Irrigation monitoring and control system for smart agriculture based on internet of things

Abstract

The application provides a smart agricultural irrigation monitoring control system based on Internet of Things, and belongs to the technical field of agricultural control, and comprises a data acquisition module, an abnormality analysis module, a dynamic reconstruction and error compensation module and a hierarchical control irrigation module, which are used for constructing a sliding window with time decay memory, generating a space-time synchronous characteristic value sequence; by comparing real pipe network flow state parameters with virtual hydraulic model deduced parameters, calculating virtual and real flow resistance distortion ratios, and combining soil humidity to construct a comprehensive abnormal evolution discrimination index; using zero flow state of water pump shutdown pressure maintaining and steady state of constant frequency operation two kinds of limit working conditions, iteratively generating a dynamic reference compensation coefficient; generating a control instruction of a corresponding topological node according to the comprehensive abnormal evolution discrimination index combined with the dynamic reference compensation coefficient, and performing irrigation control; the application effectively avoids global paralysis caused by local faults, and improves long-term operation stability of a large-area irrigation system in a complex outdoor environment.

Description

Technical Field

[0001] This invention belongs to the field of agricultural control technology, specifically relating to an Internet of Things-based smart agricultural irrigation monitoring and control system. Background Technology

[0002] Modern large-scale agricultural irrigation areas typically span vast geographical areas. In such scenarios involving long-term, complex weather changes in farmland, the physical pipeline network inevitably faces the risk of aging and damage. Currently, the most relevant conventional scheduling methods in this scenario rely primarily on single-point threshold monitoring and static topology scheduling. In practice, the system deploys pressure and flow monitoring devices at pipeline nodes and temperature and humidity monitoring nodes in the farmland. When the humidity falls below an absolute threshold, the valve is opened to supply water; when the pipeline pressure deviates from a safe threshold, an alarm is triggered to shut down the system.

[0003] Because current technology struggles to reduce the correlation between physical facility malfunctions and natural environmental changes, and because sudden leaks in the pipeline network and extreme high temperatures causing a surge in crop transpiration can both lead to drastic fluctuations in pressure or humidity at monitoring nodes, using static thresholds can easily cause system misidentification, resulting in scheduling accidents such as missed water adjustments or accidental valve closures. Furthermore, in complex and harsh field conditions, monitoring equipment inevitably experiences zero-point temperature drift after prolonged operation, and the inner walls of underground pipelines continuously accumulate sediment and form chronic scale. These long-term physical disturbances cause significant discrepancies between the system's built-in static hydraulic benchmark model and the actual physical scenario. As the operating cycle progresses, data deviations accumulate and amplify, ultimately causing the system to lose its accuracy in the later stages of operation, leading to widespread misjudgments and the complete paralysis of the entire irrigation control network. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a smart agricultural irrigation monitoring and control system based on the Internet of Things.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The data acquisition module is used to acquire raw monitoring data collected by multi-source sensors deployed at each topology node of the farmland pipeline network at different time points, perform time-series alignment processing on the raw monitoring data, and generate a sequence of synchronous feature values ​​of each topology node at a unified target time. The anomaly analysis module is used to compare the actual flow parameters in the synchronous feature value sequence with the theoretical flow parameters derived from the preset virtual hydraulic model under the same control state, calculate the flow resistance distortion amplitude of each topological node, obtain the measured soil moisture of the farmland in the area corresponding to the topological node, and generate a comprehensive anomaly evolution discrimination index for each topological node based on the flow resistance distortion amplitude and the measured soil moisture. The dynamic reconstruction and error compensation module is used to extract the deviation between the actual flow parameters and the theoretical flow parameters under the first extreme working condition and the second extreme working condition, respectively, to perform iterative compensation on the two types of deviations and generate dynamic benchmark compensation coefficients. The first extreme working condition is the zero flow condition where the water pump stops and the entire network maintains pressure, and the second extreme working condition is the steady-state condition where the water pump runs at a constant frequency and there is no valve action. The hierarchical irrigation control module is used to classify the network status into levels based on the numerical range of the comprehensive anomaly evolution discrimination index, and generate control commands for the corresponding topology nodes in combination with the dynamic benchmark compensation coefficient. This is used to adjust the pump operating frequency and solenoid valve opening for irrigation control.

[0006] Preferably, the step of performing time-series alignment processing on the original monitoring data to generate a sequence of synchronization feature values ​​for each topological node at a unified target time includes: A sliding window is constructed with a preset benchmark step size as the interval. The original monitoring data of each topology node at multiple historical time points within the sliding window are obtained. The attenuation weights are assigned according to the distance of each historical time point from the target time. The weighted original monitoring data of each historical time point are accumulated to generate the synchronization feature value sequence.

