Tree pest detection method and device based on electronic nose and hollow structure recognition

Abstract

The invention provides a tree disease and insect pest detection method and device based on electronic nose and hollow structure recognition, and relates to the technical field of disease and insect pest detection, and the method comprises the following steps: attaching a hollow structure with a built-in electronic nose to a branch detection part, positioning a spatial position, collecting candidate mark gas, and extracting types and concentration values; calculating a concentration variation coefficient and complexity to adjust the number of filtering time points, and obtaining a filtered concentration value through moving average filtering; screening marker gas in combination with a disease and insect pest associated gas database, calculating a detection-axis distance, and obtaining a priority coefficient of each part in combination with a concentration average value after filtering of the marker gas; and arranging and screening the first m bits according to a descending order of priority coefficients, setting a critical distance by taking the highest priority part as a reference point, measuring the check-base distance between other parts and the reference point, comparing, and judging the number of infection points. According to the method, the filtering window is dynamically adjusted, the data noise is reduced, and the defect that a plurality of high-concentration points are misjudged as a plurality of infection points is avoided in combination with spatial positioning and priority division.

Description

Technical Field

[0001] The present invention relates to the technical field of pest and disease detection, and in particular to a tree pest and disease detection method and device based on electronic nose and hollow structure recognition. Background Art

[0002] With technological advancements, electronic noses and hollow structure recognition technologies are bringing new insights to tree pest and disease detection. Electronic noses can rapidly detect changes in volatile organic compounds (VOCs) released by plants. When attacked by pests and diseases, plants undergo metabolic changes, releasing specific VOCs. For example, citrus fruits infected with Huanglongbing (Huanglongbing) produce specific volatile compounds that attract psyllids. Tea trees attacked by the tea looper (Tea geometrid) release large amounts of high-concentration VOCs. Analyzing these VOCs can assess tree health. However, current electronic nose sensors have limitations in sensitivity, selectivity, and stability. They are susceptible to interference from environmental factors such as temperature, humidity, and light. Furthermore, there is a lack of standardized experimental methods and data analysis methods across different studies, hindering detection accuracy and comparability of results. Hollow structures, through their specialized cavity design, possess the ability to selectively enrich target gases. This allows them to precisely capture volatile gases released from the test area of ​​the tree under test, reducing interference from unrelated gases in the surrounding environment. This significantly increases the concentration of the target gas in the detection area and enhances the electronic nose's ability to detect low-concentration characteristic gases.

[0003] In the prior art, a method for predicting tea tree pest severity based on electronic nose technology, disclosed in publication number CN119380079A, includes the following steps: pre-treating multiple groups of tea trees to infect them with different types of pests; using an electronic nose to detect volatile gases from all tea trees within a certain period of time after being infested, obtaining electronic nose response curves for tea trees with different pest types and at different pest severity levels; processing the electronic nose response curves to extract tea tree pest characteristic data; constructing and training a tea tree pest classifier based on the tea tree pest characteristic data and its corresponding pest type; constructing and training a pest severity prediction network for each pest type based on the tea tree pest characteristic data and its corresponding pest severity; collecting and processing volatile gases from the tea trees to be tested using the electronic nose to obtain pest characteristic data; inputting the pest characteristic data into the tea tree pest classifier to obtain a classification result; and, based on the classification result, inputting the pest characteristic data into a corresponding pest severity prediction network to obtain a pest severity prediction result. This method can quickly identify pest types and predict pest severity.

[0004] However, there are still some deficiencies. As can be seen from the above statement, in the gas data processing link, the existing technology only performs conventional processing on the electronic nose response curve, and does not dynamically adjust the filter window size through "concentration variation coefficient and complexity". This processing method is difficult to eliminate false signals caused by instantaneous environmental interference (such as airflow fluctuations around tea trees), making it easy for noise to be mixed into the extracted feature data. The contaminated feature data will directly affect the accuracy of subsequent data processing; Secondly, the existing technology does not spatially locate and prioritize the detection sites, which directly leads to the inability to distinguish the differences in pest and disease risks in different parts of the same plant, and the inability to judge the spread status of pests and diseases based on spatial position relationships, especially when multiple high-concentration points caused by the same infection point spreading to different parts are mistakenly identified as multiple infection points. Summary of the Invention

[0005] The purpose of the present invention is to provide a tree disease and insect pest detection method and device based on electronic nose and hollow structure recognition to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: A tree pest and disease detection method based on electronic nose and hollow structure recognition, comprising the following steps: S1. Attach the hollow structure containing the electronic nose to the detection area of ​​the branch, locate the spatial position of different detection areas, and collect candidate marker gases at the detection areas during the detection period. Extract the type and concentration of the candidate marker gases at each time point, calculate the concentration coefficient of variation and complexity, adjust the number of filtering time points accordingly, and perform sliding average filtering to obtain the filtered concentration values ​​of the candidate marker gases at each detection area at different time points. S2. Analyze the filtered concentration values ​​based on a preset pest-associated gas database to identify marker gases. Based on the spatial location of different detection locations, the vertical distance between each location and the central axis of the trunk is calculated as the detection-axis distance. Combined with the average filtered concentration of the marker gas at each detection location over the detection period, the priority coefficient for each detection location is calculated to represent the degree of pest risk. S3. Arrange the priority coefficients of all detection parts in descending order, filter out the detection parts with the top m priority coefficients, take the detection part with the highest priority as the reference point, set the critical distance with the reference point as the center, measure the straight-line distance between the other m-1 detection parts and the reference point as the detection-reference distance, compare the detection-reference distance with the critical distance one by one, and determine the number of infection points of pests and diseases based on the comparison results.

