A state monitoring method and system in a multi-machine scheduling scene of an agricultural robot

Abstract

The application discloses a kind of state monitoring method and system under the scene of multi-machine scheduling of agricultural robot, the method is established long connection with cloud platform after starting agricultural robot end, corresponding working condition parameter baseline is loaded after completing identity authentication, and real-time data channel is initialized;End side real-time identification operation mode and with working condition data synchronization report, cloud end matches dynamic alarm threshold and carries out working condition self-adaptive anomaly detection, according to single equipment and machine group two-dimensional display running situation, realize global real-time perception.Abnormal triggering hierarchical execution linkage strategy, generate structured work order and push to operation and maintenance terminal;When fault shutdown, linkage scheduling system completes remaining task re-planning, and equipment is repaired and automatically re-enters scheduling pool.After operation is completed, cloud end archives full data according to hierarchical storage architecture and generates operation report.The application solves the problems that existing agricultural machine monitoring working condition has poor adaptability, alarm accuracy is low, weak network data reliability is insufficient, multi-machine storage efficiency is low and operation and maintenance scheduling lacks linkage.

Description

Technical Field

[0001] This invention relates to the field of agricultural machinery robot technology, and in particular to a method and system for status monitoring in a multi-machine scheduling scenario of agricultural machinery robots. Background Technology

[0002] With the development of smart agriculture technology and the advancement of large-scale agricultural operations, agricultural robots are gradually replacing traditional manually driven agricultural machinery, undertaking tasks in various stages such as tilling, sowing, plant protection, and harvesting. These devices typically use batteries as a power source, relying on motors to drive the walking and working mechanisms, and achieving autonomous navigation and path planning through onboard controllers. They have become one of the core pieces of equipment in modern agricultural production. In the production scenarios of large-scale farms and agricultural cooperatives, it is often necessary to simultaneously dispatch multiple agricultural robots to carry out joint operations. This creates an urgent need for centralized monitoring of equipment operating status, anomaly warnings, and maintenance responses. Stable and reliable multi-machine status monitoring systems have become a key support for ensuring the efficiency of agricultural machinery operations and the continuity of production.

[0003] Currently, there are three main existing technical solutions for monitoring the condition of agricultural machinery:

[0004] The first type is manual inspection based on local dashboards, a common management method used by small and medium-sized agricultural machinery cooperatives. Each agricultural robot is equipped with a physical pointer instrument or a simple LED digital display screen in its cab or on the side of the vehicle to display the remaining battery power, motor operation indicator status, and radiator temperature warning signs. On-site inspectors check the status of each machine individually. This solution can only achieve local, discrete status monitoring. Data acquisition is severely delayed and lacks real-time capability. There will be monitoring blind spots of tens of minutes or even hours between two inspections. If abnormal conditions such as motor overload or sudden drop in battery voltage occur during this period, they cannot be detected in time. At the same time, the multi-machine management capability of this solution is extremely limited. When there are many agricultural robots operating simultaneously and they are scattered across different plots, the workload of manual inspection increases significantly, and operators cannot fully grasp the overall operating status of the fleet. Furthermore, key operating parameters are only stored locally on the equipment and cannot be uploaded and aggregated to form a data set. This results in fault diagnosis relying entirely on the experience judgment of on-site personnel, lacking data-supported remote operation and maintenance decision-making.

[0005] The second category consists of remote tracking products based on GPS / BeiDou positioning modules, which have been applied to some mid-to-high-end agricultural machinery or retrofitted equipment. This solution involves installing a positioning terminal on the agricultural robot. The terminal has a built-in mobile communication module and satellite positioning chip, which uploads the equipment's geographical coordinates, speed, and direction angle information to a cloud server at fixed time intervals. Operators can view the real-time location and historical routes of each machine via a mobile app or computer webpage, primarily for equipment anti-theft, operational area statistics, and general dispatching. However, this solution only obtains the equipment's location and motion status information, completely failing to perceive critical internal conditions that determine the equipment's health, such as battery electrochemical status, motor electrical parameters, and the thermodynamic state of core components. Operators can only know the equipment's location but cannot determine its operational health status. Furthermore, this solution lacks proactive early warning and maintenance linkage mechanisms. When equipment stops due to a malfunction, the positioning terminal only shows a zero speed state, unable to automatically distinguish it from a normal parking or waiting state. The system cannot proactively identify the anomaly and notify maintenance personnel, still relying on on-site personnel to report the problem by phone, failing to form a closed-loop business process from monitoring detection to fault handling.

[0006] The third category consists of existing general-purpose IoT monitoring platforms, such as mainstream commercial agricultural machinery management platforms, which can remotely monitor agricultural machinery and provide basic fault alerts. These platforms typically collect equipment data using a fixed-period polling method, relying solely on voltage as a single indicator for battery status monitoring. Alarm strategies use fixed threshold comparisons, and multi-machine management employs a unified data collection frequency and fixed storage strategy. This type of solution has several shortcomings in agricultural machinery operation scenarios. First, the accuracy of battery status assessment is insufficient. Existing platforms typically estimate battery SOC using a single coulomb counter or open-circuit voltage method, failing to consider the impact of frequent high-current surges, contact resistance changes caused by severe vibrations, and large day-night temperature differences in farmland on battery characteristics. Under actual operating conditions, the SOC estimation error is significant, far exceeding the nominal accuracy. Second, the alarm strategy lacks adaptability to operating conditions. Fixed threshold alarms cannot distinguish between parameter fluctuations caused by normal load changes and genuine abnormal degradation trends. This is particularly problematic during normal operations such as climbing slopes and deep plowing. In some cases, false alarms are prone to occur frequently, while early faults are easily missed during light-load cruise. Third, the data transmission strategy is not optimized for weak signal environments in farmland. Farmland operation areas are mostly located in areas with weak communication signal coverage. The fixed-frequency reporting strategy is prone to causing a large amount of data loss in weak signal areas, and there is a lack of effective data retransmission and consistency guarantee mechanisms after the signal is restored. Fourth, there is a lack of targeted data hierarchical storage strategy in multi-machine concurrent scenarios. When dozens of agricultural machines report data at the same time, the unified storage strategy will cause excessive writing pressure on the time-series database and significantly increase the query response time, which cannot meet the low latency requirements of real-time monitoring.

[0007] In summary, existing agricultural machinery status monitoring technologies are ill-suited to the complex farmland conditions, weak signal communication environments, and large-scale concurrent monitoring requirements of multi-machine scheduling scenarios for agricultural robots. They cannot simultaneously achieve high-precision battery status assessment, low false alarm rate anomaly detection, highly reliable data transmission, and efficient multi-machine concurrent data management. Furthermore, they struggle to form a complete solution from anomaly detection to maintenance and handling, thus hindering the maintenance efficiency and operational stability of large-scale agricultural robot operations. Summary of the Invention

[0008] To address these issues, this invention provides a status monitoring method and system for multi-machine scheduling scenarios of agricultural machinery robots, solving the problems of large battery status estimation errors, numerous false alarms and missed alarms under complex farmland conditions, poor data reliability in weak signal environments, low efficiency of multi-machine concurrent storage, and lack of operation and maintenance and scheduling linkage in existing agricultural machinery monitoring systems.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a method for status monitoring in a multi-machine scheduling scenario of agricultural machinery robots, comprising the following steps:

[0010] S1. Equipment Online and Operating Condition Initialization: After the agricultural machinery robot is started, it establishes a long communication connection with the cloud platform and reports basic equipment information and initial self-test data to the cloud platform. After the cloud platform completes the equipment identity authentication of the agricultural machinery robot, it loads the corresponding operating condition parameter baseline configuration according to the equipment model, initializes the real-time data transmission channel, and updates the equipment online status.

[0011] S2. Adaptive Real-time Monitoring: The agricultural machinery robot terminal identifies the current operation mode in real time based on the collected multi-source operation data, and reports the operation mode tag to the cloud platform along with the real-time operation data. The cloud platform matches the corresponding parameter baseline and dynamic alarm threshold according to the received operation mode tag, performs adaptive anomaly detection on the real-time operation data, and displays the operation status in two dimensions: single device and machine group, so as to achieve full-domain real-time perception of the operation status of multiple machines.

[0012] S3. Tiered Alarm and Intelligent Operation and Maintenance Linkage: When the anomaly detection result of the cloud platform exceeds the corresponding dynamic alarm threshold, the cloud platform classifies the alarm level according to the severity and scope of the anomaly, and executes the corresponding linkage handling strategy; at the same time, it generates a structured work order containing device identifier, alarm type, operating condition context and handling suggestions and pushes it to the operation and maintenance terminal, forming a full-process operation and maintenance management from alarm triggering, work order dispatch, on-site handling to confirmation;

[0013] S4. Multi-machine scheduling and linkage: When the agricultural machinery robot terminal triggers an emergency alarm due to a fault, causing the operation to be interrupted, the cloud platform will synchronize the location information of the faulty equipment and the remaining workload to the machine group scheduling system; the machine group scheduling system will combine the battery status, spatial location and operation capacity matching degree of the online equipment to perform task replanning and optimize the allocation of the remaining tasks to the remaining online equipment; after the faulty equipment is repaired and passes the self-inspection, it will be reinstated into the operation scheduling pool.

[0014] S5. Job Completion and Data Archiving: After all agricultural machinery robots have completed the current batch of operations, the cloud platform will archive all data of this operation in a hierarchical storage architecture, generate an operation report including equipment energy consumption analysis, abnormal event summary and operating condition distribution statistics, and notify the operation and maintenance terminal and the machine group scheduling system that the operation is completed, so as to realize the full life cycle management of operation data.

[0015] As a preferred method for status monitoring in multi-machine scheduling scenarios of agricultural machinery robots, in step S2, the agricultural machinery robot end identifies the current operation mode in real time based on the collected multi-source operating data, and reports the operation mode label to the cloud platform synchronously along with the real-time operating condition data. The cloud platform then uses a multi-factor adaptive correction method to estimate the battery state of charge (SOC), specifically including:

[0016] The current operation mode is determined based on the characteristics of motor current, vehicle vibration, and positioning driving speed. The operation modes include six categories: tillage, sowing, plant protection, harvesting, migration, and standby.

