Field water level monitoring method and system based on intelligent irrigation gate

Abstract

The invention relates to the technical field of intelligent irrigation water level monitoring, and discloses a field water level monitoring method and system based on an intelligent irrigation gate, and the method comprises the steps: preprocessing field crop information and field water level information, calling a crop growth model according to the field crop information and the field water level information, and obtaining crop water demand data; judging whether the current field water level meets the requirements of the crop growth stage or not, generating a water level regulation and control decision, sending the water level regulation and control decision to an upstream reservoir, obtaining an irrigation plan and controlling the reservoir to open a gate and discharge water; the cooperative control of the field upstream water source and the field crop water demand is realized.

Description

Technical Field

[0001] The invention relates to the technical field of intelligent irrigation water level monitoring, and discloses a field water level monitoring method and system based on an intelligent irrigation gate. Background Art

[0002] Traditional irrigation systems still use a relatively extensive irrigation method, relying on experience or fixed irrigation schedules for irrigation. However, the actual water demand of crops at different growth stages and the real-time water level in the field change over time. Currently, the field water level is restricted by the relationship between field irrigation facilities and natural water supply and cannot be quantified, resulting in the inability to control field irrigation. This makes field crops more dependent on natural conditions rather than human control, resulting in restricted crop growth. Traditional gate control methods cannot accurately adjust the opening according to actual water demand, resulting in difficulty in accurately controlling the amount of irrigation water. Summary of the invention

[0003] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0004] In order to solve the above technical problems, the main purpose of the present invention is to provide a field water level monitoring method and system based on intelligent irrigation gates, wherein the field water level monitoring method based on intelligent irrigation gates includes:

[0005] Preprocessing field crop information and field water level information;

[0006] According to the field crop information and field water level information, the crop growth model is retrieved to obtain the crop water demand data, determine whether the current field water level meets the needs of the crop growth stage, and generate water level control decisions;

[0007] Send water level control decisions to upstream reservoirs to obtain irrigation plans and control reservoir gate opening and water release;

[0008] The gate communication terminal receives crop water demand data and controls the gate opening and closing degree.

[0009] As a preferred solution of the field water level monitoring method based on the intelligent irrigation gate of the present invention, wherein:

[0010] Acquiring the field water level information through a field liquid level meter and preprocessing the field water level information;

[0011] The preprocessing of the field crop information includes enhancing the contrast and clarity of the collected field crop information and obtaining the growth status, plant color and density of the crops;

[0012] The multispectral data of field crop information is subjected to radiation correction and atmospheric correction, and the multispectral data is subjected to dimensionality reduction processing through principal component analysis to extract the spectral features of the field crops, and the spectral features of the field crops are used to analyze the physiological status of the crops.

[0013] As a preferred solution of the field water level monitoring method based on the intelligent irrigation gate of the present invention, wherein:

[0014] Setting an electronic tag for each field crop, wherein the electronic tag is used to store field crop information;

[0015] The crop growth model includes crop physiological characteristics, historical meteorological data, and soil condition data.

[0016] The crop growth model is also used to predict water demand data of field crops at different growth stages, including daily water demand and water level range.

[0017] As a preferred solution of the field water level monitoring method based on the intelligent irrigation gate of the present invention, wherein:

[0018] The crop growth model is used to output the water demand information of the field crop growth stage, including the name of the growth stage, the lower limit and upper limit of the suitable water level, and the daily water demand;

[0019] If the water level in the field is lower than the lower limit of the suitable water level for the current growth stage of the field crops, obtain additional water;

[0020] If the field water level is higher than the upper limit of the suitable water level for the current growth stage of the field crops, the field water level will be lowered by closing the irrigation gate.

[0021] As a preferred solution of the field water level monitoring method based on the intelligent irrigation gate of the present invention, wherein:

[0022] The water level control decision for each field with electronic tags is generated through the electronic tags and the amount of water added.

[0023] As a preferred solution of the field water level monitoring method based on the intelligent irrigation gate of the present invention, wherein:

[0024] The communication terminal of the field gate receives the field water level information and water level regulation decision through the wireless communication module. The gate control system receives the water level regulation decision and controls the opening and closing degree of the gate.

[0025] If the water level control decision is to close the irrigation gate, the gate will be completely closed. If the water level control decision is to adjust the gate opening degree, the difference between the lower limit of the field crop water level and the current field water level is obtained, and the gate opening adjustment amount is obtained. The irrigation time required to reach the required amount of supplementary water is obtained through the irrigation channel flow.

[0026] As a preferred solution of the field water level monitoring method based on the intelligent irrigation gate of the present invention, wherein:

[0027] Before retrieving the crop growth model according to the field crop information and the field water level information to obtain the crop water requirement data, the field water level monitoring method further includes:

[0028] Each field is abstracted into a node, and the adjacent relationship between two adjacent fields is abstracted into a line to form a node topology graph;

[0029] Embed a topology perception layer in the node topology graph, wherein the topology perception layer is used to monitor the sensor status and communication status corresponding to each node;

[0030] If the monitoring node of the topology perception layer fails, the field water level information of the neighboring nodes is interpolated and compensated to predict the field water level information of the failed node.

[0031] As a preferred solution of the field water level monitoring method based on the intelligent irrigation gate of the present invention, wherein:

[0032] The interpolation compensation is performed through the field water level information of the neighboring nodes to predict the field water level information of the failed node, including:

[0033] The field water level of the failed node is predicted based on the field water level information of all neighboring nodes, the distance between the failed node and all neighboring nodes, and a preset attenuation coefficient, wherein the attenuation coefficient is adjusted based on historical data errors.

[0034] As a preferred solution of the field water level monitoring method based on the intelligent irrigation gate of the present invention, wherein:

[0035] The attenuation coefficient is obtained based on historical data error adjustment, including:

[0036] Define the environment state, action space and reward function of the reinforcement learning model; the environment state includes the distribution density of neighboring nodes, crop type, and real-time meteorological conditions; the action space includes the discrete or continuous adjustment range of the attenuation coefficient; the reward function is calculated based on the error in the historical test results, the adjustment amplitude penalty coefficient, and the attenuation coefficient before and after each iterative update;

[0037] Use Q-learning or deep deterministic policy gradient algorithm to train the model through historical data so that the agent can learn to choose the optimal attenuation coefficient according to the environment.

