A control method and system of an automatic sterilization ploughing machine

Abstract

The application discloses a kind of control method and system of automatic sterilization ploughing machine, belong to intelligent agricultural equipment field.The method includes: receiving ploughing path timing information of ploughing machine, obtain fungicide nozzle coordinate timing information;Randomly configure fungicide spraying flow, pressure, angle timing information, execute spray surface prediction, obtain the coverage constraint surface timing information with fungicide nozzle coordinate as origin;Based on nozzle coordinate timing information, in combination with coverage constraint surface timing information, execute union analysis, obtain fungicide coverage prediction area;When the first area intersection ratio of coverage prediction area and target coverage area is greater than or equal to the first area intersection ratio threshold, according to ploughing path, spraying flow, pressure, angle timing information executes automatic sterilization ploughing machine control.Through accurate prediction and intelligent control, improve the intelligent degree of fungicide spraying control, realize accurate coverage control, to improve the technical effect of fungicide utilization rate.

Description

Technical Field

[0001] This invention relates to the field of intelligent agricultural equipment, specifically to a control method and system for an automatic sterilizing ploughing machine. Background Technology

[0002] With the continuous improvement of agricultural mechanization, ploughing machines, as important agricultural machinery, are playing an increasingly important role in farmland operations. To prevent soil-borne diseases and increase crop yields, simultaneous soil sterilization during ploughing has become a crucial technical means in modern precision agriculture.

[0003] Existing soil disinfection technology for plowing machines primarily employs a fixed-parameter spraying method for fungicides, uniformly spraying according to preset flow rates, pressures, and angles. This traditional approach to fungicide spraying control has a low level of intelligence, failing to dynamically adjust based on the actual plowing path and plot characteristics. This results in a mismatch between spraying parameters and actual needs, hindering precise control based on actual coverage requirements and the ability to effectively predict and analyze the actual coverage area. Consequently, it easily leads to problems such as repeated or missed spraying. These technical deficiencies directly result in low fungicide utilization, increasing agricultural production costs, potentially imposing unnecessary environmental burdens, and affecting the consistency and reliability of soil disinfection effects. Summary of the Invention

[0004] The purpose of this invention is to provide a control method and system for an automatic sterilizing and ploughing machine, so as to solve the problems mentioned in the background art, such as the low level of intelligence in the control of fungicide spraying and the inability to adjust according to actual coverage needs, resulting in low utilization rate of fungicide.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] In a first aspect, the present invention provides a control method for an automatic sterilizing ploughing machine, comprising: receiving ploughing path timing information of the ploughing machine, wherein the ploughing path of the ploughing machine has fungicide nozzle coordinate timing information; randomly configuring fungicide spray flow rate timing information, fungicide spray pressure timing information, and fungicide spray angle timing information, performing simultaneous sequential spray surface prediction to obtain coverage constraint surface timing information, wherein the coverage constraint surface is a horizontal plane constructed from a set of spatial coordinates with the fungicide nozzle coordinates as the origin; performing a union analysis of the coverage constraint surface based on the fungicide nozzle coordinate timing information and the coverage constraint surface timing information to obtain a fungicide coverage prediction area; when the first area intersection-union ratio of the fungicide coverage prediction area and the target coverage area is greater than or equal to a first area intersection-union ratio threshold, performing control of the automatic sterilizing ploughing machine according to the ploughing path timing information, the fungicide spray flow rate timing information, the fungicide spray pressure timing information, and the fungicide spray angle timing information.

[0007] Secondly, the present invention provides a control system for an automatic sterilizing tillage machine, comprising: a path receiving module for receiving tillage path timing information of the tillage machine, wherein the tillage path of the tillage machine includes sterilizing agent nozzle coordinate timing information; and a parameter configuration module for randomly configuring sterilizing agent spray flow rate timing information, sterilizing agent spray pressure timing information, and sterilizing agent spray angle timing information, performing simultaneous sequential spray surface prediction, and obtaining coverage constraint surface timing information, wherein the coverage constraint surface is a horizontal surface constructed from a spatial coordinate set with the sterilizing agent nozzle coordinates as the origin. The system includes: a sterilization prediction module, used to perform a union analysis of the coverage constraint surface based on the timing information of the sterilizer nozzle coordinates and the timing information of the coverage constraint surface, to obtain the sterilizer coverage prediction area; and a control execution module, used to control the automatic sterilization ploughing machine according to the timing information of the ploughing path, the timing information of the sterilizer spraying flow rate, the timing information of the sterilizer spraying pressure, and the timing information of the sterilizer spraying angle when the first area intersection-union ratio of the sterilizer coverage prediction area and the target coverage area is greater than or equal to the first area intersection-union ratio threshold.

[0008] The beneficial effects that this application can achieve are as follows:

[0009] The system receives the timing information of the tilling path of a tiller, which includes the timing information of the fungicide nozzle coordinates. By acquiring the real-time path and nozzle position information of the tiller, it provides basic data support for subsequent precise control. It then randomly configures the timing information of fungicide spray flow rate, spray pressure, and spray angle, and performs simultaneous spray surface prediction to obtain the timing information of the coverage constraint surface. This coverage constraint surface is a horizontal plane constructed from a set of spatial coordinates with the fungicide nozzle coordinates as the origin. Through random configuration and predictive analysis of multiple parameters, it can simulate the coverage effect under different combinations of spray parameters, providing a data foundation for intelligent control decisions. Based on the timing information of the fungicide nozzle coordinates... By combining the timing information of the coverage constraint surface, a union analysis of the coverage constraint surface is performed to obtain the predicted coverage area of ​​the fungicide, thereby accurately calculating the actual coverage range of the fungicide and avoiding the defect of traditional fixed spraying methods that cannot predict the coverage effect. When the first area intersection ratio of the predicted coverage area and the target coverage area is greater than or equal to the first area intersection ratio threshold, the automatic sterilization ploughing machine is controlled according to the timing information of the ploughing path, the timing information of the fungicide spraying flow rate, the timing information of the fungicide spraying pressure, and the timing information of the fungicide spraying angle. By setting the coverage rate threshold for intelligent judgment and automatic control, the fungicide spraying achieves the expected coverage effect while realizing the intelligence and automation of the control process.

[0010] The above technical solutions enable accurate prediction, intelligent judgment, and automatic control of fungicide spraying, effectively improving the intelligence level of fungicide spraying control and the utilization rate of fungicide. Attached Figure Description

[0011] Figure 1 A flowchart illustrating the control method for an automatic sterilizing ploughing machine provided by the present invention;

[0012] Figure 2 This is a schematic diagram of the control system of an automatic sterilizing ploughing machine provided by the present invention.

[0013] In the attached diagram, the components represented by each number are as follows:

[0014] The system includes a path receiving module 11, a parameter configuration module 12, a sterilization prediction module 13, and a control execution module 14.

[0015] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1, as Figure 1 As shown, an embodiment of the present invention provides a control method for an automatic sterilizing and tilling machine, comprising:

[0018] S1. Receive the timing information of the tilling path of the tiller, wherein the tilling path of the tiller includes the timing information of the coordinates of the fungicide nozzle.

[0019] Specifically, firstly, the system receives the plowing path timing information of the plowing machine during operation. This plowing path timing information refers to the path trajectory data recorded in chronological order when the plowing machine is performing plowing operations. This data includes the position information of the plowing machine at different times, forming the plowing path of the plowing machine.

