An agricultural data and driving behavior data fusion analysis platform and method

Abstract

This application discloses a fusion analysis platform and method for agricultural data and driving behavior data, relating to the field of fusion analysis technology. It includes generating a torque-limiting virtual throttle command based on the grass entanglement deterioration trend index when there is no clutch jump characteristic and the grass entanglement deterioration trend index is greater than a preset intervention threshold, and generating a height compensation amount based on the implement height data; combining the limit pull-up command, delayed clutch command, torque-limiting virtual throttle command, and height compensation amount into intervention command data; cyclically updating the torque rise rate and the grass entanglement deterioration trend index; and outputting a throttle recovery command smoothly matched with the current throttle opening sequence and a height reset command matched with the implement height data when the latest torque value of the power output shaft torque sequence is less than a preset safety threshold and the grass entanglement deterioration trend index is less than a preset recovery threshold. This application has the effect of improving the efficiency of fusion analysis.

Description

Technical Field

[0001] This application relates to the field of fusion analysis technology, and in particular to a fusion analysis platform and method for agricultural data and driving behavior data. Background Technology

[0002] Agricultural machinery operations are crucial for processes such as straw return to the field, rotary tillage for stubble removal, and field preparation. The power take-off shaft plays a central role in transmitting power from the tractor to the implements, and its torque and speed directly affect implement load, operational continuity, and transmission safety. Meanwhile, driving behavior, as the human control input during agricultural machinery operation, such as throttle, clutch, and implement raising / lowering, directly alters power output, feed rate, and implement posture, significantly impacting operational quality and equipment safety.

[0003] In complex field environments, crop residue density, ground undulation, and changes in driving speed all contribute to the load on the implements, causing torque fluctuations in the power output shaft to become more than just a simple mechanical reflection. Especially in conditions with thick haystacks or wet grass layers, driver actions such as accelerating, clutch engagement, and adjusting implement height often overlap with sudden load increases, creating potential risks of entanglement, stalling, and transmission shock. Therefore, the safety control of agricultural machinery cannot rely solely on single sensor data; it requires attention to the temporal correlation between changes in implement load and driving behavior.

[0004] However, existing agricultural machinery monitoring methods mostly rely on single-point threshold alarms such as low speed and torque overload, lacking synchronous analysis of power output shaft status, driving status, and cab operation data. While some systems can collect torque, speed, or throttle opening data, the time references for these data are inconsistent, making it difficult to distinguish between sudden increases in actual load and torque changes caused by driving disturbances. Furthermore, in situations where the driver rapidly depresses the clutch or accelerates, it is difficult to promptly assess the impact of remaining implement kinetic energy and changes in feed on the risk of grass entanglement, resulting in delayed intervention and abrupt recovery processes, which can easily lead to work interruptions and impacts on transmission components. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this application provides a platform and method for the fusion analysis of agricultural data and driving behavior data.

[0006] Firstly, this application provides a method for fusing and analyzing agricultural data and driving behavior data, comprising the following steps: The torque and speed sequences of the power output shaft are collected from the power output shaft sensor data, the ground speed sequence and implement height data are collected from the driving sensor data, and the throttle opening sequence and clutch travel signal are collected from the cockpit operation sensor data. Aligned data frames are generated according to the sampling timestamp. The torque sequence of the power output shaft within the aligned data frame is differentially processed, and driving disturbances are eliminated by combining the speed sequence and ground speed sequence to obtain the torque rise rate. When the torque increase rate exceeds the preset increase threshold, the active intervention cycle is entered. Within the active intervention cycle, the throttle opening sequence is differentiated to obtain the throttle change rate, and the torque increase rate, throttle change rate, and ground speed sequence are mapped to the grass entanglement deterioration trend index. The clutch jump feature is extracted from the clutch stroke signal; when the clutch jump feature exists and the residual kinetic energy value converted from the speed sequence is greater than the preset kinetic energy threshold, the limit pull-up command and the delayed clutch command are generated based on the machine height data; When there is no clutch jump characteristic and the grass entanglement deterioration trend index is greater than the preset intervention threshold, a torque-limiting virtual throttle command is generated based on the grass entanglement deterioration trend index, and a height compensation amount is generated based on the machine height data. The extreme pull-up command, delayed clutch command, torque-limiting virtual throttle command, and height compensation amount are combined into intervention command data; the torque increase rate and grass entanglement deterioration trend index are updated cyclically. When the latest torque value of the power output shaft torque sequence is less than the preset safety threshold and the grass entanglement deterioration trend index is less than the preset recovery threshold, a throttle recovery command that smoothly matches the current throttle opening sequence and a height reset command that matches the implement height data are output.

[0007] Secondly, this application provides a platform for the fusion analysis of agricultural data and driving behavior data, including: The data acquisition module is used to acquire the torque and speed sequences of the power output shaft from the power output shaft sensor data, the ground speed sequence and implement height data from the driving sensor data, and the throttle opening sequence and clutch stroke signal from the cockpit operation sensor data, and generate aligned data frames according to the sampling timestamp. The differential processing module is used to perform differential processing on the power output shaft torque sequence within the aligned data frame, and combine the speed sequence and ground speed sequence to remove driving disturbances and obtain the torque rise rate. The intervention cycle module is used to enter an active intervention cycle when the torque increase rate is greater than a preset increase threshold. Within the active intervention cycle, the throttle opening sequence is differentiated to obtain the throttle change rate, and the torque increase rate, throttle change rate, and ground speed sequence are mapped to the grass entanglement deterioration trend index. The extraction module is used to extract clutch jump characteristics from the clutch stroke signal; when clutch jump characteristics exist and the residual kinetic energy value converted from the speed sequence is greater than the preset kinetic energy threshold, the limit pull-up command and delayed clutch command are generated based on the machine height data. The judgment module is used to generate a torque-limiting virtual throttle command based on the grass entanglement deterioration trend index when there is no clutch jump characteristic and the grass entanglement deterioration trend index is greater than the preset intervention threshold, and to generate a height compensation amount based on the machine height data. The command output module is used to synthesize the limit pull-up command, delayed clutch command, torque-limiting virtual throttle command, and height compensation amount into intervention command data; it cyclically updates the torque increase rate and the grass entanglement deterioration trend index. When the latest torque value of the power output shaft torque sequence is less than the preset safety threshold and the grass entanglement deterioration trend index is less than the preset recovery threshold, it outputs a throttle recovery command that smoothly matches the current throttle opening sequence and a height reset command that matches the implement height data.

