Farmland and river ecological collaborative monitoring and regulation system and method

Abstract

The invention relates to the field of environment monitoring and ecological restoration, in particular to a farmland and river ecological collaborative monitoring and regulation system and method. The system comprises a first-stage barrage, a second-stage barrage, a gabion revetment, front-end acquisition equipment, execution equipment and a server, wherein the first-stage barrage and the second-stage barrage are arranged at intervals; the execution equipment comprises irrigation equipment, a first gate, a second gate and an ecological flocculant putting device; the server comprises a data acquisition module, a data storage module, an analysis processing module and a control execution module, inputs control parameters including a first gate opening degree, a second gate opening degree, an irrigation equipment electric valve opening degree and an ecological flocculant putting amount through an LSTM model, and predicts state variables including an upstream water level, a downstream flow, an irrigation water supply amount, a total phosphorus concentration and a total nitrogen concentration; and a multi-target optimization model including agricultural water demand, ecological flow, flood control safety and water quality purification targets is solved in combination with an optimization NSGA-II algorithm, target control parameters are acquired, and a control instruction is generated to drive execution equipment.

Description

Technical Field

[0001] The present invention relates to the field of environmental monitoring and ecological restoration, and in particular to a system and method for collaborative monitoring and control of farmland and river ecology. Background Art

[0002] In traditional farmland irrigation and river management, weirs and gabion gabion revetments are core engineering facilities for ensuring agricultural production and river safety. Weirs regulate water levels to create a stable irrigation source and control flooding. Gabion gabion revetments stabilize bank slopes and reduce soil erosion through their flexible structures, while also utilizing interstices between the rocks and plant roots to achieve preliminary water purification. Both serve the irrigation needs of farmland and protect river safety. The ecological synergy between farmland and river is essentially a coupling of material cycles and energy flows within the "water-soil-organism" system. Rivers act as water transport vehicles, providing irrigation water for farmland. Nutrients such as nitrogen and phosphorus in runoff from farmland contribute to the natural fertility cycle through river runoff. Microhabitats within river channels, such as shallows and deep pools, provide breeding grounds for natural predators of farmland insects. Farmland vegetation filters surface runoff, reducing the pollution load on the river.

[0003] However, the construction and operation of existing technologies for weirs and gabion stone cage revetments present the following problems: The engineering design only regulates single needs, such as farmland irrigation or river flood control, and lacks a holistic consideration of the ecological coordination between farmland and river channels. For example, existing intelligent irrigation systems intelligently regulate irrigation volume based on farmland soil moisture, but because they fail to monitor river hydrological data and ecological baseflow thresholds, this often leads to excessive water withdrawal, which in turn causes problems such as insufficient ecological flow, pollutant accumulation, and reduced biodiversity. Furthermore, existing water resource regulation methods are mostly based on fixed rules. Due to the lack of real-time monitoring and collaborative analysis of multi-dimensional data on river hydrology, water quality, and farmland soil moisture, it is impossible to achieve a multi-objective dynamic balance between agricultural water demand, ecological water demand, and flood control safety through intelligent monitoring and regulation systems. That is, dynamically adjusting water resource scheduling strategies based on real-time monitoring data to meet crop irrigation water needs while maintaining the river's ecological baseflow and preventing and controlling flood risks. Especially when flood risks and water pollution exceeding standards occur simultaneously during the flood season, only regulating and controlling a single demand according to priority while ignoring other demands may lead to the destruction of farmland and river biological habitats. For example, rapid flood discharge during the flood season may easily ignore the need for pollutant dilution, and when water pollution exceeds standards, there is a lack of strategies to simultaneously optimize irrigation water intake and river purification, resulting in a high risk of ecological imbalance.

[0004] Therefore, in view of the shortcomings of the above-mentioned existing technologies, there is an urgent need to build a farmland and river ecological collaborative monitoring and control system that can integrate hydrological regulation, ecological purification, data monitoring and intelligent decision-making, so as to achieve the material cycle optimization and energy flow coupling of farmland and river ecology, monitor and control the ecological environment of farmland and river, and promote the ecological and intelligent management of farmland water conservancy and river. Summary of the Invention

[0005] In response to the shortcomings of existing technologies, the present invention aims to provide a system and method for collaborative monitoring and control of farmland and river ecology, so as to solve the problem that existing dams and gabion stone cage revetments may cause water resource waste and insufficient water purification during operation due to regulation based only on a single demand, thereby leading to the destruction of farmland and river biological habitats and reduction of biodiversity.

[0006] The basic solution provided by the present invention is a system for collaboratively monitoring and controlling farmland and river ecology, comprising a first-level and second-level weirs spaced apart, and gabion stone cage revetments on both sides of the river. The system also includes front-end data acquisition equipment, execution equipment, and a server.

[0007] The execution equipment includes irrigation equipment located in the farmland irrigation area, a first gate located in the first-level barrage, a second gate located in the second-level barrage, and an ecological flocculant delivery device located on the gabion stone cage revetment. The first gate and the second gate are both electric gates.

