Data processing method and system for rice seedling transplanting energy efficiency scheduling based on dynamic load zoning

Abstract

The application discloses a rice transplanter energy efficiency scheduling data processing method and system based on dynamic load partitioning, relates to the technical field of data processing, and comprises the following steps: acquiring engine output speed data, hydraulic system working pressure data, operation travel speed data and seedling needle action beat data in rice transplanter operation, screening effective stable reference windows, generating pressure sorting position sequences, engine response difference sequences, speed deviation sequences and seedling needle frequency sequences; generating a credible load sequence according to the first offset relationship between the pressure sorting position and the seedling needle frequency, engine response and speed deviation, and then generating a processing cycle state code, a dynamic load boundary and a current load partition, and outputting an engine target speed section and a hydraulic variable pump target displacement section; and correcting subsequent control actions according to the residual errors before and after execution. The scheme is used for reducing load misjudgment and partition oscillation caused by fixed load boundaries and reducing invalid power regulation.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a data processing method and system for rice transplanting energy efficiency scheduling based on dynamic load zoning. Background Technology

[0002] When rice transplanters operate continuously in the field, the engine, hydraulic variable pump, transplanting mechanism, and traveling mechanism need to be matched to balance transplanting rhythm, operating speed, and fuel consumption. In actual operation, soil moisture content, mud depth, seedling supply status, seed needle grasping rhythm, and traveling speed all cause continuous changes in the transplanter's load. Some transplanters' energy efficiency scheduling relies on hydraulic system working pressure, engine output speed, or fixed gear boundaries. When the hydraulic system working pressure rises briefly, the power output is easily increased directly; when the hydraulic system working pressure drops briefly, the power output is easily reduced. The above approach does not adequately distinguish the sequential relationship between changes in hydraulic system working pressure, seed needle grasping frequency, engine response, and traveling speed.

[0003] For example, when a rice transplanter enters a slightly deeper muddy area, the hydraulic system's working pressure may initially rise, followed by a decrease in the seedling needle grasping frequency, and only then will the engine output speed respond. This situation typically corresponds to load changes caused by increased operating resistance. Conversely, if the seedling needle grasping frequency decreases first without a synchronous rise in the hydraulic system's working pressure, it is more likely due to discontinuous seedling supply or abnormal seedling needle movement. If the hydraulic system's working pressure and the seedling needle grasping frequency shift synchronously within adjacent processing cycles and recover quickly, it is more likely due to synchronous interference or short-term impact. If energy efficiency scheduling is still triggered according to fixed thresholds or single-point data, it will lead to frequent changes in the engine's target speed and the hydraulic variable pump's target displacement, resulting in boundary oscillations, ineffective regulation, and increased fuel consumption.

[0004] When multiple rice transplanters operate in tandem, different units experience varying levels of mud resistance and seedling supply. Simply relying on a uniform operating speed or fixed path allocation makes load balancing difficult, easily leading to situations where some units operate under sustained high loads while others operate under prolonged low loads. Therefore, a data processing method is needed that can establish dynamic boundaries based on stable historical data and incorporate load zoning based on pressure ranking position, seedling needle frequency, engine response, and operating speed. Summary of the Invention

[0005] In response to the technical problems described in the background art, the present invention provides a method and system for processing rice transplanting energy efficiency scheduling data based on dynamic load zoning, which is used to reduce the probability of load misjudgment, zoning oscillation and ineffective amplification of control actions under fixed load boundaries.

[0006] A data processing method for rice transplanting energy efficiency scheduling based on dynamic load zoning includes: acquiring engine output speed data, hydraulic system working pressure data, operating speed data, and seedling needle movement rhythm data during rice transplanter operation; filtering effective stable reference windows from historical sampling data; inserting the current hydraulic system working pressure data into the pressure sorting sequence of the effective stable reference window to generate a pressure sorting position sequence, and generating an engine response difference sequence, a speed deviation sequence, and a seedling needle frequency sequence; comparing the starting point of the pressure sorting position entering the upper deviation zone with the starting point of the seedling needle frequency entering the lower deviation zone, combining the engine response difference sequence and speed deviation sequence within the response waiting period to generate a reliable load sequence; converting the reliable load sequence into a symbolic segment and generating a processing cycle status code; generating a dynamic load boundary based on the effective stable reference window; statistically analyzing the boundary crossing return data of the load zone candidates relative to the dynamic load boundary; determining the current load zone based on the processing cycle status code and the boundary crossing return data; and determining the engine target speed segment and the hydraulic variable pump target displacement segment based on the current load zone.

[0007] Optionally, a valid stable reference window is selected from the historical sampling data, including: dividing the historical sampling data into candidate historical windows according to the same window length, forming an initial inspection interval using the manufacturer's allowable range and the calibration pressure sequence frozen in the previous operation phase; sequentially excluding candidate historical windows whose number of missing data reaches the integrity screening exclusion limit, whose number of consecutive out-of-bounds reaches the boundary screening exclusion limit, whose number of direction reversals reaches the stability screening exclusion limit, or whose consistency requirements are not met when performing readback consistency marking, and determining the remaining candidate historical windows as valid stable reference windows; wherein, the distribution statistics merge the data of each stable sub-window, and the time series statistics are only performed within each stable sub-window.

[0008] Optionally, generating a pressure sorting position sequence, an engine response difference sequence, a speed deviation sequence, and a seedling needle frequency sequence includes: arranging the pressure data within the effective stable reference window from low to high; temporarily inserting the current pressure and obtaining the pressure sorting position according to the midpoint rank of the parallel data, without writing the current pressure back into the pressure sorting sequence; generating an engine response difference based on the difference between the current actual engine speed and the engine target speed recorded in the previous processing cycle; generating speed deviation data based on the deviation between the current operating speed and the target operating speed; forming a single-row transplanting frequency based on the effective hole formation times of adjacent transplanting rows in the same transplanting row, and generating a beat confirmation mark when a complete action interval is not formed.

[0009] Optionally, the starting point of the pressure sorting position entering the upper bias section is compared with the starting point of the seedling needle frequency entering the lower bias section. A reliable load sequence is generated by combining the engine response difference sequence and speed deviation sequence within the response waiting period. This includes: performing unified clock alignment and link delay compensation on the two starting points, and determining the order only when the absolute value of the time difference between the two starting points exceeds the comprehensive time delay uncertainty; generating a beat pressure verification result based on the order of the starting points, where the pressure moves upward first and the seedling needle moves downward later, or where there is an abnormal observation or synchronous short-term offset on the seedling needle side; generating a power coverage verification result based on the engine response difference sequence and the speed residual change in the same target speed segment; and writing the corresponding period into the reliable load data, load observation data, or execution lag observation data by combining the speed deviation data.

[0010] Optionally, converting the trusted load sequence into symbol segments and generating processing cycle status codes includes: recording the pressure sorting position changes of adjacent processing cycles as upward, downward, hold, or alternating directions, merging consecutive records with the same direction to form symbol segments; generating a direction status code based on the direction and continuous length of the symbol segments; generating an amplitude status code based on the relative position of the current pressure sorting position extreme value interval and the reference pressure sorting position extreme value interval; generating a recovery status code based on the out-of-bounds duration relative to the reference stable interval, the number of sampling points required to return to the stable interval, and the stable duration after returning; generating an execution status code based on the residual changes before and after execution; and forming processing cycle status codes in a fixed order of direction, amplitude, recovery, and execution.

[0011] Optionally, dynamic load boundaries are generated based on an effective stable reference window, including: generating candidate basic sorted position boundaries based on historical effective operation data with load status, fuel consumption per unit operating area, seedling frequency, and operation quality markers; verifying candidate combinations that are sequentially increasing, have the minimum interval between adjacent boundaries, and have the minimum number of samples in each partition, to obtain three basic sorted position boundaries and initialize the dynamic load boundaries.

[0012] Optionally, after obtaining the three basic sorting position boundaries and initializing the dynamic load boundaries, the process includes: generating a theoretical response phase sequence based on the engine target speed change records and historical effective response records that have occurred before the current processing cycle; reading the most recent target strategy record and corresponding feedback when the instruction age reaches the response waiting period, and extracting the actual response phase sequence; comparing the two response phase sequences item by item, and generating boundary back-calibration results based on phase integrity, phase sequence, and post-execution residual changes, such as maintaining the boundary, extending the waiting period, freezing the boundary, and limited outward or limited inward movement; freezing the boundary first when the previous level load partition is continuously triggered and the residual does not improve, checking the execution link and pressure protection status, and only moving the boundary outward in a limited manner after continuous confirmation of false positives in the partition.

[0013] Optionally, the boundary crossing return data of the candidate load zoning relative to the dynamic load boundary is statistically analyzed, and the current load zoning is determined based on the processing cycle status code and the boundary crossing return data. This includes: generating basic load zoning candidates based on the dynamic load boundary and pressure sorting position, and correcting the candidates sequentially based on the cycle pressure verification results, power coverage verification results, and speed deviation data; recording the first crossing position, the duration of the boundary crossing, the number of sampling points required to return to the stable interval, and the duration of the stable interval after the return; maintaining the load zoning of the previous processing cycle when there is a short-term boundary crossing and a rapid return, and determining the current load zoning based on the corrected candidates when there is a continuous boundary crossing, a slow return, or insufficient recovery; the current load zoning is a mutually exclusive low load zoning, medium-low load zoning, medium-high load zoning, or high load zoning.

[0014] Optionally, the target engine speed range and the target hydraulic variable pump displacement range are determined based on the current load zone, including: querying the energy efficiency mapping table to obtain candidate engine target speed ranges and candidate hydraulic variable pump target displacement ranges from the same feasible pairing record; verifying that the estimated flow rate does not exceed the pump's allowable continuous flow rate, and verifying that the sum of the pump shaft input power and other load power not included in the pump shaft input power does not exceed the engine's available power, and that the plugged hydraulic power supplied by the hydraulic variable pump is not counted repeatedly; verifying the complete transition range covering the current speed to the target speed, during which the engine speed is first increased while maintaining or decreasing the pump displacement, and the pump displacement is then ramped up after the engine speed enters the target speed range.

[0015] A data processing system for rice transplanting energy efficiency scheduling based on dynamic load zoning is also provided. This system implements a method for rice transplanting energy efficiency scheduling data processing based on dynamic load zoning. The system includes: a data acquisition module for acquiring engine output speed data, hydraulic system working pressure data, operating speed data, and seedling needle movement rhythm data; a reference window filtering module for filtering effective and stable reference windows; a sequence generation module for generating pressure sorting position sequences, engine response difference sequences, speed deviation sequences, and seedling needle frequency sequences; a reliable load processing module for generating reliable load sequences; a symbol state processing module for generating symbol segments and processing cycle status codes; a dynamic zoning module for generating dynamic load boundaries and current load zoning; an energy efficiency strategy determination module for outputting the engine target speed range and the hydraulic variable pump target displacement range; and an execution control module for correcting subsequent control actions based on the pre-execution residual and the post-execution residual.

[0016] The beneficial effects of this invention are reflected in: In the data processing method for rice transplanting energy efficiency scheduling based on dynamic load zoning, instead of directly triggering regulation based on the working pressure data of a single hydraulic system or the output speed data of a single engine, it first filters effective stable reference windows from historical sampling data, then generates a pressure sorting position sequence based on the effective stable reference windows, and combines it with the engine response difference sequence, speed deviation sequence, and seedling needle frequency sequence to form a reliable load sequence. Furthermore, it identifies short-term disturbances, continuous offsets, insufficient recovery, and insufficient execution changes through symbol segments, processing cycle status codes, and boundary crossing return data, ensuring that the current load zoning does not frequently switch due to a single boundary violation. Simultaneously, it adjusts the dynamic load boundary using the theoretical response stage sequence and the actual response stage sequence, enabling the dynamic load boundary to update with execution feedback. Finally, it judges the execution effect of the target power output strategy by using the pre-execution residual and the post-execution residual, and adjusts the machine's operating speed through the unit load difference sequence when multiple rice transplanters are working collaboratively, thereby reducing zoning switching caused by single-point boundary violations, reducing unidirectional control actions when the post-execution residual is not improved, and providing a zoning basis for rice transplanting energy efficiency scheduling. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0018] Figure 1 This is a schematic diagram illustrating the steps of the rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning according to the present invention; Figure 2 This is a schematic diagram of some steps in S1 of the rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning of the present invention. Figure 3 This is a schematic diagram of some steps in step S2 of the rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning of the present invention; Figure 4 This is a schematic diagram of some steps in S3 of the rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning of the present invention. Figure 5 This is a schematic diagram of some steps in step S4 of the rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning of the present invention. Figure 6 This is a schematic diagram of the module structure and data flow of the rice transplanting energy efficiency scheduling data processing system based on dynamic load zoning according to the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0022] This invention provides a data processing method for rice transplanting energy efficiency scheduling based on dynamic load zoning, such as... Figure 1 As shown, in one embodiment, the method includes: S1. Acquire engine output speed data, hydraulic system working pressure data, operating speed data, and seedling needle action rhythm data during rice transplanter operation; filter effective stable reference windows from historical sampling data; insert the current hydraulic system working pressure data into the pressure sorting sequence of the effective stable reference window to generate a pressure sorting position sequence, and generate an engine response difference sequence, speed deviation sequence, and seedling needle frequency sequence. S2. Compare the starting point of the pressure sorting position entering the upper bias section with the starting point of the seedling needle frequency entering the lower bias section, and generate a reliable load sequence by combining the engine response difference sequence and speed deviation sequence within the response waiting period; convert the reliable load sequence into a symbol segment to generate a processing cycle status code. S3. Generate dynamic load boundaries based on the effective stable reference window, count the boundary crossing return data of the load partition candidates relative to the dynamic load boundaries, and determine the current load partition based on the processing cycle status code and the boundary crossing return data. S4. Obtain the current load partition, match the target power output strategy according to the current load partition, and output the target speed range of the engine and the target displacement range of the hydraulic variable pump; obtain the residual before and after the execution of the target power output strategy relative to the same specific target parameter, and correct subsequent control actions in the same direction according to the change of residual.

[0023] In this embodiment, it should be noted that in S1, this method is used to reduce load mispartitioning and partition oscillation caused by short-term pressure disturbances, unilateral cycle time abnormalities, and execution link lag. The processing cycle is the time unit in which the whole machine controller completes data merging, status updating, and strategy judgment at fixed time intervals; the pressure sorting position is the normalized empirical position of the current pressure within the stable reference pressure distribution; the seedling needle frequency is the single-row transplanting frequency formed by adjacent effective hole-forming events within the same transplanting row, which can also be recorded as the seedling needle grasping frequency in the original action interface; the engine response difference is the directional difference between the current actual engine speed and the engine target speed recorded in the previous processing cycle; the speed deviation data is the direction and magnitude of the deviation of the current operating speed from the agronomic task target speed. The above names correspond to independent data channels, and data of different dimensions are not directly added together. The whole machine performance boundary is preferentially based on the manufacturer's hydraulic schematic diagram, pump and valve rated data, engine power mapping, and measured results from calibrated sensors. Standard records are used to standardize measurement objects, operating conditions, and units, and cannot replace the calibration of rated pressure, overflow setting, available power, and control limit for specific machine models.

[0024] Specifically, S1 is applied to the period when the high-speed riding rice transplanter is in a straight, continuous transplanting state. The engine controller provides the rotational speed and target record, the pressure sensor installed at the outlet of the variable pump in the continuous transplanting circuit provides the hydraulic system working pressure, the wheel speed and satellite positioning fusion interface provides the operating speed, and the seedling needle position or drive readback interface provides the hole-forming event time with a unified clock. The system excludes data corresponding to turning, transplanting unit lifting, stopping, malfunctions, and non-transplanting states, and takes the median value of multiple valid readings within the same processing cycle to obtain synchronous operation data. This processing is not simply a collection of sensor readings, but is used to ensure that subsequent time-series comparisons point to the same physical operation stage.

[0025] Furthermore, S1 employs a two-stage reference construction process. The first stage uses the manufacturer-allowed range and the calibration pressure sequence frozen from the previous operating stage to form an initial check interval, thereby filtering candidate historical windows. The second stage uses only windows that pass integrity, continuous out-of-bounds checks, direction reversals, and readback consistency checks to form the current pressure sorting sequence. An effective stable reference window can be multiple discontinuous stable sub-windows; distribution statistics allow merging data from each sub-window, but the direction of change, adjacent action intervals, and response stages are calculated only within a single continuous sub-window. This avoids circular dependencies caused by filtering data using dynamic load boundaries that haven't yet been generated, and also avoids splicing timing sequences across turns or downtime.

[0026] For example, with an exemplary processing cycle of 0.1s, the three valid engine readings in the current cycle are 1848 r / min, 1850 r / min, and 1852 r / min, with a median of 1850 r / min; the three valid pressure readings are 13.7 MPa, 13.8 MPa, and 13.9 MPa, with a median of 13.8 MPa; and the three valid speed fusion interface readings are 1.15 m / s, 1.16 m / s, and 1.17 m / s, with a median of 1.16 m / s. Before operation, the travel distance of 20 m and the time taken of 17.2 s were recorded in the field calibration section. The distance was divided by the time to obtain the verification speed of 1.16 m / s, which was retained to two decimal places. This result is only used to verify the speed channel ratio and does not replace the current cycle readback. Six consecutive effective acupoint formation moments form five complete action intervals, with a span of 0.92 seconds between the first and last moments. Dividing the number of complete intervals by this span yields a converged needle frequency of 5.4 Hz. These data are then input into the rotation speed, pressure, velocity, and beat channels, with clear origins.

