Intelligent detection method and system for vegetable and fruit planting environment parameters based on sensor

Abstract

The invention provides an intelligent detection method and system for vegetable and fruit planting environment parameters based on a sensor, and relates to the technical field of intelligent agriculture. Controllable physical intervention is applied to a sensor monitoring area in a selected time window, so that soil reaches a preset saturated stable state; obtaining an original output signal of the sensor in the stable state; the performance drift amount of the sensor can be accurately quantified by performing cross-time dimension comparison on the signal and a corresponding signal in an initial reference state; and based on the quantized performance drift amount, reconstructing a mapping relation between an output value of the sensor and a real soil moisture physical quantity, and converting a real-time output value of the sensor by adopting the reconstructed mapping relation in a non-intervention time period to obtain calibrated soil moisture data. The problem of measurement data deviation caused by long-term operation, environment change and self-aging of the sensor in the prior art is effectively solved.

Description

Technical Field

[0001] This application relates to the field of intelligent agriculture technology, and more specifically, to a sensor-based intelligent detection method and system for environmental parameters of fruit and vegetable cultivation. Background Technology

[0002] In modern agriculture, especially in the cultivation of high-value fruits and vegetables, deploying sensor-based intelligent monitoring systems has become crucial for ensuring precise control of environmental parameters and guaranteeing high-quality crop yields. This system monitors key data such as air temperature and humidity, soil moisture, and carbon dioxide concentration in real time. Among these, the soil moisture sensor, which measures the electrical properties of the soil to guide precise irrigation, is the core component of the system.

[0003] To improve soil fertility and structure, farms commonly apply organic fertilizers for extended periods. This process promotes the activity of soil microorganisms. As microorganisms decompose organic matter, they slowly alter the soil microenvironment, producing trace amounts of slightly corrosive byproducts such as organic acids. While these substances are not highly toxic, their long-term, persistent presence poses a subtle challenge to the metal electrodes (such as stainless steel or copper alloys) of sensors buried in the soil.

[0004] Over periods of months or even years, trace amounts of corrosion react slowly with the electrode surface, gradually forming an extremely thin (nanoscale) passivation layer or oxide film. This film, invisible to the naked eye, fundamentally alters the electrical properties between the electrode and the soil—causing a systematic, non-linear, slow drift in the output signal of any sensor, whether based on capacitance or impedance principles. For example, when the actual soil moisture content is stable, the original voltage value output by the sensor may continuously and slightly decrease.

[0005] The problem is that the calibration formula of the intelligent system is a static model established based on standard conditions when the sensor is newly installed. It cannot identify and compensate for the complex drift caused by long-term physicochemical changes. Therefore, the system continuously receives raw signals that are already biased. After being converted by the old formula, the final displayed moisture value may seem "normal," but in fact, it has gradually deviated from the true value.

[0006] Because this drift is gradual and non-linear, the resulting data deviation often remains within the system's preset "normal" range (e.g., 20%-30%). For example, even if the actual moisture content has dropped to the edge of drought at 20%, the system may still display 25%, without triggering any alarms. The system's anomaly detection function thus silently fails.

[0007] This leads to a classic dilemma: "normal data, abnormal plants." During inspections, managers might observe vague symptoms of chronic water stress in crops, such as stunted growth and discolored leaves. However, when checking the system, all data appears "normal" with no alerts. Because they place more trust in the system's "hard data," managers often rule out simple irrigation problems and suspect more complex causes like pests, diseases, or nutrient imbalances. This leads to time-consuming and laborious measures such as sending samples for testing and blindly applying fertilizers and pesticides, delaying the simplest intervention: adjusting irrigation.

[0008] As a result, crops are subjected to undetected chronic water stress during critical growth periods, leading to continuous damage to physiological processes such as photosynthesis and nutrient absorption. This not only reduces crop resistance but also ultimately results in decreased yield and poorer fruit quality (e.g., low sugar content, poor taste, and poor storage tolerance), directly impacting economic benefits. Technological tools originally intended to ensure production become, due to their internal hidden faults, indirect causes of misleading decision-making and production losses. This dilemma highlights the importance of achieving long-term online self-diagnosis and adaptive calibration of sensors in dynamically changing agricultural environments. Summary of the Invention

[0009] This application provides a sensor-based intelligent detection method and system for vegetable and fruit planting environmental parameters, aiming to solve the technical problem that in the long-term operation of existing intelligent detection systems, sensor performance drift leads to data deviation, which in turn misleads management decisions and affects crop growth and quality.

[0010] On the one hand, this application provides a sensor-based intelligent detection method for environmental parameters in fruit and vegetable cultivation, comprising: Within a selected time window, controllable physical intervention is applied to the sensor monitoring area to bring the soil to a preset saturated and stable state. The original output signal of the sensor in the saturated steady state is acquired and compared with the corresponding signal in the initial reference state across the time dimension to quantify the performance drift of the sensor. Based on the performance drift, the mapping relationship between the sensor output value and the actual soil moisture physical quantity is reconstructed; During non-intervention periods, the real-time output values ​​of the sensors are converted using the reconstructed mapping relationship to obtain calibrated soil moisture data.

[0011] Optionally, the controllable physical intervention includes: Injecting supersaturated water into the monitoring area to bring the soil to a preset saturated and stable state; or, The monitoring area is divided into a core area and a buffer isolation area. While injecting supersaturated water into the core area, ventilation and / or heating are applied to the buffer isolation area to suppress the transmission interference of external environmental variables to the core area.

[0012] Optionally, the step of acquiring the original output signal of the sensor in the saturated steady state and comparing it with the corresponding signal in the initial reference state across time dimensions to quantify the performance drift of the sensor includes: Acquire the original output signal of the sensor in this saturated steady state; The algebraic difference between the original signal in the current saturated state and the reference signal in the initial reference state is calculated as the performance drift.

[0013] Optionally, the step of acquiring the original output signal of the sensor in the saturated steady state and comparing it with the corresponding signal in the initial reference state across time dimensions to quantify the performance drift of the sensor includes: In the saturated stable state, a preset electrical excitation signal is applied to the sensor; Collect and analyze the sensor's output response signal under the electrical excitation signal; Extract at least one characteristic parameter characterizing the sensor performance from the output response signal; The extracted feature parameters are compared with the corresponding feature parameters obtained under the initial baseline state, and the difference is calculated as the performance drift.

[0014] Optionally, the step of reconstructing the mapping relationship between the sensor output value and the actual soil moisture physical quantity based on the performance drift includes: Identify the soil microenvironment characteristics of the monitoring area of ​​the sensor; Based on the characteristics of the soil microenvironment, the monitoring area is divided into multiple microenvironment management units; Establish a soil microenvironment file for each of the aforementioned microenvironment management units; Based on the performance drift and the soil microenvironment profile of the currently self-calibrated microenvironment management unit, the mapping relationship between the sensor output value and the actual soil moisture physical quantity is reconstructed.

[0015] Optionally, the step of establishing a soil microenvironment profile for each of the microenvironment management units includes: Real-time monitoring of soil characteristic data of the microenvironment management unit, wherein the soil characteristic data includes at least soil electrical conductivity, pH value and redox potential; When the change in the soil characteristic data relative to the archived value is detected to exceed a preset threshold, an archive update is triggered; Based on current soil characteristic data, update the parameters related to water adsorption capacity, infiltration rate, and oxygen exchange sensitivity in the soil microenvironment profile.

