A vegetation water requirement prediction method and system for sea embankment ecological protection

Abstract

The present application relates to the technical field of resource management, in particular to a vegetation water requirement prediction method and system for sea embankment ecological protection, comprising collecting environmental monitoring data of the sea embankment area; a salt stress factor model is established based on soil data to calculate the comprehensive salt stress coefficient of the monitoring point, dynamically correct the vegetation water requirement based on multiple factors, and obtain the basic vegetation water requirement; a multi-dimensional time sequence model considering the relationship between the tidal cycle and soil water replenishment is established to obtain resource scheduling weights under different tidal stages and environmental conditions; based on the vegetation data, underground data and environmental data of different sea embankment areas, a water resource allocation network based on multi-region cooperation is used to obtain a distribution optimization scheme based on regional characteristic parameters; the basic vegetation water requirement, resource scheduling weights and distribution optimization scheme are comprehensively used to generate vegetation water requirement prediction results and hierarchical resource scheduling schemes for irrigation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of resource management, in particular to a vegetation water requirement prediction method and system for ecological protection of seawalls. BACKGROUND

[0002] At present, there is a lack of scientific and quantitative management tools and systematic decision support frameworks for resource management and organizational planning of large-scale ecological restoration projects. There are problems such as single resource allocation model and insufficient decision-making process.

[0003] In terms of workflow management, traditional project scheduling methods are mainly based on historical experience and simple linear programming models, lacking effective modeling capabilities for complex decision-making scenarios under multiple variables and multiple constraints. Existing enterprise management information systems have a lot of room for improvement in terms of data integration, prediction accuracy, and decision automation. In particular, in large-scale, long-cycle ecological protection projects, project managers have difficulty in achieving unified optimization management of multi-dimensional resources such as personnel scheduling, equipment configuration, and cost control.

[0004] With the rapid development of the ecological protection industry, enterprises have an increasingly urgent need for intelligent project management systems, and it is necessary to establish a data-driven resource optimization allocation model to improve organizational operational efficiency, reduce management costs, and achieve fine-grained management throughout the project lifecycle.

[0005] Therefore, a vegetation water requirement prediction method and system for ecological protection of seawalls are proposed. SUMMARY

[0006] The present application aims to provide a vegetation water requirement prediction method and system for ecological protection of seawalls, which includes collecting environmental monitoring data of the seawall area; establishing a salt stress factor model based on soil data to calculate the comprehensive salt stress coefficient of the monitoring points, dynamically correcting the vegetation water requirement with multiple factors, and obtaining the basic vegetation water requirement; establishing a multi-dimensional time series model considering the relationship between the tidal cycle and soil water replenishment due to geological structure differences, to obtain resource scheduling weights under different tidal stages and environmental conditions; based on the vegetation data, underground data, and environmental data of different seawall areas, and based on the water resource allocation network of multi-region collaboration, obtaining a deployment optimization scheme based on regional characteristic parameters; integrating the basic vegetation water requirement, resource scheduling weights, and deployment optimization scheme to generate vegetation water requirement prediction results and hierarchical resource scheduling schemes for irrigation.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0008] A vegetation water requirement prediction method for ecological protection of seawalls, comprising:

[0009] Collecting environmental monitoring data of the seawall area, including soil data, vegetation data, underground data and environmental data;

[0010] Based on the soil data, a salt stress factor model is established, combined with the vegetation data and the environmental data, the comprehensive salt stress coefficient of the monitoring point is calculated, the vegetation basic water requirement is obtained by dynamically correcting the vegetation water requirement according to the comprehensive salt stress coefficient;

[0011] Based on the soil data, the underground data and the environmental data, a multi-dimensional time sequence model considering the difference of geological structure is established, the relationship between the tidal period and the soil water replenishment is obtained, and the resource scheduling weight under different tidal stages and environmental conditions is obtained;

[0012] According to the vegetation data, the underground data and the environmental data of different seawall areas, based on the water resource allocation network of multi-region cooperation, the optimization scheme based on the regional characteristic parameters is obtained;

[0013] The vegetation basic water requirement, the resource scheduling weight and the optimization scheme are comprehensively considered to generate the vegetation water requirement prediction result and the hierarchical resource scheduling scheme, and irrigation is carried out.

[0014] Preferably, the soil data includes soil conductivity, soil permeability and soil moisture content;

[0015] The vegetation data includes vegetation species data, growth stage data and vegetation distribution;

[0016] The underground data includes underground water level change data and geological feature change data;

[0017] The environmental data includes tidal time data, environmental temperature data, environmental humidity data and rainfall data.