[0027] The reason for adopting the above processing logic is that hydraulic pressure drifts slowly with mud resistance, planting depth, and oil temperature, while the update time and filtering delay of different interfaces are not the same. By unifying timestamps, continuous operating condition screening, freezing calibration sequence screening, and relative sorting processing, S1 solves the comparison distortion caused by asynchronous sampling, abnormal historical data, and changes in absolute pressure baseline, so that the relative pressure position, single-row planting frequency, and engine response have a traceable time basis, thus providing a reliable input for the sequential relationship judgment of S2.

[0028] In S2, to address the issue that pressure changes and needle frequency changes may originate from different sampling links, the system uses a unified controller clock to align the pressure data and effective acupoint formation time, and compensates for delays caused by pressure sensor filtering, bus transmission, and the formation of complete action intervals. Only when the absolute value of the time difference between the entry points of two segments is greater than the comprehensive time delay uncertainty is one entry point determined to precede the other; if the difference does not exceed the comprehensive time delay uncertainty, it is determined to be synchronous or the timing needs to be confirmed, preventing communication delays from being mistaken for the order of mechanical responses.

[0029] Furthermore, when the pressure sorting position first enters the upper bias segment and the seedling needle frequency subsequently enters the lower bias segment, a clock pressure verification result is generated indicating that the pressure first moves upward and the seedling needle then moves downward; when the seedling needle frequency first enters the lower bias segment but the pressure sorting position does not enter the upper bias segment, a clock pressure verification result is generated indicating abnormal observation on the seedling needle side; when both starting points are determined to be synchronized and both return to their respective stable segments within a preset return cycle, a clock pressure verification result indicating short-term synchronous offset is generated. Subsequently, using the engine response difference sequence as the response direction input, the speed residual of the same target speed segment is checked to see if it continuously decreases within the response waiting period, thereby generating a power coverage verification result indicating whether the power coverage is effective, lagging, or ineffective. Power coverage is a state description of whether the engine actually follows the predetermined target, and does not necessarily mean that the engine's nominal power is sufficient.

[0030] For example, if the pressure shift begins 0.3 seconds earlier than the frequency shift, the overall time delay uncertainty obtained from calibration is 0.1 seconds, thus confirming that the pressure shifts upward first. In the 20 historical execution readbacks of valid response records, the 95th quantile time when the engine begins to approach the same target speed range is 0.6 seconds. Adding two processing cycles as communication and mechanical inertia margins, the response waiting period is set to 0.8 seconds. Within this waiting period, if the speed residual continuously decreases and the speed only shows a slight negative deviation of 0.04 m / s, the data is written into the reliable load data; if the frequency shifts downward on one side only, the data is written into the load observation data; if the cycle pressure relationship points to increased load but the speed residual does not improve, the data is written into the execution lag observation data, and a reliable load sequence is formed according to the processing cycle order. Subsequently, the changes in the pressure sorting position of adjacent cycles are recorded as upward, downward, hold, or alternating directions, and consecutive records with the same direction are merged to form a symbol segment; if 8 cycles within the short window move upward, the current extreme value interval is higher than the reference extreme value interval as a whole, the return process shows insufficient recovery and the execution residual is significantly reduced, then the processing cycle status codes of continuous upward, overall upward, insufficient recovery, and effective execution are formed in a fixed order.

[0031] The reason for adopting the above processing logic is that using high pressure alone cannot distinguish between continuous planting resistance, short-term mud impact, and hydraulic baseline drift; similarly, using frequency decrease alone cannot distinguish between load increase, missed readings, and local faults on the seedling needle side. S2 solves the problem of uncertain meaning of a single signal by cross-checking pressure-cycle timing, residual at the same target rotation speed, and speed deviation, allowing real continuous load increase, cycle-side anomalies, and execution link lag to enter different data paths; symbol segments and fixed-sequence status codes further preserve direction, amplitude, recovery, and execution effect, avoiding short-term repetitive changes from directly triggering load zoning switching, thereby providing physically meaningful and continuously constrained load evidence for dynamic zoning.

[0032] In S3, to address the issue of fixed pressure thresholds failing to adapt to changes in field resistance and oil temperature, the system generates three basic sorting position boundaries from historical valid operational data with load status, fuel consumption per unit operating area, seedling needle frequency, and operational quality markers, and uses these to initialize the dynamic load boundary. Operational quality markers are derived from checks performed according to agronomic tasks and applicable rice transplanter testing methods, including checks for missed planting, damaged seedlings, floating seedlings, lodging, planting depth, and uniformity. Online frequency and speed judgments are only considered permissible conditions for transplanting rhythm; acceptable rhythm is not directly equated with acceptable overall operational quality.

[0033] Furthermore, S3 forms a theoretical response stage sequence of response start, response enhancement, and response stabilization based on the engine target speed change records and historical effective response records that have occurred before the current processing cycle. The system reads the most recent target strategy record that was output before the current processing cycle and whose instruction age has reached the response waiting period, reads the corresponding engine speed and pressure feedback, and extracts the actual response start point, actual response enhancement point, and actual response stabilization point. The engine target speed record saved immediately after the previous processing cycle is only used to generate the engine response difference and is not mistakenly considered to have gone through a complete response waiting period. When extracting each stage point, in addition to checking the engine speed residual, it also checks whether the change direction of the hydraulic system working pressure is consistent with the direction of the same feasible pair of historical effective responses, and excludes data during the pressure protection period; only when the speed condition and pressure verification condition are met simultaneously is the corresponding actual response stage point confirmed. The boundary back-calibration result is the execution feedback calibration result defined in this invention, indicating that the system compares the stage integrity, stage sequence, and residual changes after execution item by item, resulting in the indications of maintaining the boundary, extending the waiting period, freezing the boundary, limited outward movement, or limited inward movement, without modifying the original sampled data. The target output record generated by S4 in this processing cycle only enters the subsequent processing cycle S3 after the instruction age reaches the response waiting period. Therefore, there is no causal inversion of adjusting the current boundary with the future execution result.

[0034] For example, the three basic sorting position boundaries obtained from historical labeled data optimization are 0.28, 0.58, and 0.84, respectively. The current pressure sorting position is 0.91, and high-load partition candidates are first formed. If a candidate only exceeds the boundary for 2 processing cycles, returns to the stable range within the next 3 processing cycles, and remains stable for 10 processing cycles, the load partition of the previous processing cycle is maintained. If a candidate exceeds the boundary for 8 consecutive processing cycles, exceeding the 6 processing cycle confirmation length determined by historical continuous load increase records, and does not recover within the return cycle, the current load partition is determined according to the candidate after correction for beat rate, power coverage, and speed deviation. The boundary crossing return data therefore includes the first crossing position, the duration of the boundary crossing, the number of sampling points required to return to the stable range, and the duration of stability after the return.

[0035] The reason for adopting the above processing logic is that the dynamic boundary needs to adapt to changes in operating conditions, but it cannot move outward and mask risks when the actuator malfunctions or there is a real overload. The boundary is maintained when the actual response phase is complete and the residual decreases after execution; the response waiting period is extended when the actual response is complete but the overall response is later than the historical effective response record; when the upper-level zone is continuously triggered and the residual does not improve, the boundary is frozen and the execution link and pressure protection status are checked. Limited outward movement is only allowed if it is confirmed to be a false positive for the zone; the boundary is moved inward when pressure protection, continuous speed decrease, or continuous frequency deviation occurs without triggering the upper-level zone. The dynamic load boundary is only used for energy efficiency scheduling and does not replace the maximum permissible continuous operating pressure, independent overflow protection setting, and absolute safety pressure limit determined based on the manufacturer's data, nor does it replace the continuous operation advance protection trigger pressure determined based on pressure calibration records and protection action tests; when the pressure reaches any corresponding protection condition, it is processed directly according to the protection sequence, without being limited by status codes, boundary crossing return data, or the current load zone.

[0036] In S4, the system determines the observation priority, primary control priority, secondary control priority, or protection priority based on the current load zone, processing cycle status code, and boundary crossing return data. The observation priority maintains the original target power output strategy and continues sampling; the primary control priority selects feasible pairings between adjacent load zones; the secondary control priority selects the higher power feasible pairing that meets the operational requirements in the current load zone; the protection priority limits travel speed, restricts pump demand, or performs unloading, and prohibits further amplification of control actions in the same direction. Priorities describe the urgency of the action and do not mechanically equate high load with simultaneously increasing engine speed and pump displacement.

[0037] Furthermore, the energy efficiency mapping table is formed by grouping historical effective operating data according to load zoning, pressure sorting position segments, speed deviation segments, and operating quality markers. Each record contains paired constraints for the engine target speed range, hydraulic variable pump target displacement range, and fuel consumption per unit operating area. Candidate engine target speed ranges and candidate hydraulic variable pump target displacement ranges are output only as a single feasible pairing record. Each feasible pairing satisfies the pump's permissible continuous flow limit and ensures that the sum of the pump shaft input power and the power of the travel, mechanical insertion, and electrical auxiliary loads not yet included in the pump shaft input power does not exceed the engine's available power at the corresponding speed. Insertion hydraulic power supplied by the variable pump is not counted repeatedly. This constraint is used to prevent excessive pump torque after displacement increases, which could exacerbate engine speed drops.

[0038] For example, historical records provide candidate speed ranges of 1900 r / min to 2050 r / min and corresponding candidate displacement ranges of 20 mL / r to 22 mL / r. The system selects the specific pairing of 1900 r / min and 20 mL / r. Based on the pump calibration records, the estimated flow rate is 30.8 L / min, obtained by processing with a transmission ratio of 0.9 and a volumetric efficiency of 0.9, which is lower than the allowable continuous flow rate of 40.0 L / min given by the thermal balance test. At a pressure of 13.8 MPa, the internally unrounded hydraulic output power is 7.08 kW. Then, based on the variable pump overall efficiency of 0.84, the internally unrounded pump shaft input power is 8.43 kW, which is recorded as 7.1 kW and 8.4 kW after retaining one decimal place, respectively. After adding the other load power of 3.0 kW obtained from the bench power balance test, which has not yet been included in the pump shaft input power, the total demand is 11.4 kW, which is lower than the available power of 14.5 kW given by the engine mapping records.

[0039] The reason for adopting the above processing logic is that high pressure does not necessarily correspond to high flow rate, and the power requirements of high pressure and low flow rate are different from those of low pressure and high flow rate. S4 makes the target strategy physically executable through paired retrieval of speed and displacement, flow rate and power constraints, feedback of residuals with the same target, and speed limits when necessary. When the difference between the residuals before and after execution enters the close range, the action is reduced; when the difference does not enter the close range and the relative improvement reaches the effective improvement threshold, the action is maintained; when the relative improvement does not reach the threshold, the action is reduced; when the residual increases in the opposite direction and the increase exceeds the close range, the action in the same direction is canceled, so that the partitioning results, power commands, and execution readback form a closed loop.

[0040] In summary, the entire data processing method for rice transplanting energy efficiency scheduling based on dynamic load zoning firstly addresses the issues of asynchronous sampling, abnormal historical windows, and absolute pressure baseline drift by using a unified clock, two-stage reference filtering, and relative pressure sorting. Simultaneously, it generates a reliable load sequence by combining the physical starting point relationship between pressure upward bias and single-row transplanting frequency downward bias, the residual of the target rotational speed, and speed deviation, thus distinguishing between real continuous load increases, seedling needle-side anomalies, and execution link lags. Furthermore, it suppresses zoning oscillations caused by short-term boundary overflows by using symbol segments, processing cycle status codes, and boundary crossing return data. The boundary back-calibration between the theoretical response stage and the actual response stage enables the dynamic load boundary to adapt to different field conditions within the cumulative adjustment envelope. As a result, under the premise of maximum allowable continuous working pressure, continuous operation early protection trigger pressure, independent overflow protection setting, and independent protection of the absolute safety pressure limit, the system outputs the engine target speed range and hydraulic variable pump target displacement range in pairs from the energy efficiency mapping table that meets the flow and power constraints of the transition range. Then, based on the residual correction control action before and after the execution of the same specific target, the system finally obtains the traceable current load partition, target power output strategy, and execution feedback results.

[0041] like Figure 2 As shown, in one embodiment, S1 includes:

[0042] S11. Collect engine output speed data, hydraulic system working pressure data, working speed data, and seedling needle action rhythm data, and perform unified timestamp marking, range verification, outlier checking, and processing cycle merging to obtain synchronous operation data.

[0043] S12. Divide the historical sampling data into candidate historical windows according to the same window length, and filter them in turn according to integrity, continuous out-of-bounds, direction reversal and execution readback consistency to obtain an effective and stable reference window.

[0044] S13. Arrange the pressure data in the effective stable reference window from low to high to form a pressure sorting sequence, insert the current hydraulic system working pressure data, obtain the pressure sorting position according to the midpoint rank of the parallel data, and form a pressure sorting position sequence.

[0045] S14. Generate engine response difference sequence, speed deviation sequence and seedling needle frequency sequence based on synchronous operation data, and generate a beat confirmation mark for processing cycles that do not form a complete action interval.

[0046] In this embodiment, it should be noted that in S11, the main data acquisition component is the machine controller and its interface circuit. Engine speed and target data are entered into the machine controller via the engine controller bus interface; pressure data is entered via the pressure sensor and analog-to-digital converter interface installed on the continuous rice transplanting circuit at the variable pump outlet; speed data is entered via the wheel speed sensor and satellite positioning fusion interface; and motion cycle data is entered via the seedling needle position sensor or drive readback interface. Each record carries the sampling time generated by the controller's unified clock, channel identifier, validity flag, and operation status flag. Before operation, the sensors perform zero-point, range, and temperature drift checks. Isolated points that exceed the sensor's range, have a rate of change exceeding the equipment's allowable value, or are not supported by adjacent readings are marked as abnormal points, but the original records are still stored in the data recording module for verification.

[0047] The anomaly detection limit is jointly calibrated using stable operating conditions and permissible step operating conditions. The system replays 20 stable records and 20 actual acceleration / deceleration or pressure step records respectively, filtering candidate upper limits for the rate of change to retain actual step changes and exclude single-point spikes. For example, the upper limits for the rate of change of engine speed, pressure, and speed are set to 800 r / min / s, 50.0 MPa / s, and 1.5 m / s², respectively. When three pressure readings with timestamps of 0s, 0.02s, and 0.04s are 13.0 MPa, 15.0 MPa, and 13.1 MPa, respectively, the pressure change in the first segment is 2.0 MPa, which, after dividing by 0.02s, yields a rate of change of 100.0 MPa / s; the pressure change in the second segment is 1.9 MPa, which, after the same processing, yields 95.0 MPa / s. Both sides of the pressure change rate exceeded the upper limit, and the difference between the first and last readings was only 0.1 MPa. Therefore, only the intermediate reading was marked as an isolated anomaly. If the subsequent reading continued to approach 15.0 MPa, it was treated as a real pressure increase and could not be deleted as an anomaly.

[0048] The sample size and single-segment duration used to determine the processing cycle were obtained through stability comparisons, rather than being directly specified. The system used 20, 30, and 40 segments of fault-free continuous rice planting records to calculate the channel update time statistics. When increasing from 30 to 40 segments, the change in each statistical value did not exceed 0.01s, so 30 segments were chosen. Comparing the single-segment durations of 30s, 60s, and 90s, 60s was sufficient to cover more than 300 independent action intervals under stable conditions with a frequency not lower than 5.9Hz, while extending it to 90s changed the update time quantile by less than 0.01s. Therefore, the example used 60s per segment.

[0049] The processing cycle is not an arbitrary preset constant. The system calculates the 95th percentile of adjacent valid update intervals for the continuous data channels of engine speed, pressure, and speed, and calculates the 95th percentile of the arrival delay from valid edge triggering to timestamped data being written to the controller for the action event channel. The maximum value from these results is then taken and rounded up to an integer multiple of the controller's base clock. For example, the update time statistics for the speed, pressure, and speed channels are 0.05s, 0.02s, and 0.1s respectively, and the action event arrival delay is 0.08s, with a maximum value of 0.1s. Therefore, the processing cycle is set to 0.1s. The physical interval of an action can be longer than the processing cycle; therefore, the system only stores the most recently confirmed frequency and freshness flag, and does not require a new complete action interval to be generated in each cycle.

[0050] Furthermore, after acquiring synchronous operation data, S12 receives historical engine output speed data, historical hydraulic system working pressure data, historical operation travel speed data, historical seedling needle action rhythm data, historical control commands, and historical execution readback data. The frozen calibration pressure sequence refers to a reference sequence that has been reviewed before the start of a continuous operation phase and is prohibited from being rewritten by current data during that phase. Upon initial generation, six valid windows that have passed all screenings are obtained from the previous calibration operation. Each window retains 50 median pressure values, forming a total of 300 reference values, which are then written to the version number. After the operation phase ends, only new windows with consistent execution readbacks, no protective actions, and passing quality verification are used for the next version update; if there are fewer than 300 new samples or the verification fails, the original version continues to be used. For example, if version 3 is used in the current phase, version 4 is not generated if there are only 250 valid update candidates, thus preventing the current abnormal pressure from back-contaminating the reference sequence used for self-judgment.

[0051] Effective load change events are used to select historical ranges, and their extraction rules are formed based on joint evidence of pressure, beat, and velocity. The system requires that the pressure ranking position change continuously in the same direction for at least six processing cycles relative to the previous freezing boundary, and that subsequent records show frequency changes of three consecutive independent action intervals or velocity changes in the same direction for five consecutive processing cycles, to be counted as an effective load change; isolated pressure transgressions are not counted. When replaying 20 records of different mud layer resistances, historical ranges with fewer than three events tend to cover only a single resistance state, while boundary estimations tend to stabilize after three events. For example, a 60-second record satisfies the above joint conditions at 12s, 31s, and 47s, thus containing three effective load changes.

