A rural domestic sewage resource utilization treatment method and system

Abstract

The present application relates to the field of environmental protection and sewage treatment technology, and discloses a kind of rural domestic sewage resource utilization processing method and system, the method is by multi-parameter sensing module real-time collection sewage quality, farmland moisture regime and crop image, in combination with convolution neural network identifies crop growth stage, and based on multilayer perception mechanism constructs water quality-fertilizer-irrigation dynamic matching model, intelligent control filter material replacement cycle, wetland plant harvesting opportunity and effluent distribution ratio.The system is composed of grid sedimentation tank, biological filter, artificial wetland three-stage treatment unit, equipped with embedded simple operation and maintenance interface and fault self-diagnosis function, support dialect voice prompt and remote data upload.The present application realizes the accurate conversion of sewage nitrogen and phosphorus resources into farmland fertilizer by the above technical scheme, reduces the operation and maintenance threshold, and improves the system stability and resource efficiency.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of environmental protection and sewage treatment, and particularly relates to a rural domestic sewage resource utilization treatment method and system. BACKGROUND

[0002] Rural domestic sewage treatment has become a key link for improving the living environment and realizing ecological circular agriculture. The current mainstream treatment mode mostly follows the urban centralized sewage treatment idea, relies on fixed process parameters and standardized discharge standards, and the design core is to regard sewage as a pollution source to be eliminated, rather than a resource to be utilized. However, the sewage generation in rural areas has significant temporal and spatial unevenness: the water quantity fluctuates dramatically due to the influence of seasons and work-rest, and the water quality is complex due to mixed sources such as washing, kitchen waste and breeding; at the same time, the demand for nitrogen and phosphorus nutrients of the surrounding farmland changes dynamically with crop types and growth stages.

[0003] The traditional system often leads to overload failure of the treatment facility in the high load period, idle waste in the low load period, and serious mismatch between the effluent nitrogen and phosphorus concentrations and the actual demand of the farmland, because it lacks real-time sensing capability for the influent water quality, farmland soil moisture and crop fertilizer demand rules. Or, due to excessively high concentration, it causes soil salinization and groundwater pollution, or due to insufficient nutrients, it forces farmers to use additional chemical fertilizers, which deviates from the original intention of resource utilization.

[0004] Among them, constructed wetlands and ecological filters are low-cost treatment technologies widely used in rural areas, and their efficiency highly depends on the synergistic effect of plant absorption, microbial degradation and substrate adsorption. The existing scheme generally adopts a fixed cycle to replace the filter material, uniformly harvest wetland plants at a certain time, and uniformly distribute irrigation water, without establishing a dynamic response mechanism of sewage quality-crop fertilizer demand-irrigation strategy. When the ammonia nitrogen or total phosphorus concentration of the influent suddenly increases, the system cannot trigger filter regeneration or adjust the plant harvesting node in advance to enhance the denitrification and phosphorus removal capacity.

[0005] During the crop jointing and booting stages, the irrigation priority of the eutrophic effluent cannot be actively improved, resulting in the loss of valuable nitrogen and phosphorus resources or environmental risks. The fundamental contradiction between this static operation logic and the dynamic demand of agricultural production makes most projects fall into the dilemma of construction compliance, low efficiency and resource waste.

[0006] Although the existing technology attempts to introduce an automatic control module, there are two major defects: first, the sensor deployment is isolated, only monitoring a single link (such as influent COD or effluent pH), without building a full-chain data closed loop covering the sewage end-treatment end-farmland end; second, the regulation algorithm is detached from the agricultural scene, simplifying the complex crop physiological demand into a fixed threshold comparison, and cannot intelligently match the nutrient supply rhythm according to the key growth stages such as rice tillering period and fruit tree fruit swelling period.

[0007] More fatally, the system interaction interface is mostly industrial-grade configuration software, which requires the operator to have professional environmental engineering knowledge. However, the actual operation and maintenance subject in rural areas is mostly ordinary farmers with limited education, who often have no way to deal with complex operations such as frequent alarms and parameter debugging, ultimately leading to a high idle rate of more than 40% of the equipment. Therefore, an intelligent rural sewage resource utilization processing system deeply integrating multi-source sensing, agronomic models and simple interaction is urgently needed to fundamentally solve the three technical bottlenecks of poor adaptability to water quality fluctuations, low resource matching accuracy and weak operation and maintenance sustainability. SUMMARY

[0008] The present application provides a rural domestic sewage resource utilization processing method and system, which realizes dynamic response to rural domestic sewage quality and quantity fluctuations, precise matching of crop fertilizer and water demand rules, and adaptive support for non-professional farmers' operation and maintenance by constructing a closed-loop control architecture integrating multi-parameter sensing module, intelligent control algorithm and simple operation and maintenance interface, thereby stably maintaining sewage treatment efficiency, maximizing nitrogen and phosphorus resource utilization efficiency, and ensuring long-term sustainable operation of the system without external professional intervention.

[0009] The multi-parameter sensing module is arranged at the inlet and outlet of the sewage treatment system and the target irrigated farmland area, and includes a water quality sensor array, a flow meter, a soil moisture sensor and a crop growth image acquisition unit. The water quality sensor array is used to collect real-time values of chemical oxygen demand, ammonia nitrogen concentration, total phosphorus concentration and pH value; the flow meter is used to measure the sewage throughput per unit time; the soil moisture sensor is buried in the plough layer of the farmland for continuous monitoring of soil volume moisture content; the crop growth image acquisition unit is composed of a visible light camera installed on a field fixed support, with a shooting frequency of once a day and a fixed shooting time of 10 am, and the collected images are used for subsequent crop growth stage identification.

[0010] The crop growth stage identification is based on a pre-trained convolutional neural network model, which inputs a sequence of seven consecutive farmland images and outputs a growth stage label of the current crop. The label includes six categories: seedling stage, tillering stage, jointing stage, heading stage, filling stage and mature stage. The model has completed transfer learning training for five main crops: rice, leafy vegetables, solanaceous vegetables, citrus fruit trees and stone fruit trees before deployment.

[0011] The intelligent regulation algorithm constructs a dynamic matching model among sewage water quality, crop fertilizer requirement and irrigation water quantity, the model takes sewage treatment system effluent water quality parameters, target farmland current crop type and its growth stage, real-time soil moisture content as input variables, takes pretreatment filter material replacement period, constructed wetland plant harvesting trigger condition, effluent irrigation distribution ratio as output control quantity.

[0012] The adjustment mechanism of the pretreatment filter material replacement period is as follows: when the chemical oxygen demand at the water inlet exceeds 300 mg / L for three consecutive days, or the ammonia nitrogen concentration exceeds 25 mg / L for three consecutive days, the system automatically shortens the current filter material replacement period to half of the original period; when the above parameters are below the threshold for seven consecutive days, the system restores the original replacement period. The filter material is a mixed filler of porous ceramic and zeolite with a mass ratio of 7:3, and the filling height is 1.2 m. The replacement operation is manually performed by farmers, and the system prompts the replacement time through the controller interface.