[0018] Preferably, the salt stress factor model includes a data preprocessing layer, a feature extraction layer and a stress calculation layer:

[0019] The data preprocessing layer normalizes the soil conductivity, uses a piecewise nonlinear regression method to establish a conversion relationship between the soil conductivity and the soil salt concentration in different concentration intervals, and corrects the standardized salt concentration data according to the soil type and temperature; the feature extraction layer queries a preset vegetation salt tolerance database according to the vegetation type data, extracts the salt sensitivity coefficient of the corresponding vegetation, calculates the growth stage correction factor combined with the growth stage data, establishes a coupling response model of temperature and humidity and salt stress based on the principles of plant physiology, and calculates the environmental influence factor; the stress calculation layer uses a nonlinear fusion algorithm to consider the interaction and threshold effect between factors, and performs multi-dimensional weighted calculation on the standardized salt concentration data, the salt sensitivity coefficient, the growth stage correction factor and the environmental influence factor to obtain a comprehensive salt stress coefficient; based on the comprehensive salt stress coefficient and the preset vegetation theoretical water requirement, a dynamic correction algorithm is used to calculate the vegetation basic water requirement;

[0020] The vegetation salt tolerance database includes the salt lethal concentration of common vegetation on seawalls, the optimal growth salt range, the salt sensitivity parameters at different growth stages, and the temperature and humidity coupling influence coefficients, the dynamic correction algorithm uses a nonlinear mapping function to segmentally correct the theoretical water requirement according to the grade range of the comprehensive salt stress coefficient.

[0021] Preferably, the multi-dimensional time sequence model includes a time sequence decomposition layer, a correlation analysis layer and a weight calculation layer:

[0022] The time sequence decomposition layer uses a Fourier transform method to periodically decompose the tidal time data, extracts the main tidal period characteristics, uses a moving average algorithm to analyze the trend of the groundwater level change data, and obtains the groundwater level response data; the correlation analysis layer establishes a permeability classification matrix based on the soil permeability data and the groundwater level response data, uses a correlation analysis method to calculate the correlation degree of the tidal period and the soil moisture content change, and combines the rainfall data to analyze the comprehensive influence of rainfall, evapotranspiration, runoff and deep seepage on soil moisture through a complete water balance model to obtain multi-factor correlation relationship data; the weight calculation layer uses an analytic hierarchy process to distribute the weights of the tidal influence, groundwater recharge, soil permeability and rainfall contribution based on the multi-factor correlation relationship data, and adjusts the regional differences combined with the geological feature change data to calculate the resource scheduling weight under different tidal stages and environmental conditions.

[0023] Preferably, the allocation optimization scheme is obtained as follows: the seawall area is divided into several sub-areas using a clustering algorithm according to vegetation distribution data, and a geological stability level is assigned to each sub-area based on the geological characteristic change data in the underground data; a water resource allocation network is constructed according to the geographical location relationship and pipeline layout conditions between the sub-areas, and the flow-pressure relationship and transportation capacity constraint parameters of each allocation channel are set based on the hydraulic principles; minimizing the total allocation cost and maximizing the balance of resource utilization are used as the objective function, and the minimum water demand guarantee of each sub-area, the pipeline transportation capacity and hydraulic balance are used as constraints, and dynamic optimization calculations are performed in combination with environmental data to generate an allocation optimization scheme based on regional characteristic parameters.

[0024] Preferably, the vegetation water demand prediction results and the hierarchical resource scheduling scheme are generated as follows: a data fusion algorithm is used to comprehensively calculate the basic vegetation water demand, resource scheduling weights and allocation optimization schemes, a multivariate prediction model is established through a machine learning method, and historical data and real-time monitoring data are combined to predict the trend of vegetation water demand changes in future periods; a prediction confidence interval is established based on the error analysis of the prediction model and the parameter uncertainty propagation theory, and a three-level early warning mechanism is established according to the confidence interval width and the prediction deviation. When the predicted water demand deviation exceeds the set threshold, the corresponding level of scheduling response is automatically triggered; a decision tree algorithm is used in combination with an expert knowledge base to generate a hierarchical resource scheduling scheme including conventional scheduling, early warning scheduling and emergency scheduling based on historical scheduling data and current prediction results, and a scheduling schedule and resource allocation plan are output.

