A method and system for planning agricultural machinery operations based on unmanned aerial vehicles

Abstract

The present invention discloses a method and system for planning agricultural machinery operations based on unmanned aerial vehicles, which relates to the technical field of agricultural operation planning. Based on the acquired geographic image frame, a complete farmland map is formed. This step helps to identify and collect the crop distribution set in the complete terrace map, and provides accurate spatial information for farmland management and operation; and a fuzzy clustering algorithm is used to classify the terrain of the terrace area, analyze the influence of different gradient terrains on the speed of agricultural machinery operation, and construct the corresponding influence coefficient Yxxs. This process provides an operation speed optimization strategy based on actual terrain conditions, thereby improving operation efficiency and utilization of farmland resources. Based on field status data information, a condition analysis model is created and trained, and the terrain complexity coefficient Dfxs in different gradient terrain areas is analyzed and obtained to grasp the conditions in different areas.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural operation planning, and in particular to an agricultural machinery operation planning method and system based on unmanned aerial vehicle (UAV) technology. Background Art

[0002] In the modernization process of agricultural production today, the application of science and technology is becoming increasingly significant. With the rapid development of aerial remote sensing technology and drone technology, drones have been widely used in the agricultural field as a tool for efficiently obtaining aerial information. Especially in the planning of agricultural machinery operations, the various sensors carried by drones can capture and analyze images in the fields in real time.

[0003] However, despite the many advantages brought by the UAV-based agricultural machinery operation planning method, there are also some challenges and shortcomings that cannot be overcome by existing technologies. First, in complex terrain areas such as terraces, existing image processing and analysis still face challenges in processing efficiency and accuracy. Especially for terraces with multi-level terrain changes, higher-level algorithms are needed to ensure accurate terrain classification and operation planning. Secondly, it is necessary to take into account the inherent conditions of each area within the terrace, so as to comprehensively analyze the impact of the terrace status on agricultural machinery operations. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides an agricultural machinery operation planning method and system based on unmanned aerial vehicles, which solves the problems in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for planning agricultural machinery operations based on unmanned aerial vehicles, comprising the following steps:

[0006] S1. Using aerial remote sensing technology, drones equipped with several sets of sensors hover over the terraces to capture video streams, and extract several sets of terrace geographic image frames from the video streams in a time series manner;

[0007] S2, based on the several groups of terraced field geographical image frames obtained in step S1, and in combination with image processing and analysis technology, geometric correction is performed on the several groups of terraced field geographical image frames, and the corrected several groups of terraced field geographical image frames are spliced ​​into a complete farmland map to identify and collect the crop distribution set in the complete terraced field map;

[0008] S3, the entire terrace area is preliminarily divided, the crop distribution set includes relevant terrain data information and field status data information in each preliminary divided area, and the entire terrace area is terrain classified according to the relevant terrain data information in each preliminary divided area and using a fuzzy clustering algorithm to re-divide several groups of preliminary divided areas, and analyze the influence of different gradient terrains on the agricultural machinery operation speed Ys to construct an influence coefficient Yxxs, and if the influence coefficient Yxxs exceeds a limit threshold R, a reminder instruction is issued externally;

[0009] S4. After receiving the reminder instruction, a condition analysis model is created based on the field status data information. After training and testing the condition analysis model, the terrain complexity coefficient Dfxs in different gradient terrain areas is analyzed and obtained. The terrain complexity coefficient Dfxs is associated with the influence coefficient Yxxs in the corresponding gradient terrain area, and after linear normalization, the operation progress index Zczs in the corresponding gradient terrain area is obtained by fitting;

[0010] S5. Pre-set the evaluation threshold Q, and compare and analyze the operation process index Zczs with the evaluation threshold Q to comprehensively analyze the execution planning strategy of the agricultural machinery when facing different gradient terrain areas, and transmit the corresponding execution planning strategy to the driving system of the agricultural machinery to perform the task operation.

[0011] Preferably, S1 specifically comprises the following steps:

[0012] S11, using a drone to hover over the terraced field, and capturing a continuous video stream according to a plurality of sets of sensors carried in the drone, wherein the plurality of sets of sensors include an optical photography sensor, a thermal infrared sensor, and a radar sensor, and the drone is provided with a GPS and an inertial measurement unit;

[0013] S12, using video processing software to extract individual terraced field geographical image frames from the continuous video stream, and setting the extraction frequency according to the frame rate of the video stream, and then saving the extracted groups of terraced field geographical image frames to cloud storage.

[0014] Preferably, S2 specifically includes the following steps:

[0015] S21, firstly, geometric correction is performed on a plurality of groups of terraced field geographical image frames, and then wavelet denoising technology is used to decompose the data information in the plurality of groups of terraced field geographical image frames through wavelet transform, and the noise component is removed in the wavelet domain, and at the same time, key feature information in the image frames is extracted through image enhancement technology, and the distribution of crops in the images is identified and analyzed;

[0016] S22. Sort the continuous terraced field geographical image frames in a corresponding order according to the shooting order of the video stream, calculate the homography matrix between the field geographical image frames through the RANSAC algorithm, and use the homography matrix to perform perspective transformation on the field geographical image frames to splice them into a complete farmland map.

[0017] Preferably, S3 specifically includes the following steps:

[0018] S31, according to the complete farmland map obtained in step S22, the actual terraced field area is divided into a plurality of groups of preliminary divided areas, each of which has an equal area;

[0019] S32, the crop distribution set includes relevant terrain data information and field status data information in each preliminary divided area;

[0020] S321, the relevant terrain data information in each preliminary divided area includes the slope value Pdz and the elevation in each preliminary divided area;

[0021] S322, field status data information includes humidity and flatness in each interval, and the humidity difference Sdz between adjacent areas is calculated by statistical algorithm jc , the mean humidity difference Sdz in the entire terrace avg , leveling difference value Pcc between adjacent areas jc and the mean value of the leveling difference within the entire terrace Pcc avg .

[0022] Preferably, S3 also includes:

[0023] S33, according to the relevant terrain data information in each preliminary divided area, and using the fuzzy clustering algorithm, the gradient terrain is divided into three groups of fuzzy sets, and the three groups of fuzzy sets include gentle gradient levels, medium gradient levels and steep gradient levels;

[0024] S331, and use the fuzzy C-means clustering algorithm to calculate the cluster center value To classify the terrain of the entire terraced area, For example, obtain it as follows:

[0025]

[0026] In the formula, represents the cluster center value at the kth slope level, reflecting the mean of the cluster slope values ​​at the kth slope level; n represents the number of locations with different slope values ​​collected in the entire terrace area, i = 1, 2, 3, ..., n, Pdz i Expressed as the slope value at the i-th position, Y ikIt is expressed as the membership of the slope value at the i-th position to the mean of the cluster slope values ​​at the k-th slope level, and the membership Y ik Update according to the distance from the slope value to the mean slope value of each cluster. The closer the distance, the higher the membership degree.