[0008] The server includes a data acquisition module, a data storage module, a data analysis and processing module and a control execution module;

[0009] The data acquisition module is used to obtain hydrological data including upstream water level, downstream flow and irrigation water supply, water quality data including total phosphorus concentration and total nitrogen concentration, soil moisture data including soil moisture content and crop stem water potential from the front-end acquisition device through the communication network, and obtain meteorological warning data including 24-hour precipitation from the server;

[0010] The data storage module is used to build a historical database, the data in the historical database includes hydrological, water quality, soil moisture and meteorological warning data obtained from the data acquisition module, and control parameters obtained from the control execution module, the control parameters including the first gate opening, the second gate opening, the irrigation equipment electric valve opening and the amount of ecological flocculant added;

[0011] The data analysis and processing module includes a prediction model submodule and a multi-objective optimization model submodule;

[0012] The prediction model submodule is used to build an LSTM prediction model. The input of the LSTM prediction model is the time series data of the control parameters, and the output is the predicted value of the state variables under the corresponding control parameter combination. The state variables include upstream water level, downstream flow, irrigation water supply, total phosphorus concentration and total nitrogen concentration. The LSTM prediction model is trained using data from the historical database;

[0013] The multi-objective optimization model submodule is used to establish a multi-objective optimization model based on the state variables predicted by LSTM. The multi-objective optimization model includes agricultural water demand targets, ecological water demand targets, flood control safety targets, and water quality purification targets with different weight coefficients. The weight coefficient of each target is dynamically allocated according to a preset scenario mode. The multi-objective optimization model obtains the target control parameters by optimizing the NSGA-II algorithm;

[0014] The control execution module is used to generate a control instruction based on the target control parameter and send it to the execution device through the communication network, drive the irrigation equipment, the first gate, the second gate and the ecological flocculant delivery device to perform corresponding operations according to the control instruction, and read the control parameters of the current execution device through the communication network and send them to the data storage module.

[0015] The principle of the present invention is as follows: hydrological data including upstream water level, downstream flow and irrigation water supply, water quality data including total phosphorus concentration and total nitrogen concentration, soil moisture data including soil moisture content and crop stem water potential, and meteorological warning data including 24-hour precipitation are obtained from the front-end collection equipment through the data acquisition module, all the acquired data are transmitted to the data storage module and the data analysis and processing module in real time, and a historical database is constructed through the data storage module. The historical database is used to store hydrological, water quality, soil moisture, meteorological warning data, and control parameters obtained from the control execution module, the control parameters including the first gate opening, the second gate opening, the irrigation equipment electric valve opening and the amount of ecological flocculant added; an LSTM prediction model is constructed through the prediction model submodule in the data analysis and processing module, and the LSTM prediction model is trained by the data in the historical database. The trained LSTM prediction model can output the predicted value of the state variable under the corresponding control parameter combination according to the time series data of the input control parameters, the state variable including the upstream water level, downstream flow, etc. , irrigation water supply, total phosphorus concentration and total nitrogen concentration. The multi-objective optimization model submodule establishes a multi-objective optimization model including agricultural water demand, ecological water demand, flood control safety and water quality purification targets according to the state variables. The weight coefficient of each target is dynamically allocated according to the preset different scenario modes, and the target control parameters are obtained by solving the optimized NSGA-II algorithm; the control execution module generates control instructions according to the target control parameters, and sends them to the execution device through the communication network to drive the irrigation equipment installed in the farmland, the first gate installed in the first-level weir, the second gate installed in the second-level weir and the ecological flocculant delivery device installed on the gabion stone cage revetment to perform corresponding operations according to the control instructions, so as to realize the regulation of farmland irrigation, river water level flow and water quality. At the same time, the control execution module reads the control parameters of the current execution device through the communication network and sends them to the data storage module; the whole process forms a closed loop, and through the real-time data collection, storage, analysis and execution and feedback of control instructions, the material cycle optimization and energy flow coupling of farmland and river ecosystems are realized, and multi-objective coordinated regulation is achieved.

[0016] Beneficial effects of the present invention:

[0017] 1. The system integrates four goals of agricultural water demand (ensuring the matching of crop water supply and demand), ecological water demand (maintaining river ecological base flow), flood safety (preventing water level from exceeding the limit), and water quality purification (preventing total phosphorus and total nitrogen concentration from exceeding the standard) through a multi-objective optimization model, and dynamically allocates the weight coefficients of each goal according to the preset scenario mode. The system can simultaneously consider irrigation water supply guarantee, river ecological flow maintenance, flood risk prevention and control, and water quality active purification, realize the coordinated regulation of the four goals, avoid water waste or excessive extraction caused by regulation for only a single target, and thus avoid ecological imbalance problems such as excessive irrigation that may damage river ecological base flow, and excessive flood discharge in flood season that may not dilute pollutants enough, causing water quality problems and affecting farmland and river ecology.

[0018] 2. The system can predict the state variables that can be achieved under the current control parameter setting according to the control parameter through the LSTM prediction model learning the mapping relationship between historical control parameters and state variables, and quickly solve the target control parameters of the multi-objective optimization model by combining the optimization NSGA-II algorithm, so as to realize fast and intelligent decision-making under data-driven, reduce waste and balance multi-objective conflicts.

[0019] 3. By adding an ecological flocculant feeding device, combined with the quantitative calculation of the water quality purification target in the multi-objective optimization model, the flocculant can be actively fed according to real-time water quality data, reducing the accumulation of pollutants in the river, and thus improving the biological habitat and alleviating the problem of reduced biodiversity caused by deteriorating water quality. Compared with the traditional passive purification relying on plant roots, the introduction of active water quality purification mechanism improves the intelligent regulation and control capability of the river pollution.

[0020] 4. Through the data acquisition module, data storage module, data analysis and processing module, and control execution module in the server, the whole process of monitoring, prediction, optimization, and execution is realized. Unlike the existing river water regulation and intelligent irrigation system, the present scheme establishes the mapping relationship between control parameters and state variables through the LSTM model, quickly solves the multi-objective optimization function by using the optimization NSGA-II algorithm, and can output target control parameters according to the current upstream water level, downstream flow, irrigation water supply, total phosphorus concentration, and total nitrogen concentration, to drive the first gate, the second gate, the irrigation equipment, and the ecological flocculant feeding device to perform corresponding operations, realizing the automation operation from data acquisition to device regulation.