[0020] Specifically, the tilling path of the tiller contains time-series coordinate information of the fungicide nozzles. That is, at each time point along the tilling path, the spatial coordinates of the fungicide nozzles mounted on the tiller are recorded. These coordinates are arranged chronologically, forming the motion trajectory data of the fungicide nozzles. The acquisition of the tilling path time-series information and the fungicide nozzle coordinate time-series information can be achieved through GPS positioning systems, inertial navigation systems, or other position sensors, ensuring that the position coordinates of the tiller and its onboard fungicide nozzles in three-dimensional space can be obtained in real time and accurately.

[0021] By receiving this timing information, the movement status of the fungicide nozzles can be monitored throughout the entire plowing process, providing basic data support for subsequent precise fungicide spraying control.

[0022] S2. Randomly configure the timing information of the spray flow rate, spray pressure, and spray angle of the fungicide, and perform simultaneous sequential spray surface prediction to obtain the timing information of the coverage constraint surface. The coverage constraint surface is a horizontal plane constructed from a set of spatial coordinates with the coordinates of the fungicide nozzle as the origin.

[0023] Specifically, the parameters for fungicide spraying are randomly configured, and predictive analysis of the spray coverage is performed based on these parameters. Specifically, three types of time-series information for spraying parameters are randomly configured: first, the time-series information for fungicide spray flow rate, i.e., the spray flow rate value of the fungicide at different times; second, the time-series information for fungicide spray pressure, i.e., the working pressure value at different times; and third, the time-series information for fungicide spray angle, i.e., the spray angle setting of the nozzle at different times. The random configuration of these three types of parameters provides a diverse parameter combination space for subsequent optimization algorithms.

[0024] Based on the aforementioned randomly configured parameter combinations, simultaneous spraying area prediction is performed. This simultaneous spraying area prediction refers to calculating and predicting the spatial area that the fungicide can cover at the current moment based on the spraying parameter settings. Through this simultaneous prediction analysis, the temporal information of the coverage constraint surface can be obtained. This coverage constraint surface is a horizontal plane constructed from the set of spatial coordinates with the fungicide nozzle coordinates as the origin. In other words, the coverage constraint surface is a two-dimensional planar region centered on the nozzle position, and all coordinate points within this region constitute the theoretical coverage range that the fungicide can reach under specific spraying parameters.

[0025] By establishing coverage constraints, the coverage effect corresponding to different combinations of spraying parameters can be quantified, providing a foundation for subsequent precise control and optimization. This dynamic prediction method based on time-series information can better adapt to the actual movement trajectory of the tiller and changes in the working environment compared to the traditional static spraying mode.

[0026] S3. Based on the time sequence information of the disinfectant nozzle coordinates and combined with the time sequence information of the coverage constraint surface, perform a union analysis of the coverage constraint surface to obtain the predicted area of ​​disinfectant coverage.

[0027] Specifically, based on the aforementioned time-series information of the fungicide nozzle coordinates and the time-series information of the coverage constraint surface, a union analysis of the coverage constraint surface is performed to obtain the predicted area of ​​fungicide coverage.

[0028] This union analysis process refers to performing a mathematical union operation on the coverage constraint surfaces at different times. Since the fungicide nozzles move continuously along the plowing path during the plowing operation, at each time t, the nozzle is located at a specific spatial coordinate position, corresponding to a coverage constraint surface with that coordinate as its origin. Over time, these coverage constraint surfaces form a series of coverage areas distributed in different locations in space.

[0029] By performing a union analysis of coverage constraint surfaces, the coverage constraint surfaces at all times are merged, which means superimposing and fusing the coverage areas at each time point in the spatial coordinate system. This union operation can eliminate the overlapping parts between coverage areas at different times, while retaining all covered spatial coordinate points, ultimately forming a complete coverage area without duplicate calculations.

[0030] The predicted fungicide coverage area, obtained after union analysis, represents the overall area that the fungicide can cover during the entire plowing operation under the current spraying parameter configuration. This predicted fungicide coverage area provides the basic data for subsequent comparative analysis with the target coverage area.

[0031] By using a dynamic union analysis method based on time-series information, compared with the traditional static coverage calculation method, the coverage effect during the actual spraying process can be reflected more accurately, providing support for improving the utilization rate of fungicides.

[0032] S4. When the first area intersection ratio between the predicted area covered by the fungicide and the target area is greater than or equal to the first area intersection ratio threshold, the automatic sterilization ploughing machine is controlled according to the ploughing path timing information of the ploughing machine, the spraying flow timing information of the fungicide, the spraying pressure timing information of the fungicide, and the spraying angle timing information of the fungicide.

[0033] Specifically, after obtaining the predicted area for fungicide coverage, the first step is to calculate the first area intersection-union ratio (IUU) between the predicted and target coverage areas. This IUU is an indicator that measures the degree of matching between the predicted coverage effect (predicted fungicide coverage area) and the expected coverage target (target coverage area). The formula is the ratio of the intersection area to the union area of ​​the predicted and target coverage areas. This ratio ranges from 0 to 1, with a value closer to 1 indicating a higher degree of matching between the two areas. The target coverage area refers to the land area that requires fungicide treatment, pre-defined according to agronomic requirements. This area is typically determined by an expert group based on factors such as soil conditions, crop planting plans, or pest and disease distribution.

[0034] When the calculated first area intersection ratio is greater than or equal to the preset first area intersection ratio threshold, it indicates that the current spraying parameter configuration can meet the coverage accuracy requirements, and it is determined that the actual spraying control operation can be executed. At this time, based on the aforementioned plowing path timing information, fungicide spraying flow rate timing information, fungicide spraying pressure timing information, and fungicide spraying angle timing information, the control of the automatic sterilizing plowing machine is executed. This control process specifically includes: controlling the plowing machine's travel trajectory according to the plowing path timing information, and simultaneously controlling the fungicide spraying working state according to the fungicide spraying flow rate timing information, fungicide spraying pressure timing information, and fungicide spraying angle timing information, ensuring accurate fungicide spraying while plowing.

[0035] This judgment mechanism based on area intersection ratio ensures that actual spraying is only performed when the coverage effect meets the expected standard, thus avoiding the waste of fungicide caused by inaccurate coverage in traditional methods and significantly improving the utilization rate of fungicide.

[0036]

[0037] Table 1. Comparison of the effects of traditional fixed spraying methods and the spraying method of this application.

[0038] As can be seen from Table 1 above (Comparison of the effects of traditional fixed spraying methods and the spraying method of this application), the spraying method provided in this application embodiment achieves significant improvements in multiple indicators compared to the traditional fixed spraying method. Regarding fungicide utilization, the traditional method, unable to dynamically adjust according to actual terrain and operating conditions, results in a utilization rate of only 65-75%. However, through intelligent time-series control and precise coverage prediction, the utilization rate is increased to 88-95%, an improvement of 20-30%. In terms of coverage accuracy, the traditional fixed spraying method is limited by the rigid constraints of preset parameters, with coverage accuracy generally within the range of 70-80%. However, through the area intersection and comparison judgment mechanism and dynamic parameter optimization, the coverage accuracy is improved to 92-98%, ensuring the comprehensiveness and effectiveness of the disinfection treatment. Simultaneously, the embodiments of this application significantly reduce the rate of repeated spraying and missed spraying. The traditional method has a repeated spraying rate as high as 15-25% and a missed spraying rate of 8-15%, causing serious resource waste and uneven treatment. Through precise coverage prediction and intelligent control, the rate of repeated spraying is reduced to 3-8%, and the rate of missed spraying is reduced to 1-3%, achieving reductions of 60-80% and 70-85% respectively.