[0007] Furthermore, the concentration of the candidate marker gas during the detection period is normalized, and then the mean of the normalized candidate marker gas concentration is calculated. The concentration variation coefficient and the complexity of the candidate marker gas are calculated according to the following formula: ; ; ; in, is the complexity of the candidate marker gas, is the number of candidate marker gases, For the The concentration variation coefficient of the candidate marker gas, For the The standard deviation of the concentration of the candidate marker gas in the detection time period, For the The mean of the concentrations of the candidate marker gases, is the mean of the concentration variation coefficients of all candidate marker gases, is the index of the candidate marker gas, ; Adjust the filter window size according to the complexity of the candidate marker gas: ; in, is the number of filtering time points, is the reference value of the number of filtering time points, is the first level threshold of complexity, is the second level threshold of complexity.

[0008] Furthermore, the specific steps for obtaining the filtered concentration value are as follows: For each candidate marker gas at each detection location, within its corresponding detection time period, for each time point, its concentration value is calculated relative to the nearest concentration value at that time point. The average of the concentration values ​​at the time points is used as the filtered concentration value at that time point.

[0009] Furthermore, the marker gases are screened out based on the preset pest-related gas database. The specific steps are as follows: From the candidate marker gases, retain the gases that match the types of pest gases in the database; Calculate the filtered mean concentration of these gases during the detection period. If the mean falls within the gas concentration range corresponding to the pests and diseases in the database, the gas is listed as a candidate gas. For a gas to be selected, if its filtered concentration value exceeds the preset threshold at more than three consecutive time points during the detection period, it will be included in the pre-selected marker gas set; From the preselected marker gas set, the gas with the highest average concentration after filtering is selected as the final marker gas.

[0010] Furthermore, based on the spatial positions of different detection parts, the vertical distances between them and the central axis of the trunk are obtained. The specific steps are as follows: Set up a cross section every 50 cm along the trunk upward direction, and measure the coordinates of the center of the cross section at different heights of the trunk; Connect the centers of the cross sections at each height to form a virtual straight line, which is the central axis of the trunk; Based on the spatial positions of different detection parts, determine the coordinates of each detection part in space; According to the distance formula from the spatial geometric midpoint to the straight line, the vertical distance from the coordinates of each detection part to the central axis of the trunk is calculated.

[0011] Furthermore, based on the filtered mean concentration of the marker gas at each detection location during the detection period and the detection-axis distance, the priority coefficient of the pest risk at different detection locations is obtained, according to the following formula: ; in, For the Priority coefficient of pest and disease risk of each detection part, For the The average concentration of the marker gas at each detection location after filtering during the detection time period, For the The detection-axis distance of each detection part, is the index of the detection part, , is the number of test sites; Where, is the weight coefficient of the filtered mean concentration of the marker gas at the detection site, is the weight coefficient of the inspection-axis distance, On the basis of ; Arrange the priority coefficients of all detection parts in descending order and select the detection parts with the top m priority coefficients. The specific steps are as follows: Set the sorting rules: Rule 1: Sort by priority coefficient from large to small; Rule 2: If there are multiple detection locations with the same priority coefficient, they are sorted by the average concentration after filtering by the marker gas, with the location with the highest average concentration being ranked first. According to rule 1, all detection parts are sorted in descending order of priority coefficient to obtain the preliminary sorting results; Check the preliminary sorting results. If there are detection locations with the same priority coefficient, extract the filtered mean concentration of the marker gas from these detection locations, perform secondary sorting according to Rule 2, adjust the order, and select the top m detection locations.

[0012] Furthermore, the inspection-base distance is compared with the critical distance one by one, and the number of infection points of pests and diseases is determined based on the comparison results. The specific steps are as follows: Compare the detection-base distance with the set critical distance one by one, and count the number of detection locations where the detection-base distance is greater than the critical distance, which is recorded as ; like , that is, if all the inspection-base distances are less than or equal to the critical distance, then it is determined that there is one pest infection point on the tree; like , that is, there is at least one detection-base distance greater than the critical distance, then it is determined that there are multiple pest and disease infection points on the tree.