[0052] The number of joint confirmations for effective load change events was determined through limited playback comparison. The system used 20 actual mud layer loading events, 20 short-duration impacts, 24 missed action readings, and 15 slippage records. Candidate combinations were compared for pressure over 4, 6, and 8 consecutive processing cycles; frequency over 2, 3, and 4 consecutive independent intervals; and velocity over 3, 5, and 7 consecutive processing cycles. With the shorter combination, 5 short-duration impacts or missed readings were mistakenly counted as load changes. With the longer combination, the average delay for actual load loading was 0.3 seconds. The combination of 6 pressure cycles, 3 frequency intervals, and 5 velocity cycles resulted in no missed actual load loading records and only 1 unconfirmed record; therefore, this combination was adopted.

[0053] The historical range and window length are determined through limited historical records. The system compares the separation results of 30s, 60s, and 90s historical ranges and 3s, 5s, and 10s windows for slow baseline changes, short-term impacts, and turning boundaries from the aforementioned 20 consecutive straight-line operation records. It selects a combination that covers at least three effective load changes without a single window crossing a turn. In the example verification, the 30s window only covers two events, and the 10s window crosses four load condition changes. Using the most recent 60s and 5s windows covers three events without crossing a turn. Twelve candidate windows are formed with a processing cycle of 0.1s, each containing 50 cycles.

[0054] The integrity screening exclusion limit was determined by manually masking 1 to 6 sampling points for each of the 20 records mentioned above and repeatedly generating sorting positions, frequency freshness, and final partitioning results. The acceptable range was defined as a pressure sorting position change not exceeding 0.02, a frequency confirmation delay increase not exceeding 0.2s, and a partitioning consistency rate not lower than 95%. These standards are derived from the upper bound of fluctuations in repeated processing without missing records. In the example, when 4 points were masked, the sorting position change was 0.01, the confirmation delay increased by 0.1s, and the partitioning consistency rate was 98%; when 5 points were masked, the corresponding results were 0.03, 0.3s, and 92%. Therefore, a single-window missing record reaching 5 processing cycles was determined as the integrity screening exclusion limit.

[0055] The initial screening interval was determined by combining candidate quantile combinations with the manufacturer's allowable range. The system compared candidate combinations from 0.5% to 99.5%, 1% to 99%, and 2% to 98% across 30 stable windows and 25 short-term shock or sustained load records. Evaluation was based on stable window retention rate, shock exclusion rate, and the number of sustained anomaly misses. The final interval was 1% to 99%, with a more stringent boundary taken from the manufacturer's range. Example frozen data yielded a pressure range of 10.6 MPa to 14.1 MPa, a velocity range of 1.1 m / s to 1.3 m / s, and a needle frequency range of 5.9 Hz to 6.3 Hz. These intervals were used only for initial screening and did not replace subsequent dynamic boundaries.

[0056] The number of consecutive out-of-bounds exclusions is determined based on the limited anomaly records for each channel. The pressure channel uses 25 short-duration impact and continuous load records; the first 5 out-of-bounds cycles may recover spontaneously, and the continuous anomaly can be stably separated after the 6th cycle, so the pressure upper limit is set at 6 processing cycles. The speed channel uses 20 slippage records; after 5 consecutive out-of-bounds cycles, it separates from the instantaneous positioning error, so 5 processing cycles are used. The frequency channel uses 24 missed reads and 18 real beat descent records; it is excluded only after 3 consecutive independent complete action intervals out of bounds. In the example, a candidate window has 6 consecutive pressure out-of-bounds cycles while its speed only has 2, but it is still excluded because the pressure reaches the corresponding upper limit.

[0057] The measurement dead zones for each channel were obtained through repeatability calibration. The pressure sorting position dead zone was derived from the maximum normalized fluctuation of 100 constant-pressure repeated samplings, rounded up (example: 0.01). The velocity dead zone was derived from the 95th percentile of the fused velocity difference between adjacent data points in 100 uniform linear records, rounded up (example: 0.02 m / s). The seedling needle frequency dead zone was derived from the 95th percentile of the single-interval frequency fluctuation in 30 stable action records (example: 0.1 Hz). Data differences between adjacent data points not exceeding the corresponding dead zone were recorded as held, without accumulating direction reversal, thus avoiding the measurement of sensor repeatability errors as true reciprocating variations.

[0058] The upper limit for direction reversal was determined through a separation test of stable and oscillation records. The system counted the number of reversals for 30 stable windows and 20 oscillation windows after measuring the dead zone of each channel, and compared the number of reversals (3, 4, and 5) to determine the number of stable windows misplaced and oscillation windows missed. In the example, a 3-reversal upper limit resulted in 4 misplaced stable windows, a 5-reversal upper limit resulted in 3 missed oscillation windows, and a 4-reversal upper limit resulted in only 1 misplaced window with no sustained oscillation missed. Therefore, a 4-reversal upper limit was determined as the stability screening exclusion upper limit.

[0059] The tolerance for consistent readback marking is determined through 20 valid control records and 10 records of disconnection or jamming. The controller first confirms that the target readback is consistent with the direction of the issued target, then checks that the difference in engine target speed readback does not exceed 10 r / min, and checks that the difference in variable pump equivalent displacement readback does not exceed 0.5 mL / r; if any readback is missing, reversed, or exceeds the tolerance, it is marked as inconsistent. In the example, the engine target is 1900 r / min, the readback is 1898 r / min, the pump target is 20 mL / r, and the equivalent readback is 19.7 mL / r. Both differences are within the tolerance, so it is marked as consistent; if the pump readback is still 18.0 mL / r, it is marked as inconsistent.

[0060] S12 excludes candidate historical windows in a fixed order based on the following criteria: the number of missing data reaches the integrity screening exclusion limit, the number of consecutive out-of-bounds checks reaches the boundary screening exclusion limit, the number of direction flips reaches the stability screening exclusion limit, or the consistency requirement is not met during readback consistency checks. The remaining candidate historical windows are then designated as valid stable reference windows. For example, out of 12 candidate windows, 3 are excluded in the integrity check, 2 in the consecutive out-of-bounds check, and 1 in the stability check. No further exclusions are made during the readback consistency check, resulting in 6 windows remaining, yielding a total of 300 reference stress data points. Each check only processes windows not yet excluded in the previous step, thus avoiding double counting.

[0061] Furthermore, S13 obtains the pressure sorting position based on the current hydraulic system working pressure and pressure sorting sequence. The obtained pressure sorting position can be represented as:

[0062]

[0063] In the above expression: Indicates the first The normalized pressure sorting position for each processing cycle is dimensionless data, with a value greater than 0 and not greater than 1, and is used to determine the relative position of the current pressure in the reference pressure distribution. This represents a positive integer cycle number generated by incrementing the controller's processing cycle. Indicates the first The median value of the effective sampled values ​​of the hydraulic system working pressure within each processing cycle, in MPa. The data comes from the pressure sensor that has completed zero-point, range and temperature drift calibration, and is within the effective range after calibration. Indicates that in the pressure sorting sequence, less than The number of reference pressure data points shall not exceed A non-negative integer, where the subscript l indicates a pressure lower than the current pressure; Indicates that the pressure sort sequence is equal to The number of reference pressure data points shall not exceed The non-negative integer, where the subscript e indicates that it is equal to the current pressure; The total number of reference pressure data included in the pressure sorting sequence is a positive integer, and the subscript r represents the reference data; the constant 1 represents the rank correction amount used when the current pressure is added as a new sorting object; the coefficient 0.5 represents the weight assigned to existing data with the same value when the midpoint rank is used for tied pressures. and This is obtained by comparing the current pressure with the reference pressure item by item. The pressure is obtained by counting from the reference pressure record; the current pressure is only temporarily inserted to obtain the rank and is not written back to the reference sequence. This expression is not executed when there is no valid reference data; the system uses the calibration pressure sequence frozen in the previous operation phase and generates a reference data update marker.

[0064] In this embodiment, the calculation logic of the expression is as follows: First, the pressure data within the effective stable reference window are arranged in ascending order of value to form a pressure sorting sequence; then, the current pressure is compared with each reference pressure item by item, and the number of data points lower than the current pressure and the number of data points equal to the current pressure are accumulated respectively; next, the data with the same value are processed using the midpoint rank and the rank correction amount corresponding to the current pressure is added; finally, the total amount of data after adding the current pressure is used to complete the normalization, and the pressure sorting position that can be consistently compared between the basic pressures of different fields is obtained.

[0065] The reason for adopting this expression is that mud layer thickness, hydraulic oil temperature, and insertion depth will change the absolute pressure baseline. Relative sorting can preserve the pressure increase trend and reduce the effect of slow drift. Midpoint rank can also avoid different results for the same pressure due to different writing order.

[0066] For example, an effective stable reference window provides 300 reference pressure data points. The three readings for the current cycle are 13.7 MPa, 13.8 MPa, and 13.9 MPa, with the median of 13.8 MPa serving as the baseline. A comparison of each item revealed that 272 reference pressures were lower than the current pressure, and 2 reference pressures were equal to the current pressure. Therefore, 272, 2, and 300 were substituted into the above expression as two comparison counts and the total number of reference data, respectively. The rank of the tied data, after being processed with a coefficient of 0.5, was 1. The numerator, after rank correction, was 274, and the denominator was 301. Dividing these values ​​and retaining two decimal places yielded a pressure ranking position of 0.91.

[0067] This calculation process solves the problem of difficulty in lateral comparison when absolute pressure values ​​drift with operating conditions, so that parallel pressures no longer have different ranks due to the order of writing, and provides a unified dimensionless input for subsequent upper section judgment and dynamic load boundary generation.

[0068] In S14, the system reads the current actual engine speed and subtracts the engine target speed record saved in the previous processing cycle from this actual speed to obtain a directional engine response difference value, which is then used to form an engine response difference value sequence according to the processing cycle. For example, if the target speed was adjusted from 1800 r / min to 1900 r / min eight processing cycles before the current cycle, and the target speed remains unchanged thereafter, the target record in the previous processing cycle immediately preceding the current cycle is still 1900 r / min; the current actual speed is 1850 r / min, and after subtraction, the engine response difference value is -50 r / min. When the absolute value of consecutive negative values ​​decreases towards 0, it indicates that the actual speed is approaching the target speed; the power coverage judgment still uses the shortest distance from the actual speed to the same target speed range, and different targets cannot be mixed.

[0069] Furthermore, speed deviation data is obtained by comparing the current speed with the target operating speed, recording the direction, absolute magnitude, and duration of the deviation. For example, if the current speed is 1.16 m / s and the agronomic task target speed is 1.2 m / s, the difference between the two yields a negative deviation of 0.04 m / s, which is then converted relative to the target speed to obtain a negative relative deviation of 3.3%. The target speed is derived from the agronomic combination of a target plant spacing of 0.2 m and a target single-row transplanting frequency of 6.0 Hz. The target speed is calculated by multiplying the plant spacing by the number of planting holes per unit time to obtain 1.2 m / s, and the speed channel ratio is verified through a 20 m measured section. Speed ​​deviation grading is determined by separating the distribution of quality-compliant straight-line operation records from records of vehicle getting stuck or slipping, rather than relying on a single subjective setting.

[0070] The seedling needle frequency data is generated based on the effective planting times of adjacent holes in the same planting row. The system first removes pulse jitter, repeated triggering, and cross-turn events, then counts the number of complete intervals formed by consecutive action moments, and divides the number of complete intervals by the duration of the first and last effective action moments. In the example, 6 effective action moments form 5 complete intervals, lasting 0.92s, resulting in an aggregated seedling needle frequency of 5.4Hz. This aggregated value is used for frequency amplitude and stability judgment, and is not directly used as the physical moment of the lower bias starting point. If a complete interval has not been formed in the current cycle, a beat confirmation mark is generated and the most recently confirmed frequency and its freshness are saved; it is updated again after a complete interval is formed. Continuous low-frequency determination must be based on 3 independent complete action intervals; repeatedly saving the same frequency value does not constitute new low-frequency evidence.

[0071] The parameters for extracting effective planting events were determined through the maximum normal planting frequency and repeated edge experiments. When the upper limit of the stable frequency is 6.3Hz, the normal interval between adjacent actions is approximately 0.16s. The system halves this interval and leaves a margin downwards to obtain a 0.08s repeated edge merging window; only the first edge is retained within this window. For example timestamps of 0s, 0.02s, 0.17s, and 0.34s, the 0.02s edge is deleted because its interval with the preceding edge is smaller than the merging window; the other three are retained. If a turning indicator changes from a straight line to a turn between two edges, the interval crossing this state change is directly cut off and not used for frequency calculation.

[0072] The frequency freshness expiration time was determined using 24 missed read records and 30 normal low-frequency records. The system compared candidate retention times of 0.3s, 0.4s, and 0.5s, prioritizing that normal actions are not prematurely invalidated and missed read data is not repeatedly used over a long period. For example, 0.4s was selected, corresponding to four processing cycles. If the most recent complete interval was formed before 0.3s, the system can retain its frequency value but add a "not latest" flag; if no new interval is formed by 0.4s, the frequency value is considered invalid and a tick confirmation flag is generated. Invalid values ​​cannot be used as new evidence of continuous downward bias.

[0073] The frequency data used for determining the starting point of the downward bias is formed based on a single complete action interval. For each new complete interval, the system immediately generates a single-interval frequency and records the midpoint of that interval as the physical event time. Downward bias is confirmed when three consecutive independent intervals are below the lower end of the stable segment. The midpoint of the first qualified interval is saved as the starting point of the downward bias, and the end time of the third interval is saved as the confirmation time. For example, if three consecutive interval frequencies are 5.8Hz, 5.7Hz, and 5.6Hz, and all are below the lower end of 5.9Hz, the midpoint of the first interval is used to determine the starting point. The aggregated value formed by the five intervals is not included in the comparison of the starting point order, thus avoiding false delays caused by the aggregation window.

[0074] like Figure 3As shown, in one embodiment, S2 includes:

[0075] S21. Determine the upper bias segment of the pressure sorting position and the lower bias segment of the seedling needle frequency based on the effective stable reference window, and perform unified clock alignment, link delay compensation and comprehensive delay uncertainty judgment on the entry point of the two segments.

[0076] S22. Generate cycle pressure verification results based on the order of the starting points, generate power coverage verification results based on the engine response difference sequence and the speed residual change of the same target speed segment, and form a reliable load sequence by combining the speed deviation sequence.

[0077] S23. Record the pressure sorting position changes of adjacent processing cycles in the reliable load sequence as upward shift, downward shift, hold, or alternating direction, and merge records with the same direction to form a symbol segment.

[0078] S24. Generate direction status code, amplitude status code, recovery status code and execution status code according to the direction and continuous length, the position of the current pressure sorting position extreme value interval relative to the reference pressure sorting position extreme value interval, the return to stability process and the execution residual change, and form the processing cycle status code in a fixed order.

[0079] In this embodiment, it should be noted that in S21, the upper deviation segment is determined jointly by the effective stable reference window and the tagged verification records. The system first converts the pressure value in the effective stable reference window into a pressure ranking position relative to the calibration sequence that was frozen before the occurrence. Then, it traverses the candidate boundaries on 30 records with qualified operation quality markings and consistent execution readbacks, using the continuous load increase recognition rate, the number of short-term impact false alarms, and the number of partition repetitions as evaluation criteria. In the example, the number of short-term impact false alarms at candidate boundaries of 0.8, 0.84, and 0.88 are 5, 2, and 1, respectively, while the number of continuous load increase missed detections are 1, 1, and 4, respectively. Therefore, the range where the pressure ranking position reaches 0.84 is determined as the upper deviation segment.

[0080] The stable and lower deviation ranges of the seedling needle frequency were determined by independent action interval records of qualified quality. The system sorted the single-interval frequencies in 30 continuous operation records, and after excluding turns, missed readings, and repeated edges, verified the impact of the candidate lower end frequencies of 5.8Hz, 5.9Hz, and 6.0Hz on the missed detection of low-frequency anomalies and the false alarms of normal fluctuations. When using 5.9Hz, there was 1 false alarm and no missed detection of real low-frequency anomalies. The system also compared the candidate upper end frequencies of 6.2Hz, 6.3Hz, and 6.4Hz with the above stable records and 12 overspeed or repeated trigger records. When using 6.2Hz, 4 normal high-frequency records were mistakenly excluded; when using 6.4Hz, 2 continuous high-frequency anomalies were missed; and when using 6.3Hz, there were no missed detections of continuous anomalies and only 1 record awaiting confirmation. Therefore, the range below 5.9Hz was defined as the lower bias segment, and the range from 5.9Hz to 6.3Hz was defined as the stable segment; this range corresponds to approximately 354 to 378 hole formation events per minute, and the example hole spacing of 0.2m is within the allowable range for high-speed machine tasks.

[0081] The number of consecutive frequency confirmations was determined through 24 complete action interval missed reading tests and 18 actual beat descent records. The system was tested for confirmation conditions of 2, 3, and 4 consecutive independent intervals. With 2 intervals, 3 missed readings were misjudged as downward deviation; with 4 intervals, the average wait time for the actual descent was 0.2 seconds longer; with 3 intervals, there were no continuous abnormal missed judgments and only 1 record to be confirmed was retained. Therefore, 3 independent intervals were used for the example. The count is accumulated according to the newly formed action intervals. Repeatedly saving the same frequency value in the processing cycle will not increase the count.

[0082] The preset return cycle is statistically analyzed based on 30 mud impacts and local undulations in the field surface. The system sequentially records the time from the synchronous offset starting point to the simultaneous return to their respective stable sections for each pressure sorting position and seedling needle frequency, comparing the 85th, 90th, and 95th percentiles as candidates; the 90th percentile is 0.3s, which covers 27 short-term events that can recover on their own, and does not retain long periods of continuous load increase as short-term offsets, so it is processed with a cycle of 0.1s for 3 cycles. This parameter only identifies short-term synchronous offsets and is not used as the engine response waiting period.