[0027] S332, the membership degree Y of the slope value at the i-th position to the mean of the cluster slope values ​​at the k-th slope level ik Obtained by the following formula:

[0028]

[0029] Where C is the fuzzification index, which is a parameter that controls the fuzziness of the membership, and C>1, m represents the number of different slope levels into which the gradient terrain is divided, k=1, 2, ..., m, It is expressed as the cluster center value at the kth slope level, It is expressed as the cluster center value at the zth slope level, reflecting the central characteristics of the slope level. Expressed as Pdz i and The Euclidean distance between Expressed as Pdz i and The Euclidean distance between

[0030] S333, pre-set the division threshold, and set the cluster center value Compare and analyze with the division threshold to obtain the division standard of gradient terrain division. The specific contents are as follows:

[0031] If the cluster center value When it is greater than the division threshold, it will be defined as a steep gradient level;

[0032] If the cluster center value When it is equal to the division threshold, it will be defined as a medium gradient level;

[0033] If the cluster center value When it is less than the division threshold, it will be defined as a gentle gradient level;

[0034] S334. According to the division standard in step S333, the several groups of preliminary divided areas are divided again. At this time, the entire terrace area is divided into regions according to the steep gradient level, the medium gradient level and the gentle gradient level, and marked as the first region, the second region, the third region, ..., the pth region, respectively.

[0035] Preferably, S3 also includes:

[0036] S33. According to the re-divided terraced area, the influence of different gradient terrains on the agricultural machinery operation speed Ys is analyzed to construct the influence coefficient Yxxs. The influence coefficient Yxxs is obtained by the following formula:

[0037]

[0038] In the formula, Yxxs represents the construction influence coefficient, Ys0 represents the operation speed when the terrain gradient is 0, The cluster center value at the kth slope level, a1 and a2 are weight coefficients;

[0039] S331. By comparing the influence coefficient Yxxs with the preset limit threshold R, it is preliminarily determined whether it is necessary to issue a reminder instruction. If the influence coefficient Yxxs exceeds the limit threshold R, a reminder instruction is issued. If the influence coefficient Yxxs does not exceed the limit threshold R, no reminder instruction is issued for the time being.

[0040] Preferably, S4 specifically comprises the following steps:

[0041] S41, using a convolutional neural network to build an initial model, and training and testing the initial model with field status data information, and using the trained initial model as a state recognition model, respectively obtaining feature information in the state recognition model, and training and testing the state recognition model with the obtained feature information, and combining the received reminder instruction, using the trained state recognition model as a conditional analysis model;

[0042] S411. Based on the conditional analysis model, the terrain complexity coefficient Dfxs in different gradient terrain areas is analyzed and obtained:

[0043]

[0044] Where α and β are weight coefficients, p represents the number of re-divided regions, j = 1, 2, 3, ..., p, Sdz jc Expressed as the humidity difference between adjacent areas, Sdz avg It is expressed as the mean of the humidity difference in the entire terrace, Pcc jc Expressed as the leveling difference between adjacent areas, Pcc avg It is expressed as the mean of the leveling difference values ​​within the entire terrace.

[0045] Preferably, S4 also includes:

[0046] S42, by associating the terrain complexity coefficient Dfxs with the influence coefficient Yxxs in the corresponding gradient terrain area, and after linear normalization processing, fitting to obtain the operation progress index Zczs in the corresponding gradient terrain area, specifically according to the following formula:

[0047]

[0048] In the formula, Zzmd represents the planting density, Dfxs represents the terrain complexity coefficient, Yxxs represents the influence coefficient in the corresponding gradient terrain area, F1 represents the weight coefficient of the influence coefficient Yxxs in the corresponding gradient terrain area, F2 represents the weight coefficient of the terrain complexity coefficient Dfxs, F3 represents the weight coefficient of the planting density Zzmd, and X represents the correction constant. By calculating the terrain complexity coefficient Dfxs, the influence coefficient Yxxs in the corresponding gradient terrain area and the planting density Zzmd, the arrangement status of the agricultural machinery operation process in the corresponding gradient terrain area is analyzed.

[0049] Preferably, S5, by comparing and analyzing the operation progress index Zczs with the evaluation threshold Q, a comprehensive analysis is made of the execution planning strategy of the agricultural machinery when facing different gradient terrain areas, and the specific contents are as follows:

[0050] If the operation process index Zczs ≥ the evaluation threshold Q, it indicates that the operation execution efficiency of the agricultural machinery in the current terrace terrain area is in a normal state. At this time, the operation speed of the agricultural machinery will be increased to speed up the operation process, save time costs, optimize the operation path according to the crop distribution, reduce the turning frequency and repeated operations, and further improve the operation efficiency;

[0051] If the operation process index Zczs is less than the evaluation threshold Q, it indicates that the operation execution efficiency of the agricultural machinery in the current terrace terrain area is in an abnormal state. At this time, the current operation speed of the agricultural machinery will be maintained to ensure the operation quality and safety.

[0052] An agricultural machinery operation planning system based on unmanned aerial vehicles includes a capture module, a processing module, a region division module, a preliminary analysis module, a comprehensive analysis module and a planning module;

[0053] The capture module uses aerial remote sensing technology to enable drones equipped with several sets of sensors to hover over the terraces in advance to capture video streams, and extracts several sets of terrace geographic image frames from the video streams in a time series manner;

[0054] A processing module, based on a plurality of terraced field geographical image frames and in combination with image processing and analysis technology, performs geometric correction on the plurality of terraced field geographical image frames, and splices the corrected plurality of terraced field geographical image frames into a complete farmland map, so as to identify and collect a crop distribution set in the complete terraced field map;

[0055] The area division module preliminarily divides the entire terrace area. The crop distribution set includes relevant terrain data information and field status data information in each preliminarily divided area. Based on the relevant terrain data information in each preliminarily divided area, and using the fuzzy clustering algorithm, the entire terrace area is terrain classified to re-divide several groups of preliminarily divided areas.