[0013] To achieve the above object, the present invention further provides the following technical solutions: A tree pest and disease detection system based on an electronic nose and hollow structure recognition, the system being configured to execute any of the above-mentioned tree pest and disease detection methods based on an electronic nose and hollow structure recognition, comprising: The data acquisition module is used to fit the hollow structure with a built-in electronic nose to the detection area of ​​the branch, locate the spatial position of different detection areas, and collect candidate marker gases at the detection area during the detection time period. The type and concentration of the candidate marker gases at each time point are extracted, and the concentration coefficient of variation and complexity are calculated. The number of filtering time points is adjusted accordingly. After sliding average filtering, the filtered concentration value of the candidate marker gas at each detection area at different time points is obtained; The data processing module is used to analyze the filtered concentration values ​​according to the preset pest-associated gas database to screen out the marker gases. Based on the spatial location of different detection sites, the vertical distance between the site and the main trunk axis is obtained as the detection-axis distance. The priority coefficient of each detection site is obtained by combining the average filtered concentration of the marker gas at each detection site within the detection time period, which is used to characterize the degree of pest risk. The judgment module is used to arrange the priority coefficients of all detection parts in descending order, screen out the detection parts with the top m priority coefficients, take the detection part with the highest priority as the reference point, set the critical distance with the reference point as the center, measure the straight-line distance between the other m-1 detection parts and the reference point, compare them one by one with the critical distance, and judge the number of infection points of pests and diseases based on the comparison results.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present invention calculates the concentration variation coefficient and complexity of candidate marker gases, dynamically adjusts the filter window size, and then obtains the filtered concentration value through sliding average filtering. This process can flexibly adapt the filter strength according to the fluctuating characteristics of gas concentration, effectively filter out false signals caused by transient environmental interference, reduce noise in characteristic data, and provide a more reliable data basis for subsequent marker gas screening and priority coefficient calculation. A complete spatial positioning and priority division system was constructed through multiple steps. The inspection-axis distance was calculated based on the spatial position of different detection parts, and the priority coefficient of each detection part was obtained by combining the average concentration of the marker gas after filtering, realizing the quantitative ranking of the risk degree, screening out the detection parts with the top m priorities, taking the detection part with the highest priority as the reference point, and setting the critical distance with the reference point as the center. The inspection-base distance of the other m-1 detection parts and the reference point was measured, and compared with the critical distance one by one to determine the number of infection points of pests and diseases. This process fully combines the spatial position relationship with the risk priority, making the judgment of the number of infection points more scientific and accurate, avoiding the disadvantage of misjudging multiple high-concentration points as multiple infection points based solely on gas concentration, and accurately distinguishing independent infection points from diffusion areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Schematic diagram of the overall method flow of the present invention; Figure 2 It is a block diagram of the module composition of the present invention; Figure 3 This is a schematic diagram of the fitting of the number of candidate marker gas types and complexity of the present invention; Figure 4 Schematic diagram of the fitting of the mean value of the concentration variation coefficient and the complexity of the present invention; Figure 5 Schematic diagram of the fitting of the concentration mean and priority coefficient after filtering in the present invention; Figure 6 This is a fitting diagram of the detection-axis distance and the priority coefficient of the present invention. DETAILED DESCRIPTION

[0016] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0017] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0018] Example 1: See also Figures 1-6 , the present invention provides a technical solution: A tree pest and disease detection method based on electronic nose and hollow structure recognition, comprising the following steps: S1. Attach the hollow structure containing the electronic nose to the detection area of ​​the branch, locate the spatial position of different detection areas, and collect candidate marker gases at the detection areas during the detection period. Extract the type and concentration of the candidate marker gases at each time point, calculate the concentration coefficient of variation and complexity, adjust the number of filtering time points accordingly, and perform sliding average filtering to obtain the filtered concentration values ​​of the candidate marker gases at each detection area at different time points. On the basis of the above embodiment, the hollow structure with the built-in electronic nose is attached to the detection part of the branch, and the spatial position of different detection parts is determined. The specific steps are as follows: 1) Using the intersection of the ground and the trunk as the coordinate origin, establish a three-dimensional rectangular coordinate system, where the X-axis is horizontal and points to the side of the trunk, the Y-axis is perpendicular to the ground and points upward, and the Z-axis is horizontal and perpendicular to the X-axis. 2) Install a small GPS module outside the hollow structure; 3) Fit the hollow structure of the built-in electronic nose tightly against the part of the branch to be inspected, and adjust the position of the hollow structure so that it is in full contact with the branch surface to avoid positioning deviation due to looseness; 4) Real-time acquisition of 3D coordinate data of hollow structures through the positioning module , continuously collect multiple times and calculate the average value as the spatial coordinate of the detection part; Based on the above embodiment, candidate marker gases are collected at the detection site during the detection period, and the types and concentrations of the candidate marker gases at each time point are extracted. The specific steps are as follows: 1) Based on the tree species and common pest and disease types, preset the detection time period (e.g., 24 consecutive hours) and sampling interval (sampling once every hour), and calibrate the detection range and sensitivity of the electronic nose; 2) After the hollow structure is aligned with the detection area and positioned, the electronic nose's gas collection function is activated. The gas released from the detection area is absorbed in real time through the structure's built-in gas sensor array. 3) Each time a sample is taken, the electronic nose system automatically records the current sampling time and forms a timestamp to ensure that the gas data corresponds to the sampling time. 4) After sampling, the electronic nose analyzes the gas sample using its built-in data analysis module. Based on the sensor array's response characteristics and the sensor's response to the changes in resistance and conductivity, it identifies and outputs candidate marker gases, such as ethanol, ethylene, and indole. 5) For each candidate marker gas identified, the sensor's response strength, including voltage signal value and current change, is compared and converted with a preset concentration calibration curve to calculate the concentration value of the gas at the current time point. The concentration calibration curve is obtained by calibrating with a standard gas of known concentration; 6) The candidate marker gas types, corresponding concentration values ​​and timestamps extracted at each time point are synchronously stored in the system database to form a raw gas data record table; 7) Repeat steps 2-6 throughout the entire detection period according to the preset sampling interval to ensure that complete gas data is collected at each time point until the detection period ends.