[0017] The temperature-current coupling efficiency factor and vibration correction factor are used to correct the coulomb meter integral results in real time to obtain a preliminary SOC value. The calculation formula is as follows:

[0018]

[0019] In the formula, Let SOC be the value at time t. The initial SOC value is determined according to the priority of the value saved from the last normal shutdown, the value obtained by reverse lookup of the open circuit voltage after static setting, and the factory default value. This refers to the battery's rated capacity. This is the instantaneous discharge current; This is the temperature-current coupling efficiency factor, used to compensate for the effects of ambient temperature and discharge current on coulombic efficiency; This is a vibration correction factor used to compensate for the impact of changes in battery contact resistance caused by bumpy farmland surfaces on the accuracy of current measurement.

[0020] The formula for calculating the temperature-current coupling efficiency factor is as follows:

[0021]

[0022] In the formula, The baseline coulomb efficiency; This refers to the actual measured temperature of the battery. For reference temperature; , This is the temperature compensation coefficient, obtained from battery characteristic calibration. This is a high-current correction factor used to compensate for the effect of high-current discharge on coulomb efficiency. This is the battery's rated discharge current.

[0023] The formula for calculating the vibration correction factor is as follows:

[0024]

[0025] In the formula, This represents the root mean square value of the vehicle's three-axis accelerometer. The baseline vibration value for the smooth road surface; This is the correction factor for the effect of vibration on contact resistance.

[0026] As a preferred method for state monitoring in multi-machine scheduling scenarios of agricultural machinery robots, the preliminary SOC value is implemented using a discrete recursive approach, and the calculation formula is as follows:

[0027]

[0028] In the formula, Let SOC be the value at the k-th sampling time. The SOC value at the previous sampling time; The median value is taken under high dynamic conditions to suppress transient spikes, and the average value is taken under steady-state conditions to improve computational efficiency. This represents the duration of the current sampling interval;

[0029] Apply physical constraints to the SOC value:

[0030]

[0031] A SOC drift correction mechanism is set up so that when the cumulative correction amount after continuous updates exceeds the preset drift threshold, SOC recalibration is triggered at the time of standby, full charge cutoff, and undervoltage protection to correct the drift error generated by integration.

[0032] As a preferred method for status monitoring in multi-machine scheduling scenarios of agricultural machinery robots, step S2 also includes an open-circuit voltage (OCV) mapping correction step:

[0033] The OCV-SOC mapping relationship is pre-stored. When the battery is detected to be in a static state and the static time meets the preset conditions, the current OCV value is read, and the SOC value is obtained through the mapping relationship. Then, the coulomb SOC value and the OCV SOC value are weighted and fused to obtain the final SOC estimation result. The fusion calculation formula is as follows:

[0034]

[0035] In the formula, The SOC value after fusion; The SOC value is calculated using the coulomb integral method; The SOC value obtained from OCV mapping; This is the weighting coefficient, with a value ranging from 0.7 to 0.9.

[0036] As a preferred method for state monitoring in multi-machine scheduling scenarios of agricultural machinery robots, in step S2, the cloud platform also uses a multi-factor evaluation method that weights and fuses capacity ratio and internal resistance ratio to calculate battery health status (SOH), specifically including:

[0037] The formula for calculating SOH using the capacity ratio method is:

[0038]

[0039] In the formula, The current available capacity of the battery is estimated by integrating the full charge and discharge cycle or by integrating under controlled current. This refers to the battery's rated capacity.

[0040] The formula for calculating SOH using the internal resistance ratio method is:

[0041]

[0042] In the formula, This refers to the battery's nominal internal resistance. The current internal resistance of the battery is calculated using the voltage and current changes under a step load. The calculation formula is as follows:

[0043]

[0044] In the formula, , The voltage and current before the load step change, , The voltage and current after the load step change;

[0045] The combined SOH is obtained by weighting and combining the capacity ratio SOH and the internal resistance ratio SOH, as shown in the formula:

[0046]

[0047] In the formula, , Let be the weighting coefficient, satisfying .

[0048] As a preferred method for status monitoring in multi-machine scheduling scenarios of agricultural robots, the agricultural robot terminal and the cloud platform adopt an adaptive data acquisition and transmission mechanism with signal quality awareness, specifically including:

[0049] The reported data is divided into low-frequency status frames, high-frequency telemetry frames, event alarm frames, and task progress frames, and a differentiated reporting strategy is adopted. The reporting frequency of high-frequency telemetry frames is dynamically adjusted according to the operation mode and signal quality, and the calculation formula is as follows:

[0050]

[0051] In the formula, Based on the reporting frequency; This is the working condition coefficient, with different values ​​set for different working modes; The signal quality coefficient is calculated using the following formula:

[0052]

[0053] In the formula, The current wireless signal strength, This serves as a good reference value for signal strength.

[0054] All reported data includes a timestamp, and the timestamp conversion relationship is as follows: Each frame of data is appended with a frame sequence number and a CRC checksum. The cloud platform detects data loss by using the sequence number continuity and verifies data integrity by using the checksum.

[0055] The agricultural machinery robot maintains a local circular buffer. When a network interruption is detected, data is written to the local buffer. After the network is restored, data is retransmitted in the order of priority of event alarm frames, status frames, and telemetry frames, and the retransmission bandwidth is limited to a preset proportion of the currently available bandwidth.

[0056] The cloud platform determines the online and offline status of devices through a keep-alive mechanism. The keep-alive timeout threshold is calculated using the following formula:

[0057]

[0058] In the formula, The current time; This refers to the time when the device last reported the keep-alive heartbeat packet; As a fault tolerance factor; when the device's last reporting time is earlier than the keep-alive timeout threshold, the device is determined to be offline and an offline alarm is triggered;

[0059] Simultaneously, a data obsolescence determination mechanism is set up, and the calculation formula is as follows:

[0060]

[0061] In the formula, This refers to the last valid data update time of the device. The data staleness duration; when the data staleness duration exceeds the preset threshold, the display status will be downgraded to expired.

[0062] As a preferred method for status monitoring in multi-machine scheduling scenarios of agricultural machinery robots, step S2 involves performing adaptive anomaly detection on real-time operating data, specifically including:

[0063] For each operating mode of each device, an independent parameter baseline distribution is maintained. The baseline includes the mean and standard deviation of each monitored parameter. The baseline parameters are updated online using an exponential moving average method, with the update formula as follows:

[0064]

[0065] In the formula, The baseline mean after the k-th update; The parameter value is the value of the kth sample. This is a smoothing coefficient that is adaptively adjusted based on the stability of the operating conditions.

[0066] For temperature parameters, calculate the rate of temperature change, and generate a dynamic temperature rise threshold by combining the operating mode baseline and the ambient temperature correction item. When the rate of temperature change exceeds the dynamic temperature rise threshold, trigger an alarm for excessively rapid temperature rise.

[0067] The formula for calculating the rate of temperature change is:

[0068]

[0069] In the formula, This represents the temperature value from the kth sample. To calculate the number of span points; The sampling interval;

[0070] The formula for calculating the dynamic temperature rise threshold is:

[0071]

[0072] In the formula, This represents the baseline mean of the temperature rise rate under the corresponding operating mode; This represents the baseline standard deviation of the temperature rise rate under the corresponding operating mode; Confidence coefficient; This is an ambient temperature correction item.

[0073] As a preferred method for status monitoring in multi-machine scheduling scenarios of agricultural machinery robots, step S2 includes performing adaptive anomaly detection on real-time operating data, which also includes voltage drop detection, specifically including:

[0074] The battery terminal voltage is monitored in real time, and the voltage change rate is calculated using the finite difference method. The calculation formula is as follows:

[0075]

[0076] In the formula, Let be the terminal voltage at the k-th sampling time; Let be the terminal voltage at the km-th sampling time; The number of span points;

[0077] Before calculation, the voltage data is denoised using a moving average. The moving average formula is as follows:

[0078]

[0079] In the formula, This is the filtered voltage value; The length of the sliding window;

[0080] By combining current limiting conditions to exclude normal load fluctuations, a voltage descent alarm is triggered when the voltage drop rate exceeds a preset threshold.

[0081] Meanwhile, the window drop amplitude is used as a supplementary criterion, and the calculation formula is as follows:

[0082]

[0083] In the formula, A sliding window voltage sequence of length L; This represents the maximum voltage value within the window. The minimum voltage value within the window; This represents the voltage drop within the window.

[0084] As a preferred solution for status monitoring in multi-machine scheduling scenarios of agricultural robots, it also includes a multi-dimensional correlation anomaly detection and alarm deduplication mechanism:

[0085] When a single-dimensional parameter falls within a suspected anomaly range, a multi-dimensional correlation anomaly score calculation is initiated. The baseline deviation of all multi-dimensional parameters is considered to determine whether a multi-parameter collaborative anomaly alarm should be triggered. The anomaly score calculation formula is as follows:

[0086]

[0087] In the formula, The weight of the i-th parameter; Let be the measured value of the i-th parameter; This represents the baseline mean of the i-th parameter under the corresponding work mode; This represents the baseline standard deviation of the i-th parameter under the corresponding work mode;

[0088] Alarms are divided into three levels: emergency, important, and general. Emergency alarms trigger power limiting or shutdown protection actions of the equipment and are pushed synchronously through multiple channels; important alarms generate maintenance work orders and push them to the responsible personnel; general alarms are temporarily stored in the observation queue.

[0089] The alarm deduplication suppression window mechanism is adopted, and the formula for calculating the suppression window start time is as follows:

[0090]

[0091] In the formula, This is the current alarm trigger time; The base duration for deduplication suppression window; The start time of the deduplication window; If there are similar unprocessed alarms within the interval, repeat pushes will be suppressed.