[0021] Further, the LSTM model training steps are as follows:

[0022] S1. Obtain the control parameters in the historical database and perform normalization processing;

[0023] S2. Extract the control parameter time series data with a time interval of N and a time step of T after normalization as sample input, extract the state variable with a time interval of N after K time steps of the control parameter is applied as output, and train through the training set, validation set and test set divided by multiple groups of samples, where N is determined according to the sampling frequency of the data acquisition module.

[0024] Under this scheme, by normalizing the control parameter time series, we avoid model training deviations caused by data scale differences. By extracting the control parameter time series of T time steps as input, the LSTM model can accurately learn the dynamic relationship between control parameters and state variables from the input data of uniform scale, providing a reliable prediction basis for subsequent intelligent control. The LSTM model can capture long-term dependencies in the data, improve the prediction accuracy of state variables, and provide a reliable input basis for multi-objective optimization, thereby improving the accuracy and responsiveness of system control.

[0025] Furthermore, the objective function of the multi-objective optimization model is:

[0026] min(α·f irr +β·f eco +γ·f flood +δ·f water )

[0027] Among them, f irr is the irrigation water supply deviation rate, which is used to reflect the degree of deviation between the actual water supply of farmland and the water demand of crops, α is the agricultural water demand target weight coefficient, and f eco is the ecological flow deviation rate, which is used to measure the deviation between the measured river flow and the ecological flow threshold, β is the ecological water demand target weight coefficient, and f flood is the water level exceeding limit risk, which is used to quantify the risk of the measured water level exceeding the flood control limit water level, γ is the flood control safety target weight coefficient, and f water is the water pollution exceeding the standard, which is used to comprehensively evaluate the degree of exceeding the standard for total phosphorus and total nitrogen concentrations. δ is the water purification target weight coefficient, which satisfies α+β+γ+δ=1. The target weight coefficients are dynamically allocated according to the preset scenario mode.

[0028] Irrigation water supply deviation rate f irr The calculation formula is as follows:

[0029]

[0030] Among them, w irr_act The irrigation water supply output by the LSTM model, w irr_need The preset crop water requirement;

[0031] Ecological flow deviation rate f eco The calculation formula is as follows:

[0032]

[0033] Among them, Q act is the downstream flow output by the LSTM model, Q threshold is the preset ecological base flow threshold;

[0034] Risk of water level exceeding limit flood The calculation formula is as follows:

[0035]

[0036] Among them, H act H is the upstream water level output by the LSTM model. safe To set water levels for pre-set flood safety limits;

[0037]

[0038] Among them, C TP 、C TN are the total phosphorus concentration and total nitrogen concentration output by the LSTM model, C TP0 、C TN0 They are the preset standard total phosphorus limiting concentration and total nitrogen limiting concentration respectively.

[0039] Under this plan, the agricultural water demand target, ecological water demand target, flood control safety target and water quality purification target are quantified through the irrigation water supply deviation rate, ecological flow deviation rate, water level over-limit risk and water quality pollution exceeding the scale. By integrating the above four major goals through linear weighting, the complex ecological coordination problem is transformed into a computable optimization problem, which is convenient for algorithm solution and engineering implementation.

[0040] Furthermore, the scenario modes include a drought resistance and irrigation mode, a flood prevention priority mode, a water quality emergency mode, and an ecological maintenance mode. The rules for the preset scenario modes are as follows:

[0041] Drought resistance and irrigation mode: when the crop stem water potential is obtained Less than the preset value And the soil moisture content θ is less than the preset value θ0, and the weight coefficient satisfies α>max(β, γ, δ);

[0042] Flood control priority mode: When the 24-hour rainfall R is greater than the preset value R0 or the water level H upstream of the first-level dam act Greater than the preset flood control safety limit water level H safe When triggered, the weight coefficient satisfies γ>max(α, β, δ);

[0043] Water quality emergency mode: When the total phosphorus concentration C TP Greater than the preset standard total phosphorus limit concentration CTP0 Or total nitrogen concentration C TN Greater than the preset standard total nitrogen limit concentration C TN0 And it is triggered when the duration is greater than the preset value, and the weight coefficient satisfies δ>max(α, β, γ);

[0044] Ecological maintenance mode: triggered during non-drought, non-flood, and non-pollution periods when the above three modes are not met, and the weight coefficient satisfies β>max(α, γ, δ);

[0045] Among them, α is the target weight coefficient of agricultural water demand, β is the target weight coefficient of ecological water demand, γ is the target weight coefficient of flood control safety, and δ is the target weight coefficient of water quality purification. When the trigger conditions of multiple modes are met at the same time, the weight coefficients of each mode that meets the trigger conditions are weighted averaged to obtain the final weight coefficient. The specific calculation is as follows:

[0046]

[0047] Where i is the number of modes that meet the trigger conditions at the same time, i∈[1,3], α i , β i , γ i , δ i are the weight coefficients of each target in the i-th mode that meets the trigger conditions.

[0048] Under this solution, the four preset scenario modes cover typical working conditions in agricultural and river environments. Through clear trigger conditions, automatic identification of working conditions is achieved to avoid delays in manual judgment. Differentiated weight adjustments are made according to the scenario modes, allowing the system to focus on core target requirements at different stages and improve resource allocation efficiency.