[0008] In summary, this application includes at least one of the following beneficial technical effects: 1. This application provides a method for fusing and analyzing agricultural data and driving behavior data. By generating aligned data frames based on sampling timestamps from the power output shaft torque sequence, speed sequence, ground speed sequence, implement height data, throttle opening sequence, and clutch stroke signal, the implement load status, driving status, and driving behavior during agricultural operations are incorporated into the same time reference for fusion analysis. On this basis, the power output shaft torque sequence is differentially processed, and driving disturbances are eliminated by combining the speed sequence and ground speed sequence, resulting in a torque rise rate that better reflects the actual changes in operating load. Compared to the traditional judgment method that relies solely on a decrease in speed or an over-limit in torque, this application further enters an active intervention cycle after the torque increase rate exceeds a preset threshold. The throttle change rate obtained by differentiating the throttle opening sequence is mapped together with the torque increase rate and the ground speed sequence to form a grass entanglement deterioration trend index. This establishes a calculable correlation between the driver's rapid acceleration, changes in unit feed, and sudden increases in implement load, thereby enabling earlier identification of grass entanglement development trends under conditions such as thick grass piles and wet grass layers. This reduces misjudgments caused by fluctuations in a single sensor value and improves the stability and specificity of risk identification in complex field environments. 2. By extracting clutch jump characteristics from the clutch stroke signal and converting the residual kinetic energy value into the speed sequence, the rotational energy that the cutter shaft can still use when the driver quickly depresses the clutch is included in the intervention judgment; when the residual kinetic energy value is greater than the preset kinetic energy threshold, the limit lifting command and the delayed clutch command are generated based on the machine height data, so that the machine lifting and clutch delay are executed in coordination under the same risk state, reducing the grass entanglement and transmission shock caused by the instantaneous disconnection of power before stalling. For situations where there is no clutch jump characteristic but the grass entanglement deterioration trend index has exceeded the preset intervention threshold, this application generates a torque-limiting virtual throttle command based on the grass entanglement deterioration trend index and generates a height compensation amount based on the implement height data, so that the power output reduction and feed adjustment are coordinated, rather than controlling the engine or suspension mechanism in isolation; then, various commands are combined into intervention command data, and after cyclically updating the torque rise rate and grass entanglement deterioration trend index, a throttle recovery command and a height reset command are output only when the latest torque value is lower than the preset safety threshold and the grass entanglement deterioration trend index is lower than the preset recovery threshold. This makes the intervention entry, execution and exit form a closed loop, which not only improves the timeliness of grass entanglement risk handling, but also improves the smoothness and continuity of operation in the recovery phase. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating the method for fusing and analyzing agricultural data and driving behavior data according to an embodiment of this application.

[0011] Figure 2 This is a schematic diagram of a system for the fusion analysis of agricultural data and driving behavior data according to an embodiment of this application. Detailed Implementation

[0012] The following description, in conjunction with the implementation of the present invention, is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention, and all such modifications and additions should fall within the protection scope of the present invention.

[0013] Example 1: This application is mainly applied to agricultural implements driven by the tractor's power take-off shaft, such as straw returning machines and rotary tillers. In this scenario, when the implement suddenly enters a thick haystack, the torque of the power take-off shaft will increase in a short period of time, and the driver may simultaneously perform actions such as sudden acceleration and clutch depressing. Relying solely on the decrease in power take-off shaft speed to determine blockage often misses the optimal intervention opportunity. This implementation processes power take-off shaft sensor data, driving sensor data, and cockpit operation sensor data on the same time axis. By utilizing the correspondence between torque increase rate, throttle change rate, ground speed sequence, and clutch stroke signal, it distinguishes the risk of hay entanglement after the superposition of sudden load increase and driving behavior, and generates intervention command data when the risk is still under control.

[0014] This application discloses a method for fusing and analyzing agricultural data and driving behavior data.

[0015] Reference Figure 1 A method for fusing and analyzing agricultural data and driving behavior data includes the following steps: The torque and speed sequences of the power output shaft are collected from the power output shaft sensor data, the ground speed sequence and implement height data are collected from the driving sensor data, and the throttle opening sequence and clutch travel signal are collected from the cockpit operation sensor data. Aligned data frames are generated according to the sampling timestamp. Furthermore, generating aligned data frames based on sampling timestamps includes: Read the sampling timestamps of the power output shaft torque sequence, speed sequence, ground speed sequence, implement height data, throttle opening sequence, and clutch stroke signal respectively; Using the sampling timestamp of the power output shaft torque sequence as the reference time axis, coaxial interpolation is performed on the speed sequence, ground speed sequence, and implement height data to generate an agricultural state aligned sequence; The throttle opening sequence and clutch travel signal are compensated for in advance. The amount of compensation is determined by the sampling interval of the cockpit operation sensor data and the response delay of the power output shaft torque sequence. The agricultural state alignment sequence, the compensated throttle opening sequence, and the compensated clutch stroke signal are combined at the same time node to generate an aligned data frame.

[0016] In one specific implementation, a data acquisition process, an alignment processing process, an active intervention loop process, and an instruction generation process can be set within the tractor vehicle controller; The data acquisition process includes collecting torque and speed sequences from the power take-off shaft sensor data. The torque sequence, output by the power take-off shaft torque sensor, is measured in N·m, with a sampling period of 10ms to 20ms. The speed sequence, output by the power take-off shaft speed sensor, is measured in r / min. Ground speed and implement height data are collected from the driving sensor data. The ground speed sequence, measured in m / s, can come from wheel speed sensors, radar ground speed sensors, or existing ground speed data in the vehicle bus. The implement height data, measured in mm, can come from the suspension position sensor. Throttle opening and clutch travel signals are collected from the cockpit operation sensor data. The throttle opening sequence can be expressed as a percentage, ranging from 0% to 100%, and the clutch travel signal can be expressed as a percentage of pedal travel, also ranging from 0% to 100%. After the above data enters the alignment process, an alignment data frame is generated according to the sampling timestamp. An alignment data frame includes at least the power output shaft torque sequence value, speed sequence value, ground speed sequence value, implement height data value, throttle opening sequence value, and clutch stroke signal value at the same time node.

[0017] Specifically, generating aligned data frames based on sampling timestamps can employ a more granular timing processing method. The data acquisition process reads the sampling timestamps of the power output shaft torque sequence, speed sequence, ground speed sequence, implement height data, throttle opening sequence, and clutch stroke signal, respectively. The power output shaft torque sequence is usually most directly related to load changes; therefore, the sampling timestamp of the power output shaft torque sequence is used as the reference time axis. For example, the reference time axis is... to The interval between adjacent time nodes is 10ms. For rotational speed sequences, ground speed sequences, and implement height data, if the sampling timestamp is not completely consistent with the reference time axis, the alignment process uses coaxial interpolation to generate an agricultural state aligned sequence. Taking the ground speed sequence as an example, if the reference time node... Ground velocity sampling time and Between, and The speed at time is , The speed at time is ,but The speed of the ground can be pressed at any time. Obtained, among which, These are the interpolated ground speed sequence values. and These are the values ​​of adjacent original ground velocity sequences.