[0013] The adjustment mechanism of the constructed wetland plant harvesting time is as follows: when the target crop enters the high fertilizer demand peak period, and the total phosphorus concentration of the effluent of the sewage treatment system is higher than 80% of the upper limit of the recommended fertilizer concentration of the current stage of the crop, the system triggers the harvesting instruction, and the instruction content includes the harvesting area number, the recommended harvesting area ratio and the recommended harvesting tool type. The constructed wetland plant is a mixed community of reed and cattail, and the harvesting area ratio is dynamically calculated according to the crop fertilizer requirement gap, and the calculation formula is: the harvesting area ratio is equal to the crop fertilizer requirement gap divided by the total nitrogen and phosphorus storage in unit area of wetland plant, and the calculation result is rounded to one decimal place.

[0014] The adjustment mechanism of the effluent distribution ratio is as follows: the system calculates the comprehensive water and fertilizer requirement index of each field according to the crop type, current growth stage and soil moisture content of each target farmland, the index is obtained by weighted summation of crop stage water requirement coefficient, stage fertilizer requirement coefficient and soil water deficiency coefficient, and the weight coefficients are 0.4, 0.4 and 0.2 respectively. The system distributes all the treated effluent of the day according to the proportion of the comprehensive water and fertilizer requirement index of each field to the total, and the distribution instruction is executed by the electric valve group, and the valve opening degree is adjusted by the proportional-integral-derivative controller, and the control period is 10 minutes.

[0015] The simple operation and maintenance interface of the application is composed of an embedded graphic controller, the screen size of the controller is seven inches, the resolution is 800*480 pixels, the interface layout adopts a three-row four-column icon matrix, the icons in the first row correspond to crop type selection, including five types of rice, leaf vegetables, eggplants, citrus and stone fruits; the icons in the second row correspond to system state viewing, including water quality trend chart, soil moisture chart, filter material life bar and plant harvesting countdown; the icons in the third row correspond to fault handling guide, including sensor calibration, valve reset, power restart and contact after-sales. After the farmer selects the crop type through the physical key, the system automatically loads the control parameter set of the corresponding crop, and the parameter set includes the water and fertilizer threshold table of each growth stage of the crop, the recommended harvesting time window table and the filter replacement reference period table.

[0016] The fault alarm function of the application includes sensor failure detection and water quality exceeding standard early warning. The sensor failure detection is realized through consistency verification of adjacent sensor data, when the deviation of the reading of a certain sensor from the average value of other sensors in the same area exceeds 30% and the duration exceeds one hour, the system determines that the sensor fails, the interface displays that sensor No. X is abnormal, and suggests cleaning the probe or replacing the battery. The water quality exceeding standard early warning is realized by comparing the real-time monitoring data of the water outlet with the current stage allowed maximum fertilization concentration of the farmland, when any water quality parameter exceeds the threshold value, the system immediately closes the corresponding water outlet valve, the interface displays that the water nitrogen and phosphorus exceeds the standard, irrigation is suspended, the wetland plant density is checked, and the standby water storage tank is started to temporarily store the exceeding standard water.

[0017] The sewage treatment system of the application includes a three-stage treatment unit of a grid sedimentation tank, a biological filter tank and a constructed wetland. The grid sedimentation tank is arranged downstream of the water inlet, the grid spacing is 10 mm, and the hydraulic retention time of the sedimentation zone is four hours; the biological filter tank is filled with mixed filter material of porous ceramic and zeolite, the filter material particle size range is 3 mm to 8 mm, the filter layer thickness is 1.5 m, the bottom is provided with a perforated aeration pipe, and the aeration amount is dynamically adjusted according to the ammonia nitrogen load of the inlet water; the constructed wetland is a horizontal subsurface flow type structure, the length-width ratio of the wetland bed body is 3:1, the filling substrate is a mixture of gravel and coarse sand, the surface is planted with reed and cattail, and the hydraulic load of the wetland is 0.1 m 3 / m 2 ·d.

[0018] The system of the application performs data backup and model fine-tuning operations at two o'clock every morning, the backup content includes all sensor raw data of the day, control instruction execution record and farmer operation log; the model fine-tuning adopts an online learning mechanism, takes the deviation between the actual crop growth response and the predicted fertilization effect of the day as a feedback signal, adjusts the weight parameters of the crop fertilizer requirement calculation module in the dynamic matching model, the learning rate is set to 0.001, the single iteration step number is 50 steps, and the fine-tuning process does not affect the normal operation of the system.

[0019] The farmland irrigation distribution execution mechanism comprises an electric butterfly valve group, a pressure-compensated drip irrigation belt and a flow metering module. The electric butterfly valve is installed at the starting end of each branch pipe, the driving voltage is 24V DC, and the response time is less than 5s. The pressure-compensated drip irrigation belt is laid in the crop row, the drip head spacing is 0.3m, and the rated working pressure is 0.1MPa. The flow metering module adopts an electromagnetic flow meter, the measurement accuracy is ±1%, the data sampling frequency is once per minute, and the metering result is used for checking the accuracy of the distribution proportion execution.

[0020] The crop growth image acquisition unit is equipped with an automatic cleaning mechanism which comprises a miniature air pump, an air jet nozzle and a wiper blade. Before daily image acquisition, the cleaning program is executed as follows: firstly, the air pump is started to spray compressed air to the lens surface for 3s to remove floating dust; then, the wiper blade reciprocates along the horizontal direction twice to remove water stains and insect traces; after the completion of the cleaning program, the image acquisition is started after a delay of 10s to ensure that the imaging quality meets the identification requirements.

[0021] The system is provided with a remote data uploading interface, the interface protocol adopts general packet radio service technology, the operation data of the previous day is automatically uploaded to the cloud server after compression and encryption at 8am every day, and the uploading content includes the average of water quality monitoring data, filter replacement records, plant harvesting execution, fault alarm times and irrigation water volume distribution statistics. The cloud server is deployed with a data analysis module for generating a monthly operation report, and the report content includes a system processing efficiency trend chart, a resource utilization total statistics table and a common fault type distribution chart. The report is pushed to the designated farmer's mobile phone through a short message.

[0022] The biological filter aeration amount adjusting mechanism is as follows: the initial value of the aeration amount is determined according to the ammonia nitrogen concentration of the inlet water, the interval of the table lookup is 5mg / L per grade; the system detects the dissolved oxygen concentration of the filter outlet water every 30 minutes, if the detection value is lower than 2mg / L for three times in succession, the aeration amount is increased by 20% based on the current aeration amount; if the detection value is higher than 5mg / L for three times in succession, the aeration amount is reduced by 15%; the adjustment range is not more than 25% at a time, and the cumulative adjustment times per day are not more than eight, so as to prevent the biological membrane from falling off due to violent disturbance.

