Automatic navigation and positioning method for agricultural equipment

Abstract

The invention provides an automatic navigation and positioning method for agricultural equipment, and the method comprises the steps: obtaining the root system residue distribution data of a previous crop and the soil hardening difference, building a correlation model of the crop root system residue distribution and the soil hardening difference, and determining a target tillage depth plan for the sowing of a next crop according to the correlation model; the operation track and depth parameters meeting the soil texture improvement requirements are applied to an automatic navigation and positioning system of agricultural equipment, and the operation process of the agricultural equipment is automatically controlled and dynamically adjusted; and based on the adjusted operation track and depth parameters, analyzing the long-term influence of an automatic navigation and positioning system of agricultural equipment on soil texture improvement in different growing seasons by tracking soil quality indexes, crop yields and root system distribution.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an automatic navigation and positioning method for agricultural equipment. Background Art

[0002] In agricultural production, the adoption of a crop rotation system can effectively improve soil fertility and pest control effects. However, due to differences in root distribution and soil compaction among different crops, it is difficult to match the tillage depth planning of subsequent crop sowing with the root distribution of the previous crop. Specifically, during the growth of the previous crop, the roots will form distributions of different depths and densities in the soil, and these root residues will affect the physical structure and compaction degree of the soil. When sowing the next crop, if the tillage depth does not take into account the influence of the root distribution of the previous crop, the tillage may be too deep or too shallow, thereby affecting the emergence and growth of the next crop. Therefore, when agricultural machinery is performing tillage operations, how to dynamically adjust the tillage trajectory and depth parameters according to information such as the type, growth period and harvest time of the previous crop to adapt to the differences in soil compaction degree in different regions. However, during the crop rotation process in different seasons, the distribution of crop root residues varies greatly, resulting in uneven soil compaction, which brings challenges to the tillage depth planning of the next crop. At the same time, when agricultural machinery is operating in the field, it is necessary to dynamically adjust the operation trajectory and depth parameters according to historical farming data to adapt to the soil compaction conditions in different regions. In summary, how to dynamically adjust the operation trajectory and depth parameters of agricultural machinery according to historical farming data to adapt to the differences in soil compaction in different regions under the rotation system is a technical problem that needs to be solved urgently. Summary of the invention

[0003] The present invention provides an automatic navigation and positioning method for agricultural equipment, which mainly includes:

[0004] Obtain the root residue distribution data of the previous crop and the difference in soil compaction, establish a correlation model between the root residue distribution of crops and the difference in soil compaction, and determine the target tillage depth plan for sowing the next crop based on the correlation model;

[0005] Obtain historical tillage data, including tillage depth, equipment used, trajectory, and changes in soil texture over time. Analyze historical tillage data to obtain the time series characteristics of soil texture changes under different tillage depths and trajectories, and establish a dynamic mapping relationship between tillage depth, trajectory, and soil texture.

[0006] According to the target tillage depth planning for the next crop sowing, the adjusted operation parameters of the agricultural machinery equipment are obtained in combination with the dynamic mapping relationship, and the operation trajectory and operation depth parameters of the agricultural machinery equipment are dynamically adjusted according to the dynamic mapping relationship to make the operation depth match the target tillage depth planning;

[0007] Determine whether the impact of the test evaluation data of the adjusted operation trajectory and depth parameters on soil texture improvement reaches the expected goal. If not, continue to dynamically adjust the operation trajectory and operation depth parameters of the agricultural machinery equipment until the soil texture improvement requirements are met;

[0008] Apply the operation trajectory and depth parameters that meet the requirements of soil texture improvement to the automatic navigation and positioning system of agricultural equipment to automatically control and dynamically adjust the operation process of agricultural machinery and equipment;

[0009] After each growing season, the crop root residue distribution data and soil texture data are obtained, and the association model is updated using the incremental learning algorithm to adjust the relevant operating parameters of the agricultural machinery navigation system, including the operating trajectory and depth parameters;

[0010] Based on the adjusted operation trajectory and depth parameters, by tracking soil quality indicators, crop yield, and root distribution, the long-term impact of the automatic navigation and positioning system of agricultural machinery on soil texture improvement in different growing seasons is analyzed.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0012] The present invention discloses an automatic navigation and positioning method for agricultural equipment. The method establishes a correlation model between the residual distribution of the previous crop roots and the difference in soil compaction, and combines the time series characteristics of soil texture changes obtained by analyzing historical tillage data to construct a dynamic mapping model of tillage depth, trajectory and soil texture. According to the target tillage depth planning for the next crop sowing, the present invention dynamically adjusts the operation trajectory and depth parameters of agricultural machinery and equipment until the soil texture improvement requirements are met. Automatic control and dynamic adjustment are achieved through the automatic navigation and positioning system of agricultural equipment, and the incremental learning algorithm is used to update the model and parameters after each growing season. Through long-term tracking and analysis, the present invention realizes the precise improvement of soil texture by agricultural machinery and equipment, improves tillage efficiency and crop yield, and provides a new technical solution for intelligent precision agriculture. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The present invention is a flowchart of an automatic navigation and positioning method for agricultural equipment. DETAILED DESCRIPTION

[0014] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0015] like Figure 1 In this embodiment, an automatic navigation and positioning method for agricultural equipment may specifically include:

[0016] S101. Obtain root residue distribution data of the previous crop and soil compaction differences, establish a correlation model between crop root residue distribution and soil compaction differences, and determine a target tillage depth plan for sowing the next crop based on the correlation model.

[0017] A soil resistivity detector is used to obtain soil resistivity data, and soil moisture data is collected through a wireless sensor. The soil density value is calculated according to the soil resistivity data and the soil moisture data. A root distribution image is obtained from the soil profile according to an underground biomass detector, and the root distribution image is processed by a grayscale threshold segmentation method to obtain the root residual area ratio within the soil layer depth interval. A soil compaction evaluation function is established for the soil layer depth interval, and a soil compaction prediction model is trained according to the soil density value and the root residual area ratio. Each soil layer is evaluated by the soil compaction prediction model to obtain soil compaction grade data, and the lowest depth interval of the soil compaction grade is searched under the condition of satisfying the depth constraint according to the minimum tillage depth requirement of the next crop, to obtain the tillage depth value.