[0038] The present invention discloses a field water level monitoring system based on an intelligent irrigation gate, wherein:

[0039] Information collection module, used to collect field crop information and field water level information;

[0040] A data processing module, used for preprocessing the collected field crop information and field water level information;

[0041] The crop analysis unit is used to receive the pre-processed field crop information and field water level information, and to establish field crop label quantities with field water level information for each field, and to retrieve the crop water requirement data of the crop growth model through the label quantities;

[0042] The gate control module is used to receive crop water demand data, generate water level regulation decisions, and output them to the upstream reservoir. It is also used to receive irrigation plans output by the upstream reservoir.

[0043] The upstream reservoir is used to receive water level regulation decisions, formulate irrigation plans based on the water level regulation decisions, and output them to the gate control module.

[0044] As a preferred solution of the field water level monitoring system based on the intelligent irrigation gate of the present invention, wherein:

[0045] The gate control module receives an irrigation plan and controls the gate opening according to the crop water requirement, irrigation time and flow parameters of the irrigation system included in the irrigation plan;

[0046] The water level control decision includes the field number of each field crop, the water level data of each field, the crop data and the water demand.

[0047] Beneficial effects of the present invention: The present application sets a crop analysis unit to output crop water demand data, a gate control module to output water level control decisions, and links the upstream reservoir to realize sequential control of the field gates. By real-time monitoring of crop growth status and field water level, and combining crop growth model data, it is possible to accurately judge the crop water demand and output accurate water level control decisions. When the field needs irrigation, it is possible to apply for an irrigation plan from the upstream reservoir, and according to the irrigation plan fed back by the reservoir, accurately control the opening and closing degree of the field gate, so that the irrigation process is carried out efficiently and orderly, and reduce the impact of natural conditions on the field water level. According to the crop water demand data, the field water level difference and the irrigation system parameters, the gate opening adjustment amount and the irrigation time are accurately calculated to realize precise control of the gate, and ensure that the field water level can quickly and accurately reach the range suitable for crop growth. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0049] Figure 1 It is a flow chart of the field water level monitoring method based on the intelligent irrigation gate of the present invention;

[0050] Figure 2 This is a composition diagram of the field water level monitoring system based on the intelligent irrigation gate of the present invention;

[0051] Figure 3 It is a schematic diagram of field crop irrigation in the field water level monitoring method based on intelligent irrigation gates of the present invention. DETAILED DESCRIPTION

[0052] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0053] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0054] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0055] Example 1

[0056] like Figure 1 As shown, the field water level monitoring method based on the intelligent irrigation gate includes:

[0057] Preprocessing field crop information and field water level information;

[0058] Collecting field crop information through an image sensor, wherein the field crop information includes crop growth status, plant color and density;

[0059] Acquiring the field water level information through a field liquid level meter and preprocessing the field water level information;

[0060] The preprocessing of the field crop information includes enhancing the contrast and clarity of the collected field crop information and obtaining the growth status, plant color and density of the crops;

[0061] The multispectral data of field crop information is subjected to radiation correction and atmospheric correction, and the multispectral data is subjected to dimensionality reduction processing through principal component analysis to extract the spectral features of the field crops, and the spectral features of the field crops are used to analyze the physiological status of the crops.

[0062] Furthermore, a specific implementation method for field water level information collection and preprocessing includes:

[0063] Install multiple image sensors in the field and fix them on brackets that are higher than the crops to ensure that representative images of the crop area can be captured. Use a distributed layout to fully cover the entire field, and determine the number and location of sensors based on the size and shape of the field.

[0064] The collection frequency is determined by the crop growth stage and actual needs. In the early stage of crop growth, the growth changes are relatively slow, so the collection frequency is appropriately reduced, such as collecting once every 2-3 days. In the vigorous growth period, crops change rapidly, and image data is collected once a day to capture changes in crop growth status in a timely manner.

[0065] The field crop information collected by the image sensor includes crop growth status, plant color and density. For example, the crop growth status can be judged by the overall shape of the crop in the image, the degree of leaf stretch, and whether there are signs of pests and diseases. The plant color information is obtained by analyzing the color values ​​of the image pixels. Different colors reflect the nutritional status of the crop or whether it is suffering from diseases. The plant density information is obtained by counting the plants in the image or estimating the number of plants per unit area using image processing algorithms.

[0066] The histogram equalization method is applied to redistribute the grayscale values ​​of the image to make the grayscale histogram distribution of the image more uniform, thereby enhancing the contrast of the image. For a darker and low-contrast crop image, after histogram equalization, the originally difficult to distinguish details become clearer, which is conducive to the subsequent analysis of the crop growth status, color and density.

[0067] The image is smoothed by Gaussian filtering to remove noise interference in the image, while retaining the edge and detail information of the image, further improving the clarity of the image. Gaussian filtering performs weighted averaging on each pixel in the image and the pixels in its neighborhood. The weight is determined by the Gaussian function, which effectively reduces random noise such as salt and pepper noise in the image.

[0068] The multispectral data collected by the image sensor is affected by factors such as the sensor's own characteristics and lighting conditions, resulting in inaccurate spectral reflectance data. Radiation correction eliminates the influence of these factors by calibrating the sensor's response characteristics. For example, a standard reference plate with known reflectivity is used to obtain its multispectral data under the same lighting conditions as the crop image. The crop multispectral data is corrected through the mean model so that the data can accurately reflect the crop's true spectral reflectance characteristics.