[0083] The overall time delay uncertainty was determined through bench synchronous triggering tests. The system inputs a common trigger signal to the pressure sampling link and the seedling needle motion sampling link, measuring the earliest and latest differences caused by sensor filtering, analog-to-digital conversion, bus queuing, timestamp quantization, single-interval midpoint time assignment, and data arrival. In 100 tests, a maximum arrival difference of 0.08s was obtained covering 95% of the records. Adding the 0.02s timestamp margin provided by the controller clock configuration, the overall time delay uncertainty was calculated to be 0.1s. Only when the absolute value of the difference between two starting points corrected for physical event times exceeds this value is the order of arrival determined; otherwise, synchronization is considered. If any data point carries a time-series confirmation mark, it is preferentially considered to have a time sequence confirmation requirement.

[0084] Furthermore, S22 generates three types of mutually exclusive clock pressure verification results. When the pressure ranking position first enters the upper bias segment and the seedling needle frequency subsequently enters the lower bias segment, a result is generated where the pressure first shifts upward and the seedling needle then shifts downward; when the seedling needle frequency first enters the lower bias segment but the pressure ranking position does not enter the upper bias segment, an abnormal observation result on the seedling needle side is generated; when the two starting points shift synchronously and recover within a preset return cycle, a synchronous short-time shift result is generated. Other records that do not meet the complete time evidence are retained as time series pending confirmation and are not directly included in the reliable load data. This ensures that each result has clear input conditions and avoids obtaining conflicting clock pressure conclusions in the same cycle.

[0085] The three types of beat pressure verification results can be reproduced through recordings at different times. For example, in the first set of records, the pressure upward deviation started at 12.0s, and the frequency downward deviation started at 12.3s. After correction, the difference exceeded 0.1s, generating a result where the pressure shifted upward first and the seedling needle shifted downward later. In the second set of records, the frequency entered the downward deviation zone at 20.0s, and the pressure ranking position was below 0.84 within the following 0.8s, generating an abnormal observation result on the seedling needle side. In the third set of records, the difference between the two starting points was 0.06s, and both returned to the stable zone within 0.3s, generating a synchronous short-time offset result. The three sets of data entered different paths and will not obtain two results simultaneously.

[0086] The response waiting period is determined jointly by consistent historical execution readback records for both engines and variable pumps. The system filters the most recent 20 target adjustments of the same type for engines and 20 displacement step changes for variable pumps. For engines, the system records the time from target issuance to the start of a continuous decrease in the speed residual; for variable pumps, it records the time from target issuance to stable actual current readback and an independent outlet flow trend consistent with the calibration direction. After sorting, the 95th percentile is taken, and a margin of 2 processing cycles is added. In the example, the engine percentile time is 0.6s, which becomes 0.8s after adding two 0.1s cycles; the variable pump percentile time is 0.4s, which becomes 0.6s after adding the same margin. The system takes the larger of the two results, 0.8s, which is equivalent to 8 processing cycles. If either type of historical sample has fewer than 20 instances, the more conservative option between the manufacturer's response upper limit and the previous calibration version is used and marked as needing an update, avoiding shortening the waiting time with a few incidental records.

[0087] Power coverage verification uses the engine response difference sequence to confirm the direction of actual speed change, and then uses the shortest distance from the actual speed to the same target speed segment to confirm the degree of improvement. When the actual speed is lower than the lower end of the target segment, the actual speed is subtracted from the lower end of the target segment; when it is within the target segment, the residual is recorded as 0 r / min; when it is higher than the upper end of the target segment, the upper end of the target segment is subtracted from the actual speed. Power coverage is considered valid when the residual continuously decreases within the response waiting period and meets the effective improvement requirements obtained from historical verification; power coverage is considered lagging when the residual starts to decrease but still does not meet the requirements at the end of the waiting period; and power coverage is considered invalid when the residual does not decrease, increases in the opposite direction, or the execution readback is inconsistent. These three types of results are mutually exclusive.

[0088] Effective improvement requirements for power coverage and the number of consecutive reductions were determined through records of 20 successful responses, 10 hysteresis responses, and 15 invalid responses. The system tested improvement levels of 0.1, 0.2, and 0.3, and tested the confirmation criteria for decline over 1, 2, and 3 consecutive processing cycles. Evaluation was based on the separation rate between success and hysteresis and the number of false noise judgments. An example was selected where the improvement level reached 0.2 and decreased for at least 2 consecutive cycles. Effective improvement was defined as a decrease in residual speed from 100 r / min to 80 r / min and then to 50 r / min; hysteresis was defined as a decrease from 100 r / min to 95 r / min and then to 85 r / min with an improvement of only 0.15 at the end of the waiting period; and invalid improvement was defined as an increase from 100 r / min to 105 r / min and then to 120 r / min.

[0089] Speed ​​deviation zones are determined using 40 straight-line records of acceptable operation quality and 15 records of slippage or getting stuck. The system iterates through candidate boundaries according to the percentage deviation relative to the target speed, ensuring that acceptable records are kept in the normal or slightly negative deviation zones as much as possible, and that continuous stall records are placed in the severely negative deviation zone. Example verification results classify deviations not exceeding 2.0% as normal, negative deviations exceeding 2.0% but not exceeding 5.0% as slightly negative deviations, negative deviations exceeding 5.0% but not exceeding 10.0% as load observations, and negative deviations exceeding 10.0% for five consecutive processing cycles as severely negative deviations. For example, a negative relative deviation of 3.3% is considered slightly negative and can serve as supplementary evidence of increased load; severely negative deviations are transferred to protection observation, and power increases are no longer solely based on pressure ranking position.

[0090] Trusted load data, load observation data, and execution lag observation data are mutually exclusive custom record types. Trusted load data is written when pressure shifts upward first and seedling needles then deflect downward, power coverage is effective, and velocity has not entered a severely negative bias. Load observation data is written when there are abnormal observations on the seedling needle side, short-term synchronization offsets, or timing issues requiring confirmation. Execution lag observation data is written when the cycle pressure relationship supports increased load but power coverage is ineffective, execution readback is inconsistent, or residuals increase in the opposite direction. In the example, records with pressure starting 0.3 seconds earlier, improvement of 0.5, and velocity negative bias of 3.3% are entered into trusted load data; records with frequency deflection on one side and pressure position of 0.6 are entered into load observation data; records with residuals increasing from 50 r / min to 70 r / min are entered into execution lag observation data. Each record must at least store the cycle number, pressure sorting position, two types of verification results, velocity status, frequency freshness, and original data index, and then construct a trusted load sequence by cycle number.

[0091] In S23, a symbol segment is a time-series object formed by merging consecutive records in the same direction, containing a start point, end point, direction, and continuous length. The system first uses a pressure sorting position dead zone of 0.01 to eliminate minute changes within the sensor's repeatability range, and then records changes in adjacent cycles as upward, downward, or hold; when upward and downward movements repeatedly occur within a short period and meet the condition for direction reversal, it is recorded as direction alternation. Consecutive records in the same direction are only merged within the same valid continuous time period and cannot cross data missing, turning, or tick confirmation sections. The symbol segment formed in this way retains the direction of change and continuity, rather than treating each sampling point as an independent load event.

[0092] The short window length and the number of main directions within the short window are determined using 20 continuous load records and 20 short-duration impact records. The system compares candidate windows with 8, 10, and 12 processing cycles, and compares the number of records in the same direction with different proportions within each window, using confirmation delay and false alarms as evaluation criteria. In the example, 4 false alarms were observed with 8 cycles, an average increase of 0.2 seconds in confirmation time was observed with 12 cycles, and 1 false alarm was observed with no missed detections due to continuous load when using 10 cycles and at least 8 records in the same direction. Therefore, this combination is selected as the short window and its main direction condition.

[0093] The length of the long window and the number of background main directions were determined using the slow-changing conditions portion of the same validation set. The system compared the coverage capabilities of 40, 60, and 80 processing cycles for slowly changing mud layers, with the conditions of not crossing bends and being able to distinguish between short-term upward movement and continuous background. In the example, the 40-cycle test misclassified slow loading as short-term impacts three times, and the 80-cycle test had two instances of windows crossing condition switches. The 60-cycle test, with at least 31 records in the same direction, did not have these problems and was therefore determined as the long window condition. When the current long window has 39 upward movement records, the background main direction is recorded as upward movement.

[0094] The number of directional alternations was determined using 30 stable windows and 20 oscillation windows. After applying the measurement dead zone, the system tested conditions for 3, 4, and 5 alternations of upward and downward movement within the short window. Using 3 alternations, 3 stable windows were mistakenly recorded as alternating; using 5 alternations, 2 true oscillation windows were missed; and using 4 alternations, only 1 unconfirmed record was retained. Therefore, at least 4 alternations were determined as the directional alternation condition. A continuous upward movement background was defined when there were 8 upward movements in the current short window and 39 upward movements in the long window; a short upward movement was defined as a short-term upward movement when the short window moved upwards while the long window remained mainly stationary.

[0095] In S24, the direction status code is generated based on the symbol segment direction and continuous length, including continuous upward movement, continuous downward movement, hold, direction alternation, and recovery return. The conditions for recovery return are determined through 30 short-term impact recovery records: the direction status code changes to recovery return only when the previous symbol segment has continuously moved upward or downward, the current reverse symbol segment forms and re-enters the reference stable interval, and remains within this interval for at least two consecutive processing cycles. In the example, when the pressure sorting position moves upward for five consecutive cycles, then downward for three consecutive cycles, and returns to the reference stable interval in the last two cycles, it is recorded as recovery return; if it falls back into the interval for only one cycle, it is still recorded as pending confirmation to avoid single-point fall triggering recovery.

[0096] The amplitude status code forms mutually exclusive branches based on the endpoint relationship between the current extreme value interval and the reference extreme value interval. When the lower end of the current interval is higher than the upper end of the reference interval, it is recorded as an overall upward shift; when the upper end of the current interval is higher than the upper end of the reference interval and the remaining parts overlap, it is recorded as a local upward shift; when both ends of the current interval are within the reference interval, it is recorded as an intra-reference fluctuation; when the lower end of the current interval is lower than the lower end of the reference interval but the remaining parts overlap, it is recorded as a local downward shift; when the upper end of the current interval is lower than the lower end of the reference interval, it is recorded as an overall downward shift. For example, when the reference interval is 0.12 to 0.82, the current intervals 0.86 to 0.91, 0.78 to 0.88, 0.2 to 0.8, 0.08 to 0.2, and 0.02 to 0.1 respectively enter the above five branches, thus avoiding the problem of uncovered endpoint relationships.

[0097] The recovery status code is generated based on the duration of the out-of-bounds movement relative to the reference stable interval, the number of sampling points required for return, and the duration of stability after return. It includes fast return, slow return, and insufficient recovery. The relevant boundaries are obtained by comparing 30 short-duration impacts with 25 continuous load records. A fast return is defined as an out-of-bounds movement of no more than 2 cycles, returning within 3 cycles, and stabilizing for at least 10 cycles after return. A slow return is defined as an out-of-bounds movement of 3 to 5 cycles, or a return requiring 4 to 5 cycles, but stabilizing for at least 10 cycles after return. Insufficient recovery is defined as an out-of-bounds movement of 6 cycles, a return exceeding 5 cycles, or stabilizing for less than 10 cycles after return. For example, three sets of records are generated for out-of-bounds movement of 2 cycles, a return of 3 cycles, and stabilization for 10 cycles; out-of-bounds movement of 4 cycles, a return of 5 cycles, and stabilization for 10 cycles; and out-of-bounds movement of 6 cycles, a return of 5 cycles, and stabilization for 7 cycles, respectively.

[0098] The execution status code only uses the most recent control action from the response waiting cycle that was completed before the current processing cycle, and does not use feedback that has not yet been generated in S4 of the current cycle. Insufficient execution change is generated when the difference between the residuals before and after execution is within the near-range of residuals; valid execution is generated when the decrease in residuals after execution exceeds the near-range and the relative improvement meets the requirements for effective improvement; execution direction mismatch is generated when the residuals after execution increase and exceed the near-range, or when the execution readback direction is reversed. In the example, under the same objective, a decrease in residuals from 50 r / min to 10 r / min is recorded as valid execution, a decrease from 50 r / min to 45 r / min as insufficient execution change, and an increase from 50 r / min to 70 r / min as execution direction mismatch, thus avoiding future feedback from participating in the current status code.

[0099] The recovery status code is generated in this step for the reference stability interval formed by the effective stable reference window, while the boundary crossing return data of S3 is generated for the current dynamic load boundary. Since the two objects are different, there is no loop about whether the status code or the dynamic boundary comes first. The processing cycle status codes are always written in a fixed order: direction status code, amplitude status code, recovery status code, and execution status code. For example, continuous upward movement, overall upward movement, insufficient recovery, and effective execution are combined into a single status record in this order, without concatenating values ​​of different dimensions.

[0100] The window length of the current pressure ranking extreme value interval is determined by recording 20 short-term impacts and 20 continuous load increases. The system compares the minimum and maximum values ​​of the most recent 6, 10, and 14 processing cycles. With 6 cycles, 4 single-point impacts are recorded as overall offsets; with 14 cycles, the average delay for continuous load confirmation is 0.3 seconds; and with 10 cycles, only 1 instance awaits confirmation and there are no missed continuous load increases. Therefore, the most recent 10 processing cycles are used. For example, the minimum and maximum values ​​of the current window are 0.86 and 0.91, respectively, which form the current extreme value interval.

[0101] The quantile combinations of the reference extreme value intervals are determined through independent stability verification windows. After the stress sorting sequence is frozen, the system collects 20 valid stability verification windows that have passed integrity, continuous out-of-bounds, direction reversal, and readback consistency checks. Each position is obtained relative to the same frozen sequence, and then the two candidate groups of 5% to 95% and 10% to 90% are compared. Although the latter combination has one less false alarm in 30 short-term impact replays, it has three more missed detections in 25 sustained loads, so the former combination is adopted; the example endpoints are 0.12 and 0.82. This process does not make clustering inferences about the rank distribution of the reference data itself to avoid statistical self-reference.

[0102] like Figure 4 As shown, in one embodiment, S3 includes:

[0103] S31. Generate three basic sorting position boundaries based on historical data with load status, fuel consumption per unit working area, seedling frequency and working quality markers. Initialize the dynamic load boundary with the basic sorting position boundaries and generate the theoretical response stage sequence.

[0104] S32. Read the most recent engine target speed segment and hydraulic variable pump target displacement segment records that were output before the current processing cycle and whose instruction age has reached the response waiting period. Read the corresponding engine output speed data and hydraulic system working pressure data. Extract the actual response stage sequence. Compare the theoretical response stage sequence and the actual response stage sequence item by item. Generate boundary back calibration results based on stage integrity, stage sequence and residual changes after execution.

[0105] S33. Based on the dynamic load boundary, generate the basic load partition candidate, the load partition candidate after the cycle time correction, the load partition candidate after the power coverage correction, and the current load partition candidate after the speed deviation correction in sequence.

[0106] S34. Record the location of the first crossing of the dynamic load boundary by the current load partition candidate, the duration of the crossing, the number of sampling points required to return to the stable interval, and the duration of the stable interval after the return, to form boundary crossing return data and determine the current load partition.

[0107] In this embodiment, it should be noted that in S31, the historical load status is generated by independent instrumentation calibration, without using the pressure sorting position boundary to be optimized for reverse generation. The system sets different mud resistance and operating depths in at least five fields, collecting flywheel output power, pump shaft torque, travel speed, single-row transplanting frequency, and quality inspection results. The flywheel output power relative to the available power at the same rotational speed is used as the primary label, and verified by the direction of pump shaft torque change. For example, after sorting 300 instrumentation records by power occupancy, intervals that can stably separate four groups of records are formed around 0.3, 0.6, and 0.8, with a cross-field boundary variation not exceeding 0.02, thus sequentially labeling them as low load, low-medium load, medium-high load, and high load; records with a power occupancy of 0.7 and consistent pump shaft torque direction are labeled as medium-high load. Production operation data is only matched to verified operating condition units, without redefining historical labels, thereby avoiding cyclic labeling.

[0108] The quality indicators for the operation are generated jointly by the applicable test methods and the agronomic task documents, and cannot be inferred solely from frequency and speed. The system extracts continuous planting samples according to the task specifications for each historical valid operation segment, recording missed planting, damaged seedlings, floating seedlings, lodging, planting depth, and uniformity of planting spacing. Any indicator exceeding the task's allowable value is marked as unqualified. For example, the task document specifies upper limits of 5.0% for missed planting rate, 5.0% for damaged seedlings, 3.0% for floating seedlings, and 3.0% for lodging rate, respectively; the allowable planting depth ranges from 15mm to 25mm; and at least 95 out of 100 inspection holes must have a planting spacing between 0.18m and 0.22m. In a certain work section, 100 holes were inspected and 4 holes were found to be missing, 3 holes were damaged, 2 holes were floating, and 1 hole was overturned. Since the total sample size was 100 holes, the corresponding proportions were 4.0%, 3.0%, 2.0%, and 1.0% respectively. The depth of each hole was within the allowable range and 96 holes were of acceptable spacing. Therefore, a quality qualified mark was generated. The above values ​​are only examples in the task document and do not replace the limits of specific contracts and applicable standards.

[0109] The three basic ranking position boundaries are not directly taken as fixed quantiles. The system pre-collects valid operational data from no fewer than five fields of the same machine model and sensor calibration version. First, historical data is screened into a valid and stable reference window according to rule S12. Then, the samples within the window are labeled with the aforementioned load status, fuel consumption per unit operating area, seedling needle frequency, and operational quality. The pressure ranking position of each sample is obtained relative to the calibration pressure sequence frozen before the occurrence of that sample; the distribution is not inferred after self-ranking the same batch of calibration samples. After excluding records of turning, malfunctions, non-planting, and inconsistent execution readbacks, a limited traversal is performed on the candidate boundary combinations. Combinations with fewer than 20 samples in any partition are excluded. Then, the dispersion of fuel consumption, frequency dispersion, number of mixed quality non-compliant records, and number of erroneous switches within each partition are compared, and combinations with stable verification performance are selected. For example, if the sample screening results in 300 calibration records, the three basic boundaries are 0.28, 0.58, and 0.84, respectively.