[0018] Specifically, soil resistivity detectors were used to obtain soil resistivity data from the previous crop area, and soil profiles were scanned by geological radar scanners. Soil hardness distribution data was obtained based on the soil resistivity and soil hardness comparison conversion table. Soil moisture data was then collected by wireless sensors. The soil density data was calculated using the hardness and humidity compensation formula h = k1 × r + k2 × w, where h is soil density, r is soil hardness, w is soil moisture content, and k1 and k2 are corresponding compensation coefficients. The root residue distribution image was obtained from the soil profile using the underground biomass detector, and the root area was extracted using the pixel grayscale threshold segmentation method. Pixels with grayscale values ​​less than the threshold t were marked as root areas. The segmented images were marked with connected areas, and the root residue area percentage p in each soil layer depth interval d was calculated. For each soil layer depth interval d, a soil compaction evaluation function b = f (h, p) was established, where b is soil compaction, h is soil density, p is root residue area percentage, and f is the evaluation function. The function parameters were obtained using the least squares fitting method. Soil texture data is obtained from the soil sample library, sample data is calculated in combination with the soil compaction evaluation function, and the soil compaction prediction model m(x) is trained using the random forest algorithm. The input variable x includes soil texture, soil density, and root residue ratio. Each soil layer is evaluated according to the soil compaction prediction model to obtain soil compaction grade data. The minimum tillage depth requirement dmin for the next crop is retrieved from the crop planting parameter library. The depth search algorithm is used to find the depth interval with the lowest soil compaction grade under the dmin constraint condition to obtain the final tillage depth value. In farmland soil management, the soil resistivity detector reflects the soil structure state by measuring the current conduction capacity in the soil. When the soil moisture content is 25%, the resistivity of compacted soil is usually between 85 and 120 ohm-meters, while the resistivity of loose soil is between 45 and 75 ohm-meters. The geological radar scan uses a 400MHz antenna, the scanning depth is set at 80 cm, and a set of data is recorded every 10 cm to obtain the electromagnetic wave reflection signal intensity of the soil profile. Soil hardness and resistivity show a significant positive correlation. By establishing a regional soil hardness comparison table, when the resistivity is 100 ohm-meter, the corresponding soil hardness value is 2.8 MPa. The distribution state of plant root residues in the soil directly affects the soil structure. The underground biomass detector uses near-infrared band imaging. The root tissue shows a lower reflectivity in the near-infrared image. By setting the grayscale threshold to 75, the image is divided into root area and non-root area. In the 0 to 20 cm soil layer, the root residue area of ​​corn after harvest accounts for 32%, while it decreases to 15% in the 20 to 40 cm soil layer. This distribution feature is closely related to the degree of soil compaction. In the soil compaction evaluation function, soil density and root residue area ratio are two key input variables. The function is fitted using the measured data of the experimental field. When the soil density is 2.5 MPa and the root residue area accounts for 25%, the calculated soil compaction degree is 0.72.Soil texture also has an important impact on compaction. For every 10% increase in clay content, the soil compaction degree increases by an average of 0.15. The training samples of the soil compaction prediction model include soils of different texture types, of which sandy loam samples account for 35%, loam samples account for 45%, and clay samples account for 20%. After model training, the prediction accuracy reaches 88%. In actual application, the soil compaction level of the middle layer of a farmland is level 3, the soil texture is loam, and soybeans are planned to be planted. The minimum tillage depth required for soybean growth is 25 cm. Through depth search, it is found that the soil compaction level in the depth range of 30 to 35 cm is the lowest, which is level 1. Based on this, the tillage depth is determined to be 32 cm. In the actual application scenario, taking the preparation for planting soybeans after corn harvest as an example, the soil resistivity detection data shows that the resistivity of the surface 20 cm is 95 ohm-meter, which is converted to a soil hardness of 2.6 MPa, a soil water content of 22%, and a calculated soil density of 2.3 MPa. At the same time, it was detected that the residual area of ​​corn roots in this layer of soil accounted for 28%. Combined with the local soil texture data, it was predicted that the soil compaction level was 2. Through deep search, it was finally determined that the tillage depth was 28 cm.

[0019] S102. Obtain historical tillage data, which includes tillage depth, equipment used, trajectory, and changes in soil texture over time. Analyze the historical tillage data to obtain the time series characteristics of soil texture changes under different tillage depths and different trajectories, and establish a dynamic mapping relationship between tillage depth, trajectory, and soil texture.

[0020] A sequence of agricultural machinery trajectory coordinate points is obtained from the agricultural machinery operation database, and trajectory density distribution data is obtained by using a trajectory density calculation formula based on the trajectory point sequence; a soil particle size analyzer is used to obtain the sand content, silt content and clay content in the soil sampling data, and soil particle composition distribution data is obtained by using a Kriging interpolation algorithm based on the soil particle content; a tillage depth time series data set is established based on the tillage depth data in the agricultural machinery operation record, and the tillage depth time series data set includes a time series and the tillage depth value at the corresponding moment; soil texture type data is obtained based on the soil particle composition distribution data using a soil texture triangulation method, and a soil texture prediction model is established using a long short-term memory neural network, and the input variables of the prediction model include the tillage depth time series data set, the trajectory density distribution data and the soil particle composition distribution data.