[0069] Gas molecules and aerosols in the atmosphere will scatter and absorb light, thus affecting the accuracy of multispectral data. According to the meteorological conditions when the data was collected (such as atmospheric humidity, temperature, aerosol content, etc.), the multispectral data is corrected to remove the influence of the atmosphere on the spectrum. For example, atmospheric radiation transfer models such as MODTRAN are used, relevant meteorological parameters are input, the influence of the atmosphere on spectra in different bands is calculated, and the collected crop multispectral data is corrected.

[0070] Multispectral data after radiation correction and atmospheric correction usually contains multiple bands with high data dimension, which increases the complexity and computational complexity of data processing. The principal component analysis method is used to convert the high-dimensional multispectral data into a set of new, mutually orthogonal low-dimensional data, namely, principal components. The principal components retain the variance information of the original data. For example, assuming that the original multispectral data contains 10 bands, it is converted into 3-5 principal components through principal component analysis. While reducing the data dimension, the key spectral features used to analyze the physiological state of crops are extracted, such as the reflectance changes of specific band combinations, which are used to reflect the moisture content of crops.

[0071] The application of histogram equalization method to enhance image contrast can make the indistinguishable details in the originally dark and low-contrast crop images clear, which helps to extract valuable information from the image more accurately and observe the crop growth conditions more clearly. By setting atmospheric correction, the data is corrected according to meteorological conditions to effectively remove the atmospheric influence. The design of the information acquisition module comprehensively considers the accuracy, efficiency and scientific analysis of field crop information acquisition, providing strong data support for precision agriculture decision-making.

[0072] Setting an electronic tag for each field crop, wherein the electronic tag is used to store field crop information;

[0073] The crop growth model includes multiple factors such as crop physiological characteristics, historical meteorological data, soil condition data, etc.

[0074] The crop growth model is also used to predict water demand data of field crops at different growth stages, specifically including daily water demand and water level range.

[0075] The crop growth model is used to output the water demand information of the field crop growth stage, including the name of the growth stage, the lower limit and upper limit of the suitable water level, and the daily water demand;

[0076] If the water level in the field is lower than the lower limit of the suitable water level for the current growth stage of the field crops, obtain additional water;

[0077] If the field water level is higher than the upper limit of the suitable water level for the current growth stage of the field crops, the field water level will be lowered by closing the irrigation gate.

[0078] The water level control decision for each field with electronic tags is generated through the electronic tags and the amount of water added.

[0079] Provide a method for generating water level control decisions:

[0080] For example, when used in rice fields, a unique electronic tag is set for each piece of rice planted in the farmland. The electronic tag has a storage function and is used to record specific information of the rice in the farmland, such as rice variety, planting time, farmland number, etc.

[0081] The rice growth model is constructed by comprehensively considering multiple factors such as rice physiological characteristics, historical meteorological data, and soil condition data. The rice physiological characteristics include the growth patterns and water requirements of rice in different growth stages; the historical meteorological data include information such as precipitation, temperature, light, and evaporation over the years; and the soil condition data include soil texture, fertility, and water holding capacity, which determine the soil's ability to provide water for rice.

[0082] By using the acquired multi-source data and through data analysis and modeling techniques, the rice growth model can predict the water requirement data of field rice at different growth stages, including daily water requirement and suitable water level range. For example, for rice, the daily water requirement during the tillering stage is X cubic meters per mu, and the suitable water level range is between Y1-Y2 centimeters.

[0083] The system reads the information in the electronic tags and retrieves the water requirement information of the current growth stage of the rice in the corresponding farmland from the rice growth model, including the name of the growth stage, the lower limit of the suitable water level, the upper limit of the suitable water level and the daily water requirement.

[0084] The real-time field water level is compared with the suitable water level range output from the rice growth model. If the field water level is lower than the lower limit of the suitable water level, it indicates that the rice is short of water and the amount of supplementary water needs to be calculated; if the field water level is higher than the upper limit of the suitable water level, it means that there is too much water in the field and measures need to be taken to lower the water level.

[0085] When the field water level is lower than the lower limit of the suitable water level, the amount of supplementary water is calculated. The calculation method of the supplementary water amount is: the difference between the lower limit of the suitable water level and the current field water level multiplied by the farmland area to obtain the total volume of water that needs to be supplemented. For example, if the lower limit of the suitable water level is 5 cm, the current field water level is 3 cm, and the farmland area is 10 mu (1 mu = 666.67 square meters), the amount of water required to be supplemented is (5-3)×0.01×10×666.67=133.334 cubic meters (0.01 is the coefficient for converting centimeters to meters), and the water level control decision for the farmland is generated.

[0086] When the field water level is higher than the upper limit of the suitable water level, the generated water level control decision is to close the irrigation gate to prevent more water from flowing in. At the same time, drainage facilities can be opened to lower the field water level according to actual conditions.

[0087] An actual production implementation process includes:

[0088] Use a dedicated electronic tag writing device to write information such as the rice variety "Rice No. 2", planting time "May 10, 20xx" and farmland number "001" into the electronic tag, and then firmly install the electronic tag in a conspicuous and non-damageable position in the farmland.

[0089] By consulting agricultural scientific research materials and conducting field test observations, we collected data on the physiological characteristics of rice in the region and in similar planting environments over the years, collected historical meteorological data provided by the local meteorological department, and obtained soil condition data by testing and analyzing farmland soil.

[0090] Use data analysis software (such as Python's related data analysis library or professional agricultural modeling software) to integrate and analyze the collected multi-source data to build a rice growth model. During the model construction process, continuously adjust parameters and algorithms to improve the accuracy of the model's prediction of rice water demand data.

[0091] After the end of each planting season, the actual rice growth data, meteorological data, and soil data of that year are fed back into the model to retrain the model so that it can better adapt to the changing environment and rice growth characteristics.

[0092] Install high-precision water level sensors in each farmland to monitor field water level data in real time and transmit the data to the central control system. The installation location of the water level sensor should be representative and able to accurately reflect the overall water level conditions of the farmland.

[0093] The central control system regularly reads the information in the electronic tags through wireless radio frequency identification (RFID) technology or other communication methods, and retrieves the water requirement information corresponding to the current growth stage of the rice from the rice growth model database based on the read farmland number, rice variety and planting time.