[0023] The artificial wetland plant harvesting execution is completed manually by the farmers according to the system prompt, after triggering the harvesting instruction, the system displays a planar schematic diagram of the harvesting area on the controller interface, and the diagram is marked with a recommended harvesting boundary line, a recommended advancing direction arrow and a safety icon. A backpack type brush cutter is recommended for the harvesting tool, the blade rotating speed is set to 3,000 revolutions per minute, the harvesting height is 20cm from the surface of the wetland, and the harvested materials are concentrated and placed in a designated collection point, and the system records the weight of the harvested materials for subsequent nitrogen and phosphorus release amount accounting.

[0024] The simple operation and maintenance interface has a voice broadcast function, when the system detects abnormal key parameters or needs farmers to perform operations, in addition to screen display, pre-recorded voice prompts are played synchronously, the voice content includes operation step number, tool preparation list, estimated time consumption, safety protection requirements. The voice speed is set to 120 words per minute, the volume can be manually adjusted, the broadcast language is the local dialect version, the system has five main dialect voice libraries built-in, the farmers select the matching dialect type through the menu when using for the first time.

[0025] The system sets a backup power module, which is composed of a lead-acid storage battery and a solar charging controller. The rated capacity of the storage battery is 200A, and the nominal voltage is 12V. The maximum input power of the solar charging controller is 300W, and it has overcharge protection, overdischarge protection and short circuit protection functions. When the city power is interrupted, the system automatically switches to backup power supply, prioritizing power supply for sensor array, controller and electric valve, and the aeration equipment and image acquisition unit enter low-power standby mode. The system wakes up every two hours to detect the city power recovery state during standby.

[0026] The farmland soil moisture content sensor adopts frequency domain reflection principle, the probe length is 30cm, the measurement range is 5% to 50%, and the measurement accuracy is ±2%. The sensor automatically executes a self-calibration program every six hours, the calibration process is completed by the built-in reference capacitor, data uploading is suspended during calibration, and calibration coefficients are generated after calibration and applied to subsequent measurement value correction. When calibration fails, the system records an error code and displays that the soil sensor needs to be manually reset on the interface.

[0027] The crop fertilizer requirement calculation module in the dynamic matching model adopts a segmented linear interpolation method, taking the crop growth stage as the horizontal coordinate and the daily nitrogen requirement per unit area and the daily phosphorus requirement as the vertical coordinate to construct a two-dimensional lookup table. The lookup table data comes from the crop fertilization guidelines published by agricultural research institutions and is recorded into the system after field verification. When the crop is between two standard growth stages, the system calculates the current fertilizer requirement by time proportion interpolation, and the interpolation formula is: current fertilizer requirement equals the fertilizer requirement of the previous stage plus the fertilizer requirement increment of the stage interval multiplied by the time proportion within the stage.

[0028] The system performs an initialization configuration process when it is installed for the first time, which includes: the farmer selects the climate type of the area through the controller, the system loads the crop growth cycle reference parameters of the corresponding area; the farmer inputs the area, soil texture type and perennial crop type of each irrigated farmland; the system generates an initial set of control parameters based on the input information and prompts the farmer to install soil sensors and image acquisition units in each farmland, and performs sensor position registration and number binding operation after installation is completed, and the binding information is stored in the non-volatile memory.

[0029] Compared with the prior art, the application has the advantages and positive effects that:

[0030] 1、 The application realizes the synchronous real-time monitoring of sewage quality, farmland moisture, and crop growth through a multi-parameter sensing module, solving the problem of lagging regulation and control caused by the lack of dynamic sensing capability in traditional rural sewage treatment systems;

[0031] 2、 By constructing a dynamic matching model of sewage quality and crop fertilizer requirement, the accurate conversion of nitrogen and phosphorus resources from pollutants to fertilizers is realized, avoiding resource waste and secondary pollution;

[0032] 3、 Through a simple operation and maintenance interface and a fault self-diagnosis function, the system operation complexity is reduced to a level that ordinary farmers can independently complete, solving the problem of system idling caused by the lack of professional operation and maintenance personnel in rural areas;

[0033] 4、 Through adaptive adjustment of filter replacement period and optimization of wetland plant harvesting time, the service life of the core treatment unit is extended, and the long-term operation cost is reduced;

[0034] 5、 Through precise water distribution of electric valve group and drip irrigation belt, the irrigation water is distributed on demand, improving the water resource utilization efficiency;

[0035] 6、 The overall system can stably operate for more than five years without external professional support, the annual average fertilizer replacement amount reaches 80 kg per mu, and the sewage treatment compliance rate remains above 98%. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 is a schematic diagram of the overall technical scheme architecture of a rural domestic sewage resource utilization treatment method and system proposed by the application;

[0037] Figure 2 is a schematic diagram of the core principle framework of the sewage quality-crop fertilizer-irrigation water dynamic matching model in the application;

[0038] Figure 3 is a logic flow framework diagram of the multi-parameter sensing module and intelligent control algorithm cooperative closed-loop control in the application;

[0039] Figure 4 is a schematic diagram of the multi-level interaction relationship and data flow of the terminal farmer operation and maintenance interface and cloud data service in the application;

[0040] Figure 5 is a resource linkage control framework diagram of the sewage treatment unit and farmland irrigation execution mechanism in the application; DETAILED DESCRIPTION

[0041] Please refer to Figures 1-5The application provides a rural domestic sewage resource utilization treatment method and system, which realizes dynamic response to rural domestic sewage quality and quantity fluctuation, accurate matching of farmland crop fertilizer and water demand law, and adaptive support for non-professional farmer operation and maintenance by constructing a closed-loop control architecture of a three-in-one of a multi-parameter sensing module, an intelligent regulation and control algorithm and a simple operation and maintenance interface, so that the sewage treatment efficiency is stably maintained without external professional intervention, the nitrogen and phosphorus resource utilization efficiency is maximized, and the long-term sustainable operation of the system is ensured.

[0042] The method comprises the following steps:

[0043] S1, through the multi-parameter sensing module arranged at the water inlet and water outlet of the sewage treatment system and the target irrigated farmland area, real-time collection of sewage chemical oxygen demand, ammonia nitrogen concentration, total phosphorus concentration, pH, flow, soil volume water content and crop growth stage image data is realized;

[0044] S2, based on a pre-trained convolutional neural network model, a seven-day farmland image sequence is recognized, and a current crop growth stage label is output;

[0045] S3, sewage quality parameters, crop type, growth stage and soil water content are input into a dynamic matching model, and the pretreatment filter replacement period, the artificial wetland plant harvesting trigger condition and the water irrigation distribution ratio are calculated;

[0046] S4, according to the calculation results, the execution mechanism is driven to complete the filter replacement prompt, the plant harvesting instruction issuing and the electric valve opening degree adjustment;

[0047] S5, through an embedded graphic controller, three types of operation interfaces of crop selection, state viewing and fault handling are provided to farmers, and voice broadcast and screen warning are triggered when an abnormality occurs;

[0048] S6, data backup and model fine-tuning are performed every day at 2 o'clock in the morning, and running data is synchronously uploaded to a cloud server to generate a monthly report.