[0056] The preliminary analysis module analyzes the influence of different gradient terrains on the operation speed Ys of agricultural machinery to construct the influence coefficient Yxxs. If the influence coefficient Yxxs exceeds the limit threshold R, a reminder instruction is issued externally;

[0057] The comprehensive analysis module, after receiving the reminder instruction, creates a conditional analysis model based on the field status data information, and after training and testing the conditional analysis model, analyzes and obtains the terrain complexity coefficient Dfxs in different gradient terrain areas, and obtains the operation progress index Zczs in the corresponding gradient terrain area by correlating the terrain complexity coefficient Dfxs with the influence coefficient Yxxs in the corresponding gradient terrain area and performing linear normalization processing;

[0058] The planning module pre-sets the evaluation threshold Q, and compares and analyzes the operation process index Zczs with the evaluation threshold Q to comprehensively analyze the execution planning strategy of the agricultural machinery when facing different gradient terrain areas, and transmits the corresponding execution planning strategy to the driving system of the agricultural machinery to perform the task operation.

[0059] The present invention provides a method and system for planning agricultural machinery operations based on drones, which have the following beneficial effects:

[0060] (1) Based on the acquired geographic image frames, a complete farmland map is formed. This step helps to identify and collect the crop distribution set in the complete terrace map, providing accurate spatial information for farmland management and operation; and the fuzzy clustering algorithm is used to classify the terrain of the terrace area, analyze the influence of different gradient terrain on the operation speed of agricultural machinery, and construct the corresponding influence coefficient Yxxs. This process provides an operation speed optimization strategy based on actual terrain conditions, thereby improving operation efficiency and the utilization rate of farmland resources. Based on the field status data information, a condition analysis model is created and trained to analyze and obtain the terrain complexity coefficient Dfxs in different gradient terrain areas to grasp the conditions in different areas. Finally, the operation process index Zczs obtained can not only quantify the impact of terrain complexity on operation efficiency, but also provide a scientific basis for decision-making, and comprehensively analyze the execution planning strategy of agricultural machinery in different gradient terrain areas. The optimized execution planning strategy is transmitted to the driving system of the agricultural machinery in real time, so that the agricultural machinery can perform task operations intelligently and improve operation efficiency and the utilization efficiency of farmland resources. In short, this method can not only classify the different gradients of terraces, but also further analyze the impact on agricultural machinery operations through the classified fields, thereby realizing intelligent processing of farmland management and ensuring the accuracy and efficiency of agricultural production.

[0061] (2) Through the precise understanding of terraced terrain and crop distribution, combined with the reclassification of terrain by fuzzy clustering algorithm and the construction of influence coefficient Yxxs, it is possible to evaluate the impact of different terrains on the complexity and execution efficiency of agricultural machinery operations, so as to optimize the operation path and speed planning of agricultural machinery, which can further reduce duplication and waste in operations and improve the efficiency of agricultural machinery operations and resource utilization.

[0062] (3) Use convolutional neural networks to build an initial model, and use field status data information to train and test the initial model. The trained initial model is used as a state recognition model to obtain feature information in different terraced terrain areas, and the state recognition model is further trained and tested to improve the accuracy and adaptability of the model. This step ensures that the state recognition model can effectively reflect the actual state of the farmland. By calculating the operation process index Zczs, the system can accurately evaluate the efficiency and safety of agricultural machinery operations under different terrain conditions. This precise evaluation and analysis helps agricultural managers optimize agricultural machinery operation paths and speed planning, improve operation efficiency and reduce resource waste.

[0063] (4) The system can dynamically adjust the operating speed and path planning of agricultural machinery according to the actual conditions of different terraced terrain areas. In this way, whether in normal or abnormal operating conditions, the system can provide the best execution planning strategy to ensure the continuity and stability of operations. This dynamic response mechanism can reduce the damage to farmland and crops caused by agricultural machinery during operation, and also reduce the probability of accidents involving agricultural machinery. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A schematic diagram of the steps of a method for planning agricultural machinery operations based on a drone according to the present invention;

[0065] Figure 2 The present invention is a flowchart of an agricultural machinery operation planning system based on unmanned aerial vehicles. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] Example 1

[0068] See also Figure 1 The present invention provides an agricultural machinery operation planning method based on unmanned aerial vehicle, comprising the following steps:

[0069] S1. Using aerial remote sensing technology, drones equipped with several sets of sensors hover over the terraces to capture video streams, and extract several sets of terrace geographic image frames from the video streams in a time series manner;

[0070] S2, based on the several groups of terraced field geographical image frames obtained in step S1, and in combination with image processing and analysis technology, geometric correction is performed on the several groups of terraced field geographical image frames, and the corrected several groups of terraced field geographical image frames are spliced ​​into a complete farmland map to identify and collect the crop distribution set in the complete terraced field map;

[0071] S3, the entire terrace area is preliminarily divided, the crop distribution set includes relevant terrain data information and field status data information in each preliminary divided area, and the entire terrace area is terrain classified according to the relevant terrain data information in each preliminary divided area and using a fuzzy clustering algorithm to re-divide several groups of preliminary divided areas, and analyze the influence of different gradient terrains on the agricultural machinery operation speed Ys to construct an influence coefficient Yxxs, and if the influence coefficient Yxxs exceeds a limit threshold R, a reminder instruction is issued externally;

[0072] S4. After receiving the reminder instruction, a condition analysis model is created based on the field status data information. After training and testing the condition analysis model, the terrain complexity coefficient Dfxs in different gradient terrain areas is analyzed and obtained. The terrain complexity coefficient Dfxs is associated with the influence coefficient Yxxs in the corresponding gradient terrain area, and after linear normalization, the operation progress index Zczs in the corresponding gradient terrain area is obtained by fitting;

[0073] S5. Pre-set the evaluation threshold Q, and compare and analyze the operation process index Zczs with the evaluation threshold Q to comprehensively analyze the execution planning strategy of the agricultural machinery when facing different gradient terrain areas, and transmit the corresponding execution planning strategy to the driving system of the agricultural machinery to perform the task operation.