[0016] Optionally, the mapping relationship between the reconstructed sensor output value and the actual soil moisture physical quantity includes any of the following methods: The performance drift is used as a fixed offset to linearly correct the sensor output value; or... The sensor output value under saturated steady state is combined with the actual soil moisture physical quantity to form new data points, and the mapping function is refitted by combining the historical data point set.

[0017] Optionally, the step of converting the real-time output value of the sensor using the reconstructed mapping relationship to obtain calibrated soil moisture data during the non-intervention period includes: During non-intervention periods, when converting the real-time output value of the sensor using the reconstructed mapping relationship, the corresponding mapping relationship is selected to convert the real-time output value based on the location of the micro-environment management unit.

[0018] Optionally, the method for determining the selected time window includes: Obtain information on the current growth stage of the planted crop; Based on the preset mapping relationship between crop growth stages and calibration cycles, the trigger frequency of the selected time window is dynamically adjusted, wherein the trigger frequency during the fruiting period is higher than that during the seedling period.

[0019] On the other hand, this application provides a sensor-based intelligent detection system for environmental parameters in fruit and vegetable cultivation, the system comprising: The physical intervention module is used to apply controllable physical intervention to the sensor monitoring area within a selected time window so that the soil reaches a preset saturated and stable state. The signal acquisition module is used to acquire the original output signal of the sensor in this saturated steady state; The drift calculation module is used to compare the original output signal with the corresponding signal under the initial reference state across the time dimension in order to quantify the performance drift of the sensor. The mapping relationship reconstruction module is used to reconstruct the mapping relationship between the sensor output value and the actual soil moisture physical quantity based on the performance drift amount. The real-time monitoring module is used to convert the real-time output values ​​of the sensors using the reconstructed mapping relationship during non-intervention periods to obtain calibrated soil moisture data.

[0020] This application relates to a sensor-based intelligent detection method and system for environmental parameters in fruit and vegetable cultivation. By applying controllable physical intervention to the sensor monitoring area within a selected time window, the soil reaches a preset saturated and stable state, thereby acquiring the sensor's original output signal under this stable state. By comparing this signal with the corresponding signal under the initial reference state across time dimensions, the performance drift of the sensor can be accurately quantified. Based on the quantified performance drift, this application can reconstruct the mapping relationship between the sensor output value and the actual soil moisture physical quantity, and use the reconstructed mapping relationship to convert the sensor's real-time output value during non-intervention periods to obtain calibrated soil moisture data.

[0021] Through the above technical solution, this application effectively solves the problem of measurement data deviation caused by long-term operation, environmental changes, and aging of sensors in existing technologies. Existing systems, due to their static calibration models, cannot identify and compensate for long-term, nonlinear signal drift caused by changes in the surface characteristics of sensor electrodes, leading to the contradiction of "normal" displayed data but "abnormal" crop growth. This application, by introducing periodic physical intervention and a dynamic mapping relationship reconstruction mechanism, can capture and correct sensor performance drift in real time and accurately, ensuring the authenticity and reliability of soil moisture data. This avoids misjudgments by managers due to erroneous data, enabling timely and correct irrigation measures to ensure appropriate water supply for crops during critical growth stages, ultimately significantly improving crop health, yield, and quality, overcoming the technical dilemma of data distortion and misleading decision-making in existing intelligent detection systems. Attached Figure Description

[0022] To illustrate this application more clearly, the accompanying drawings used in the embodiments will be briefly described below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0023] Figure 1 The diagram above illustrates a flowchart of a sensor-based intelligent detection method for environmental parameters in fruit and vegetable cultivation. Figure 2 The diagram above illustrates a schematic of a sensor-based intelligent detection system for environmental parameters in fruit and vegetable cultivation.

[0024] Figure reference numerals: 100, Intelligent detection system for vegetable and fruit planting environment parameters based on sensors; 10, Physical intervention module; 20, Signal acquisition module; 30, Drift calculation module; 40, Mapping relationship reconstruction module; 50, Real-time monitoring module. Detailed Implementation

[0025] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0026] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0027] In modern agriculture, especially in the cultivation of high-value fruits and vegetables, precise monitoring and control of environmental parameters are crucial for ensuring healthy crop growth and high-quality yields. However, existing sensor-based intelligent detection systems experience gradual degradation of sensor performance over long-term operation, leading to deviations in measurement data. These deviations are often difficult for existing systems to detect, potentially misleading management decisions and ultimately impacting crop growth and economic benefits. The calibration correction formulas for the initial sensor measurements in traditional intelligent detection systems are based on the standard performance of new sensors and initial environmental conditions, failing to effectively identify and compensate for long-term, nonlinear signal drift caused by changes in sensor electrode surface characteristics. This static calibration model cannot adapt to gradual, nonlinear sensor performance degradation, resulting in seemingly normal system data while crop growth continues to experience problems.

[0028] like Figure 1 The diagram illustrates an exemplary flowchart of a sensor-based intelligent detection method for environmental parameters in fruit and vegetable cultivation. This application proposes a sensor-based intelligent detection method for environmental parameters in fruit and vegetable cultivation, comprising: S10, within the selected time window, apply controllable physical intervention to the sensor monitoring area to make the soil reach a preset saturated and stable state; In this context, "sensor" typically refers to various sensors used to measure soil moisture content, such as capacitive sensors, resistive sensors, time-domain reflectometry (TDR) sensors, or frequency-domain reflectometry (FDR) sensors. These sensors indirectly infer soil moisture content by measuring the electrical properties of the soil (such as dielectric constant, resistance, or impedance). The monitoring area refers to the actual soil region where sensors are deployed and data is collected. A saturated steady state refers to a state where, after controlled physical intervention, the soil moisture content reaches saturation and remains relatively stable over a certain period. This provides a unified benchmark for evaluating sensor performance under known steady-state conditions.

[0029] S20: Acquire the original output signal of the sensor in the saturated steady state and compare it with the corresponding signal in the initial reference state across the time dimension to quantify the performance drift of the sensor. The initial reference state refers to the reference point where the sensor's performance is at its best or in a known standard state after initial installation or rigorous calibration, and the corresponding signal data is used as the reference for subsequent performance drift calculations.

[0030] S30, Based on the performance drift, reconstruct the mapping relationship between the sensor output value and the actual soil moisture physical quantity; S40, during non-intervention periods, uses the reconstructed mapping relationship to convert the real-time output value of the sensor to obtain calibrated soil moisture data.

[0031] Within a selected time window, controlled physical interventions are applied to the sensor-monitored area to bring the soil to a preset saturation and stable state. The implementation of these controlled physical interventions can be varied. For example, supersaturated water can be injected into the monitoring area using manual or automated irrigation systems until the soil is fully saturated and excess water begins to seep out, at which point the soil moisture content reaches its maximum and tends to stabilize. Another approach is to construct a temporary sealed enclosure above the monitoring area and then spray water mist into the enclosure using a misting device, gradually wetting the soil surface until it reaches saturation. Alternatively, drainage ditches can be set up around the monitoring area, continuously supplying water until the ditches begin to discharge water steadily, at which point the soil moisture reaches a dynamic equilibrium saturation state. All these methods aim to create a known and stable soil moisture environment for accurate subsequent evaluation of sensor performance.