[0039] Overall, the embodiments of this application achieve a 15-25% reduction in fungicide consumption, solving the technical problems of low intelligence in fungicide spraying control and inability to accurately control according to actual coverage needs in the prior art. This achieves the technical effect of improving the intelligence of fungicide spraying control, realizing precise coverage control, and thus improving the utilization rate of fungicide.

[0040] Furthermore, the tillage path timing information of the tillage machine is received, wherein the tillage path of the tillage machine includes fungicide nozzle coordinate timing information, and the tillage path timing information characterizes the tillage shovel coordinate timing information, including:

[0041] S11. Obtain the spatial position of the plow blade and the spatial position of the disinfectant nozzle for the preset sterilization plow model;

[0042] S12. Using the spatial position of the plow shovel as the origin, and combining it with the spatial position of the fungicide nozzle, construct the relative distribution position of the fungicide nozzle;

[0043] S13. Associate and store the preset sterilizing ploughing machine model with the relative distribution position of the sterilizing agent nozzles to construct a sterilizing agent nozzle coordinate identification table;

[0044] S14. Input the model of the sterilizing ploughing machine into the coordinate identification table of the sterilizing agent nozzles, output the relative distribution position identification of the sterilizing agent nozzles, convert the timing information of the ploughing path of the ploughing machine, and obtain the timing information of the coordinates of the sterilizing agent nozzles.

[0045] In a preferred embodiment, by establishing a mapping relationship between the machine model and the distribution position of the spray nozzles, the accurate conversion from the tillage path timing information of the tiller to the coordinate timing information of the fungicide spray nozzles can be achieved.

[0046] First, obtain the spatial location information of the components of the preset sterilizing ploughing machine model, including the spatial location of the ploughing blade and the spatial location of the fungicide nozzles. The preset sterilizing ploughing machine model refers to a specific sterilizing ploughing machine model that is pre-configured and supported. This model includes standardized information such as the equipment's structural parameters and component configuration. The spatial location of the ploughing blade refers to its three-dimensional coordinate position within the overall structure of the ploughing machine, typically serving as the positioning reference for the main working components. The spatial location of the fungicide nozzles refers to the three-dimensional coordinate position of each fungicide nozzle within the ploughing machine structure; these nozzles are usually distributed at different locations within the ploughing machine to achieve comprehensive coverage. Then, establish a relative coordinate system with the ploughing blade spatial location as the origin. Combined with the fungicide nozzle spatial locations, construct the relative distribution position of the fungicide nozzles. This relative distribution position describes the spatial positional relationship of each fungicide nozzle on the ploughing blade of the preset sterilizing ploughing machine model relative to the ploughing blade, including geometric parameters such as relative distance, azimuth angle, and height difference. By establishing this relative positional relationship, the absolute spatial coordinates of each nozzle can be accurately calculated given the location of the ploughing shovel. Subsequently, the preset sterilizing ploughing machine model and the corresponding relative distribution positions of the fungicide nozzles are associated and stored, constructing a fungicide nozzle coordinate identification table. This fungicide nozzle coordinate identification table is a database structure that uses the preset sterilizing ploughing machine model as the index key to store the nozzle distribution configuration information for the corresponding model. Different ploughing machine models may have different nozzle distribution patterns due to structural design differences; by establishing this identification table structure, compatibility with multiple model devices can be supported.

[0047] Subsequently, the location information of the fungicide nozzles can be obtained through model matching and coordinate transformation. Specifically, the model of the currently used fungicide plowing machine is input into the fungicide nozzle coordinate identification table, and the corresponding relative distribution position identification of the fungicide nozzles is output. Based on this identification information, coordinate transformation calculation is performed on the plowing path timing information of the plowing machine, that is, the movement trajectory of the plowing shovel is converted into the movement trajectory of each nozzle, and finally the coordinate timing information of the fungicide nozzles is obtained.

[0048] By using a coordinate transformation method based on relative positional relationships, not only is the versatility and scalability of the system improved, but the accuracy of the nozzle position information is also ensured, providing reliable basic data support for subsequent precise spraying control.

[0049] Furthermore, by randomly configuring the timing information of fungicide spray flow rate, fungicide spray pressure, and fungicide spray angle, and performing simultaneous spray surface prediction, the timing information of the coverage constraint surface is obtained, including:

[0050] S21. Configure multiple sets of data according to the preset sterilization plow machine model. Each set of the multiple sets of data includes sterilizer spray flow rate record data, sterilizer spray pressure record data, sterilizer spray angle record data, and a label indicating the coverage constraint surface with the coordinates of the sterilizer nozzle as the origin.

[0051] S22. Using the label of the coverage constraint surface with the coordinates of the fungicide nozzle as the origin as supervision, and the fungicide spray flow rate record data, the fungicide spray pressure record data, and the fungicide spray angle record data as input, multiple sets of data are retrieved, and several coverage constraint surface predictors are constructed through machine learning.

[0052] S23. Perform union fitting on the outputs of the plurality of coverage constraint surface predictors to obtain a coverage constraint surface prediction model, and store it in association with a preset sterilization and ploughing machine model to construct a coverage constraint surface prediction model library.

[0053] S24. Using the coverage constraint surface prediction model library, based on the model of the bactericidal tillage machine, perform simultaneous spraying surface prediction on the timing information of the bactericidal spraying flow rate, the timing information of the bactericidal spraying pressure, and the timing information of the bactericidal spraying angle to obtain the timing information of the coverage constraint surface.

[0054] In a preferred embodiment, firstly, a training dataset is configured according to a preset sterilizing and tilling machine model. Specifically, multiple sets of data are configured for the preset sterilizing and tilling machine model, each set constituting a complete training sample. Each set of data includes four key elements: first, fungicide spray flow rate recording data, recording specific values ​​under different flow rate settings; second, fungicide spray pressure recording data, recording specific values ​​under different pressure settings; third, fungicide spray angle recording data, recording specific values ​​under different angle settings; and fourth, a label identifying the coverage constraint surface with the fungicide nozzle coordinates as the origin, which records the actual measured boundary coordinates of the coverage area under specific parameter combinations.

[0055] Then, a coverage constraint surface predictor is constructed using supervised learning. The label of the coverage constraint surface, with the origin at the fungicide nozzle coordinates, is used as the supervision signal. The fungicide spray flow rate record, fungicide spray pressure record, and fungicide spray angle record are used as input features. Multiple sets of data are retrieved for machine learning training. Several coverage constraint surface predictors are constructed using machine learning algorithms. For example, a multilayer perceptron (MLP) network is used, specifically including: an input layer containing three neurons, corresponding to the three input features of fungicide spray flow rate record, fungicide spray pressure record, and fungicide spray angle record; a hidden layer with a 2-3 layer structure, each layer containing 32-64 neurons, using the ReLU activation function to handle the nonlinear feature relationships in the multiple sets of data; and an output layer with a regression structure, outputting the coverage constraint surface boundary coordinate parameters corresponding to each set of data labels. A Support Vector Regression (SVR) model was employed, using a Radial Basis Function (RBF) kernel function to establish a nonlinear mapping relationship between fungicide spray flow rate records, fungicide spray pressure records, fungicide spray angle records, and coverage constraint surface labels from multiple datasets. Specifically, the kernel parameter γ was set to a range of 0.1-1.0 to accommodate the feature distribution of different datasets; the regularization parameter C was set to a range of 10-100 to balance model complexity and fitting accuracy; and the tolerance error ε was set to a range of 0.001-0.01 to control the fitting accuracy of the label data. A Random Forest Regression model was then used, specifically involving the construction of 50-200 decision trees based on multiple datasets. Each decision tree was trained using randomly selected dataset samples, with a maximum depth of 5-15 to accommodate the complexity of different datasets. The minimum number of leaf nodes was set to 3-10 to ensure effective learning of the label data from each dataset. Prediction results from each dataset were randomly selected from the three spray parameters and integrated to improve the accuracy and stability of the coverage constraint surface label prediction. By constructing these machine learning models with different architectures, we can effectively learn the complex mapping relationship between spraying parameters and coverage constraint surface labels in multiple sets of data, and obtain several coverage constraint surface predictors to accurately predict the coverage effect under different parameter combinations.