[0019] Based on the above embodiment, the concentration of the candidate marker gas during the detection period is normalized, and then the mean of the normalized candidate marker gas concentration is calculated, and the concentration variation coefficient is calculated according to the following formula: ; ; in, is the number of candidate marker gases, For the The concentration variation coefficient of the candidate marker gas, For the The standard deviation of the concentration of the candidate marker gas in the detection time period, For the The mean of the concentrations of the candidate marker gases, is the mean of the concentration variation coefficients of all candidate marker gases, is the index of the candidate marker gas, ; Table 1. Changes in the complexity of candidate marker gases with the number of species and the mean value of the concentration coefficient of variation

[0020] According to Table 1, as the number of candidate marker gases increases from 1 to 15, the mean of the concentration coefficient of variation increases from 0.2 to 3.0, and the complexity increases from 0.6 to 9.0 accordingly. The change ranges of the three always maintain a fixed proportion, which conforms to the linear law. The complexity of the candidate marker gases is linearly positively correlated with the number of species and the mean of the concentration coefficient of variation.

[0021] according to Figure 3-Figure 4 It can be seen that the complexity of candidate marker gases shows a linear positive correlation with two types of parameters: The fitting line between the complexity of candidate marker gases and the number of types is a straight line. The complexity increases linearly with the number of candidate marker gases. The more types there are, the more complex the gas composition is. The fitting line between the complexity of candidate marker gases and the mean value of concentration variation coefficient is a straight line. The complexity increases linearly with the increase of the mean value of concentration variation coefficient. The higher the concentration variation coefficient, the higher the composition complexity.

[0022] Based on the above embodiment, the complexity of the candidate marker gas is calculated according to the following formula: ; in, The complexity of the candidate marker gas is used to evaluate the complexity of the gas composition by combining the number of candidate marker gas types and the mean value of the concentration variation coefficient. The greater the complexity, the higher the complexity of the gas composition. On the basis of the above, it should be noted that: Number of types Refers to the number of candidate marker gases related to tree diseases and insect pests. When it increases, it directly reflects that the metabolic disorder caused by diseases and insect pests in trees has intensified, the diversity of gas components has increased, and the complexity Increase, therefore, and Is positively correlated.

[0023] Mean of the coefficient of variation of concentration It is an indicator to measure the degree of fluctuation of candidate marker gas at multiple acquisition moments. When it increases, it reflects that the gas release rhythm of trees is disordered due to diseases and insect pests, the concentration fluctuation is aggravated, and the complexity Increase, therefore, and Is positively correlated.

[0024] The mean of the number of species and the mean of the coefficient of variation of concentration describe gas composition characteristics from two different dimensions. Both are positively correlated with "gas composition complexity," but they are independent parameters, reflecting the complexity of gas composition from the perspectives of gas composition diversity and concentration volatility, respectively. The two cannot replace each other. Therefore, they can be integrated into a comprehensive indicator that can be quantified and compared through linear combination.

[0025] In summary, the above function form is set to express the functional relationship between complexity and the number of candidate marker gas types and the mean value of the concentration variation coefficient.

[0026] Adjust the filter window size according to the complexity of the candidate marker gas: ; in, is the number of filtering time points, is the reference value of the number of filtering time points, is the first level threshold of complexity, is the second level threshold of complexity.

[0027] On the basis of the above, it should be noted that: By analyzing the complexity of a large number of labeled samples (gas composition data with known health, mild abnormality, and severe abnormality) Perform analysis to find the natural dividing point: Collect three types of samples (healthy sample set , mildly abnormal sample set ), severely abnormal sample set ), calculate the complexity of each sample separately ; statistics 、 、 The distribution range of medium complexity shows obvious "clustering characteristics": The complexity is concentrated in the lower range; The complexity is concentrated in the medium range; The complexity is concentrated in the higher range; Pick Maximum value and The midpoint between the minimum and the normal value, ensuring healthy and mildly abnormal Values ​​are clearly separated; Similarly, take Maximum value and The dividing point between the minimum and maximum values ​​to avoid mild and severe abnormalities Values ​​overlap.