[0049] In step S1, the multi-parameter sensing module is composed of a water quality sensor array, a flowmeter, a soil moisture sensor and a crop growth image acquisition unit. The water quality sensor array is arranged at the water inlet and water outlet positions, and is used to collect sewage chemical oxygen demand, ammonia nitrogen concentration, total phosphorus concentration and pH values, with a sampling frequency of once every 10 minutes. After analog-digital conversion, the data is stored in a local cache area.

[0050] The flowmeter is installed on the main water pipeline and measures the sewage flow per unit time by electromagnetic induction principle, with a measurement accuracy of ±1%. The data is uploaded to the central processing unit after being synchronously marked with a time stamp.

[0051] The soil moisture sensor is buried 20 cm below the plough layer of each target farmland, measures the soil volume water content using the frequency domain reflection principle, the probe length is 30 cm, the measurement range is 5% to 50%, an automatic self-calibration program is executed every six hours, and an error code is recorded and an interface prompt is triggered when calibration fails.

[0052] The crop growth image acquisition unit is composed of a visible light camera on a fixed support, the shooting frequency is once a day, the shooting period is fixed at 10 o'clock in the morning, the lens is equipped with an automatic cleaning mechanism, the cleaning program includes three seconds of compressed air injection and two times of wiper reciprocating motion, the image acquisition is started after a delay of 10 s after cleaning is completed, and the imaging clarity meets the subsequent identification requirements.

[0053] In step S2, the crop growth stage recognition is performed based on a pre-trained convolutional neural network model. The model input is a sequence of seven consecutive days of farmland images, each image has a resolution of no less than 1280x720 pixels, and is input into the network after grayscale normalization and size cropping. The network structure includes five convolutional layers, three max pooling layers, two fully connected layers, and the final output layer uses a soft max function to generate a six-class growth stage probability distribution, and the highest probability is selected as the current stage label. The six labels are seedling stage, tillering stage, jointing stage, heading stage, filling stage, and mature stage.

[0054] The model has completed transfer learning training for five main crops of rice, leafy vegetables, solanaceous vegetables, citrus fruit trees, and stone fruit trees before deployment, and the training samples come from a historical image library labeled by agricultural research institutions. The recognition results are stored in the non-volatile memory after being bound with the current date and crop type, serving as input parameters for the subsequent dynamic matching model.

[0055] In step S3, the dynamic matching model takes the effluent water quality parameters of the sewage treatment system, the current crop type and its growth stage of the target farmland, and the real-time soil moisture content as input variables, and takes the pretreatment filter replacement period, the artificial wetland plant harvesting trigger condition, and the effluent irrigation distribution ratio as output control variables. The model core is a three-layer multilayer perceptron structure, the number of hidden layer neurons is 128, 64, and 32 respectively, the activation function is the rectified linear unit function, and the loss function is the mean square error function.

[0056] The training data is derived from a set of correlation samples between water quality parameters, crop growth stages, and final fertilization effects in historical operation records, with a total of more than 50,000 samples covering different seasons, climates, and soil conditions. The input layer of the model receives a standardized parameter vector, including normalized values of chemical oxygen demand, ammonia nitrogen concentration, total phosphorus concentration, pH value, flow rate, soil moisture content, crop type code, and growth stage code. The output layer generates three control instruction vectors corresponding to the filter replacement cycle scaling factor, the plant harvesting area ratio, and the irrigation water distribution weight for each field.

[0057] The adjustment mechanism for the pre-treatment filter replacement cycle is as follows:

[0058] The system continuously monitors the chemical oxygen demand and ammonia nitrogen concentration at the inlet. When the chemical oxygen demand exceeds 300 mg / L for three consecutive days or the ammonia nitrogen concentration exceeds 25 mg / L for three consecutive days, the central processing unit shortens the current filter replacement cycle to half of the original cycle, flashes a red warning symbol at the first row and third column of the controller interface, and plays a voice prompt: "Filter load is too high, please replace within seven days."

[0059] When the above parameters are below the threshold for seven consecutive days, the system restores the original replacement cycle, and the interface warning symbol turns green and constant. The filter is a mixed filler of porous ceramic and zeolite with a mass ratio of 7:3, and the filling height is 1.2 m. The replacement operation is performed manually by the farmer, and the system records the replacement time and operator number for subsequent life statistics and responsibility tracing.

[0060] The adjustment mechanism for the artificial wetland plant harvesting time is as follows: When the target crop enters the high fertilizer demand peak period, and the total phosphorus concentration of the wastewater treatment system effluent is higher than 80% of the upper limit of the recommended fertilization concentration for the current stage of the crop, the system triggers the harvesting instruction. The instruction content includes the harvesting area number, the recommended harvesting area ratio, and the recommended harvesting tool type.

[0061] The harvesting area ratio is dynamically calculated based on the crop fertilizer demand gap, and the calculation formula is: the harvesting area ratio equals the crop fertilizer demand gap divided by the total nitrogen and phosphorus storage in the wetland plants per unit area. The crop fertilizer demand gap is output by the crop fertilizer demand calculation module in the dynamic matching model, which uses the piecewise linear interpolation method with the crop growth stage as the horizontal coordinate and the daily nitrogen and phosphorus demand per unit area as the vertical coordinate to construct a two-dimensional lookup table. The lookup table data is derived from the crop fertilization guidelines published by agricultural research institutions and is recorded in the system after field verification.

[0062] When the crop is between two standard growth stages, the system calculates the current fertilizer requirement by time proportion interpolation, the interpolation formula is: current fertilizer requirement equals the previous stage fertilizer requirement plus the fertilizer requirement increment of the stage interval multiplied by the proportion of time within the stage. The total nitrogen and phosphorus storage in wetland plants per unit area is calculated by historical harvest records, the initial value is set to 0.8 kg of nitrogen and 0.3 kg of phosphorus per square meter, and the subsequent dynamic update is based on the actual harvested weight and laboratory test results. The calculation result is rounded to one decimal place, if the result is less than 5%, no harvesting is triggered, if it is greater than 80%, it is executed in batches, and the single harvesting does not exceed 40%, and the interval is greater than 15 days.

[0063] The adjustment mechanism of water distribution ratio is as follows: the system calculates the comprehensive water and fertilizer requirement index of each field according to the crop type, current growth stage and soil water content of each target farmland. The index is obtained by weighted summation of crop stage water requirement coefficient, stage fertilizer requirement coefficient and soil water deficit coefficient, and the weight coefficients are 0.4, 0.4 and 0.2 respectively.