[0074] In this embodiment, the terraced field geographical images captured by aerial remote sensing technology and drones, combined with image processing and analysis technology, can accurately identify and collect the distribution of crops in the farmland. These data help agricultural managers better understand the real-time status of each piece of land, thereby optimizing planting strategies and agricultural machinery operation planning, and further improving the efficiency and quality of agricultural production; the terraced field terrain is classified using a fuzzy clustering algorithm, the influence of different gradient terrains on the operation speed of agricultural machinery is analyzed, and the influence coefficient Yxxs is constructed, which helps to accurately plan the operation path and speed of agricultural machinery. This personalized operation planning not only saves time and resources, but also reduces unnecessary agricultural machinery operation and operation costs. When the influence coefficient Yxxs exceeds the preset threshold, the system can issue a reminder instruction in time, prompting the farm manager or agricultural machinery operator to adjust the operation strategy. At the same time, based on the establishment and training of the conditional analysis model, combined with the comparison analysis of the operation process index Zczs, intelligent support is provided for the real-time decision-making of agricultural machinery to ensure the safety and efficiency of the operation process. In short, this method can not only improve the efficiency and quality of agricultural production, but also effectively solve many challenges in existing agricultural management through data protection and rational use, and promote the realization of sustainable agricultural development and intelligent management.

[0075] Example 2

[0076] Please refer to Figure 1 Specifically: S1 specifically includes the following steps:

[0077] S11, using a drone to hover over the terraced field, and capturing a continuous video stream according to a plurality of sets of sensors carried in the drone, wherein the plurality of sets of sensors include an optical photography sensor, a thermal infrared sensor, and a radar sensor, and the drone is provided with a GPS and an inertial measurement unit;

[0078] S12. Use video processing software, such as OpenCV or FFmpeg, to extract individual terraced field geographic image frames from a continuous video stream, and set the extraction frequency according to the frame rate of the video stream, such as extracting one frame per second (frame rate is 1), extracting one frame every 5 seconds (frame rate is 0.2), etc., and then save the extracted groups of terraced field geographic image frames to cloud storage for subsequent processing and analysis. At the same time, the extracted image frames can also be further analyzed, such as terraced field shape recognition, crop classification, soil texture analysis, etc. This usually requires the use of image processing and computer vision algorithms, such as edge detection, image segmentation and feature extraction techniques.

[0079] In this embodiment, a drone equipped with an optical photography sensor, a thermal infrared sensor, and a radar sensor can be used to capture continuous video streams above the terraces. The optical photography sensor is used to obtain image information within the visible spectrum, the thermal infrared sensor can detect the heat distribution of farmland vegetation, and the radar sensor is used to detect the elevation and structure of the ground. The comprehensive application of these sensors makes it possible to obtain multi-dimensional information about farmland, which helps to fully understand the growth status and environmental conditions of farmland. These image frames can not only be used for subsequent geographic correction and splicing, but also for further image processing and computer vision algorithm applications such as terrace shape recognition, crop classification, and soil texture analysis. Such real-time processing and analysis capabilities improve the accuracy and response speed of farmland management. Through image processing and computer vision algorithms, combined with multi-dimensional information captured by sensor data, intelligent management of farmland is achieved. For example, based on the results of terrace shape recognition and crop classification, accurate agricultural machinery operation plans and fertilization and irrigation strategies can be formulated to optimize the utilization efficiency and operation quality of farmland resources.

[0080] Example 3

[0081] Please refer to Figure 1 Specifically: S2 specifically includes the following steps:

[0082] S21, firstly, geometric correction is performed on a plurality of groups of terraced field geographical image frames, and then wavelet denoising technology is used to decompose the data information in the plurality of groups of terraced field geographical image frames through wavelet transform, and the noise component is removed in the wavelet domain, and at the same time, key feature information in the image frames is extracted through image enhancement technology, and the distribution of crops in the images is identified and analyzed;

[0083] The application of geometric correction and wavelet denoising technology ensures the geometric accuracy and data quality stability of the generated farmland geographic image frame. These processing steps effectively remove noise and deformation in the image, making the final generated farmland map have high accuracy and resolution, providing reliable visual data support for agricultural decision-making;

[0084] S22. Sort the continuous terraced field geographical image frames in a corresponding order according to the shooting order of the video stream, calculate the homography matrix between the field geographical image frames through the RANSAC algorithm, and use the homography matrix to perform perspective transformation on the field geographical image frames to splice them into a complete farmland map.

[0085] By using homography matrix calculation and perspective transformation technology, multiple terraced field geographic image frames are spliced ​​into a complete farmland map, achieving a display from local to panoramic perspective. Such a panoramic perspective can help agricultural managers fully understand the terrain characteristics and crop distribution of the terraced fields, and provide more detailed and comprehensive data basis for agricultural machinery operation path planning and resource management.

[0086] In this embodiment, geometric correction is first performed on the extracted groups of terraced field geographical image frames to ensure the accuracy and consistency of the image. Then, the wavelet transform technology is applied to process the image, decompose the image data and remove the noise component in the wavelet domain, which helps to improve the clarity and quality of the image. At the same time, the image enhancement technology is used to extract key feature information from the processed image frames, such as the distribution of crops, to provide a reliable data basis for subsequent analysis and decision-making. According to the shooting order of the video stream, the continuous terraced field geographical image frames are sorted in sequence. The RANSAC algorithm is used to calculate the homography matrix between the field geographical image frames, which can effectively describe the geometric transformation relationship between different image frames. Using the calculated homography matrix, each field geographical image frame is perspective transformed to realize the splicing of the image into a complete farmland map. Such a processing process can provide farmland topography and crop distribution information under a panoramic perspective, and provide more comprehensive and accurate visual data support for agricultural management and operation planning.

[0087] Example 4

[0088] Please refer to Figure 1 Specifically: S3 specifically includes the following steps:

[0089] S31, according to the complete farmland map obtained in step S22, the actual terraced field area is divided into a plurality of groups of preliminary divided areas, each of which has an equal area;

[0090] S32, the crop distribution set includes relevant terrain data information and field status data information in each preliminary divided area;

[0091] S321, the relevant terrain data information in each preliminary divided area includes the slope value Pdz and the elevation in each preliminary divided area;

[0092] S322, field status data information includes humidity and flatness in each interval, and the humidity difference Sdz between adjacent areas is calculated by statistical algorithmjc , the mean humidity difference Sdz in the entire terrace avg , leveling difference value Pcc between adjacent areas jc and the mean value of the leveling difference within the entire terrace Pcc avg .