[0028] On the basis of the above, it should be noted that: When the complexity is low (i.e. ): There are few types of gases in the signal, the concentration fluctuates smoothly, and the noise interference is weak. At this time, there is no need for an overly large filter window. Use the reference value It can accurately retain the details of the original signal while suppressing a small amount of noise; When the complexity is medium (i.e. ): New signature gases appear in the signal, the concentration fluctuation increases, and the degree of confusion between noise and effective features increases. At this time, a larger window is needed Enhance noise suppression capabilities while avoiding over-smoothing that may mask slightly abnormal features; When the complexity is high (i.e. ): There are many types of gases in the signal, the concentration fluctuates violently, and the noise and effective features are highly mixed. In this case, a larger window must be used. Powerfully smooth noise to ensure that the extracted core features are stable and reliable, and to prevent noise from being misidentified as pest and disease signals.

[0029] The window size increases with the complexity ( ), which conforms to the rule that "the higher the complexity, the more chaotic the signal, and the stronger the filtering required".

[0030] Based on the above embodiment, the specific steps for obtaining the filtered concentration value are as follows: For each candidate marker gas at each detection location, within its corresponding detection time period, for each time point, its concentration value is calculated relative to the nearest concentration value at that time point. The average of the concentration values ​​at the time points is used as the filtered concentration value at that time point.

[0031] S2. Analyze filtered concentration values ​​based on a preset pest-associated gas database to identify marker gases. Based on the spatial location of each detection site, calculate its vertical distance from the central axis of the trunk as the detection-axis distance. Combined with the average filtered concentration of the marker gas at each detection site over the detection period, calculate a priority coefficient for each detection site to characterize the pest risk level. Both the detection-axis distance and the average filtered concentration are normalized before calculating the priority coefficient for each detection site. On the basis of the above embodiment, the marker gas is screened out according to the preset pest-related gas database. The specific steps are as follows: From the candidate marker gases, retain the gases that match the types of pest gases in the database; Calculate the filtered mean concentration of these gases during the detection period. If the mean falls within the gas concentration range corresponding to the pests and diseases in the database, the gas is listed as a candidate gas. For a gas to be selected, if its filtered concentration value exceeds the preset threshold at more than three consecutive time points during the detection period, it will be included in the pre-selected marker gas set; From the preselected marker gas set, the gas with the highest average concentration after filtering is selected as the final marker gas.

[0032] On the basis of the above embodiment, based on the spatial positions of different detection parts, the vertical distances between the parts and the central axis of the trunk are obtained. The specific steps are as follows: Set up a cross section every 50 cm along the trunk upward direction, and measure the coordinates of the center of the cross section at different heights of the trunk; Connect the centers of the cross sections at each height to form a virtual straight line, which is the central axis of the trunk; Based on the spatial positions of different detection parts, determine the coordinates of each detection part in space , They are the X-axis coordinate value, Y-axis coordinate value, and Z-axis coordinate value of the detection part respectively; According to the distance formula from the spatial geometric midpoint to the straight line, the vertical distance from the coordinates of each detection part to the central axis of the trunk is calculated.

[0033] Table 2. Changes in priority coefficients with filtered concentration mean and detection-axis distance

[0034] Table 2 shows that as the filtered mean concentration increases from 0.5 to 6.5, the priority coefficient increases from 0.38 to 5.08. Furthermore, when the detector-axis distance remains relatively stable, the priority coefficient increases with each increase in the mean concentration, reflecting the positive contribution of the mean concentration to the priority coefficient. Therefore, the priority coefficient is positively correlated with the filtered mean concentration.

[0035] As the distance between the detector and the axis decreases from 5.0 to 0.8, the priority coefficient increases. Especially at smaller distances, a small change in distance leads to a more significant increase in the priority coefficient, reflecting that closer distances have a more significant impact on priority. Therefore, the priority coefficient is negatively correlated with the distance between the detector and the axis.

[0036] according to Figure 5-Figure 6 It can be seen that the fitting line of the mean concentration after filtering and the priority coefficient is a straight line, which is linearly positively correlated. That is, the larger the mean concentration after filtering, the higher the priority coefficient. The change trends of the two are uniform and synchronized. The mean concentration has a stable positive driving effect on the priority.

[0037] The fitting line of the inspection-axis distance and the priority coefficient is a curve, which shows a nonlinear negative correlation. The smaller the inspection-axis distance, the higher the priority coefficient. When the distance is small, a small change in the distance will cause a more obvious change in the priority coefficient, reflecting the reverse effect of distance on priority and its nonlinear characteristics.