[0032] The raw output signal of the sensor in its saturated steady-state state is acquired and compared with the corresponding signal in the initial reference state across time dimensions to quantify the sensor's performance drift. Several methods can be used to acquire the raw output signal of the sensor in its saturated steady-state state. For example, the analog or digital signals such as voltage, current, or frequency output in the saturated steady-state state can be read directly through the sensor's data acquisition interface. Alternatively, the sensor can be connected to a data logger, and the output signal can be continuously recorded at a preset sampling frequency over a period of time in the saturated steady-state state. The average or stable value can then be taken as the raw output signal. After acquiring the raw output signal in the current saturated steady-state state, it needs to be compared with the corresponding signal in the initial reference state. The signal in the initial reference state can be a reference value measured under standard saturated soil conditions when the sensor is newly manufactured or calibrated for the first time. The comparison method can be a simple algebraic subtraction of the current raw signal and the initial reference signal to obtain a difference, which is the performance drift. For example, if the initial reference signal is 2.5V and the raw signal in the current saturated steady-state state is 2.3V, then the performance drift is -0.2V.

[0033] Based on performance drift, the mapping relationship between sensor output values ​​and actual soil moisture physical quantities is reconstructed. The method for reconstructing the mapping relationship can be selected according to actual needs and sensor type. A simple approach is to use the performance drift as a fixed offset to linearly correct the real-time sensor output value. For example, if the drift is -0.2V, then in subsequent real-time monitoring, 0.2V is added to each voltage value output by the sensor to compensate for the drift. Another more complex but more accurate approach is to use the sensor output value under the current saturated steady state (the corresponding actual soil moisture physical quantity is known to be saturated) as a new data point, and combine it with a set of historical calibration data points to refit a new mapping function. For example, statistical methods such as least squares, polynomial fitting, or nonlinear regression can be used to update the original calibration curve with the new data points, thereby obtaining a mapping relationship that better reflects the current sensor performance.

[0034] During non-intervention periods, the reconstructed mapping relationship is used to convert the real-time sensor output values ​​to obtain calibrated soil moisture data. After the mapping relationship reconstruction is completed, the raw signals from the sensor's real-time output are continuously received during routine non-intervention monitoring periods. At this time, the old, uncalibrated mapping relationship is no longer used; instead, the newly reconstructed mapping relationship is used to process these real-time signals. For example, if the reconstructed mapping relationship is a linear function Y = aX + b (where X is the sensor output value and Y is the calibrated soil moisture data), the real-time acquired X value is substituted into this function to calculate the calibrated Y value. If the reconstructed mapping relationship is a more complex nonlinear function or lookup table, it will be converted accordingly. In this way, even if sensor performance drifts, its output real-time data can be accurately converted into the true physical quantity of soil moisture, thus providing a reliable basis for subsequent irrigation decisions.

[0035] The overall working principle of this application is to periodically apply controllable physical intervention to the sensor monitoring area until it reaches a preset saturation and stable state, thereby providing a unified and repeatable benchmark for evaluating sensor performance. In this stable state, the sensor's raw output signal is acquired and compared across time dimensions with the corresponding signal in the initial benchmark state. This allows for the precise quantification of the performance drift caused by long-term operation and environmental interaction. This drift directly reflects sensor performance degradation, indicating the degree of deviation between the sensor output and the actual physical quantity. Based on the quantified performance drift, the mapping relationship between the sensor output value and the actual soil moisture physical quantity can be dynamically reconstructed. This reconstruction can be a simple linear correction or a refitting of a more complex mapping function using new data points, enabling the calibration model to adapt to gradual changes in sensor performance. Finally, during non-intervention periods, the reconstructed mapping relationship is used to convert the sensor's real-time output value, ensuring that the obtained soil moisture data is calibrated and accurate, thus avoiding data distortion caused by sensor drift and providing reliable data support for precision management of vegetable and fruit cultivation.

[0036] The core innovation of this application lies in the introduction of a "controllable physical intervention" and a "performance drift quantification" mechanism, and on this basis, the realization of "dynamic reconstruction of the mapping relationship." Traditional intelligent detection systems use static calibration correction formulas for the original sensor measurements during deployment, failing to effectively identify and compensate for long-term, nonlinear signal drift caused by changes in the surface characteristics of the sensor electrodes. This static calibration model cannot adapt to gradual, nonlinear sensor performance degradation, resulting in seemingly normal data but persistent problems in crop growth. In contrast, this application applies controllable physical intervention to the sensor monitoring area within a selected time window, bringing the soil to a preset saturated stable state, providing a unified and repeatable benchmark for evaluating sensor performance. Based on this, by acquiring the sensor's original output signal in this saturated stable state and comparing it across time dimensions with the corresponding signal in the initial benchmark state, the performance drift of the sensor can be accurately quantified. This dynamic drift calculation mechanism enables real-time sensing of sensor performance changes. Furthermore, based on the quantified performance drift, this application can dynamically reconstruct the mapping relationship between the sensor output value and the actual soil moisture physical quantity, thereby enabling the calibration model to adapt to gradual changes in sensor performance. This dynamic calibration mechanism enables the conversion of the sensor's real-time output value using the reconstructed mapping relationship during non-intervention periods, thereby obtaining calibrated soil moisture data. This effectively solves the problem of inaccurate data caused by performance drift during long-term sensor operation.

[0037] In some embodiments, the controllable physical intervention includes: Injecting supersaturated water into the monitoring area to bring the soil to a preset saturated and stable state; or, The monitoring area is divided into a core area and a buffer isolation area. While injecting supersaturated water into the core area, ventilation and / or heating are applied to the buffer isolation area to suppress the transmission interference of external environmental variables to the core area.

[0038] Specifically, the first method involves directly applying excessive moisture to the sensor's monitoring area to ensure the soil pores are fully filled, thereby achieving a preset saturated stable state. This method is simple to operate and suitable for scenarios with relatively low environmental control requirements. The second method is more refined, its core being the improvement of the accuracy and anti-interference capability of the saturated stable state through regional division and differentiated intervention. Specifically, the monitoring area is divided into a core area and a buffer isolation area. The core area is the region actually monitored by the sensor and needs to reach a saturated stable state, while the buffer isolation area surrounds the core area. It aims to form a physical barrier by applying ventilation and / or heating to effectively suppress the conduction interference of external environmental factors (such as temperature fluctuations, humidity changes, evaporation, etc.) on the moisture state of the core area. Ventilation can promote moisture evaporation in the buffer isolation area or regulate its internal gas composition, while heating can maintain the temperature gradient between the buffer isolation area and the core area, further reducing the impact of external heat or moisture on the core area.

[0039] The technical solution of this application effectively improves the accuracy and stability of achieving a stable soil saturation state by providing two controllable physical intervention methods. The first method, through direct injection of supersaturated water, can quickly saturate the soil, providing a basis for subsequent signal acquisition. The second method introduces the concepts of a core zone and a buffer zone, and applies aeration and / or heating to the buffer zone to actively construct a controlled microenvironment. It is precisely because of the inhibitory effect of the buffer zone on external environmental variables that the soil in the core zone can be maintained more stably and uniformly at the preset saturation state, thus providing a reliable benchmark condition for the accurate quantification of sensor performance drift.