[0021] Specifically, the historical tillage operation records are read from the agricultural machinery operation database, and the spatial coordinate point sequence of the agricultural machinery driving trajectory is extracted. The trajectory points are corrected for accuracy through the global positioning solution tool. The trajectory density calculation formula d=n / s is used, where n is the number of trajectory points and s is the cultivated area, to obtain the tillage trajectory density distribution map. According to the soil monitoring database, the historical sampling data is retrieved, and the soil particle composition at different time nodes is obtained through the soil particle size analyzer, including the content of sand, silt, and clay. The Kriging interpolation algorithm is used to generate the soil particle composition distribution map. According to the agricultural machinery models and supporting agricultural tool specifications recorded in the tillage machinery equipment parameter library, the actual tillage depth records during the historical operation are extracted to establish the tillage depth time series data set d(t), where t is the time series and d is the tillage depth value at the corresponding moment. Based on the soil particle composition distribution map, the soil texture triangulation map classification method is used to divide the soil texture types. Combined with the tillage trajectory density distribution map, the soil texture change data of the trajectory coverage area is obtained through spatial superposition operation. The soil texture time series prediction model m(x) was constructed using a long short-term memory neural network. The input variable x included tillage depth, trajectory density, and soil particle composition. The model parameters were optimized by gradient boosting tree to obtain the dynamic mapping relationship between tillage depth, trajectory, and soil texture. The spatial distribution characteristics of agricultural machinery operation trajectories reflect the degree of refinement of tillage operations. In a farmland with an area of ​​10 hectares, the number of agricultural machinery operation trajectory points reached 2,000, and the calculated trajectory density was 200 points / hectare. This density level indicates that the tillage operation is relatively uniform. The spatial distribution of trajectory points was interpolated by Kriging to form a continuous density distribution map, showing the spatial variation characteristics of tillage intensity. The trajectory density in the middle area of ​​the farmland reached 250 points / hectare, while the edge area decreased to 150 points / hectare. Soil particle composition is a key indicator to characterize soil texture. The data obtained by soil particle size analysis showed that the sand content of the soil before tillage was 45%, the silt content was 35%, and the clay content was 20%, which belongs to the sandy loam type. After three years of continuous tillage, the sand content of the soil at the same location decreased to 40%, the silt content increased to 38%, and the clay content increased to 22%, and the soil texture changed significantly. This change is closely related to the tillage depth and track distribution. In areas with higher track density, the degree of soil particle fragmentation is higher. The temporal changes in tillage depth reflect the intensity of agricultural machinery operations. The average tillage depth of a certain farmland is 25 cm in spring and increases to 30 cm in autumn. This seasonal change is consistent with the requirements of crop planting. There are differences in the tillage depth capabilities of different agricultural machinery and equipment. The maximum tillage depth of large deep tillers can reach 45 cm, while the tillage depth of ordinary rotary tillers is generally around 20 cm. The dynamic changes in soil texture are affected by multiple factors. Under the conditions of a track density of 200 points / hectare and a tillage depth of 30 cm, the soil particle fragmentation rate increases by 15%, resulting in an increase in the content of fine particles.By learning historical data, the long short-term memory neural network found that when the tillage depth increases by 5 cm, the fine particle content of the soil increases by an average of 2%, and this change is more significant in areas with higher trajectory density. In practical applications, the tillage data of a farm for five consecutive years showed that high-frequency deep tillage caused the sand content of the surface soil to decrease from 50% to 42%, while the content of silt and clay increased accordingly. Based on these historical data, combined with the changes in tillage depth and trajectory distribution characteristics, the soil texture prediction model accurately predicts that if the current tillage method is maintained, the soil sand content will further decrease to 38% in the next two years, which provides a basis for timely adjustment of tillage parameters. The dynamic mapping model reveals the influence of different tillage methods on soil texture and helps to achieve the optimal regulation of tillage parameters.

[0022] S103. According to the target tillage depth plan for sowing the next crop, the adjusted operating parameters of the agricultural machinery equipment are obtained in combination with the dynamic mapping relationship, and the operating trajectory and operating depth parameters of the agricultural machinery equipment are dynamically adjusted according to the dynamic mapping relationship to match the operating depth with the target tillage depth plan.

[0023] The soil density and water content data collected by the soil detector are obtained, and the actual maximum tillage depth limit value of the agricultural machinery equipment is calculated based on the soil density and water content data; the boundary coordinates and terrain elevation data of the cultivated land area are obtained by using a plot surveying instrument, and the minimum partitioning algorithm is executed through the boundary coordinates and terrain elevation data to obtain an operation sub-area with uniform slope, and the A-star algorithm is run according to the operation sub-area to generate an optimal operation trajectory route; the track coverage density is calculated from the optimal operation trajectory route, and the operation parameters that meet the expected soil texture are calculated using an online optimizer based on the track coverage density and the soil texture prediction function; the operation status of the agricultural machinery equipment is collected in real time using a wireless data sensor. If there is a deviation between the operation status and the expected operation parameters, a correction amount is obtained based on the deviation, and the control parameters of the agricultural machinery equipment are corrected in real time based on the correction amount.

[0024] Specifically, the agricultural machinery type parameters, including the maximum tillage depth, nominal power, and operating width, are obtained from the agricultural machinery database. Based on the soil density and water content data collected by the soil detector, the actual maximum tillage depth limit r(h,w) of each agricultural machinery under the current soil conditions is calculated using a deep neural network, where h is the soil density and w is the water content. The plot mapping instrument is used to obtain the boundary coordinates and terrain elevation data of the cultivated land area. The plot is divided into operating sub-areas with uniform slopes using the minimum partitioning algorithm. The A-star algorithm is used to generate the optimal operating trajectory route for each sub-area, and the trajectory coverage density c(x,y) is calculated, where x and y are spatial coordinates. The soil texture prediction function s(d,c) is obtained from the dynamic mapping relationship, where d is the tillage depth and c is the trajectory coverage density. Combined with the target tillage depth planning and the actual maximum tillage depth limit r of the agricultural machinery, an online optimizer is used to calculate the operating parameters that meet the expected soil texture, including the driving speed v, engine speed n, and operating width w. According to the prediction results of agricultural machinery operation parameters and soil texture, the genetic algorithm is used to optimize the trajectory coverage density, output the optimal trajectory route l(t), where t is the time series and the operation depth sequence d(t), and generate the operation control instruction set. Wireless data sensors are used to collect the operation status of agricultural machinery in real time, including the actual operation depth da and the actual trajectory ca. The correction value δ is obtained through the parameter deviation calculation function p(da,ca), and the control parameters of agricultural machinery, including speed, speed and width, are corrected in real time. The dynamic adjustment of agricultural machinery operation parameters involves the coordinated optimization of multiple key factors. Taking a large deep tiller as an example, its nominal maximum tillage depth is 50 cm, the nominal power is 180 kW, and the operation width is 3.4 meters. The maximum tillage depth during actual operation is significantly affected by the soil conditions. When the soil density is 2.6 MPa and the water content is 22%, the actual maximum tillage depth limit calculated by the deep neural network is 45 cm. This limit ensures the safe operation of the equipment under complex soil conditions. Land surveying data showed that a 15-hectare farmland had obvious terrain undulations, with slopes ranging from 3% to 8%. The minimum partitioning algorithm divided the land into five sub-areas, with the largest sub-area being 4.2 hectares and the smallest sub-area being 2.1 hectares. The A-star algorithm generates operating trajectories based on terrain factors, so that the track coverage density reaches 220 points / hectare in flat areas and decreases to 180 points / hectare in areas with larger slopes. The soil texture prediction function reflects the influence of tillage operations on soil structure. When the tillage depth is 35 cm and the track coverage density is 200 points / hectare, it is predicted that the content of fine particles in the soil will increase by 5%. Based on this prediction result, the online optimizer calculates the optimal operating parameters of agricultural machinery: the driving speed is 6 km / hectare.