[0093] S3 also includes:

[0094] S33, according to the relevant terrain data information in each preliminary divided area, and using the fuzzy clustering algorithm, the gradient terrain is divided into three groups of fuzzy sets, and the three groups of fuzzy sets include gentle gradient levels, medium gradient levels and steep gradient levels;

[0095] S331, and use the fuzzy C-means clustering algorithm to calculate the cluster center value To classify the terrain of the entire terraced area, For example, obtain it as follows:

[0096]

[0097] In the formula, represents the cluster center value at the kth slope level, reflecting the mean of the cluster slope values ​​at the kth slope level; n represents the number of locations with different slope values ​​collected in the entire terrace area, i = 1, 2, 3, ..., n, Pdz i Expressed as the slope value at the i-th position, Y ik It is expressed as the membership of the slope value at the i-th position to the mean of the cluster slope values ​​at the k-th slope level, and the membership Y ik Update according to the distance from the slope value to the mean slope value of each cluster. The closer the distance, the higher the membership degree.

[0098] The above slope value Pdz can be monitored and obtained through an inertial measurement unit (IMU), a global positioning system or a lidar;

[0099] S332, the membership degree Y of the slope value at the i-th position to the mean of the cluster slope values ​​at the k-th slope level ik Obtained by the following formula:

[0100]

[0101] Where C is the fuzzification index, which is a parameter that controls the fuzziness of the membership, and C>1, m represents the number of different slope levels into which the gradient terrain is divided, k=1, 2, ..., m, It is expressed as the cluster center value at the kth slope level, It is expressed as the cluster center value at the zth slope level, reflecting the central characteristics of the slope level. Expressed as Pdz i and The Euclidean distance between Expressed as Pdz i and The Euclidean distance between

[0102] The fuzzification index C usually takes a real value greater than 1, and is usually set to 2, because in this case the objective function is easier to solve and can effectively distinguish different clusters.

[0103] S333, pre-set the division threshold, and set the cluster center value Compare and analyze with the division threshold to obtain the division standard of gradient terrain division. The specific contents are as follows:

[0104] If the cluster center value When it is greater than the division threshold, it will be defined as a steep gradient level;

[0105] If the cluster center value When it is equal to the division threshold, it will be defined as a medium gradient level;

[0106] If the cluster center value When it is less than the division threshold, it will be defined as a gentle gradient level;

[0107] S334. According to the division standard in step S333, the several groups of preliminary divided areas are divided again. At this time, the entire terrace area is divided into regions according to the steep gradient level, the medium gradient level and the gentle gradient level, and marked as the first region, the second region, the third region, ..., the pth region, respectively.

[0108] In this embodiment, according to the complete farmland map obtained in the aforementioned step S22, the actual terrace area is divided into several groups of preliminary division areas to ensure that the areas of each preliminary division area are equal. Each preliminary division area contains rich terrain data information and field status data information, which are the basis for subsequent analysis and decision-making. Through the detailed analysis of fuzzy clustering algorithm and terrain data, the refined classification and regional division of terrace terrain is realized. This refined division can help agricultural managers to have a deep understanding of the characteristics and changes of different terrain areas, and provide a more accurate basis for operation planning and resource allocation. Analyzing the influence of different gradient terrains on the operation speed of agricultural machinery and constructing the influence coefficient Yxxs can help optimize the operation path and speed adjustment of agricultural machinery, improve operation efficiency and resource utilization efficiency. According to the analysis of the status data information in the terrace, such as humidity and flatness, the operation strategy is further optimized to reduce resource waste and environmental impact. By setting the division threshold and comparison analysis, the intelligent classification and evaluation of different gradient terrains are realized. Once the influence coefficient Yxxs exceeds the set threshold, the system will automatically issue a reminder instruction to help agricultural managers adjust the operation strategy in time to ensure operation safety and efficiency. Based on advanced data collection and analysis technology, agricultural management has been made data-driven and scientific. This method not only improves the efficiency of acquiring and processing farmland data, but also promotes the progress of modern agricultural management and provides important support for sustainable agricultural development.

[0109] Example 5

[0110] Please refer to Figure 1 , specifically: S3 also includes:

[0111] S33. According to the re-divided terraced area, the influence of different gradient terrains on the agricultural machinery operation speed Ys is analyzed to construct the influence coefficient Yxxs. The influence coefficient Yxxs is obtained by the following formula:

[0112]

[0113] In the formula, Yxxs represents the construction influence coefficient, Ys0 represents the operation speed when the terrain gradient is 0, The cluster center value at the kth slope level, a1 and a2 are both weight coefficients, where 0<a1≤1, 0<a2≤1, and a1+a2=1.

[0114] S331. By comparing the influence coefficient Yxxs with the preset limit threshold R, it is preliminarily determined whether it is necessary to issue a reminder instruction. If the influence coefficient Yxxs exceeds the limit threshold R, a reminder instruction is issued. If the influence coefficient Yxxs does not exceed the limit threshold R, no reminder instruction is issued for the time being.

[0115] In this embodiment, by constructing the influence coefficient Yxxs, the system can quantify the influence of different gradient terrains on the speed of agricultural machinery operation, which helps agricultural managers and agricultural machinery operators to accurately understand and predict the operating efficiency under different terrain conditions, thereby optimizing the operation path and speed planning, and improving the operating efficiency and farmland utilization. By comparing the influence coefficient Yxxs with the preset limit threshold R, the system can judge in real time whether the current terrace terrain is suitable for the current operating speed. By issuing reminder instructions, agricultural machinery operators can adjust the operation strategy in time to cope with terrain changes and operating challenges under different conditions, and realize effective resource management and real-time adjustment of operations. By systematically analyzing and quantifying the impact of different terrains on agricultural machinery operations, data-driven decision-making can effectively improve the accuracy and quality of farmland operations. This refined operation management method helps to reduce resource waste, improve the stability and sustainability of the crop growth environment, and thus promote the continuous growth of agricultural production efficiency. Through the above method, the method can not only improve the intelligence level and technical content of agricultural production, but also effectively cope with complex farmland terrain and environmental conditions, and provide important support for the sustainable development of modern agriculture.

[0116] Example 6

[0117] Please refer to Figure 1 Specifically: S4 specifically includes the following steps:

[0118] S41, using a convolutional neural network to build an initial model, and training and testing the initial model with field status data information, and using the trained initial model as a state recognition model, respectively obtaining feature information in the state recognition model, and training and testing the state recognition model with the obtained feature information, and combining the received reminder instruction, using the trained state recognition model as a conditional analysis model;

[0119] S411. Based on the conditional analysis model, the terrain complexity coefficient Dfxs in different gradient terrain areas is analyzed and obtained:

[0120]

[0121] Where α and β are weight coefficients, p represents the number of re-divided regions, j = 1, 2, 3, ..., p, Sdz jc Expressed as the humidity difference between adjacent areas, Sdz avg It is expressed as the mean of the humidity difference in the entire terrace, Pcc jc Expressed as the leveling difference between adjacent areas, Pcc avg It is expressed as the mean of the leveling difference values ​​within the entire terrace.