[0040] Through the above technical solution, this application can significantly improve the control accuracy and anti-interference ability of soil saturation stability. In particular, by dividing the core area and buffer isolation area and applying ventilation and / or heating, the interference of the external environment on the moisture state of the core area is effectively isolated, ensuring that the sensor acquires signals under more ideal saturation conditions. As a result, the acquired raw output signal can more accurately reflect the performance state of the sensor itself, thereby making the quantification of performance drift more precise, and ultimately improving the reliability and accuracy of the calibrated soil moisture data.

[0041] For example, suppose a soil moisture sensor needs to be calibrated periodically in a vegetable and fruit greenhouse. To ensure calibration accuracy, a second controllable physical intervention method can be used. Specifically, a ring-shaped buffer zone is first set up around the sensor monitoring area. This buffer zone can be constructed of porous material and integrates a micro fan and heating wire. While injecting supersaturated water into the core area to achieve a stable saturation state, the micro fan in the buffer zone is activated to ventilate and accelerate water evaporation, and the heating wire is turned on to maintain the temperature difference between the buffer zone and the external environment. In this way, the influence of air humidity and temperature fluctuations in the greenhouse on soil moisture evaporation and infiltration in the core area can be effectively prevented, ensuring that the soil in the core area remains in a highly stable saturated state throughout the calibration time window, thus providing a solid foundation for accurately calculating the sensor performance drift.

[0042] In some embodiments, the step of acquiring the original output signal of the sensor in the saturated steady state and comparing it with the corresponding signal in the initial reference state across time dimensions to quantify the performance drift of the sensor includes: Acquire the original output signal of the sensor in this saturated steady state; The algebraic difference between the original signal in the current saturated state and the reference signal in the initial reference state is calculated as the performance drift.

[0043] The process involves acquiring the sensor's raw output signal under saturated steady-state conditions. This is the direct electrical or digital signal generated by the sensor after controlled physical intervention is applied to the sensor's monitoring area to bring the soil to a preset saturated steady-state, without any calibration or processing. This signal reflects the sensor's instantaneous response under specific saturated conditions. Further, the algebraic difference between the current saturated steady-state raw signal and the initial reference signal is calculated as the performance drift. This is achieved by comparing the raw output signal acquired under saturated steady-state conditions with the corresponding reference signal recorded during initial installation or calibration (i.e., the initial reference state). This comparison is typically performed using simple subtraction, and the resulting difference is quantified as the sensor's performance drift. The initial reference signal can be the ideal output value measured under standard saturated conditions before the sensor is put into use, or a reference value measured under the same saturated conditions by a known high-performance sensor. This algebraic difference directly reflects the deviation in sensor performance over time or due to environmental changes.

[0044] The technical solution of this application directly acquires the original output signal of the sensor in a saturated steady state and calculates the algebraic difference between it and the reference signal in the initial reference state, thus directly and quantitatively reflecting the actual changes in sensor performance. This quantification method based on algebraic difference simplifies the calculation process of drift, making the identification of performance drift more intuitive and efficient. By measuring in a saturated steady state, the consistency of soil moisture conditions can be ensured, thereby eliminating the influence of soil moisture changes on the sensor output signal, allowing the calculated drift to be more accurately attributed to the sensor's own performance degradation or long-term deviations caused by environmental factors.

[0045] The above technical solution enables a direct and easily implemented method to accurately quantify sensor performance drift. This algebraic difference-based calculation method avoids complex model building or multi-parameter fitting, reducing computational complexity and resource consumption while ensuring the accuracy and reliability of drift assessment. This provides precise input for subsequently reconstructing the mapping relationship between sensor output values ​​and actual soil moisture physical quantities, thereby effectively improving the accuracy of soil moisture data calibration.

[0046] In some embodiments, the step of acquiring the original output signal of the sensor in the saturated steady state and comparing it with the corresponding signal in the initial reference state across time dimensions to quantify the performance drift of the sensor includes: In the saturated stable state, a preset electrical excitation signal is applied to the sensor; Collect and analyze the sensor's output response signal under the electrical excitation signal; Extract at least one characteristic parameter characterizing the sensor performance from the output response signal; The extracted feature parameters are compared with the corresponding feature parameters obtained under the initial baseline state, and the difference is calculated as the performance drift.

[0047] Specifically, a specific form of electrical signal is applied to the sensor to achieve the effect of applying a preset electrical excitation signal, such as AC voltage, AC current, pulse signal, or frequency sweep signal, to the sensor in the saturated steady state. The purpose is to actively detect the electrical characteristics of the sensor, enabling it to produce an analyzable response under controlled conditions, thereby revealing its internal state in greater depth.

[0048] The acquisition and analysis of the sensor's output response signal under the electrical excitation signal can be understood as obtaining its response data by measuring changes in voltage, current, impedance, capacitance, frequency, or phase after receiving the electrical excitation signal. These response signals are then analyzed, for example, through spectral analysis, time-domain analysis, or parameter fitting, to identify patterns or values ​​related to sensor performance.

[0049] In practical applications, at least one characteristic parameter characterizing the sensor's performance is extracted from the output response signal. Specifically, this involves filtering out quantities from the complex response signal that can sensitively reflect changes in the sensor's physical or chemical properties. For example, for capacitive sensors, the capacitance value, loss tangent (tan δ), or quality factor (Q value) at a specific frequency can be extracted; for resistive sensors, the resistance value or temperature coefficient can be extracted; and for resonant sensors, the resonant frequency or bandwidth can be extracted. The selection of these characteristic parameters should be based on the specific type of sensor and its potential degradation modes.

[0050] Furthermore, the characteristic parameters measured by the sensor in its current saturated steady state are compared with the same characteristic parameters measured by the sensor in its initial reference state when it was brand new or well-calibrated. This comparison can be a simple algebraic difference, percentage change, or a more complex statistical difference analysis. The resulting difference directly and quantitatively reflects the degree to which the sensor performance changes over time or in the operating environment, i.e., the performance drift.

[0051] The technical solution of this application, by applying a preset electrical excitation signal to the sensor under saturated steady-state conditions and collecting and analyzing its output response signal, can actively and deeply probe the internal electrical characteristics of the sensor. Compared with passive monitoring that relies solely on the raw output signal, this active excitation method enables the sensor to generate richer and more diagnostically valuable response data under controlled conditions. By extracting specific characteristic parameters from these response signals, minute changes in the sensor's internal structure or materials can be captured more accurately, such as electrode oxidation and dielectric constant drift of dielectric materials. These changes are the root causes of sensor performance drift. It is precisely because these deep-seated performance indicators can be identified that the quantification of performance drift becomes more accurate and comprehensive, thereby overcoming the limitations of traditional methods in capturing complex degradation mechanisms.

[0052] The above technical solution enables more accurate and comprehensive quantification of sensor performance drift. This method not only detects the overall drift of the sensor output signal but also identifies the specific physical or chemical changes causing the drift by analyzing characteristic parameters, thereby improving the accuracy and reliability of drift calculation. This refined drift quantification allows for the establishment of a calibration model that better reflects the sensor's current actual state when reconstructing the mapping relationship between sensor output values ​​and real soil moisture physical quantities. This significantly improves the accuracy and stability of calibrated soil moisture data, providing more reliable data support for precision irrigation and environmental management in vegetable and fruit cultivation.