[0025] hours, the engine speed is 1800 rpm, and the working width is adjusted to 3.2 meters. When the genetic algorithm optimizes the track coverage density, it makes differentiated adjustments to the working intensity in different areas while maintaining the overall working efficiency. The optimized track route increases the number of repeated operations in areas with severe soil compaction, so that the track coverage density is increased to 240 points / hectare, while the working intensity is appropriately reduced in areas with loose soil. The optimal working depth sequence shows that the tillage depth should change dynamically with soil conditions during the operation, ranging from 28 to 42 cm. Real-time monitoring data plays a key role in the dynamic adjustment of operating parameters. When it is detected that the actual working depth deviates from the target value by more than 3 cm, the parameter deviation calculation function immediately generates a correction instruction. In a certain operation, it was found that the actual tillage depth was 32 cm, which was lower than the target value of 38 cm. The system calculated that the engine speed needed to be increased to 1950 rpm and the driving speed needed to be reduced to 5.2 km / h. This real-time adjustment ensures that the working quality meets the requirements. During the operation, the number of equipment parameter adjustments reached an average of 12 times per hectare, reflecting the rapid response capability of the control system.

[0026] S104. Determine whether the impact of the test evaluation data of the adjusted operating trajectory and depth parameters on soil texture improvement reaches the expected goal. If not, continue to dynamically adjust the operating trajectory and operating depth parameters of the agricultural machinery equipment until the soil texture improvement requirements are met.

[0027] A soil analyzer is used to obtain soil particle composition data, the soil particle composition data includes a sand particle ratio, a silt particle ratio and a clay particle ratio, and a soil texture improvement index is obtained through a neural network based on the soil particle composition data; based on the soil texture improvement index, an improvement target value is obtained from a soil texture standard library, spatial distribution data of agricultural machinery operation tracks and operation depth data are collected through a wireless sensor network, and soil density data of the operation area are measured using a soil density detector; based on the spatial distribution data and operation depth data of the agricultural machinery operation tracks, an operation uniformity index is calculated, and an operation quality assessment value is generated in combination with the soil density data of the operation area; based on the deviation between the soil texture improvement index and the improvement target value and the operation quality assessment value, an operation parameter adjustment amount is obtained through a deep optimization calculator, and agricultural machinery operation control parameters are calculated using a parameter optimizer, and the agricultural machinery operation control parameters include track spacing, driving speed and operation depth.

[0028] Specifically, soil samples were collected from each block of the farmland according to the soil sampling equipment, and the soil particle composition data, including the sand ratio ps, silt ratio pm, and clay ratio pc, were measured by the soil analyzer. The soil texture improvement index r = f(ps, pm, pc) was calculated using a neural network, and the improvement target value rt was obtained from the soil texture standard library. The spatial distribution map of the agricultural machinery operation trajectory was obtained using a wireless sensor network, the operation depth time series data d(t) was recorded by a depth sensor, and the soil density detector was used to measure the soil density m(x, y) in the operation area, where x and y are spatial coordinates. According to the operation trajectory distribution and depth data, the multi-source data fusion algorithm was used to calculate the operation uniformity index h = g(c, d), where c is the track coverage density and d is the operation depth. The operation quality assessment value q was generated in combination with the soil density. Support vector regression was used to predict the trend of soil texture changes. Based on the texture improvement index deviation δr = rt-r and the operation quality assessment value q, the operation parameter adjustment amount was generated by the depth optimization calculator, including the track spacing adjustment value Δs, the speed adjustment value Δv, and the depth adjustment value Δd. According to the adjustment of the operating parameters, the parameter optimizer is used to calculate the new agricultural machinery operation control parameters, track spacing sn = s0 + Δs, driving speed vn = v0 + Δv, and operating depth dn = d0 + Δd, and the update command is sent to the agricultural machinery equipment through the control data channel. The data comparator is used to determine whether the soil texture improvement index r meets the target value rt. When the deviation δr is greater than the preset threshold σ, it returns to continue collecting operating data and updating the control parameters. When δr is less than or equal to σ, the parameter adjustment process is completed. In the evaluation of farmland soil texture improvement, soil particle composition is a key indicator. Taking the sampling analysis of a farmland soil as an example, the initial state of sand ratio is 52%, silt ratio is 31%, and clay ratio is 17%. The texture improvement index calculated by the neural network is 0.65, while the optimal soil texture improvement index target value of the cultivated crops in this area in the soil texture standard library is 0.82. This difference indicates that the soil structure needs to be adjusted, and the soil particle composition can be optimized by changing the tillage method. Monitoring of the agricultural machinery operation process shows that in a 12-hectare farmland, the average spacing between the operation tracks is 2.8 meters, and the operation depth recorded by the depth sensor fluctuates between 28 and 35 centimeters. The soil density detection data shows that the soil density in the cultivated area shows obvious spatial differences, with the density in the central area reaching 2.8 MPa, while the edge area is 2.2 MPa. This unevenness directly affects the effect of soil texture improvement. The multi-source data fusion algorithm comprehensively considers the spatiotemporal distribution characteristics of track coverage density and operation depth. When the track coverage density is 180 points / hectare and the average operation depth is 32 cm, the calculated operation uniformity index is 0.78, and the operation quality assessment value obtained by combining the soil density data is 0.72. Support vector regression prediction shows that under the existing operation parameters, the improvement rate of the soil texture improvement index is gradually slowing down.The results of the depth optimization calculation show that the operating parameters need to be appropriately adjusted. The track spacing is reduced by 0.3 meters to increase the coverage density, the driving speed is reduced by 0.8 kilometers per hour to increase the tillage intensity, and the operating depth is increased by 3 centimeters to improve the soil structure. These adjustments will help speed up the process of soil texture improvement. The adjusted agricultural machinery operation control parameters are track spacing of 2.5 meters, driving speed of 5.5 kilometers per hour, and operating depth of 35 centimeters. After one operation cycle, the soil texture improvement index rose to 0.75, with a deviation of 0.07 from the target value of 0.82, which is greater than the preset threshold of 0.05, indicating that the operating parameters need to be further adjusted. The new round of monitoring data shows that the soil particle composition has undergone positive changes, with the proportion of sand particles reduced to 48%, the proportion of silt particles increased to 33%, and the proportion of clay particles increased to 19%. This trend of change is in line with the texture improvement expectations. The data comparison results guide the agricultural machinery equipment to make a second round of parameter adjustments, and finally achieve the expected soil texture through continuous optimization.