[0018] The response time of cockpit operation sensor data is typically earlier than the mechanical response of the power output shaft torque sequence. To establish a more accurate correspondence between driving behavior and load changes, the alignment process performs time advance compensation on the throttle opening sequence and clutch travel signal. The compensation amount is determined by the sampling interval of the cockpit operation sensor data and the response delay of the power output shaft torque sequence. In actual calibration, the driver can perform a step throttle action under no-load conditions, recording the time when the throttle opening sequence begins to change and the time when the power output shaft torque sequence shows a corresponding change; the difference between the two is taken as the response delay of the power output shaft torque sequence. If the sampling interval of the cockpit operation sensor data is 10ms and the response delay is 60ms, the compensation amount can be taken as 60ms and converted into 6 sampling points according to the sampling interval. The compensated throttle opening sequence and the compensated clutch travel signal are then combined with the agricultural state alignment sequence at the same time node to form an aligned data frame. In the aligned data frame obtained in this way, the driver operation data and the implement load data are close to each other in causal time, and the subsequent weed entanglement deterioration trend index will not be diluted due to the mechanical response lag.

[0019] The torque sequence of the power output shaft within the aligned data frame is differentially processed, and driving disturbances are eliminated by combining the speed sequence and ground speed sequence to obtain the torque rise rate. Furthermore, the torque sequence of the power output shaft within the aligned data frame is differentially processed, and driving disturbances are removed by combining the speed sequence and ground speed sequence to obtain the torque rise rate, including: Extract the rotational speed sequence from the aligned data frame, and determine the sliding window length based on the current rotational speed value in the rotational speed sequence; The torque sequence of the power output shaft is processed by moving average according to the sliding window length to generate a smooth torque sequence; Calculate the torque difference between adjacent sampling points in the smoothed torque sequence, and divide the torque difference by the time interval between adjacent sampling points to obtain the initial torque change rate sequence; Extract the ground velocity difference between adjacent sampling points in the ground velocity sequence, and divide the ground velocity difference by the time interval between adjacent sampling points to obtain the ground velocity change rate sequence; Extract the correspondence between the ground speed change rate sequence and the initial torque change rate sequence from the safety data segment before the torque rise rate is triggered, and generate the driving disturbance change rate sequence. The initial torque change rate sequence is subtracted point by point from the driving disturbance change rate sequence to obtain the torque increase rate after removing the driving disturbance.

[0020] In one specific implementation, after forming the aligned data frame, the alignment process hands over the power output shaft torque sequence within the aligned data frame to the active intervention loop process for differential processing. Since tractor acceleration, deceleration, and changes in ground adhesion can cause short-term fluctuations in the power output shaft torque, the active intervention loop process does not directly use the difference between adjacent torques as the basis for blockage judgment. Instead, it combines the speed sequence and ground speed sequence to eliminate driving disturbances and obtain the torque increase rate. For example, in the current aligned data frame, the power output shaft torque sequence increases from 480 N·m to 610 N·m in a time interval of 0.1 s, indicating a relatively high initial torque change rate. If the ground speed sequence shows the tractor decreasing from 1.2 m / s to 0.8 m / s at this time, it indicates that the torque change includes a component caused by changes in the unit's driving state. The active intervention loop process will subtract this component to avoid misjudging normal driving disturbances as the risk of being entangled in thick haystacks.

[0021] Specifically, the calculation of the torque rise rate can be completed in the active intervention loop process. The active intervention loop process extracts the rotational speed sequence from the aligned data frame and determines the sliding window length based on the current rotational speed value of the sequence. The higher the rotational speed, the higher the frequency of periodic mechanical vibration in the power output shaft torque sequence, and the sliding window length can be shortened accordingly; at lower rotational speeds, the sliding window length can be appropriately increased to reduce the impact of single-point fluctuations. For example, the current rotational speed value is divided into several intervals. For instance, when the current rotational speed is below 400 r / min, the sliding window length is 7 sampling points; when the current rotational speed is between 400 r / min and 700 r / min, the sliding window length is 5 sampling points; and when the current rotational speed is above 700 r / min, the sliding window length is 3 sampling points. This sliding window length is then used for the moving average processing of the power output shaft torque sequence to generate a smooth torque sequence. If the first... The sliding window length for each sampling point is The values ​​in the smoothed torque sequence You can click within the window Calculation of the average value of the torque sequence of each power output shaft Representing the The smooth torque sequence values ​​at each sampling point.

[0022] After obtaining the smoothed torque sequence, the cyclic process is actively intervened to calculate the torque difference between adjacent sampling points in the smoothed torque sequence. This torque difference is then divided by the time interval between adjacent sampling points to obtain the initial torque change rate sequence. The calculation relationship can be written as follows: ,in, For the first The initial torque change rate sequence values ​​at each sampling point and These are the smoothed torque sequence values ​​for two adjacent sampling points. and These are the corresponding sampling timestamps. During the active intervention loop process, the ground speed difference between adjacent sampling points in the ground speed sequence is extracted, and the ground speed difference is divided by the time interval between adjacent sampling points to obtain the ground speed change rate sequence. The ground speed change rate sequence is used to characterize the impact of unit speed changes on the load.

[0023] To eliminate driving disturbances, the active intervention cycle extracts the correspondence between the ground speed change rate sequence and the initial torque change rate sequence from the safe data segment before the torque escalation rate is triggered. The safe data segment can be selected from the 1 to 3 seconds before entering the active intervention cycle, where the power output shaft torque sequence is below a preset safety threshold and the throttle change rate is below a certain minimum value. Within this safe data segment, if there is an approximately linear correspondence between the ground speed change rate sequence and the initial torque change rate sequence, a proportionality coefficient can be used. This represents the contribution of driving disturbances to the initial torque change rate sequence. This can be obtained by using the least squares relationship between the initial torque change rate sequence and the ground speed change rate sequence in the safety data segment. Subsequently, a driving disturbance change rate sequence is generated, and the first... The number can be 1. Multiply by the first The initial torque change rate sequence is calculated. The active intervention cycle subtracts the driving disturbance change rate sequence from the initial torque change rate sequence point by point, resulting in the torque rise rate after removing driving disturbances. This process ensures that the torque rise rate reflects changes in implement feeding resistance more accurately than changes in the additional load caused by the tractor's own motion.

[0024] When the torque increase rate exceeds the preset increase threshold, the active intervention cycle is entered. Within the active intervention cycle, the throttle opening sequence is differentiated to obtain the throttle change rate, and the torque increase rate, throttle change rate, and ground speed sequence are mapped to the grass entanglement deterioration trend index. Furthermore, the torque increase rate, throttle change rate, and ground speed sequence are mapped to a grass entanglement deterioration trend index, including: The torque increase rate is normalized within a range to generate a load surge component. The throttle change rate is normalized within a range to generate a driving acceleration component; Extract the current ground speed value and the rate of change of ground speed from the ground speed sequence, and combine the current ground speed value and the rate of change of ground speed to generate the feed state component; The fusion weights are assigned according to the numerical values ​​of the load surge component, driving acceleration component, and feeding state component in the current aligned data frame. The load surge component, driving acceleration component, and feeding state component, after being assigned and fused with weights, are weighted and superimposed to generate the grass entanglement deterioration trend index.