[0110] The basic boundary candidate set is generated from the labeled samples themselves. The system arranges 300 calibration records in ascending order of stress ranking. Whenever two adjacent records have different independent load states, the median position between the two stress ranking positions is taken as the original candidate. When the distance between the original candidates does not exceed the stress ranking dead zone of 0.01, they are merged into one group, and the median position within the group is taken. The example data generates 18, 22, and 15 original candidates in the three label transition directions, respectively. After merging, 7, 8, and 6 candidate positions are retained, respectively. The system only combines candidates with increasing order, adjacent intervals of not less than 0.1, and at least 20 records in each partition, resulting in 36 verifiable combinations. Thus, the input set, generation order, and exclusion conditions for finite traversal can all be reproduced from the calibration records.

[0111] The number of plots and the minimum sample size per partition were determined through cross-plot holdout validation. The system established boundaries using 3, 5, and 7 plots respectively, and used plots not involved in modeling as validation data. When using 3 plots, the maximum variation of the third boundary among different holdout groups was 0.05; after using 5 plots, the maximum variation decreased to 0.02; and when increasing to 7 plots, the improvement was less than 0.01, therefore a minimum of 5 plots was required. The system also compared the results with at least 10, 20, and 30 records per partition. When using 10 records, there were 4 instances of paired sort flips; after using 20 records, there was only 1 instance; and increasing to 30 records did not change the boundary further. Therefore, the example used 20 records as the minimum sample size.

[0112] The selection of multiple indicators for the basic boundary adopts a fixed priority order. The system first excludes combinations with substandard quality records, violations of flow or power constraints, and insufficient samples in any zone. Then, it compares the number of erroneous switches. If the number of erroneous switches is the same, it compares the dispersion of fuel consumption per unit operating area and the dispersion of seedling needle frequency in that order. If they are still tied, the conservative boundary farther from the protection zone is selected. For example, the sample numbers for the four zones of candidate combinations 0.25, 0.55, and 0.82 are 55, 80, 85, and 80 respectively, with 7 erroneous switches and a fuel consumption dispersion of 0.6 kg / hm². 2 The sample sizes for candidate combinations 0.28, 0.58, and 0.84 were 65, 78, 82, and 75 respectively, with 4 incorrect switching occurrences and a dispersion of 0.4 kg / hm². 2 Therefore, the latter combination was chosen.

[0113] Fuel consumption per unit operating area is recorded by measuring fuel consumption mass, operating width, and effective travel distance within the same effective operating section, and then normalized to kg / hm². 2 Save; if the engine controller only provides the cumulative fuel injection quantity, first convert it to mass according to the manufacturer's fuel injection quantity calibration. Example: The working width is 1.8m and the effective travel distance is 1000m; multiplying these two gives a working area of ​​1800m². 2 After conversion, it is 0.18hm 2 The fuel mass for this section is 0.86 kg. Dividing the fuel mass by the working area and rounding to one decimal place yields 4.8 kg / hm². 2 Each boundary candidate is compared only with records under the same machine type, width, field conditions, and quality compliance conditions to avoid the influence of speed and width on fuel consumption per unit operating time. The example group contains 40 records, with a median value of 4.8 kg / hm. 2 The median value for the other pair in the same group was 5.3 kg / hm. 2 Therefore, the previous pair is entered into the energy efficiency mapping table.

[0114] The theoretical response phase sequence is generated based on records of engine target speed changes and historical effective response records that have occurred before the current processing cycle, including response initiation, response enhancement, and response stabilization. Response initiation is the starting point where the speed residual first decreases for two consecutive cycles after consistent readback; response enhancement is the starting point where the rate of decrease in residual per unit time enters the middle of the historical effective response and remains there for two consecutive cycles; response stabilization is the first point where the actual speed enters the same target speed range and remains there for five consecutive processing cycles. The number of consecutive cycles is determined through verification using 20 effective responses and 10 noise disturbance records, selecting the minimum number that can eliminate single-point fluctuations and does not significantly delay control confirmation. The theoretical phase is built using only historical feedback that has already occurred, and does not presuppose results that have not yet been executed.

[0115] The historical effective response midpoints are obtained by sorting the residual reduction rates. The system calculates the residual reduction rate for each adjacent processing cycle in 20 effective responses, excludes protection actions and inconsistent readback records, and selects the 25th to 75th percentiles to form the midpoints; in the example, the midpoints are 40 r / min / s to 120 r / min / s. In a record with a processing cycle of 0.1s and a target of 1900 r / min, the residuals are 100 r / min, 98 r / min, 94 r / min, 84 r / min, 74 r / min, 64 r / min, and 0 r / min, and the adjacent reduction rates are 20 r / min / s, 40 r / min / s, 100 r / min / s, 100 r / min / s, 100 r / min / s, and 640 r / min / s, respectively. The first point after a continuous descent is confirmed as the start of the response. The first point of entering the middle section after reducing the speed for two consecutive cycles is considered as the response enhancement. The first point of maintaining a residual of 0 r / min for five consecutive cycles is considered as the response stabilization. Therefore, the example satisfies the sequential conditions of the three stages.

[0116] The historical pressure response direction for the same feasible pair is extracted from the pressure records of completed actions. The system selects at least 20 historical actions with consistent readback, no pressure protection trigger, and effective speed response. The pressure sequence is aligned with the target issuance time. A 0.1 MPa pressure dead zone is used to record minor changes as sustained. The percentage of rising, sustained, and falling records during the response from start to stabilization is then calculated. The system compares the 60%, 70%, and 80% directional consistency requirements. At 60%, two pairs show mislabeling of opposite directions; at 80%, five valid pairs cannot form a direction; and at 70%, there are no mislabelings and only one pair awaiting confirmation. Therefore, 70% is adopted. In the example pair, 15 of the 20 records show pressure rises, 3 show sustained pressures, and 2 show falls. Therefore, the historical pressure rise direction is saved. If the consistency requirement is not met, the actual response stage is not confirmed by pressure direction.

[0117] Furthermore, in S32, the system extracts the actual response start point, actual response enhancement point, and actual response stability point from the most recent target strategy record where the command age has reached the response waiting period, and forms the actual response phase sequence. At each phase point, the engine speed residual condition and the direction of hydraulic system working pressure change are checked simultaneously. The pressure direction should be consistent with the direction of the same feasible pair in the historical effective response. Data during the pressure protection period is not used for phase confirmation. If the actual phases are complete, sequential, and the residuals decrease after execution, the dynamic load boundary is maintained. If the actual phases are complete but the overall starting points are later than the 95th percentile time of the historical effective response records, the subsequent response waiting period is extended. If the upper-level load partition is triggered continuously and the residuals do not improve after execution, the boundary is frozen first, and the engine target readback, pump control current, actual pressure, speed, and frequency are checked. Only when the pressure is lower than the continuous operation early protection trigger pressure, the cycle time is normal, the speed is normal, and three consecutive events are confirmed as false positives for the partition, is the boundary adjusted to the side away from the current pressure ranking position. If the upper-level partition is not triggered but subsequent pressure protection, continuous speed decrease, or continuous frequency deviation occurs, the boundary is adjusted to the side closer to the current pressure ranking position. The number of three consecutive confirmations comes from 18 execution link anomalies and 20 real false positive replays, which can prevent a single execution failure from pushing the boundary outward.

[0118] The pressure direction dead zone and lag criterion during the actual response phase are determined through synchronous bench recording. Pressure hold is defined as the absolute value of adjacent changes in initial pressure not exceeding 0.1 MPa. This dead zone is derived from the 95th quantile of 100 constant-pressure repeated samplings, rounded up. Only after exceeding the dead zone is the response direction compared to historical valid responses. For example, the response start, enhancement, and stabilization times at the historical 95th quantile are 0.6 s, 0.8 s, and 1.2 s, respectively. The actual response phase occurs at 0.7 s, 0.9 s, and 1.4 s, respectively, with consistent pressure directions. Therefore, this is considered a complete phase but overall lag. If only the start and enhancement points are present without five consecutive stable cycles, the phase is considered missing, and boundary shifting based on this is not permitted.

[0119] The boundary backcalibration results can be reproduced through four sets of feedback records. When the current third boundary is 0.84, the actual stage is complete and the residual decreases from 80 r / min to 20 r / min, so the output maintains the boundary. When the actual stage times are 0.7s, 0.9s, and 1.4s respectively, which are all later than the corresponding historical quantiles, the output is extended and waits. When three consecutive events meet the conditions of pressure below 15.5MPa, normal frequency and speed, and residual increasing from 50 r / min to 55 r / min, and are confirmed by playback to be false positives, the boundary can be moved outward to 0.86. When the upper-level partition is not triggered but the pressure reaches 15.5MPa or the speed and frequency continue to deteriorate, the boundary is moved inward to 0.82. Any execution link failure will only freeze the boundary and enter protection, and will not be used as a basis for outward movement.

[0120] The boundary adjustment step size was determined through a limited search using 30 sets of replay data. The system tested sorting position step sizes of 0.01, 0.02, and 0.03, evaluating the number of missed detections, the number of erroneous switches, and the stabilization time after adjustment. A step size of 0.01 required an average of 3 windows to stabilize, a step size of 0.03 resulted in 4 cross-zone erroneous switches, and a step size of 0.02 stabilized within 2 windows and generated only 1 pending confirmation record. Therefore, 0.02 was selected for the example. This step size is only used in boundary calibration that meets safety requirements and is not applied to the maximum permissible continuous operating pressure, overflow protection settings, or absolute safety pressure limits.

[0121] The minimum interval between adjacent boundaries is determined through partitioned sample coverage verification. The system tests candidate intervals of 0.08, 0.1, and 0.12 respectively, and checks whether each partition still retains at least 20 verification samples after adjustment. At an interval of 0.08, two partitions switch back and forth during playback; at an interval of 0.12, the middle partition has one instance of insufficient samples; at an interval of 0.1, all four partitions meet the sample size requirement and have the lowest erroneous switching. Therefore, an interval of 0.1 is selected for the example. After each adjustment, the boundary order and the mutual exclusion relationship of the four intervals are rechecked. If the minimum interval is violated, the original boundaries are maintained and the system is put into observation.

[0122] The maximum number of adjustments per valid window is determined through continuous noise playback. The system compares the results of allowing 1, 2, and 3 adjustments per 5-second window on 20 feedback noise records. Allowing 2 or 3 adjustments resulted in 3 and 6 repeated adjustments within the same window, respectively, while allowing 1 adjustment resulted in no such repeated adjustments. Therefore, each valid window can only be adjusted a maximum of 1 time. It should be noted that this parameter limits a single valid window, not the entire operation phase; subsequent windows must still meet the requirements of a complete response, false positive confirmation, and safety check before adjustments can be made.

[0123] The cumulative adjustment envelope of the dynamic boundary is determined through 30 sets of cross-window replays. The system compares candidate envelopes with cumulative allowable variations of 0.04, 0.06, and 0.08 relative to the basic boundary. With 0.04, calibration fails in 3 slow-changing conditions; with 0.08, boundary approaches adjacent partitions in 2 cases; and with 0.06, there are no missed detections and the minimum interval is maintained. Therefore, the example limits the cumulative variation of each basic boundary to ±0.06. Each dynamic boundary must also be within a normalized effective domain greater than 0 and less than 1. If any condition of boundary order, minimum interval, effective domain, or cumulative envelope is violated after adjustment, the original boundary is maintained and the system is moved to observation.

[0124] In S33, the basic load zoning candidates are obtained from the pressure ranking position and three dynamic load boundaries: low load candidates are generated when the load is below the first boundary, medium-low load candidates are generated when the load is between the first and second boundaries, medium-high load candidates are generated when the load is between the second and third boundaries, and high load candidates are generated when the load reaches or exceeds the third boundary. Subsequently, in a fixed order, cycle pressure verification results are used to generate cycle correction candidates, power coverage verification results are used to generate power coverage correction candidates, and speed deviation is used to generate current load zoning candidates. This order ensures that unilateral frequency anomalies do not increase the current load zoning, power demand is not amplified when power coverage is invalid, and protection judgment is prioritized when severe speed drop is accompanied by high absolute pressure.

[0125] For example, a pressure ranking position of 0.91 is higher than the third dynamic boundary of 0.84, forming a high-load baseline candidate. The clockwise pressure result, with pressure shifting upwards first and frequency then downwards, supports this candidate. Effective power coverage indicates that the engine is following the target, and the negative speed deviation of 0.04 m / s is a slight negative deviation. Therefore, the current load partition candidate remains at a high load. If the same pressure ranking position is accompanied by abnormal observations on the seedling side while the speed is normal, the load will not be increased during the clockwise correction phase. If power coverage is invalid and execution readback is inconsistent, the current record will enter the execution lag observation phase, and the same-direction strategy amplification will be frozen. Each correction only uses the data generated in the previous steps and does not write back to modify the original samples.

[0126] In S34, the system records the position where the current load partition candidate first crosses the dynamic load boundary, counts the number of consecutive cycles on the boundary side as the boundary duration, counts the number of cycles required from the last boundary cycle to return to the stable interval as the number of sampling points required to return to the stable interval, and counts the number of consecutive cycles maintained after returning as the stable duration after returning. The judgment parameters for fast return, continuous boundary crossing, slow return, and insufficient recovery are derived from the aforementioned short-term impact and continuous load increase verification set. For example, if the boundary crossing does not exceed 2 cycles and returns and stabilizes for 10 cycles within 3 cycles, the partition of the previous cycle is maintained; if the boundary crossing reaches 6 cycles, returns for more than 3 cycles, or stabilizes for less than 10 cycles after returning, the partition is determined according to the current candidate. When the boundary crossing return data indicates a short-term boundary crossing and a fast return, the load partition of the previous processing cycle is maintained; when the boundary crossing return data indicates a continuous boundary crossing, a slow return, or insufficient recovery, the current load partition is determined according to the current load partition candidate.

[0127] The current load zones include mutually exclusive low-load, medium-low-load, medium-high-load, and high-load zones. Only one zone can be assigned to each processing cycle. The maximum permissible continuous operating pressure, continuous operation advance protection trigger pressure, independent overflow protection setting, and absolute safety pressure limit are all independent of the four zones and do not constitute a fifth load zone. The absolute safety pressure limit is the minimum short-time peak permissible pressure of the pump, valve, pipeline, and pressure sensor as confirmed by the manufacturer, and cannot be replaced by the maximum permissible continuous operating pressure. The independent overflow protection setting must be lower than this absolute safety pressure limit, and dynamic load boundaries must not replace any hardware or software protection boundaries.

[0128] The pressure sensor range for the example model is 0 to 25.0 MPa, and the maximum permissible continuous operating pressure given in the hydraulic datasheet is 16.0 MPa. The short-time peak permissible pressures for the pump, valves, pipelines, and pressure sensor are 20.0 MPa, 21.0 MPa, 22.0 MPa, and 25.0 MPa, respectively. The minimum value is taken as the absolute safe pressure limit of 20.0 MPa, and the independent overflow protection is set to 18.0 MPa, which is lower than this limit. The maximum measurement deviation from 30 pressure calibration records is rounded up to 0.2 MPa, and the maximum pressure increment from trigger to execution in 20 protection action tests is rounded up to 0.3 MPa. After subtracting two margins from the maximum permissible continuous operating pressure, the continuous operation advance protection trigger pressure of 15.5 MPa is obtained. When the pressure reaches 15.5 MPa, the pump is immediately limited, the speed is reduced, or the load is unloaded. When the pressure reaches 16.0 MPa, continuous load increase is prohibited, and the hardware overflow and absolute limit continue to operate independently according to the manufacturer's logic.

[0129] Pressure signal failure is handled using fail-safe procedures. The system generates a pressure validity flag based on sensor diagnostic positions, range checks, and consecutive missing data counts. If pressure data is missing for two processing cycles, exceeds the 25.0 MPa range, or a diagnostic fault occurs, increasing engine speed and variable pump displacement is prohibited, dynamic load boundaries are frozen, and the system enters manufacturer's safety mode. In the example, if there are no pressure updates for two consecutive cycles, even if the most recent pressure ranking position is only 0.6, the system will not continue to amplify power actions. Pressure ranking positions are only used for energy efficiency scheduling and must not replace mechanical relief valves or manufacturer's safety controls.

[0130] The number of consecutive missing stress data confirmations was determined using limited communication delay records and actual disconnection records. The system replayed 20 communication delay records showing only a single missing frame that recovered in the next cycle, and 15 records of continuous disconnection. Candidate conditions for failure to update for 1, 2, and 3 consecutive processing cycles were compared, with the number of times normal data mistakenly entered safe mode and the time to resume loading after a disconnection used as evaluation criteria. With 1 cycle, 4 single-frame delays were mistakenly identified as faults; with 3 cycles, the delay after a real disconnection was 0.2 seconds longer than candidate 1; with 2 cycles, no continuous disconnections were missed, and only 1 record awaiting confirmation was retained. Therefore, this example defines consecutive missing data reaching 2 processing cycles as a stress failure confirmation condition.

[0131] like Figure 5 As shown, in one embodiment, S4 includes:

[0132] S41. Determine the observation priority, first-level control priority, second-level control priority, or protection priority based on the current load zoning, processing cycle status code, and boundary crossing return data. Query the energy efficiency mapping table to obtain the candidate engine target speed range, candidate hydraulic variable pump target displacement range, and fuel consumption records per unit operating area for paired constraints.

[0133] S42. Verify candidate pairs based on pump allowable continuous flow rate, pump shaft input power, and engine available power, and select and output specific target parameters from the verified pairs.

[0134] S43. Before and after the execution of the target power output strategy, obtain the residual before execution and the residual after execution based on the same specific target parameter, and maintain, reduce or cancel subsequent control actions in the same direction according to the changes in the residual.