[0029] S105. Apply the operation trajectory and depth parameters that meet the requirements of soil texture improvement to the automatic navigation and positioning system of agricultural equipment to automatically control and dynamically adjust the operation process of the agricultural machinery equipment.

[0030] Acquire differential positioning data and real-time position coordinates of agricultural machinery, wherein the real-time position coordinates are calculated by a real-time dynamic positioning method based on the differential positioning data; generate a smooth trajectory curve using a deep neural network based on the real-time position coordinates and the real-time value of the tillage depth collected by the depth sensor, wherein the curvature change rate of the smooth trajectory curve is less than a preset curvature threshold; calculate a steering control amount using a state feedback controller based on the smooth trajectory curve and the roll angle and pitch angle data collected by the attitude sensor, wherein the steering control amount is used to adjust the driving direction of the agricultural machinery; generate a hydraulic lifting control amount using a proportional-integral controller based on the deviation between the real-time value of the tillage depth and the target depth value, wherein the hydraulic lifting control amount is used to adjust the tillage depth of the agricultural machinery.

[0031] Specifically, the real-time position data p(x, y) of agricultural machinery is collected by satellite positioning receiver, and the differential correction data δp is obtained through the differential positioning station. The real-time dynamic positioning algorithm is used to calculate the actual position coordinates of agricultural machinery pc=p+δp, and the driving direction angle θ and driving speed v are calculated from the position coordinate sequence. The trajectory planner is used to load the target operation trajectory curve l(x, y), and the deep neural network is used to generate a smooth trajectory curve ls(x, y) according to the real-time value d of the tillage depth detected by the depth sensor and the target depth value dt, and the curvature change rate is less than the preset threshold k. The roll angle α and pitch angle β of agricultural machinery are collected by the attitude sensor, and the steering control amount u=f(α, β, δθ) is calculated by the state feedback controller in combination with the angular deviation δθ=θ-θt, θt is the target angular deviation. According to the position deviation e(x, y) and directional deviation δθ between the actual trajectory of agricultural machinery and the target trajectory, the linear quadratic regulator is used to generate the speed correction δv, and the speed control instruction vt=v+δv is issued through the control data interface. The depth sensor is used to detect the tillage depth value d in real time, and the depth deviation δd=d-dt is calculated. The hydraulic lifting control value h=g(δd) is generated by the proportional integral controller to realize the dynamic adjustment of the tillage depth of agricultural machinery. The navigation and depth control parameters of agricultural machinery are optimized online by using the random forest algorithm to generate the optimized control parameter group, including the steering control gain ku, the speed control gain kv, and the depth control gain kh. The coordinated adjustment of the motion state of agricultural machinery is realized through the control signal interface. In the process of automatic navigation and positioning of agricultural machinery, the real-time positioning accuracy directly affects the operation quality. When operating in open land, the original coordinate accuracy obtained by the satellite positioning receiver is in the meter range. The correction data of the differential positioning station can improve the accuracy to the centimeter level. In actual operation, the original position error of a certain agricultural machinery during driving reached 1.2 meters. After differential correction, the error was reduced to 2.8 centimeters. The calculated driving direction angle was 275 degrees and the driving speed was 6.5 kilometers per hour. Trajectory planning has an important impact on the operation quality. The target trajectory curve requires good smoothness to avoid sharp turns. In an irregular farmland, the maximum curvature change rate of the original trajectory curve reached 0.08 / meter, which was reduced to 0.03 / meter after deep neural network smoothing. At the same time, the depth sensor detected that the current tillage depth was 28 cm, which was 4 cm away from the target depth of 32 cm. The attitude of agricultural machinery has a significant impact on the accuracy of navigation control. During operation, the attitude sensor detected that the roll angle fluctuation range was plus or minus 2.5 degrees, and the pitch angle fluctuation range was plus or minus 3.8 degrees. When the heading angle of the agricultural machinery deviates by 8.5 degrees from the target direction, the state feedback controller generates a corresponding steering command to adjust the rudder angle by 12 degrees.Position tracking control requires agricultural machinery to accurately follow the planned trajectory. When operating in a straight line, the maximum lateral deviation of a certain agricultural machinery from the target trajectory is 15 cm, and the directional deviation is 5.2 degrees. The linear quadratic regulator calculates that the travel speed needs to be reduced from 7.2 km / h to 6.8 km / h. This speed adjustment helps to improve the trajectory tracking accuracy. During the depth dynamic control process, the sensor detects that the tillage depth gradually decreases from the target value of 35 cm to 31 cm. The proportional integral controller generates a hydraulic lifting control signal based on this, extending the hydraulic cylinder by 28 mm, and gradually adjusting the tillage depth to the target value. During the operation, the depth control accuracy is maintained within the range of plus or minus 2 cm. The random forest algorithm optimizes the control parameters in real time by analyzing the historical motion data of agricultural machinery. In a certain optimization, the steering control gain was adjusted from 1.2 to 1.5, the speed control gain was adjusted from 0.8 to 1.1, and the depth control gain was adjusted from 1.0 to 1.3. The optimized parameter combination made the movement of agricultural machinery smoother, and the trajectory tracking and depth control accuracy were improved. The online optimization mechanism can adapt to the operation requirements under different field conditions.

