≤1 corresponding to the abnormal area according to L jct The emission order from large to small is placed after the target abnormal area, and the target abnormal area and 0.7<L jct ≤1 corresponding abnormal area is deleted from the abnormal area screened out in S102, and the target abnormal area and 0.7<L jct The abnormal areas corresponding to ≤1 are regarded as the first echelon of agricultural management;

[0032] S302: Repeat the operations of S103, S202, and S301 to determine the second agricultural situation management echelon, the third agricultural situation management echelon, and so on, until the abnormal area screened out in S102 cannot be used as the target abnormal area;

[0033] For abnormal areas that cannot be used as target abnormal areas, agricultural situation management will be carried out in descending order of real-time agricultural situation risk index, and abnormal areas that cannot be used as target abnormal areas will be used as target agricultural situation management echelons;

[0034] S303: Carry out agricultural situation management on the first agricultural situation management echelon, the second agricultural situation management echelon..., and the target agricultural situation management echelon in turn.

[0035] Furthermore, the S40 includes:

[0036] S401: For the first agricultural condition management echelon, the pest control time W1, humidity control time W2, and temperature control time W3 of the target abnormal area are collected;

[0037] The pest control time M of the abnormal area numbered j stored in the first agricultural situation management echelon 1j =W1*(Q 1jt / Q 1ct ), when M 1j When ≥W1, the pest control time D of the abnormal area numbered j stored in the first agricultural situation management echelon 1j = Pest control time G1 in target abnormal area, when M 1j When W1 is less than D 1j =G1+[W1-W1*(Q 1jt / Q 1ct )];

[0038] The humidity management duration M of the abnormal area numbered j stored in the first agricultural management echelon 2j =W2*(Q 2jt / Q 2ct ), when M 2j When ≥W2, the humidity management time D of the abnormal area numbered j stored in the first agricultural management echelon 2j = Humidity control time G2 of target abnormal area, when M 2jWhen <W2, D 2j =G2+[W2- W2*(Q 2jt / Q 2ct )];

[0039] The temperature management duration M of the abnormal area numbered j stored in the first agricultural management echelon 3j =W3*(Q 3jt / Q 3ct ), when M 3j When ≥W3, the temperature management time D of the abnormal area numbered j stored in the first agricultural situation management echelon 3j = Humidity control time G3 of target abnormal area, when M 3j When <W3, D 3j =G3+[W3- W3*(Q 3jt / Q 3ct )];

[0040] S402: Based on the operation of S401, the management time of each abnormal cause is predicted for each stored abnormal area for the second agricultural management echelon, the third agricultural management echelon, etc.;

[0041] S403: Collect the agricultural risk index of each abnormal area stored in the target agricultural risk management echelon at time t, and record the abnormal area corresponding to the maximum value of the collected agricultural risk index as γ, γ = 1, 2, ..., n, and the fertilization management time D of the abnormal area numbered j stored in the target agricultural risk management echelon 4j =G γ +[W γ -W γ *(H j / H γ ), G γ Indicates the fertilization management time of the abnormal area numbered γ, W γ Indicates the duration of fertilization treatment in the abnormal area numbered γ;

[0042] S404: Perform agricultural condition management on each abnormal area according to the predicted management time.

[0043] An agricultural situation perception assistance system based on a multimodal large model, the system includes an agricultural situation risk index prediction module, a correlation analysis module, an agricultural situation management strategy determination module, and an agricultural situation management time prediction module;

[0044] The agricultural risk index prediction module is used to predict the real-time agricultural risk index of each abnormal area and determine the processing order of each abnormal area based on the prediction results;

[0045] The correlation analysis module is used to analyze the correlation between the abnormal areas according to the distribution characteristic coefficients corresponding to the abnormal causes in the abnormal areas;

[0046] The agricultural situation management strategy determination module is used to adjust the processing order of each abnormal area according to the correlation between the abnormal areas, and determine the agricultural situation management strategy for each abnormal area based on the adjustment result;

[0047] The agricultural condition management time prediction module is used to predict the agricultural condition management time of each abnormal area, and perform agricultural condition management processing on each abnormal area according to the predicted management time.

[0048] Furthermore, the agricultural risk index prediction module includes an agricultural risk perception three-dimensional model acquisition unit, an abnormality index calculation unit, an agricultural risk index prediction unit, and an abnormal area processing order determination unit;

[0049] The agricultural condition perception three-dimensional model acquisition unit performs a three-dimensional simulation of the growth conditions and growth environment of crops planted in the target area in real time based on the monitoring data output by the agricultural condition monitoring equipment arranged in each divided area, and obtains a real-time agricultural condition perception three-dimensional model of the target area;

[0050] The abnormality index calculation unit constructs a calculation formula to calculate the abnormality index of each divided area based on the real-time agricultural sentiment perception three-dimensional model of the target area;

[0051] The agricultural risk index prediction unit predicts the real-time agricultural risk index of each abnormal area selected according to the constructed prediction model;

[0052] The abnormal area processing sequence determining unit determines the processing sequence of each abnormal area according to the order of the real-time agricultural risk index of each abnormal area from large to small.