[0056] Subsequently, several cover constraint surface predictors were integrated and fused. A model fusion strategy was employed to perform union fitting on the outputs of multiple cover constraint surface predictors. The prediction results from different predictors were effectively fused through methods such as weighted averaging, voting mechanisms, or stacking fusion. Specifically, corresponding weight coefficients were assigned to each cover constraint surface predictor based on its performance, with higher-performing predictors receiving higher weights. Finally, a cover constraint surface prediction model was obtained through weighted fusion. This model combines the advantages of multiple individual predictors, improving prediction accuracy and the model's generalization ability. The obtained cover constraint surface prediction model was then associated and stored with the corresponding preset sterilizing and tilling machine models to construct a cover constraint surface prediction model library. Different models of sterilizing and tilling machines exhibit significantly different spraying characteristics due to differences in structural parameters, nozzle configurations, etc. Therefore, it is necessary to establish model-specific prediction models to ensure prediction accuracy and applicability.

[0057] Subsequently, in practical applications, the coverage constraint surface prediction model library is used to first identify the model of the currently used sterilizing and tilling machine, and then call the corresponding coverage constraint surface prediction model. For the input fungicide spray flow rate time series information, fungicide spray pressure time series information, and fungicide spray angle time series information, these are input into the coverage constraint surface prediction model one by one in chronological order, and simultaneous sequential spray surface prediction calculations are performed. Through this prediction mechanism, the corresponding coverage constraint surface can be quickly and accurately predicted at each moment based on the current spraying parameter settings, obtaining complete coverage constraint surface time series information, providing data support for subsequent coverage area analysis and control decisions.

[0058] Compared to traditional empirical formulas or fixed-parameter methods, machine learning-based prediction methods can more accurately establish the complex nonlinear relationship between spraying parameters and coverage effect, thus improving the accuracy and reliability of coverage prediction.

[0059] Furthermore, the outputs of the plurality of coverage constraint surface predictors are subjected to union fitting to obtain a coverage constraint surface prediction model, including:

[0060] S231. Constructing the union fitting rule:

[0061] S232. Obtain a plurality of predicted coverage constraint surfaces from the plurality of coverage constraint surface predictors;

[0062] S233. Perform coordinate triggering frequency statistics on the several predicted coverage constraint surfaces to obtain multiple coordinate triggering frequencies;

[0063] S234. Based on the multiple coordinate trigger frequencies, extract the coordinates in the multiple prediction coverage constraint surfaces whose coordinate trigger frequencies are greater than or equal to the coordinate trigger frequency threshold, and add them to the union coordinates.

[0064] S235. Construct a fitting coverage constraint surface based on the union coordinates;

[0065] S236. Based on the union fitting rule, perform union fitting on the outputs of the plurality of coverage constraint surface predictors to obtain the coverage constraint surface prediction model.

[0066] In a preferred embodiment, firstly, a union fitting rule is constructed. This union fitting rule defines the fusion strategy and judgment criteria for the outputs of multiple coverage constraint surface predictors, including the principle for setting the coordinate trigger frequency threshold, the rule for extracting the union coordinates, and the method for constructing the fitted coverage constraint surface. The union fitting rule provides technical standards and operational specifications for the subsequent fusion process of multiple coverage constraint surface predictors. Then, several predicted coverage constraint surfaces are obtained from several coverage constraint surface predictors. Specifically, for the same set of spraying parameter inputs, different coverage constraint surface predictors (such as MLP networks, SVR models, random forest models, etc.) output their respective prediction results, and each coverage constraint surface predictor generates a predicted coverage constraint surface, resulting in several predicted coverage constraint surfaces. These predicted coverage constraint surfaces represent different region boundaries in the spatial coordinate system. Due to the differences in the algorithm principles and model structures of different predictors, their prediction results may have certain differences.

[0067] Then, coordinate trigger frequency statistics are performed on several predicted coverage constraint surfaces. This statistical process involves dividing the spatial coordinate system into grid cells and counting the number of times each grid coordinate appears in different predicted coverage constraint surfaces, i.e., the frequency of being covered by different predicted coverage constraint surfaces. By traversing all predicted coverage constraint surfaces, the number of predicted coverage constraint surfaces that determine each coordinate point as a point within the coverage area is calculated, thus obtaining multiple coordinate trigger frequencies. Subsequently, a filtering process is performed based on the coordinate trigger frequencies. Specifically, coordinate points in several predicted coverage constraint surfaces whose coordinate trigger frequencies are greater than or equal to a preset coordinate trigger frequency threshold are extracted, and these coordinate points are added to the union coordinates. The coordinate trigger frequency threshold is typically set to a certain proportion of the coverage constraint surface predictors; for example, if more than half of the coverage constraint surface predictors believe that a coordinate should be covered, that coordinate point is included in the union coordinates. This coordinate filtering method can effectively eliminate abnormal prediction results from individual coverage constraint surface predictors and improve the reliability of the fusion results.

[0068] Next, a fitted coverage constraint surface is constructed based on the union coordinates. By performing boundary fitting and region reconstruction on the selected union coordinates, a fitted coverage constraint surface that integrates the advantages of multiple coverage constraint surface predictors is constructed. This fitting process can employ geometric algorithms such as convex hull algorithm, boundary smoothing, or region filling to ensure that the generated coverage constraint surface has a reasonable geometric shape and continuity. Subsequently, the final model is constructed based on the union fitting rules. The above union fitting process is formalized into a unified coverage constraint surface prediction model, which can receive spraying parameter inputs and output coverage constraint surface prediction results fused from multiple predictors. The coverage constraint surface prediction model obtained through this union fitting method has higher prediction accuracy and stronger robustness compared to a single coverage constraint surface predictor.

[0069] By using a union fitting method based on coordinate trigger frequency statistics, the prediction advantages of multiple predictors are effectively integrated, the prediction error of a single coverage constraint surface predictor is reduced, and the accuracy and reliability of coverage prediction are improved.

[0070] Furthermore, based on the time-series information of the fungicide nozzle coordinates, combined with the time-series information of the coverage constraint surface, a union analysis of the coverage constraint surface is performed to obtain the predicted fungicide coverage area, including:

[0071] S31. Based on the time sequence information of the bactericide nozzle coordinates and combined with the time sequence information of the coverage constraint surface, perform simultaneous time sequence coverage constraint surface analysis to obtain the set of spray coverage areas.