[0007] Preferably, the attenuation weights are calculated and allocated according to a negative exponential decay function with the natural logarithm as the base, and the attenuation weights corresponding to historical time points further away from the target time are smaller.

[0008] Preferably, the calculation of the flow resistance distortion amplitude of each topology node includes: Obtain the absolute value of the difference between the measured pressure drop of the upstream and downstream pressure sensors of the topological node and the theoretical pressure drop output by the virtual hydraulic model. After normalizing the absolute value of the difference, multiply it by the ratio of the measured instantaneous flow velocity at the topological node to the theoretical flow velocity output by the virtual hydraulic model to obtain the flow resistance distortion amplitude.

[0009] Preferably, the step of generating a comprehensive anomaly evolution discrimination index for each topological node based on the flow resistance distortion amplitude and the measured soil moisture includes: The first ratio is denoted as the relative ratio of the flow resistance distortion amplitude of each neighboring node in the spatially associated neighborhood set of the current topological node to the baseline value of the flow resistance distortion amplitude during the historical normal operation of the corresponding neighboring node. The second ratio is denoted as the relative ratio of the difference between the target soil moisture and the measured soil moisture in the corresponding area of ​​each neighboring node to the target soil moisture. The first ratio and the second ratio are multiplied and then weighted and accumulated according to the spatial distance from each neighboring node to the current topological node to generate the comprehensive anomaly evolution discrimination index.

[0010] Preferably, the spatial distance weights from each neighboring node to the current topology node are generated based on the reciprocal of the total pipeline length between each neighboring node and the current topology node.

[0011] Preferably, the generation of dynamic benchmark compensation coefficients includes: The relative deviation between the measured hydrostatic pressure of each topology node under the zero-flow condition and the theoretical hydrostatic pressure calculated based on the elevation difference is obtained; the relative deviation between the measured pressure drop and the theoretical pressure drop of each topology node under at least one of the steady-state conditions is obtained, and a weighted average is performed after allocating time forgetting decay weights according to the occurrence time of each steady-state condition; the relative deviation of the hydrostatic pressure and the relative deviation of the measured pressure drop after weighting are superimposed, and the benchmark compensation coefficient of the previous reconstruction cycle is iteratively updated to generate the dynamic benchmark compensation coefficient of the current reconstruction cycle.

[0012] Preferably, the step of hierarchically dividing the pipeline network status based on the numerical range of the comprehensive anomaly evolution discrimination index, and generating control commands for corresponding topology nodes in combination with dynamic benchmark compensation coefficients, includes: When the comprehensive anomaly evolution discrimination index is less than the first preset threshold, the hierarchical irrigation control module determines that the pipeline network is in a normal aging state, multiplies the preset basic control quantity by the dynamic benchmark compensation coefficient of the current reconstruction cycle, and generates the control command; when the comprehensive anomaly evolution discrimination index is greater than or equal to the first preset threshold and less than the second preset threshold, the hierarchical irrigation control module determines that the pipeline network is in a water shortage response state, generates a water demand urgency adjustment coefficient based on the measured soil moisture, multiplies the standby basic control quantity by the water demand urgency adjustment coefficient, and generates the control command; when the comprehensive anomaly evolution discrimination index is greater than or equal to the second preset threshold, the hierarchical irrigation control module determines that the pipeline network is in a fault isolation state, and sets the control command to zero.

[0013] Preferably, the multi-source sensors include pressure sensors, electromagnetic flow meters, and soil moisture sensors deployed at each topology node and in farmland areas.

[0014] Preferably, the virtual hydraulic model is constructed based on the topology connectivity map of the pipeline geographic information system and the pipeline elevation difference data. It is used to deduce the theoretical pressure, theoretical flow velocity and theoretical pressure drop of each topological node through fluid dynamics equations under the given valve opening degree and pump frequency.

[0015] The IoT-based smart agricultural irrigation monitoring and control system provided by this invention has the following beneficial effects: By comparing real flow parameters with theoretical parameters of the virtual hydraulic model, the flow resistance distortion amplitude is calculated. This is then integrated with measured soil moisture to generate a comprehensive anomaly evolution discrimination index. This index can clearly distinguish between two fundamentally different anomalies: physical pipeline damage and natural ecological water shortage. This solves the problem of frequent false alarms or missed detections in existing technologies due to the inability to identify the root cause of anomalies using static thresholds. Furthermore, by utilizing two extreme operating conditions—zero flow and steady state—two types of aging deviations—sensor temperature drift and pipe scaling or wear—dynamic benchmark compensation coefficients are iteratively generated to continuously correct the benchmark parameters of the virtual hydraulic model. This further improves the accuracy of the virtual-real comparison, ensuring the anomaly index remains reliable over the long term. This addresses the problem of traditional fixed models accumulating and amplifying deviations due to facility aging, eventually leading to complete failure. Through the aforementioned progressive and interconnected operations, the smart irrigation system can maintain high-precision irrigation decisions autonomously even in the face of multiple interferences such as sensor drift, pipe scaling, sudden pipe bursts, and extreme droughts over a long operating cycle of several years. It strictly limits the impact of single-point physical failures to local topological areas while ensuring normal water supply to crops in non-faulty areas. This effectively overcomes the shortcomings of existing technologies where a single-point failure can lead to the paralysis of the entire network, and achieves long-term operational stability of large-area smart irrigation systems in complex field environments. Attached Figure Description