[0053] Furthermore, the correlation analysis module includes a distribution characteristic coefficient calculation unit and a correlation analysis unit;

[0054] The distribution characteristic coefficient calculation unit calculates the distribution characteristic coefficient corresponding to each abnormal cause in each abnormal area based on the monitoring data output by the agricultural monitoring equipment arranged in each abnormal area within a random period of time;

[0055] The correlation analysis unit classifies the abnormal regions according to the distribution characteristic coefficients corresponding to the abnormal causes in the abnormal regions, and calculates the correlation between the abnormal regions based on the classification results.

[0056] Furthermore, the agricultural situation management strategy determination module includes an agricultural situation management echelon determination unit and an agricultural situation management unit;

[0057] The agricultural situation management echelon determination unit determines whether the abnormal areas stored in the first abnormal area subset meet the set conditions, and determines the first agricultural situation management echelon, the second agricultural situation management echelon, etc., and the target agricultural situation management echelon according to the judgment result;

[0058] The agricultural situation management unit conducts agricultural situation management on the first agricultural situation management echelon, the second agricultural situation management echelon..., and the target agricultural situation management echelon in turn.

[0059] Furthermore, the agricultural condition management time prediction module includes a management time prediction unit and a fertilization management time prediction unit;

[0060] The control time prediction unit predicts the control time for each abnormal area stored in each agricultural condition management echelon according to the pest control time, humidity control time and temperature control time of the target abnormal area in each agricultural condition management echelon, and performs agricultural condition management processing on each abnormal area according to the predicted control time;

[0061] The fertilization management time prediction unit collects the real-time agricultural risk index of each abnormal area stored in the target agricultural management echelon, predicts the fertilization management time of each abnormal area stored in the target agricultural management echelon based on the collected information, and performs agricultural management on each abnormal area according to the predicted management time.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] 1. The present invention uses a real-time three-dimensional agricultural condition perception model of the target area to screen out abnormal areas in the target area. Based on the monitoring data output by agricultural condition monitoring equipment arranged in each abnormal area within a random period of time, the distribution characteristic coefficient corresponding to each abnormal cause in each abnormal area is predicted, thereby achieving accurate analysis of the causes of abnormal crop growth. This process does not require human participation, further improving analysis efficiency.

[0064] 2. The present invention classifies abnormal areas according to the correlation between them, which is conducive to ensuring that abnormal areas within each agricultural situation management echelon are applicable to the same agricultural situation management strategy, and is conducive to the rapid determination of agricultural situation management strategies for abnormal areas.

[0065] 3. The present invention predicts the control time for various abnormal causes in each abnormal area stored in each agricultural condition management echelon based on the pest control time, humidity control time and temperature control time in the target abnormal area in each agricultural condition management echelon, and predicts the fertilization control time for each abnormal area stored in the target agricultural condition management echelon based on the real-time agricultural condition risk index of each abnormal area stored in the target agricultural condition management echelon, thereby realizing a reasonable distribution of the control time for crops planted in the abnormal area, which is conducive to effective control of the abnormal area in a short time. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 This is a schematic diagram of the workflow of an agricultural sentiment perception assistance system and method based on a multimodal large model of the present invention;

[0067] Figure 2 The present invention is a schematic structural diagram of the working principle of an agricultural sentiment perception auxiliary system and method based on a multimodal large model. DETAILED DESCRIPTION

[0068] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0069] Example: Figure 1 and Figure 2 As shown, the present invention provides a system and method for assisting agricultural sentiment perception based on a multimodal large model, and a method for assisting agricultural sentiment perception based on a multimodal large model, the method comprising:

[0070] S10: Several sets of agricultural monitoring equipment are evenly arranged in the target area. Abnormal areas in the target area are screened out based on the monitoring data output by each set of agricultural monitoring equipment arranged in the target area. The processing order of each abnormal area is determined based on the real-time agricultural risk index of each abnormal area.

[0071] The S10 includes:

[0072] S101: Divide the target area by distance d to obtain a plurality of square areas with a side length of d. Each square area is provided with a set of agricultural monitoring equipment, the set of agricultural monitoring equipment including a soil sensor, a meteorological monitoring sensor, a seedling condition detector (the seedling condition detector monitors the growth of crops in the target area via a camera), and an insect pest detection and reporting instrument. Monitoring data output by each set of agricultural monitoring equipment arranged in the target area is collected. Based on the collected monitoring data, including the moisture content of the soil surface and the soil deep layer X meters from the soil surface as monitored by the soil sensor, the temperature and humidity as monitored by the meteorological monitoring sensor, the plant height and leaf area of ​​the crops as monitored by the seedling condition detector, and the number of pests as monitored by the insect pest detection and reporting instrument, a real-time three-dimensional simulation of the growth condition and growth environment of the crops planted in the target area is performed to obtain a real-time agricultural condition perception three-dimensional model of the target area, where X represents a constant and 0<X<the average burial depth of the crops planted in the target area.