[0092] The suppression window duration is dynamically adjusted, and the calculation formula is as follows:

[0093]

[0094] In the formula, This is the base suppression duration corresponding to the alarm level; The growth coefficient is the factor that suppresses repetition. This sets the number of times the same type of alarm will be triggered within a preset period; if the same type of alarm is triggered frequently, the suppression window will be automatically extended and the alarm level will be upgraded.

[0095] This invention also provides a status monitoring system for multi-machine scheduling scenarios of agricultural machinery robots, employing the aforementioned status monitoring method for multi-machine scheduling scenarios of agricultural machinery robots, including:

[0096] The device online and working condition initialization module is used to establish a long communication connection with the cloud platform after the agricultural machinery robot is started, and to report basic device information and initial self-test data to the cloud platform. After the cloud platform completes the device identity authentication of the agricultural machinery robot, it loads the corresponding working condition parameter baseline configuration according to the device model, initializes the real-time data transmission channel and updates the device online status.

[0097] The adaptive real-time monitoring module is used by the agricultural machinery robot to identify the current operation mode in real time based on the collected multi-source operation data, and to report the operation mode tag to the cloud platform along with the real-time operation data. The cloud platform matches the corresponding parameter baseline and dynamic alarm threshold according to the received operation mode tag, performs adaptive anomaly detection on the real-time operation data, and displays the operation status in two dimensions: single device and machine group, so as to achieve full-domain real-time perception of the operation status of multiple machines.

[0098] The hierarchical alarm and intelligent operation and maintenance linkage module is used to classify alarm levels according to the severity and scope of impact of the anomaly detection results of the cloud platform when the anomaly detection results exceed the corresponding dynamic alarm threshold, and to execute the corresponding linkage handling strategy. At the same time, it generates a structured work order containing device identifier, alarm type, operating condition context and handling suggestions and pushes it to the operation and maintenance terminal, forming a full-process operation and maintenance management from alarm triggering, work order dispatch, on-site handling to confirmation.

[0099] The multi-machine scheduling and linkage module is used to synchronize the location information and remaining workload of the faulty equipment to the machine group scheduling system when the operation is interrupted due to an emergency alarm triggered by a fault on the agricultural machinery robot end. The machine group scheduling system performs task replanning by combining the battery status, spatial location and operation capacity matching degree of the online equipment, and optimizes the allocation of the remaining operation tasks to the remaining online equipment. After the faulty equipment is repaired and passes the self-test, it is reinstated into the operation scheduling pool.

[0100] The task completion and data archiving module is used to archive all data of the current batch of agricultural machinery robots in a hierarchical storage architecture after all agricultural machinery robots have completed the current batch of tasks. This generates a task report that includes equipment energy consumption analysis, abnormal event summary and working condition distribution statistics. The report then notifies the operation and maintenance terminal and the machine group scheduling system that the task has been completed, thereby realizing full lifecycle management of task data.

[0101] The present invention has the following advantages:

[0102] This invention utilizes a multi-factor Coulomb calculation method that combines operating condition identification with temperature-current coupling efficiency compensation and vibration contact resistance correction, along with a static open-circuit voltage fusion correction mechanism. This effectively counteracts the interference of complex operating conditions such as large day-night temperature differences in farmland, bumpy roads, and intermittent discharge of large currents, significantly improving the estimation accuracy of battery state of charge and health status, and providing a reliable data foundation for equipment operation and maintenance and scheduling decisions.

[0103] This invention uses a dynamic baseline threshold mechanism based on operating modes to automatically distinguish between normal operating load fluctuations and abnormal equipment degradation, thereby reducing false alarms at the source. Combined with a multi-dimensional correlation anomaly scoring mechanism, it can capture early compound degradation trends where single indicator changes are not significant, thus upgrading from reactive alarms to proactive early warnings.

[0104] This invention employs signal quality-aware adaptive reporting frequency adjustment, combined with local caching for network outage resumption and hierarchical priority retransmission strategies. It adapts to the working environment with weak communication coverage in farmland, effectively reducing data loss caused by network fluctuations, ensuring the continuity and consistency of monitoring data, and the data retransmission does not occupy the bandwidth of real-time data streams.

[0105] This invention significantly shortens the time from anomaly detection to maintenance response through hierarchical alarms, automatic work order arrangement, and closed-loop management throughout the entire lifecycle, thereby significantly improving fault handling efficiency. At the same time, the fault information is linked with the machine group scheduling system, which can automatically redistribute the remaining tasks, ensuring the continuity of work tasks during the busy farming season, reducing production losses caused by equipment failures, and realizing full-process automation of monitoring, alarms, maintenance, and scheduling. Attached Figure Description

[0106] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0107] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0108] Figure 1 This is a flowchart of the status monitoring method for multi-machine scheduling of agricultural robots provided in this embodiment of the invention;

[0109] Figure 2 This is a schematic diagram of the technical architecture of the status monitoring method for multi-machine scheduling of agricultural robots provided in this embodiment of the invention;

[0110] Figure 3 This is a schematic diagram of the status monitoring system architecture for a multi-machine scheduling scenario of agricultural machinery robots provided in an embodiment of the present invention. Detailed Implementation

[0111] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0112] Example 1

[0113] See Figure 1 and Figure 2 Embodiment 1 of the present invention provides a status monitoring method for multi-machine scheduling of agricultural machinery robots, including the following steps:

[0114] S1. Equipment Online and Operating Condition Initialization: After the agricultural machinery robot is started, it establishes a long communication connection with the cloud platform and reports basic equipment information and initial self-test data to the cloud platform. After the cloud platform completes the equipment identity authentication of the agricultural machinery robot, it loads the corresponding operating condition parameter baseline configuration according to the equipment model, initializes the real-time data transmission channel, and updates the equipment online status.

[0115] Specifically, in large-scale agricultural machinery operations, a single farm often operates multiple models of agricultural robots simultaneously. These different models exhibit significant differences in parameters such as battery capacity, motor power, and component heat resistance thresholds. Using a uniform monitoring benchmark would directly lead to systematic deviations in subsequent status estimation and anomaly detection. Therefore, after device authentication, the cloud platform matches a dedicated baseline configuration of operating parameters for each device model, ensuring from the outset that the monitoring standards match the device's characteristics. Simultaneously, it employs a long-connection access method to meet the low-latency requirements of real-time farmland monitoring. Compared to the traditional short-connection polling mode, long connections enable instant data uploads and instant command issuance, providing a stable communication link for real-time operating condition data transmission and emergency protection command issuance. Upon initialization, the online status of the devices is updated synchronously, allowing dispatchers to quickly grasp the scale of the dispatchable robot fleet.

[0116] S2. Adaptive Real-time Monitoring: The agricultural machinery robot terminal identifies the current operation mode in real time based on the collected multi-source operation data, and reports the operation mode tag to the cloud platform synchronously along with the real-time operation data; the cloud platform matches the corresponding parameter baseline and dynamic alarm threshold according to the received operation mode tag, performs adaptive anomaly detection on the real-time operation data, and displays the operation status in two dimensions: single device and machine group, so as to achieve full-domain real-time perception of the operation status of multiple machines.

[0117] Specifically, agricultural robots exhibit fundamentally different operating characteristics under different work modes. In tillage and harvesting modes, the load is heavy, current fluctuations are large, and component temperatures rise rapidly, representing highly dynamic operating conditions. In sowing and plant protection modes, the load is moderate and stable. In relocation modes, the walking load is light, and the working mechanism has no load. In standby mode, there is almost no power output. If a uniform alarm threshold is used, either frequent false alarms will occur under high load conditions, or early faults will be missed under light load conditions. Therefore, this step first identifies the operating mode on the edge and tags it along with the data. The cloud then automatically matches the corresponding operating condition's parameter baseline and dynamic threshold, ensuring that anomaly detection is always performed under the same operating condition benchmark. This fundamentally distinguishes between normal load fluctuations and abnormal equipment degradation, significantly improving detection accuracy. Simultaneously, a dual-dimensional display of single device and cluster views is adopted, satisfying both the needs of maintenance personnel to view detailed parameters of individual devices and the needs of dispatchers to grasp the overall cluster operation status of the entire work area. This achieves full coverage perception from single-point devices to the entire cluster, adapting to the global management needs of multi-machine scheduling.

[0118] S3. Tiered Alarm and Intelligent Operation and Maintenance Linkage: When the anomaly detection result of the cloud platform exceeds the corresponding dynamic alarm threshold, the cloud platform classifies the alarm level according to the severity and scope of the anomaly and executes the corresponding linkage handling strategy; at the same time, it generates a structured work order containing device identifier, alarm type, operating condition context and handling suggestions and pushes it to the operation and maintenance terminal, forming a full-process operation and maintenance management from alarm triggering, work order dispatch, on-site handling to confirmation.

[0119] Specifically, the risk levels of abnormal faults in agricultural machinery operation scenarios vary greatly. Severe faults such as battery overheating and sudden voltage drops can directly lead to equipment burnout, requiring immediate triggering of protection mechanisms. Mid-term anomalies with parameters consistently deviating from the baseline require investigation by maintenance personnel; minor, occasional exceedances only require observation without intervention. If all alarms are pushed and handled in the same way, urgent faults will be overwhelmed by a large number of low-priority alarms, and maintenance personnel will be frequently disturbed by invalid information. This step classifies alarm levels according to the severity of the anomaly and matches them with differentiated linkage strategies, ensuring that high-risk faults receive the fastest response while avoiding inefficient use of maintenance resources. The automatic generation of structured work orders with operating condition context effectively transforms alarm information into directly executable maintenance tasks. The work order includes the operating mode, environmental conditions, parameter trends, and suggested handling steps at the time of the fault. After receiving the order, maintenance personnel no longer need to sift through large amounts of historical data or rely on personal experience to determine the cause of the fault; they can directly follow the instructions to handle the situation, significantly improving fault handling efficiency and achieving full traceability from fault discovery to handling.