[0032] S106. After each growing season, obtain crop root residue distribution data and soil texture data, and use the incremental learning algorithm to update the association model and adjust the relevant operating parameters of the agricultural machinery navigation system, including operating trajectory and depth parameters.

[0033] The root spatial distribution image and soil profile samples are obtained, and the sand content, silt content and clay content in the soil samples are determined by a soil particle analyzer to obtain a soil texture characteristic vector; based on the soil texture characteristic vector and the root distribution image, an incremental Bayesian network is used to construct a root distribution and soil texture association model, and the historical operation trajectory vector and depth record vector are read from the operation database; the soil texture characteristic vector and the operation trajectory vector are mapped and optimized by a deep autoencoder to obtain an optimized operation parameter vector; for the optimized operation parameter vector, a parameter optimizer is used to calculate the compensation correction amount in combination with the trajectory accuracy index and the depth accuracy index to obtain the operation control parameters.

[0034] Specifically, the spatial distribution image of crop roots is collected by underground biomass detector, soil profile samples are obtained by soil sampler, and soil particle composition data are measured by soil particle analyzer, including sand content ps, silt content pm, and clay content pc, to generate soil texture feature vector s = (ps, pm, pc). The incremental Bayesian network is used to construct the root distribution and soil texture association model m1(r, s), where r is the root distribution vector and s is the soil texture feature vector. The historical operation trajectory l(t) and depth record d(t) are read from the agricultural machinery operation database. The soil improvement index q = f(s) is calculated based on the soil texture feature vector s, where f is the improvement degree calculation function. The time series data miner is used to extract the association features between the operation parameter vector p = (l, d) and the soil texture change vector Δs. The newly collected soil texture data sn and the historical data sh are feature fused to generate a combined feature vector sc. The weight of the association model is updated through the neural network trainer to obtain the updated model parameter θ1. The deep autoencoder is used to optimize the mapping relationship between soil texture and operation parameters. The input variables include soil feature vector sc and operation parameter vector p, and the optimized operation parameter vector po is output. According to the optimized operation parameter vector po, combined with the historical operation evaluation indicators, including trajectory accuracy e1 and depth accuracy e2), the compensation correction amount Δp is calculated by the parameter optimizer to generate the operation control parameter pc=po+Δp for the next growing season. The soil texture assessment after crop harvest is of great significance for guiding the cultivation of the next growing season. Taking the farmland after corn harvest as an example, the underground biomass detector scans the soil layer from 0 to 60 cm to obtain the root distribution image. The soil sampling analysis results show that the sand content in the surface layer from 0 to 20 cm is 48%, the silt content is 35%, and the clay content is 17%. Based on this, the soil texture feature vector is generated to reflect the soil particle composition. The incremental Bayesian network updates the model parameters by continuously learning new data. In the observation of a farmland for three consecutive growing seasons, 15 soil samples were collected each season for analysis, and 45 sets of root distribution and soil texture data pairs were accumulated. The model learning results show that the soil texture in the root-dense area has changed significantly, and the clay content has increased by an average of 3.2%. At the same time, historical records extracted from the operation database show that the tillage depth adopted in the area varies between 32 and 38 cm. The soil improvement index reflects the degree of improvement in soil texture. When the soil particle composition is close to the optimal state for crop growth, the improvement index is close to 1. In actual application, the initial improvement index of corn field soil is 0.72. By optimizing the tillage parameters, the sand content is reduced to 45%, the powder content is increased to 37%, and the improvement index rises to 0.85. Time series data mining found that for every 5 cm increase in tillage depth, the content of fine particles in the soil increases by an average of 1.8%.When the newly collected soil data is integrated with the historical data, the time weighting method is adopted, with the weight of the most recent season data being 0.6, the weight of the previous season data being 0.3, and the weight of the previous season data being 0.1. When the neural network updates the weight, 90 sets of historical data and 15 sets of new data are input into the training, and the model prediction accuracy is improved from 85% to 89%. The deep autoencoder further optimizes the mapping relationship, and the output operation parameters show that the tillage depth should be controlled at around 35 cm. Historical operation evaluation data show that the track deviation of agricultural machinery equipment is less than 10 cm and the depth deviation is less than 2 cm when operating on flat ground, but the accuracy is significantly reduced in sloping areas. The parameter optimization calculation results show that in areas with a slope greater than 5 degrees, the driving speed needs to be reduced by 20% to ensure the operation accuracy. In the final generated operation control parameters, the track spacing is set to 2.8 meters and the standard tillage depth is set to 35 cm. These parameters will guide the tillage operations in the next growing season.

[0035] S107. Based on the adjusted operation trajectory and depth parameters, by tracking soil quality indicators, crop yield, and root distribution, the long-term impact of the automatic navigation and positioning system of agricultural machinery on soil texture improvement in different growing seasons is analyzed.

[0036] A multifunctional soil detector is used to obtain soil quality indicators, wherein the soil quality indicators include soil density, water content and nutrient content. Based on the soil quality indicators, crop growth indicators are collected through biosensors, wherein the crop growth indicators include aboveground biomass and yield per unit area. With respect to the crop growth indicators, a root distribution scanner is used to obtain a root spatial distribution image, and a root morphological feature vector is extracted from the root spatial distribution image through an image segmentation algorithm, wherein the root morphological feature vector includes root depth, root density and distribution range. Based on the soil quality indicators, the crop growth indicators and the root morphological feature vector, a support vector regressor is used to establish a coupling model to obtain a soil texture improvement trend function.