[0053] For example, assuming a capacitive soil moisture sensor is used, its performance drift may manifest as changes in effective area due to electrode corrosion or changes in dielectric constant due to aging of the dielectric material. To accurately quantify this drift, a sinusoidal alternating current excitation signal with a frequency of 1 MHz and a peak-to-peak value of 1 V can be applied to the sensor under saturated steady-state conditions. Subsequently, the sensor's output response signal under the excitation signal is acquired using an impedance analyzer, such as its complex impedance spectrum. From the complex impedance spectrum, the equivalent capacitance value and loss tangent value at a specific frequency point (e.g., 1 MHz) can be extracted as characteristic parameters characterizing the sensor's performance. The currently measured capacitance value and loss tangent value are compared with the corresponding values ​​recorded by the sensor under initial reference conditions (e.g., during factory calibration), and the difference or percentage change is calculated. For example, if the initial capacitance value is 100 pF and the currently measured value is 95 pF, the capacitance drift is -5 pF; if the initial loss tangent value is 0.01 and the currently measured value is 0.012, the loss tangent drift is 0.002. These differences are used as the performance drift of the sensor for subsequent mapping reconstruction. In this way, even minute degradations within the sensor can be accurately captured and quantified, ensuring the accuracy of calibration.

[0054] In some embodiments, the step of reconstructing the mapping relationship between the sensor output value and the actual soil moisture physical quantity based on the performance drift includes: Identify the soil microenvironment characteristics of the monitoring area of ​​the sensor; Based on the characteristics of the soil microenvironment, the monitoring area is divided into multiple microenvironment management units; Establish a soil microenvironment file for each of the aforementioned microenvironment management units; Based on the performance drift and the soil microenvironment profile of the currently self-calibrated microenvironment management unit, the mapping relationship between the sensor output value and the actual soil moisture physical quantity is reconstructed.

[0055] Specifically, a detailed analysis of the soil within the monitoring area is conducted to determine its physical, chemical, and biological properties. These properties affect the soil's moisture retention capacity and the sensor's response, enabling the identification of the soil microenvironment characteristics within the sensor's monitoring area. For example, soil texture (the ratio of sandy soil, loam, and clay), organic matter content, bulk density, pH value, electrical conductivity, and cation exchange capacity can be analyzed. These characteristics can be obtained through initial soil sampling, laboratory analysis, or real-time monitoring using auxiliary soil property sensors.

[0056] Based on the similarities or differences in the identified soil microenvironment characteristics, the entire sensor monitoring area is logically divided into several sub-regions with relatively homogeneous soil characteristics. This allows the monitoring area to be divided into multiple microenvironment management units according to the soil microenvironment characteristics. This division can be performed using the spatial analysis functions of a Geographic Information System (GIS), clustering algorithms, or by combining the experience of agronomic experts. The aim is to ensure that the soil environment within each microenvironment management unit is relatively stable, thereby making the mapping relationship established within that unit more representative.

[0057] In practical applications, a dedicated data record or configuration file is created for each defined microenvironment management unit, establishing a soil microenvironment profile for each unit. This profile stores soil microenvironmental characteristic data specific to that unit, such as soil type, typical moisture retention curve parameters, infiltration rate, and other local environmental factors that may affect sensor performance. This profile serves as the basic reference data for sensor calibration within that unit.

[0058] Furthermore, based on the performance drift and the soil microenvironment profile of the currently self-calibrated microenvironment management unit, the mapping relationship between the sensor output value and the actual soil moisture physical quantity is reconstructed. In reconstructing this mapping relationship, a generic model is no longer used; instead, two key pieces of information are incorporated: first, the sensor's own performance drift, reflecting the deviation caused by changes in the sensor over time or the operating environment; and second, the specific soil microenvironment profile of the microenvironment management unit where the sensor is located, providing the inherent characteristics of the soil in that area. By comprehensively considering these two factors, a more targeted and accurate local mapping model can be established, thereby accurately converting the sensor's output value into the actual soil moisture physical quantity under that specific microenvironment.

[0059] The technical solution of this application introduces the identification of soil microenvironment characteristics in the sensor monitoring area and divides the monitoring area into multiple microenvironment management units accordingly, thereby establishing a unique soil microenvironment profile for each unit. Due to the differences in soil microenvironment characteristics, the response characteristics of the sensor may exhibit subtle variations in different areas. By establishing a profile for each microenvironment management unit and reconstructing the mapping relationship by incorporating performance drift, the calibration process can fully consider the influence of the local soil environment on the sensor output. Therefore, the mapping relationship between the sensor output value and the actual soil moisture physical quantity is no longer a single global model, but rather a personalized adjustment for each microenvironment management unit, thus more accurately reflecting the actual soil moisture status of that unit.

[0060] The above technical solution effectively addresses the problem of insufficient sensor calibration accuracy caused by the heterogeneity of the soil microenvironment in traditional methods. By precisely identifying and managing different soil microenvironment units and establishing customized mapping relationships for each unit, the accuracy and regional applicability of soil moisture data calibration are significantly improved. This enables the provision of more reliable and refined soil moisture monitoring data in complex planting environments, providing a more solid data foundation for precision irrigation and environmental management of vegetables and fruits, thereby helping to optimize water resource utilization efficiency and promote healthy crop growth.

[0061] For example, suppose a large vegetable and fruit growing area is monitored by sensors. First, soil sampling and analysis of the area identify two main soil microenvironment characteristics: one part is sandy loam, and the other is clay loam. Based on these characteristics, the monitoring area is divided into two microenvironment management units: a "sandy loam unit" and a "clay loam unit." Subsequently, soil microenvironment profiles are established for each unit, recording key parameters such as water retention characteristics and infiltration rates for each soil type. When a sensor is self-calibrated, its corresponding microenvironment management unit is first determined (e.g., it is located in the "sandy loam unit"). Then, combining the sensor's performance drift with the soil microenvironment profile of the "sandy loam unit," a mapping relationship between the sensor output value and the actual soil moisture physical quantity, applicable to this specific sandy loam environment, is reconstructed. When the sensor monitors in real time during non-intervention periods, its output value is converted using this customized mapping relationship, resulting in more accurate calibrated soil moisture data for that sandy loam environment. Similarly, the sensors located in the "clay loam unit" will also use their corresponding files to reconstruct mapping relationships and convert data.

[0062] In some embodiments, the step of establishing a soil microenvironment profile for each of the microenvironment management units includes: Real-time monitoring of soil characteristic data of the microenvironment management unit, wherein the soil characteristic data includes at least soil electrical conductivity, pH value and redox potential; When the change in the soil characteristic data relative to the archived value is detected to exceed a preset threshold, an archive update is triggered; Based on current soil characteristic data, update the parameters related to water adsorption capacity, infiltration rate, and oxygen exchange sensitivity in the soil microenvironment profile.

[0063] Specifically, the soil characteristic data of the microenvironment management unit is monitored in real time. Data related to soil physicochemical properties is continuously collected through auxiliary or integrated sensors deployed within each microenvironment management unit. This soil characteristic data includes at least soil electrical conductivity, pH value, and redox potential. Soil electrical conductivity reflects soil salinity, pH value indicates soil acidity / alkalinity, and redox potential is closely related to soil aeration and microbial activity. These parameters are key factors affecting soil moisture adsorption, infiltration, and sensor response.

[0064] Specifically, when the detected change in soil characteristic data relative to the recorded values ​​exceeds a preset threshold, a record update is triggered. This means that the real-time monitored soil characteristic data is continuously compared with the historical data recorded in the soil microenvironment record of the microenvironment management unit. Once the change in any key parameter (such as conductivity, pH value, or redox potential) exceeds the preset allowable range, a significant change in the soil microenvironment is identified, and the record update process is automatically initiated. The preset threshold can be set empirically based on crop type, soil type, and planting management strategy, or dynamically adjusted through a machine learning model to ensure the timeliness and accuracy of record updates.