[0073] S102: Calculate the abnormal index of each divided area based on the real-time agricultural sentiment perception three-dimensional model of the target area. The specific calculation formula is:

[0074] F i =a*(g i / G i )+b*(s i / S i );

[0075] Where i = 1, 2, ..., m, represents the number of each square area in the target area, m represents the total number of square areas in the target area, a and b represent the proportional coefficient and a + b = 1, s i represents the total area of ​​abnormal leaves of crops monitored by the seedling condition detectors arranged in the i-th square area (abnormal leaves refer to yellow leaves or broken leaves), S i G represents the total leaf area of ​​crops monitored by the seedling condition detectors arranged in the i-th square area, i represents the total number of crops planted in the i-th square area, g i represents the total number of abnormal crops in the i-th square area (abnormal crops refer to crops with yellowed and damaged leaves, crops with wilted and drooping leaves, and crops with growth heights lower than the average growth height), F i represents the abnormal index of the ith square area;

[0076] If 0.4<F i ≤1, the ith square area is considered to be an abnormal area. If 0≤F i ≤0.4, the i-th square area is considered not to be an abnormal area;

[0077] S103: Predicting the real-time agricultural risk index of each abnormal area screened out. The specific prediction formula is: , where j = 1, 2, ..., n, represents the number corresponding to each abnormal area, n represents the total number of abnormal areas screened out, t represents the time value, y jt It represents the number of pests detected by the insect monitoring device in the abnormal monitoring area numbered j at time t, e represents a constant and e>1, v=1,2,3,4, when v=1, w vjt It represents the moisture content of the soil surface detected by the soil sensor in the abnormal monitoring area numbered j at time t. When v=2, w vjt It represents the moisture content of the deep soil layer X meters away from the soil surface detected by the soil sensor arranged in the abnormal monitoring area numbered j at time t. When v=3, 4, w vjt They represent the temperature and humidity monitored by the meteorological monitoring sensor in the abnormal monitoring area numbered j at time t. If w vjt If it can meet the growth needs of crops, , if w vjt If it cannot meet the growth needs of crops, , H jt represents the agricultural risk index of the abnormal area numbered j at time t;

[0078] The processing order of each abnormal area is determined according to the order of the real-time agricultural risk index of each abnormal area from large to small;

[0079] S20: analyzing the correlation between the abnormal areas according to the distribution characteristic coefficients corresponding to the abnormal causes in the abnormal areas;

[0080] The S20 includes:

[0081] S201: collecting monitoring data output by agricultural monitoring equipment arranged in each abnormal area within the time period [t0, t];

[0082] At time t, according to Calculate the distribution characteristic coefficient corresponding to the abnormal cause of pest infestation in the abnormal area numbered j, where F jt represents the abnormal index of the abnormal area numbered j at time t, t0 represents the most recent start-up time of the agricultural monitoring equipment from time t, and k1 represents the relationship coefficient between the number of pests and the degree of damage to crops planted in the square area obtained based on experimental data;

[0083] At time t, according to Calculate the distribution characteristic coefficient corresponding to the abnormal cause of humidity abnormality in the abnormal area numbered j;

[0084] Among them, k2 represents the relationship coefficient between humidity and the degree of damage to crops planted in the square area obtained based on experimental data. If 0.5*(w´ 1jt +w´ 2jt )≤0.5*(w 1jt +w 2jt )≤0.5*(w´´ 1jt +w´´ 2jt ),but , if 0.5*(w 1jt +w 2jt )>0.5*(w´´ 1jt +w´´ 2jt ) or 0.5*(w 1jt +w 2jt )<0.5*(w´ 1jt +w´ 2jt ),but or ,w´ 1jt 、w´´ 1jt They represent the lower limit of soil surface moisture content and the upper limit of soil surface moisture content temperature corresponding to the normal growth of crops planted in the abnormal area numbered j at time t, w´ 2jt 、w´´ 2jt They represent the lower limit of soil moisture content at a distance of X meters from the soil surface and the upper limit of soil moisture content at a distance of X meters from the soil surface when the crops planted in the abnormal area numbered j grow normally at time t, respectively. p1=1 or p1=-1. When 0.5*(w 1jt +w 2jt )>0.5*(w´´ 1jt +w´´ 2jt ), p1=-1, when 0.5*(w 1jt +w 2jt )<0.5*(w´ 1jt +w´ 2jt ), p1=1;

[0085] At time t, according to The distribution characteristic coefficient corresponding to the abnormal cause of temperature abnormality in the abnormal area numbered j is calculated, where k3 represents the relationship coefficient between the temperature and the degree of damage to crops planted in the square area obtained based on experimental data, T jt It represents the temperature value monitored by the meteorological monitoring sensor in the abnormal area numbered j at time t. If T´ jt ≤T jt ≤T´´ jt ,but , if T jt>T´´ jt or T jt <T´ jt ,but or , T´ jt 、T´´ jt They represent the lower and upper temperature limits of the crops grown in the abnormal area numbered j at time t when they grow normally, p2=1 or p2=-1. When T jt >T´´ jt When p2=-1, when T jt <T´ jt When p2=1;