[0120] S4. Multi-machine scheduling and linkage: When the agricultural machinery robot is interrupted due to a fault triggering an emergency alarm, the cloud platform will synchronize the location information of the faulty equipment and the remaining workload to the machine group scheduling system. The machine group scheduling system will combine the battery status, spatial location and work capacity matching degree of the online equipment to perform task replanning and optimize the allocation of the remaining work tasks to the remaining online equipment. After the faulty equipment is repaired and passes the self-inspection, it will be reinstated into the work scheduling pool.

[0121] Specifically, agricultural production is highly time-sensitive, with planting and harvesting windows often lasting only a few days. If manual coordination is relied upon to fill in the gaps after a single agricultural machine malfunctions, the response is slow, and task allocation can easily become unreasonable, delaying the overall work progress. In this step, after the monitoring system detects an emergency fault, it automatically synchronizes the location and remaining workload of the faulty equipment to the dispatch system. The dispatch system automatically completes the optimal allocation of remaining tasks based on principles such as ensuring battery life, minimizing idling losses by choosing the nearest available location, and matching work capacity to ensure work quality. This quickly fills the work gaps without manual intervention. After the faulty equipment passes its repair self-test, it is automatically reinstated into the dispatch pool, eliminating the need for manual addition of equipment and task reset. This reduces the operational burden on personnel, ensuring that the equipment can be put into operation immediately after recovery, maximizing the overall operational efficiency of the fleet.

[0122] S5. Job Completion and Data Archiving: After all agricultural machinery robots have completed the current batch of operations, the cloud platform will archive all data of this operation in a hierarchical storage architecture, generate an operation report including equipment energy consumption analysis, abnormal event summary and operating condition distribution statistics, and notify the operation and maintenance terminal and the machine group scheduling system that the operation is completed, so as to realize the full life cycle management of operation data.

[0123] Specifically, real-time monitoring demands low latency for data read and write operations. However, storing all historical data in high-performance storage media is not only costly but also slows down real-time queries. Therefore, a tiered storage architecture is adopted. During job execution, a hot-layer cache ensures real-time read and write performance. After job completion, data is migrated to a warm-layer time-series database and a cold-layer object storage based on access frequency. This reduces long-term storage costs and prevents historical data from crowding out real-time computing resources, ensuring the monitoring performance of subsequent batches of jobs. Simultaneously, multi-dimensional job reports are generated, transforming operational data into optimization criteria. By analyzing the energy consumption level, frequency of abnormal events, and operating condition distribution of individual devices, potential health risks can be identified, leading to optimized maintenance plans. Overall cluster job data statistics allow for optimization of subsequent task scheduling strategies and cluster configurations. This ensures that monitoring data not only serves current fault handling but also provides data support for long-term operation and maintenance optimization and scheduling upgrades. Finally, synchronized notifications to the operation and maintenance and scheduling ends facilitate timely monitoring of job progress and coordinated planning for the next batch of tasks.

[0124] In one possible embodiment, in step S2, the agricultural machinery robot identifies the current operating mode in real time based on the collected multi-source operating data, and after synchronously reporting the operating mode label to the cloud platform along with the real-time operating condition data, the cloud platform uses a multi-factor operating condition adaptive correction method to estimate the battery state of charge (SOC), specifically including:

[0125] The current operation mode is determined based on the characteristics of motor current, vehicle vibration, and positioning driving speed. The operation modes include six categories: tillage, sowing, plant protection, harvesting, migration, and standby.

[0126] The temperature-current coupling efficiency factor and vibration correction factor are used to correct the coulomb meter integral results in real time to obtain a preliminary SOC value. The calculation formula is as follows:

[0127]

[0128] In the formula, Let SOC be the value at time t. The initial SOC value is determined according to the priority of the value saved from the last normal shutdown, the value obtained by reverse lookup of the open circuit voltage after static setting, and the factory default value. This refers to the battery's rated capacity. This is the instantaneous discharge current; This is the temperature-current coupling efficiency factor, used to compensate for the effects of ambient temperature and discharge current on coulombic efficiency; This is a vibration correction factor used to compensate for the impact of changes in battery contact resistance caused by bumpy farmland surfaces on the accuracy of current measurement.

[0129] Specifically, battery SOC estimation is based on the coulomb integration method. This method calculates the discharged capacity by integrating the discharge current over time and then derives the remaining capacity based on the rated capacity. However, the accuracy of this method is highly dependent on the coulomb efficiency and the accuracy of current sampling. In farmland operations, the diurnal temperature range can exceed 30°C, causing significant changes in battery electrochemical activity with temperature, leading to a deviation of the coulomb efficiency from the nominal value. Frequent high-current surges during operation exacerbate battery polarization, further reducing coulomb efficiency. Simultaneously, the severe bumps on unpaved farmland surfaces cause transient changes in the contact resistance of the battery terminals and internal tabs, resulting in deviations in current sampling. Most existing technologies use fixed-parameter coulomb integration, completely neglecting these unique farmland operating conditions, thus leading to significant errors in actual operation. This solution introduces a temperature-current coupling efficiency factor to correct for operating condition fluctuations in coulomb efficiency, and introduces a vibration correction factor to compensate for vibration interference in current sampling. It improves the integration accuracy under complex operating conditions from two dimensions: efficiency correction and sampling correction. The initial SOC adopts a three-level priority strategy, which can obtain the initial value closest to the true value under different startup scenarios, reduce the cumulative impact of initial error on integration, and reduce estimation deviation from the source.

[0130] The formula for calculating the temperature-current coupling efficiency factor is as follows:

[0131]

[0132] In the formula, The baseline coulomb efficiency; This refers to the actual measured temperature of the battery. For reference temperature; , This is the temperature compensation coefficient, obtained from battery characteristic calibration. This is a high-current correction factor used to compensate for the effect of high-current discharge on coulomb efficiency. The rated discharge current of the battery is used. Coulombic efficiency exhibits a non-linear relationship with temperature; at low temperatures, electrochemical activity decreases and side reactions intensify, leading to reduced efficiency. At high temperatures, accelerated side reactions also cause efficiency shifts, indicating a non-linear change. A quadratic polynomial is used to fit the effect of temperature on efficiency. The first-order coefficient reflects the linear temperature coefficient, while the quadratic coefficient captures the non-linear shift, ensuring temperature compensation covers a wide temperature range in agricultural settings and maintaining accuracy even under severe diurnal temperature variations. During high-current discharge, concentration polarization and ohmic polarization significantly increase, and usable capacity decreases with increasing discharge rate. Furthermore, polarization loss is approximately proportional to the square of the current. A high-current correction term is constructed using the square of the current rate, accurately reflecting the capacity decay effect caused by high-current impacts during deep plowing and hill climbing in agricultural machinery operations. Multiplying these two corrections achieves the coupling effect of temperature and current, closely reflecting the real-world operating conditions where temperature and load change simultaneously in agricultural machinery operations, resulting in higher accuracy compared to single-factor corrections.

[0133] The formula for calculating the vibration correction factor is as follows:

[0134]

[0135] In the formula, This represents the root mean square value of the vehicle's three-axis accelerometer. The baseline vibration value for the smooth road surface; This is a correction factor for the impact of vibration on contact resistance. The root mean square value of vehicle body vibration acceleration can quantitatively reflect the degree of road bumpiness. When the vibration intensity is lower than the baseline of a smooth road surface, the contact resistance of the battery connection structure is in a stable state and does not require correction. Threshold-triggered correction is implemented to avoid overcompensation introducing additional errors under stable operating conditions. When the vibration intensity exceeds the baseline, the bumps can cause fretting wear and momentary poor contact in the battery terminals and connectors, increasing the equivalent contact resistance and causing measurement deviation in the current sampling circuit. The degree of deviation is approximately linearly positively correlated with the vibration intensity. A linear coefficient is used... The vibration increment is proportionally corrected to offset the current measurement error caused by the change in contact resistance and improve the accuracy of current integration.

[0136] In one possible embodiment, the initial SOC value is implemented using a discrete recursive method, and the calculation formula is as follows:

[0137]

[0138] In the formula, Let SOC be the value at the k-th sampling time. The SOC value at the previous sampling time; The median value is taken under high dynamic conditions to suppress transient spikes, and the average value is taken under steady-state conditions to improve computational efficiency. This represents the duration of the current sampling interval;

[0139] Apply physical constraints to the SOC value:

[0140]

[0141] A SOC drift correction mechanism is set up so that when the cumulative correction amount after continuous updates exceeds the preset drift threshold, SOC recalibration is triggered at the time of standby, full charge cutoff, and undervoltage protection to correct the drift error generated by integration.

[0142] Specifically, discrete recursion is an engineering implementation of the continuous integral formula, adapted to the discrete sampling processing mode of vehicle controllers and cloud streaming computing. Each sampling cycle only needs to be calculated based on the result of the previous moment and the current sampled value, resulting in low computational load and high real-time performance, making it suitable for deployment in embedded devices and cloud real-time computing scenarios. The equivalent current adopts a condition-differentiated strategy, balancing accuracy and computing power. Under high-dynamic conditions such as farming and harvesting, the current has many transient spikes; directly using instantaneous values ​​will lead to integration deviations. Median filtering can effectively suppress pulse interference and retain the overall trend of load changes. Under steady-state conditions such as transition and standby, the current is stable, and the mean calculation has sufficient accuracy and higher computational efficiency, achieving a dynamic balance between accuracy and computing power under different conditions. The clip physical constraint is based on the physical properties of battery SOC; the remaining charge cannot exceed the range of 0-100%. This enforces a constraint on the boundary overflow of the integration calculation, avoiding physically meaningless values. The drift correction mechanism is used to solve the inherent error accumulation problem of the coulomb integral method. After long-term operation, the sampling error will continue to accumulate, causing SOC drift. Therefore, recalibration is performed at three times when the true value of SOC is known or can be accurately measured: standby, full charge, and undervoltage. This periodically eliminates the accumulated error and ensures the estimation accuracy of the equipment during long-term continuous operation.