[0064] The crop stage water requirement coefficient and stage fertilizer requirement coefficient are derived from the built-in parameter table of the dynamic matching model, and the soil water deficit coefficient calculation formula is: soil water deficit coefficient equals the lower limit of suitable water content of crops minus the current measured water content divided by the lower limit of suitable water content. The system distributes all the treated water on the same day according to the proportion of the comprehensive water and fertilizer requirement index of each field to the total, and the distribution instruction is executed through the electric valve group.

[0065] The electric butterfly valve is installed at the starting end of each branch pipe, the driving voltage is 24V DC, the response time is less than five seconds, the opening degree is adjusted by proportional integral derivative controller, and the control period is 10 minutes. The flow measurement module uses electromagnetic flowmeter, the measurement accuracy is ±1%, the data sampling frequency is once per minute, and the measurement result is used to verify the execution accuracy of the distribution ratio, if the deviation between the measured flow and the instruction flow exceeds 10% and the duration exceeds 30 minutes, the system records the deviation event and triggers the valve reset program.

[0066] In step S4, the actuator includes a filter replacement prompt module, a plant harvesting instruction issuing module and an electric valve adjustment module. The filter replacement prompt module generates a flashing icon and voice broadcast on the controller interface, and sends a short message to the farmer's mobile phone, the short message content includes filter type, replacement step number, required tool list and estimated time. The plant harvesting instruction issuing module displays a planar schematic diagram of the harvesting area on the controller interface when the trigger condition is met, and the diagram marks the recommended harvesting boundary line, recommended direction arrow and safety icon.

[0067] The harvesting tool is recommended to use a backpack brush cutter, the blade rotation speed is set to 3,000 rpm, the harvesting height is 20 cm above the wetland surface, and the harvested materials are concentrated and stacked at the designated collection point, and the system records the weight of the harvested materials for subsequent nitrogen and phosphorus release calculation. The electric valve adjustment module receives the pulse width modulation signal output by the proportional-integral-derivative controller, drives the motor to rotate to the specified angle, and the valve position sensor feedbacks the opening value in real time, forming a closed loop control. If the valve does not reach the target opening within five seconds, the system determines that the mechanical jam has occurred, immediately closes the corresponding branch and triggers the valve reset fault code, the interface third row second column icon is highlighted, and the voice broadcast is: valve No. 3 is jammed, please manually reset and restart the system.

[0068] In step S5, the simple operation and maintenance interface is composed of an embedded graphic controller, the controller screen size is 7 inches, the resolution is 800x480 pixels, and the interface layout adopts a three-row four-column icon matrix.

[0069] The first row of icons corresponds to crop type selection, including rice, leaf vegetables, eggplants, citrus, and stone fruits. After the farmer selects through the physical button, the system automatically loads the corresponding crop control parameter set, which includes the water and fertilizer threshold table for each growth stage of the crop, the recommended harvesting time window table, and the filter replacement reference period table.

[0070] The second row of icons corresponds to system status viewing, including water quality trend chart, soil moisture chart, filter life bar, and plant harvesting countdown. All charts are dynamically updated in line or column form, with data refresh frequency of every 30 seconds.

[0071] The third row of icons corresponds to fault handling instructions, including sensor calibration, valve reset, power restart, and contact after-sales. Each icon is associated with a pre-recorded operation video and text steps, which are automatically played after the farmer clicks.

[0072] The voice broadcast function is started simultaneously when the system detects abnormal key parameters or when the farmer needs to perform an operation. The voice content includes operation step number, tool preparation list, estimated time consumption, and safety protection requirements. The speech speed is set to 120 words per minute, the volume can be manually adjusted, and the broadcast language is the local dialect version. The system has five main dialect voice libraries, and the farmer selects the matching dialect type through the menu for the first time.

[0073] In step S6, the system performs data backup and model fine-tuning operations every day at 2 a.m. The backup content includes all sensor raw data, control instruction execution records, and farmer operation logs. The data is compressed and encrypted and stored on the local solid state disk, with a retention period of 365 days.

[0074] The model fine-tuning adopts an online learning mechanism, taking the deviation between the actual crop growth response and the predicted fertilization effect on the same day as the feedback signal to adjust the weight parameters of the crop fertilizer requirement calculation module in the dynamic matching model. The learning rate is set to 0.001, and the number of single iteration steps is 50 steps. The fine-tuning process does not affect the normal operation of the system. The remote data upload interface uses General Packet Radio Service technology. At 8:00 am every day, the previous day's operation data is compressed and encrypted and then uploaded to the cloud server. The uploaded content includes water quality monitoring data mean, filter replacement record, plant harvesting execution, fault alarm number, and irrigation water volume distribution statistics.

[0075] The cloud server deploys a data analysis module to generate a monthly operation report. The report content includes system processing efficiency trend chart, resource utilization total statistics table, and common fault type distribution chart. The report is pushed to the designated farmer's mobile phone by SMS at 10:00 am on the 5th of each month.

[0076] The system includes a grid sedimentation tank, a biological filter tank, and a three-stage treatment unit of constructed wetland. The grid sedimentation tank is arranged downstream of the water inlet, the grid spacing is 10 mm, the hydraulic retention time in the sedimentation zone is 4 hours, a sludge hopper is arranged at the bottom of the tank, and the sludge is automatically discharged once a day. The sludge discharge amount is dynamically adjusted according to the suspended solids concentration of the influent. The biological filter tank is filled with mixed filter material of porous ceramic and zeolite, the filter material particle size range is 3-8 mm, the filter layer thickness is 1.5 m, and a perforated aeration pipe is arranged at the bottom. The aeration amount is dynamically adjusted according to the ammonia nitrogen load of the influent.

[0077] The initial value of the aeration amount is determined according to the ammonia nitrogen concentration of the influent by table lookup, and the table lookup interval is 5 mg / L per grade. The system detects the dissolved oxygen concentration of the filter tank effluent every 30 minutes. If the detected value is lower than 2 mg / L for three consecutive times, the aeration amount is increased by 20% based on the current aeration amount. If the detected value is higher than 5 mg / L for three consecutive times, the aeration amount is reduced by 15%. The adjustment range of each time does not exceed 25%, and the cumulative adjustment times per day do not exceed eight times to prevent the biological membrane from falling off due to severe disturbance.

[0078] The constructed wetland is a horizontal subsurface flow type structure, the length-width ratio of the wetland bed is 3:1, the filling substrate is a mixture of gravel and coarse sand, the surface is planted with reed and cattail, and the hydraulic load of the wetland is 0.1 m 3 / m 2 ·d. The wetland effluent is irradiated by ultraviolet light in a disinfection tank and then enters the clear water tank. A liquid level sensor is arranged in the clear water tank. When the liquid level is lower than 0.5 m, a low water level alarm is triggered. When the liquid level is higher than 2.5 m, an overflow valve is started to discharge into the standby water storage tank.