[0016] To more clearly illustrate the embodiments and design schemes of the present invention, the accompanying drawings required for this embodiment will be briefly described below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the steps of an IoT-based smart agricultural irrigation monitoring and control system according to an exemplary embodiment of the present invention. Figure 2 This is a schematic diagram of the feedback loop of the dynamic reconstruction and error compensation module of the IoT-based smart agricultural irrigation monitoring and control system provided by the present invention according to an exemplary embodiment. Figure 3 This is a schematic diagram illustrating the state transition of a hierarchical control irrigation module in an IoT-based smart agricultural irrigation monitoring and control system provided by the present invention, according to an exemplary embodiment. Detailed Implementation

[0018] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be construed as limiting the scope of protection of the present invention.

[0019] The technical solutions provided by the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0020] First, this invention provides a smart agricultural irrigation monitoring and control system based on the Internet of Things, specifically as follows: Figure 1 As shown, it includes the following modules: The data acquisition module is used to acquire raw monitoring data collected by multi-source sensors deployed at each topology node of the farmland pipeline network at different time points, perform time-series alignment processing on the raw monitoring data, and generate a sequence of synchronous feature values ​​of each topology node at a unified target time.

[0021] It should be noted that modern large-scale agricultural irrigation areas typically cover hundreds to thousands of acres of geographical space. The IoT sensing devices deployed on-site exhibit highly fragmented characteristics, with data update frequencies varying greatly from different sources: macro-level internet weather forecast data (such as rainfall probability and temperature prediction) are usually updated on an hourly basis; soil temperature and humidity sensors located at the far end of farmland are limited by the battery life of low-power wide area networks and often only wake up to sample once every tens of minutes or even hours; while industrial-grade pressure transmitters and electromagnetic flowmeters deployed in main pipelines and pumping stations continuously report data at high frequencies of seconds or even milliseconds via wired or 4G high-speed networks.

[0022] Existing conventional data analysis models directly input this raw data, which lacks strict spatiotemporal alignment, into logical operations, leading to failure issues. When a cloud server attempts to calculate the hydraulic balance equation at a certain moment, the node pressure data extracted by the system might be from 10:00:05, while the flow velocity data might still be from 10:00:00, and the soil moisture data might be even earlier, from 08:30. This asynchronous clock difference is amplified infinitely in fluid dynamics calculations. Even a small sampling time difference can cause the system to erroneously calculate a much larger fluid loss rate than actual, resulting in frequent false leak alarms. Simultaneously, the processing of isolated data lacking spatial correlation makes it easy for control commands to generate temporal logic conflicts, leading to water hammer effects or unexpected pressure buildup in pipelines.

[0023] By establishing a sliding window with a time decay mechanism, all physical quantities of different dimensions are mapped to a unified sequence with absolute timestamps. In other words, the discrete data from each sensor at different time points are output as topological node feature values ​​that have been mathematically smoothed and synchronously reconstructed at a unified time.

[0024] Specifically, the formula for constructing the spatiotemporal synchronization feature value sequence is as follows: In the formula, Represents topology nodes At any moment Synchronized and aligned feature values, such as pressure, flow rate, or soil moisture; Represents topology nodes At historical junctures The raw monitoring data was acquired in real time through analog sensors deployed on-site. This represents the k-th time alignment decay weight, obtained by the system based on the distance of historical time points from the target alignment time. The distance is automatically generated, and the older the data, the lower the weight. This represents the total number of historical time points within the sliding window; This indicates the preset sampling reference step size, such as a fixed 10 seconds.

[0025] It should be noted that after this operation, the raw data from multiple heterogeneous nodes such as farmland, pipelines, and water sources are forcibly arranged to the same moment in time. This mathematical relationship, mapped onto real agricultural scenarios, ensures that regardless of fluctuations in the communication status of on-site sensors, the water pressure, flow rate, and moisture content of the pipelines are always real data occurring at the same instant, thus eliminating errors caused by communication delays.