[0065] In practical applications, based on current soil characteristic data, the parameters related to water adsorption capacity, infiltration rate, and oxygen exchange sensitivity in the soil microenvironment profile are updated. This means that after a profile update is triggered, the latest soil characteristic data is used to recalculate or adjust parameters related to soil moisture retention capacity (such as water-holding curve parameters), the rate at which water moves through the soil (such as saturated hydraulic conductivity), and the oxygen supply in the soil (affecting root respiration and microbial activity). These updated parameters more accurately reflect the actual response of the soil to moisture, thus providing a more precise input for subsequently reconstructing the mapping relationship between sensor output values ​​and actual soil moisture physical quantities.

[0066] The technical solution of this application effectively solves the problem of soil microenvironment archives becoming outdated due to environmental changes in basic technical solutions by introducing a real-time monitoring and dynamic updating mechanism. It is precisely because of the real-time monitoring of key characteristic data such as soil conductivity, pH value, and redox potential, and the setting of thresholds to trigger archive updates, that the soil microenvironment archive can always maintain a high degree of consistency with actual soil conditions. When soil properties change, the parameters related to water adsorption capacity, infiltration rate, and oxygen exchange sensitivity in the archive can be promptly sensed and updated, thereby ensuring that the soil microenvironment information used to reconstruct the mapping relationship between sensor output values ​​and actual soil moisture physical quantities is the most up-to-date and accurate. This dynamic adaptability avoids inaccurate mapping relationships caused by changes in the soil environment, significantly improving the reliability of calibrated soil moisture data.

[0067] Through the above technical solution, this application overcomes the limitations of traditional methods that rely on static or untimely updates to soil microenvironment profiles. By monitoring soil characteristic data in real time and dynamically updating the profiles according to preset thresholds, the accuracy and timeliness of the soil microenvironment profiles are ensured. This allows for full consideration of the dynamic changes in soil physicochemical properties when reconstructing the mapping relationship between sensor output values ​​and actual soil moisture physical quantities, thereby significantly improving the accuracy and reliability of calibrated soil moisture data. This provides more precise environmental parameter data for vegetable and fruit cultivation, facilitating refined water and fertilizer management and optimizing crop growth.

[0068] For example, suppose the soil microenvironment profile of a microenvironment management unit initially records a pH of 6.5 and an electrical conductivity of 0.8 mS / cm. During the planting process, due to changes in fertilization or irrigation water quality, real-time monitoring shows that the soil pH of this unit gradually decreases to 5.8, and the electrical conductivity increases to 1.2 mS / cm. If the preset pH change threshold is 0.5 and the electrical conductivity change threshold is 0.3 mS / cm, then the changes in both pH and electrical conductivity exceeding the preset thresholds will be detected, triggering a profile update. Based on the current pH of 5.8 and electrical conductivity of 1.2 mS / cm, the parameters in the microenvironment management unit profile related to soil moisture adsorption capacity (e.g., acidic soils may affect the adsorption of certain ionic water), infiltration rate (e.g., high salinity may reduce infiltration rate), and oxygen exchange sensitivity (e.g., pH changes affect microbial activity and soil structure) will be reassessed and updated. For example, the moisture adsorption capacity parameter may be adjusted to reflect the actual water retention capacity of the soil in more acidic, high-salinity environments, and the infiltration rate parameter may be adjusted accordingly. The updated archive will be used to reconstruct the mapping between sensor output values ​​and actual soil moisture physical quantities, ensuring that the accuracy of sensor calibration can be maintained even if the soil environment changes significantly.

[0069] In some embodiments, the mapping relationship between the reconstructed sensor output value and the actual soil moisture physical quantity includes any of the following methods: The performance drift is used as a fixed offset to linearly correct the sensor output value; or... The sensor output value under saturated steady state is combined with the actual soil moisture physical quantity to form new data points, and the mapping function is refitted by combining the historical data point set.

[0070] After determining the sensor's performance drift, this drift is treated as a constant error term. Specifically, this fixed offset can be directly subtracted from or added to the sensor's real-time output value to correct systematic deviations caused by factors such as sensor aging and environmental changes. For example, if the performance drift indicates that the sensor output value is generally too high, this offset will be used to adjust the real-time output value downward during the conversion process, thereby obtaining soil moisture data closer to the true value. The advantages of this method are its simple calculation, strong real-time performance, and suitability for scenarios where the drift characteristics are relatively stable and exhibit linear changes.

[0071] On the other hand, a new data point is formed by combining the sensor output value under saturated steady-state conditions with the actual soil moisture physical quantity. This data point is then combined with a historical data point set to refit the mapping function. Reliable data points obtained under controlled physical intervention are used to update or re-establish the conversion model between the sensor output and the actual physical quantity. Specifically, when the soil reaches a preset saturated steady-state, the sensor output signal is known, and the actual soil moisture physical quantity (e.g., saturated water content) is also determined. This allows for the formation of a precise calibration point. This new data point is then integrated into the existing historical calibration data point set, for example, by recalculating or adjusting the parameters of the mapping function using least squares, regression analysis, or other curve fitting algorithms. This method can more flexibly adapt to sensor nonlinear drift or complex changes in its response characteristics over time, thereby improving the accuracy and robustness of the calibration.

[0072] The technical solution of this application aims to effectively address the performance drift problem that may occur in sensors during long-term use by providing two different mapping relationship reconstruction methods. When using the performance drift amount as a fixed offset for linear correction, its working principle is based on the assumption that the sensor drift mainly manifests as an overall shift in the output signal. By directly compensating for this offset, the real-time output value of the sensor can be quickly and effectively adjusted to a more accurate range. This method is based on a simplified assumption about the drift mode and is suitable for situations where the drift behavior is relatively linear and stable.

[0073] Furthermore, the working principle is more refined when a new data point is formed by combining the sensor output value under saturated steady-state conditions with the actual soil moisture physical quantity, and then refitting the mapping function using a set of historical data points. By obtaining a highly reliable calibration point under controlled conditions, this technical solution can use this new information to correct or update the original calibration curve of the sensor. This new data point serves as an "anchor point" for the current sensor state, participating in the refitting process of the mapping function together with historical data. This allows the reconstructed mapping function to better reflect the sensor's current actual response characteristics, thereby overcoming the problem of nonlinear drift or changes in the shape of the response curve that simple linear correction may not be able to solve. Thus, regardless of whether the sensor drift is linear or nonlinear, the technical solution of this application can provide a corresponding effective strategy for calibration.

[0074] Through the above technical solutions, this application can flexibly select appropriate mapping relationship reconstruction strategies based on the specific drift characteristics of the sensor and actual application requirements. Using a linear correction method enables rapid and low-computational-cost calibration, particularly suitable for scenarios with high real-time requirements and relatively simple drift patterns, effectively maintaining the basic accuracy of soil moisture data. Furthermore, employing a method that combines new data points to refit the mapping function significantly improves the accuracy and adaptability of calibration, especially when sensor performance undergoes complex nonlinear changes. By fully utilizing historical data and the latest saturated stable-state data, the reconstructed mapping relationship can more accurately reflect the complex correspondence between the sensor and the actual soil moisture physical quantity, thereby ensuring the reliability and effectiveness of long-term monitoring data and providing a more solid data foundation for the refined management of vegetable and fruit growing environments.