[0135] S44. Under the condition of multi-machine collaborative operation, invalid unit broadcast data is screened out to form a unit load difference sequence, and the machine operation speed adjustment command is output within the allowable conditions of rice transplanting rhythm and the overall machine limit.

[0136] In this embodiment, it should be noted that in S41, the mechanical relief valve always operates independently and in parallel according to the 18.0MPa setting, does not participate in the software priority mutual exclusion selection, and is not masked by any load partitioning results. The software first checks for pressure sensor faults; if the signal is valid, it then checks the absolute safety pressure limit of 20.0MPa, the independent relief protection linkage value of 18.0MPa, the maximum allowable continuous working pressure of 16.0MPa, and the continuous operation early protection trigger pressure of 15.5MPa in sequence according to the severity of the risk. When the absolute safety pressure limit is reached, the manufacturer's specified emergency unloading and shutdown interlock are executed; when the relief linkage value is reached, software unloading is executed simultaneously with mechanical relief; when the maximum allowable continuous working pressure is reached, continuous load increase is prohibited and pumps are limited or unloaded; when the early protection trigger pressure is reached, load increase is first frozen and pumps are limited, and if necessary, the speed is reduced according to the overall machine limit. When any of the above safety conditions are triggered, the software uniformly outputs the protection priority and overrides the energy efficiency scheduling results.

[0137] If no pressure or sensor safety conditions are triggered, the system selects a software result according to the following priority levels: protection priority, secondary control priority, primary control priority, and observation priority. Protection priority also applies to situations where speed drops significantly, frequency deviates continuously, or execution direction is incompatible; the action is to limit speed, limit pump speed, or unload and prohibit amplification of actions in the same direction. Observation priority applies to low or medium-low loads where the status code shows hold or rapid return; the action is to maintain the original strategy. Primary control priority applies to medium-low or medium-high loads where the status code shows continuous deviation but execution is effective; the action is to select an adjacent zone for pairing. Secondary control priority applies to medium-high or high loads where the status code shows continuous deviation or insufficient recovery; the action is to select the higher power pair within the current zone that meets the operational requirements.

[0138] Priority conditions and conflict sequence are determined through 10 short-term disturbances, 10 continuous load increases, 10 execution anomalies, and 10 pressure protection replays, with evaluation based on not bypassing protection, reducing invalid switching, and maintaining cycle time. For example, when the pressure replay is 15.6 MPa, the early protection branch is entered; when the pressure replay reaches 16.0 MPa, continuous load increases are prohibited; when the pressure replay reaches 18.0 MPa, mechanical overflow and software unloading occur in parallel; and when the pressure replay reaches 20.0 MPa, emergency interlocking is executed. If no safety conditions are triggered, pressure position 0.91, high load, and insufficient recovery enters the secondary control priority; pressure position 0.7, medium-high load, continuous upward movement, and effective execution enters the primary control priority; and pressure position 0.2, low load, and rapid return enters the observation priority. This example illustrates that safety protection always takes precedence over energy efficiency actions.

[0139] The minimum sample size and statistics for the energy efficiency mapping table are determined through hold-out validation. The system groups valid records of the same model and calibration version by partition, pressure sorting position segment, speed deviation segment, and quality marker. The stability of pair selection is tested when each group has at least 20, 30, and 40 records. Once the sample size reaches 40 records, the pairing sorting on the hold-out data no longer changes frequently; therefore, the minimum sample size for the example is 40 records. Each pair stores the engine speed segment, displacement segment, median fuel consumption per unit operating area, proportion of frequency-stable records, and the operating quality qualification marker. When the sample size is insufficient, conservative pairings based on adjacent calibrated operating conditions are reverted; no new strategies are created based on a single low fuel consumption record.

[0140] The target engine speed range and the target displacement range of the hydraulic variable pump are formed by grouping historical control gears and continuous target values. The system first aggregates target records according to the same control gear, then merges records with adjacent speed differences not exceeding 50 r / min and displacement differences not exceeding 1.0 mL / r, and then performs flow rate, power, cycle time, and operation quality verification. Pairs that pass the verification and have pump shaft input power higher than adjacent low-load units are called higher power feasible pairs. In the example, 40 records are retained for each of the 4 adjacent target points, for a total of 160 records, which are merged to obtain the speed range of 1900 r / min to 2050 r / min and its paired displacement range of 20 mL / r to 22 mL / r. When the sample is insufficient, conservative pairing selects a pair with lower flow rate and power and qualified quality from adjacent calibrated units with lower pressure positions, without independently splicing the interval endpoints.

[0141] The target record merging threshold is determined jointly by controlling resolution and the stability of the retained records. The system compares 30 sets of records with the same load gear at speed differences of 25 r / min, 50 r / min, and 75 r / min, and also compares displacement differences of 0.5 mL / r, 1.0 mL / r, and 1.5 mL / r. The minimum combination breaks down the same continuous gear into 6 excessively small units, while the maximum combination merges 3 sets of records with different fuel consumption and cadence performance. The threshold is adopted because the paired sorting of retained records at 50 r / min and 1.0 mL / r is stable and each set reaches the minimum sample size. This merging only forms candidate segments and cannot be used to skip subsequent physical feasibility checks.

[0142] Furthermore, in S42, the target speed range of the candidate engine and the target displacement range of the candidate hydraulic variable pump can only be selected from the same feasible pairing record. The permissible continuous flow rate is determined through equipment data and high-temperature thermal balance tests: the rated continuous flow rate given in the manufacturer's pump data is 48.0 L / min, the permissible flow rate given in the valve assembly data is 45.0 L / min, and the upper limit of the flow rate required to meet oil temperature stability in 10 high-temperature continuous operation tests is 40.0 L / min. The minimum value among these three is taken, therefore the example permissible continuous flow rate is 40.0 L / min. The system sequentially multiplies the pump displacement, the transmission ratio of the pump shaft speed to the engine shaft speed, the engine speed, and the pump volumetric efficiency, then reduces the value by a factor of 1000 based on the displacement and flow rate units to obtain the estimated flow rate; the working pressure is multiplied by the estimated flow rate and reduced by a factor of 60 based on the minutes and seconds units to obtain the hydraulic output power; finally, the hydraulic output power is divided by the overall efficiency of the variable pump to obtain the pump shaft input power.

[0143] The transmission ratio, volumetric efficiency, and overall efficiency of the variable pump are all derived from equipment calibration and are not directly preset. The transmission ratio is defined as the ratio of the pump shaft speed to the engine shaft speed, determined by recording the synchronous speeds of the engine shaft and pump shaft. The results of 10 stable points are concentrated at 0.9, so 0.9 is used in the example. The volumetric efficiency is obtained by comparing the measured flow rate with the theoretical displacement flow rate at 5 speed points and 3 times at each point. The median value under the current operating conditions is 0.9. The overall efficiency of the variable pump is obtained by bench recording of pump shaft torque, pump shaft speed, outlet pressure, and measured flow rate. The example value is 0.84. The available engine power is derived from the flywheel end available record after deducting the necessary internal consumption of the engine from the engine mapping. Other load power is added separately from the engine-side power balance test, including the transmission loss between the engine flywheel and the pump shaft that has not yet been included in the pump shaft input power, as well as the travel, mechanical insertion, and electrical auxiliary loads that have not yet been deducted from the available engine power. The insertion hydraulic power powered by the variable pump should not be included again.

[0144] The engine's available power, other load power, and transient power margin are obtained through bench power balancing at the same speed and operating mode. For example, a stable flywheel power of 16.0 kW was measured at 1900 r / min. After deducting the engine cooling, charging, and internal necessary consumption recorded by the controller (1.5 kW), the available power is 14.5 kW. Pump drive losses, travel load, mechanical insertion load, and electrical auxiliary load, not yet factored in pump shaft input power, are 0.4 kW, 1.2 kW, 0.9 kW, and 0.5 kW respectively. Adding these together yields an other load power of 3.0 kW. The maximum unmodeled transient increment from 20 acceleration tests is 0.8 kW, rounded up to retain a 1.0 kW margin. The aforementioned 3.0 kW is only used for pairing within the same mode; other speeds or modes require recalibration.

[0145] Feasibility verification covers the entire transition range from the current speed to the target speed, rather than just checking the target steady-state point. The pressure used is the more stringent value between the upper limit of the candidate operating range and the continuous operation early protection trigger pressure. The variable pump overall efficiency uses the lower verification value under the corresponding oil temperature, pressure, and displacement conditions. The available power of the engine mapped at 1850 r / min, 1875 r / min, and 1900 r / min is 14.0 kW, 14.2 kW, and 14.5 kW, respectively. Other loads recorded in the power balance of the same mode are 2.8 kW, 2.9 kW, and 3.0 kW, respectively. During verification, the lowest available power in the range and the upper limit of other loads in the range are used, respectively. A conservative check is performed between the calibration grids using lower available power and higher load; if any point is not satisfied, the pairing is not output.

[0146] The transition control first increases the engine speed while maintaining or decreasing the pump displacement, and then ramps up the displacement once the engine speed reaches the target range. For example, the current equivalent displacement readback is 18.0 mL / r, and this displacement is maintained during the increase from 1850 r / min to 1900 r / min. Checking with the upper limit of the stage (1900 r / min), a more stringent pressure of 15.5 MPa, a transmission ratio of 0.9, a volumetric efficiency of 0.9, and a combined efficiency of 0.84, the maximum estimated flow rate for this stage is 27.7 L / min, the internally unrounded hydraulic output power is 7.16 kW, and the pump shaft input power is 8.52 kW. Adding the upper limit of other loads (3.0 kW) and a transient margin of 1.0 kW, the total is 12.52 kW, which is lower than the minimum available power of 14.0 kW in the transition range. The displacement ramp was adjusted to 20 mL / r only after the speed reached 1900 r / min. The steady-state point of the candidate target at the current pressure of 13.8 MPa yielded a flow rate of 30.8 L / min, an internally unrounded hydraulic output power of 7.08 kW, and a pump shaft input power of 8.43 kW. After verification at a more stringent pressure of 15.5 MPa, the values ​​were 7.95 kW and 9.47 kW, respectively. After adding other loads and transient margins, the result was 13.47 kW, which is lower than the available power of 14.5 kW at the target point. Furthermore, the flow rates in both stages were below 40.0 L / min. Therefore, the complete transition process was feasible.

[0147] The current points and number of repetitions for the control current-displacement calibration table were determined through coverage and interpolation error verification. The manufacturer's control data specifies a continuous permissible control current range of 0.4A to 1.2A for the current pump valve. The system compares 3, 5, and 7 equidistant current points. With 3 points, the maximum displacement interpolation error is 1.2 mL / r; with 5 points, this decreases to 0.4 mL / r; and with 7 points, it only improves by another 0.1 mL / r. Therefore, 5 points are used: 0.4A, 0.6A, 0.8A, 1.0A, and 1.2A. Each point is compared 1, 3, and 5 times. After 3 repetitions, the median value change does not exceed 0.1 mL / r; therefore, both the current increase and decrease are measured 3 times.

[0148] In implementations where the actual displacement of the variable pump cannot be directly read, the system obtains the equivalent displacement readback through the aforementioned control current-displacement calibration table. During calibration, the displacement, pressure, and oil temperature at each current point are recorded, and the upward and downward lookup values ​​are saved for the hysteresis interval. During operation, the actual current readback obtained from the pump control valve current sensor is used to look up the table, and the flow trend given by the independent pump outlet flow sensor is required to be consistent with the calibration direction at the same speed. Both constitute the execution readback, and the issued command value cannot be used directly to prove execution. For example, in the upward table, 0.6A corresponds to 17.0 mL / r, 0.8A corresponds to 19.0 mL / r, and 1.0A corresponds to 21.0 mL / r. The actual current readback of 0.7A obtains the current equivalent displacement of 18.0 mL / r based on adjacent positions, and the actual current readback of 0.9A corresponds to the target equivalent displacement of 20.0 mL / r. The downward direction uses a separately saved downward table. The permissible readback error of 20 mL / r is taken from the 95th percentile of 30 repeated measurements and rounded up, for example, 0.5 mL / r.

[0149] The pump outlet flow trend judgment limit is determined through repeatable recording by the flow sensor. The system obtains three continuous readings at each of 30 stable flow points, and calculates the maximum deviation of the median value from the calibrated flow rate. In the example, the maximum deviation is 0.2 L / min. When comparing two consecutive median flow values, the deviations on both sides are added and rounded up, using 0.4 L / min as the direction confirmation dead zone. For example, at 1900 r / min and before displacement adjustment, the flow readings are 27.5 L / min, 27.6 L / min, and 27.7 L / min, with a median of 27.6 L / min. After adjustment, the readings are 30.6 L / min, 30.7 L / min, and 30.8 L / min, with a median of 30.7 L / min, representing an increase of 3.1 L / min, exceeding the direction confirmation dead zone and consistent with the calibrated direction of the increased displacement. If the actual current is missing or the flow trend is inconsistent, a readback is performed to mark it as inconsistent, and the displacement residual must not be set to 0.

[0150] Specific target points are selected from the candidate segment through point-by-point verification and fixed sorting. The system only enumerates the speed-displacement pairs that actually appear in the energy efficiency mapping table and have the minimum number of samples. Points with unqualified flow rate, power, cycle time, or quality are first excluded, and then sorted from low to high according to the median fuel consumption per unit operating area. When fuel records are the same, points with lower pump shaft input power and smaller execution residuals are given priority. In the example candidate segment, there are four verified points: 1900 r / min and 20 mL / r, 1950 r / min and 21 mL / r, 2000 r / min and 22 mL / r, and 2050 r / min and 22 mL / r, with corresponding median fuel consumption values ​​of 4.8 kg / hm². 2 4.9 kg / hm 2 5.0 kg / hm2 and 5.1 kg / hm 2 Therefore, the first specific pair is chosen instead of arbitrarily selecting values ​​from the two intervals.

[0151] In S43, the system selects specific target parameters from feasible pairs corresponding to candidate speed ranges and candidate displacement ranges. Before the target strategy is output, the pre-execution residual between the current controlled parameter and the specific target parameter is calculated; after the response waiting period, the post-execution residual between the execution readback data and the same specific target parameter is calculated. S22 uses the interval residual for power coverage verification with the target speed range as the object; S43 uses the execution residual for control action correction. After selecting the specific target parameter, both the pre-execution residual and the post-execution residual use the same specific target point as the object, and the two must not be mixed. The engine speed residual and the variable pump displacement residual are calculated using their respective specific target parameters, and residuals of different dimensions are not directly added together.

[0152] The relative improvement threshold for residuals was determined using 20 successful records and 15 invalid or overshoot records. The system iterated through three candidate values: 0.1, 0.2, and 0.3, evaluating the time to reach the target tolerance and the number of subsequent overshoots. When using 0.1, four instances of insufficient change were mistakenly judged as valid; when using 0.3, three normal responses were delayed in confirmation; and when using 0.2, only one instance required confirmation and no reverse actions were missed. Therefore, a relative improvement of 0.2 was determined as the effective execution requirement.

[0153] The residual proximity intervals are determined based on the repeatability and allowable tolerance of the controlled object. In 30 stable repeated measurements of the engine target readback, the 95th percentile difference is rounded up to 10 r / min; in 30 repeated measurements of the equivalent displacement of the variable pump, the corresponding result is rounded up to 0.5 mL / r. Therefore, a residual difference not exceeding the corresponding value is considered proximity. For example, when the engine residual decreases from 50 r / min to 45 r / min, the difference is within the proximity interval; similarly, when the displacement residual decreases from 2.0 mL / r to 1.6 mL / r, it is also within the proximity interval.

[0154] The specific target parameters allow for tolerances using repeatable results and are verified through operational testing. The system compares engine speed tolerances of 10 r / min, 20 r / min, and 30 r / min, and displacement tolerances of 0.5 mL / r, 1.0 mL / r, and 1.5 mL / r in 30 stable target tests. Three of the larger combinations prematurely determined the unstable readouts to be stable. The smallest combination matched the 95th percentile repeatability without adding any quality non-compliance records. Therefore, the allowable tolerances for engine speed and displacement are set at 10 r / min and 0.5 mL / r, respectively. For example, when the specific targets are 1900 r / min and 20 mL / r, engine speed readouts between 1890 r / min and 1910 r / min, and displacement readouts between 19.5 mL / r and 20.5 mL / r, are considered to fall within the allowable tolerance range for the specific targets.

[0155] The action reduction ratio is determined by the recovery time, overshoot count, and repetition adjustment count recorded above. The system was tested by reducing the next action increment to 25%, 50%, and 75% of the original increment. Using 25% required an average of two additional adjustments, using 75% resulted in three overshoots, and using 50% resulted in no overshoot and only an average increase of one adjustment. Therefore, 50% was chosen for this example. This ratio only applies to subsequent increments that have not yet been executed and does not undo targets that have already been safely executed and readback completed.

[0156] The system first determines whether the difference between the pre-execution residual and the post-execution residual falls into the residual proximity range. If it does, the subsequent control increment is reduced. If it does not fall into the range and the post-execution residual decreases, the system continues to determine whether the relative improvement reaches 0.2. If it does, the subsequent control action is maintained; otherwise, the subsequent control increment is reduced. If the post-execution residual increases and the increase exceeds the proximity range, the subsequent control action in the same direction is canceled and the execution link is checked. If the actual value has already entered the allowable tolerance of the specific target parameter, the current target is maintained and no further increase in the same-direction control increment is added. Each branch executes mutually exclusively in the above order, with engine speed and pump displacement being determined separately.