[0079] The system is provided with a backup power supply module, which is composed of a lead-acid storage battery and a solar charging controller. The rated capacity of the storage battery is 200A, and the nominal voltage is 12V. The maximum input power of the solar charging controller is 300W, and it has overcharge protection, overdischarge protection, and short circuit protection functions. When the city power is interrupted, the system automatically switches to backup power supply, prioritizing the power supply of the sensor array, controller, and electric valve, and the aeration equipment and image acquisition unit enter low-power standby mode. During standby, the system wakes up every two hours to detect the city power recovery state. If the backup power supply is less than 20%, the system shuts down unnecessary modules, only maintains basic monitoring and alarm functions, and sends a text message to the farmer's mobile phone that the power is about to run out, please check the city power or replace the battery.

[0080] The system performs an initialization configuration process when it is first installed. The process includes:

[0081] The farmer selects the climate type of the region through the controller, and the system loads the corresponding regional crop growth cycle reference parameters. The farmer inputs the area of each irrigated farmland, the soil texture type, and the perennial crop type. The system generates an initial set of control parameters based on the input information and prompts the farmer to install soil sensors and image acquisition units in each farmland. After installation, the system performs sensor position registration and number binding operations, and the binding information is stored in the non-volatile memory.

[0082] The registration process includes sensor self-check, signal strength test, and data consistency verification. If any step fails, the system prompts to reinstall. After the system completes the initialization, it automatically enters the trial operation mode and collects baseline data for 72 hours without executing any control instructions. After 72 hours, the system generates an initial operation report, which is confirmed by the farmer before being officially enabled.

[0083] The fault alarm function includes sensor failure detection and water quality exceeding standard early warning. Sensor failure detection is achieved through consistency verification of adjacent sensor data. When the deviation of a sensor reading from the average of other sensors in the same area exceeds 30% and the duration exceeds one hour, the system determines that the sensor is failed, the interface displays sensor number X abnormal, and suggests cleaning the probe or replacing the battery. The system also records the failure time and location for subsequent maintenance and scheduling.

[0084] Water quality exceeding standard early warning is achieved by comparing real-time monitoring data at the water outlet with the maximum allowable fertilization concentration at the current stage of the farmland. When any water quality parameter exceeds the threshold, the system immediately closes the corresponding water outlet valve, the interface displays water nitrogen and phosphorus exceeding standard, irrigation is suspended, the wetland plant density is checked, and the standby reservoir is started to temporarily store the exceeding standard water. The standby reservoir has a capacity of 50% of the daily treatment capacity, and a stirrer and a reprocessing pump are installed in the reservoir. The exceeding standard water is diluted and filtered again before being reprocessed until the water quality meets the standard.

[0085] The farmland irrigation distribution actuator is composed of an electric butterfly valve set, a pressure-compensated drip irrigation tape, and a flow metering module. The pressure-compensated drip irrigation tape is laid in the crop rows with a dripper spacing of 0.3 m and a rated working pressure of 0.1 MPa. The dripper flow error is not more than ±5%. The system performs a self-check of the drip irrigation tape pressure every morning at 6 o'clock. By adjusting the main pump speed, the pressure of each branch is stabilized at the set value. If the pressure of a certain branch continuously deviates by more than 15%, the system determines that the dripper is clogged or the pipe is broken, closes the branch, and triggers an abnormal drip irrigation alarm. The flow metering module data is used to generate an irrigation log, which includes the actual irrigation volume, theoretical water requirement, deviation rate, and execution time of each field. The log is archived daily and used to generate a monthly report.

[0086] The automatic cleaning mechanism of the crop growth image acquisition unit is composed of a miniature air pump, a jet nozzle, and a wiper blade. The air pump has a working pressure of 0.3 MPa, and the jet nozzle angle is adjustable to ensure that the airflow covers the entire lens surface. The wiper blade is made of silicone and has a weather resistance of not less than five years. The wiper movement is driven by a miniature stepper motor with a travel accuracy of ±0.5 mm. The cleaning program records the cleaning time, air pump working time, and wiper movement frequency. If the image recognition confidence level is still below 90% after three consecutive cleanings, the system determines that the lens is heavily contaminated and triggers an alarm for manual cleaning.

[0087] The crop fertilizer requirement calculation module in the dynamic matching model uses the piecewise linear interpolation method, with the crop growth stage as the horizontal coordinate and the daily nitrogen requirement per unit area and the daily phosphorus requirement per unit area as the vertical coordinate to construct a two-dimensional lookup table. The lookup table data comes from the crop fertilization guidelines published by agricultural research institutions and is recorded into the system after field verification.

[0088] The system sets up a remote data upload interface, which uses the General Packet Radio Service (GPRS) technology. The system automatically uploads the previous day's operation data to the cloud server after compression and encryption at 8 o'clock in the morning. The uploaded content is verified for integrity using the Hash algorithm. If the transmission fails, it will automatically retry three times. If all three attempts fail, it will record the transmission error and wait for the next upload period. The cloud server deploys a data analysis module that uses time series analysis and clustering algorithms to generate a monthly operation report.

[0089] The report content includes a system processing efficiency trend chart, with the horizontal axis representing the date and the vertical axis representing the chemical oxygen demand removal rate, ammonia nitrogen removal rate, and total phosphorus removal rate. The report also includes a resource utilization total amount statistics table, which lists the nitrogen and phosphorus replacement amounts and water saving amounts for each crop type. The report also includes a common fault type distribution chart, which displays the proportion of four types of faults, including sensor failure, valve jamming, power interruption, and image recognition failure, in the form of a pie chart. The report is pushed to the designated farmer's mobile phone via SMS, and the push content includes a report summary and a complete report download link. Farmers can view detailed data through a mobile browser.

[0090] The biological filter aeration amount adjustment mechanism is as follows: the initial value of the aeration amount is determined according to the ammonia nitrogen concentration of the influent, and the interval of the table is 5 mg / L per grade. The system detects the dissolved oxygen concentration of the filter effluent every 30 minutes, and the detection point is located 50 cm above the filter layer. An optical dissolved oxygen sensor is used, and the measurement range is 0 to 10 mg / L, with an accuracy of ± 2%. If the detection value is less than 2 mg / L for three consecutive times, the aeration amount will be increased by 20% based on the current aeration amount. If the detection value is higher than 5 mg / L for three consecutive times, the aeration amount will be reduced by 15%. The adjustment range is not more than 25% at a time, and the cumulative adjustment times per day are not more than eight times, to prevent the biological membrane from falling off due to violent disturbance. The aeration amount adjustment record includes adjustment time, pre-adjustment value, post-adjustment value, and trigger condition, which is used for subsequent operation analysis and parameter optimization.