[0143] In one possible embodiment, step S2 further includes an open-circuit voltage (OCV) mapping correction step:

[0144] The OCV-SOC mapping relationship is pre-stored. When the battery is detected to be in a static state and the static time meets the preset conditions, the current OCV value is read, and the SOC value is obtained through the mapping relationship. Then, the coulomb SOC value and the OCV SOC value are weighted and fused to obtain the final SOC estimation result. The fusion calculation formula is as follows:

[0145]

[0146] In the formula, The SOC value after fusion; The SOC value is calculated using the coulomb integral method; The SOC value obtained from OCV mapping; This is the weighting coefficient, with a value ranging from 0.7 to 0.9.

[0147] Specifically, the open-circuit voltage (OCV) is the terminal voltage of a battery when it reaches thermodynamic equilibrium. It has a one-to-one inherent mapping relationship with the state of charge (SOC), determined by the thermodynamic properties of the battery electrode materials. Static measurement offers high accuracy, but requires the battery to be left to stand for a sufficiently long time to eliminate polarization voltage and obtain the true OCV. Therefore, it cannot be directly used for real-time estimation during dynamic operations. The coulomb integral method has good dynamic response and can track changes in charge in real time, but it suffers from cumulative drift. The OCV method offers high static accuracy and can be used as a calibration benchmark, but it is only effective when the battery is stationary. Weighted fusion assigns higher weight to the coulomb meter results. Since agricultural machinery operates under dynamic conditions for most of the time, the real-time performance of the coulomb meter is a core requirement. OCV is only used as an error correction term, calibrating cumulative drift during stationary periods. This ensures both real-time dynamic monitoring and effectively suppresses long-term drift, achieving a balance between dynamic accuracy and long-term stability.

[0148] In one possible embodiment, in step S2, the cloud platform further calculates the battery health status (SOH) using a multi-factor evaluation method that weights and fuses capacity ratio and internal resistance ratio, specifically including:

[0149] The formula for calculating SOH using the capacity ratio method is:

[0150]

[0151] In the formula, The current available capacity of the battery is estimated by integrating the full charge and discharge cycle or by integrating under controlled current. This refers to the battery's rated capacity.

[0152] The formula for calculating SOH using the internal resistance ratio method is:

[0153]

[0154] In the formula, This refers to the battery's nominal internal resistance. The current internal resistance of the battery is calculated using the voltage and current changes under a step load. The calculation formula is as follows:

[0155]

[0156] In the formula, , The voltage and current before the load step change, , The voltage and current after the load step change;

[0157] The combined SOH is obtained by weighting and combining the capacity ratio SOH and the internal resistance ratio SOH, as shown in the formula:

[0158]

[0159] In the formula, , Let be the weighting coefficient, satisfying .

[0160] Specifically, capacity decay directly determines battery life, but complete capacity testing has a long cycle and is difficult to update frequently, making it suitable as a long-term health assessment indicator. Increased internal resistance reflects the intensification of ohmic and electrochemical polarization in the battery, which is directly related to power output capability and heat generation risk. It can also be quickly measured online via load step, making it suitable for capturing early degradation trends. The principle of measuring internal resistance using the step load method is that the battery terminal voltage consists of the equilibrium voltage, ohmic polarization voltage, and concentration polarization voltage. At the moment of load step, concentration polarization has not yet been established, and the voltage change is entirely caused by ohmic internal resistance. Therefore, the DC internal resistance can be quickly estimated by the ratio of the instantaneous voltage change to the current change, without the need for long-term testing, making it suitable for online monitoring. The weighted fusion method combines the advantages of both types of indicators. The capacity indicator has a higher weight, aligning with the core requirements of agricultural machinery for battery life, while the internal resistance indicator serves as a supplement, reflecting early battery degradation trends and heat generation risk. This enables a multi-dimensional, full-cycle assessment of battery health, providing a comprehensive basis for maintenance, replacement, and lifespan prediction.

[0161] In one possible embodiment, the agricultural machinery robot terminal and the cloud platform employ a signal quality-aware adaptive data acquisition and transmission mechanism, specifically including:

[0162] The reported data is divided into low-frequency status frames, high-frequency telemetry frames, event alarm frames, and task progress frames, and a differentiated reporting strategy is adopted. The reporting frequency of high-frequency telemetry frames is dynamically adjusted according to the operation mode and signal quality, and the calculation formula is as follows:

[0163]

[0164] In the formula, Based on the reporting frequency; This is the working condition coefficient, with different values ​​set for different working modes; The signal quality coefficient is calculated using the following formula:

[0165]

[0166] In the formula, The current wireless signal strength, This serves as a good reference value for signal strength.

[0167] All reported data includes a timestamp, and the timestamp conversion relationship is as follows: Each frame of data is appended with a frame sequence number and a CRC checksum. The cloud platform detects data loss by using the sequence number continuity and verifies data integrity by using the checksum.

[0168] The agricultural machinery robot maintains a local circular buffer. When a network interruption is detected, data is written to the local buffer. After the network is restored, data is retransmitted in the order of priority of event alarm frames, status frames, and telemetry frames, and the retransmission bandwidth is limited to a preset proportion of the currently available bandwidth.

[0169] The cloud platform determines the online and offline status of devices through a keep-alive mechanism. The keep-alive timeout threshold is calculated using the following formula:

[0170]

[0171] In the formula, The current time; This refers to the time when the device last reported the keep-alive heartbeat packet; As a fault tolerance factor; when the device's last reporting time is earlier than the keep-alive timeout threshold, the device is determined to be offline and an offline alarm is triggered;

[0172] Simultaneously, a data obsolescence determination mechanism is set up, and the calculation formula is as follows:

[0173]

[0174] In the formula, This refers to the last valid data update time of the device. The data staleness duration; when the data staleness duration exceeds the preset threshold, the display status will be downgraded to expired.

[0175] Specifically, this mechanism is designed for communication environments with weak signals and large network fluctuations in farmland. It classifies data according to importance and dynamically adjusts the reporting speed according to network quality, ensuring the reliability of critical data under limited bandwidth and solving the problems of severe packet loss and poor data integrity in existing fixed-frequency reporting under weak signal conditions. The principle of data classification and reporting is that different data have different real-time and reliability requirements. Event alarm frames are related to fault handling and must be delivered immediately and reliably; status frames reflect the overall status of the equipment and can be reported at a low frequency; telemetry frames are used for detailed trend analysis and their frequency can be dynamically adjusted. In the dual-factor adjustment of the reporting frequency, the operating condition coefficient increases the sampling density in high-dynamic operating scenarios to ensure that operating details are not lost; the signal coefficient reduces the reporting frequency of non-critical data under weak signal conditions to avoid retransmission storms caused by a large number of packet losses, and the lower limit of 0.3 ensures the minimum monitoring density and avoids complete loss of equipment status awareness.

[0176] Timestamps, sequence numbers, and CRC checks are fundamental guarantees of data integrity. A unified timestamp ensures time-series alignment of multi-source data, sequence numbers detect packet loss, and CRC checks identify transmission errors. Local ring buffers and prioritized retransmission address data retention issues in network outage scenarios. The ring structure automatically overwrites the oldest data, adapting to limited local storage. Prioritized retransmission ensures priority recovery of critical data such as fault alarms, while bandwidth limiting prevents retransmission traffic from crowding out real-time data bandwidth, ensuring current monitoring remains unaffected. The fault tolerance factor in the keep-alive mechanism resists instantaneous network jitter, preventing false offline judgments due to brief packet loss. Data staleness determination distinguishes between online but stagnant data and normal online status, preventing schedulers from making incorrect decisions based on expired data.

[0177] In one possible embodiment, step S2, performing adaptive anomaly detection on the real-time operating data, specifically includes:

[0178] For each operating mode of each device, an independent parameter baseline distribution is maintained. The baseline includes the mean and standard deviation of each monitored parameter. The baseline parameters are updated online using an exponential moving average method, with the update formula as follows:

[0179]

[0180] In the formula, The baseline mean after the k-th update; The parameter value is the value of the kth sample. This is a smoothing coefficient that is adaptively adjusted based on the stability of the operating conditions.

[0181] For temperature parameters, calculate the rate of temperature change, and generate a dynamic temperature rise threshold by combining the operating mode baseline and the ambient temperature correction item. When the rate of temperature change exceeds the dynamic temperature rise threshold, trigger an alarm for excessively rapid temperature rise.

[0182] The formula for calculating the rate of temperature change is:

[0183]

[0184] In the formula, This represents the temperature value from the kth sample. To calculate the number of span points; The sampling interval;

[0185] The formula for calculating the dynamic temperature rise threshold is:

[0186]

[0187] In the formula, This represents the baseline mean of the temperature rise rate under the corresponding operating mode; This represents the baseline standard deviation of the temperature rise rate under the corresponding operating mode; Confidence coefficient; This is an ambient temperature correction item.

[0188] Specifically, different operating modes have fundamentally different load levels and parameter fluctuation ranges. For example, high motor current and rapid temperature rise are normal phenomena in tillage mode. Using a uniform threshold would inevitably generate a large number of false alarms. Therefore, baselines are established separately for each equipment and operating mode to ensure that the alarm threshold always matches the normal fluctuation range of the current operating condition, thereby reducing the false alarm rate at its source. An exponential moving average is used to update the baseline, allowing it to slowly adapt to parameter drift caused by normal aging and wear of the equipment, while avoiding significant deviations due to single abnormal sampling values, ensuring the stability and adaptability of the baseline. Adaptive smoothing coefficients can reduce the update speed to improve baseline stability when the operating condition is stable, and accelerate the update speed to quickly adapt to new operating conditions when the operating condition changes.