[0038] On the basis of the above embodiment, the priority coefficients of pest and disease risks at different detection locations are obtained according to the filtered average concentration of the marker gas at each detection location during the detection period and the detection-axis distance, and the formula is as follows: ; in, For the The priority coefficient of the pest and disease risk of each detection part is used to combine the two index parameters of the filtered concentration mean and the detection-axis distance to comprehensively evaluate the pest and disease risk of the detection part. The larger the priority coefficient, the higher the risk of the detection part being damaged by pests and diseases. For the The average concentration of the marker gas at each detection location after filtering during the detection time period, For the The detection-axis distance of each detection part, is the index of the detection part, , is the number of test sites; On the basis of the above, it should be noted that: When pests and diseases move in tree branches, they will continuously release specific marker gases. The more active the activity and the higher the population density, the greater the amount of gas released, and the gas concentration around the detection site will also increase accordingly. Therefore, the average concentration of the marker gas after filtering It can directly reflect the activity level of pests and diseases. The larger the number, the more intense the pest and disease activity in that area, and the higher the risk level. It is positively correlated with the priority coefficient.

[0039] The central axis of the tree trunk is the core channel for the transportation of nutrients and water. The closer the branches are to the central axis, the closer they are to the central axis. The smaller it is, the closer its physiological relationship with the trunk is. If pests and diseases occur in this part, they will not only easily invade the trunk and affect the overall metabolism, but also accelerate the spread of the disease because of its proximity to the core channel. On the contrary, the farther away from the central axis, that is, The larger the distance, the lower the threat to the tree as a whole. The priority coefficient is negatively correlated.

[0040] The function uses addition to superimpose the two sub-items, which can achieve a comprehensive evaluation of two independent indicators: when and When the change is simultaneous ( While increasing The positive contributions of the two items will be superimposed, making More significantly reflect the trend of increasing risk; When the two change in opposite directions ( Increase but The additive form can balance the influence of the two by adjusting the weight coefficient, thus avoiding the one-sidedness of risk assessment by a single indicator.

[0041] This form of linear superposition is simple and intuitive, and can flexibly reflect the relative importance of the two parameters in risk assessment, meeting the actual needs of "judging risks by integrating multiple indicators."

[0042] In summary, the above function form is used to express the functional relationship between the priority coefficient and the filtered concentration mean and the detection-axis distance.

[0043] Where, is the weight coefficient of the filtered mean concentration of the marker gas at the detection site, is the weight coefficient of the inspection-axis distance; Mean concentration of marker gas after filtering It is a core indicator that directly reflects the intensity of pest activity. The presence and activity of pests will directly lead to an increase in the concentration of marker gases. The concentration is strongly correlated with the population density and degree of damage of pests. It is the "direct evidence" for judging risks. It reflects more on the "potential impact range" of the risk. The risk spreads more to the parts close to the trunk, but this impact must be based on the "existence of pests and diseases". Very low, almost no sign of gas release, even if Therefore, the direct risk evidence is should be given higher weight, so On the basis of ; As an implementation method, The value range is 0.5-1.0, The value range is 0-0.5. The specific value is set by technical personnel according to actual conditions and is not limited here.

[0044] S3. Arrange the priority coefficients of all detection parts in descending order, filter out the detection parts with the top m priority coefficients, take the detection part with the highest priority as the reference point, set the critical distance with the reference point as the center, measure the straight-line distance between the other m-1 detection parts and the reference point as the detection-reference distance, compare the detection-reference distance with the critical distance one by one, and determine the number of infection points of pests and diseases based on the comparison results.

[0045] On the basis of the above embodiment, the priority coefficients of all detection parts are arranged in descending order, and the detection parts with the top m priority coefficients are screened out. The specific steps are as follows: Set the sorting rules: Rule 1: Sort by priority coefficient from large to small; Rule 2: If there are multiple detection locations with the same priority coefficient, they are sorted by the average concentration after filtering by the marker gas, with the location with the highest average concentration being ranked first. According to rule 1, all detection parts are sorted in descending order of priority coefficient to obtain the preliminary sorting results; Check the preliminary sorting results. If there are detection locations with the same priority coefficient, extract the filtered mean concentration of the marker gas from these detection locations, perform secondary sorting according to Rule 2, adjust the order, and select the top m detection locations.

[0046] Based on the above embodiment, the highest priority detection part is used as the reference point, and the reference point is used as the center to set the critical distance. The straight-line distance between the other m-1 detection parts and the reference point is measured as the detection-base distance. The specific steps are as follows: From the first m detected parts, select the first one, that is, the highest priority detection part as the reference point, and retrieve its spatial coordinates in the three-dimensional rectangular coordinate system. , They are the X-axis coordinate value, Y-axis coordinate value, and Z-axis coordinate value of the reference point respectively; The specific value of the critical distance is determined based on the biological characteristics of the target pests and diseases (adult activity range, larval dispersal radius), the growth characteristics of tree branches (branch diameter, branch density), and historical detection data. For example, for trunk-boring pests with a small activity range, the critical distance can be set at 5-10 cm; for diseases easily spread by airflow, the critical distance can be appropriately increased to 15-20 cm. From the first m detected parts, extract the spatial coordinates of the remaining m-1 detection parts except the reference point, and record them as , and are associated one by one with the corresponding detection part identification, Respectively The X-axis coordinate value, Y-axis coordinate value, and Z-axis coordinate value of each detection part; According to the distance formula between two points in three-dimensional space, the straight-line distance between each detection part and the reference point is calculated as the detection-base distance. ,in, For the The straight-line distance between the detection part and the reference point, .