[0094] If the rice currently planted in the farmland is in the tillering stage, the suitable water level obtained from the model is in the range of 4-7 cm, while the real-time monitored field water level is 3 cm. It is determined that the field water level is below the lower limit of the suitable water level and water needs to be supplemented.

[0095] If the calculation shows that 200 cubic meters of water needs to be replenished and the irrigation pipe flow rate is 50 cubic meters per hour, the planned irrigation time is 200÷50=4 hours. Then, the system sends instructions to the irrigation equipment to adjust the irrigation gate opening and irrigation time to ensure sufficient water replenishment.

[0096] If the field water level is higher than the upper limit of the appropriate water level, the central control system immediately sends a closing command to the irrigation gate to stop irrigation. At the same time, check whether the farmland is equipped with drainage facilities (such as drainage pipes, drainage pumps, etc.). If there are drainage facilities, according to the degree of excessive field water level and the drainage capacity of the drainage facilities, a drainage plan is formulated to open the drainage pump for a certain period of time or open the drainage pipe gate to a specific opening to reduce the field water level to an appropriate range.

[0097] A unique electronic tag is set for each rice planted in each rice field, and specific information such as rice variety, planting time, and field number is recorded. This enables the system to accurately locate the rice in each field and realize personalized management of rice in different fields. The growth model is constructed by integrating rice physiological characteristics, historical meteorological data, and soil condition data, and various key factors affecting rice growth and water demand are fully considered. By obtaining the field water level in real time and comparing it with the suitable water level range output by the growth model, it can timely detect abnormal moisture conditions during rice growth and quickly make corresponding water level control decisions. When the field water level is lower than the lower limit of the suitable water level, the supplementary water volume is obtained to control the irrigation gate, making irrigation decisions more accurate. Linking the upstream reservoir irrigation plan helps to reasonably arrange irrigation resources and avoid waste or excessive use of water resources. When the water level is higher than the upper limit, the irrigation gate is closed in time and drainage facilities are considered to be opened, which can effectively prevent water accumulation in the field from damaging rice and ensure the normal growth of rice.

[0098] Send water level control decisions to upstream reservoirs, apply for irrigation plans and control reservoir gate opening and water release;

[0099] Send the water level control decision to the upstream reservoir. If the decision is to increase the amount of irrigation water, the application content includes the amount of water required and the irrigation time range. After receiving the application, the upstream reservoir management will formulate an irrigation plan;

[0100] The upstream reservoir feeds back the irrigation plan to the field gate communication terminal, and opens the gate to release water by generating the irrigation plan. The field gate communication terminal receives the crop water demand data and water level control decisions, and controls the opening and closing degree of the gate.

[0101] If the decision is to close the irrigation gate, the gate will be completely closed. If the decision is to increase the amount of irrigation water, the gate opening will be controlled through the irrigation plan including crop water demand, irrigation time and flow parameters of the irrigation system.

[0102] like Figure 3 As shown, a specific implementation method of linking farmland irrigation with upstream water supply includes:

[0103] Water level sensors, irrigation gates with communication terminals and control devices, and field control units are deployed in the field. The water level sensors are responsible for real-time monitoring of the field water level and transmitting the data to the field control unit. The irrigation gates receive instructions from the field control unit to control the degree of opening and closing. The field control unit is used to collect water level data, read crop-related information, communicate with the upstream reservoir and send instructions to the gates. A management system has been established in the upstream reservoir, which is used to receive water level regulation decision applications in the field, formulate irrigation plans, feed back the irrigation plans to the fields, and connect to the reservoir's gate control system to perform water release operations.

[0104] The field control unit determines whether to adjust the water level through real-time monitoring of field water level data and crop water requirement data obtained by retrieving the crop growth model. If the field water level is lower than the lower limit of the water level suitable for the current growth stage of the crop, it calculates how much water needs to be added, and then combines the actual water delivery capacity of the irrigation system and previous irrigation experience to estimate a reasonable irrigation time range and generate a decision to increase the amount of irrigation water. If the field water level is higher than the upper limit of the suitable water level, it generates a decision to close the irrigation gate.

[0105] The field control unit sends water level control decisions to the upstream reservoir management system through the wireless communication network in accordance with the standardized protocol (XML or JSON format). For example, the decision to increase the amount of irrigation water contains information such as the farmland number, decision type (increase water volume), amount of water to be supplemented, and irrigation time range in a specific format.

[0106] After receiving the water level control decision from the field, the upstream reservoir management system will first check and analyze the application information. If the decision is to increase the amount of irrigation water, a detailed irrigation plan will be formulated based on the current water storage situation of the reservoir, the water demand of other places, and the amount of water and time range applied for by the field. For example, if the water in the reservoir is sufficient and there is no conflict in the water use of other places, water will be released at a specific time at the flow rate required by the rice in the field.

[0107] The upstream reservoir management system feeds back the formulated irrigation plan to the field control unit through the same wireless communication network. The irrigation plan includes key information such as irrigation start time, end time and flow rate.

[0108] After receiving the irrigation plan and water level control decision feedback from the upstream reservoir, the field control unit analyzes and operates the gate opening.

[0109] If the irrigation plan is to increase the amount of irrigation water, the field control unit determines the opening of each irrigation gate based on the crop water demand, irrigation time, and known irrigation system flow parameters in the irrigation plan. For example, the total water demand, irrigation time, and total pipeline flow are obtained, and the amount of water that needs to pass through the gate during the irrigation time is calculated based on the proportion of the flow that a certain gate is responsible for, so that the gate should be adjusted to an opening that allows the corresponding flow of water to pass through. The field control unit sends a command to the corresponding irrigation gate communication terminal to adjust the gate to the specified opening.

[0110] If the irrigation plan is to close the irrigation gate, the field control unit directly sends a closing command to the irrigation gate communication terminal to completely close the gate.