[0122] The above humidity values ​​can be monitored and obtained through humidity sensors;

[0123] Flatness can be monitored and obtained through laser radar or laser rangefinder;

[0124] S4 also includes:

[0125] S42, by associating the terrain complexity coefficient Dfxs with the influence coefficient Yxxs in the corresponding gradient terrain area, and after linear normalization processing, fitting to obtain the operation progress index Zczs in the corresponding gradient terrain area, specifically according to the following formula:

[0126]

[0127] Wherein, Zzmd represents the planting density, Dfxs represents the terrain complexity coefficient, Yxxs represents the influence coefficient in the corresponding gradient terrain area, F1 represents the weight coefficient of the influence coefficient Yxxs in the corresponding gradient terrain area, F2 represents the weight coefficient of the terrain complexity coefficient Dfxs, and F3 represents the weight coefficient of the planting density Zzmd. Among them, 0<F1≤1, 0<F2≤1, 0<F3≤1, and F1+F2+F3=1, and X represents the correction constant. By calculating the terrain complexity coefficient Dfxs, the influence coefficient Yxxs in the corresponding gradient terrain area and the planting density Zzmd, the arrangement status of the agricultural machinery operation process in the corresponding gradient terrain area is analyzed.

[0128] The above-mentioned planting density Zzmd can be monitored and obtained by optical photography sensors;

[0129] In this embodiment, through the comprehensive analysis of the terrain complexity coefficient Dfxs and the influence coefficient Yxxs, accurate suggestions for agricultural machinery operation arrangements are provided for the operation progress index Zczs, which helps to optimize the path planning and operation speed setting of agricultural machinery, and improve agricultural production efficiency and land utilization. Using deep learning models and data analysis technology, the system can obtain and analyze the status information of the terraced area in real time. This information is not only used for the calculation of the operation progress index Zczs, but also provides decision support and real-time feedback for farm managers, helping to quickly respond to changes in the farmland environment. Through the scientific calculation of the operation progress index Zczs, farms can effectively manage agricultural machinery resources and human resources, reduce operation time and cost, while maximizing the use of farmland resources and promoting sustainable agricultural development.

[0130] Example 7

[0131] Please refer to Figure 1 Specifically: S5. By comparing and analyzing the operation process index Zczs with the evaluation threshold Q, the execution planning strategy of agricultural machinery in the face of different gradient terrain areas is comprehensively analyzed. The specific contents are as follows:

[0132] If the operation process index Zczs ≥ the evaluation threshold Q, it indicates that the operation execution efficiency of the agricultural machinery in the current terrace terrain area is in a normal state. At this time, the operation speed of the agricultural machinery will be increased to speed up the operation process, save time costs, optimize the operation path according to the crop distribution, reduce unnecessary turns and repeated operations, and further improve the operation efficiency;

[0133] If the operation process index Zczs is less than the evaluation threshold Q, it indicates that the operation execution efficiency of the agricultural machinery in the current terrace terrain area is in an abnormal state. At this time, the current operation speed of the agricultural machinery will be maintained to ensure the quality and safety of the operation, avoid equipment damage or failure, reduce the operation intensity, reduce the impact on farmland and crops, and avoid unnecessary losses and waste.

[0134] In this embodiment, by comparing and analyzing the operation progress index Zczs with the evaluation threshold Q, the system can monitor the operation execution status of agricultural machinery in different gradient terrain areas in real time. When the operation progress index Zczs ≥ the evaluation threshold Q, the system determines that the operation efficiency of the agricultural machinery is normal and automatically increases the operation speed of the agricultural machinery. Through this method of dynamically adjusting the operation speed, the operation process of the agricultural machinery can be significantly accelerated, saving a lot of time costs, especially in the busy farming season. This time saving is of great significance to the overall efficiency improvement of agricultural production. While improving the operation speed, the system also optimizes the operation path of the agricultural machinery according to the crop distribution. By reducing unnecessary turns and repeated operations, the system can reduce fuel consumption and equipment wear, and further improve the operation efficiency. This strategy of optimizing the path not only reduces resource waste, but also reduces the negative impact on the environment, which is in line with the sustainable development concept of modern agriculture. When the operation process index Zczs is less than the evaluation threshold Q, the system determines that the operation efficiency of agricultural machinery is in an abnormal state. At this time, the current operation speed of agricultural machinery will be maintained. In this way, the system prioritizes the operation quality and equipment safety, avoiding equipment damage or failure caused by excessive speed increase. At the same time, reducing the operation intensity helps to reduce the impact on farmland and crops, and avoid unnecessary losses and waste. This conservative strategy is particularly important in complex terrain or severe weather conditions, and can effectively reduce the operation risk. By real-time monitoring and analyzing the operation process index Zczs, the system can flexibly adjust the operation speed and path of agricultural machinery according to actual conditions. This real-time response mechanism makes agricultural machinery operations more intelligent and efficient, and can cope with different terrain and environmental changes, and provide the best operation plan. Intelligent scheduling not only improves the efficiency of agricultural machinery, but also improves the intelligence and automation level of agricultural production. The strategy of dynamically adjusting the operation speed and optimizing the operation path can not only save fuel and reduce equipment wear, but also reduce the overall cost of agricultural machinery operations. Reducing resource waste and environmental impact not only helps to improve the economic benefits of agricultural production, but also promotes the development of green agriculture.

[0135] Example 8

[0136] Please refer to Figure 1 and Figure 2 ,Specifically: An agricultural machinery operation planning system based on ,UAV, includes a capture module, a processing module, a region division module, a preliminary ,analysis module, a comprehensive analysis module and a planning module;

[0137] The capture module uses aerial remote sensing technology to enable drones equipped with several sets of sensors to hover over the terraces in advance to capture video streams, and extracts several sets of terrace geographic image frames from the video streams in a time series manner;

[0138] A processing module, based on a plurality of terraced field geographical image frames and in combination with image processing and analysis technology, performs geometric correction on the plurality of terraced field geographical image frames, and splices the corrected plurality of terraced field geographical image frames into a complete farmland map, so as to identify and collect a crop distribution set in the complete terraced field map;

[0139] The area division module preliminarily divides the entire terrace area. The crop distribution set includes relevant terrain data information and field status data information in each preliminarily divided area. Based on the relevant terrain data information in each preliminarily divided area, and using the fuzzy clustering algorithm, the entire terrace area is terrain classified to re-divide several groups of preliminarily divided areas.