[0026] It should be noted that the reference step size in this embodiment... Set to 30 seconds, decay weight A negative exponential decay function with the natural logarithm as the base is used for allocation. The underlying logic for setting this artificial parameter is to balance the slow dynamic changes in the agricultural environment (such as soil moisture evaporation) with the fast dynamic changes in pipeline fluids (such as sudden changes in water pressure caused by pump start-up and shutdown). Setting the parameters too high will cause a smooth transition, resulting in a severe lag in the system's detection of sudden pipe bursts and leaks, thus losing its protection mechanism; setting them too low will introduce a huge computational load and communication network jitter noise.

[0027] The anomaly analysis module is used to compare the actual flow parameters in the synchronous feature value sequence with the theoretical flow parameters derived from the preset virtual hydraulic model under the same control state, calculate the flow resistance distortion amplitude of each topological node, obtain the measured soil moisture of the farmland in the area corresponding to the topological node, and generate a comprehensive anomaly evolution discrimination index for each topological node based on the flow resistance distortion amplitude and the measured soil moisture.

[0028] It should be noted that agricultural irrigation networks are large-scale fluid networks that are constantly and dynamically changing. During normal scheduling, due to differences in water demand in different regions, variable frequency pumps frequently adjust their output power, and the solenoid valves of each branch line are constantly switching on and off according to the irrigation rotation plan. These routine control actions inevitably lead to drastic global fluctuations in the static and dynamic pressure of the entire network. The core logic of current conventional fault diagnosis models is to set fixed upper and lower absolute thresholds for each pressure sensor. For example, a pressure below 0.2 MPa is considered a leak. This static threshold mechanism has obvious defects: if the variable frequency pump actively reduces its head to respond to the low-flow drip irrigation demand of distant farmland, the network pressure will naturally drop and touch the low-pressure threshold. The system will then misjudge the normal scheduling action as a serious leak across the entire network and trigger an emergency shutdown. If the alarm threshold is set too broadly to avoid false alarms, when a real local rupture and leak occurs in the terminal branch pipe, the pressure drop at the rupture point may not reach the broad lower alarm limit because the main pump is still supplying pressure, ultimately causing the hidden pipe rupture to be completely missed.

[0029] To address the aforementioned issues, this embodiment compares flow data collected by real sensors with theoretical flow data calculated from a virtual pipeline network under the same valve opening conditions. This directly eliminates the global pressure reference offset caused by pump start-up and shutdown, and extracts the flow resistance distortion ratio of the physical pipeline itself. Subsequently, this purely physical distortion amplitude is combined with the soil water shortage status of surrounding farmland for calculation. Analysis using real fluid parameters, virtual hydraulic model predicted parameters, and measured farmland humidity yields a comprehensive abnormal evolution discrimination index. This index clearly distinguishes between the characteristics of physical facility failures and natural ecological evolution, providing a precise decision-making basis for subsequent control strategy diversion.

[0030] Specifically, by calling the built-in hydraulic topology equations of the pipe network, the theoretical flow parameters of the virtual pipe network under the current valve opening and pump frequency are derived. The specific formula for calculating the flow resistance distortion amplitude by combining the actual flow parameters of each node at a unified moment is as follows: In the formula, This represents the magnitude of the flow resistance distortion of topological node i at time t; This represents the pressure drop measured by the pressure sensors upstream and downstream of topology node i, which is obtained by subtracting the measured pressure data of the two physical pressure sensors upstream and downstream of topology node i. This represents the reference pressure drop output by the virtual hydraulic model under the same control conditions. The cloud server calculates this by substituting the valve opening and pump frequency of topology node i at time t into the built-in pipe network hydraulic topology equation. This represents the instantaneous flow velocity measured by the electromagnetic flowmeter at topological node i; This represents the predicted flow velocity output by the virtual hydraulic model. It represents a very small positive number.

[0031] Specifically, the formula for generating the comprehensive anomaly evolution discrimination index for each topological node by integrating the measured soil moisture of farmland in the corresponding area is as follows: In the formula, The comprehensive anomaly evolution discrimination index represents the topological node i at time t; This represents the set of spatially related neighborhoods of topological node i, extracted from the pre-stored pipeline network GIS topology connectivity map in the system; Representing neighboring nodes The spatial distance attenuation weight is calculated based on the reciprocal of the total pipeline length from neighboring node j to topology node i. This represents the magnitude of flow resistance distortion at neighboring node j at time t; This represents the average value of the flow resistance distortion amplitude of neighboring node j during normal operation within the historical reference time period; This indicates the target soil moisture level corresponding to the crop planted in the neighborhood node j; This represents the soil moisture at time t measured by the TDR moisture sensor at neighboring node j. It represents a very small positive number.

[0032] The historical reference time period is selected from a period of 7 consecutive days after the system is put into operation during which there are no alarm records and the water pump frequency fluctuation is less than 5%. The flow resistance distortion amplitude data of each topology node during this period are collected and the arithmetic mean is calculated as the historical reference time period. The benchmark value is dynamically updated once a week.