[0047] On the basis of the above embodiment, the detection-base distance is compared with the critical distance one by one, and the number of infection points of the pests and diseases is determined according to the comparison results. The specific steps are as follows: Compare the detection-base distances of m-1 detection parts with the set critical distance one by one, and count the number of detection parts with detection-base distance greater than the critical distance, which is recorded as ; like , that is, if all the inspection-base distances are ≤ the critical distance, then it is determined that there is only one pest infection point on the tree; like , that is, there is at least one detection-base distance greater than the critical distance, then it is determined that there are multiple points of disease and insect pest infection on the tree.

[0048] See also Figure 2 , the present invention also provides a technical solution: A tree pest and disease detection system based on an electronic nose and hollow structure recognition, the system being configured to execute any of the above-mentioned tree pest and disease detection methods based on an electronic nose and hollow structure recognition, comprising: The data acquisition module is used to fit the hollow structure with a built-in electronic nose to the detection area of ​​the branch, locate the spatial position of different detection areas, and collect candidate marker gases at the detection area during the detection time period. The type and concentration of the candidate marker gases at each time point are extracted, and the concentration coefficient of variation and complexity are calculated. The number of filtering time points is adjusted accordingly. After sliding average filtering, the filtered concentration value of the candidate marker gas at each detection area at different time points is obtained; The data processing module is used to analyze the filtered concentration values ​​according to the preset pest-associated gas database to screen out the marker gases. Based on the spatial location of different detection sites, the vertical distance between the site and the main trunk axis is obtained as the detection-axis distance. The priority coefficient of each detection site is obtained by combining the average filtered concentration of the marker gas at each detection site within the detection time period, which is used to characterize the degree of pest risk. The judgment module is used to arrange the priority coefficients of all detection parts in descending order, screen out the detection parts with the top m priority coefficients, take the detection part with the highest priority as the reference point, set the critical distance with the reference point as the center, measure the straight-line distance between the other m-1 detection parts and the reference point, compare them one by one with the critical distance, and judge the number of infection points of pests and diseases based on the comparison results.

[0049] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0050] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by computer software, electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution.

[0051] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0052] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A tree pest and disease detection method based on electronic nose and hollow structure recognition, characterized in that: The specific steps include: S1. Attach the hollow structure containing the electronic nose to the detection area of ​​the branch, locate the spatial position of different detection areas, and collect candidate marker gases at the detection areas during the detection period. Extract the type and concentration of the candidate marker gases at each time point, calculate the concentration coefficient of variation and complexity, adjust the number of filtering time points accordingly, and perform sliding average filtering to obtain the filtered concentration values ​​of the candidate marker gases at each detection area at different time points. S2. Analyze the filtered concentration values ​​based on a preset pest-associated gas database to identify marker gases. Based on the spatial location of different detection locations, the vertical distance between each location and the central axis of the trunk is calculated as the detection-axis distance. Combined with the average filtered concentration of the marker gas at each detection location over the detection period, the priority coefficient for each detection location is calculated to represent the degree of pest risk. S3. Arrange the priority coefficients of all detection parts in descending order, filter out the detection parts with the top m priority coefficients, take the detection part with the highest priority as the reference point, set the critical distance with the reference point as the center, measure the straight-line distance between the other m-1 detection parts and the reference point as the detection-reference distance, compare the detection-reference distance with the critical distance one by one, and determine the number of infection points of pests and diseases based on the comparison results.

2. The tree pest and disease detection method based on electronic nose and hollow structure recognition according to claim 1 is characterized in that: Normalize the concentration of the candidate marker gas during the detection period, then calculate the mean of the normalized candidate marker gas concentration, and calculate the concentration variation coefficient and the complexity of the candidate marker gas according to the following formula: ； ； ； in, is the complexity of the candidate marker gas, is the number of candidate marker gases, For the The concentration variation coefficient of the candidate marker gas, For the The standard deviation of the concentration of the candidate marker gas in the detection time period, For the The mean of the concentrations of the candidate marker gases, is the mean of the concentration variation coefficients of all candidate marker gases, is the index of the candidate marker gas, ; Adjust the filter window size according to the complexity of the candidate marker gas: ； in, is the number of filtering time points, is the reference value of the number of filtering time points, is the first level threshold of complexity, is the second level threshold of complexity.