[0049] Furthermore, the step of optimizing the NSGA-II algorithm to solve and obtain the target control parameters includes:

[0050] S1. Initialize the random solutions with a population size of 50. Each solution contains four decision variables: the first gate opening μ1, the second gate opening μ2, the irrigation equipment electric valve opening μ3, and the amount of ecological flocculant μ4, where μ1, μ2, μ3∈[0,1], 0 represents fully closed, 1 represents fully open, and μ4∈[0,1], where μ4 represents the amount of ecological flocculant as a percentage of the maximum capacity of the ecological flocculant delivery device;

[0051] S2, generate Pareto optimal solution set through fast non-dominated sorting and congestion calculation;

[0052] S3. After 30 iterations, the solution closest to the ideal point is selected as the output solution of the target control parameters. The ideal point is the point where [μ1, μ2, μ3, μ4] = [0, 0, 0, 0].

[0053] S4. Perform boundary constraint processing on the current output solution according to different scenario modes to ensure that the target control parameters are within a reasonable range under the current model scenario;

[0054] The boundary constraints include: in the drought resistance and irrigation mode, μ1 ≥ μ 1_eco And μ2≥μ 2_eco , where μ 1_eco and μ 2_eco They are the preset minimum first gate opening and second gate opening in drought resistance and irrigation mode to ensure the maintenance of river ecological base flow; in flood control priority mode, μ4≥μ 4_flood , where μ 4_flood The preset minimum amount of ecological flocculant in the flood prevention priority mode to ensure that the water quality meets the standard; in the water quality emergency mode, μ2≤μ 2_water And μ3≤μ 3_water , where μ 2_water The maximum second gate opening in the water quality emergency mode is preset to slow down the water flow and ensure that the ecological flocculant can fully play its role. 3_water The maximum electric valve opening of the irrigation equipment is preset in the water quality emergency mode to reduce the pollution load of the farmland discharge water.

[0055] In this scheme, the parameter settings of population size 50 and iteration 30 generations enable the algorithm to run efficiently and meet real-time control requirements. The non-dominated sorting method retains diverse solutions to avoid falling into local optimality, ensuring that an equilibrium solution is found when multiple objectives conflict. Boundary constraint processing ensures that the algorithm output solution is within a reasonable range of target control parameters under the current mode scenario. In the drought resistance and irrigation mode, the algorithm maintains the minimum ecological flow while giving priority to meeting irrigation needs; in the flood prevention priority mode, the water level is quickly lowered while controlling the content of flood pollutants; in the water quality emergency mode, the retention time is increased while the purification effect is enhanced while the high-intensity purification is carried out, and irrigation is reduced to reduce farmland runoff pollution.

[0056] The present invention also discloses a method for collaboratively monitoring and regulating farmland and river ecology, which is applied to the above-mentioned collaborative monitoring and regulating system for farmland and river ecology and includes the following steps:

[0057] S1. Real-time acquisition of hydrological data including upstream water levels, downstream flow, and irrigation water supply; water quality data including total phosphorus and total nitrogen concentrations; soil moisture data including soil moisture content and crop stem water potential; meteorological warning data including 24-hour precipitation; and data on the first and second gate openings, irrigation equipment electric valve openings, and ecological flocculant dosage, and the construction of a historical database based on the above data.

[0058] S2, construct an LSTM model for establishing a mapping relationship between control parameters and state variables, the control parameters including a first gate opening degree, a second gate opening degree, an irrigation equipment electric valve opening degree and an ecological flocculant dosage, and the state variables including an upstream water level, a downstream flow, an irrigation water supply, a total phosphorus concentration and a total nitrogen concentration, the LSTM model being trained by using data in a historical database;

[0059] S3, a multi-objective optimization model is constructed based on the state variables, the multi-objective optimization model including agricultural water demand targets, ecological water demand targets, flood control safety targets and water quality purification targets with different weight coefficients, and the weight coefficients of the targets being dynamically allocated according to different scenario modes, the scenario modes being determined according to obtained hydrological, water quality and entropy data and preset scenario mode determination rules;

[0060] S4, state variables under different control parameters are output by the LSTM model, the multi-objective optimization model is solved by using an optimization NSGA-II algorithm, the output solution is boundary-constrained according to the current scenario mode, and target control parameters are generated;

[0061] S5, the target control parameters are converted into control instructions and sent to an execution device through a communication network to perform corresponding operations. BRIEF DESCRIPTION OF DRAWINGS

[0062] Figure 1 It is a system block diagram of the embodiment 1 of the ecological collaborative monitoring and regulation system for farmland and riverway of the application;

[0063] Figure 2 It is a riverway structure plan view of the embodiment 1 of the application;

[0064] In the attached drawings of the specification, the attached marks include a first river blocking dam 1, a second river blocking dam 2, a gabion revetment 3, an ecological flocculant dosing device 4 and a farmland irrigation area 5. DETAILED DESCRIPTION

[0065] The following will be further described in detail through specific embodiments:

[0066] The embodiment 1 is basically as shown in the attached Figure 1 and Figure 2 :

[0067] An ecological collaborative monitoring and regulation system for farmland and riverway, comprising a first river blocking dam 1 and a second river blocking dam 2 arranged at intervals of 250 meters, and gabion revetments 3 on both sides of the riverway, the ecological collaborative monitoring and regulation system for farmland and riverway further comprising a front-end acquisition device, an execution device and a server.