[0033] It should be noted that, This represents the relative ratio of the flow resistance distortion amplitude to the historical normal flow resistance distortion amplitude. This indicates the severity of chronic water shortage at the environmental and ecological level; when a physical pipe bursts, a large amount of water is lost through an undesigned breach, causing a sharp increase in local pressure drop and distortion of flow velocity, leading to... Far exceeding historical benchmarks At this time, regardless of soil moisture Whether the requirements are met or not, the comprehensive anomaly evolution discrimination index calculated by multiplying the two will spike dramatically, thus classifying the scene as a sudden and severe physical damage; conversely, when the scene experiences gradual high temperature and drought, although the farmland is extremely short of water (leading to...) (Very large), but the fluid resistance characteristics of the physical pipe network have not changed, that is, the current... Basically equal to historical benchmark The ratio term approaches 1, and the product of the two terms will maintain the comprehensive abnormal evolution discrimination index at a stable low level. This mathematical relationship, when mapped to the physical scenario, directly endows the control system with the ability to distinguish between true damage and false water shortage.

[0034] It should be noted that the spatial distance attenuation weight in this embodiment... The calculation rules stipulate that the weights of nodes exceeding three topological levels (i.e., spanning three branch three-way valves) are forcibly reset to zero. The underlying logic for setting this artificial parameter is that water hammer oscillations and local depressurization effects in fluid dynamics exhibit strong spatial damping characteristics in actual pipe networks. If this parameter is not restricted or is set improperly, it can lead to minor fluctuations in extremely distant, unrelated branches throughout the entire Daejeon network being included in the local calculation, thereby triggering a large-scale chain of misjudgments and redundant system calculations.

[0035] The dynamic reconstruction and error compensation module is used to extract the deviation between the actual flow parameters and the theoretical flow parameters under the first extreme working condition and the second extreme working condition, respectively, and to perform iterative compensation for the two types of deviations to generate dynamic benchmark compensation coefficients. The first extreme working condition is a zero flow condition in which the water pump stops and the entire network maintains pressure, and the second extreme working condition is a steady-state condition in which the water pump runs at a constant frequency and there is no valve action.

[0036] It should be noted that all irrigation pipe networks and IoT sensing devices deployed in the field during agricultural irrigation will undergo irreversible physical degradation over time. During operation cycles lasting months or even years, the inner walls of underground pipe networks will continuously accumulate microbial films, sediment, and calcium and magnesium ion scale, causing the absolute inner diameter of the pipes to gradually decrease, and the physical friction resistance along the water flow will inevitably increase month by month. At the same time, industrial pressure sensors exposed to the field will inevitably experience a slow, chronic temperature drift in their internal circuit zero-point voltage after hundreds or thousands of cycles of drastic temperature changes between day and night and prolonged exposure to sunlight.

[0037] Existing conventional models fail in this scenario primarily because their underlying physical parameters, such as pipe wall roughness coefficients and sensor theoretical zero points, are often fixed in the database during the initial construction phase of the project and are not adjusted as the equipment status changes. When facilities have been in operation for more than six months and show obvious signs of aging, the virtual model continues to extrapolate theoretical data based on the standards of brand-new pipes and sensors. Over time, the accumulated deviations lead to a significant, persistent residual difference between the virtually calculated pressure drop and the actual measured pressure drop, which is then incorrectly and continuously identified by the system as chronic leakage or partial blockage.

[0038] This embodiment utilizes the unique physical conditions inherent in irrigation projects to extract aging parameters in reverse: during zero-flow periods without irrigation scheduling tasks, such as pump shutdown and network-wide pressure maintenance, the static pressure deviation is extracted, completely isolating and eliminating pure circuit temperature drift from sensors; during steady-state periods when pumps maintain a constant frequency for extended periods, the pressure drop deviation is extracted, isolating and quantifying the increase in pipeline friction resistance. By using virtual and real pressure difference data under both zero-flow and steady-state extreme conditions, a dynamic benchmark compensation coefficient with long-term memory characteristics is calculated, comprehensively solving the problem of system control accuracy failure caused by the aging of physical facilities.

[0039] Specifically, under the zero-flow condition of pump shutdown and network-wide pressure maintenance, and the steady-state condition of pump constant frequency operation and no valve action, the deviations between the actual flow parameters and the virtual theoretical flow parameters are extracted. The two types of deviations are iteratively stripped and cumulatively compensated to generate a dynamic benchmark compensation coefficient with long-term memory characteristics. The specific formula is as follows: In the formula, This represents the dynamic baseline compensation coefficient of topology node i in the (n+1)th reconstruction cycle. Represents the dynamic baseline compensation coefficient of topology node i in the current nth reconstruction cycle; This represents the absolute hydrostatic pressure data measured by the pressure sensor at topology node i under zero-flow conditions; This represents the theoretical hydrostatic pressure data of topological node i calculated based on the GIS elevation difference model under zero flow conditions; The weighting for time-related forgetting decay is assigned based on the length of historical records, with the most recent steady-state condition having the largest weight. This represents the measured pressure drop data of the pressure sensor at topology node i under steady-state conditions; This represents the theoretical pressure drop number output by the virtual hydraulic model for topological node i under steady-state conditions. This represents the preset hyperparameters; this implementation uses the preset hyperparameters. This is used to prevent the denominator from being 0.