[0111] Furthermore, during the entire irrigation process, the water level sensor monitors the changes in the field water level in real time. At the same time, the control device of the irrigation gate will also feed back the actual gate opening and water flow conditions to the field control unit, so that the field control unit can know the irrigation status at any time.

[0112] If the actual water level change is found to be different from the expected one during irrigation, or there is other situations such as sudden rainfall, equipment failure, etc., and the irrigation plan needs to be adjusted, the field control unit will regenerate the water level control decision and send it to the upstream reservoir management system to request adjustment of the irrigation plan. After receiving the new request, the upstream reservoir management system will re-evaluate and adjust the irrigation plan and then feed it back to the field control unit. The field control unit will adjust the opening and closing degree of the irrigation gate again according to the new plan to ensure that the field water level is always within the range suitable for crop growth.

[0113] Furthermore, Figure 3 In the field gate, the field gate includes irrigation gate and drainage gate. The irrigation gate is located in the river water area, and the drainage gate is located on the level with low slope in the river water area. The opening of the irrigation gate is controlled to control the amount of water for field crops. If the field water level demand decreases, the drainage gate is opened to lower the water level.

[0114] The water level control decision is sent to the upstream reservoir to obtain the required supplementary water volume and irrigation time range, so that the upstream reservoir can accurately understand the water demand of rice in the field, which is helpful for the reservoir management department to reasonably allocate water resources among many water demands according to actual needs, improve the utilization efficiency of water resources, and avoid waste or over-allocation of water resources; further, the upstream reservoir formulates an irrigation plan according to the received application and promptly feeds back to the field gate communication terminal to ensure that when the field needs irrigation, it can quickly obtain water supply to meet the water demand of rice growth, reduce the negative impact of untimely irrigation on rice growth, and ensure the normal growth and development of rice. Through clear information transmission and decision-making execution process, the close coordination between field irrigation demand and upstream reservoir water supply is achieved. The upstream reservoir formulates irrigation plan according to field demand. Irrigation plans are received and feedback is given, and the gates are controlled according to the plan in the field. The whole process is closely linked, which improves the coordination and stability of the entire irrigation system and reduces irrigation problems caused by poor information or uncoordinated operations. From the issuance of water level control decisions, to the upstream reservoir formulating and feedback of irrigation plans, to the field gates executing operations according to the plan, each link has clear tasks and information interaction, which enhances the reliability of the system and ensures that the irrigation system can operate stably under various circumstances and provide reliable water resources for rice planting. The upstream reservoir uniformly receives water level control decisions and formulates irrigation plans, which facilitates centralized management and macro decision-making. The reservoir management department comprehensively considers the water demand of multiple fields, the water storage situation of the reservoir and other relevant factors to formulate a more reasonable and optimized irrigation plan to improve the irrigation management level of the entire region.

[0115] The field gate communication terminal only needs to control the gate opening and closing degree according to the received crop water demand data, water level control decisions and irrigation plans. The operation is relatively simple and clear, which reduces the complexity of the work of field operators and reduces irrigation problems caused by human operational errors. It also facilitates the maintenance and management of irrigation equipment.

[0116] Example 2

[0117] like Figure 2 As shown in the figure, the field water level monitoring system based on the intelligent irrigation gate includes:

[0118] Information collection module, used to collect field crop information and field water level information;

[0119] A data processing module, used for preprocessing the collected field crop information and field water level information;

[0120] The crop analysis unit is used to receive the pre-processed field crop information and field water level information, and to establish field crop label quantities with field water level information for each field, and to retrieve the crop water requirement data of the crop growth model through the label quantities;

[0121] The gate control module is used to receive crop water demand data, generate water level regulation decisions, and output them to the upstream reservoir. It is also used to receive irrigation plans output by the upstream reservoir.

[0122] The gate control module receives an irrigation plan and controls the gate opening according to the crop water requirement, irrigation time and flow parameters of the irrigation system included in the irrigation plan;

[0123] The water level control decision includes the serial number of each field, the water level data of each field, the crop data and the water demand.

[0124] The upstream reservoir is used to receive water level regulation decisions, formulate irrigation plans based on the water level regulation decisions, and output them to the gate control module.

[0125] Example 3

[0126] Field water level monitoring methods based on smart irrigation gates also include:

[0127] Each field is abstracted into a node, and the adjacent relationship between two adjacent fields is abstracted into a line to form a node topology graph;

[0128] Embed a topology perception layer in the node topology graph, wherein the topology perception layer is used to monitor the sensor status and communication status corresponding to each node;

[0129] If the monitoring node of the topology perception layer fails, the field water level information of the neighboring nodes is interpolated and compensated to predict the field water level information of the failed node.

[0130] The specific implementation methods of the field water level monitoring method based on the intelligent irrigation gate include:

[0131] Each irrigation unit (field) is defined as a topological node. The node attributes include metadata such as geographic coordinates, area, crop type, and historical water level data. The Delaunay triangulation algorithm is used to build a neighbor relationship network to ensure that the distance between any adjacent nodes does not exceed 200 meters (to adapt to the transmission radius of the wireless sensor). The LoRaWAN protocol is used to realize real-time broadcast of node status, and supports self-reconstruction of the topological structure when adding / removing fields.

[0132] The topology awareness layer includes:

[0133] Deploy an STM32F4 series processor at each node, integrate sensor health detection circuits (such as power supply voltage monitoring and signal strength detection), monitor link quality through the RSSI value of the Zigbee module, set -85dBm as the communication failure threshold, and use 3-bit binary code to represent the node status (such as 001 for sensor failure, 010 for communication interruption, and 100 for power supply abnormality);

[0134] Predicted field water level information includes:

[0135] Conduct 72 hours of continuous monitoring before the irrigation season, establish a baseline database of water levels at each node, use historical data to train the LSTM network, optimize the interpolation algorithm parameters until the prediction error is less than 5%, and use the TDMA time slot allocation protocol to ensure that the communication time slots of adjacent nodes are staggered to avoid conflicts. The topology map coloring status (green normal / yellow warning / red fault) is displayed on the central platform. When a node failure is detected, the data collection of the three neighboring nodes is automatically triggered, the Kriging interpolation algorithm is called to generate the prediction value, and the compensation data is written to the database through the MQTT protocol. Based on the above design, the compensation data is automatically compared with the manual measurement value every week, and the model parameters are updated.