[0075] In some embodiments, the step of converting the real-time output value of the sensor using the reconstructed mapping relationship to obtain calibrated soil moisture data during the non-intervention period includes: During non-intervention periods, when converting the real-time output value of the sensor using the reconstructed mapping relationship, the corresponding mapping relationship is selected to convert the real-time output value based on the location of the micro-environment management unit.

[0076] Specifically, the aforementioned microenvironment management unit refers to multiple relatively homogeneous sub-regions into which the sensor's monitoring area is divided based on soil microenvironment characteristics. Each microenvironment management unit can be assigned a unique identifier or coordinate information to clarify its spatial location within the entire monitoring area. When the sensor's real-time output value needs to be converted, the location of the microenvironment management unit where the sensor is located is first identified. Subsequently, based on this location information, a mapping relationship corresponding to that specific microenvironment management unit is selected from a pre-established mapping relationship library. This corresponding mapping relationship is reconstructed based on the soil microenvironment profile and performance drift of the microenvironment management unit, thus more accurately reflecting the relationship between the sensor output and the actual soil moisture physical quantities within the unit.

[0077] The technical solution of this application effectively solves the calibration accuracy problem caused by the spatial heterogeneity of the soil microenvironment by introducing the concept of a microenvironment management unit and selecting a corresponding mapping relationship based on its location. Specifically, since the soil characteristics of different microenvironment management units may differ, the sensor's response to the same real soil moisture physical quantity may vary in different units. By establishing a dedicated mapping relationship for each microenvironment management unit and selecting it according to the specific unit where the sensor is located during real-time conversion, it can be ensured that the applied mapping relationship is highly matched with the actual working environment of the sensor. This makes the calibration process more targeted, avoids the errors that may be caused by a "one-size-fits-all" global mapping, and thus improves the local accuracy of soil moisture data.

[0078] Through the above technical solution, this application can significantly improve the spatial accuracy and reliability of calibrated soil moisture data acquired during non-intervention periods. Because each microenvironment management unit applies a mapping relationship that matches its own characteristics, it can more accurately reflect the true soil moisture status of a local area. This is of great significance for implementing precision agriculture management, such as zoned irrigation and precision fertilization, helping to optimize water resource utilization efficiency, improve crop yield and quality, and reduce resource waste and environmental burden caused by inaccurate data.

[0079] For example, suppose a large vegetable and fruit growing base is divided into multiple microenvironment management units, such as Zone A, Zone B, and Zone C, based on soil type, topography, or historical planting data. During the calibration phase, corresponding mapping relationships MA, MB, and MC are reconstructed for the soil microenvironment characteristics and sensor performance drift of Zones A, B, and C, respectively. When a sensor located in Zone A outputs a signal in real time, the system identifies the sensor as being in Zone A and automatically selects mapping relationship MA to convert its output value, obtaining the calibrated soil moisture data for Zone A. Similarly, when a sensor in Zone B or Zone C outputs a signal, MB or MC is selected for conversion, respectively. In this way, even if the soil characteristics of different areas differ, it ensures that the soil moisture data for each area has undergone targeted high-precision calibration.

[0080] In some embodiments, the method for determining the selected time window includes: Obtain information on the current growth stage of the planted crop; Based on the preset mapping relationship between crop growth stages and calibration cycles, the trigger frequency of the selected time window is dynamically adjusted, wherein the trigger frequency during the fruiting period is higher than that during the seedling period.

[0081] Specifically, obtaining information on the current growth stage of the planted crop refers to determining the specific growth stage of the crop, such as seedling stage, vegetative growth stage, flowering stage, and fruiting stage, through various means, such as visual observation, image recognition technology, growth model prediction, or sensor data analysis. Crops at different growth stages exhibit significant differences in their soil moisture requirements and sensitivity. Furthermore, based on a preset mapping relationship between crop growth stages and calibration cycles, the trigger frequency of the selected time window is dynamically adjusted. A database or rule set is pre-established, defining the sensor calibration frequencies required for different crops at different growth stages. For example, in growth stages where crops are more sensitive to water changes or have high water requirements (such as the fruiting stage), the calibration trigger frequency is set higher to ensure the accuracy of soil moisture data; while in stages with relatively low water requirements or less sensitivity to water changes (such as the seedling stage), the calibration frequency can be appropriately reduced. This dynamic adjustment mechanism allows calibration activities to more accurately serve the actual growth needs of the crop.

[0082] The technical solution of this application dynamically adjusts the trigger frequency of a selected time window by acquiring information on the current growth stage of the planted crop and combining it with a preset mapping relationship between crop growth stages and calibration cycles. This is because crops at different growth stages have different sensitivities and requirements for soil moisture. For example, during the fruiting stage, crops have higher accuracy requirements for water and respond more severely to water stress, thus requiring more frequent sensor calibration to ensure real-time data accuracy. Conversely, during the seedling stage, crops may have relatively high tolerance to water changes or lower water requirements, allowing for a more appropriate reduction in calibration frequency. This dynamic adjustment mechanism based on crop growth stages makes sensor calibration no longer fixed but intelligently adaptable to the physiological needs of the crop, thus avoiding unnecessary frequent calibrations and preventing data inaccuracies caused by insufficient calibration during critical periods.

[0083] The above technical solutions can significantly improve the relevance and effectiveness of sensor calibration. Specifically, by closely linking the calibration frequency to the crop growth stage, higher-precision soil moisture data can be obtained during critical growth periods (such as the fruiting stage), providing a reliable basis for precision irrigation, thereby optimizing the crop growth environment and improving yield and quality. Simultaneously, appropriately reducing the calibration frequency during non-critical periods effectively reduces unnecessary physical intervention and resource consumption, lowers operating costs, and achieves a balance between calibration efficiency and resource utilization efficiency.

[0084] For example, suppose a tomato planting base uses the intelligent detection method of this application. During the seedling stage of tomatoes, since they are relatively insensitive to changes in water content, sensor calibration can be set to be performed every two weeks. When tomatoes enter the flowering and fruiting stages, their water demand increases significantly, and they become more sensitive to water stress. At this time, the calibration frequency will be automatically adjusted to once a week or even once every three days according to a preset mapping relationship. For example, during the fruiting stage, controlled physical intervention will be triggered more frequently to obtain the original output signal under saturated and stable conditions, and to calculate the performance drift and reconstruct the mapping relationship. This dynamic adjustment ensures that the accuracy of soil moisture data is maximized during the most critical period of tomato growth, thereby guiding precision irrigation and avoiding the impact of insufficient or excessive water on fruit quality and yield.

[0085] This application also proposes a sensor-based intelligent detection system for environmental parameters in fruit and vegetable cultivation, such as... Figure 2 As shown, a sensor-based intelligent detection system 100 for vegetable and fruit growing environment parameters includes: The physical intervention module 10 is used to apply controllable physical intervention to the sensor monitoring area within a selected time window so that the soil reaches a preset saturated and stable state. The signal acquisition module 20 is used to acquire the original output signal of the sensor in the saturated steady state; The drift calculation module 30 is used to compare the original output signal with the corresponding signal in the initial reference state across the time dimension in order to quantify the performance drift of the sensor. The mapping relationship reconstruction module 40 is used to reconstruct the mapping relationship between the sensor output value and the actual soil moisture physical quantity based on the performance drift amount. The real-time monitoring module 50 is used to convert the real-time output value of the sensor using the reconstructed mapping relationship during non-intervention periods to obtain calibrated soil moisture data.