[0091] The artificial wetland plant harvesting is executed manually by farmers according to system prompts. After triggering the harvesting instruction, the system displays a plan view of the harvesting area on the controller interface, with suggested harvesting boundary lines, recommended direction arrows, and safety icon. The recommended harvesting tool is a backpack-type brush cutter, with a blade speed setting of 3000 rpm, and the harvesting height is 20 cm above the wetland surface to ensure the retention of root systems for regeneration. The harvested plants are stacked at designated collection points, which are equipped with weighing platforms. The system automatically records the weight of the harvested plants, which is used for subsequent nitrogen and phosphorus release calculation. The calculation formula is: nitrogen release amount equals the weight of the harvested plants multiplied by the nitrogen content per unit weight, and phosphorus release amount equals the weight of the harvested plants multiplied by the phosphorus content per unit weight. The nitrogen and phosphorus contents are retrieved from the built-in database according to plant species and growth stage, with initial values of 2.5% and 0.8% respectively, and are dynamically updated based on laboratory test results.

[0092] The simple operation and maintenance interface has a voice broadcast function. When the system detects abnormal key parameters or requires farmers to perform operations, in addition to screen display, pre-recorded voice prompts are also played simultaneously. The voice content includes operation step number, tool preparation list, estimated time consumption, and safety requirements. The voice speed is set to 120 words per minute, the volume can be manually adjusted, and the broadcast language is in local dialect version. The system has five main dialect voice libraries built-in, and farmers can choose the matching dialect type through the menu when using it for the first time. The voice broadcast log records the broadcast time, trigger condition, and broadcast content summary.

[0093] The system is equipped with a backup power supply module, which is composed of a lead-acid battery and a solar charging controller. The rated capacity of the battery is 200A, the nominal voltage is 12V, and it adopts a sealed maintenance-free design with a cycle life of more than 500 times. The maximum input power of the solar charging controller is 300W, with a maximum power point tracking function. When the power is interrupted, the system automatically switches to backup power supply, prioritizing power supply for sensors, controllers, and electric valves, while the aeration equipment and image acquisition unit enter low-power standby mode.

[0094] During standby, the power recovery state is detected every two hours, and if the power is restored, the main power is automatically switched back and the standby module is restarted. The standby power supply power monitoring module samples the voltage value every 10 minutes, triggers a low power alarm when the voltage is lower than 11V, and closes unnecessary modules when the voltage is lower than 10V, only maintaining basic monitoring and alarm functions.

[0095] The farmland soil moisture sensor uses the frequency domain reflection principle, the probe length is 30cm, the measurement range is 5% to 50%, and the measurement accuracy is ±2%. The sensor automatically executes a self-calibration program every six hours, the calibration process is completed by the built-in reference capacitor, the data upload is suspended during calibration, and the calibration coefficient is generated after calibration and applied to subsequent measurement value correction.

[0096] The calibration coefficient is stored in the internal memory of the sensor and automatically loaded each time the power is turned on. When calibration fails, the system records an error code and displays on the interface that the soil sensor needs to be manually reset, and sends a reset instruction SMS to the farmer's mobile phone. The reset operation includes power-off restart, probe cleaning, and position adjustment. After the operation is completed, the system re-executes the self-check and calibration.

[0097] The crop fertilizer requirement calculation module in the dynamic matching model uses the piecewise linear interpolation method, taking the crop growth stage as the horizontal coordinate and the daily nitrogen requirement per unit area and the daily phosphorus requirement per unit area as the vertical coordinate to construct a two-dimensional lookup table. The lookup table data comes from the crop fertilization guidelines published by agricultural research institutions and is recorded into the system after field verification. The calculation result is checked by the boundary, and if it exceeds the physiological limit value, it is automatically truncated to the limit value to prevent abnormal instructions.

[0098] The system performs an initialization configuration process when it is first installed. The process includes:

[0099] The farmer selects the climate type of the area through the controller, and the system loads the corresponding crop growth cycle reference parameters of the area. The parameters include the standard growth stage start and end dates of each crop, the suitable temperature range, and the precipitation requirement. The farmer inputs the area, soil texture type, and perennial crop type of each irrigated farmland, and the system generates an initial set of control parameters based on the input information. The parameter set includes a filter replacement reference period table, such as a 90-day period for sandy soil and a 120-day period for clay soil; a recommended harvesting time window table, such as a 45-day to 50-day window for rice harvesting; and a water and fertilizer requirement threshold table for each growth stage of crops, such as a water requirement coefficient of 0.6 and a fertilizer requirement coefficient of 0.4 for leafy vegetables in the seedling stage. The system prompts the farmer to install soil sensors and image acquisition units in each farmland, and after installation, performs sensor position registration and number binding operations. The binding information is stored in the non-volatile memory.

[0100] The registration process includes sensor self-check, signal strength test, data consistency check, any failure of any link will prompt to re-install. After the system completes initialization, it automatically enters the trial operation mode, collects baseline data for 72 hours, during which no control instructions are executed, and generates an initial operation report after 72 hours, which is confirmed by the farmer before being officially put into use.

[0101] It should be noted that the relational terms herein, such as first and second, and the like, are used solely to distinguish one from another entity or action without necessarily requiring or implying any actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0102] While embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, combinations, and variations of the embodiments can be made by those skilled in the art without departing from the spirit and scope of the present application, which is defined by the appended claims and their equivalents.

Claims

1. A rural domestic sewage resource utilization treatment method, characterized in that, The application relates to a sewage treatment system and a method for dynamically matching the sewage treatment system. Real-time acquisition of sewage chemical oxygen demand, ammonia nitrogen concentration, total phosphorus concentration, pH value and flow data of a sewage treatment system inlet and outlet through a multi-parameter sensing module, and simultaneous acquisition of soil volume water content and crop growth image sequences of a target irrigated farmland area; Based on a pre-trained convolutional neural network model, a seven-day continuous crop growth image sequence is identified to output a current crop growth stage label, which includes six types of seedling stage, tillering stage, jointing stage, heading stage, filling stage and mature stage; The sewage chemical oxygen demand, ammonia nitrogen concentration, total phosphorus concentration, pH value, flow, soil volume water content, crop type and growth stage label are input into a dynamic matching model to calculate a pretreatment filter replacement cycle scaling coefficient, a constructed wetland plant harvesting area ratio and a farmland outlet irrigation distribution weight; According to the pretreatment filter replacement cycle scaling coefficient, a filter replacement prompt instruction is triggered, according to the constructed wetland plant harvesting area ratio, a harvesting area and tool recommendation instruction are triggered, and according to the farmland outlet irrigation distribution weight, an electric valve group is driven to adjust the opening degree of each branch to realize water demand distribution; Through an embedded graphics controller, three types of operation interfaces of crop type selection, system state viewing and fault handling guidance are provided to farmers, and when a parameter is abnormal or operation is triggered, a dialect voice broadcast function is simultaneously started; Sensor data backup and dynamic matching model online fine tuning are performed at two o'clock every morning, and running data is uploaded to a cloud server through a general packet radio service technology at eight o'clock every morning to generate a monthly operation report; Seven-day continuous standardized input images are input into a convolutional neural network model containing five convolutional layers, three maximum pooling layers and two fully connected layers to generate a six-class growth stage probability distribution; The core of the dynamic matching model is a multi-layer perception machine structure, and the number of hidden layers is three.