[0136] When a field node loses data due to sensor failure or communication interruption (such as equipment damaged by heavy rain), the system interpolates the water level data of neighboring nodes to predict the failed node information, avoiding the problem of local monitoring blind spots caused by traditional single node failures and ensuring the continuity of water level monitoring; the embedded topological perception layer monitors the status of each node device in real time (such as sensor health and communication quality), can warn of potential failures in advance, reduce the impact of sudden failures on the system, and associate the water level changes of adjacent fields into the topological network, using geographic spatial relationships (such as terrain elevation and soil permeability) for interpolation compensation, which is more in line with the actual water potential distribution law than independent monitoring of a single node. Through sub-node monitoring and centralized data processing, local autonomy is guaranteed and efficient coordination of global water resource scheduling is achieved.

[0137] The interpolation compensation is performed through the field water level information of the neighboring nodes to predict the field water level information of the failed node, including:

[0138] The field water level of the failed node is predicted based on the field water level information of all neighboring nodes, the distance between the failed node and all neighboring nodes, and a preset attenuation coefficient, wherein the attenuation coefficient is adjusted based on historical data errors.

[0139] A specific implementation method for predicting the field water level information of a failed node by interpolating and compensating the field water level information of adjacent nodes includes:

[0140] With the failed node as the center, adjacent nodes are dynamically screened based on a preset radius (e.g., 200 meters), giving priority to nodes that are in the same irrigation branch canal as the failed node, have similar soil types, or have high correlation with historical water level fluctuations.

[0141] The distance weight is adjusted according to the water flow direction between nodes (such as upstream / downstream relationship). For example, the influence coefficient of the downstream node on the upstream failed node is reduced by 30%. Based on the inverse proportional function of the Euclidean distance (such as a=1 / (1+d)), d is the distance between nodes, and the influence coefficient is mapped by the Euclidean distance a. The attenuation coefficient is determined, and the error between the historical interpolation prediction value and the actual value is recorded. The exponentially weighted moving average method (EWMA) is used to dynamically correct the attenuation coefficient. If the water level of a neighboring node deviates from the mean by more than 2 standard deviations, the node data is temporarily excluded.

[0142] Specific implementation methods include:

[0143] Water level sensors (such as float or pressure sensors) are deployed in the field, with a spacing of 50-100 meters between sensor nodes. Data is transmitted to the edge gateway via LoRa or NB-IoT. Each node records coordinates and builds a topological relationship diagram (marking water flow direction and channel connection relationships). It stores at least one year of water level data, including water level changes under different irrigation cycles and weather conditions, for training the attenuation coefficient adjustment model.

[0144] Through time series analysis (such as ARIMA model), the periodic law of water level changes is identified as the background reference for interpolation compensation. The gateway monitors the node heartbeat signal. If there is no response for three consecutive times, it is judged as failed, triggering the interpolation compensation process, smoothing filtering (Savitzky-Golay filtering) on ​​the data of adjacent nodes, eliminating instantaneous noise interference, obtaining the node prediction value, and calling the attenuation coefficient optimization results in the historical database.

[0145] When the failed node is restored, the predicted value is compared with the actual value, the attenuation coefficient is updated and fed back to the database.

[0146] The location of the failed node is marked on the monitoring platform, the predicted water level value is displayed (the actual value is distinguished by a dotted line), and a confidence interval (such as ±5cm) is generated. If the predicted water level exceeds the channel safety threshold (such as 90% of the channel design water level), a text message alarm is triggered and it is recommended to adjust the gate opening.

[0147] Quantifying geographic proximity through distance attenuation coefficients (such as inverse distance weighting) is more in line with the physical law of water level diffusion with distance. The attenuation coefficient is adjusted based on historical error feedback and can adapt to different regional characteristics (such as soil permeability and crop type). For example, in sandy soil areas, water levels fluctuate quickly, and the system automatically increases the attenuation coefficient to respond more quickly to changes in neighboring nodes. The outlier rejection mechanism avoids the impact of occasional sensor failures or short-term environmental interference (such as animals touching sensors) on predictions, ensuring data reliability. When a node fails, redundant compensation of multiple neighboring node data is used to maintain monitoring continuity and avoid interruptions to irrigation decisions caused by single point failures. Through spatial correlation modeling, dynamic parameter optimization, and abnormal robustness processing, "low-cost, high-precision, and adaptive" field water level monitoring is achieved, providing key technical support for the reliable operation of intelligent irrigation systems and refined management of water resources.

[0148] The attenuation coefficient is obtained based on historical data error adjustment, including:

[0149] Define the environment state, action space and reward function of the reinforcement learning model; the environment state includes the distribution density of neighboring nodes, crop type, and real-time meteorological conditions; the action space includes the discrete or continuous adjustment range of the attenuation coefficient; the reward function is calculated based on the error in the historical test results, the adjustment amplitude penalty coefficient, and the attenuation coefficient before and after each iterative update;

[0150] Use Q-learning or deep deterministic policy gradient algorithm to train the model through historical data so that the agent can learn to choose the optimal attenuation coefficient in a specific environment.

[0151] A specific implementation method for selecting an optimal attenuation coefficient includes:

[0152] The design of environmental states calculates the spatial density (such as the number of nodes per square meter) or topological density (such as the average connection distance between nodes) of neighboring nodes through the sensor network. For example, the Gaussian kernel density estimation method is used to calculate the local density index based on the node coordinates, and the crop types are encoded as discrete features (such as one-hot encoding or embedding layer), and the crop growth cycle parameters (such as the difference in water requirements during the growth period and maturity period) are introduced as auxiliary states to obtain real-time meteorological conditions including dynamic parameters such as temperature, humidity, and light intensity, which need to be converted into numerical features that can be processed by the model through normalization (such as Min-Max or Z-Score).