[0025] In one specific implementation, when the torque increase rate after eliminating driving disturbances exceeds a preset increase threshold, the active intervention cycle process enters the active intervention loop. The preset increase threshold can be calibrated according to the type of agricultural implement and the working width; for example, it can be set to 60 N·m / s to 150 N·m / s for a straw returning machine. After entering the active intervention loop, the active intervention loop process differentiates the throttle opening sequence to obtain the throttle change rate, which represents how quickly the driver increases the throttle opening per unit time. For example, if the throttle opening sequence increases from 35% to 70% within 0.2s, then the throttle change rate is 175% / s. The active intervention loop process then maps the torque increase rate, throttle change rate, and ground speed sequence to a grass entanglement deterioration trend index. This index can be represented by a dimensionless value from 0 to 1; the larger the value, the stronger the grass entanglement deterioration trend formed by the superposition of sudden increase in power output shaft load, driving acceleration action, and feeding state.

[0026] Specifically, after obtaining the torque increase rate, the mapping processing of the grass entanglement deterioration trend index can be further refined. Active intervention in the cyclic process performs interval normalization on the torque increase rate, generating a load surge component. The normalization interval can be determined by a preset increase threshold and the upper limit of the allowable change rate of the implement's rated torque. For example, when the torque increase rate equals the preset increase threshold, the load surge component is 0; when it reaches the upper limit of the allowable change rate, the load surge component is 1. The intermediate interval is converted linearly and limited to 0 to 1. The throttle change rate is also performed interval normalization to generate a driving acceleration component. When the throttle change rate is below 20% / s, the driving acceleration component can be 0; when it reaches 200% / s, the driving acceleration component can be 1. Intermediate values ​​are obtained proportionally.

[0027] The active intervention cycle extracts the current ground speed value and the rate of change of ground speed from the ground speed sequence, and combines the current ground speed value and the rate of change of ground speed to generate the feeding state component. A higher current ground speed value indicates a greater amount of straw entering the implement per unit time; a negative rate of change of ground speed indicates that the unit is experiencing obstruction, and the likelihood of a thick straw pile causing an increase in load increases. The feeding state component can be obtained by combining the normalized result of the current ground speed value and the normalized result of the rate of change of ground speed. For example, the normalized value corresponding to the current ground speed value is denoted as... The normalized value of the hysteresis corresponding to the rate of change of ground velocity is denoted as Feeding status component Desirable ,in, For the feed state component, This is the normalized result of the current ground speed value. The normalized result of the hindrance is obtained by converting the ground velocity change rate.

[0028] Subsequently, fusion weights are assigned based on the numerical values ​​of the load surge component, driving acceleration component, and feeding status component in the current aligned data frame. If the driving acceleration component is significantly higher than the other components, its corresponding weight is increased; if the load surge component is dominant, its corresponding weight is also increased. The load surge component, driving acceleration component, and feeding status component, after being assigned fusion weights, are then weighted and superimposed to generate a grass entanglement deterioration trend index.

[0029] Among them, the fusion weight can be confirmed frame by frame during the active intervention loop process: first, read the load surge component in the current aligned data frame. Driving acceleration component and feeding status components All three values ​​are dimensionless values ​​between 0 and 1; then, they are multiplied by their respective calibration priority coefficients to obtain the load priority value. Driving priority value and feed priority value ,in , and The calibration coefficients are written according to the type of agricultural implement, such as those in a straw returning machine. Take 1.2, Take 1.0, Take 0.8; then calculate the sum of priority values. ,when When the load surge component exceeds the minimum calculated value, the fusion weight is determined as follows: The fusion weight of the driving acceleration component is determined as follows: The fusion weights of the fed-in state components are confirmed as follows: To ensure the sum of the three fusion weights remains at 1; to prevent a single component's short-term spike from completely dominating the entanglement trend index, each fusion weight can be limited to between 0.15 and 0.60. Weights exceeding the upper limit are redistributed proportionally to the priority values ​​of the other two components, and weights below the lower limit are supplemented to the lower limit before the remaining weights are renormalized; when When the weights are not greater than the minimum calculated value, the fusion weights of the previous aligned data frame are used. After adopting this confirmation method, the weights of the load surge component, driving acceleration component, and feeding state component change with the numerical relationship within the current data frame, while also being constrained by the calibration priority coefficient and upper and lower limits, which can avoid excessive weight jumps.

[0030] The clutch jump feature is extracted from the clutch stroke signal; when the clutch jump feature exists and the residual kinetic energy value converted from the speed sequence is greater than the preset kinetic energy threshold, the limit pull-up command and the delayed clutch command are generated based on the machine height data; When there is no clutch jump characteristic and the grass entanglement deterioration trend index is greater than the preset intervention threshold, a torque-limiting virtual throttle command is generated based on the grass entanglement deterioration trend index, and a height compensation amount is generated based on the machine height data. Furthermore, clutch jump characteristics are extracted from the clutch stroke signal, and residual kinetic energy values ​​are calculated from the speed sequence, including: The clutch travel signal is processed by differential processing of adjacent sampling points to obtain the clutch travel rate sequence; Extract sampling points in the clutch stroke change rate sequence that are greater than a preset jump threshold as clutch jump feature points; Read the speed sequence values ​​of the clutch jump characteristic points at corresponding time nodes to obtain the clutch jump speed values; The clutch jump speed value is converted into the cutter shaft angular velocity data, and the residual kinetic energy value is generated by combining it with the pre-calibrated equivalent rotational inertia of the cutter shaft. The residual kinetic energy value is compared with the preset kinetic energy threshold to form the ultimate lifting determination result.

[0031] Furthermore, based on the machine height data, limit lifting commands and delayed clutch commands are generated, including: When there is a clutch jump characteristic and the residual kinetic energy value is greater than the preset kinetic energy threshold, read the machine height data corresponding to the clutch jump characteristic point as the machine height at the jump moment. The maximum lifting range is calculated based on the machine height at the moment of transition and the preset upper limit of safe height. The ultimate pull-up range is matched with the residual kinetic energy value to generate the ultimate pull-up command; The remaining time of clutch disengagement is calculated based on the clutch stroke change rate sequence, and the remaining time of clutch disengagement and the residual kinetic energy value are jointly mapped to a delay time. Write the delay time into the clutch instruction queue to generate delayed clutch instructions.

[0032] Furthermore, based on the grass entanglement deterioration trend index, a torque-limiting virtual throttle command is generated, including: Extract the throttle opening sequence values ​​at the moment of triggering the active intervention cycle as the throttle reference opening; The throttle reduction amount is generated based on the difference between the grass entanglement deterioration trend index and the preset intervention threshold; Subtract the throttle reduction amount from the reference throttle opening to obtain the target throttle opening; The slope of the target throttle opening is limited to generate a torque-limited virtual throttle command. The torque-limiting virtual throttle command is compared with the current throttle opening sequence. When the difference is less than the preset opening difference threshold, the torque-limiting virtual throttle command is written into the intervention command data.