[0086] S202: Classify the abnormal area according to the distribution characteristic coefficient corresponding to each abnormal cause in the abnormal area. The specific classification processing method is as follows:

[0087] The abnormal area corresponding to the maximum value of the real-time agricultural risk index is taken as the target abnormal area, and the number of the selected abnormal area is c, c = 1, 2, ..., n and c ≠ j, where Q 1ct , Q 2ct , Q 3ct All are not equal to 0;

[0088] When Q 1jt , Q 2jt , Q 3jt When both are not equal to 0, , L jct The correlation between the abnormal area numbered j and the abnormal area numbered c at time t is calculated, q = 1, 2, 3. At this time, number j is put into set A to obtain the first abnormal area subset;

[0089] When Q 1jt , Q 2jt , Q 3jt There is a 0, , now put number j into set A;

[0090] When Q 1jt , Q 2jt , Q 3jt When there are two 0s, , now put number j into set A;

[0091] When Q 1jt , Q 2jt , Q 3jt When both are equal to 0, , at this time, it is considered that the abnormal cause of the crops planted in the abnormal area numbered j is abnormal fertilization, and the number j is put into the set A;

[0092] S30: adjusting the processing order of each abnormal area according to the correlation between the abnormal areas, and determining the agricultural situation management strategy for each abnormal area based on the adjustment result;

[0093] The S30 includes:

[0094] S301: For an abnormal region stored in the first abnormal region subset, if the condition 0.7<L is satisfied jct ≤1, then 0.7<L jct ≤1 corresponding to the abnormal area according to L jct The emission order from large to small is placed after the target abnormal area, and the target abnormal area and 0.7<L jct ≤1 corresponding abnormal area is deleted from the abnormal area screened out in S102, and the target abnormal area and 0.7<L jct The abnormal areas corresponding to ≤1 are regarded as the first echelon of agricultural management;

[0095] S302: Repeat the operations of S103, S202, and S301 to determine the second agricultural situation management echelon, the third agricultural situation management echelon, and so on, until the abnormal area screened out in S102 cannot be used as the target abnormal area;

[0096] For abnormal areas that cannot be used as target abnormal areas, agricultural situation management will be carried out in descending order of real-time agricultural situation risk index, and abnormal areas that cannot be used as target abnormal areas will be used as target agricultural situation management echelons;

[0097] S303: Conducting agricultural situation management on the first agricultural situation management echelon, the second agricultural situation management echelon, etc., and the target agricultural situation management echelon in sequence;

[0098] S40: predicting the agricultural condition management time of each abnormal area, and performing agricultural condition management processing on each abnormal area according to the predicted management time;

[0099] S40 includes:

[0100] S401: For the first agricultural condition management echelon, the pest control time W1, humidity control time W2, and temperature control time W3 of the target abnormal area are collected;

[0101] The pest control time M of the abnormal area numbered j stored in the first agricultural situation management echelon 1j =W1*(Q 1jt / Q 1ct ), when M 1j When ≥W1, the pest control time D of the abnormal area numbered j stored in the first agricultural situation management echelon 1j = Pest control time G1 in target abnormal area, when M 1j When W1 is less than D1j =G1+[W1-W1*(Q 1jt / Q 1ct )], pest control time G1 of the target abnormal area = the time corresponding to the abnormality index of the target abnormal area being 0.3;

[0102] The humidity management duration M of the abnormal area numbered j stored in the first agricultural management echelon 2j =W2*(Q 2jt / Q 2ct ), when M 2j When ≥W2, the humidity management time D of the abnormal area numbered j stored in the first agricultural management echelon 2j = Humidity control time G2 of target abnormal area, when M 2j When <W2, D 2j =G2+[W2- W2*(Q 2jt / Q 2ct )], the humidity control time G2 of the target abnormal area = the time corresponding to when the humidity monitored by the meteorological monitoring sensor arranged in the target abnormal area is 30% and the moisture content of the soil surface monitored by the soil sensor arranged in the target abnormal area is less than 30%, or the humidity control time G2 of the target abnormal area = the time corresponding to when the moisture content of the deep soil layer at a distance of X meters from the soil surface monitored by the soil sensor arranged in the target abnormal area is greater than 60%;

[0103] The temperature management duration M of the abnormal area numbered j stored in the first agricultural management echelon 3j =W3*(Q 3jt / Q 3ct ), when M 3j When ≥W3, the temperature management time D of the abnormal area numbered j stored in the first agricultural situation management echelon 3j = Humidity control time G3 of target abnormal area, when M 3j When <W3, D 3j =G3+[W3- W3*(Q 3jt / Q 3ct )], the temperature control time G2 of the target abnormal area = the time corresponding to when the temperature monitored by the meteorological monitoring sensor arranged in the target abnormal area is 30°C, or the temperature control time G2 of the target abnormal area = the time corresponding to when the temperature monitored by the meteorological monitoring sensor arranged in the target abnormal area is 3°C;