[0189] Among these methods, temperature anomalies are detected using the rate of temperature rise rather than absolute value detection. The principle is that early equipment failures often manifest as increased heating and a faster temperature rise, at which point the absolute temperature may still be within a safe range, making it impossible to detect with an absolute value alarm. By calculating the rate of temperature rise through differential calculation of multiple sampling points, abnormal temperature rising trends can be captured, enabling early fault warnings. The dynamic threshold is designed based on the 3σ principle of a normal distribution. Setting the threshold to 3 can cover 99.7% of normal fluctuations, resulting in an extremely low false alarm rate under normal operating conditions. An ambient temperature correction term is added because the equipment's base temperature is high in high-temperature environments, and the normal temperature rise rate will also increase accordingly. Appropriately relaxing the threshold can avoid false alarms caused by environmental factors, allowing the threshold to adapt to both operating conditions and environmental variables.

[0190] In one possible embodiment, step S2, which involves performing adaptive anomaly detection on the real-time operating data, further includes voltage descent detection, specifically including:

[0191] The battery terminal voltage is monitored in real time, and the voltage change rate is calculated using the finite difference method. The calculation formula is as follows:

[0192]

[0193] In the formula, Let be the terminal voltage at the k-th sampling time; Let be the terminal voltage at the km-th sampling time; The number of span points;

[0194] Before calculation, the voltage data is denoised using a moving average. The moving average formula is as follows:

[0195]

[0196] In the formula, This is the filtered voltage value; The length of the sliding window;

[0197] By combining current limiting conditions to exclude normal load fluctuations, a voltage descent alarm is triggered when the voltage drop rate exceeds a preset threshold.

[0198] Meanwhile, the window drop amplitude is used as a supplementary criterion, and the calculation formula is as follows:

[0199]

[0200] In the formula, A sliding window voltage sequence of length L; This represents the maximum voltage value within the window. The minimum voltage value within the window; This represents the voltage drop within the window.

[0201] Specifically, voltage sag is a typical symptom of serious faults such as internal battery short circuits, loose connections, and poor contact. These faults develop rapidly and pose a high risk, requiring quick identification. Therefore, a multi-dimensional voltage sag detection mechanism is designed. The finite difference method is used to calculate the voltage change rate, quantifying the speed of voltage drop and distinguishing between slow discharge and sudden sags. A moving average filter is used to eliminate sampling noise and small fluctuations, avoiding false triggers caused by noise. Combined with current limiting conditions, a sudden increase in normal load can also cause a momentary drop in battery terminal voltage. This voltage drop is a normal operating condition fluctuation; by determining whether the current increases synchronously, normal voltage drops caused by load changes can be excluded, significantly reducing the false alarm rate. The window drop amplitude serves as a supplementary criterion. Some faults manifest as a short-duration deep drop followed by a small rebound. A single rate of change criterion may miss the peak value due to the sampling interval. The window amplitude criterion can capture the maximum drop depth within the sliding window, covering different forms of voltage anomalies. The two criteria can be flexibly configured in parallel (triggering when either is met, improving sensitivity) or in series (triggering only when both are met, reducing the false alarm rate), adapting to monitoring needs of different risk levels.

[0202] One possible implementation also includes a multi-dimensional correlation anomaly detection and alarm deduplication mechanism:

[0203] When a single-dimensional parameter falls within a suspected anomaly range, a multi-dimensional correlation anomaly score calculation is initiated. The baseline deviation of all multi-dimensional parameters is considered to determine whether a multi-parameter collaborative anomaly alarm should be triggered. The anomaly score calculation formula is as follows:

[0204]

[0205] In the formula, The weight of the i-th parameter; Let be the measured value of the i-th parameter; This represents the baseline mean of the i-th parameter under the corresponding work mode; This represents the baseline standard deviation of the i-th parameter under the corresponding work mode;

[0206] Alarms are divided into three levels: emergency, important, and general. Emergency alarms trigger power limiting or shutdown protection actions of the equipment and are pushed synchronously through multiple channels; important alarms generate maintenance work orders and push them to the responsible personnel; general alarms are temporarily stored in the observation queue.

[0207] The alarm deduplication suppression window mechanism is adopted, and the formula for calculating the suppression window start time is as follows:

[0208]

[0209] In the formula, This is the current alarm trigger time; The base duration for deduplication suppression window; The start time of the deduplication window; If there are similar unprocessed alarms within the interval, repeat pushes will be suppressed.

[0210] The suppression window duration is dynamically adjusted, and the calculation formula is as follows:

[0211]

[0212] In the formula, This is the base suppression duration corresponding to the alarm level; The growth coefficient is the factor that suppresses repetition. This sets the number of times the same type of alarm will be triggered within a preset period; if the same type of alarm is triggered frequently, the suppression window will be automatically extended and the alarm level will be upgraded.

[0213] Specifically, the multidimensional correlation anomaly detection system is designed to address the concealment of early-stage faults. In the early stages of equipment degradation, multiple parameters often deviate slightly from their normal ranges simultaneously, but none of the individual parameters reach the alarm threshold, making traditional single-parameter detection ineffective. This anomaly scoring formula, based on the chi-square test, standardizes the deviation of each parameter into a relative deviation in units of standard deviation, squares the deviations, and then sums them with weights to comprehensively reflect the overall degree of deviation. When the score exceeds the threshold, it indicates a coordinated degradation trend among multiple parameters, triggering an alarm even if a single parameter does not exceed its limit. This enables early detection of faults and improves the foresight of anomaly discovery.

[0214] The alarm grading mechanism is designed based on the severity of the fault and the priority of handling, and allocates maintenance resources differently: emergency alarms directly trigger on-site protection to prevent equipment damage from escalating; important alarms drive closed-loop handling of maintenance work orders; and general alarms are only recorded for observation to avoid excessive occupational maintenance effort. An alarm deduplication and suppression window is used to prevent repeated pushes of the same persistent fault, reducing interference with maintenance personnel; it only suppresses unhandled alarms to ensure that new alarms can be pushed normally when the fault recurs after handling. The dynamic window adjustment mechanism works by the fact that frequent triggering of similar alarms indicates that the fault persists and may worsen. Extending the suppression window avoids information bombardment, while upgrading the alarm level prompts maintenance personnel to pay attention to the degradation trend, achieving linkage and adaptation between alarm frequency and alarm level.

[0215] The application scenarios of this invention are as follows:

[0216] Multi-machine collaborative operation scenario in large-scale farms / agricultural cooperatives:

[0217] This scenario utilizes contiguous farmland of thousands of acres or more as its framework, where farms or cooperatives simultaneously dispatch dozens to hundreds of agricultural robots to collaboratively complete the entire process of plowing, planting, managing, and harvesting. The core requirement is centralized and visualized unified management and control of the machine fleet. Traditional manual inspections and simple positioning systems cannot support real-time status perception of large-scale machine fleets, easily leading to problems such as monitoring blind spots, data asynchrony, and delayed fault response. This invention provides unified access to all agricultural machinery equipment through a cloud platform, relying on a hot-warm-cold tiered storage architecture to support high-concurrency data writing and low-latency queries. The real-time operating status is displayed on a large status screen in both individual device and machine fleet dimensions. When a single device fails and stops, the automatic linkage scheduling system redistributes the remaining tasks, achieving centralized and automated management and control of hundreds of machine fleets, significantly reducing the labor costs of machine fleet management and ensuring the overall progress of large-scale operations.

[0218] Complex terrain and harsh working conditions in farmland operations:

[0219] This scenario encompasses hilly terrain, farmland with uneven soil hardness, and northern and plateau farmland with significant diurnal temperature variations. During agricultural machinery operations, load fluctuations are drastic, and the operating conditions of core components such as batteries and motors frequently change with terrain and soil conditions. Traditional fixed-threshold monitoring solutions cannot distinguish between normal load fluctuations and equipment anomalies, resulting in a very high false alarm rate. Furthermore, fixed-parameter battery power estimation has large errors, making it impossible to accurately predict battery life and easily leading to equipment breakdowns due to power outages. This invention uses an end-side operating condition identification module to determine operating modes such as tillage, climbing, and relocation in real time. It employs a multi-factor adaptive SOC algorithm to simultaneously correct for interference from ambient temperature, road vibration, and high-current surges on battery power estimation. Simultaneously, it dynamically adjusts alarm thresholds based on the operating condition baseline, automatically eliminating parameter changes caused by normal load fluctuations. Even under complex operating conditions, it maintains high-precision condition assessment and low-false-alarm anomaly detection, accurately predicting equipment battery life and avoiding the risk of downtime during operations.

[0220] Scenarios of farmland operations in remote areas with weak signal coverage:

[0221] This scenario corresponds to remote farmland, reclamation areas, and forest edge work areas far from villages and towns. These areas have weak mobile communication base station coverage, low and fluctuating signal strength, and frequent network outages are common. Traditional fixed-frequency data reporting strategies suffer significant packet loss under weak signal conditions, resulting in direct data loss after network outages, leading to monitoring interruptions and failures to be reported in a timely manner. The adaptive communication module of this invention can sense the current signal quality in real time and dynamically adjust the reporting frequency of non-critical data, avoiding retransmission storms under weak signal conditions. Simultaneously, it relies on a local ring buffer to retain operational data during network outages. After network recovery, data is retransmitted according to the priority of alarms, status, and telemetry, and the retransmission bandwidth does not encroach on real-time transmission resources. This ensures data integrity and monitoring continuity even in farmland environments with extremely low signal strength, solving the problem of difficult monitoring in remote areas.

[0222] High-efficiency operation support scenarios during the busy farming season:

[0223] This scenario corresponds to the busy farming seasons such as spring plowing, summer planting, and autumn harvest. Agricultural production is highly seasonal, with operational windows typically lasting only a few days. Equipment failures causing downtime directly delay farming, resulting in reduced yields. Existing monitoring solutions can only provide post-failure alarms, relying on on-site personnel to report faults, leading to long response times. Furthermore, post-failure task replacement depends on manual coordination, resulting in low efficiency. This invention utilizes mechanisms such as temperature rise rate detection and multi-dimensional correlation anomaly scoring to proactively identify early equipment degradation trends, enabling pre-failure proactive warnings. Upon fault triggering, alarms are pushed out in stages, automatically generating maintenance work orders with handling guidelines. Emergency faults directly trigger on-site protection actions. Simultaneously, fault information is synchronized to the machine group scheduling system, automatically completing the optimal allocation of remaining tasks, minimizing the impact of faults on operational progress, and ensuring operational continuity during busy farming seasons.