[0040] It should be noted that, The ratio representing the cumulative amount of environmental temperature drift and This represents the weighted average ratio of the accumulated fluid scaling resistance. In a physical scenario, if long-term scaling on the pipe wall causes the actual frictional force to steadily exceed the theoretical frictional force calculated by the factory model, the steady-state error term will remain positive. This will cause the overall factor within the square brackets to be less than 1, thus causing the value of the dynamic reference compensation coefficient to begin decreasing relative to the previous cycle. This mathematical attenuation correction instruction, mapped to the system's underlying logic, forces the virtual hydraulic reference model in the above steps to continuously approach the current, naturally aging physical reality. This allows the system to resolve long-term environmental disturbances caused by mechanical wear and electronic drift without human intervention or recalibration.

[0041] Please see Figure 2 The diagram illustrates the feedback loop of the dynamic reconstruction and error compensation module of an IoT-based smart agricultural irrigation monitoring and control system. The left side shows two extreme condition extraction channels: under zero-flow condition, temperature drift deviation is extracted from the measured hydrostatic pressure and the theoretical hydrostatic pressure based on the elevation difference; under steady-state condition, scaling resistance deviation is extracted from the measured pressure drop and the theoretical pressure drop with a time-forgotten attenuation weight. The two deviations are merged at the superposition node and enter the iterative update stage, updating the benchmark compensation coefficient of the previous reconstruction cycle and outputting the dynamic benchmark compensation coefficient for the current reconstruction cycle. The output is fed back through the feedback path to correct the benchmark parameters of the virtual hydraulic model and serves as the input for the next iteration. The iterative update stage is marked with a bold border to highlight the key improvement of long-term memory self-correction.

[0042] The hierarchical control irrigation module is used to classify the network status into levels based on the numerical range of the comprehensive anomaly evolution discrimination index, and generate control commands for the corresponding topology nodes in combination with the dynamic benchmark compensation coefficient. This is used to adjust the pump operating frequency and solenoid valve opening for irrigation control.

[0043] It should be noted that irrigation problems in actual production are often complex and highly dynamic. When localized pipe network leaks or severe droughts occur in certain areas, traditional programmable logic controllers (PLCs), limited by simple relay logic, can only adopt rigid lock-up strategies and cannot adapt to the control requirements of complex scenarios. Existing conventional control models will completely fail in such scenarios. The core reason is that these models rely solely on a single high-pressure or low-pressure threshold to determine pipe network anomalies. Once the judgment condition is triggered, the only action taken is to shut down the pumps and close the main water supply pipe. This one-size-fits-all approach can have disastrous consequences in agricultural fields: in addition to the faulty branch, hundreds of acres of normally growing crops will be cut off from water. A single localized fault can directly paralyze the entire irrigation network, leaving the system completely defenseless against risks.

[0044] The improved hierarchical adaptive control execution unit designed in this embodiment regulates the irrigation system based on the comprehensive anomaly evolution discrimination index and dynamic benchmark compensation coefficient obtained in the above steps. The input data of this control logic includes the comprehensive anomaly evolution discrimination index matrix of each topological node, the soil water shortage vector, and the backup route topology relationship of the pipeline network GIS. The final output is the frequency division and calibration action command issued to each water pump inverter and node solenoid valve, forming a complete closed loop at the execution level. Even in the face of macro-environmental deterioration or micro-facilities damage, the system can still operate, ensuring the survival rate of crops through intelligent resource allocation.

[0045] Specifically, the pipeline network status is hierarchically divided based on the numerical range of the comprehensive anomaly evolution discrimination index. Control commands for corresponding topology nodes are generated using dynamic benchmark compensation coefficients. The specific method for irrigation control by adjusting pump operating frequency and solenoid valve opening is as follows: in, This represents the valve opening degree or pump operating frequency command issued by topology node i at time t. This indicates the default valve base opening or water pump operating frequency of topology node i, which is obtained by querying the system's crop routine irrigation schedule table. Represents the dynamic baseline compensation coefficient of topology node i in the current nth reconstruction cycle; and These represent the preset first-level and second-level graded response thresholds, respectively. The basic alternative opening coefficient of topology node i is calculated by the system's network topology routing algorithm based on the alternative path impedance. This represents a water urgency adjustment function generated based on the current measured soil moisture.