[0033] Furthermore, a height compensation amount is generated based on the machine height data, including: Extract the machine height data at the moment of active intervention cycle triggering as the machine reference height; The level of feed reduction is determined based on the numerical range of the grass entanglement deterioration trend index. The feed reduction level is mapped to the machine lifting range, and the machine lifting range is added to the machine reference height to generate the target machine height; Subtract the current height value from the target machine height data to obtain the height compensation amount; The height compensation amount is subjected to a single incremental limit process to generate the height compensation amount that is written into the intervention command data.

[0034] In one specific implementation, within the same active intervention cycle, the active intervention cycle process also extracts clutch jump characteristics from the clutch stroke signal. Clutch jump characteristics reflect the driver's action of rapidly depressing the clutch within a short period. If clutch jump characteristics exist, the active intervention cycle process reads the rotational speed sequence values ​​at the same time point and calculates the residual kinetic energy value from the rotational speed sequence. When the residual kinetic energy value is greater than a preset kinetic energy threshold, it indicates that the cutter shaft still has rotational energy that can be used for grass removal and detachment from thick haystacks. At this time, the command generation process generates a limit lift command and a delayed clutch command based on the implement height data. The limit lift command describes the data command for rapidly raising the implement from its current height to near a safe height, and the delayed clutch command describes the time data of the clutch disengagement action relative to the driver's pedal action. If the active intervention cycle process does not extract clutch jump characteristics, and the grass entanglement deterioration trend index is greater than a preset intervention threshold (e.g., a preset intervention threshold of 0.65), the command generation process generates a torque-limiting virtual throttle command based on the grass entanglement deterioration trend index and generates a height compensation amount based on the implement height data. The torque-limiting virtual throttle command does not simply set the throttle opening to zero, but rather reduces the driver's current throttle opening according to the risk level, so that the engine torque demand remains within a relatively smooth range of change; the height compensation amount is used to reduce the instantaneous feed amount of the implement.

[0035] Specifically, for processing the clutch travel signal, the active intervention loop process performs differential processing on adjacent sampling points of the clutch travel signal to obtain a clutch travel change rate sequence. If the clutch travel signal is expressed as a percentage, the clutch travel at the i-th sampling point is... The previous sampling point was The sampling time interval is Then the first of the clutch travel rate change sequence The number is The system actively intervenes in the cyclic process to extract sampling points in the clutch travel rate change sequence that exceed a preset jump threshold, using these points as clutch jump feature points. The preset jump threshold can be set from 150% / s to 300% / s to distinguish between normal, slow stamping and panic stamping.

[0036] After obtaining the clutch jump characteristic point, the loop process is actively intervened to read the speed sequence values ​​at the corresponding time nodes of the clutch jump characteristic point, thus obtaining the clutch jump speed value. If the clutch jump speed value... Tool shaft angular velocity data expressed in r / min Press Calculation, where Pi This is the clutch speed jump value. This refers to the angular velocity data of the tool shaft, in rad / s. This is then combined with the pre-calibrated equivalent moment of inertia of the tool shaft. Generate residual kinetic energy value The calculation formula is ,in, The equivalent rotational inertia of the tool shaft, in units of E is the residual kinetic energy value, in units of The equivalent moment of inertia of the cutter shaft can be obtained from an unloaded acceleration test during the factory calibration stage of the agricultural implement, or it can be calculated based on the mass distribution of the cutter shaft, blades, and connecting parts and then written into the controller calibration area.

[0037] The active intervention cycle compares the residual kinetic energy value with a preset kinetic energy threshold to form a limit lift determination result. The preset kinetic energy threshold can be set according to the implement lifting response time and the cutter shaft's grass removal requirements, for example, 20% to 35% of the no-load kinetic energy at rated speed. When the limit lift determination result is true, the command generation process enters the generation process of the limit lift command and the delayed clutch command.

[0038] In the absence of clutch abrupt change characteristics and when the grass entanglement worsening trend index exceeds the preset intervention threshold, the generation process of the torque-limiting virtual throttle command can be implemented as follows: The command generation process extracts the throttle opening sequence values ​​at the moment the active intervention cycle is triggered, as the throttle reference opening. This throttle reference opening reflects the throttle request already given by the driver when entering the risk state; typically, the instantaneous throttle opening after the active intervention cycle is not directly used to avoid a sudden increase in control requests due to sudden acceleration. The command generation process generates the throttle reduction amount based on the difference between the grass entanglement worsening trend index and the preset intervention threshold. For example, if the grass entanglement worsening trend index is 0.78 and the preset intervention threshold is 0.65, the difference is 0.13. The throttle reduction amount can be calculated at a ratio of 0.5% to 1.5% of the throttle opening for every 0.01 difference. If the throttle reference opening is 60% and the throttle reduction amount is 12%, then the target throttle opening is 48%.

[0039] After the target throttle opening is generated, the command generation process applies slope limiting processing to the target throttle opening to generate a torque-limited virtual throttle command. The slope limiting processing can limit the opening change within each sampling period to between 2% and 5%, preventing the torque-limited virtual throttle command from suddenly dropping from a high opening and causing a sudden change in engine torque request.

[0040] Subsequently, the command generation process compares the torque-limiting virtual throttle command with the current throttle opening sequence. When the difference is less than a preset opening difference threshold, the torque-limiting virtual throttle command is written into the intervention command data. The preset opening difference threshold can be 30% to 50% to exclude situations where the driver's throttle opening is completely abnormal, avoiding an unstable recovery process caused by excessive deviation between the torque-limiting virtual throttle command and the current throttle opening sequence. After the intervention command data is written, the engine torque request participates in subsequent control according to the torque-limiting virtual throttle command.

[0041] The generation of height compensation is coordinated with the virtual throttle command for torque limiting. The command generation process extracts the implement height data at the moment of triggering the active intervention cycle as the implement reference height. The implement reference height can be the suspension position corresponding to the normal working depth of the implement, for example, 150mm. The command generation process determines the feeding reduction level based on the numerical range of the weed entanglement deterioration trend index. When the weed entanglement deterioration trend index is between 0.65 and 0.75, it can be determined as a level 1 feeding reduction level; when it is between 0.75 and 0.85, it can be determined as a level 2 feeding reduction level; when it exceeds 0.85, it can be determined as a level 3 feeding reduction level. Then, the feeding reduction level is mapped to the implement lifting range, for example, level 1 corresponds to 20mm to 30mm, level 2 corresponds to 30mm to 50mm, and level 3 corresponds to 50mm to 80mm. The command generation process adds the implement lifting range to the implement reference height to generate the target implement height, and then subtracts the target implement height from the current height value in the implement height data to obtain the height compensation amount. If the reference height of the implement is 150mm, the current height is 155mm, and the implement lifting range corresponding to the secondary feed reduction level is 40mm, then the target implement height is 190mm, and the height compensation is 35mm. To prevent excessive suspension movement, the instruction generation process performs single-increment limit processing on the height compensation, for example, increasing it by a maximum of 10mm per sampling cycle before generating the height compensation amount to be written into the intervention instruction data.