3. The tree pest and disease detection method based on electronic nose and hollow structure recognition according to claim 2 is characterized in that: The specific steps to obtain the filtered concentration value are as follows: For each candidate marker gas at each detection location, within its corresponding detection time period, for each time point, its concentration value is calculated relative to the nearest concentration value at that time point. The average of the concentration values ​​at the time points is used as the filtered concentration value at that time point.

4. The tree pest and disease detection method based on electronic nose and hollow structure recognition according to claim 3 is characterized in that: The marker gases are screened out based on the preset pest-related gas database. The specific steps are as follows: From the candidate marker gases, retain the gases that match the types of pest gases in the database; Calculate the filtered mean concentration of these gases during the detection period. If the mean falls within the gas concentration range corresponding to the pests and diseases in the database, the gas is listed as a candidate gas. For a gas to be selected, if its filtered concentration value exceeds the preset threshold at more than three consecutive time points during the detection period, it will be included in the pre-selected marker gas set; From the preselected marker gas set, the gas with the highest average concentration after filtering is selected as the final marker gas.

5. The tree pest and disease detection method based on electronic nose and hollow structure recognition according to claim 1 is characterized in that: Based on the spatial position of different detection parts, the vertical distance between them and the central axis of the trunk is obtained. The specific steps are as follows: Set up a cross section every 50 cm along the trunk upward direction, and measure the coordinates of the center of the cross section at different heights of the trunk; Connect the centers of the cross sections at each height to form a virtual straight line, which is the central axis of the trunk; Based on the spatial positions of different detection parts, determine the coordinates of each detection part in space; According to the distance formula from the spatial geometric midpoint to the straight line, the vertical distance from the coordinates of each detection part to the central axis of the trunk is calculated.

6. The tree pest and disease detection method based on electronic nose and hollow structure recognition according to claim 5, characterized in that: According to the filtered mean concentration of the marker gas at each detection location during the detection period and the detection-axis distance, the priority coefficient of the pest risk at different detection locations is obtained according to the following formula: ； in, For the Priority coefficient of pest and disease risk of each detection part, For the The average concentration of the marker gas at each detection location after filtering during the detection time period, For the The detection-axis distance of each detection part, is the index of the detection part, , is the number of test sites; Where, is the weight coefficient of the filtered mean concentration of the marker gas at the detection site, is the weight coefficient of the inspection-axis distance, On the basis of ; Arrange the priority coefficients of all detection parts in descending order and select the detection parts with the top m priority coefficients. The specific steps are as follows: Set the sorting rules: Rule 1: Sort by priority coefficient from large to small; Rule 2: If there are multiple detection locations with the same priority coefficient, they are sorted by the average concentration after filtering by the marker gas, with the location with the highest average concentration being ranked first. According to rule 1, all detection parts are sorted in descending order of priority coefficient to obtain the preliminary sorting results; Check the preliminary sorting results. If there are detection locations with the same priority coefficient, extract the filtered mean concentration of the marker gas from these detection locations, perform secondary sorting according to Rule 2, adjust the order, and select the top m detection locations.

7. The tree pest and disease detection method based on electronic nose and hollow structure recognition according to claim 6, characterized in that: Compare the inspection-base distance with the critical distance one by one, and determine the number of infection points of pests and diseases based on the comparison results. The specific steps are as follows: Compare the detection-base distance with the set critical distance one by one, and count the number of detection locations where the detection-base distance is greater than the critical distance, which is recorded as ; like , that is, if all the inspection-base distances are less than or equal to the critical distance, then it is determined that there is one pest infection point on the tree; like , that is, there is at least one detection-base distance greater than the critical distance, then it is determined that there are multiple pest and disease infection points on the tree.

8. A tree pest and disease detection system based on electronic nose and hollow structure recognition, the system being used to execute the tree pest and disease detection method based on electronic nose and hollow structure recognition according to any one of claims 1 to 7, characterized in that: include: The data acquisition module is used to fit the hollow structure with a built-in electronic nose to the detection area of ​​the branch, locate the spatial position of different detection areas, and collect candidate marker gases at the detection area during the detection time period. The type and concentration of the candidate marker gases at each time point are extracted, and the concentration coefficient of variation and complexity are calculated. The number of filtering time points is adjusted accordingly. After sliding average filtering, the filtered concentration value of the candidate marker gas at each detection area at different time points is obtained; The data processing module is used to analyze the filtered concentration values ​​according to the preset pest-associated gas database to screen out the marker gases. Based on the spatial location of different detection sites, the vertical distance between the site and the main trunk axis is obtained as the detection-axis distance. The priority coefficient of each detection site is obtained by combining the average filtered concentration of the marker gas at each detection site within the detection time period, which is used to characterize the degree of pest risk. The judgment module is used to arrange the priority coefficients of all detection parts in descending order, screen out the detection parts with the top m priority coefficients, take the detection part with the highest priority as the reference point, set the critical distance with the reference point as the center, measure the straight-line distance between the other m-1 detection parts and the reference point, compare them one by one with the critical distance, and judge the number of infection points of pests and diseases based on the comparison results.

Images