[0104] S402: Based on the operation of S401, the management time of each abnormal cause is predicted for each stored abnormal area for the second agricultural management echelon, the third agricultural management echelon, etc.;

[0105] S403: Collect the agricultural risk index of each abnormal area stored in the target agricultural risk management echelon at time t, and record the abnormal area corresponding to the maximum value of the collected agricultural risk index as γ, γ = 1, 2, ..., n, and the fertilization management time D of the abnormal area numbered j stored in the target agricultural risk management echelon 4j =G γ +[W γ -W γ *(H j / H γ ), G γ Indicates the fertilization management time of the abnormal area numbered γ, W γ Indicates the duration of fertilization treatment in the abnormal area numbered γ;

[0106] S404: Perform agricultural condition management on each abnormal area according to the predicted management time. The specific method of agricultural condition management is the existing technology. For example, when performing humidity management, if the humidity in the abnormal area is too low, humidity management is achieved by watering the crops planted in the abnormal area.

[0107] An agricultural situation perception assistance system based on a multimodal large model, the system includes an agricultural situation risk index prediction module, a correlation analysis module, an agricultural situation management strategy determination module, and an agricultural situation management time prediction module;

[0108] The agricultural risk index prediction module is used to predict the real-time agricultural risk index of each abnormal area and determine the processing order of each abnormal area based on the prediction results;

[0109] The agricultural risk index prediction module includes an agricultural risk perception three-dimensional model acquisition unit, an abnormality index calculation unit, an agricultural risk index prediction unit, and an abnormal area processing order determination unit;

[0110] The agricultural condition perception 3D model acquisition unit performs a 3D simulation of the growth conditions and growth environment of crops planted in the target area in real time based on the monitoring data output by the agricultural condition monitoring equipment arranged in each divided area, and obtains a real-time agricultural condition perception 3D model of the target area;

[0111] The abnormal index calculation unit constructs a calculation formula to calculate the abnormal index of each divided area based on the real-time agricultural sentiment perception three-dimensional model of the target area;

[0112] The agricultural risk index prediction unit predicts the real-time agricultural risk index of each abnormal area selected based on the constructed prediction model;

[0113] The abnormal area processing order determination unit determines the processing order of each abnormal area according to the order of the real-time agricultural risk index of each abnormal area from large to small;

[0114] The correlation analysis module is used to analyze the correlation between abnormal areas based on the distribution characteristic coefficients corresponding to the abnormal causes in the abnormal areas;

[0115] The correlation analysis module includes a distribution characteristic coefficient calculation unit and a correlation analysis unit;

[0116] The distribution characteristic coefficient calculation unit calculates the distribution characteristic coefficient corresponding to each abnormal cause in each abnormal area based on the monitoring data output by the agricultural monitoring equipment arranged in each abnormal area within a random period of time;

[0117] The correlation analysis unit classifies the abnormal areas according to the distribution characteristic coefficients corresponding to the abnormal causes in the abnormal areas, and calculates the correlation between the abnormal areas based on the classification results;

[0118] The agricultural situation management strategy determination module is used to adjust the processing order of each abnormal area according to the correlation between the abnormal areas, and determine the agricultural situation management strategy for each abnormal area based on the adjustment results;

[0119] The agricultural situation management strategy determination module includes the agricultural situation management echelon determination unit and the agricultural situation management unit;

[0120] The agricultural situation management echelon determination unit determines whether the abnormal area stored in the first abnormal area subset meets the set conditions, and determines the first agricultural situation management echelon, the second agricultural situation management echelon, etc., and the target agricultural situation management echelon based on the judgment result;

[0121] The agricultural situation management unit conducts agricultural situation management on the first agricultural situation management echelon, the second agricultural situation management echelon, etc., and the target agricultural situation management echelon in turn;

[0122] The agricultural condition management time prediction module is used to predict the agricultural condition management time of each abnormal area, and perform agricultural condition management on each abnormal area according to the predicted management time;

[0123] The agricultural management time prediction module includes a management time prediction unit and a fertilization management time prediction unit;

[0124] The control time prediction unit predicts the control time for each abnormal area stored in each agricultural condition management echelon according to the pest control time, humidity control time and temperature control time of the target abnormal area in each agricultural condition management echelon, and performs agricultural condition management on each abnormal area according to the predicted control time;

[0125] The fertilization management time prediction unit collects the real-time agricultural risk index of each abnormal area stored in the target agricultural management echelon, predicts the fertilization management time of each abnormal area stored in the target agricultural management echelon based on the collected information, and performs agricultural management on each abnormal area according to the predicted management time.