[0153] The design of the action space includes dividing the attenuation coefficient into discrete values ​​with fixed step sizes (such as {0.1, 0.2, ..., 1.0}). Each action corresponds to an adjustment amplitude. The output layer uses the Tanh activation function to map the action range to [-1, 1], and then linearly transforms it to the actual attenuation coefficient range (such as 0.1~1.0).

[0154] Further tanh function settings are adjusted according to actual production requirements.

[0155] The reward function design includes the need to balance error reduction, adjust stability and resource consumption. The mean square error (MSE) between the predicted value and the actual value based on historical data is Rerror=−α×MSE(ypred,ytrue), where α is the error weight coefficient, MSE(ypred,ytrue) is the error between the predicted value and the actual value, ypred is the predicted value, and ytrue is the actual value.

[0156] In order to prevent frequent and large adjustments, the adjustment amplitude penalty term is used to limit excessive adjustments of the reward function. The calculation method of the adjustment amplitude penalty term includes Rpenalty=−β×| − ∣, β is the penalty coefficient, is the reward function adjustment value at time t, is the adjustment value of the reward function at time t-1, Rpenalty is the adjustment range limit coefficient, and finally, the smoothness index of the attenuation coefficient change trend (such as the first-order difference variance) is introduced as an additional reward item. A hierarchical reward mechanism is adopted to decompose the overall goal into phased sub-goals.

[0157] The design of the Q-learning algorithm includes hashing the multidimensional environmental state (such as crop type + meteorological data) into a unique key value as the index of the Q table, constructing a state-action matrix with an initial value of zero or a small random number, setting a 10% probability of randomly selecting an action, a 90% probability of selecting the current optimal action, and updating the Q value.

[0158] Furthermore, the multi-dimensional environment sample content is stored, and the Q table is updated by randomly sampling small batches of data to break the data correlation. Finally, the change range of the Q value is monitored, and the training is terminated when the update amount is lower than the threshold.

[0159] Furthermore, the Q value is updated through the Bellman equation.

[0160] Specific implementation methods include:

[0161] Multidimensional environmental parameters (such as crop type, temperature, humidity, etc.) are discretized. For example, continuous meteorological data is binned by intervals, crop types are mapped to enumeration values, and finally all dimensions are concatenated into a unique string as a hash key. Each hash key corresponds to an independent state row in the Q table.

[0162] According to business needs, a set of executable actions (such as agricultural operations such as irrigation) is clearly defined. Each action corresponds to a column in the Q table, forming a two-dimensional state-action matrix.

[0163] Use zero initialization or small random numbers (such as ±0.01) to fill the matrix to ensure that the strategy in the initial exploration phase is random and avoid the local optimal trap.

[0164] Design a probability threshold controller: 10% probability of randomly selecting an action (exploring unknown strategies), 90% probability of selecting the action with the highest Q value in the current state (using the known optimal strategy). This probability can be dynamically adjusted through the attenuation mechanism, focusing on exploration in the early stage and biased towards utilization in the later stage.

[0165] Create a first-in-first-out (FIFO) storage queue to record the four-tuple (current state, action, reward, new state) of each interaction. Randomly extract small batch samples (such as 32 groups) from the pool each time an update is made to break the data time series correlation and improve training stability.

[0166] For each sample, calculate the target Q value: current reward + discount factor × maximum estimated Q value in the new state. Multiply the difference between the target value and the original Q value by the learning rate, and gradually correct the corresponding units in the matrix.

[0167] Track the Q table update range in real time and calculate the average change of the entire table or hot spot area.

[0168] Set a dynamic threshold. When the change in N consecutive updates is lower than the dynamic threshold, the model is judged to have converged and the training is terminated.

[0169] Cluster and group high-dimensional sparse states (such as merging similar weather conditions) to reduce the dimension of the Q table and improve query efficiency. In a multi-threaded environment, split the Q table area to allow different threads to batch process action updates of non-conflicting states and accelerate the training process.

[0170] A specific implementation method of applying the Deep Deterministic Policy Gradient (DDPG) algorithm includes:

[0171] Use the deep deterministic policy gradient algorithm to set the network structure, obtain noise perturbation and normalization.

[0172] Furthermore, the network structure includes a 3-layer fully connected network, the number of hidden layer nodes is 256, the activation function is ReLU, and the output layer is Tanh.

[0173] The input state and action are concatenated and passed through a 3-layer fully connected network to output the Q value.

[0174] By obtaining OU noise (Ornstein-Uhlenbeck Process) on the Actor output action, inertial disturbance is simulated.

[0175] Normalization is used to perform batch normalization on the input state to accelerate training convergence.

[0176] The environmental state integrates the density of neighboring nodes (reflecting the dynamics of network topology), crop types (reflecting scene specificity) and real-time meteorological conditions (external interference factors), enabling the model to perceive multi-dimensional dynamic variables. The action space supports discrete or continuous adjustment (such as the DDPG algorithm). The continuous space allows precision adjustment at the 0.1dB level, avoiding the quantization error caused by traditional discretization.

[0177] Importantly, it should be noted that the construction and arrangement of the present application shown in a plurality of different exemplary embodiments are only exemplary. Although only two embodiments are described in detail in this disclosure, it should be readily understood by those who refer to this disclosure that many modifications are possible, for example, the size, scale, structure, shape and ratio of various elements, and parameter values ​​(e.g., temperature, pressure, etc.), installation arrangement, use of materials, color, directional changes, etc., without substantially departing from the novel teachings and advantages of the subject matter described in the application. For example, the element shown as integrally formed can be composed of multiple parts or elements, the position of the element can be inverted or otherwise changed, and the nature or number or position of the discrete element can be changed or changed. Therefore, all such modifications are intended to be included in the scope of the present invention. The order or sequence of any process or method steps can be changed or reordered according to alternative embodiments. Any "device plus function" clause is intended to cover the structure of the execution function described in this article, and is not only structurally equivalent but also equivalent structure. Without departing from the scope of the present invention, other replacements, modifications, changes and omissions can be made in the design, operating conditions and arrangement of the exemplary embodiment. Therefore, the invention is not limited to a specific embodiment, but extends to numerous modifications still falling within the scope of the appended claims.