[0068] The front-end data collection equipment includes a water level meter located upstream of the first-level barrage 1, an electromagnetic flowmeter located downstream of the second-level barrage 2, a turbine flowmeter located at the water intake of the irrigation equipment, a water quality sensor located at the bottom of the gabion stone cage revetment 3, a soil moisture sensor and a crop stem water potential sensor located in the farmland irrigation area 5, and the data collection frequency of each sensor is unified to be once every 30 minutes;

[0069] The execution equipment includes irrigation equipment located in the farmland irrigation area 5, a first gate located in the first-level barrage 1, a second gate located in the second-level barrage 2, and an ecological flocculant delivery device 4 located on the gabion stone cage revetment 3. The first gate and the second gate are both electric gates. The ecological flocculant delivery device 4 is located between the first-level barrage 1 and the second-level barrage 2.

[0070] The server includes a data acquisition module, a data storage module, a data analysis and processing module and a control execution module.

[0071] The data acquisition module is used to obtain hydrological data including upstream water level, downstream flow and irrigation water supply, water quality data including total phosphorus concentration and total nitrogen concentration, soil moisture data including soil moisture content and crop stem water potential from the front-end acquisition device through the communication network, and obtain meteorological warning data including 24-hour precipitation from the server;

[0072] The data storage module is used to build a historical database, the data in the historical database includes hydrological, water quality, soil moisture and meteorological warning data obtained from the data acquisition module, and control parameters obtained from the control execution module, the control parameters including the first gate opening, the second gate opening, the irrigation equipment electric valve opening and the amount of ecological flocculant added;

[0073] The data analysis and processing module includes a prediction model submodule and a multi-objective optimization model submodule.

[0074] The prediction model submodule is used to build an LSTM prediction model. The input of the LSTM prediction model is the time series data of the control parameters, and the output is the predicted value of the state variables under the corresponding control parameter combination. The state variables include upstream water level, downstream flow, irrigation water supply, total phosphorus concentration and total nitrogen concentration. The LSTM prediction model is trained using data from the historical database;

[0075] The LSTM model training steps are as follows:

[0076] S1. Obtain control parameters from the historical database and perform normalization processing;

[0077] S2. Extract the control parameter time series data with a time interval of N and a time step of T after normalization as sample input, extract the state variable with a time interval of N after K time steps of the control parameter is applied as output, and train through the training set, validation set and test set divided by multiple groups of samples, where N is determined according to the sampling frequency of the data acquisition module.

[0078] In this embodiment, the acquisition frequency of each sensor is unified as once every 30 minutes, so the time interval N is 30 minutes, T is preset to 6, and K is preset to 6. Each sample data covers 3 hours of control parameter time series data and state variables after 3 hours. Samples are obtained from the historical database and divided into training set, validation set and test set in a ratio of 6:3:1. The weight parameters of the LSTM model are optimized by the gradient descent method. The model structure consists of 2 LSTM layers (128 neurons each), 1 fully connected layer, the activation function is ReLU, the optimizer is Adam, and the learning rate is 0.001, until the mean square error of the model on the validation set converges to the preset threshold value of 0.1. The trained LSTM model can output the predicted value of the state variable under the action of this set of control parameters based on the input control parameters.

[0079] The multi-objective optimization model submodule is used to establish a multi-objective optimization model based on the state variables predicted by LSTM. The multi-objective optimization model includes agricultural water demand targets, ecological water demand targets, flood control safety targets, and water quality purification targets with different weight coefficients. The weight coefficient of each target is dynamically allocated according to a preset scenario mode. The multi-objective optimization model obtains the target control parameters by optimizing the NSGA-II algorithm;

[0080] Furthermore, the objective function of the multi-objective optimization model is:

[0081] min(α·f irr +β·f eco +γ·f flood +δ·f water )

[0082] Among them, f irr is the irrigation water supply deviation rate, which is used to reflect the degree of deviation between the actual water supply of farmland and the water demand of crops, α is the agricultural water demand target weight coefficient, and f eco is the ecological flow deviation rate, which is used to measure the deviation between the measured river flow and the ecological flow threshold, β is the ecological water demand target weight coefficient, and f flood is the water level exceeding limit risk, which is used to quantify the risk of the measured water level exceeding the flood control limit water level, γ is the flood control safety target weight coefficient, and f wateris the water pollution exceeding the standard, which is used to comprehensively evaluate the degree of exceeding the standard for total phosphorus and total nitrogen concentrations. δ is the water purification target weight coefficient, which satisfies α+β+γ+δ=1. The target weight coefficients are dynamically allocated according to the preset scenario mode.

[0083] Irrigation water supply deviation rate f irr The calculation formula is as follows:

[0084]

[0085] Among them, w irr_act The irrigation water supply output by the LSTM model, w irr_heed The preset crop water requirement;

[0086] Ecological flow deviation rate f eco The calculation formula is as follows:

[0087]

[0088] Among them, Q act is the downstream flow output by the LSTM model, Q threshold is the preset ecological base flow threshold;

[0089] Risk of water level exceeding limit flood The calculation formula is as follows:

[0090]

[0091] Among them, H act H is the upstream water level output by the LSTM model. safe To set water levels for pre-set flood safety limits;

[0092]

[0093] Among them, C TP 、C TN are the total phosphorus concentration and total nitrogen concentration output by the LSTM model, C TP0 、C TN0 are the preset standard total phosphorus limiting concentration and total nitrogen limiting concentration, C TP0 、C TN0 The preset values ​​are 0.3mg / L and 1.2mg / L respectively.