[0157] For example, if the target engine speed is 1900 r / min, and the actual speed before strategy output is 1850 r / min, the difference forms a pre-execution residual of 50 r / min. After 0.8 seconds, the readback shows 1890 r / min, and the difference with the same target forms a post-execution residual of 10 r / min. First, subtracting the post-execution residual from the pre-execution residual yields a reduction of 40 r / min. Then, processing relative to the pre-execution residual yields an improvement of 0.8, higher than the verified 0.2, and the reduction exceeds the upper limit of the near-range, so the current action is maintained. If another record shows a pre-execution residual of 50 r / min and a post-execution residual of 45 r / min, with the difference falling within the 10 r / min near-range, the next action is scaled down. If the post-execution residual is 70 r / min, with the increase exceeding the upper limit of the near-range, the action in the same direction is canceled.

[0158] In S44, a collaborative work group is a set of rice transplanters that have established communication associations within the same task number, field number, and work period. Each machine broadcasts its current load zone, pressure ranking position segment, speed segment, fuel consumption per unit work area, and work status indicator. The system excludes data related to turning, malfunctions, non-transplanting, and communication timeouts, and only allows units of the same model and calibration version to participate in the current comparison. Broadcasts from different models or calibration versions are directly marked as version incompatible and excluded. This implementation does not compare the original pressure across models or automatically convert the load of different models. The unit load difference sequence is a record of the difference between the current load zone of the machine and the majority load zone of the effective units as a new broadcast batch is generated. For example, there are 4 effective units of the same model and calibration version. Their current load zones are high load zone, medium-high load zone, medium-high load zone, and medium-high load zone, respectively. The majority load zone is the medium-high load zone, and the machine is in the high load zone. The current difference record is that the machine is one adjacent zone higher than the majority load zone. The system only adds this record when a new timestamp valid batch is received.

[0159] The operation status identifier consists of the turning status, the planting mechanism enable bit, the fault bit, and the broadcast timestamp. A line-lined state, planting enabled, no fault, and the timestamp not expired are marked as valid planting; if any condition is not met, the specific reason for rejection is recorded. In the example, the second unit's turning status is turning, the planting mechanism is still enabled, and the fault bit is normal, so it is marked as turning and rejected; the fifth unit maintains a straight line and is fault-free, but the latest timestamp has passed 1.5 seconds, so it is marked as communication timeout and rejected. The identifier field is saved simultaneously with the original broadcast; only the final valid or invalid conclusion should not be saved.

[0160] The broadcast period was determined through field communication experiments. The system recorded 100 broadcast arrival intervals under different inter-machine distances, different obstructions, and different bus loads, comparing candidate periods of 0.3s, 0.5s, and 0.8s. 0.3s significantly increased bus occupancy, 0.8s increased the delay in confirming load changes, and 0.5s covered 95% of the arrival intervals without bus congestion; therefore, 0.5s was adopted. Even if the same broadcast data was read by multiple local processing cycles, only one batch number was retained, preventing repeated accumulation as continuous evidence.

[0161] The number of communication timeouts was determined through packet loss replay. The system tested timeout conditions of 2, 3, and 4 consecutive times without updates. With 2 times, 4 short-term packet losses resulted in the premature exclusion of valid units; with 4 times, 2 instances of old operating data continued to be included in the comparison; with 3 times, there was no misuse of old data and only 1 instance requiring confirmation, therefore 3 consecutive times were used. When the broadcast period is 0.5s, the corresponding time is 1.5s; after this time, unit data is no longer included in the current majority statistics.

[0162] The majority of units is determined by a strict majority rule, obtained by increasing the number of valid units by 1, taking half the result, and rounding up. This rule replays records from 3 to 8 units to avoid ties for even-numbered units; for example, when there are 4 valid units, at least 3 must agree to constitute a majority opinion, and this unit is included in the 4 valid units. Continuous differences are only confirmed when 3 new valid broadcast batches are received consecutively and the first and last batches span 1.0 s. This confirmation count comes from 20 short-term path differences and 20 continuous load difference replays; the same remote snapshot must not be counted repeatedly at a local 0.1 s cycle.

[0163] The speed adjustment step size was determined through operational quality verification. The system conducted continuous straight-line tests on candidate step sizes of 0.04 m / s, 0.06 m / s, and 0.08 m / s to check predicted planting spacing, missed planting, seedling drift, engine overshoot, and ride smoothness. When using 0.04 m / s, load balancing required an average of two adjustments; when using 0.08 m / s, the predicted planting spacing approached the task boundary three times; and when using 0.06 m / s, it did not exceed the allowable range and only required one adjustment. Therefore, 0.06 m / s was selected for the example.

[0164] The ramp time is determined through a combination of manufacturer-permitted acceleration and quality testing. For the example model, the maximum permissible smooth deceleration under continuous transplanting conditions is 0.08 m / s². With a velocity change of 0.06 m / s, the theoretical shortest time is less than 1.0 s. The system then compares ramps of 0.8 s, 1.0 s, and 1.2 s. The 0.8 s ramp exhibits two beat fluctuations, the 1.2 s ramp results in an unbalanced load, and the 1.0 s ramp has no overshoot; therefore, 1.0 s is adopted. After each adjustment, at least one ramp time and a new broadcast batch must be waited before further adjustments are permitted.

[0165] The permitted transplanting rhythm conditions are the online control access conditions defined in this invention and are not equivalent to the complete operation quality being qualified. The system requires that the frequency readback is still fresh, there are no missing action timestamps, the transplanting mechanism is fault-free, the predicted hole spacing is within the agronomic task's allowable range, the speed command meets the machine's acceleration limit, and manual intervention is not activated. In the example, the independent closed-loop transplanting drive confirms the frequency remains at 5.4Hz, the speed is adjusted from 1.16m / s to 1.1m / s, and dividing the speed by the frequency yields a predicted hole spacing of approximately 0.2m, which falls within the 0.18m to 0.22m range, thus allowing execution. If the frequency timestamp exceeds the 0.4s freshness limit, the original speed is maintained even if the predicted hole spacing is within the range. When the transplanting frequency and travel speed are mechanically linked, the new frequency after linkage must be used for re-judgment; coordinated speed adjustment must not be executed if an effective frequency cannot be obtained. Any coordinated command must comply with the priorities of machine protection, path safety, and manual intervention; the final operation quality is still based on the complete index check results of the applicable test method.

[0166] This invention also provides a rice transplanting energy efficiency scheduling data processing system based on dynamic load zoning, such as... Figure 6 As shown, the system is used to implement a data processing method for rice transplanting energy efficiency scheduling based on dynamic load zoning. The system includes the following physical and logical modules:

[0167] The system comprises the following modules: a data acquisition module for acquiring engine output speed data, hydraulic system working pressure data, operating speed data, and seedling needle movement cycle data; a reference window filtering module for filtering valid and stable reference windows; a sequence generation module for generating pressure sorting position sequences, engine response difference sequences, speed deviation sequences, and seedling needle frequency sequences; a reliable load processing module for generating reliable load sequences; a symbol state processing module for generating symbol segments and processing cycle status codes; a dynamic partitioning module for generating dynamic load boundaries and current load partitions; an energy efficiency strategy determination module for outputting the engine target speed range and the hydraulic variable pump target displacement range; and an execution control module for correcting subsequent control actions based on the pre-execution residuals and post-execution residuals.

[0168] In one implementation, the data acquisition module includes an engine controller bus interface, a pressure sensor interface, a pump control valve current readback interface, a pump outlet flow sensor interface, a speed data interface, and a seedling needle motion sensor interface. The pressure and flow sensors are located at the variable pump outlet and reflect the load position of the continuous transplanting loop, avoiding non-target pressure and flow pulses generated solely by steering or the raising and lowering of the transplanting section. The pump control valve current readback interface receives the actual drive current independent of the issued command. The speed data interface receives the fusion result of wheel speed and satellite positioning. The motion sensor interface receives valid planting events for the same transplanting row. Each interface writes data with timestamps and validity flags into the machine controller memory. The processor executes reference window filtering, sequence generation, reliable load processing, symbol state processing, dynamic partitioning, and energy efficiency strategy determination logic.

[0169] In one implementation, the energy efficiency strategy determination module sends the target engine speed to the engine controller via the overall machine bus, and sends the target displacement or corresponding control current to the pump control valve via the variable pump control interface. The actual engine speed, target readback, actual pump control valve current, pump outlet pressure, pump outlet flow rate, and speed are returned to the execution control module via the corresponding interface. The execution control module returns the pre-execution residual, post-execution residual, and actual response stage to the dynamic partitioning module for subsequent processing cycles to generate boundary back-calibration results, thereby forming a data and control closed loop from sensors, processor, control bus, engine, and variable pump to the readback interface.

[0170] In one implementation, the system further includes a collaborative scheduling module, a human-machine interaction module, and a data logging module. The collaborative scheduling module receives valid unit broadcast data from the same collaborative work group, generates unit load difference sequences, and outputs ramp-limited local speed adjustment commands. The human-machine interaction module displays the current load partition, processing cycle status code, target power output strategy, maximum permissible continuous operating pressure status, continuous operation advance protection status, independent overflow protection status, absolute safe pressure status, and execution verification results. The data logging module saves original samples, validity flags, reference window numbers, four types of sequences, reliable load data, symbol segments, status codes, dynamic boundaries, current partitions, target strategies, execution residuals, and response stages. The data logging module retains the correspondence between original values ​​and processed values ​​to support the updating of frozen calibration sequences and fault tracing in the next work stage.

[0171] The following uses a specific application scenario to integrate and verify the calculation logic and data flow process of each step in the above implementation method.

[0172] Assume that at a certain moment, a high-speed riding rice transplanter that has completed checks on the same model and calibration version is in a continuous straight-line transplanting state. The agronomic task document gives a target hill spacing of 0.2m and a target single-row transplanting frequency of 6.0Hz. Multiplying these two values ​​gives a target speed of 1.2m / s. Based on the conversion of a time unit containing 60s per minute, multiplying 6.0Hz by 60 gives 360 hill-forming events per minute, which is higher than the classification benchmark of 350 events per minute for high-speed rice transplanters in the "Guidance Catalogue for Industrial Structure Adjustment (2024 Edition)". The target hill spacing is also within the range of 0.18m to 0.22m given in the task document. The equipment datasheet specifies the pressure sensor's range as 0 to 25.0 MPa, the maximum permissible continuous operating pressure as 16.0 MPa, and the short-term peak permissible pressures for the pump, valves, pipelines, and sensors as 20.0 MPa, 21.0 MPa, 22.0 MPa, and 25.0 MPa, respectively. The minimum value among these is taken as the absolute safe pressure limit of 20.0 MPa, and the equipment overflow is set to 18.0 MPa, which is lower than this limit. Pressure calibration and protection action tests provide a measurement margin of 0.2 MPa and a control overshoot margin of 0.3 MPa, respectively. Subtracting these margins sequentially from the maximum permissible continuous operating pressure yields the continuous operation early protection trigger pressure of 15.5 MPa.

[0173] In this example, all power boundaries are derived from equipment calibration under the same operating mode. The pump synchronous speed record gives a transmission ratio of 0.9 between the pump shaft speed and the engine shaft speed; the flow calibration gives a volumetric efficiency of 0.9; the bench records of pump shaft torque, pump shaft speed, outlet pressure, and flow rate give a combined efficiency of 0.84 for the variable pump; and the high-temperature thermal balance test gives an allowable continuous flow rate of 40.0 L / min. Engine mapping at 1850 r / min, 1875 r / min, and 1900 r / min yields usable power outputs of 14.0 kW, 14.2 kW, and 14.5 kW, respectively; the stable flywheel power at 1900 r / min is 16.0 kW, which, after deducting the necessary internal consumption of 1.5 kW, results in 14.5 kW. Under the same mode, the pump drive loss, travel load, mechanical insertion load, and electrical auxiliary load are 0.4kW, 1.2kW, 0.9kW, and 0.5kW respectively. Adding these together yields an upper limit of 3.0kW for other loads not included in the pump shaft input power. The maximum unmodeled increment recorded in 20 accelerated tests is 0.8kW, which is rounded up and used as a transient margin of 1.0kW. These values ​​are used as the equipment boundaries in this example and are not replaced by dynamic load boundaries.

[0174] The processing cycle is derived from the channel statistics of the aforementioned 30 fault-free records. The 95th percentile update intervals for the continuous engine speed, pressure, and velocity channels are 0.05s, 0.02s, and 0.1s, respectively. The 95th percentile delay from action event triggering to data entry is 0.08s. The maximum value is taken and merged into the controller's base clock to obtain a processing cycle of 0.1s. The control log shows that the engine target was adjusted from 1800r / min to 1900r / min eight processing cycles prior to the current cycle. The command age corresponding to these eight processing cycles is 0.8s. After that, the target remains unchanged, and the target record of the previous processing cycle currently saved is still 1900r / min.

[0175] The initial engine readings for the current cycle are 1848 r / min, 1850 r / min, and 1852 r / min, with the median value being 1850 r / min; the initial pressure readings are 13.7 MPa, 13.8 MPa, and 13.9 MPa, with the median value being 13.8 MPa; the speed fusion interface readings are 1.15 m / s, 1.16 m / s, and 1.17 m / s, with the median value being 1.16 m / s. The pump and valve control log shows a current upward control current of 0.7 A. From the adjacent positions in the 5-point calibration table (0.6 A corresponding to 17.0 mL / r and 0.8 A corresponding to 19.0 mL / r), the current equivalent displacement is obtained as 18.0 mL / r. Before operation, the length of the field verification section was calibrated to 20m with a measuring tape. The timing record was 17.2s. Dividing the distance by the time and retaining two decimal places yielded 1.16m / s. This average value was only used to verify the speed channel ratio and did not replace the current cycle readback. The 6 valid moments recorded by the motion sensor formed 5 complete intervals with a first and last span of 0.92s. Dividing the number of intervals by the span and retaining one decimal place yielded a clustered seedling needle frequency of 5.4Hz.

[0176] In S1, the historical range of 60s and the window length of 5s are derived from multiple combinations of 20 records for verification. Dividing the historical range by the window length yields 12 candidate windows, and dividing the window length by the 0.1s processing cycle yields 50 cycles per window. Integrity checks use a 5-cycle exclusion limit determined by manual masking verification, excluding 3 windows. Continuous boundary checks use a 6-cycle pressure limit determined by 25 pressure shocks and continuous load records, excluding 2 more windows; in this example, no exclusions occurred for speeds reaching 5 cycles or frequencies reaching 3 independent intervals. Direction reversal checks use a 4-cycle limit determined by stable and oscillating windows, excluding 1 more window. No new exclusions were added during readback consistency checks. After successively subtracting the aforementioned 3, 2, and 1 windows from the 12 candidate windows, 6 valid stable reference windows are retained, each with 50 pressure data points, resulting in 300 reference pressure data points. Continuity is maintained within each window, and only distribution statistics are merged between windows; direction and action intervals are not calculated across windows.

[0177] The system compares the current pressure of 13.8 MPa with 300 reference pressures one by one. The counting results show that 272 reference values ​​are lower than the current pressure, and 2 reference values ​​are equal to the current pressure. Substituting 272, 2, and 300 into the unique expression in S13, the parallel data contributes one rank according to the midpoint rank. After adding the correction of the current value, the numerator is 274, and the denominator after adding the current data is 301. Dividing them and keeping two decimal places, the pressure sorting position is 0.91. The current actual engine speed of 1850 r / min minus the target recorded speed of 1900 r / min in the previous processing cycle gives the engine response difference of −50 r / min; the current speed of 1.16 m / s deviates negatively by 0.04 m / s from the target speed of 1.2 m / s. The aggregation needle frequency of 5.4 Hz is used for amplitude judgment, and the single interval frequency and its midpoint time are used for the downward deviation starting point judgment, thus forming four types of sequences.

[0178] In S2, the pressure upper bias boundary of 0.84 Hz comes from boundary optimization of 300 tagged calibration records, the stable lower frequency of 5.9 Hz comes from 30 segments of qualified continuous operation records, and the continuous downward bias requiring 3 independent action intervals comes from 24 missed reads and 18 real beat descent verifications. The unified timestamp records show that, after taking the midpoint of the first qualified low-frequency interval as the physical starting point, the starting point of the pressure sorting position entering the upper bias segment is still 0.3 s earlier than the starting point of the lower frequency bias; this difference is not generated by the aggregated frequency window spanning 0.92 s. The arrival difference of 0.08 s from 100 synchronous trigger tests on the bench is obtained. After adding a timestamp margin of 0.02 s, the overall time delay uncertainty is 0.1 s. After correction, the starting point difference exceeds this value, thus generating the result of pressure shifting upward first and then downward biasing. The 95th quantile time given by the most recent 20 effective engine response records is 0.6s, which becomes 0.8s after adding a margin of 2 processing cycles. The corresponding quantile time given by the most recent 20 effective variable pump step records is 0.4s, which becomes 0.6s after adding the same margin. Taking the larger value of the two results, the final response waiting period is 0.8s.

[0179] The current feedback verification is based on the specific speed target of 1900 r / min, which was output and maintained 0.8 seconds ago, not the target that was first output in the previous processing cycle. In the completed response records, the actual speed has moved closer to the current value of 1850 r / min from 1800 r / min, and the residual relative to the same target has decreased from 100 r / min to 50 r / min. The reduction of 50 r / min is divided by the initial residual to obtain an improvement of 0.5, which exceeds the 0.2 threshold determined by 20 successful, 10 hysteresis, and 15 invalid records. Therefore, a valid result for power coverage is generated. Dividing the negative speed bias of 0.04 m / s by the target speed of 1.2 m / s and converting it to a percentage yields 3.3%, which falls within the slightly negative bias range determined by 40 quality qualified records and 15 slip records. Therefore, the current data is written into the reliable load data.