[0072] S32. Perform a region intersection-union ratio analysis on the set of sprayed coverage areas to obtain a second area intersection-union ratio;

[0073] S33. When the second area intersection ratio is greater than or equal to the second area intersection ratio threshold, update the timing information of the fungicide spraying flow rate, the timing information of the fungicide spraying pressure, and the timing information of the fungicide spraying angle and execute the loop.

[0074] S34. Otherwise, perform a union analysis of the coverage constraint surface on the set of sprayed coverage areas to obtain the predicted coverage area of ​​the bactericide.

[0075] In a preferred embodiment, firstly, based on the time-series information of the fungicide nozzle coordinates and combined with the time-series information of the coverage constraint surface, a simultaneous coverage constraint surface analysis is performed. Specifically, the coordinate position of the fungicide nozzle at each moment is matched and combined with the coverage constraint surface at the corresponding moment, and a time synchronization mechanism ensures a one-to-one correspondence between the fungicide nozzle coordinate information and the coverage constraint surface information. At each time node t, the fungicide nozzle coordinates are used as the positioning origin of the coverage constraint surface to generate the actual spraying coverage area at that moment. By traversing the entire time-series process, a series of spraying coverage areas arranged in chronological order are obtained, forming a set of spraying coverage areas. This set reflects the coverage status of the fungicide spraying at each moment during the entire operation.

[0076] Subsequently, a region crossover and union ratio analysis was performed on the sprayed coverage area set to assess the degree of overlap between covered areas at different times. The ratio of the intersection area to the union area between covered areas at all times was calculated to obtain a second area crossover and union ratio. This second area crossover and union ratio reflects the degree of overlapping coverage during the spraying process: a high second area crossover and union ratio indicates more overlapping spraying, which may lead to waste of fungicide; a low second area crossover and union ratio indicates less overlap between covered areas and relatively higher spraying efficiency. Through this quantitative analysis, the rationality of the current spraying parameter configuration can be objectively assessed.

[0077] When the second area overlap ratio is greater than or equal to a preset second area overlap ratio threshold, it indicates that the current spraying parameter configuration has led to excessive overlapping coverage, resulting in low fungicide utilization. At this point, a parameter optimization mechanism is triggered, updating the fungicide spray flow rate timing information, fungicide spray pressure timing information, and fungicide spray angle timing information, and returning to step S2 to perform iterative optimization. This iterative process reduces overlapping coverage and improves fungicide utilization efficiency by adjusting the spraying parameters. Iteration continues until a parameter combination that meets the overlap requirements is found.

[0078] When the second area intersection-union ratio is less than the second area intersection-union ratio threshold, it indicates that the current spraying parameter configuration is reasonable in terms of repeated coverage control and there is no excessive overlap problem. At this point, a coverage constraint surface union analysis is performed on the sprayed coverage area set. The sprayed coverage areas at each time point are spatially unioned to eliminate overlapping parts, retaining all covered spatial coordinate points, and finally obtaining the fungicide coverage prediction area. This fungicide coverage prediction area represents the overall area that fungicide spraying can cover under the optimized spraying parameter configuration.

[0079] Dynamic optimization based on overlap detection ensures that the generation process of the predicted fungicide coverage area avoids resource waste caused by excessive repeated spraying while guaranteeing the integrity and effectiveness of the coverage. A threshold control mechanism based on the second area intersection-to-union ratio can find the optimal balance between coverage effect and resource utilization, thereby improving the utilization efficiency of the fungicide.

[0080] Furthermore, the embodiments of this application also include: when the first area intersection ratio between the predicted area covered by the fungicide and the target area is less than the first area intersection ratio threshold, updating the timing information of the fungicide spraying flow rate, the timing information of the fungicide spraying pressure, and the timing information of the fungicide spraying angle is performed in a loop.

[0081] In a preferred embodiment, when the first area overlap ratio between the predicted coverage area and the target coverage area is less than a first area overlap ratio threshold, it indicates that the current spraying parameter configuration cannot meet the expected coverage requirements, and there is a problem of insufficient coverage or coverage deviation. This first area overlap ratio reflects the degree of matching between the actual predicted coverage effect and the target coverage requirement. When the first area overlap ratio is lower than the first area overlap ratio threshold, it indicates that the coverage accuracy has not met agronomic requirements or operational standards.

[0082] At this point, a parameter reconfiguration mechanism is triggered to update the timing information of the fungicide spray flow rate, the timing information of the fungicide spray pressure, and the timing information of the fungicide spray angle, and then return to step S2 to re-execute the coverage prediction and analysis process. This cyclic optimization process aims to improve the coverage effect by adjusting the spraying parameters, including strategies such as increasing the spray flow rate to expand the coverage area, adjusting the spray pressure to change the coverage shape, and optimizing the spray angle to improve the coverage direction. Through the cyclic optimization mechanism based on coverage accuracy feedback, the problem of insufficient coverage can be automatically identified and corrected, ensuring that the final spraying parameter configuration can meet the requirements of the target coverage area. This mechanism complements the aforementioned repeated coverage detection mechanism (step S33), together constructing a complete two-way optimization control system: avoiding resource waste caused by excessive repeated spraying and preventing poor disinfection effect caused by insufficient coverage.

[0083] By employing a dual-judgment and cyclical optimization scheme, the intelligence level of fungicide spraying control has been improved, achieving synergistic optimization of coverage effect and resource utilization rate, ensuring that fungicide operations can both meet agronomic requirements and maximize the utilization efficiency of fungicides.

[0084] Furthermore, the process of updating the timing information of the fungicide spray flow rate, the timing information of the fungicide spray pressure, and the timing information of the fungicide spray angle in a loop includes:

[0085] S331. When the timing information of the spray flow rate of the bactericide, the timing information of the spray pressure of the bactericide, and the timing information of the spray angle of the bactericide are randomly updated a preset number of times, a number of historical bactericidal spray control particles are obtained.

[0086] S332. Constructing the fitness function:

[0087] S333. When the update is triggered by the second area crossover ratio being greater than or equal to the second area crossover ratio threshold, the fitness function is inversely proportional to the ratio of the second area crossover ratio to the second area crossover ratio threshold.

[0088] S334. When the update is triggered when the first area intersection ratio of the bactericide-covered predicted area and the target covered area is less than the first area intersection ratio threshold, the fitness function is proportional to the ratio of the first area intersection ratio to the first area intersection ratio threshold.

[0089] S335. Based on the fitness function, configure the fitness of several historical particles of the several historical sterilization spray control particles.

[0090] S336. Based on the fitness of the aforementioned historical particles, a first preset number of particles with smaller fitness are selected as natural enemy particles, and a second preset number of particles with larger fitness are selected as food particles.

[0091] S337. Using the food particles as the approaching targets and the enemy particles as the distant targets, perform Levi flight according to the long search step size constraint interval and the short search step size constraint interval to obtain an expanded particle set, and execute the loop.

[0092] In a preferred embodiment, firstly, a number of historical disinfection spray control particles are obtained through a random update mechanism. When the timing information of the disinfectant spray flow rate, disinfectant spray pressure, and disinfectant spray angle is randomly updated a preset number of times, the parameter combination of each update is recorded as a historical disinfection spray control particle. Each particle contains complete three-dimensional parameter information (flow rate, pressure, angle), representing a possible spray control strategy. Through multiple random updates, a number of historical disinfection spray control particles are obtained, forming a set of sampling points in the parameter space, providing basic data for subsequent intelligent optimization.