[0094] The scenario modes include drought resistance and irrigation mode, flood prevention priority mode, water quality emergency mode and ecological maintenance mode. The rules of the preset scenario modes are as follows:

[0095] Drought resistance and irrigation mode: when the crop stem water potential is obtained Less than the preset value and the soil moisture content θ is less than a preset value θ0, the weight coefficient satisfies α>max(β, γ, δ);

[0096] flood prevention priority mode: when the acquired 24-hour rainfall R is greater than a preset value R0 or the water level H act of the upstream of the first-level dam is greater than a preset flood control safety limit water level H safe , the weight coefficient satisfies γ>max(α, β, δ);

[0097] water quality emergency mode: when the acquired total phosphorus concentration C TP is greater than a preset standard total phosphorus limit concentration C TP0 or the total nitrogen concentration C TN is greater than a preset standard total nitrogen limit concentration C TN0 and the duration is greater than a preset value, the weight coefficient satisfies δ>max(α, β, γ);

[0098] ecological maintenance mode: triggered in a period that does not meet the above three modes, that is, non-drought, non-flood, and non-pollution, the weight coefficient satisfies β>max(α, γ, δ);

[0099] wherein α is the weight coefficient of the agricultural water demand target, β is the weight coefficient of the ecological water demand target, γ is the weight coefficient of the flood control safety target, and δ is the weight coefficient of the water quality purification target. When the triggering conditions of multiple modes are met at the same time, the weight coefficients of each mode that meets the triggering condition are weighted and averaged to obtain the final weight coefficient, which is calculated as follows:

[0100]

[0101] wherein i is the number of modes that meet the triggering condition at the same time, i∈[1, 3], α i , β i , γ i , and δ i are the weight coefficients of each target in the i-th mode that meets the triggering condition;

[0102] Specifically, in this embodiment, the preset crop stem water potential preset value is -1.5 MPa, the soil moisture content preset value θ0 is 20%, and the 24-hour rainfall preset value R0 is 50 mm. The rules of the preset scenario mode are as follows:

[0103] drought resistance and irrigation preservation mode: when the acquired crop stem water potential and the soil moisture content θ<20% are triggered, the weight coefficients are α=0.6, β=0.2, γ=0.1, and δ=0.1;

[0104] flood prevention priority mode: when the acquired 24-hour rainfall R>50mm or the water level H act>2.8m, weight coefficients satisfy α=0.15, β=0.1, γ=0.7, δ=0.05;

[0105] Water quality emergency mode: when the total phosphorus concentration C TP >0.3mg / L or total nitrogen concentration C TN >1.2mg / L and the duration is greater than 2 hours, weight coefficients satisfy α=0.1, β=0.2, γ=0.1, δ=0.6;

[0106] Ecological maintenance mode: triggered in the non-drought, non-flood, non-pollution period when the above three modes are not met, weight coefficients satisfy α=0.4, β=0.3, γ=0.1, δ=0.2;

[0107] Specifically, in the present embodiment, when the measured θ<20%, C TP >0.3mg / L, the conditions for triggering the drought resistance and water quality emergency modes are met at the same time, at this time the final weight coefficients are obtained by weighted average according to the weight coefficients of each mode that meets the triggering condition, as follows:

[0108]

[0109] Finally, α=0.35, β=0.2, γ=0.1, δ=0.35.

[0110] The step of solving the target control parameters by the optimized NSGA-II algorithm includes:

[0111] S1, initialize a random solution with a population size of 50, each solution containing four decision variables: first gate opening μ1, second gate opening μ2, irrigation equipment electric valve opening μ3 and ecological flocculant dosage μ4, wherein μ1, μ2, μ3∈[0,1], 0 represents complete closing and 1 represents complete opening, μ4∈[0,1], wherein μ4 represents the ecological flocculant dosage as a percentage of the maximum capacity of the ecological flocculant dosing device;

[0112] S2, generate a Pareto optimal solution set by fast non-dominated sorting and congestion calculation;

[0113] S3, after 30 generations of iteration, select the solution closest to the ideal point as the output solution of the target control parameters, the ideal point being the point [μ1, μ2, μ3, μ4] = [0, 0, 0, 0];

[0114] S4, boundary constraint processing is performed on the current output solution according to different scenario modes, to ensure that the target control parameters are within a reasonable range under the current mode scenario;

[0115] The boundary constraints include: in the drought resistance and irrigation mode, μ1 ≥ μ 1_eco And μ2≥μ 2_eco , where μ 1_eco and μ 2_eco are the preset minimum first gate opening and second gate opening in drought resistance and irrigation mode, respectively. In this embodiment, both are 0.3 to ensure the maintenance of river ecological base flow; in flood control priority mode, μ4≥μ 4_flood , where μ 4_flood is the preset minimum amount of ecological flocculant in the flood prevention priority mode to ensure that the water quality meets the standard, which is 0.5 in this embodiment; in the water quality emergency mode, μ2≤μ 2_water And μ3≤μ 3_water , where μ 2_water μ is the preset maximum second gate opening in the water quality emergency mode to slow down the water flow and ensure that the ecological flocculant can fully play its role. In this embodiment, it is 0.5; 3_water It is the preset maximum electric valve opening of the irrigation equipment in the water quality emergency mode to reduce the pollution load of the farmland drainage water, which is 0.4 in this embodiment.

[0116] Specifically, in this embodiment, the drought resistance and irrigation mode and the water quality emergency mode are triggered simultaneously, and the target control parameters [μ1, μ2, μ3, μ4] = [0.4, 0.3, 0.3, 0.5] are obtained by optimizing the NSGA-II algorithm.