[0126] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A method for assisting agricultural sentiment perception based on a multimodal large model, characterized by: The method comprises: S10: Several sets of agricultural monitoring equipment are evenly arranged in the target area. Abnormal areas in the target area are screened out based on the monitoring data output by each set of agricultural monitoring equipment arranged in the target area. The processing order of each abnormal area is determined based on the real-time agricultural risk index of each abnormal area. The S10 includes: S101: Divide the target area by a distance d to obtain a plurality of square areas with a side length of d, wherein a set of agricultural monitoring equipment is arranged in each square area, wherein the set of agricultural monitoring equipment includes a soil sensor, a meteorological monitoring sensor, a seedling condition detector, and an insect pest detection and reporting instrument. Monitoring data output by each set of agricultural monitoring equipment arranged in the target area is collected. Based on the collected monitoring data, the monitoring data includes the moisture content of the soil surface layer and the soil deep layer at a distance of X meters from the soil surface monitored by the soil sensor, the temperature and humidity monitored by the meteorological monitoring sensor, the plant height and leaf area of ​​the crops monitored by the seedling condition detector, and the number of pests monitored by the insect pest detection and reporting instrument. A three-dimensional simulation of the growth conditions and growth environment of the crops planted in the target area is performed in real time to obtain a real-time agricultural condition perception three-dimensional model of the target area, where X represents a constant and 0<X<the average burial depth of the crops planted in the target area. S102: Calculate the abnormal index of each divided area based on the real-time agricultural sentiment perception three-dimensional model of the target area. The specific calculation formula is: F i =a*(g i / G i )+b*(s i / S i ); Where i = 1, 2, ..., m, represents the number of each square area in the target area, m represents the total number of square areas in the target area, a and b represent the proportional coefficient and a + b = 1, s i represents the total area of ​​abnormal leaves of crops monitored by the seedling condition detectors arranged in the i-th square area, S i G represents the total leaf area of ​​crops monitored by the seedling condition detectors arranged in the i-th square area, i represents the total number of crops planted in the i-th square area, g i F represents the total number of abnormal crops in the i-th square area, i represents the abnormal index of the ith square area; If 0.4<F i ≤1, the ith square area is considered to be an abnormal area. If 0≤F i ≤0.4, the i-th square area is considered not to be an abnormal area; S103: Predicting the real-time agricultural risk index of each abnormal area screened out. The specific prediction formula is: , where j = 1, 2, ..., n, represents the number corresponding to each abnormal area, n represents the total number of abnormal areas screened out, t represents the time value, y jt It represents the number of pests detected by the insect monitoring device in the abnormal monitoring area numbered j at time t, e represents a constant and e>1, v=1,2,3,4, when v=1, w vjt It represents the moisture content of the soil surface detected by the soil sensor in the abnormal monitoring area numbered j at time t. When v=2, w vjt It represents the moisture content of the deep soil layer X meters away from the soil surface detected by the soil sensor arranged in the abnormal monitoring area numbered j at time t. When v=3, 4, w vjt They represent the temperature and humidity monitored by the meteorological monitoring sensor in the abnormal monitoring area numbered j at time t. If w vjt If it can meet the growth needs of crops, , if w vjt If it cannot meet the growth needs of crops, , H jt represents the agricultural risk index of the abnormal area numbered j at time t; The processing order of each abnormal area is determined according to the order of the real-time agricultural risk index of each abnormal area from large to small; S20: analyzing the correlation between the abnormal areas according to the distribution characteristic coefficients corresponding to the abnormal causes in the abnormal areas; S30: adjusting the processing order of each abnormal area according to the correlation between the abnormal areas, and determining the agricultural situation management strategy for each abnormal area based on the adjustment result; S40: Predicting the agricultural condition management time of each abnormal area, and performing agricultural condition management processing on each abnormal area according to the predicted management time.