[0037] Specifically, soil samples from multiple growing seasons were collected using soil sampling equipment, and soil quality indicators q = (h, w, n) were measured using a multifunctional soil detector, where h is soil density, w is water content, and n is nutrient content. Biosensors were used to collect crop growth indicators p = (b, y), where b is aboveground biomass and y is yield per unit area. A root distribution scanner was used to obtain spatial distribution images of crop roots, and an image segmentation algorithm was used to extract root morphological feature vectors r = (d, ρ, s), where d is root depth, ρ is root density, and s is distribution range. A deep convolutional neural network was used to generate a root growth curve g(t), where t is growth time. The operating parameter sequence a(t) was read from the agricultural machinery navigation and positioning recorder, including track coverage c, tillage depth d, and driving speed v. The corresponding relationship m(a, q) between operating parameters and soil quality indicators was established using a spatiotemporal database. According to the soil quality index sequence qt and crop growth index sequence pt in the continuous growing season, the soil texture improvement rate v = dq / dt and the crop growth response function f(q,p) were calculated by the time series analyzer. The coupling model k(q,p,r) of soil quality, crop yield and root distribution was established by the support vector regressor, and the soil texture improvement trend function T(t) was calculated in combination with the operation parameter sequence a(t). The trend predictor was used to analyze the long-term characteristics of soil texture improvement, and the improvement prediction curve L(t) and confidence interval B(t) were generated. The contribution index of agricultural machinery navigation operation to soil texture improvement was calculated by the time series evaluator. The long-term monitoring of soil quality indicators is of great significance for evaluating the effect of agricultural machinery operation. In the observation of a certain farmland for five consecutive growing seasons, the soil density decreased from the initial 2.8 MPa to 2.2 MPa, the water content was stable between 22% and 25%, and the effective nutrient content increased by 15%. At the same time, crop growth indicators show that the aboveground biomass of corn increased from 2.8 kg to 3.4 kg per square meter, and the yield per unit area increased by 18%. This trend reflects the continuous improvement of soil quality. Root distribution characteristics are an important indicator to measure the effect of soil improvement. The images obtained by the root distribution scanner in the range of 0 to 60 cm soil layer show that the maximum depth of corn roots extends from 42 cm to 48 cm, the root density increases by 25% in the 20 to 40 cm soil layer, and the horizontal distribution range expands by 15%. The deep convolutional neural network draws a root growth curve by analyzing the root images of consecutive growing seasons, showing that the roots grow most vigorously between 45 and 75 days after sowing. The operation parameters of agricultural machinery navigation are closely related to the effect of soil improvement. When the track coverage rate remains above 95%, the control accuracy of the operation depth reaches plus or minus 2 cm, and the driving speed is stable at 6 km / h. The spatiotemporal database records show that the soil quality indicators in the high-precision navigation operation area are improved 35% faster than those in the conventional operation area, especially the improvement of soil density is more significant.The results of time series analysis showed that the rate of soil texture improvement was faster in the first two growing seasons, with soil density decreasing by an average of 0.2 MPa per season, and then leveling off after the third season. The response of crops to soil improvement was manifested in the continuous increase of biomass and yield, among which the yield growth was significantly positively correlated with the soil nutrient content. The coupling model revealed the complex relationship between soil quality, crop growth and root development. When the soil density dropped below 2.4 MPa, the root system extended faster into the deep soil layer, and the growth rate of crop biomass increased. Agricultural machinery navigation operations continuously improved soil structure by precisely controlling tillage depth and trajectory distribution. The trend prediction results showed that under the condition of maintaining the existing operating parameters, the soil texture improvement index would continue to increase by 8% to 12% in the next two growing seasons, and the width of the confidence interval indicated that the prediction had high reliability. The contribution index of navigation operations to soil improvement reached 0.82, indicating that precision operations play a key role in improving soil quality.

[0038] Based on the above embodiments of the present invention, relevant personnel can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. An automatic navigation and positioning method for agricultural equipment, characterized in that: The method comprises: Obtain the root residue distribution data of the previous crop and the difference in soil compaction, establish a correlation model between the root residue distribution of crops and the difference in soil compaction, and determine the target tillage depth plan for sowing the next crop based on the correlation model; Obtain historical tillage data, including tillage depth, equipment used, trajectory, and changes in soil texture over time. Analyze historical tillage data to obtain the time series characteristics of soil texture changes under different tillage depths and trajectories, and establish a dynamic mapping relationship between tillage depth, trajectory, and soil texture. According to the target tillage depth planning for the next crop sowing, the adjusted operation parameters of the agricultural machinery equipment are obtained in combination with the dynamic mapping relationship, and the operation trajectory and operation depth parameters of the agricultural machinery equipment are dynamically adjusted according to the dynamic mapping relationship to make the operation depth match the target tillage depth planning; Determine whether the impact of the test evaluation data of the adjusted operation trajectory and depth parameters on soil texture improvement reaches the expected goal. If not, continue to dynamically adjust the operation trajectory and operation depth parameters of the agricultural machinery equipment until the soil texture improvement requirements are met; Apply the operation trajectory and depth parameters that meet the requirements of soil texture improvement to the automatic navigation and positioning system of agricultural equipment to automatically control and dynamically adjust the operation process of agricultural machinery and equipment; After each growing season, the crop root residue distribution data and soil texture data are obtained, and the association model is updated using the incremental learning algorithm to adjust the relevant operating parameters of the agricultural machinery navigation system, including the operating trajectory and depth parameters; Based on the adjusted operation trajectory and depth parameters, by tracking soil quality indicators, crop yield, and root distribution, the long-term impact of the automatic navigation and positioning system of agricultural machinery on soil texture improvement in different growing seasons is analyzed.

2. The method according to claim 1, characterized in that The method of obtaining the root residue distribution data of the previous crop and the soil compaction difference, establishing a correlation model between the root residue distribution of crops and the soil compaction difference, and determining the target tillage depth planning for sowing the next crop according to the correlation model includes: A soil resistivity detector is used to obtain soil resistivity data, a wireless sensor is used to collect soil moisture content data, and a soil density value is calculated based on the soil resistivity data and the soil moisture content data; The root distribution image is obtained from the soil profile by using an underground biomass detector, and the root distribution image is processed by a grayscale threshold segmentation method to obtain the root residual area ratio within the soil layer depth range; A soil compaction evaluation function is established for each soil depth interval, and a soil compaction prediction model is trained based on the soil density value and the root residual area ratio. Each soil layer is evaluated by the soil compaction prediction model to obtain soil compaction grade data, and the lowest depth interval of the soil compaction grade is searched under the condition of satisfying the depth constraint according to the minimum tillage depth requirement of the next crop to obtain the tillage depth value.