[0117] The control execution module is configured to generate a control instruction based on the target control parameter and transmit it to the execution device via the communication network, thereby driving the irrigation equipment, the first gate, the second gate, and the eco-flocculant delivery device to perform corresponding operations according to the control instruction, and to read the control parameters of the current execution device via the communication network and transmit them to the data storage module. In this embodiment, the opening of the first gate is adjusted to 0.4, the opening of the second gate is adjusted to 0.3, the opening of the irrigation electric valve is adjusted to 0.3, and the eco-flocculant delivery device delivers the eco-flocculant at 50% of its maximum capacity.

[0118] The above is only an embodiment of the present application, and the common knowledge of specific structures and characteristics in the scheme is not described in detail, and the ordinary skilled person in the art knows all the ordinary technical knowledge in the technical field of the present application before the application date or the priority date, can know all the prior art in the field, and has the ability to apply conventional experimental means before that date, and the ordinary skilled person in the art can improve and implement the present scheme under the guidance of the present application, and some typical known structures or known methods should not be an obstacle for the ordinary skilled person in the art to implement the present application. It should be pointed out that for those skilled in the art, without departing from the structure of the present application, a number of modifications and improvements can be made, which should also be considered as the protection scope of the present application, which will not affect the effect and practicality of the patent. The protection scope of the present application should be subject to the content of its claims, and the specific implementation mode and the like in the specification can be used to explain the content of the claims.

Claims

1. A farmland and river ecological coordinated monitoring and control system, comprising a primary and secondary weirs spaced apart, and gabion stone cage revetments on both sides of the river, characterized by: The farmland and river ecological coordinated monitoring and control system also includes front-end acquisition equipment, execution equipment and servers; The execution equipment includes irrigation equipment located in the farmland irrigation area, a first gate located in the first-level barrage, a second gate located in the second-level barrage, and an ecological flocculant delivery device located on the gabion stone cage revetment. The first gate and the second gate are both electric gates. The server includes a data acquisition module, a data storage module, a data analysis and processing module and a control execution module; The data acquisition module is used to obtain hydrological data including upstream water level, downstream flow and irrigation water supply, water quality data including total phosphorus concentration and total nitrogen concentration, soil moisture data including soil moisture content and crop stem water potential from the front-end acquisition device through the communication network, and obtain meteorological warning data including 24-hour precipitation from the server; The data storage module is used to build a historical database, the data in the historical database includes hydrological, water quality, soil moisture and meteorological warning data obtained from the data acquisition module, and control parameters obtained from the control execution module, the control parameters including the first gate opening, the second gate opening, the irrigation equipment electric valve opening and the amount of ecological flocculant added; The data analysis and processing module includes a prediction model submodule and a multi-objective optimization model submodule; The prediction model submodule is used to build an LSTM prediction model. The input of the LSTM prediction model is a time series sequence of control parameters, and the output is the predicted value of the state variable under the corresponding control parameter combination. The state variables include upstream water level, downstream flow, irrigation water supply, total phosphorus concentration and total nitrogen concentration. The LSTM prediction model is trained using data from a historical database; The multi-objective optimization model submodule is used to establish a multi-objective optimization model based on the state variables predicted by LSTM. The multi-objective optimization model includes agricultural water demand targets, ecological water demand targets, flood control safety targets, and water quality purification targets with different weight coefficients. The weight coefficient of each target is dynamically allocated according to a preset scenario mode. The multi-objective optimization model obtains the target control parameters by optimizing the NSGA-II algorithm; The control execution module is used to generate a control instruction based on the target control parameter and send it to the execution device through the communication network, drive the irrigation equipment, the first gate, the second gate and the ecological flocculant delivery device to perform corresponding operations according to the control instruction, and read the control parameters of the current execution device through the communication network and send them to the data storage module.

2. The farmland and river ecological coordinated monitoring and control system according to claim 1 is characterized by: The LSTM model training steps are as follows: S1. Obtain control parameters from the historical database and perform normalization processing; S2. Extract the control parameter time series sequence with a time interval of N and a time step of T after normalization as the sample input, extract the state variable with a time interval of N after K time steps of the control parameter is applied as the output, and train through the training set, validation set and test set divided by multiple groups of samples, where N is determined according to the sampling frequency of the data acquisition module.

3. The farmland and river ecological coordinated monitoring and control system according to claim 2 is characterized by: The objective function of the multi-objective optimization model is: min(α·f irr +β·f eco +γ·f flood +δ·f water ) Among them, f irr is the irrigation water supply deviation rate, which is used to reflect the degree of deviation between the actual water supply of farmland and the water demand of crops, α is the agricultural water demand target weight coefficient, and f eco is the ecological flow deviation rate, which is used to measure the deviation between the measured river flow and the ecological flow threshold, β is the ecological water demand target weight coefficient, and f flood is the water level exceeding limit risk, which is used to quantify the risk of the measured water level exceeding the flood control limit water level, γ is the flood control safety target weight coefficient, and f water is the water pollution exceeding the standard, which is used to comprehensively evaluate the degree of exceeding the standard for total phosphorus and total nitrogen concentrations. δ is the water purification target weight coefficient, which satisfies α+β+γ+δ=1. The target weight coefficients are dynamically allocated according to the preset scenario mode. Irrigation water supply deviation rate f irr The calculation formula is as follows: Among them, w irr_act The irrigation water supply output by the LSTM model, w irr_need The preset crop water requirement; Ecological flow deviation rate f eco The calculation formula is as follows: Among them, Q act is the downstream flow output by the LSTM model, Q threshold is the preset ecological base flow threshold; Risk of water level exceeding limit flood The calculation formula is as follows: Among them, H act H is the upstream water level output by the LSTM model. safe To set water levels for pre-set flood safety limits; Among them, C TP 、C TN are the total phosphorus concentration and total nitrogen concentration output by the LSTM model, They are the preset standard total phosphorus limiting concentration and total nitrogen limiting concentration respectively.