[0093] Then, a fitness function is constructed as a quantitative evaluation standard for the optimization objective. The fitness function measures the performance of each historical fungicide spray control particle, thus transforming the complex spray control effect into a quantifiable numerical indicator. The function value directly reflects the comprehensive performance of the corresponding parameter configuration in terms of fungicide utilization and coverage accuracy. Specifically, the fitness function is constructed using a piecewise design strategy, employing corresponding evaluation logic based on different optimization trigger conditions. The fitness function uses the area intersection-union ratio (IUU) as the evaluation index, converting the coverage effect into a fitness score by setting an ideal value target and a deviation penalty mechanism. The fitness function design follows the principle of rewarding excellence and penalizing poor performance: a higher fitness score is given when a parameter combination improves the current problem (such as reducing redundant coverage or improving coverage accuracy); a lower fitness score is given when a parameter combination exacerbates the current problem. The fitness function can also incorporate constraints and boundary limits to ensure that the parameter combinations generated during optimization are within a practically feasible range. By introducing a smooth transition mechanism and normalization processing, the fitness function can establish a reasonable mapping relationship between evaluation indicators of different orders of magnitude, avoiding optimization deviations caused by differences in numerical scales. By comprehensively considering two different optimization trigger conditions in the function design, the optimization algorithm can automatically adjust its search direction according to the specific problem it faces (over-coverage or under-coverage), achieving intelligent parameter optimization. The first trigger condition is over-coverage triggering, which occurs when the second area intersection-union ratio (IUU) is greater than or equal to the second IUU threshold. This indicates excessive overlap between sprayed areas at different times, leading to repeated spraying of fungicides, resulting in resource waste and low utilization. The optimization goal here is to reduce repeated coverage and improve fungicide utilization efficiency. The second trigger condition is insufficient coverage accuracy triggering, which occurs when the first IUU ratio between the predicted fungicide coverage area and the target coverage area is less than the first IUU threshold. This indicates a mismatch between the actual spraying coverage effect and the expected target coverage requirements, resulting in under-coverage, coverage deviation, or missed spraying. The optimization goal here is to improve coverage accuracy and ensure that the spraying range meets agronomic requirements.

[0094] Subsequently, for optimization scenarios with excessive overlapping coverage, an inverse fitness mechanism is designed. When the update process is triggered by a second area crossover ratio (OCR) greater than or equal to a second OCR threshold, it indicates that there is an issue of excessive overlapping spraying. In this case, the fitness function is inversely proportional to the ratio of the second OCR to the second OCR threshold; that is, the larger the ratio (higher degree of overlap), the smaller the fitness (worse performance). This design ensures that the optimization algorithm tends to select parameter combinations that reduce overlapping coverage, guiding the search process towards reducing overlap. For optimization scenarios with insufficient coverage, a positive fitness mechanism is designed. When the update process is triggered by a first area OCR between the fungicide coverage prediction area and the target coverage area being less than a first OCR threshold, it indicates that there is an issue of insufficient coverage accuracy. In this case, the fitness function is directly proportional to the ratio of the first OCR to the first OCR threshold; that is, the larger the ratio (higher coverage matching degree), the greater the fitness (better performance). This design ensures that the optimization algorithm tends to select parameter combinations that improve coverage accuracy, guiding the search process towards improving coverage performance.

[0095] Subsequently, the performance of historical sterilization spray control particles was evaluated based on a fitness function. By inputting the parameter information of each historical sterilization spray control particle into the fitness function, the corresponding historical particle fitness values ​​were calculated and configured, resulting in several historical particle fitness values. These historical particle fitness values ​​quantify the merits of different parameter combinations, providing a basis for subsequent particle classification and search strategies. Then, based on the ranking of the historical particle fitness values, a first preset number of particles with lower fitness were selected as enemy particles. These particles represent poor-performing parameter combinations and need to be avoided during optimization. Simultaneously, a second preset number of particles with higher fitness were selected as food particles. These particles represent superior-performing parameter combinations and are targets to be approached during optimization. Specifically, the selection strategy for enemy particles involves choosing a certain number of particles with the lowest fitness from the end of the fitness ranking results. The parameter combinations corresponding to these particles perform the worst under the current optimization objective, potentially leading to severe duplicate spraying or poor coverage. During subsequent search processes, these areas will be actively avoided to prevent the search direction from developing towards a poor solution space. The food particle selection strategy involves choosing a certain number of particles with the highest fitness from the top of the fitness ranking results. The parameter combinations corresponding to these particles perform optimally under the current optimization objective, effectively solving the problems of duplicate coverage or coverage accuracy. During the search process, the search will preferentially move closer to these regions, increasing the search density near high-quality solutions.

[0096] Next, the Lévy flight optimization algorithm is executed to achieve intelligent search of the fungicide spraying parameter space. This algorithm simulates the behavior of organisms in nature, which tend to move towards food and avoid predators, and is suitable for solving multi-objective optimization problems of fungicide spraying parameters. The core feature of the Lévy flight optimization algorithm is a step size selection mechanism with a heavy-tailed distribution, which can achieve a dynamic balance between local fine-tuning and global large-scale changes in fungicide flow rate, pressure, and angle parameters. Specifically, a bidirectional guided parameter search mechanism is constructed, with food particles representing high-quality spraying effects as the approach target and predator particles representing low-quality spraying effects as the distance target. For the current fungicide spraying parameter combination position, the attraction vector moving towards the high-quality parameter combination (food particles) and the repulsion vector moving away from the low-quality parameter combination (predator particles) are calculated, and the main direction of fungicide spraying parameter adjustment is determined by vector synthesis. Based on this, Lévy flight-style parameter updates are executed according to long search step size constraint intervals and short search step size constraint intervals. The step size distribution of Lévy flight follows a power-law characteristic, corresponding to different adjustment ranges of fungicide spraying parameters. Short step sizes (typically 1-5% of the parameter range) are used for fine-tuning of fungicide spraying parameters. Small adjustments are made near the current flow, pressure, and angle parameters to deeply explore the optimal spraying configuration near the current parameter combination, enhancing the algorithm's accuracy in areas with known high-quality parameters. Long step sizes (typically 20-50% of the parameter range) are used for large-scale jumps in fungicide spraying parameters. This allows for large-scale movements across a broad parameter space of flow, pressure, and angle, escaping the current parameter configuration area to explore more possible spraying parameter combinations. This enhances the algorithm's ability to escape local optima and avoids getting trapped in suboptimal spraying control schemes. The alternating long and short step size update mechanism ensures that fungicide spraying parameter optimization can perform a sufficiently fine search within the effective parameter range while also promptly escaping local optima for global exploration. Through Levy flight, an expanded set of fungicide spraying control particles is generated, containing new candidate combinations of fungicide flow, pressure, and angle parameters generated based on the current optimal spraying parameter information. These new control particles inherit the excellent spraying parameter characteristics while introducing moderate parameter randomness, providing a richer starting point for parameter search in the next round of fungicide spraying control optimization. Subsequently, the fungicide spraying control particles in these expanded particle sets are included in the next loop to continue the process of coverage constraint surface prediction, fungicide coverage area analysis, fitness assessment, and search optimization based on the new parameter combinations, until the optimal fungicide spraying parameter combination that can both avoid duplicate spraying and meet coverage accuracy requirements is found.

[0097] By intelligently optimizing the spraying parameters of fungicide, and simulating the efficient foraging strategies in nature, the system achieves adaptive optimization search for parameters such as flow rate, pressure, and angle of fungicide. This not only ensures the globality of the parameter search but also improves the efficiency of converging to the optimal spraying control scheme, thereby enhancing the intelligence level and parameter optimization accuracy of the fungicide spraying control system.