[0046] It should be noted that when the comprehensive anomaly evolution discrimination index is below the first-level response threshold, it indicates that the current pipeline network is healthy but aging. The system implements fault-tolerant pressure fine-tuning by multiplying by the dynamic benchmark compensation coefficient. When the comprehensive anomaly evolution discrimination index exceeds the first-level response threshold but does not reach the limit threshold, it indicates that the scenario has entered a state of severe drought and water shortage but the pipeline network has not ruptured. At this time, the system activates the triggering actuator to transfer the valve base opening or water pump operating frequency to the bypass vector conventionally preset. At the same time, it triggers the water demand urgency adjustment function generated by the current measured soil moisture. This means that the system actively cuts off the water supply branches for those planting low-value-added crops and allocates the remaining limited water resources and water pump pressure to the high-value core economic forest areas that are on the verge of dying through the bypass trunk line with precise cross-regional tilting. When the comprehensive anomaly evolution discrimination index exceeds the second-level response threshold of the life-or-death line, the dynamic benchmark compensation coefficient is forcibly reset to zero. This means that when the pipeline network encounters a devastating physical pipe burst, the system decisively executes full valve closure for the specific faulty physical node, thereby preventing the leakage of precious water resources.

[0047] It should be noted that in this embodiment, the first-level response threshold is set to 1.5, and the second-level response threshold for extreme fault isolation is set to 3.0. If the second-level response threshold is set too low, even slight water hammer vibrations caused by tractors running over the pipes in the field will frequently trigger the highest-level network disconnection and reconfiguration operation. This will not only exacerbate the mechanical fatigue of the electromagnetic valve assembly on site but also disrupt the established irrigation rhythm of the entire farmland. Introducing this tiered threshold enables macro-micro coordination and self-healing linkage of hardware resources. Under the premise of ensuring the uninterrupted supply of the main lifeline of the entire system's water network, it achieves seamless removal and isolation of physically damaged branches in the farmland irrigation area and reconfiguration of water resources at the most economical software logic cost.

[0048] Please see Figure 3 This diagram illustrates the state transitions of a hierarchical control irrigation module in an IoT-based smart agriculture irrigation monitoring and control system. Nodes represent three network states: normal aging state (control commands are generated by multiplying the basic control quantity by the dynamic baseline compensation coefficient); water shortage response state (water urgency adjustment coefficient is generated based on measured soil moisture, and the backup basic control quantity is adjusted); and fault isolation state (control commands are set to zero). Transitions between states are indicated by arrows: when the comprehensive anomaly evolution discrimination index reaches the first preset threshold, it transitions from the normal aging state to the water shortage response state; when it reaches the second preset threshold, it transitions to the fault isolation state; and when the discrimination index falls back below the corresponding threshold, it reverses. The normal aging state is indicated by a bold border as the initial state.

[0049] This concludes the embodiment.

[0050] It should be noted that the specific embodiments described above enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. Therefore, although the present invention has been described in detail in this specification, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention; and all technical solutions and improvements that do not depart from the spirit and scope of the present invention are covered within the protection scope of the patent of the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A smart agricultural irrigation monitoring and control system based on the Internet of Things, characterized in that, The system includes: The data acquisition module is used to acquire raw monitoring data collected by multi-source sensors deployed at each topology node of the farmland pipeline network at different time points, perform time-series alignment processing on the raw monitoring data, and generate a sequence of synchronous feature values ​​of each topology node at a unified target time. The anomaly analysis module is used to compare the actual flow parameters in the synchronous feature value sequence with the theoretical flow parameters derived from the preset virtual hydraulic model under the same control state, calculate the flow resistance distortion amplitude of each topological node, obtain the measured soil moisture of the farmland in the area corresponding to the topological node, and generate a comprehensive anomaly evolution discrimination index for each topological node based on the flow resistance distortion amplitude and the measured soil moisture. The dynamic reconstruction and error compensation module is used to extract the deviation between the actual flow parameters and the theoretical flow parameters under the first extreme working condition and the second extreme working condition, respectively, to perform iterative compensation on the two types of deviations and generate dynamic benchmark compensation coefficients. The first extreme working condition is the zero flow condition where the water pump stops and the entire network maintains pressure, and the second extreme working condition is the steady-state condition where the water pump runs at a constant frequency and there is no valve action. The hierarchical control irrigation module is used to classify the network status into levels based on the numerical range of the comprehensive anomaly evolution discrimination index, and generate control commands for the corresponding topology nodes in combination with the dynamic benchmark compensation coefficient. This is used to adjust the pump operating frequency and solenoid valve opening for irrigation control.