4. The farmland and river ecological coordinated monitoring and control system according to claim 3 is characterized by: The scenario modes include drought resistance and irrigation mode, flood prevention priority mode, water quality emergency mode and ecological maintenance mode. The rules of the preset scenario modes are as follows: Drought resistance and irrigation mode: when the crop stem water potential is obtained Less than the preset value And the soil moisture content θ is less than the preset value θ0, and the weight coefficient satisfies α>max(β, γ, δ); Flood control priority mode: When the 24-hour rainfall R is greater than the preset value R0 or the water level H upstream of the first-level dam act Greater than the preset flood control safety limit water level H safe When triggered, the weight coefficient satisfies γ>max(α, β, δ); Water quality emergency mode: When the total phosphorus concentration C TP Greater than the preset standard total phosphorus limit concentration Or total nitrogen concentration C TN Greater than the preset standard total nitrogen limit concentration And it is triggered when the duration is greater than the preset value, and the weight coefficient satisfies δ>max(α, β, γ); Ecological maintenance mode: triggered during non-drought, non-flood, and non-pollution periods when the above three modes are not met, and the weight coefficient satisfies β>max(α, γ, δ); Among them, α is the target weight coefficient of agricultural water demand, β is the target weight coefficient of ecological water demand, γ is the target weight coefficient of flood control safety, and δ is the target weight coefficient of water quality purification. When the trigger conditions of multiple modes are met at the same time, the weight coefficients of each mode that meets the trigger conditions are weighted averaged to obtain the final weight coefficient. The specific calculation is as follows: Where i is the number of modes that meet the trigger conditions at the same time, i∈[1,3], α i , β i , γ i , δ i are the weight coefficients of each target in the i-th mode that meets the trigger conditions.

5. The farmland and river ecological coordinated monitoring and control system according to claim 4 is characterized by: The step of optimizing the NSGA-II algorithm to obtain the target control parameters includes: S1. Initialize the random solutions with a population size of 50. Each solution contains four decision variables: the first gate opening μ1, the second gate opening μ2, the irrigation equipment electric valve opening μ3, and the amount of ecological flocculant μ4, where μ1, μ2, μ3∈[0,1], 0 represents fully closed, 1 represents fully open, and μ4∈[0,1], where μ4 represents the amount of ecological flocculant as a percentage of the maximum capacity of the ecological flocculant delivery device; S2, generate Pareto optimal solution set through fast non-dominated sorting and congestion calculation; S3. After 30 iterations, the solution closest to the ideal point is selected as the output solution of the target control parameters. The ideal point is the point where [μ1, μ2, μ3, μ4] = [0, 0, 0, 0]. S4. Perform boundary constraint processing on the current output solution according to different scenario modes to ensure that the target control parameters are within a reasonable range under the current model scenario; The boundary constraints include: in the drought resistance and irrigation mode, μ1 ≥ μ 1_eco And μ2≥μ 2_eco , where μ 1_eco and μ 2_eco They are the preset minimum first gate opening and second gate opening in drought resistance and irrigation mode to ensure the maintenance of river ecological base flow; in flood control priority mode, μ4≥μ 4_flood , where μ 4_flood The preset minimum amount of ecological flocculant in the flood prevention priority mode to ensure that the water quality meets the standard; in the water quality emergency mode, μ2≤μ 2_water And μ3≤μ 3_water , where μ 2_water The maximum second gate opening in the water quality emergency mode is preset to slow down the water flow and ensure that the ecological flocculant can fully play its role. 3_water The maximum electric valve opening of the irrigation equipment is preset in the water quality emergency mode to reduce the pollution load of the farmland discharge water.

6. A method for collaborative monitoring and control of farmland and river ecology, characterized by: The farmland and river ecological coordinated monitoring and control system according to any one of claims 1 to 5 comprises the following steps: S1. Real-time acquisition of hydrological data including upstream water levels, downstream flow, and irrigation water supply; water quality data including total phosphorus and total nitrogen concentrations; soil moisture data including soil moisture content and crop stem water potential; meteorological warning data including 24-hour precipitation; and data on the first and second gate openings, irrigation equipment electric valve openings, and ecological flocculant dosage, and the construction of a historical database based on the above data. S2. Build an LSTM model to establish a mapping relationship between control parameters and state variables. The control parameters include the first gate opening, the second gate opening, the irrigation equipment electric valve opening, and the amount of ecological flocculant added. The state variables include the upstream water level, the downstream flow, the irrigation water supply, the total phosphorus concentration, and the total nitrogen concentration. The LSTM model is trained using data from a historical database. S3. Constructing a multi-objective optimization model based on the state variables, wherein the multi-objective optimization model includes agricultural water demand targets, ecological water demand targets, flood control safety targets, and water quality purification targets with different weight coefficients. The weight coefficients of the various targets are dynamically assigned according to different scenario modes, and the scenario modes are determined based on the acquired hydrological, water quality, and entropy data and preset scenario mode judgment rules; S4. Output the state variables under different control parameters through the LSTM model, use the optimized NSGA-II algorithm to solve the multi-objective optimization model, perform boundary constraint processing on the output solution according to the current scenario mode, and generate the target control parameters; S5. Convert the target control parameters into control instructions and send them to the execution device through the communication network to perform corresponding operations.

Images