2. The method for assisting agricultural sentiment perception based on a multimodal large model according to claim 1, characterized in that: The S20 includes: S201: collecting monitoring data output by agricultural monitoring equipment arranged in each abnormal area within the time period [t0, t]; At time t, according to Calculate the distribution characteristic coefficient corresponding to the abnormal cause of pest infestation in the abnormal area numbered j, where F jt represents the abnormal index of the abnormal area numbered j at time t, t0 represents the most recent start-up time of the agricultural monitoring equipment from time t, and k1 represents the relationship coefficient between the number of pests and the degree of damage to crops planted in the square area obtained based on experimental data; At time t, according to Calculate the distribution characteristic coefficient corresponding to the abnormal cause of humidity abnormality in the abnormal area numbered j; Among them, k2 represents the relationship coefficient between humidity and the degree of damage to crops planted in the square area obtained based on experimental data. If 0.5*(w´ 1jt +w´ 2jt )≤0.5*(w 1jt +w 2jt )≤0.5*(w´´ 1jt +w´´ 2jt ),but , if 0.5*(w 1jt +w 2jt )>0.5*(w´´ 1jt +w´´ 2jt ) or 0.5*(w 1jt +w 2jt )<0.5*(w´ 1jt +w´ 2jt ),but or ,w´ 1jt 、w´´ 1jt They represent the lower limit of soil surface moisture content and the upper limit of soil surface moisture content temperature corresponding to the normal growth of crops planted in the abnormal area numbered j at time t, w´ 2jt 、w´´ 2jt They represent the lower limit of soil moisture content at a distance of X meters from the soil surface and the upper limit of soil moisture content at a distance of X meters from the soil surface when the crops planted in the abnormal area numbered j grow normally at time t, respectively. p1=1 or p1=-1. When 0.5*(w 1jt +w 2jt )>0.5*(w´´ 1jt +w´´ 2jt ), p1=-1, when 0.5*(w 1jt +w 2jt )<0.5*(w´ 1jt +w´ 2jt ), p1=1; At time t, according to The distribution characteristic coefficient corresponding to the abnormal cause of temperature abnormality in the abnormal area numbered j is calculated, where k3 represents the relationship coefficient between the temperature and the degree of damage to crops planted in the square area obtained based on experimental data, T jt It represents the temperature value monitored by the meteorological monitoring sensor in the abnormal area numbered j at time t. If T´ jt ≤T jt ≤T´´ jt ,but , if T jt >T´´ jt or T jt <T´ jt ,but or , T´ jt 、T´´ jt They represent the lower and upper temperature limits of the crops grown in the abnormal area numbered j at time t when they grow normally, p2=1 or p2=-1. When T jt >T´´ jt When p2=-1, when T jt <T´ jt When p2=1; S202: Classify the abnormal area according to the distribution characteristic coefficient corresponding to each abnormal cause in the abnormal area. The specific classification processing method is as follows: The abnormal area corresponding to the maximum value of the real-time agricultural risk index is taken as the target abnormal area, and the number of the selected abnormal area is c, c = 1, 2, ..., n and c ≠ j, where Q 1ct , Q 2ct , Q 3ct All are not equal to 0; When Q 1jt , Q 2jt , Q 3jt When both are not equal to 0, , L jct The correlation between the abnormal area numbered j and the abnormal area numbered c at time t is calculated, q = 1, 2, 3. At this time, number j is put into set A to obtain the first abnormal area subset; When Q 1jt , Q 2jt , Q 3jt There is a 0, , now put number j into set A; When Q 1jt , Q 2jt , Q 3jt When there are two 0s, , now put number j into set A; When Q 1jt , Q 2jt , Q 3jt When both are equal to 0, At this time, it is considered that the abnormal cause of the crops planted in the abnormal area numbered j is abnormal fertilization, and the number j is put into set A.

3. The method for assisting agricultural sentiment perception based on a multimodal large model according to claim 2, characterized in that: The S30 includes: S301: For an abnormal region stored in the first abnormal region subset, if the condition 0.7<L is satisfied jct ≤1, then 0.7<L jct ≤1 corresponding to the abnormal area according to L jct The emission order from large to small is placed after the target abnormal area, and the target abnormal area and 0.7<L jct ≤1 corresponding abnormal area is deleted from the abnormal area screened out in S102, and the target abnormal area and 0.7<L jct The abnormal areas corresponding to ≤1 are regarded as the first echelon of agricultural management; S302: Repeat the operations of S103, S202, and S301 to determine the second agricultural situation management echelon, the third agricultural situation management echelon, and so on, until the abnormal area screened out in S102 cannot be used as the target abnormal area; For abnormal areas that cannot be used as target abnormal areas, agricultural situation management will be carried out in descending order of real-time agricultural situation risk index, and abnormal areas that cannot be used as target abnormal areas will be used as target agricultural situation management echelons; S303: Carry out agricultural situation management on the first agricultural situation management echelon, the second agricultural situation management echelon..., and the target agricultural situation management echelon in turn.

4. The method for assisting agricultural sentiment perception based on a multimodal large model according to claim 3, characterized in that: The S40 includes: S401: For the first agricultural condition management echelon, the pest control time W1, humidity control time W2, and temperature control time W3 of the target abnormal area are collected; The pest control time M of the abnormal area numbered j stored in the first agricultural situation management echelon 1j =W1*(Q 1jt / Q 1ct ), when M 1j When ≥W1, the pest control time D of the abnormal area numbered j stored in the first agricultural situation management echelon 1j = Pest control time G1 in target abnormal area, when M 1j When W1 is less than D 1j =G1+[W1-W1*(Q 1jt / Q 1ct )]; The humidity management duration M of the abnormal area numbered j stored in the first agricultural management echelon 2j =W2*(Q 2jt / Q 2ct ), when M 2j When ≥W2, the humidity management time D of the abnormal area numbered j stored in the first agricultural management echelon 2j = Humidity control time G2 of target abnormal area, when M 2j When <W2, D 2j =G2+[W2- W2*(Q 2jt / Q 2ct )]; The temperature management duration M of the abnormal area numbered j stored in the first agricultural management echelon 3j =W3*(Q 3jt / Q 3ct ), when M 3j When ≥W3, the temperature management time D of the abnormal area numbered j stored in the first agricultural situation management echelon 3j = Temperature control time G3 of target abnormal area, when M 3j When <W3, D 3j =G3+[W3- W3*(Q 3jt / Q 3ct )]; S402: Based on the operation of S401, the management time of each abnormal cause is predicted for each stored abnormal area for the second agricultural management echelon, the third agricultural management echelon, etc.; S403: Collect the agricultural risk index of each abnormal area stored in the target agricultural risk management echelon at time t, and record the abnormal area corresponding to the maximum value of the collected agricultural risk index as γ, γ = 1, 2, ..., n, and the fertilization management time D of the abnormal area numbered j stored in the target agricultural risk management echelon 4j =G γ +[W γ -W γ *(H j / H γ ), G γ Indicates the fertilization management time of the abnormal area numbered γ, W γ Indicates the duration of fertilization treatment in the abnormal area numbered γ; S404: Perform agricultural condition management on each abnormal area according to the predicted management time.