[0178] Additionally, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment may not be described (ie, those features that are not relevant to the best mode presently contemplated for carrying out the invention or those features that are not relevant to implementing the invention).

[0179] It should be understood that in the development of any actual implementation, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but for those of ordinary skill having the benefit of this disclosure, the development effort will be a routine task of design, fabrication, and production without undue experimentation.

[0180] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A field water level monitoring method based on intelligent irrigation gates, wherein the entire field area to be monitored is divided into a plurality of plots, each plot being equipped with sensors for monitoring field crop information and field water level information; the method is characterized in that: The field water level monitoring method comprises: Preprocessing field crop information and field water level information; According to the field crop information and field water level information, the crop growth model is retrieved to obtain the crop water demand data, determine whether the current field water level meets the needs of the crop growth stage, and generate water level control decisions; Send water level control decisions to upstream reservoirs to obtain irrigation plans and control reservoir gate opening and water release; The gate communication terminal receives crop water demand data and controls the gate opening and closing degree.

2. The field water level monitoring method based on intelligent irrigation gate according to claim 1 is characterized in that: Collecting field crop information through an image sensor, wherein the field crop information includes crop growth status, plant color and density; Acquiring the field water level information through a field liquid level meter and preprocessing the field water level information; The preprocessing of the field crop information includes enhancing the contrast and clarity of the collected field crop information and obtaining the growth status, plant color and density of the crops; The multispectral data of field crop information is subjected to radiation correction and atmospheric correction, and the multispectral data is subjected to dimensionality reduction processing through principal component analysis to extract the spectral features of the field crops, and the spectral features of the field crops are used to analyze the physiological status of the crops.

3. The field water level monitoring method based on intelligent irrigation gate according to claim 2 is characterized in that: Setting an electronic tag for each field crop, wherein the electronic tag is used to store field crop information; The crop growth model includes crop physiological characteristics, historical meteorological data, and soil condition data; The crop growth model is also used to predict water demand data of field crops at different growth stages, including daily water demand and water level range.

4. The field water level monitoring method based on intelligent irrigation gate according to claim 3 is characterized in that: The crop growth model is used to output the water demand information of the field crop growth stage, including the name of the growth stage, the lower limit and upper limit of the suitable water level, and the daily water demand; If the water level in the field is lower than the lower limit of the suitable water level for the current growth stage of the field crops, obtain additional water; If the field water level is higher than the upper limit of the suitable water level for the current growth stage of the field crops, the field water level will be lowered by closing the irrigation gate.

5. The field water level monitoring method based on intelligent irrigation gate according to claim 4 is characterized in that: The water level control decision for each field with electronic tags is generated through the electronic tags and the amount of supplemented water.

6. The field water level monitoring method based on intelligent irrigation gate according to claim 5 is characterized in that: The communication terminal of the field gate receives the field water level information and water level regulation decision through the wireless communication module. The gate control system receives the water level regulation decision and controls the opening and closing degree of the gate. If the water level control decision is to close the irrigation gate, the gate will be completely closed. If the water level control decision is to adjust the gate opening degree, the difference between the lower limit of the field crop water level and the current field water level is obtained, and the gate opening adjustment amount is obtained. The irrigation time required to reach the required amount of supplementary water is obtained through the irrigation channel flow.

7. The field water level monitoring method based on intelligent irrigation gate according to claim 1 is characterized in that: Before retrieving the crop growth model according to the field crop information and the field water level information to obtain the crop water requirement data, the field water level monitoring method further includes: Each field is abstracted into a node, and the adjacent relationship between two adjacent fields is abstracted into a line to form a node topology graph; Embed a topology perception layer in the node topology graph, wherein the topology perception layer is used to monitor the sensor status and communication status corresponding to each node; If the monitoring node of the topology perception layer fails, the field water level information of the neighboring nodes is interpolated and compensated to predict the field water level information of the failed node.

8. The field water level monitoring method based on intelligent irrigation gate according to claim 7 is characterized in that: The interpolation compensation is performed through the field water level information of the neighboring nodes to predict the field water level information of the failed node, including: The field water level of the failed node is predicted based on the field water level information of all neighboring nodes, the distance between the failed node and all neighboring nodes, and a preset attenuation coefficient, wherein the attenuation coefficient is adjusted based on historical data errors.

9. The field water level monitoring method based on intelligent irrigation gate according to claim 8 is characterized in that: The attenuation coefficient is obtained based on historical data error adjustment, including: Define the environment state, action space and reward function of the reinforcement learning model; the environment state includes the distribution density of neighboring nodes, crop type, and real-time meteorological conditions; the action space includes the discrete or continuous adjustment range of the attenuation coefficient; the reward function is calculated based on the error in the historical test results of field water level treatment, the adjustment amplitude penalty coefficient, and the attenuation coefficient before and after each iterative update; Use Q-learning or deep deterministic policy gradient algorithm to train the model through historical data so that the agent can learn to choose the optimal attenuation coefficient according to the environment.

10. A field water level monitoring system based on an intelligent irrigation gate, used to implement the field water level monitoring method based on an intelligent irrigation gate according to any one of claims 1 to 9, characterized in that: include: Information collection module, used to collect field crop information and field water level information; A data processing module, used for preprocessing the collected field crop information and field water level information; The crop analysis unit is used to receive the pre-processed field crop information and field water level information, and to establish a field crop label quantity with field water level information for each field, and to retrieve the crop water requirement data of the crop growth model through the label quantity; The gate control module is used to receive crop water demand data, generate water level regulation decisions, and output them to the upstream reservoir. It is also used to receive irrigation plans output by the upstream reservoir. The upstream reservoir is used to receive water level regulation decisions, formulate irrigation plans based on the water level regulation decisions, and output them to the gate control module.

Images