[0042] When clutch engagement jump characteristics exist and the residual kinetic energy value exceeds the preset kinetic energy threshold, the generation of the ultimate lifting command and delayed clutch command emphasizes timing. The command generation process reads the implement height data corresponding to the clutch engagement jump characteristic point as the implement height at the jump moment. The implement height at the jump moment is compared with the preset safe height upper limit to calculate the ultimate lifting amplitude. The preset safe height upper limit can be calibrated based on the maximum allowable lifting position of the implement, the universal joint transmission angle, and the safe working distance. For example, if the normal working height is 150mm, the preset safe height upper limit is 240mm, and the implement height at the jump moment is 160mm, then the ultimate lifting amplitude is 80mm. The command generation process matches the ultimate lifting amplitude with the residual kinetic energy value to generate the ultimate lifting command. If the residual kinetic energy value is high, the ultimate lifting command can use a higher lifting speed and a larger single-cycle height increment; if the residual kinetic energy value is only slightly higher than the preset kinetic energy threshold, the ultimate lifting command can prioritize reaching an intermediate height that reduces the feed amount and limit the single-cycle height increment to prevent the lifting action from consuming too much time.

[0043] The delayed clutch command is determined by the clutch travel rate of change sequence and the residual kinetic energy value. The active intervention cycle calculates the remaining clutch disengagement time based on the clutch travel rate of change sequence. If the clutch travel signal is 90% at complete disengagement and 35% at the clutch jump characteristic point, with a clutch travel rate of change of 250% / s, then the remaining clutch disengagement time is approximately (90%-35%) / 250% / s, or 0.22s. The command generation process maps the remaining clutch disengagement time and the residual kinetic energy value together to a delay time. The higher the residual kinetic energy value, the longer the delay time can be within a safe range to allow the cutter shaft to discharge any trapped forage; the shorter the remaining clutch disengagement time, the shorter the delay time should not be, to prevent excessive deviation between the driver's pedal action and the control action. The delay time can be calibrated within the range of 0.1s to 0.6s, and then written into the clutch command queue to generate the delayed clutch command. The clutch instruction queue stores the execution timestamp, delay time, and corresponding clutch jump feature point number of clutch-related data instructions, which facilitates tracing the intervention action after the active intervention cycle ends.

[0044] The extreme pull-up command, delayed clutch command, torque-limiting virtual throttle command, and height compensation amount are combined into intervention command data; the torque increase rate and grass entanglement deterioration trend index are updated cyclically. When the latest torque value of the power output shaft torque sequence is less than the preset safety threshold and the grass entanglement deterioration trend index is less than the preset recovery threshold, a throttle recovery command that smoothly matches the current throttle opening sequence and a height reset command that matches the implement height data are output.

[0045] Furthermore, the cyclical update of torque increase rate and grass entanglement deterioration trend index includes: Within the active intervention cycle, new power output shaft torque sequence, new speed sequence, new ground speed sequence, and new throttle opening sequence are read at a fixed sampling period; The new power output shaft torque sequence is differentially processed to obtain the updated torque rise rate; The updated torque increase rate, the throttle change rate corresponding to the new throttle opening sequence, and the new ground speed sequence are mapped to obtain the updated grass entanglement deterioration trend index. The updated torque rise rate is compared with the torque rise rate of the previous sampling period to generate a load fall-off flag. The load drop marker, the latest torque value, and the updated grass entanglement deterioration trend index are all written into the cycle status data.

[0046] Furthermore, the output includes a throttle recovery command that smoothly matches the current throttle opening sequence and a height reset command that matches the implement height data, including: When the latest torque value of the power output shaft torque sequence is less than the preset safety threshold and the grass entanglement deterioration trend index is less than the preset recovery threshold, read the current throttle opening sequence value and the torque-limiting virtual throttle command value. The difference between the current throttle opening sequence value and the torque-limiting virtual throttle command value is split into multiple opening increments; Multiple opening increments are sequentially superimposed according to the sampling period to generate a throttle recovery command; Read the machine height data at the moment of active intervention cycle triggering and the current machine height data, and calculate the height reset difference; The height reset difference is split into multiple height callback values, and these multiple height callback values ​​are sequentially superimposed according to the sampling period to generate a height reset command.

[0047] In one specific implementation, the instruction generation process synthesizes the limit pull-up instruction, delayed clutch instruction, torque-limiting virtual throttle instruction, and height compensation amount into intervention instruction data. During actual execution, it is not required that all four types of instructions in the same intervention instruction data be valid values. For example, if there is a clutch jump characteristic and the residual kinetic energy value meets the conditions, the limit pull-up instruction and delayed clutch instruction have valid values, and the torque-limiting virtual throttle instruction can retain the value from the previous frame; if there is no clutch jump characteristic and the grass entanglement deterioration trend index exceeds the preset intervention threshold, the torque-limiting virtual throttle instruction and height compensation amount have valid values. The active intervention cycle process updates the torque rise rate and grass entanglement deterioration trend index cyclically at fixed intervals and reads the latest torque value of the power output shaft torque sequence. When the aforementioned latest torque value is less than the preset safety threshold and the grass entanglement deterioration trend index is less than the preset recovery threshold, the instruction generation process outputs a throttle recovery instruction that smoothly matches the current throttle opening sequence and a height reset instruction that matches the implement height data. The preset safety threshold can be calibrated at 70% to 85% of the rated torque of the agricultural implement, and the preset recovery threshold can be set at 0.25 to 0.4. The two together constrain the timing of exiting the active intervention cycle, so that the recovery process does not occur prematurely due to fluctuations in a single data point.

[0048] Specifically, the update process of the active intervention cycle runs throughout the entire intervention phase. Within the active intervention cycle, the data acquisition process reads the new power output shaft torque sequence, the new speed sequence, the new ground speed sequence, and the new throttle opening sequence at a fixed sampling period. The fixed sampling period can be consistent with the sampling period of the aligned data frame, for example, 10ms, or 20ms when the controller's computing power is limited. The new power output shaft torque sequence enters the aforementioned differential processing process to obtain the updated torque rise rate; the derivative of the new throttle opening sequence is used to obtain the corresponding throttle change rate; the updated torque rise rate, the throttle change rate corresponding to the new throttle opening sequence, and the new ground speed sequence are then mapped to obtain the updated grass entanglement deterioration trend index. The active intervention cycle process compares the updated torque rise rate with the torque rise rate of the previous sampling period to generate a load fall flag. If the updated torque rise rate is less than the torque rise rate of the previous sampling period, and the difference exceeds the minimum resolution set in the controller, for example, 5 N·m / s, then the load fall flag can be written as 1; otherwise, it is written as 0. The load reduction marker, the latest torque value, and the updated grass entanglement deterioration trend index are all written into the cycle status data. The cycle status data can be stored in a circular buffer for the most recent 0.5s to 2s, which is used to determine whether the load change has entered a stable recovery phase.