[0098] Example 2, as Figure 2 As shown, based on the same inventive concept as the control method of the automatic sterilizing ploughing machine provided in Embodiment 1, this embodiment of the invention also provides a control system for the automatic sterilizing ploughing machine, including:

[0099] The path receiving module 11 is used to receive the timing information of the plowing path of the plowing machine, wherein the plowing path of the plowing machine includes the timing information of the coordinates of the fungicide nozzles.

[0100] The parameter configuration module 12 is used to randomly configure the timing information of the spray flow rate, the timing information of the spray pressure, and the timing information of the spray angle of the fungicide, and to perform simultaneous sequential spray surface prediction to obtain the timing information of the coverage constraint surface. The coverage constraint surface is a horizontal surface constructed by a set of spatial coordinates with the coordinates of the fungicide nozzle as the origin.

[0101] The bactericidal prediction module 13 is used to perform a union analysis of the coverage constraint surface based on the time sequence information of the bactericidal nozzle coordinates and the time sequence information of the coverage constraint surface to obtain the bactericidal coverage prediction area.

[0102] The control execution module 14 is used to control the automatic sterilization ploughing machine according to the ploughing path timing information, the spraying flow timing information, the spraying pressure timing information, and the spraying angle timing information of the ploughing machine when the first area intersection ratio between the predicted area covered by the fungicide and the target area is greater than or equal to the first area intersection ratio threshold.

[0103] Furthermore, the path receiving module 11 includes the following execution steps:

[0104] Obtain the spatial position of the plow blade and the spatial position of the fungicide nozzle for the preset sterilizing plow model;

[0105] Using the spatial position of the plow shovel as the origin, and combining it with the spatial position of the fungicide nozzle, the relative distribution position of the fungicide nozzle is constructed;

[0106] The preset sterilizing ploughing machine model is associated with the relative distribution position of the sterilizing agent nozzles and stored to construct a sterilizing agent nozzle coordinate identification table.

[0107] Input the model of the sterilizing ploughing machine into the coordinate identification table of the sterilizing agent nozzles, output the relative distribution position identification of the sterilizing agent nozzles, convert the timing information of the ploughing path of the ploughing machine, and obtain the timing information of the coordinates of the sterilizing agent nozzles.

[0108] Furthermore, the parameter configuration module 12 includes the following execution steps:

[0109] According to the preset sterilizing ploughing machine model, multiple sets of data are configured. Each set of data includes fungicide spray flow rate record data, fungicide spray pressure record data, fungicide spray angle record data, and a label indicating the coverage constraint surface with the coordinates of the fungicide nozzle as the origin.

[0110] Using labels indicating the coverage constraint surface originating from the fungicide nozzle coordinates as supervision, and taking the fungicide spray flow rate record data, the fungicide spray pressure record data, and the fungicide spray angle record data as input, multiple sets of data are retrieved, and several coverage constraint surface predictors are constructed through machine learning.

[0111] The outputs of the several cover constraint surface predictors are subjected to union fitting to obtain the cover constraint surface prediction model, which is then associated with and stored with a preset sterilizing and ploughing machine model to construct a cover constraint surface prediction model library.

[0112] Using the aforementioned coverage constraint surface prediction model library, and based on the model of the bactericide ploughing machine, the timing information of the bactericide spraying flow rate, the timing information of the bactericide spraying pressure, and the timing information of the bactericide spraying angle are used to perform simultaneous timing spraying surface prediction, thereby obtaining the timing information of the coverage constraint surface.

[0113] Furthermore, the parameter configuration module 12 also includes the following execution steps:

[0114] Constructing the union fitting rule:

[0115] Several predicted coverage constraint surfaces are obtained from the aforementioned coverage constraint surface predictors;

[0116] The coordinate triggering frequency of the several predicted coverage constraint surfaces is statistically analyzed to obtain multiple coordinate triggering frequencies;

[0117] Based on the multiple coordinate trigger frequencies, extract the coordinates in the several predicted coverage constraint surfaces whose coordinate trigger frequencies are greater than or equal to the coordinate trigger frequency threshold, and add them to the union coordinates.

[0118] Based on the union coordinates, construct the fitting coverage constraint surface;

[0119] Based on the union fitting rule, the outputs of the plurality of coverage constraint surface predictors are subjected to union fitting to obtain the coverage constraint surface prediction model.

[0120] Furthermore, the sterilization prediction module 13 includes the following execution steps:

[0121] Based on the time sequence information of the fungicide nozzle coordinates, combined with the time sequence information of the coverage constraint surface, simultaneous time sequence coverage constraint surface analysis is performed to obtain the set of spray coverage areas.

[0122] Perform a region crossover and union ratio analysis on the set of sprayed coverage areas to obtain a second area crossover and union ratio;

[0123] When the second area intersection ratio is greater than or equal to the second area intersection ratio threshold, the timing information of the fungicide spraying flow rate, the timing information of the fungicide spraying pressure, and the timing information of the fungicide spraying angle are updated and the loop is executed.

[0124] Otherwise, perform a coverage constraint surface union analysis on the set of sprayed coverage areas to obtain the predicted coverage area of ​​the fungicide.

[0125] Furthermore, when the first area intersection ratio between the predicted area covered by the fungicide and the target area is less than the first area intersection ratio threshold, the timing information of the fungicide spraying flow rate, the timing information of the fungicide spraying pressure, and the timing information of the fungicide spraying angle are updated in a loop.

[0126] Furthermore, the sterilization prediction module 13 also includes the following execution steps:

[0127] When the timing information of the spray flow rate of the bactericide, the timing information of the spray pressure of the bactericide, and the timing information of the spray angle of the bactericide are randomly updated a preset number of times, a number of historical bactericidal spray control particles are obtained.

[0128] Construct the fitness function:

[0129] When the update is triggered by the second area crossover ratio being greater than or equal to the second area crossover ratio threshold, the fitness function is inversely proportional to the ratio of the second area crossover ratio to the second area crossover ratio threshold;

[0130] When the update is triggered when the first area intersection ratio of the predicted area covered by the fungicide and the target area is less than the first area intersection ratio threshold, the fitness function is proportional to the ratio of the first area intersection ratio to the first area intersection ratio threshold.

[0131] Based on the fitness function, configure the fitness of several historical particles of the several historical sterilization spray control particles.

[0132] Based on the fitness of the aforementioned historical particles, a first preset number of particles with smaller fitness are selected as natural enemy particles, and a second preset number of particles with larger fitness are selected as food particles.

[0133] Using the food particles as the approaching targets and the enemy particles as the distant targets, Levi's flight is executed according to the long search step size constraint interval and the short search step size constraint interval to obtain an expanded particle set, and then the loop is executed.

[0134] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A control method for an automatic sterilizing and tilling machine, characterized in that, The method includes: Receive plowing path timing information of a tiller, wherein the plowing path timing information of the tiller includes the coordinate timing information of the fungicide nozzle; Randomly configure the timing information of fungicide spray flow rate, fungicide spray pressure, and fungicide spray angle, and perform simultaneous spray surface prediction to obtain the timing information of the coverage constraint surface. The coverage constraint surface is a horizontal plane constructed from a set of spatial coordinates with the coordinates of the fungicide nozzle as the origin. Based on the time-series information of the fungicide nozzle coordinates, combined with the time-series information of the coverage constraint surface, a union analysis of the coverage constraint surface is performed to obtain the predicted area of ​​fungicide coverage. When the first area intersection ratio between the predicted area covered by the fungicide and the target area is greater than or equal to the first area intersection ratio threshold, the control of the automatic sterilization ploughing machine is executed according to the ploughing path timing information of the ploughing machine, the spraying flow timing information of the fungicide, the spraying pressure timing information of the fungicide, and the spraying angle timing information of the fungicide.