[0224] Multi-aircraft mixed formation collaborative operation scenario:

[0225] In this scenario, multiple types of agricultural robots, including those for tilling, sowing, plant protection, and harvesting, are deployed simultaneously within the same work area. The hardware parameters, load characteristics, and operating thresholds of each model vary significantly, necessitating coordinated operations and unified scheduling. General-purpose IoT platforms, employing a unified monitoring standard, cannot adapt to the characteristics of different robot models, easily leading to issues such as overly strict alarms for some models and alarm failures for others. This invention, during the device initialization phase, automatically loads a dedicated baseline of operating parameters based on the device model, independently maintaining the operating mode baseline and alarm thresholds for each type of robot. Simultaneously, it displays the operational status and task progress of multiple robot models on a unified platform, supporting collaborative scheduling of equipment across different operational stages and achieving precise and unified management of mixed multi-model formations.

[0226] It should be noted that the method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method in this embodiment, and the multiple devices will interact with each other to complete the method described.

[0227] It should be noted that the above description describes some embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0228] Example 2

[0229] See Figure 3 Embodiment 2 of the present invention also provides a status monitoring system for multi-machine scheduling scenarios of agricultural machinery robots, employing the status monitoring method for multi-machine scheduling scenarios of agricultural machinery robots described in the above embodiments, including:

[0230] The device online and working condition initialization module 100 is used to establish a long communication connection with the cloud platform after the agricultural machinery robot is started, and to report basic device information and initial self-test data to the cloud platform. After the cloud platform completes the device identity authentication of the agricultural machinery robot, it loads the corresponding working condition parameter baseline configuration according to the device model, initializes the real-time data transmission channel and updates the device online status.

[0231] The adaptive real-time monitoring module 200 is used by the agricultural machinery robot to identify the current operation mode in real time based on the collected multi-source operation data, and to synchronously report the operation mode label to the cloud platform along with the real-time operation data; the cloud platform matches the corresponding parameter baseline and dynamic alarm threshold according to the received operation mode label, performs adaptive anomaly detection on the real-time operation data, and displays the operation status in two dimensions: single device and machine group, so as to achieve full-domain real-time perception of the operation status of multiple machines.

[0232] The hierarchical alarm and intelligent operation and maintenance linkage module 300 is used to classify alarm levels according to the severity and scope of impact of the anomaly detection results of the cloud platform when the anomaly detection results of the cloud platform exceed the corresponding dynamic alarm threshold, and execute the linkage handling strategy of the corresponding level; at the same time, it generates a structured work order containing device identifier, alarm type, operating condition context and handling suggestions and pushes it to the operation and maintenance terminal, forming a full-process operation and maintenance management from alarm triggering, work order dispatch, on-site handling to confirmation;

[0233] The multi-machine scheduling and linkage module 400 is used to synchronize the location information and remaining workload of the faulty equipment to the machine group scheduling system when the operation is interrupted due to an emergency alarm triggered by a fault on the agricultural machinery robot end. The machine group scheduling system performs task replanning by combining the battery status, spatial location and operation capacity matching degree of the online equipment, and optimizes the allocation of the remaining operation tasks to the remaining online equipment. After the faulty equipment is repaired and passes the self-test, it is reinstated into the operation scheduling pool.

[0234] The task completion and data archiving module 500 is used to archive all data of the current batch of agricultural machinery robots in a hierarchical storage architecture after all agricultural machinery robots have completed the current batch of tasks. This generates a task report that includes equipment energy consumption analysis, abnormal event summary and working condition distribution statistics. The report notifies the operation and maintenance terminal and the machine group scheduling system that the task has been completed, so as to realize the full life cycle management of task data.

[0235] It should be noted that the information interaction and execution process between the modules of the above-mentioned device are based on the same concept as the method embodiment in Embodiment 1 of this application, and the resulting technical effects are the same as those in the method embodiment of this application. For details, please refer to the description in the method embodiment shown above in this application, and it will not be repeated here.

[0236] Example 3

[0237] Embodiment 3 of the present invention provides a non-transitory computer-readable storage medium, wherein the computer-readable storage medium stores program code for a state monitoring method in a multi-machine scheduling scenario of agricultural machinery robots, and the program code includes instructions for executing the state monitoring method in a multi-machine scheduling scenario of agricultural machinery robots as described in Embodiment 1 or any possible implementation thereof.

[0238] Computer-readable storage media can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

[0239] Example 4

[0240] Embodiment 4 of the present invention provides an electronic device, including: a memory and a processor;

[0241] The processor and the memory communicate with each other via a bus; the memory stores program instructions that can be executed by the processor, and the processor can call the program instructions to execute the status monitoring method for multi-machine scheduling of agricultural robots in Embodiment 1 or any possible implementation thereof.

[0242] Specifically, a processor can be implemented in hardware or software. When implemented in hardware, the processor can be a logic circuit, an integrated circuit, etc. When implemented in software, the processor can be a general-purpose processor that reads software code stored in memory. This memory can be integrated into the processor or located outside the processor and exist independently.

[0243] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable system. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0244] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing systems. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Optionally, they can be implemented using program code executable by a computing system, thereby storing them in a storage system for execution by the computing system. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0245] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for status monitoring in a multi-machine scheduling scenario for agricultural machinery robots, characterized in that, Includes the following steps: S1. Equipment Online and Operating Condition Initialization: After the agricultural machinery robot is started, it establishes a long communication connection with the cloud platform and reports basic equipment information and initial self-test data to the cloud platform. After the cloud platform completes the equipment identity authentication of the agricultural machinery robot, it loads the corresponding operating condition parameter baseline configuration according to the equipment model, initializes the real-time data transmission channel, and updates the equipment online status. S2. Adaptive real-time monitoring of working conditions: The agricultural machinery robot end identifies the current working mode in real time based on the collected multi-source operating data, and reports the working mode label to the cloud platform along with the real-time working condition data. The cloud platform matches the corresponding parameter baseline and dynamic alarm threshold based on the received operation mode label, performs adaptive anomaly detection on the real-time operation data, and displays the operation status in two dimensions: single device and cluster, so as to achieve real-time perception of the operation status of multiple machines across the entire domain. S3. Tiered Alarm and Intelligent Operation and Maintenance Linkage: When the anomaly detection result of the cloud platform exceeds the corresponding dynamic alarm threshold, the cloud platform classifies the alarm level according to the severity and scope of the anomaly, and executes the corresponding linkage handling strategy; at the same time, it generates a structured work order containing device identifier, alarm type, operating condition context and handling suggestions and pushes it to the operation and maintenance terminal, forming a full-process operation and maintenance management from alarm triggering, work order dispatch, on-site handling to confirmation; S4. Multi-machine scheduling and linkage: When the agricultural machinery robot terminal triggers an emergency alarm due to a fault, causing the operation to be interrupted, the cloud platform will synchronize the location information of the faulty equipment and the remaining workload to the machine group scheduling system; the machine group scheduling system will combine the battery status, spatial location and operation capacity matching degree of the online equipment to perform task replanning and optimize the allocation of the remaining tasks to the remaining online equipment; after the faulty equipment is repaired and passes the self-inspection, it will be reinstated into the operation scheduling pool. S5. Job Completion and Data Archiving: After all agricultural machinery robots have completed the current batch of operations, the cloud platform will archive all data of this operation in a hierarchical storage architecture, generate an operation report including equipment energy consumption analysis, abnormal event summary and operating condition distribution statistics, and notify the operation and maintenance terminal and the machine group scheduling system that the operation is completed, so as to realize the full life cycle management of operation data.

2. The status monitoring method for multi-machine scheduling of agricultural machinery robots according to claim 1, characterized in that, In step S2, the agricultural machinery robot identifies the current operating mode in real time based on the collected multi-source operating data. After synchronously reporting the operating mode label to the cloud platform along with the real-time operating condition data, the cloud platform uses a multi-factor operating condition adaptive correction method to estimate the battery state of charge (SOC), specifically including: The current operation mode is determined based on the characteristics of motor current, vehicle vibration, and positioning driving speed. The operation modes include six categories: tillage, sowing, plant protection, harvesting, migration, and standby. The temperature-current coupling efficiency factor and vibration correction factor are used to correct the coulomb meter integral results in real time to obtain a preliminary SOC value. The calculation formula is as follows: In the formula, Let SOC be the value at time t. The initial SOC value is determined according to the priority of the value saved from the last normal shutdown, the value obtained by reverse lookup of the open circuit voltage after static setting, and the factory default value. This refers to the battery's rated capacity. This is the instantaneous discharge current; This is the temperature-current coupling efficiency factor, used to compensate for the effects of ambient temperature and discharge current on coulombic efficiency; This is a vibration correction factor used to compensate for the impact of changes in battery contact resistance caused by bumpy farmland surfaces on the accuracy of current measurement. The formula for calculating the temperature-current coupling efficiency factor is as follows: In the formula, The baseline coulomb efficiency; This refers to the actual measured temperature of the battery. For reference temperature; , This is the temperature compensation coefficient, obtained from battery characteristic calibration. This is a high-current correction factor used to compensate for the effect of high-current discharge on coulomb efficiency. This is the battery's rated discharge current. The formula for calculating the vibration correction factor is as follows: In the formula, This represents the root mean square value of the vehicle's three-axis accelerometer. The baseline vibration value for the smooth road surface; This is a correction factor for the effect of vibration on contact resistance.

3. The status monitoring method for multi-machine scheduling of agricultural machinery robots according to claim 2, characterized in that, The preliminary SOC value is obtained using a discrete recursive method, and the calculation formula is as follows: In the formula, Let SOC be the value at the k-th sampling time. The SOC value at the previous sampling time; The median value is taken under high dynamic conditions to suppress transient spikes, and the average value is taken under steady-state conditions to improve computational efficiency. This represents the duration of the current sampling interval; Apply physical constraints to the SOC value: A SOC drift correction mechanism is set up so that when the cumulative correction amount after continuous updates exceeds the preset drift threshold, SOC recalibration is triggered at the time of standby, full charge cutoff, and undervoltage protection to correct the drift error generated by integration.