[0049] When the latest torque value of the power output shaft torque sequence is less than a preset safety threshold and the grass entanglement deterioration trend index is less than a preset recovery threshold, the active intervention cycle process does not immediately restore the control data to the driver's current input. Instead, the command generation process generates smooth recovery data. The command generation process reads the current throttle opening sequence value and the torque-limiting virtual throttle command value, and breaks down the difference between the two into multiple opening increments. If the current throttle opening sequence value is 55% and the torque-limiting virtual throttle command value is 42%, the difference is 13%, which can be broken down into 7 to 13 segments with an opening increment of 1% to 2% per sampling period. Then, multiple opening increments are sequentially superimposed according to the sampling period to generate a throttle recovery command. Each frame of the throttle recovery command maintains a limited difference from the previous frame, thus smoothly transitioning the engine torque request to the level matched by the current throttle opening sequence.

[0050] The height reset command employs a similar segmented callback method. The command generation process reads the machine height data at the moment the active intervention cycle is triggered and the current machine height data, and calculates the height reset difference. If the machine height data at the moment the active intervention cycle is triggered is 150mm and the current machine height data is 205mm, then the height reset difference is 55mm. This height reset difference is broken down into multiple height callback amounts, for example, 5mm to 10mm per sampling cycle, and these multiple height callback amounts are sequentially superimposed according to the sampling cycle to generate the height reset command. The height reset command causes the machine to gradually return from the raised state to the working height, avoiding a new feeding impact caused by a one-time drop. After both the throttle recovery command and the height reset command are completed, the active intervention cycle process clears the temporary markers in the current cycle status data, and retains the alignment data frame, the peak value of the grass entanglement deterioration trend index, the limit pull command, the delayed clutch command, the torque-limiting virtual throttle command, and the height compensation amount from the current intervention process for subsequent calibration and fault review.

[0051] Example 2: This application also discloses a platform for the fusion analysis of agricultural data and driving behavior data.

[0052] Reference Figure 2 A platform for the fusion and analysis of agricultural data and driving behavior data, comprising: The data acquisition module is used to acquire the torque and speed sequences of the power output shaft from the power output shaft sensor data, the ground speed sequence and implement height data from the driving sensor data, and the throttle opening sequence and clutch stroke signal from the cockpit operation sensor data, and generate aligned data frames according to the sampling timestamp. The differential processing module is used to perform differential processing on the power output shaft torque sequence within the aligned data frame, and combine the speed sequence and ground speed sequence to remove driving disturbances and obtain the torque rise rate. The intervention cycle module is used to enter an active intervention cycle when the torque increase rate is greater than a preset increase threshold. Within the active intervention cycle, the throttle opening sequence is differentiated to obtain the throttle change rate, and the torque increase rate, throttle change rate, and ground speed sequence are mapped to the grass entanglement deterioration trend index. The extraction module is used to extract clutch jump characteristics from the clutch stroke signal; when clutch jump characteristics exist and the residual kinetic energy value converted from the speed sequence is greater than the preset kinetic energy threshold, the limit pull-up command and delayed clutch command are generated based on the machine height data. The judgment module is used to generate a torque-limiting virtual throttle command based on the grass entanglement deterioration trend index when there is no clutch jump characteristic and the grass entanglement deterioration trend index is greater than the preset intervention threshold, and to generate a height compensation amount based on the machine height data. The command output module is used to synthesize the limit pull-up command, delayed clutch command, torque-limiting virtual throttle command, and height compensation amount into intervention command data; it cyclically updates the torque increase rate and the grass entanglement deterioration trend index. When the latest torque value of the power output shaft torque sequence is less than the preset safety threshold and the grass entanglement deterioration trend index is less than the preset recovery threshold, it outputs a throttle recovery command that smoothly matches the current throttle opening sequence and a height reset command that matches the implement height data.

[0053] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention, they should all fall within the protection scope of the present invention.

[0054] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0055] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A method for fusing and analyzing agricultural data and driving behavior data, characterized in that, Includes the following steps: The torque and speed sequences of the power output shaft are collected from the power output shaft sensor data, the ground speed sequence and implement height data are collected from the driving sensor data, and the throttle opening sequence and clutch travel signal are collected from the cockpit operation sensor data. Aligned data frames are generated according to the sampling timestamp. The torque sequence of the power output shaft within the aligned data frame is differentially processed, and driving disturbances are eliminated by combining the speed sequence and ground speed sequence to obtain the torque rise rate. When the torque increase rate exceeds the preset increase threshold, an active intervention cycle is initiated. Within the active intervention cycle, the throttle opening sequence is differentiated to obtain the throttle change rate, and the torque increase rate, throttle change rate and ground speed sequence are mapped to the grass entanglement deterioration trend index. Extract clutch jump features from clutch travel signals; When there is a clutch abrupt change characteristic and the residual kinetic energy value calculated from the speed sequence is greater than the preset kinetic energy threshold, the limit pull-up command and the delayed clutch command are generated based on the machine height data. When there is no clutch jump characteristic and the grass entanglement deterioration trend index is greater than the preset intervention threshold, a torque-limiting virtual throttle command is generated based on the grass entanglement deterioration trend index, and a height compensation amount is generated based on the machine height data. The extreme pull-up command, delayed clutch command, torque-limiting virtual throttle command, and height compensation amount are combined into intervention command data; the torque increase rate and grass entanglement deterioration trend index are updated cyclically. When the latest torque value of the power output shaft torque sequence is less than the preset safety threshold and the grass entanglement deterioration trend index is less than the preset recovery threshold, a throttle recovery command that smoothly matches the current throttle opening sequence and a height reset command that matches the implement height data are output.

2. The method for fusing and analyzing agricultural data and driving behavior data according to claim 1, characterized in that, Generate aligned data frames based on the sampling timestamp, including: Read the sampling timestamps of the power output shaft torque sequence, speed sequence, ground speed sequence, implement height data, throttle opening sequence, and clutch stroke signal respectively; Using the sampling timestamp of the power output shaft torque sequence as the reference time axis, coaxial interpolation is performed on the speed sequence, ground speed sequence, and implement height data to generate an agricultural state aligned sequence; The throttle opening sequence and clutch travel signal are compensated for in advance. The amount of compensation is determined by the sampling interval of the cockpit operation sensor data and the response delay of the power output shaft torque sequence. The agricultural state alignment sequence, the compensated throttle opening sequence, and the compensated clutch stroke signal are combined at the same time node to generate an aligned data frame.

3. The method for fusing and analyzing agricultural data and driving behavior data according to claim 1, characterized in that, The torque sequence of the power output shaft within the aligned data frame is differentially processed, and driving disturbances are removed by combining the speed sequence and ground speed sequence to obtain the torque rise rate, including: Extract the rotational speed sequence from the aligned data frame, and determine the sliding window length based on the current rotational speed value in the rotational speed sequence; The torque sequence of the power output shaft is processed by moving average according to the sliding window length to generate a smooth torque sequence; Calculate the torque difference between adjacent sampling points in the smoothed torque sequence, and divide the torque difference by the time interval between adjacent sampling points to obtain the initial torque change rate sequence; Extract the ground velocity difference between adjacent sampling points in the ground velocity sequence, and divide the ground velocity difference by the time interval between adjacent sampling points to obtain the ground velocity change rate sequence; Extract the correspondence between the ground speed change rate sequence and the initial torque change rate sequence from the safety data segment before the torque rise rate is triggered, and generate the driving disturbance change rate sequence. The initial torque change rate sequence is subtracted point by point from the driving disturbance change rate sequence to obtain the torque increase rate after removing the driving disturbance.