2. The method as described in claim 1, characterized in that, Receive tillage path timing information of a tiller, wherein the tillage path timing information of the tiller includes timing information of the fungicide nozzle coordinates, and the tillage path timing information of the tiller represents the timing information of the tillage shovel coordinates, including: Obtain the spatial position of the plow blade and the spatial position of the fungicide nozzle for the preset sterilizing plow model; Using the spatial position of the plow shovel as the origin, and combining it with the spatial position of the fungicide nozzle, the relative distribution position of the fungicide nozzle is constructed; The preset sterilizing ploughing machine model is associated with the relative distribution position of the sterilizing agent nozzles and stored to construct a sterilizing agent nozzle coordinate identification table. Input the model of the sterilizing ploughing machine into the coordinate identification table of the sterilizing agent nozzles, output the relative distribution position identification of the sterilizing agent nozzles, convert the timing information of the ploughing path of the ploughing machine, and obtain the timing information of the coordinates of the sterilizing agent nozzles.

3. The method as described in claim 1, characterized in that, Randomly configure the timing information of fungicide spray flow rate, fungicide spray pressure, and fungicide spray angle, and perform simultaneous spray surface prediction to obtain the timing information of the coverage constraint surface, including: According to the preset sterilizing ploughing machine model, multiple sets of data are configured. Each set of data includes fungicide spray flow rate record data, fungicide spray pressure record data, fungicide spray angle record data, and a label indicating the coverage constraint surface with the coordinates of the fungicide nozzle as the origin. Using labels indicating the coverage constraint surface originating from the fungicide nozzle coordinates as supervision, and taking the fungicide spray flow rate record data, the fungicide spray pressure record data, and the fungicide spray angle record data as input, multiple sets of data are retrieved, and several coverage constraint surface predictors are constructed through machine learning. The outputs of the several cover constraint surface predictors are subjected to union fitting to obtain the cover constraint surface prediction model, which is then associated with and stored with a preset sterilizing and ploughing machine model to construct a cover constraint surface prediction model library. Using the aforementioned coverage constraint surface prediction model library, and based on the model of the bactericide ploughing machine, the timing information of the bactericide spraying flow rate, the timing information of the bactericide spraying pressure, and the timing information of the bactericide spraying angle are used to perform simultaneous timing spraying surface prediction, thereby obtaining the timing information of the coverage constraint surface.

4. The method as described in claim 3, characterized in that, The outputs of the plurality of coverage constraint surface predictors are subjected to union fitting to obtain a coverage constraint surface prediction model, including: Constructing the union fitting rule: Several predicted coverage constraint surfaces are obtained from the aforementioned coverage constraint surface predictors; The coordinate triggering frequency of the several predicted coverage constraint surfaces is statistically analyzed to obtain multiple coordinate triggering frequencies; Based on the multiple coordinate trigger frequencies, extract the coordinates in the several predicted coverage constraint surfaces whose coordinate trigger frequencies are greater than or equal to the coordinate trigger frequency threshold, and add them to the union coordinates. Based on the union coordinates, construct the fitting coverage constraint surface; Based on the union fitting rule, the outputs of the plurality of coverage constraint surface predictors are subjected to union fitting to obtain the coverage constraint surface prediction model.

5. The method as described in claim 1, characterized in that, Based on the time-series information of the fungicide nozzle coordinates, combined with the time-series information of the coverage constraint surface, a union analysis of the coverage constraint surface is performed to obtain the predicted fungicide coverage area, including: Based on the time sequence information of the fungicide nozzle coordinates, combined with the time sequence information of the coverage constraint surface, simultaneous time sequence coverage constraint surface analysis is performed to obtain the set of spray coverage areas. Perform a region crossover and union ratio analysis on the set of sprayed coverage areas to obtain a second area crossover and union ratio; When the second area intersection ratio is greater than or equal to the second area intersection ratio threshold, the timing information of the fungicide spraying flow rate, the timing information of the fungicide spraying pressure, and the timing information of the fungicide spraying angle are updated and the loop is executed. Otherwise, perform a coverage constraint surface union analysis on the set of sprayed coverage areas to obtain the predicted coverage area of ​​the fungicide.

6. The method as described in claim 1, characterized in that, Also includes: When the first area intersection ratio between the predicted area covered by the fungicide and the target area is less than the first area intersection ratio threshold, the timing information of the fungicide spraying flow rate, the timing information of the fungicide spraying pressure, and the timing information of the fungicide spraying angle are updated and the loop is executed.

7. The method as described in claim 5 or 6, characterized in that, The process of updating the timing information of the fungicide spray flow rate, the timing information of the fungicide spray pressure, and the timing information of the fungicide spray angle is looped, including: When the timing information of the spray flow rate of the bactericide, the timing information of the spray pressure of the bactericide, and the timing information of the spray angle of the bactericide are randomly updated a preset number of times, a number of historical bactericidal spray control particles are obtained. Construct the fitness function: When the update is triggered by the second area crossover ratio being greater than or equal to the second area crossover ratio threshold, the fitness function is inversely proportional to the ratio of the second area crossover ratio to the second area crossover ratio threshold; When the update is triggered when the first area intersection ratio of the predicted area covered by the fungicide and the target area is less than the first area intersection ratio threshold, the fitness function is proportional to the ratio of the first area intersection ratio to the first area intersection ratio threshold. Based on the fitness function, configure the fitness of several historical particles of the several historical sterilization spray control particles. Based on the fitness of the aforementioned historical particles, a first preset number of particles with smaller fitness are selected as natural enemy particles, and a second preset number of particles with larger fitness are selected as food particles. Using the food particles as the approaching targets and the enemy particles as the distant targets, Levi's flight is executed according to the long search step size constraint interval and the short search step size constraint interval to obtain an expanded particle set, and then the loop is executed.

8. A control system for an automatic sterilizing and tilling machine, characterized in that, The system includes: The path receiving module is used to receive the timing information of the tilling path of the tiller, wherein the timing information of the tilling path of the tiller includes the timing information of the coordinates of the fungicide nozzle. The parameter configuration module is used to randomly configure the timing information of the spray flow rate, the timing information of the spray pressure, and the timing information of the spray angle of the fungicide, and to perform simultaneous sequential spray surface prediction to obtain the timing information of the coverage constraint surface. The coverage constraint surface is a horizontal plane constructed by a set of spatial coordinates with the coordinates of the fungicide nozzle as the origin. The bactericidal prediction module is used to perform a union analysis of the coverage constraint surface based on the time sequence information of the bactericidal nozzle coordinates and the time sequence information of the coverage constraint surface to obtain the bactericidal coverage prediction area. The control execution module is used to control the automatic sterilizing ploughing machine according to the ploughing path timing information, the spraying flow timing information, the spraying pressure timing information, and the spraying angle timing information of the ploughing machine when the first area intersection ratio of the predicted area covered by the fungicide and the target area covered by the fungicide is greater than or equal to the first area intersection ratio threshold.

Images