[0086] The overall working principle of this application is that the system periodically applies controllable physical intervention to the sensor monitoring area through a physical intervention module, bringing it to a preset saturation and stable state, thereby providing a unified and repeatable benchmark for evaluating sensor performance. In this stable state, the signal acquisition module acquires the sensor's raw output signal, and the drift calculation module compares it across time with the corresponding signal of the sensor in the initial benchmark state, accurately quantifying the performance drift caused by long-term operation and environmental interaction. This drift is a direct manifestation of sensor performance degradation, reflecting the degree of deviation between the sensor output and the actual physical quantity. Based on the quantified performance drift, the mapping reconstruction module can dynamically reconstruct the mapping relationship between the sensor output value and the actual soil moisture physical quantity. This reconstruction can be a simple linear correction or a refitting of a more complex mapping function using new data points, enabling the calibration model to adapt to gradual changes in sensor performance. Finally, during non-intervention periods, the real-time monitoring module uses the reconstructed mapping relationship to convert the real-time output values ​​of the sensors, ensuring that the obtained soil moisture data is calibrated and accurate, thereby avoiding data distortion caused by sensor drift and providing reliable data support for the precision management of vegetable and fruit cultivation.

[0087] The sensor-based intelligent detection system for vegetable and fruit growing environmental parameters proposed in this application aims to solve the common problem in existing technologies where long-term sensor performance drift leads to inaccurate data and misleading decision-making. Traditional intelligent detection systems typically rely on static calibration models, which are established in the early stages of sensor deployment but cannot adapt to the gradual and non-linear performance degradation of sensors due to environmental interactions and aging. This results in seemingly normal system data, but persistent problems in crop growth, making it difficult for managers to accurately assess and implement effective irrigation measures.

Claims

1. A sensor-based intelligent detection method for environmental parameters in fruit and vegetable cultivation, characterized in that, include: Within a selected time window, controllable physical intervention is applied to the sensor monitoring area to bring the soil to a preset saturated and stable state. The original output signal of the sensor in the saturated steady state is acquired and compared with the corresponding signal in the initial reference state across the time dimension to quantify the performance drift of the sensor. Based on the performance drift, the mapping relationship between the sensor output value and the actual soil moisture physical quantity is reconstructed; During non-intervention periods, the real-time output values ​​of the sensors are converted using the reconstructed mapping relationship to obtain calibrated soil moisture data.

2. The intelligent detection method for vegetable and fruit planting environment parameters based on sensors according to claim 1, characterized in that, The controlled physical intervention includes: Injecting supersaturated water into the monitoring area to bring the soil to a preset saturated and stable state; or, The monitoring area is divided into a core area and a buffer isolation area. While injecting supersaturated water into the core area, ventilation and / or heating are applied to the buffer isolation area to suppress the transmission interference of external environmental variables to the core area.

3. The intelligent detection method for vegetable and fruit planting environment parameters based on sensors according to claim 1, characterized in that, The step of acquiring the original output signal of the sensor in the saturated steady state and comparing it with the corresponding signal in the initial reference state across time dimensions to quantify the performance drift of the sensor includes: Acquire the original output signal of the sensor in this saturated steady state; The algebraic difference between the original signal in the current saturated state and the reference signal in the initial reference state is calculated as the performance drift.

4. The intelligent detection method for vegetable and fruit planting environment parameters based on sensors according to claim 1, characterized in that, The step of acquiring the original output signal of the sensor in the saturated steady state and comparing it with the corresponding signal in the initial reference state across time dimensions to quantify the performance drift of the sensor includes: In the saturated stable state, a preset electrical excitation signal is applied to the sensor; Collect and analyze the sensor's output response signal under the electrical excitation signal; Extract at least one characteristic parameter characterizing the sensor performance from the output response signal; The extracted feature parameters are compared with the corresponding feature parameters obtained under the initial baseline state, and the difference is calculated as the performance drift.

5. The intelligent detection method for vegetable and fruit planting environment parameters based on sensors according to claim 1, characterized in that, The step of reconstructing the mapping relationship between the sensor output value and the actual soil moisture physical quantity based on the performance drift includes: Identify the soil microenvironment characteristics of the monitoring area of ​​the sensor; Based on the characteristics of the soil microenvironment, the monitoring area is divided into multiple microenvironment management units; Establish a soil microenvironment file for each of the aforementioned microenvironment management units; Based on the performance drift and the soil microenvironment profile of the currently self-calibrated microenvironment management unit, the mapping relationship between the sensor output value and the actual soil moisture physical quantity is reconstructed.

6. The intelligent detection method for vegetable and fruit planting environment parameters based on sensors according to claim 5, characterized in that, The step of establishing a soil microenvironment file for each of the microenvironment management units includes: Real-time monitoring of soil characteristic data of the microenvironment management unit, wherein the soil characteristic data includes at least soil electrical conductivity, pH value and redox potential; When the change in the soil characteristic data relative to the archived value is detected to exceed a preset threshold, an archive update is triggered; Based on current soil characteristic data, update the parameters related to water adsorption capacity, infiltration rate, and oxygen exchange sensitivity in the soil microenvironment profile.

7. The intelligent detection method for vegetable and fruit planting environment parameters based on sensors according to claim 1, characterized in that, The mapping relationship between the reconstructed sensor output value and the actual soil moisture physical quantity includes any of the following methods: The performance drift is used as a fixed offset to linearly correct the sensor output value; or... The sensor output value under saturated steady state is combined with the actual soil moisture physical quantity to form new data points, and the mapping function is refitted by combining the historical data point set.

8. The intelligent detection method for vegetable and fruit planting environment parameters based on sensors according to claim 1, characterized in that, The step of converting the real-time output value of the sensor using the reconstructed mapping relationship to obtain calibrated soil moisture data during the non-intervention period includes: During non-intervention periods, when converting the real-time output value of the sensor using the reconstructed mapping relationship, the corresponding mapping relationship is selected to convert the real-time output value based on the location of the micro-environment management unit.

9. The intelligent detection method for vegetable and fruit planting environment parameters based on sensors according to claim 1, characterized in that, The methods for determining the selected time window include: Obtain information on the current growth stage of the planted crop; Based on the preset mapping relationship between crop growth stages and calibration cycles, the trigger frequency of the selected time window is dynamically adjusted, wherein the trigger frequency during the fruiting period is higher than that during the seedling period.

10. A sensor-based intelligent detection system for environmental parameters in fruit and vegetable cultivation, characterized in that, The system includes: The physical intervention module is used to apply controllable physical intervention to the sensor monitoring area within a selected time window so that the soil reaches a preset saturated and stable state. The signal acquisition module is used to acquire the original output signal of the sensor in this saturated steady state; The drift calculation module is used to compare the original output signal with the corresponding signal under the initial reference state across the time dimension in order to quantify the performance drift of the sensor. The mapping relationship reconstruction module is used to reconstruct the mapping relationship between the sensor output value and the actual soil moisture physical quantity based on the performance drift amount. The real-time monitoring module is used to convert the real-time output values ​​of the sensors using the reconstructed mapping relationship during non-intervention periods to obtain calibrated soil moisture data.

Images