[0140] The preliminary analysis module analyzes the influence of different gradient terrains on the operation speed Ys of agricultural machinery to construct the influence coefficient Yxxs. If the influence coefficient Yxxs exceeds the limit threshold R, a reminder instruction is issued externally;

[0141] The comprehensive analysis module, after receiving the reminder instruction, creates a conditional analysis model based on the field status data information, and after training and testing the conditional analysis model, analyzes and obtains the terrain complexity coefficient Dfxs in different gradient terrain areas, and obtains the operation progress index Zczs in the corresponding gradient terrain area by correlating the terrain complexity coefficient Dfxs with the influence coefficient Yxxs in the corresponding gradient terrain area and performing linear normalization processing;

[0142] The planning module pre-sets the evaluation threshold Q, and compares and analyzes the operation process index Zczs with the evaluation threshold Q to comprehensively analyze the execution planning strategy of the agricultural machinery when facing different gradient terrain areas, and transmits the corresponding execution planning strategy to the driving system of the agricultural machinery to perform the task operation.

[0143] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for planning agricultural machinery operations based on unmanned aerial vehicles, characterized in that: The following steps are included: S1. Using aerial remote sensing technology, drones equipped with several sets of sensors hover over the terraces to capture video streams, and extract several sets of terrace geographic image frames from the video streams in a time series manner; S2, based on the several groups of terraced field geographical image frames obtained in step S1, and in combination with image processing and analysis technology, geometric correction is performed on the several groups of terraced field geographical image frames, and the corrected several groups of terraced field geographical image frames are spliced ​​into a complete farmland map to identify and collect the crop distribution set in the complete terraced field map; S3, the entire terrace area is preliminarily divided, the crop distribution set includes relevant terrain data information and field status data information in each preliminary divided area, and the entire terrace area is terrain classified according to the relevant terrain data information in each preliminary divided area and using a fuzzy clustering algorithm to re-divide several groups of preliminary divided areas, and analyze the influence of different gradient terrains on the agricultural machinery operation speed Ys to construct an influence coefficient Yxxs, and if the influence coefficient Yxxs exceeds a limit threshold R, a reminder instruction is issued externally; S3 also includes: S33, according to the relevant terrain data information in each preliminary divided area, and using the fuzzy clustering algorithm, the gradient terrain is divided into three groups of fuzzy sets, and the three groups of fuzzy sets include gentle gradient levels, medium gradient levels and steep gradient levels; S331, and use the fuzzy C-means clustering algorithm to calculate the cluster center value , to classify the terrain of the entire terraced area, specifically For example, obtain it as follows: In the formula, represents the cluster center value at the kth slope level, reflecting the mean of the cluster slope values ​​at the kth slope level; n represents the number of locations with different slope values ​​collected in the entire terrace area, i = 1, 2, 3, ..., n, It is expressed as the slope value at the i-th position, It is expressed as the membership of the slope value at the i-th position to the mean of the cluster slope values ​​at the k-th slope level. Update according to the distance from the slope value to the mean slope value of each cluster. The closer the distance, the higher the membership degree. S332, the degree of membership of the slope value at the i-th position to the mean value of the cluster slope value at the k-th slope level Obtained by the following formula: In the formula, Expressed as a fuzzification index, it is a parameter that controls the degree of fuzziness of membership, and >1, m represents the number of different slope levels into which the gradient terrain is divided, k = 1, 2, ..., m, It is expressed as the cluster center value at the kth slope level, It is expressed as the cluster center value at the zth slope level, reflecting the central characteristics of the slope level. Expressed as and The Euclidean distance between Expressed as and The Euclidean distance between S333, pre-set the division threshold, and set the cluster center value Compare and analyze with the division threshold to obtain the division standard of gradient terrain division. The specific contents are as follows: If the cluster center value When it is greater than the division threshold, it will be defined as a steep gradient level; If the cluster center value When it is equal to the division threshold, it will be defined as a medium gradient level; If the cluster center value When it is less than the division threshold, it will be defined as a gentle gradient level; S334, according to the division standard in step S333, re-dividing the plurality of groups of preliminary divided areas, at this time, dividing the entire terrace area into regions according to the steep gradient level, the medium gradient level and the gentle gradient level, and marking them as the first region, the second region, the third region, ..., the pth region respectively; S3 also includes: S33. According to the re-divided terraced area, the influence of different gradient terrains on the agricultural machinery operation speed Ys is analyzed to construct the influence coefficient Yxxs. The influence coefficient Yxxs is obtained by the following formula: In the formula, Yxxs represents the construction influence coefficient, It is expressed as the operating speed when the terrain gradient is 0, The cluster center value at the k-th slope level, and All are weight coefficients; S331, by comparing the influence coefficient Yxxs with the preset limit threshold R, preliminarily judging whether it is necessary to issue a reminder instruction externally, if the influence coefficient Yxxs exceeds the limit threshold R, then issuing a reminder instruction externally, if the influence coefficient Yxxs does not exceed the limit threshold R, then temporarily not issuing a reminder instruction externally; S4. After receiving the reminder instruction, a condition analysis model is created based on the field status data information. After training and testing the condition analysis model, the terrain complexity coefficient Dfxs in different gradient terrain areas is analyzed and obtained. The terrain complexity coefficient Dfxs is associated with the influence coefficient Yxxs in the corresponding gradient terrain area, and after linear normalization, the operation progress index Zczs in the corresponding gradient terrain area is obtained by fitting; S4 specifically includes the following steps: S41, using a convolutional neural network to build an initial model, and training and testing the initial model with field status data information, and using the trained initial model as a state recognition model, respectively obtaining feature information in the state recognition model, and training and testing the state recognition model with the obtained feature information, and combining the received reminder instruction, using the trained state recognition model as a conditional analysis model; S411. Based on the conditional analysis model, the terrain complexity coefficient Dfxs in different gradient terrain areas is analyzed and obtained: In the formula, and are weight coefficients, p represents the number of re-divided areas, j = 1, 2, 3, ..., p, It is expressed as the humidity difference between adjacent areas. It is expressed as the mean of the humidity difference in the entire terrace. It is expressed as the difference in leveling between adjacent areas. It is expressed as the mean of the leveling difference values ​​within the entire terrace; S4 also includes: S42, by associating the terrain complexity coefficient Dfxs with the influence coefficient Yxxs in the corresponding gradient terrain area, and after linear normalization processing, fitting to obtain the operation progress index Zczs in the corresponding gradient terrain area, specifically according to the following formula: In the formula, It is expressed as planting density, Dfxs is expressed as the terrain complexity coefficient, and Yxxs is expressed as the influence coefficient in the corresponding gradient terrain area. It is expressed as the weight coefficient of the influence coefficient Yxxs in the corresponding gradient terrain area, It is expressed as the weight coefficient of the terrain complexity coefficient Dfxs, Planting density The weight coefficient of It is expressed as a correction constant, which is calculated by the terrain complexity coefficient Dfxs, the influence coefficient Yxxs in the corresponding gradient terrain area and the planting density. Calculations can be made to analyze the arrangement of agricultural machinery operation processes in the corresponding gradient terrain area; S5. Pre-set the evaluation threshold Q, and compare and analyze the operation process index Zczs with the evaluation threshold Q to comprehensively analyze the execution planning strategy of the agricultural machinery when facing different gradient terrain areas, and transmit the corresponding execution planning strategy to the driving system of the agricultural machinery to perform the task operation.