3. The method according to claim 1, characterized in that The historical tillage data are obtained, including tillage depth, equipment used, trajectory, and changes in soil texture over time. The historical tillage data are analyzed to obtain the time series characteristics of soil texture changes under different tillage depths and different trajectories, and to establish a dynamic mapping relationship between tillage depth, trajectory, and soil texture, including: Acquire a track coordinate point sequence of the agricultural machinery from the agricultural machinery operation database, and obtain track density distribution data using a track density calculation formula according to the track point sequence; A soil particle size analyzer is used to obtain the sand content, silt content and clay content in the soil sampling data, and the soil particle composition distribution data is obtained through the Kriging interpolation algorithm based on the soil particle content; Establishing a tillage depth time series data set according to the tillage depth data in the agricultural machinery operation record, wherein the tillage depth time series data set includes a time series and a tillage depth value at a corresponding time; According to the soil particle composition distribution data, the soil texture type data is obtained by using the soil texture triangulation method, and a soil texture prediction model is established by using a long short-term memory neural network. The input variables of the prediction model include the tillage depth time series data set, the trajectory density distribution data and the soil particle composition distribution data.

4. The method according to claim 1, characterized in that The method of obtaining adjusted operation parameters of agricultural machinery equipment according to the target tillage depth planning for sowing the next crop in combination with the dynamic mapping relationship, and dynamically adjusting the operation trajectory and operation depth parameters of the agricultural machinery equipment according to the dynamic mapping relationship so that the operation depth matches the target tillage depth planning includes: Acquire soil density and water content data collected by a soil detector, and calculate the actual maximum tillage depth limit value of the agricultural machinery equipment according to the soil density and water content data; A plot surveying instrument is used to obtain the boundary coordinates and terrain elevation data of the cultivated land area, and a minimum partitioning algorithm is executed based on the boundary coordinates and terrain elevation data to obtain an operation sub-area with uniform slope, and an A-star algorithm is executed based on the operation sub-area to generate an optimal operation trajectory route; The track coverage density is calculated from the optimal operation track route, and the operation parameters that meet the expected soil texture are calculated using an online optimizer according to the track coverage density and the soil texture prediction function; Wireless data sensors are used to collect the operating status of agricultural machinery in real time. If there is a deviation between the operating status and the expected operating parameters, a correction amount is obtained according to the deviation, and the control parameters of the agricultural machinery are corrected in real time according to the correction amount.

5. The method according to claim 1, characterized in that The test evaluation data of the adjusted operation trajectory and depth parameters are used to determine whether the effect on soil texture improvement reaches the expected goal. If not, the operation trajectory and operation depth parameters of the agricultural machinery equipment are continuously and dynamically adjusted until the soil texture improvement requirements are met, including: A soil analyzer is used to obtain soil particle composition data, wherein the soil particle composition data includes a sand particle ratio, a silt particle ratio, and a clay particle ratio, and a soil texture improvement index is obtained through a neural network according to the soil particle composition data; According to the soil texture improvement index, the improvement target value is obtained from the soil texture standard library, the spatial distribution data of the agricultural machinery operation trajectory and the operation depth data are collected through the wireless sensor network, and the soil density data of the operation area is measured using a soil density detector; Calculate the operation uniformity index based on the spatial distribution data of the agricultural machinery operation trajectory and the operation depth data, and generate an operation quality assessment value in combination with the soil density data of the operation area; According to the deviation between the soil texture improvement index and the improvement target value and the operation quality evaluation value, the operation parameter adjustment amount is obtained through a deep optimization calculator, and the agricultural machinery operation control parameters are calculated using a parameter optimizer. The agricultural machinery operation control parameters include track spacing, driving speed and operation depth.

6. The method according to claim 1, characterized in that The operation trajectory and depth parameters that meet the soil texture improvement requirements are applied to the automatic navigation and positioning system of agricultural equipment to automatically control and dynamically adjust the operation process of agricultural machinery and equipment, including: Acquire differential positioning data and real-time position coordinates of agricultural machinery and equipment, wherein the real-time position coordinates are calculated by a real-time dynamic positioning method based on the differential positioning data; A deep neural network is used to generate a smooth trajectory curve according to the real-time position coordinates and the real-time value of the tillage depth collected by the depth sensor, wherein the curvature change rate of the smooth trajectory curve is less than a preset curvature threshold; According to the smooth trajectory curve and the roll angle and pitch angle data collected by the attitude sensor, a steering control amount is calculated by a state feedback controller, and the steering control amount is used to adjust the driving direction of the agricultural machinery; According to the deviation between the real-time tillage depth value and the target depth value, a proportional-integral controller is used to generate a hydraulic lifting control amount, and the hydraulic lifting control amount is used to adjust the tillage depth of the agricultural machinery.

7. The method according to claim 1, characterized in that After each growing season, the crop root residue distribution data and soil texture data are obtained, and the association model is updated using an incremental learning algorithm to adjust the relevant operating parameters of the agricultural machinery navigation system. The relevant operating parameters include operating trajectory and depth parameters, including: Obtain root spatial distribution images and soil profile samples, measure the sand content, silt content and clay content in the soil samples using a soil particle analyzer, and obtain the soil texture feature vector; According to the soil texture feature vector and the root distribution image, an incremental Bayesian network is used to construct a root distribution and soil texture association model, and a historical operation trajectory vector and a depth record vector are read from an operation database; Mapping and optimizing the soil texture feature vector and the operation trajectory vector is performed through a deep autoencoder to obtain an optimized operation parameter vector; For the optimized operation parameter vector, the trajectory accuracy index and the depth accuracy index are combined, and a parameter optimizer is used to calculate the compensation correction amount to obtain the operation control parameters.

8. The method according to claim 1, characterized in that Based on the adjusted operation trajectory and depth parameters, the long-term impact of the automatic navigation and positioning system of agricultural machinery on soil texture improvement in different growing seasons is analyzed by tracking soil quality indicators, crop yields, and root distribution, including: A multifunctional soil detector is used to obtain soil quality indicators, wherein the soil quality indicators include soil density, water content and nutrient content; According to the soil quality index, collecting crop growth indexes through biosensors, wherein the crop growth indexes include aboveground biomass and yield per unit area; According to the crop growth index, a root system spatial distribution image is obtained by using a root system distribution scanner, and a root system morphological feature vector is extracted from the root system spatial distribution image by using an image segmentation algorithm, wherein the root system morphological feature vector includes root system depth, root system density and distribution range; According to the soil quality index, the crop growth index and the root morphology characteristic vector, a coupling model is established using a support vector regressor to obtain a soil texture improvement trend function.

Images