[0180] The current processing log shows that in the last 10 processing cycles, 8 pressure sorting positions have moved up, reaching the number of main directions determined by short window verification; in the last 60 processing cycles, 39 have moved up, exceeding the 31 required for long window to form the background main direction. The minimum and maximum pressure sorting positions in the last 10 cycles are 0.86 and 0.91, respectively. Relative to the 5th and 95th quantile endpoints of the 20 independent and valid stable verification windows obtained from the same frozen pressure sequence, they are 0.12 and 0.82, respectively. Therefore, the current extreme value range is generally higher than the reference extreme value range. The out-of-bounds error has exceeded the recovery parameter and has not yet returned. The residual of the previous completed control action has also decreased significantly. Therefore, the status codes are recorded in a fixed order as continuous upward movement, overall upward movement, insufficient recovery, and effective execution.

[0181] In S3, 300 historical data points from at least 5 fields, labeled with load, fuel, frequency, and operation quality, were validated using a finite boundary combination, resulting in three basic boundaries: 0.28, 0.58, and 0.84. The current pressure ranking position of 0.91 exceeds the third boundary, making it a basic candidate for high load. The cycle pressure results support actual load increases, power coverage results indicate engine response, and a slight negative speed bias provides supplementary evidence; therefore, the current candidate remains high load. Boundary crossing records show that this candidate has crossed the boundary for 8 consecutive processing cycles. Multiplying the number of cycles by 0.1s yields a duration of 0.8s, exceeding the 6-cycle confirmation length determined by 25 continuous load increases, and failing to recover within the 3-cycle rapid return range determined by 30 short-duration impact records. Therefore, the current load zone is determined to be a high load zone.

[0182] This application scenario uses the current processing cycle as the time base. S3 only uses the target strategy feedback when the instruction age has reached 0.8s. The theoretical response phase is formed based on 20 historical valid responses and 10 noise records. The response start and response enhancement each require confirmation for 2 consecutive cycles, and the response stability requires maintaining it continuously for 5 cycles within the target allowable range. The current feedback shows that the speed residual is continuously decreasing and the pressure change is still consistent with the historical direction after exceeding the 0.1MPa dead zone determined by 100 constant pressure samples. However, stable evidence for 5 consecutive cycles has not yet been obtained. Therefore, the actual response stability phase is recorded as pending confirmation and the dynamic boundary is frozen. The current pressure of 13.8MPa is lower than the continuous operation early protection trigger pressure of 15.5MPa, and also lower than the maximum allowable continuous working pressure of 16.0MPa, the overflow setting of 18.0MPa, and the absolute safety pressure limit of 20.0MPa. Therefore, pressure protection is not triggered in this example. The output and feedback of this cycle S4 are only used by the subsequent processing cycle S3 after the instruction age reaches the response waiting period and are not written back to the boundary of this cycle.

[0183] In S4, firstly, the high-load zoning, insufficient recovery state, and continuous out-of-bounds data collectively satisfy the secondary control conditions determined by 40 priority playback groups, and the pressure protection condition is not met, so the secondary control priority is output. Secondly, the system queries the energy efficiency mapping table grouped with the same model, calibration version, and field conditions; the candidate segment contains 4 adjacent target points, each with 40 valid records, for a total of 160 records, all reaching the minimum stable number determined by the leave verification, merging to form the speed segment 1900r / min to 2050r / min and its paired displacement segment 20mL / r to 22mL / r. The working width of 1.8m in the record multiplied by the effective distance of 1000m gives 1800m. 2 According to 1hm 2 Includes 10000m 2 The unit conversion is 0.18hm. 2 The fuel mass of 0.86 kg is divided by the area and rounded to one decimal place to obtain the fuel consumption per unit operating area of ​​4.8 kg / hm². 2 The other three verified points were 1950 r / min and 21 mL / r, 2000 r / min and 22 mL / r, and 2050 r / min and 22 mL / r, respectively, corresponding to a median fuel consumption of 4.9 kg / hm². 2 5.0 kg / hm 2 and 5.1 kg / hm 2 Therefore, 1900 r / min and 20 mL / r were specifically paired according to a fixed order.

[0184] Secondly, the system completes the transition range verification in two stages. In the first stage, when the speed increases from the current 1850 r / min to the lower end of the target range of 1900 r / min, the current equivalent displacement is maintained at 18.0 mL / r. The maximum estimated flow rate is 27.7 L / min, processed with the upper speed of the stage, transmission ratio of 0.9, and volumetric efficiency of 0.9. Using a more stringent pressure of 15.5 MPa, the internal unrounded hydraulic output power is 7.16 kW. Dividing this by the overall efficiency of 0.84, the pump shaft input power is 8.52 kW. After adding the upper limit of other loads of 3.0 kW and the transient margin of 1.0 kW, the result is 12.52 kW, which is lower than the minimum usable power of 14.0 kW in the transition range. In the second stage, the displacement ramp was adjusted to 20 mL / r only after the engine speed reached 1900 r / min, resulting in an estimated flow rate of 30.8 L / min, which is lower than the thermal balance limit of 40.0 L / min. At the current pressure of 13.8 MPa, the candidate target steady-state hydraulic output power was 7.08 kW and the pump shaft input power was 8.43 kW. Verifying under a more stringent pressure of 15.5 MPa yielded 7.95 kW and 9.47 kW, respectively. Adding the latter, other loads, and transient margin, the result is 13.47 kW, which is lower than the target usable power of 14.5 kW. Therefore, the complete transition process is feasible.

[0185] Finally, the system performs feedback on the engine speed residual and the variable pump displacement residual respectively. The engine speed saved before the target strategy output is 1850 r / min, and the difference from the specific target of 1900 r / min forms the pre-execution residual of 50 r / min. After 0.8 seconds, obtained from historical response statistics, the engine readback is 1890 r / min, and the difference from the same target forms the post-execution residual of 10 r / min. Subtracting the post-execution residual from the pre-execution residual yields a reduction of 40 r / min, which represents an improvement of 0.8 compared to the pre-execution residual processing. This improvement is higher than the verified 0.2, and the reduction exceeds the 10 r / min proximity range determined by 30 stable repeated measurements. Therefore, the engine control action is maintained. Subsequently, the engine log recorded 1900 r / min, 1902 r / min, 1903 r / min, 1901 r / min and 1900 r / min in 5 consecutive processing cycles, all of which entered the target speed range and specific target allowable tolerance; the number of confirmations in 5 consecutive cycles came from the separation verification of historical effective response and noise records, and the first 1900 r / min record thus became the actual response stability point.

[0186] After the engine enters the target speed range, the system outputs the target control current of the variable pump to 0.9A. The actual current reading before output is 0.7A, and the equivalent displacement is obtained from the table as 18.0mL / r. Relative to the specific target of 20.0mL / r, there is a displacement residual of 2.0mL / r before execution. At the same speed, the three readings of the outlet flow rate before adjustment are 27.5L / min, 27.6L / min, and 27.7L / min, with a median value of 27.6L / min. After 0.8s, which is jointly determined by the historical response time of the engine and the variable pump, the pump controller returns the execution completion flag. The actual reading of the independent current sensor is 0.9A, and the three readings of the outlet flow rate are 30.6L / min, 30.7L / min, and 30.8L / min, with a median value of 30.7L / min. The median flow rate increased by 3.1 L / min, exceeding the dead zone identified by 30 stable points at 0.4 L / min, which is consistent with the calibration direction. Therefore, the equivalent displacement of 20.0 mL / r was obtained by referring to the table using the actual current of 0.9 A. After execution, the displacement residual was 0 mL / r. The displacement residual decreased by 2.0 mL / r, representing an improvement of 1.0 compared to the residual before execution, which is higher than the effective improvement threshold of 0.2. The reduction also exceeded the 0.5 mL / r range, so the variable pump control was maintained. Feedback from both the engine and variable pump was written to the data logging module for use in the subsequent processing cycle S3 after the response waiting period expired, without writing back to the current processing cycle boundary.

[0187] In optional collaborative processing, the task registration for the same field results in 6 rice transplanters of the same model. The broadcast cycle is 0.5s, determined by 100 communication tests. One of the transplanters is in a turning state, and another has not been updated for 3 consecutive broadcast cycles. The cumulative broadcast cycle reaches 1.5s and reaches the timeout condition. After excluding these two, 4 valid units remain, including the current transplanter. The majority number is 3 units from 4 units according to the strict majority rule. When 3 new timestamp-valid broadcast batches are received consecutively, and the first and last batches cumulatively span 1.0s from the 0.5s cycle, and each batch shows that the current transplanter is in the high-load zone and the other 3 are in the medium-high load zone, it is confirmed that the current transplanter is continuously above the majority load zone and a speed limit suggestion is output. The same broadcast snapshot is not counted repeatedly in multiple local processing cycles.

[0188] The speed step size determined by quality verification is 0.06 m / s. Subtracting this step size from the current speed of 1.16 m / s yields the target speed of 1.1 m / s, which is then gradually increased according to the manufacturer's allowed deceleration and the 1.0 s ramp determined by quality testing. The planting mechanism mode record shows that the machine uses independent closed-loop drive, the frequency timestamp does not exceed the 0.4 s freshness limit, and the readback confirms that the frequency remains at 5.4 Hz. Before adjustment, dividing 1.16 m / s by 5.4 Hz and retaining two decimal places yields a predicted plant spacing of 0.21 m. After adjustment, dividing 1.1 m / s by the same frequency and retaining one decimal place yields a predicted plant spacing of 0.2 m. Both are within the allowable range, thus meeting the allowable planting rhythm conditions. The system must wait at least one 1.0 s ramp time and a new 0.5 s broadcast batch before allowing further adjustments. If the machine model has a speed-frequency mechanical linkage, frequency failure, or manual intervention activation, the original speed is maintained.

[0189] The preferred embodiments of this disclosure have been described in detail above with reference to the accompanying drawings. However, this disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this disclosure, various simple modifications can be made to the technical solutions of this disclosure, and these simple modifications all fall within the protection scope of this disclosure.

[0190] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, this disclosure will not describe the various possible combinations separately.

[0191] Furthermore, various different embodiments of this disclosure can be combined in any way, as long as they do not violate the spirit of this disclosure, they should also be regarded as the content disclosed in this disclosure.

[0192] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A data processing method for rice transplanting energy efficiency scheduling based on dynamic load zoning, characterized in that, The methods include: Acquire engine output speed data, hydraulic system working pressure data, operating speed data, and seedling needle movement rhythm data during rice transplanter operation, and filter effective stable reference windows from historical sampling data; The current hydraulic system working pressure data is inserted into the pressure sorting sequence of the effective stable reference window to generate a pressure sorting position sequence, and an engine response difference sequence, speed deviation sequence, and needle frequency sequence are generated. The starting point of the pressure sorting position entering the upper bias section is compared with the starting point of the seedling needle frequency entering the lower bias section. A reliable load sequence is generated by combining the engine response difference sequence and speed deviation sequence within the response waiting period. The reliable load sequence is then converted into a symbol segment and a processing cycle status code is generated. A dynamic load boundary is generated based on the effective stable reference window. Boundary crossing return data of load partition candidates relative to the dynamic load boundary is counted. The current load partition is determined based on the processing cycle status code and the boundary crossing return data. The target engine speed range and the target hydraulic displacement range are determined based on the current load zone.

2. The rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning according to claim 1, characterized in that, The process of selecting a valid and stable reference window from historical sampling data includes: Historical sampling data are divided into candidate historical windows with the same window length, and an initial inspection interval is formed using the manufacturer's allowed range and the calibration pressure sequence frozen in the previous operation phase. Candidate historical windows that are excluded in sequence due to the number of missing data reaching the integrity screening exclusion limit, the number of consecutive out-of-bounds data reaching the boundary screening exclusion limit, the number of direction flips reaching the stability screening exclusion limit, or the failure to meet the consistency requirements when performing readback consistency marking, are determined as valid stable reference windows. In this process, distributed statistics merge the data from each stable sub-window, while time-series statistics are performed only within each stable sub-window.

3. The rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning according to claim 1, characterized in that, The generation of the pressure sorting position sequence, engine response difference sequence, speed deviation sequence, and seedling frequency sequence includes: The pressure data within the effective stable reference window is arranged from low to high. The current pressure is temporarily inserted and its sorting position is obtained according to the midpoint rank of the parallel data, without writing the current pressure back into the pressure sorting sequence. The engine response difference is generated based on the difference between the current actual engine speed and the engine target speed recorded in the previous processing cycle. Speed ​​deviation data is generated based on the deviation between the current operating speed and the target operating speed; The single-row transplanting frequency is formed based on the effective hole formation time of adjacent transplanting rows in the same transplanting row. When a complete action interval is not formed, a beat confirmation mark is generated.

4. The rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning according to claim 1, characterized in that, The comparison of the starting point of the pressure sorting position entering the upper bias segment and the starting point of the seedling needle frequency entering the lower bias segment, combined with the engine response difference sequence and speed deviation sequence within the response waiting period, generates a reliable load sequence, including: Perform unified clock alignment and link delay compensation on the two starting points, and determine the order only when the absolute value of the time difference between the two starting points exceeds the comprehensive delay uncertainty; Based on the order of the starting points, the cycle pressure verification results are generated, which show that the pressure moves upward first and the seedling needle deviates downward later, and the abnormal observation or synchronous short-term deviation of the seedling needle side is observed. Based on the engine response difference sequence and the speed residual change of the same target speed segment, the power coverage verification results are generated. Combined with the speed deviation data, the corresponding cycle is written into the reliable load data, load observation data or execution lag observation data.

5. The rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning according to claim 1, characterized in that, Converting the trusted payload sequence into symbol segments and generating processing cycle status codes includes: Changes in pressure sorting position between adjacent processing cycles are recorded as upward movement, downward movement, holding, or alternating directions. Records with consecutive directions in the same direction are merged to form symbolic segments. A directional status code is generated based on the direction and continuous length of the symbol segment. An amplitude status code is generated based on the relative position of the extreme value interval of the current pressure sorting position and the extreme value interval of the reference pressure sorting position. A recovery status code is generated based on the out-of-bounds duration relative to the reference stable interval, the number of sampling points required to return to the stable interval, and the stable duration after returning. An execution status code is generated based on the change in residuals before and after execution. The processing cycle status code is formed in a fixed order of direction, amplitude, recovery, and execution.

6. The rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning according to claim 1, characterized in that, Generating dynamic load boundaries based on the effective stable reference window includes: Generate basic sorting position boundary candidates based on historical valid operation data with load status, fuel consumption per unit operating area, seedling needle frequency and operation quality tags; The candidate combinations with sequentially increasing order, minimum interval between adjacent boundaries, and minimum number of samples in each partition are validated to obtain three basic sorting position boundaries and initialize dynamic load boundaries.

7. The rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning according to claim 6, characterized in that, After obtaining the three basic sorting position boundaries and initializing the dynamic load boundaries, the process includes: The theoretical response phase sequence is generated based on records of engine target speed changes that have occurred before the current processing cycle and historical valid response records. Read the most recent target strategy record and corresponding feedback when the instruction age reaches the response waiting period, and extract the actual response phase sequence; The two response phase sequences are compared item by item, and boundary back-calibration results are generated based on phase integrity, phase sequence, and post-execution residual changes, including maintaining boundaries, extending waiting time, freezing boundaries, limited outward movement, or limited inward movement. If the upper-level load partition is triggered continuously and the residual does not improve, the boundary is frozen first. The execution link and stress protection status are checked, and the boundary is moved outward only after the partition false positives are confirmed continuously.

8. The rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning according to claim 1, characterized in that, The process involves statistically analyzing the boundary crossing return data of candidate load partitions relative to the dynamic load boundary, and determining the current load partition based on the processing cycle status code and the boundary crossing return data, including: Based on the dynamic load boundary and pressure sorting position, a candidate basic load zone is generated, and the candidate is corrected in turn based on the cycle pressure verification result, the power coverage verification result and the speed deviation data. Record the initial crossing location, the duration of the out-of-bounds crossing, the number of sampling points required to return to the stable interval, and the duration of the stable interval after the return. When a short-term out-of-bounds error occurs and a quick return occurs, the load partition of the previous processing cycle is maintained. When an out-of-bounds error continues, a slow return occurs, or the recovery is insufficient, the current load partition is determined based on the corrected candidate. The current load zones are mutually exclusive low load zones, medium-low load zones, medium-high load zones, or high load zones.

9. The rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning according to claim 1, characterized in that, Based on the current load zone, the target engine speed range and the target hydraulic variable pump displacement range are determined, including: Query the energy efficiency mapping table to obtain the target speed range of candidate engines and the target displacement range of candidate hydraulic variable pumps from the same feasible pairing record; Verify that the estimated flow rate does not exceed the pump's allowable continuous flow rate, and verify that the sum of the pump shaft input power and other load power not included in the pump shaft input power does not exceed the engine's available power. The insertion hydraulic power supplied by the hydraulic variable pump is not counted repeatedly. The calibration covers the complete transition range from the current speed to the target speed. During the transition, the engine speed is increased first while maintaining or decreasing the pump displacement. Once the engine speed enters the target speed range, the pump displacement is adjusted on a ramp.

10. A rice transplanting energy efficiency scheduling data processing system based on dynamic load zoning, characterized in that, The system is used to implement the rice transplanting energy efficiency scheduling data processing method based on dynamic load zoning as described in any one of claims 1 to 9, and the system includes: The data acquisition module is used to acquire engine output speed data, hydraulic system working pressure data, working speed data, and seedling needle movement rhythm data; The reference window filtering module is used to filter valid and stable reference windows. The sequence generation module is used to generate pressure sorting position sequences, engine response difference sequences, speed deviation sequences, and seedling frequency sequences. A trusted load processing module is used to generate trusted load sequences; The symbol state processing module is used to generate symbol segments and processing cycle status codes; The dynamic partitioning module is used to generate dynamic load boundaries and current load partitions; The energy efficiency strategy determination module is used to output the target speed range of the engine and the target displacement range of the hydraulic variable pump; The execution control module is used to correct subsequent control actions based on the residuals before and after execution.

Images