2. The rural domestic sewage resource utilization treatment method according to claim 1, characterized in that, Based on a pre-trained convolutional neural network model, a seven-day continuous crop growth image sequence is identified to output a current crop growth stage label, which includes: After performing an automatic cleaning program on the farmland image collected at ten o'clock every morning, the image is subjected to gray scale normalization and size cropping to generate a standardized input image; The highest probability is selected as the current crop growth stage label, and the identification result is bound with the crop type and stored in a non-volatile memory.

3. The rural domestic sewage resource utilization treatment method according to claim 2, characterized in that, The sewage chemical oxygen demand, ammonia nitrogen concentration, total phosphorus concentration, pH value, flow, soil volume water content, crop type and growth stage label are input into a dynamic matching model to calculate a pretreatment filter replacement cycle scaling coefficient, a constructed wetland plant harvesting area ratio and a farmland outlet irrigation distribution weight, including: After the input parameters are subjected to normalization, a three-layer multi-layer perception machine structure is input, the number of hidden layer neurons is 128, 64 and 32 respectively, and a rectified linear unit function is used as an activation function; An output layer generates three control instruction vectors, which correspond to a filter replacement cycle scaling coefficient, a plant harvesting area ratio and a farmland irrigation water distribution weight respectively.

4. The rural domestic sewage resource utilization treatment method according to claim 3, characterized in that, According to the pretreatment filter replacement cycle scaling coefficient, a filter replacement prompt instruction is triggered, including: When the chemical oxygen demand of the water inlet exceeds 300 mg / L for three consecutive days or the ammonia nitrogen concentration exceeds 25 mg / L for three consecutive days, the current filter material replacement cycle is shortened to one-half of the original cycle; When the above parameters are below the threshold for seven consecutive days, the original replacement cycle is restored, and a flashing icon and a voice prompt are displayed on the controller interface: "Filter material load is too high, please complete the replacement within seven days." 5. The rural domestic sewage resource utilization treatment method according to claim 4, characterized in that, According to the proportion of the harvested area of the constructed wetland plants, a harvesting area and tool recommendation instruction is triggered, including: When the target crop enters the peak fertilization period and the total phosphorus concentration of the effluent is higher than 80% of the upper limit of the recommended fertilization concentration of the current stage, the harvesting instruction is triggered. The proportion of the harvested area is calculated according to the formula: crop nitrogen and phosphorus deficiency amount divided by the total nitrogen and phosphorus storage in unit area of wetland plants, rounded to one decimal place, less than 5% does not trigger, more than 80% is executed in batches.

6. The rural domestic sewage resource utilization treatment method according to claim 5, characterized in that, According to the water irrigation distribution weight of each field, an electric valve group is driven to adjust the opening degree of each branch to achieve water distribution according to demand, including: Calculate the comprehensive water and fertilizer demand index of each field, which is obtained by weighting the crop stage water demand coefficient, stage fertilizer demand coefficient, and soil water deficiency coefficient, with weights of 0.4, 0.4, and 0.2, respectively. Distribute all the treated effluent of the day according to the proportion of the comprehensive water and fertilizer demand index of each field to the total, and adjust the opening degree of the electric butterfly valve every 10 minutes through a proportional-integral-derivative controller, with a valve response time of less than 5 seconds.

7. The rural domestic sewage resource utilization treatment method according to claim 6, characterized in that, Through the embedded graphic controller, three types of operation interfaces are provided to farmers: crop type selection, system status viewing, and fault handling guidance, including: The interface uses a three-row and four-column icon matrix layout, with the first row corresponding to the selection of five types of crops: rice, leafy vegetables, eggplants, citrus, and stone fruits. The second row corresponds to four types of state charts: water quality trend chart, soil moisture chart, filter material life bar, and plant harvesting countdown. The third row corresponds to four types of fault handling instructions: sensor calibration, valve reset, power restart, and contact after-sales, which automatically play text and video operation steps after being clicked. 8.The rural domestic sewage resource utilization treatment method according to claim 7, characterized in that, Sensor data backup and online fine-tuning of dynamic matching model are performed every morning at 2 am, including: Backup all sensor raw data, control instruction execution records, and farmer operation logs on the same day, and store them in the local solid state disk after compression and encryption. Use the deviation between the actual crop growth response and the predicted fertilization effect as the feedback signal to adjust the weight parameters of the crop fertilizer demand calculation module, with a learning rate of 0.001 and 50 steps per iteration.

9. A rural domestic sewage resource utilization treatment system, characterized in that, Including: A multi-parameter sensing module for real-time acquisition of sewage chemical oxygen demand, ammonia nitrogen concentration, total phosphorus concentration, pH, and flow data at the inlet and outlet of the sewage treatment system, as well as soil volume moisture content and crop growth image sequences in the target irrigated farmland area; A convolutional neural network model including five convolutional layers, three max pooling layers, and two fully connected layers; A crop growth stage recognition module for identifying the current crop growth stage label based on a pre-trained convolutional neural network model on seven consecutive crop growth image sequences; A dynamic matching model with a multi-layer perceptron structure, with three hidden layers. The dynamic matching calculation module is configured to input the sewage quality parameters, the soil moisture content, the crop type and the growth stage label into a dynamic matching model to calculate a pretreatment filter replacement period scaling coefficient, a constructed wetland plant harvesting area proportion and a water distribution weight of each field; The execution control module is configured to trigger a filter replacement prompt, a plant harvesting instruction and an electric valve opening degree adjustment instruction according to the calculation results; The farmer interaction module is configured to provide three types of operation interfaces of crop selection, state viewing and fault handling through an embedded graphic controller, and start a dialect voice broadcast when an abnormality occurs. The data management module is configured to perform data backup and model fine-tuning every morning, and upload operation data to a cloud server to generate a monthly report at 8 o'clock in the morning.

10. The rural domestic sewage resource utilization treatment system according to claim 9, characterized in that, The multi-parameter sensing module is configured to: Deploy a water quality sensor array at the water inlet and the water outlet, with a sampling frequency of once every 10 minutes; Install an electromagnetic flowmeter on the main water pipeline, with a measurement accuracy of ±1%; Bury a frequency domain reflection principle soil moisture sensor at a depth of 20 cm below the plough layer, and perform self-calibration every six hours; Configure a visible light camera with an automatic cleaning mechanism on a field support, collect images at 10 o'clock in the morning every day, and the cleaning program includes three-second compressed air injection and two times of wiper reciprocating motion.

Images