4. The method for fusing and analyzing agricultural data and driving behavior data according to claim 3, characterized in that, The torque increase rate, throttle change rate, and ground speed sequence are mapped to a grass entanglement deterioration trend index, including: The torque increase rate is normalized within a range to generate a load surge component. The throttle change rate is normalized within a range to generate a driving acceleration component; Extract the current ground speed value and the rate of change of ground speed from the ground speed sequence, and combine the current ground speed value and the rate of change of ground speed to generate the feed state component; The fusion weights are assigned according to the numerical values ​​of the load surge component, driving acceleration component, and feeding state component in the current aligned data frame. The load surge component, driving acceleration component, and feeding state component, after being assigned and fused with weights, are weighted and superimposed to generate the grass entanglement deterioration trend index.

5. The method for fusing and analyzing agricultural data and driving behavior data according to claim 1, characterized in that, The clutch abrupt change characteristics are extracted from the clutch stroke signal, and the residual kinetic energy value is calculated from the speed sequence, including: The clutch travel signal is processed by differential processing of adjacent sampling points to obtain the clutch travel rate sequence; Extract sampling points in the clutch stroke change rate sequence that are greater than a preset jump threshold as clutch jump feature points; Read the speed sequence values ​​of the clutch jump characteristic points at corresponding time nodes to obtain the clutch jump speed values; The clutch jump speed value is converted into the cutter shaft angular velocity data, and the residual kinetic energy value is generated by combining it with the pre-calibrated equivalent rotational inertia of the cutter shaft. The residual kinetic energy value is compared with the preset kinetic energy threshold to form the ultimate pull-up determination result.

6. The method for fusing and analyzing agricultural data and driving behavior data according to claim 5, characterized in that, Based on the grass entanglement deterioration trend index, a torque-limiting virtual throttle command is generated, including: Extract the throttle opening sequence values ​​at the moment of triggering the active intervention cycle as the throttle reference opening; The throttle reduction amount is generated based on the difference between the grass entanglement deterioration trend index and the preset intervention threshold; Subtract the throttle reduction amount from the reference throttle opening to obtain the target throttle opening; The slope of the target throttle opening is limited to generate a torque-limited virtual throttle command. The torque-limiting virtual throttle command is compared with the current throttle opening sequence. When the difference is less than the preset opening difference threshold, the torque-limiting virtual throttle command is written into the intervention command data.

7. The method for fusing and analyzing agricultural data and driving behavior data according to claim 1, characterized in that, Height compensation is generated based on the equipment height data, including: Extract the machine height data at the moment of active intervention cycle triggering as the machine reference height; The level of feed reduction is determined based on the numerical range of the grass entanglement deterioration trend index. The feed reduction level is mapped to the machine lifting range, and the machine lifting range is added to the machine reference height to generate the target machine height; Subtract the current height value from the target machine height data to obtain the height compensation amount; The height compensation amount is subjected to a single incremental limit process to generate the height compensation amount that is written into the intervention command data.

8. The method for fusing and analyzing agricultural data and driving behavior data according to claim 1, characterized in that, Based on the machine height data, generate limit lifting commands and delayed clutch commands, including: When there is a clutch jump characteristic and the residual kinetic energy value is greater than the preset kinetic energy threshold, read the machine height data corresponding to the clutch jump characteristic point as the machine height at the jump moment. The maximum lifting range is calculated based on the machine height at the moment of transition and the preset upper limit of safe height. The ultimate pull-up range is matched with the residual kinetic energy value to generate the ultimate pull-up command; The remaining time of clutch disengagement is calculated based on the clutch stroke change rate sequence, and the remaining time of clutch disengagement and the residual kinetic energy value are jointly mapped to a delay time. Write the delay time into the clutch instruction queue to generate delayed clutch instructions.

9. The method for fusing and analyzing agricultural data and driving behavior data according to claim 1, characterized in that, Output a throttle recovery command that smoothly matches the current throttle opening sequence and a height reset command that matches the implement height data, including: When the latest torque value of the power output shaft torque sequence is less than the preset safety threshold and the grass entanglement deterioration trend index is less than the preset recovery threshold, read the current throttle opening sequence value and the torque-limiting virtual throttle command value. The difference between the current throttle opening sequence value and the torque-limiting virtual throttle command value is split into multiple opening increments; Multiple opening increments are sequentially superimposed according to the sampling period to generate a throttle recovery command; Read the machine height data at the moment of active intervention cycle triggering and the current machine height data, and calculate the height reset difference; The height reset difference is split into multiple height callback values, and these multiple height callback values ​​are sequentially superimposed according to the sampling period to generate a height reset command.

10. A platform for the fusion analysis of agricultural data and driving behavior data, applied to the fusion analysis method of agricultural data and driving behavior data as described in any one of claims 1-9, characterized in that, include: The data acquisition module is used to acquire the torque and speed sequences of the power output shaft from the power output shaft sensor data, the ground speed sequence and implement height data from the driving sensor data, and the throttle opening sequence and clutch stroke signal from the cockpit operation sensor data, and generate aligned data frames according to the sampling timestamp. The differential processing module is used to perform differential processing on the power output shaft torque sequence within the aligned data frame, and combine the speed sequence and ground speed sequence to remove driving disturbances and obtain the torque rise rate. The intervention cycle module is used to enter an active intervention cycle when the torque increase rate is greater than a preset increase threshold. Within the active intervention cycle, the throttle opening sequence is differentiated to obtain the throttle change rate, and the torque increase rate, throttle change rate and ground speed sequence are mapped to the grass entanglement deterioration trend index. The extraction module is used to extract clutch jump features from the clutch stroke signal; When there is a clutch abrupt change characteristic and the residual kinetic energy value calculated from the speed sequence is greater than the preset kinetic energy threshold, the limit pull-up command and the delayed clutch command are generated based on the machine height data. The judgment module is used to generate a torque-limiting virtual throttle command based on the grass entanglement deterioration trend index when there is no clutch jump characteristic and the grass entanglement deterioration trend index is greater than the preset intervention threshold, and to generate a height compensation amount based on the machine height data. The command output module is used to synthesize the limit pull-up command, delayed clutch command, torque-limiting virtual throttle command, and height compensation amount into intervention command data; it cyclically updates the torque increase rate and the grass entanglement deterioration trend index. When the latest torque value of the power output shaft torque sequence is less than the preset safety threshold and the grass entanglement deterioration trend index is less than the preset recovery threshold, it outputs a throttle recovery command that smoothly matches the current throttle opening sequence and a height reset command that matches the implement height data.

Images