4. The status monitoring method for multi-machine scheduling of agricultural machinery robots according to claim 2, characterized in that, Step S2 also includes an open-circuit voltage (OCV) mapping correction step: The OCV-SOC mapping relationship is pre-stored. When the battery is detected to be in a static state and the static time meets the preset conditions, the current OCV value is read, and the SOC value is obtained through the mapping relationship. Then, the coulomb SOC value and the OCV SOC value are weighted and fused to obtain the final SOC estimation result. The fusion calculation formula is as follows: In the formula, The SOC value after fusion; The SOC value is calculated using the coulomb integral method; The SOC value obtained from OCV mapping; This is the weighting coefficient, with a value ranging from 0.7 to 0.

9.

5. The status monitoring method for multi-machine scheduling of agricultural machinery robots according to claim 1, characterized in that, In step S2, the cloud platform also uses a multi-factor evaluation method that weights and fuses capacity ratio and internal resistance ratio to calculate battery health status (SOH), specifically including: The formula for calculating SOH using the capacity ratio method is: In the formula, The current available capacity of the battery is estimated by integrating the full charge and discharge cycle or by integrating under controlled current. This refers to the battery's rated capacity. The formula for calculating SOH using the internal resistance ratio method is: In the formula, This refers to the battery's nominal internal resistance. The current internal resistance of the battery is calculated using the voltage and current changes under a step load. The calculation formula is as follows: In the formula, , The voltage and current before the load step change, , The voltage and current after the load step change; The combined SOH is obtained by weighting and combining the capacity ratio SOH and the internal resistance ratio SOH, as shown in the formula: In the formula, , For the weighting coefficients, satisfying .

6. The status monitoring method for multi-machine scheduling of agricultural machinery robots according to claim 1, characterized in that, The agricultural machinery robot terminal and the cloud platform employ an adaptive data acquisition and transmission mechanism with signal quality awareness, specifically including: The reported data is divided into low-frequency status frames, high-frequency telemetry frames, event alarm frames, and task progress frames, and a differentiated reporting strategy is adopted. The reporting frequency of high-frequency telemetry frames is dynamically adjusted according to the operation mode and signal quality, and the calculation formula is as follows: In the formula, Based on the reporting frequency; This is the working condition coefficient, with different values ​​set for different working modes; The signal quality coefficient is calculated using the following formula: In the formula, The current wireless signal strength, This serves as a good reference value for signal strength. All reported data includes a timestamp, and the timestamp conversion relationship is as follows: Each frame of data is appended with a frame sequence number and a CRC checksum. The cloud platform detects data loss by using the sequence number continuity and verifies data integrity by using the checksum. The agricultural machinery robot maintains a local circular buffer. When a network interruption is detected, data is written to the local buffer. After the network is restored, data is retransmitted in the order of priority of event alarm frames, status frames, and telemetry frames, and the retransmission bandwidth is limited to a preset proportion of the currently available bandwidth. The cloud platform determines the online and offline status of devices through a keep-alive mechanism. The keep-alive timeout threshold is calculated using the following formula: In the formula, The current time; This refers to the time when the device last reported the keep-alive heartbeat packet; As a fault tolerance factor; when the device's last reporting time is earlier than the keep-alive timeout threshold, the device is determined to be offline and an offline alarm is triggered; Simultaneously, a data obsolescence determination mechanism is set up, and the calculation formula is as follows: In the formula, This refers to the last valid data update time of the device. The data staleness duration; when the data staleness duration exceeds the preset threshold, the display status will be downgraded to expired.

7. The status monitoring method for multi-machine scheduling of agricultural machinery robots according to claim 1, characterized in that, In step S2, anomaly detection based on adaptive operating conditions is performed on the real-time operating condition data, specifically including: For each operating mode of each device, an independent parameter baseline distribution is maintained. The baseline includes the mean and standard deviation of each monitored parameter. The baseline parameters are updated online using an exponential moving average method, with the update formula as follows: In the formula, The baseline mean after the k-th update; The parameter value is the value of the kth sample. This is a smoothing coefficient that is adaptively adjusted based on the stability of the operating conditions. For temperature parameters, calculate the rate of temperature change, and generate a dynamic temperature rise threshold by combining the operating mode baseline and the ambient temperature correction item. When the rate of temperature change exceeds the dynamic temperature rise threshold, trigger an alarm for excessively rapid temperature rise. The formula for calculating the rate of temperature change is: In the formula, This represents the temperature value from the kth sample. To calculate the number of span points; The sampling interval; The formula for calculating the dynamic temperature rise threshold is: In the formula, This represents the baseline mean of the temperature rise rate under the corresponding operating mode; This represents the baseline standard deviation of the temperature rise rate under the corresponding operating mode; Confidence coefficient; This is an ambient temperature correction item.

8. The status monitoring method for multi-machine scheduling of agricultural machinery robots according to claim 7, characterized in that, In step S2, anomaly detection based on operating condition adaptation is performed on the real-time operating condition data, which also includes voltage descent detection, specifically including: The battery terminal voltage is monitored in real time, and the voltage change rate is calculated using the finite difference method. The calculation formula is as follows: In the formula, Let be the terminal voltage at the k-th sampling time; Let be the terminal voltage at the km-th sampling time; The number of span points; Before calculation, the voltage data is denoised using a moving average. The moving average formula is as follows: In the formula, This is the filtered voltage value; The length of the sliding window; By combining current limiting conditions to exclude normal load fluctuations, a voltage descent alarm is triggered when the voltage drop rate exceeds a preset threshold. Meanwhile, the window drop amplitude is used as a supplementary criterion, and the calculation formula is as follows: In the formula, A sliding window voltage sequence of length L; This represents the maximum voltage value within the window. The minimum voltage value within the window; This represents the voltage drop within the window.

9. The status monitoring method for multi-machine scheduling of agricultural machinery robots according to claim 1, characterized in that, It also includes a multi-dimensional correlation anomaly detection and alarm deduplication mechanism: When a single-dimensional parameter falls within a suspected anomaly range, a multi-dimensional correlation anomaly score calculation is initiated. The baseline deviation of all multi-dimensional parameters is considered to determine whether a multi-parameter collaborative anomaly alarm should be triggered. The anomaly score calculation formula is as follows: In the formula, The weight of the i-th parameter; Let be the measured value of the i-th parameter; This represents the baseline mean of the i-th parameter under the corresponding work mode; This represents the baseline standard deviation of the i-th parameter under the corresponding work mode; Alarms are divided into three levels: emergency, important, and general. Emergency alarms trigger power limiting or shutdown protection actions of the equipment and are pushed synchronously through multiple channels; important alarms generate maintenance work orders and push them to the responsible personnel; general alarms are temporarily stored in the observation queue. The alarm deduplication suppression window mechanism is adopted, and the formula for calculating the suppression window start time is as follows: In the formula, This is the current alarm trigger time; The base duration for deduplication suppression window; The start time of the deduplication window; If there are similar unprocessed alarms within the interval, repeat pushes will be suppressed. The suppression window duration is dynamically adjusted, and the calculation formula is as follows: In the formula, This is the base suppression duration corresponding to the alarm level; The growth coefficient is the factor that suppresses repetition. This sets the number of times the same type of alarm will be triggered within a preset period; if the same type of alarm is triggered frequently, the suppression window will be automatically extended and the alarm level will be upgraded.

10. A status monitoring system for multi-machine scheduling of agricultural machinery robots, employing the status monitoring method for multi-machine scheduling of agricultural machinery robots as described in any one of claims 1 to 9, characterized in that, include: The device online and working condition initialization module is used to establish a long communication connection with the cloud platform after the agricultural machinery robot is started, and to report basic equipment information and initial self-test data to the cloud platform. After the cloud platform completes the device identity authentication of the agricultural machinery robot, it loads the corresponding working condition parameter baseline configuration according to the device model, initializes the real-time data transmission channel, and updates the device online status. The adaptive real-time monitoring module is used by the agricultural machinery robot to identify the current working mode in real time based on the collected multi-source operating data, and to synchronously report the working mode label to the cloud platform along with the real-time working condition data. The cloud platform matches the corresponding parameter baseline and dynamic alarm threshold based on the received operation mode label, performs adaptive anomaly detection on the real-time operation data, and displays the operation status in two dimensions: single device and cluster, so as to achieve real-time perception of the operation status of multiple machines across the entire domain. The hierarchical alarm and intelligent operation and maintenance linkage module is used to classify alarm levels according to the severity and scope of impact of the anomaly detection results of the cloud platform when the anomaly detection results exceed the corresponding dynamic alarm threshold, and to execute the corresponding linkage handling strategy. At the same time, it generates a structured work order containing device identifier, alarm type, operating condition context and handling suggestions and pushes it to the operation and maintenance terminal, forming a full-process operation and maintenance management from alarm triggering, work order dispatch, on-site handling to confirmation. The multi-machine scheduling and linkage module is used to synchronize the location information and remaining workload of the faulty equipment to the machine group scheduling system when the operation is interrupted due to an emergency alarm triggered by a fault on the agricultural machinery robot end. The machine group scheduling system performs task replanning by combining the battery status, spatial location and operation capacity matching degree of the online equipment, and optimizes the allocation of the remaining operation tasks to the remaining online equipment. After the faulty equipment is repaired and passes the self-test, it is reinstated into the operation scheduling pool. The task completion and data archiving module is used to archive all data of the current batch of agricultural machinery robots in a hierarchical storage architecture after all agricultural machinery robots have completed the current batch of tasks. This generates a task report that includes equipment energy consumption analysis, abnormal event summary and working condition distribution statistics. The report then notifies the operation and maintenance terminal and the machine group scheduling system that the task has been completed, thereby realizing full lifecycle management of task data.

Images