5. A system for assisting agricultural sentiment perception based on a multimodal large model, applied to the method for assisting agricultural sentiment perception based on a multimodal large model according to any one of claims 1 to 4, characterized in that: The system includes an agricultural risk index prediction module, a correlation analysis module, an agricultural management strategy determination module, and an agricultural management time prediction module; The agricultural risk index prediction module is used to predict the real-time agricultural risk index of each abnormal area and determine the processing order of each abnormal area based on the prediction results; The correlation analysis module is used to analyze the correlation between the abnormal areas according to the distribution characteristic coefficients corresponding to the abnormal causes in the abnormal areas; The agricultural situation management strategy determination module is used to adjust the processing order of each abnormal area according to the correlation between the abnormal areas, and determine the agricultural situation management strategy for each abnormal area based on the adjustment result; The agricultural condition management time prediction module is used to predict the agricultural condition management time of each abnormal area, and perform agricultural condition management processing on each abnormal area according to the predicted management time.

6. The agricultural sentiment perception assistance system based on a multimodal large model according to claim 5 is characterized in that: The agricultural risk index prediction module includes an agricultural risk perception three-dimensional model acquisition unit, an abnormality index calculation unit, an agricultural risk index prediction unit and an abnormal area processing order determination unit; The agricultural condition perception three-dimensional model acquisition unit performs a three-dimensional simulation of the growth conditions and growth environment of crops planted in the target area in real time based on the monitoring data output by the agricultural condition monitoring equipment arranged in each divided area, and obtains a real-time agricultural condition perception three-dimensional model of the target area; The abnormality index calculation unit constructs a calculation formula to calculate the abnormality index of each divided area based on the real-time agricultural sentiment perception three-dimensional model of the target area; The agricultural risk index prediction unit predicts the real-time agricultural risk index of each abnormal area selected according to the constructed prediction model; The abnormal area processing sequence determining unit determines the processing sequence of each abnormal area according to the order of the real-time agricultural risk index of each abnormal area from large to small.

7. The agricultural sentiment perception assistance system based on a multimodal large model according to claim 6 is characterized in that: The association analysis module includes a distribution characteristic coefficient calculation unit and an association analysis unit; The distribution characteristic coefficient calculation unit calculates the distribution characteristic coefficient corresponding to each abnormal cause in each abnormal area based on the monitoring data output by the agricultural monitoring equipment arranged in each abnormal area within a random period of time; The correlation analysis unit classifies the abnormal regions according to the distribution characteristic coefficients corresponding to the abnormal causes in the abnormal regions, and calculates the correlation between the abnormal regions based on the classification results.

8. The agricultural sentiment perception assistance system based on a multimodal large model according to claim 7 is characterized in that: The agricultural situation management strategy determination module includes an agricultural situation management echelon determination unit and an agricultural situation management unit; The agricultural situation management echelon determination unit determines whether the abnormal areas stored in the first abnormal area subset meet the set conditions, and determines the first agricultural situation management echelon, the second agricultural situation management echelon, etc., and the target agricultural situation management echelon according to the judgment result; The agricultural situation management unit conducts agricultural situation management on the first agricultural situation management echelon, the second agricultural situation management echelon..., and the target agricultural situation management echelon in turn.

9. The agricultural sentiment perception assistance system based on a multimodal large model according to claim 8, characterized in that: The agricultural management time prediction module includes a management time prediction unit and a fertilization management time prediction unit; The control time prediction unit predicts the control time for each abnormal area stored in each agricultural condition management echelon according to the pest control time, humidity control time and temperature control time of the target abnormal area in each agricultural condition management echelon, and performs agricultural condition management processing on each abnormal area according to the predicted control time; The fertilization management time prediction unit collects the real-time agricultural risk index of each abnormal area stored in the target agricultural management echelon, predicts the fertilization management time of each abnormal area stored in the target agricultural management echelon based on the collected information, and performs agricultural management on each abnormal area according to the predicted management time.

Images