2. The method for planning agricultural machinery operations based on unmanned aerial vehicles according to claim 1, characterized in that: S1 specifically includes the following steps: S11, using a drone to hover over the terraced field, and capturing a continuous video stream according to a plurality of sets of sensors carried in the drone, wherein the plurality of sets of sensors include an optical photography sensor, a thermal infrared sensor, and a radar sensor, and the drone is provided with a GPS and an inertial measurement unit; S12, using video processing software to extract individual terraced field geographical image frames from the continuous video stream, and setting the extraction frequency according to the frame rate of the video stream, and then saving the extracted groups of terraced field geographical image frames to cloud storage.

3. The method for planning agricultural machinery operations based on unmanned aerial vehicles according to claim 1, characterized in that: S2 specifically includes the following steps: S21, firstly, geometric correction is performed on a plurality of groups of terraced field geographical image frames, and then wavelet denoising technology is used to decompose the data information in the plurality of groups of terraced field geographical image frames through wavelet transform, and the noise component is removed in the wavelet domain, and at the same time, key feature information in the image frames is extracted through image enhancement technology, and the distribution of crops in the images is identified and analyzed; S22. Sort the continuous terraced field geographical image frames in a corresponding order according to the shooting order of the video stream, calculate the homography matrix between the field geographical image frames through the RANSAC algorithm, and use the homography matrix to perform perspective transformation on the field geographical image frames to splice them into a complete farmland map.

4. The method for planning agricultural machinery operations based on unmanned aerial vehicles according to claim 1, characterized in that: S3 specifically includes the following steps: S31, according to the complete farmland map obtained in step S22, the actual terraced field area is divided into a plurality of groups of preliminary divided areas, each of which has an equal area; S32, the crop distribution set includes relevant terrain data information and field status data information in each preliminary divided area; S321, the relevant terrain data information in each preliminary divided area includes the slope value Pdz and the elevation in each preliminary divided area; S322, field status data information includes humidity and flatness in each interval, and the humidity difference between adjacent areas is calculated through statistical algorithms , the average humidity difference in the entire terrace , the leveling difference between adjacent areas and the average value of the leveling difference within the entire terrace .

5. The method for planning agricultural machinery operations based on unmanned aerial vehicles according to claim 1, characterized in that: S5. By comparing and analyzing the operation process index Zczs with the evaluation threshold Q, the execution planning strategy of agricultural machinery in the face of different gradient terrain areas is comprehensively analyzed. The specific contents are as follows: If the operation process index Zczs ≥ the evaluation threshold Q, it indicates that the operation execution efficiency of the agricultural machinery in the current terrace terrain area is in a normal state. At this time, the operation speed of the agricultural machinery will be increased to speed up the operation process, save time costs, optimize the operation path according to the crop distribution, reduce the turning frequency and repeated operations, and further improve the operation efficiency; If the operation process index Zczs is less than the evaluation threshold Q, it indicates that the operation execution efficiency of the agricultural machinery in the current terrace terrain area is in an abnormal state. At this time, the current operation speed of the agricultural machinery will be maintained to ensure the operation quality and safety.

6. An agricultural machinery operation planning system based on a drone, comprising an agricultural machinery operation planning method based on a drone as described in any one of claims 1 to 5, characterized in that: It includes capture module, processing module, area division module, preliminary analysis module, comprehensive analysis module and planning module; The capture module uses aerial remote sensing technology to enable drones equipped with several sets of sensors to hover over the terraces in advance to capture video streams, and extracts several sets of terrace geographic image frames from the video streams in a time series manner; A processing module, based on a plurality of terraced field geographical image frames and in combination with image processing and analysis technology, performs geometric correction on the plurality of terraced field geographical image frames, and splices the corrected plurality of terraced field geographical image frames into a complete farmland map, so as to identify and collect a crop distribution set in the complete terraced field map; The area division module preliminarily divides the entire terrace area. The crop distribution set includes relevant terrain data information and field status data information in each preliminarily divided area. Based on the relevant terrain data information in each preliminarily divided area, and using the fuzzy clustering algorithm, the entire terrace area is terrain classified to re-divide several groups of preliminarily divided areas. The preliminary analysis module analyzes the influence of different gradient terrains on the operation speed Ys of agricultural machinery to construct the influence coefficient Yxxs. If the influence coefficient Yxxs exceeds the limit threshold R, a reminder instruction is issued externally; The comprehensive analysis module, after receiving the reminder instruction, creates a conditional analysis model based on the field status data information, and after training and testing the conditional analysis model, analyzes and obtains the terrain complexity coefficient Dfxs in different gradient terrain areas, and obtains the operation progress index Zczs in the corresponding gradient terrain area by correlating the terrain complexity coefficient Dfxs with the influence coefficient Yxxs in the corresponding gradient terrain area and performing linear normalization processing; The planning module pre-sets the evaluation threshold Q, and compares and analyzes the operation process index Zczs with the evaluation threshold Q to comprehensively analyze the execution planning strategy of the agricultural machinery when facing different gradient terrain areas, and transmits the corresponding execution planning strategy to the driving system of the agricultural machinery to perform the task operation.

Images