2. The smart agricultural irrigation monitoring and control system based on the Internet of Things according to claim 1, characterized in that, The step of performing time-series alignment processing on the original monitoring data to generate a sequence of synchronization feature values ​​for each topological node at a unified target time includes: A sliding window is constructed with a preset benchmark step size as the interval. The original monitoring data of each topology node at multiple historical time points within the sliding window are obtained. The attenuation weights are assigned according to the distance of each historical time point from the target time. The weighted original monitoring data of each historical time point are accumulated to generate the synchronization feature value sequence.

3. The IoT-based smart agricultural irrigation monitoring and control system according to claim 2, characterized in that, The attenuation weights are calculated and allocated according to a negative exponential attenuation function with the natural logarithm as the base, and the attenuation weights are smaller for historical time points further away from the target time.

4. The IoT-based smart agricultural irrigation monitoring and control system according to claim 1, characterized in that, The calculation of the flow resistance distortion amplitude of each topology node includes: Obtain the absolute value of the difference between the measured pressure drop of the upstream and downstream pressure sensors of the topological node and the theoretical pressure drop output by the virtual hydraulic model. After normalizing the absolute value of the difference, multiply it by the ratio of the measured instantaneous flow velocity at the topological node to the theoretical flow velocity output by the virtual hydraulic model to obtain the flow resistance distortion amplitude.

5. The IoT-based smart agricultural irrigation monitoring and control system according to claim 4, characterized in that, The comprehensive anomaly evolution discrimination index for each topological node, generated based on the flow resistance distortion amplitude and the measured soil moisture, includes: The first ratio is denoted as the relative ratio of the flow resistance distortion amplitude of each neighboring node in the spatially associated neighborhood set of the current topological node to the baseline value of the flow resistance distortion amplitude during the historical normal operation of the corresponding neighboring node. The second ratio is denoted as the relative ratio of the difference between the target soil moisture and the measured soil moisture in the corresponding area of ​​each neighboring node to the target soil moisture. The first ratio and the second ratio are multiplied and then weighted and accumulated according to the spatial distance from each neighboring node to the current topological node to generate the comprehensive anomaly evolution discrimination index.

6. The IoT-based smart agricultural irrigation monitoring and control system according to claim 5, characterized in that, The spatial distance weights from each neighboring node to the current topology node are generated based on the reciprocal of the total pipeline length between each neighboring node and the current topology node.

7. The IoT-based smart agricultural irrigation monitoring and control system according to claim 1, characterized in that, The generation of dynamic benchmark compensation coefficients includes: The relative deviation between the measured hydrostatic pressure of each topology node under the zero-flow condition and the theoretical hydrostatic pressure calculated based on the elevation difference is obtained; the relative deviation between the measured pressure drop and the theoretical pressure drop of each topology node under at least one of the steady-state conditions is obtained, and a weighted average is performed after allocating time forgetting decay weights according to the occurrence time of each steady-state condition; the relative deviation of the hydrostatic pressure and the relative deviation of the measured pressure drop after weighting are superimposed, and the benchmark compensation coefficient of the previous reconstruction cycle is iteratively updated to generate the dynamic benchmark compensation coefficient of the current reconstruction cycle.

8. The smart agricultural irrigation monitoring and control system based on the Internet of Things according to claim 1, characterized in that, The process of hierarchically classifying the pipeline network status based on the numerical range of the comprehensive anomaly evolution discrimination index, and generating control commands for corresponding topology nodes in conjunction with dynamic benchmark compensation coefficients, includes: When the comprehensive anomaly evolution discrimination index is less than the first preset threshold, the hierarchical irrigation control module determines that the pipeline network is in a normal aging state, multiplies the preset basic control quantity by the dynamic benchmark compensation coefficient of the current reconstruction cycle, and generates the control command; when the comprehensive anomaly evolution discrimination index is greater than or equal to the first preset threshold and less than the second preset threshold, the hierarchical irrigation control module determines that the pipeline network is in a water shortage response state, generates a water demand urgency adjustment coefficient based on the measured soil moisture, multiplies the standby basic control quantity by the water demand urgency adjustment coefficient, and generates the control command; when the comprehensive anomaly evolution discrimination index is greater than or equal to the second preset threshold, the hierarchical irrigation control module determines that the pipeline network is in a fault isolation state, and sets the control command to zero.

9. The IoT-based smart agricultural irrigation monitoring and control system according to claim 1, characterized in that, The multi-source sensors include pressure sensors, electromagnetic flow meters, and soil moisture sensors deployed at each topology node, as well as soil moisture sensors deployed in farmland areas.

10. The smart agricultural irrigation monitoring and control system based on the Internet of Things according to claim 1, characterized in that, The virtual hydraulic model is constructed based on the topology connectivity map of the pipeline geographic information system and pipeline elevation difference data. It is used to deduce the theoretical pressure, theoretical flow velocity and theoretical pressure drop of each topological node under the condition of given valve opening and pump frequency through fluid dynamics equations.

Images