[0025] A vegetation water demand prediction system for seawall ecological protection, comprising:

[0026] Data acquisition unit, collecting environmental monitoring data of the seawall area;

[0027] The vegetation basic water requirement acquisition unit establishes a salt stress factor model based on soil data, calculates the comprehensive salt stress coefficient of the monitoring point, and performs multi-factor dynamic correction on the vegetation water requirement to obtain the vegetation basic water requirement;

[0028] The tidal-environmental joint unit builds a multidimensional time series model based on soil data, subsurface data, and environmental data to obtain resource scheduling weights under different tidal stages and environmental conditions;

[0029] The allocation optimization scheme acquisition unit obtains allocation optimization schemes based on regional characteristic parameters based on vegetation data, underground data, and environmental data of different seawall areas and a multi-regional coordinated water resource allocation network;

[0030] The scheduling scheme generating unit comprehensively considers the basic water demand of vegetation, resource scheduling weight and allocation optimization scheme, generates vegetation water demand prediction results and hierarchical resource scheduling scheme for irrigation.

[0031] Compared with the prior art, the present application has the beneficial effects that:

[0032] 1、The present application proposes a salt stress factor model, which converts soil conductivity into standardized salt concentration through a multi-layer data processing structure, and combines vegetation species, growth stages and environmental temperature and humidity to construct a coupling response model, and then calculates a comprehensive salt stress coefficient. Starting from the physiological salt tolerance of vegetation, the theoretical water requirement is dynamically corrected, which is more in line with the actual water requirement state in the high salt and high variation environment of the seawall area. In addition, the present application considers the interaction and threshold effect of multiple factors, has sensitivity and robustness in the high salt stress interval, effectively improves the accuracy and applicability of the prediction results, and provides support for the irrigation management of saline-alkali seawall ecological vegetation.

[0033] 2、The present application considers the special tidal dynamics and groundwater recharge characteristics of the seawall ecological area, and constructs a multi-dimensional time series model. The tidal period is extracted by Fourier transform, and the coupling influence of factors such as tide, water level, soil permeability and rainfall on soil moisture content is quantitatively revealed. A variety of hydrological environmental factors are fused and processed to form a complete soil water dynamic regulation mechanism. By introducing the difference adjustment weight of geological structure, dynamic weight redistribution can be realized according to different geological and climatic conditions, further improving the scheduling accuracy, and optimizing the irrigation control strategy.

[0034] 3、The present application constructs a water resource allocation network based on clustering algorithm and geological data, divides the resource scheduling unit at the regional level, and generates a personalized optimization scheme with the goal of minimizing the total allocation cost and maximizing the resource utilization balance. The present application can automatically trigger a three-level response mechanism according to the prediction deviation, match the normal, early warning and emergency scheduling modes, realize the intelligentization, dynamization and hierarchical management of water resource allocation, and enhance the response ability of the water resource system to sudden changes. While ensuring the health of vegetation, it realizes the efficient use of resources and the long-term stability of the ecological system. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 A vegetation water requirement prediction method flowchart for seawall ecological protection provided by the present application;

[0036] Figure 2 A vegetation water requirement prediction system structure diagram for seawall ecological protection provided by the present application;

[0037] Figure 3 A salt stress factor model structure diagram provided by the present application embodiment. DETAILED DESCRIPTION

[0038] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0039] The present application provides a vegetation water requirement prediction method for seawall ecological protection, which is applied to a vegetation water requirement prediction system for seawall ecological protection. The system comprises five modules, namely a data acquisition unit, a vegetation basic water requirement acquisition unit, a tide-environment joint unit, an allocation optimization scheme acquisition unit and a scheduling scheme generation unit. The method flow chart and system structure chart are shown in Figure 1 and Figure 2 ; the specific technical solutions are as follows:

[0040] The environmental monitoring data of the seawall area are collected, and the environmental monitoring data are standardized, including soil data, vegetation data, underground data and environmental data.

[0041] Based on the soil data, a salt stress factor model is established, the vegetation data and the environmental data are combined, the comprehensive salt stress coefficient of the monitoring point is calculated, the vegetation water requirement is dynamically corrected according to the comprehensive salt stress coefficient, and the vegetation basic water requirement is obtained.

[0042] Based on the soil data, the underground data and the environmental data, a multi-dimensional time sequence model considering the relationship between the tide cycle and the soil water replenishment considering the geological structure difference is established, and the resource scheduling weight under different tide stages and environmental conditions is obtained.

[0043] According to the vegetation data, the underground data and the environmental data of different seawall areas, based on the water resource allocation network of multi-region cooperation, the allocation optimization scheme based on the regional characteristic parameters is obtained.

[0044] The vegetation water requirement prediction result and the hierarchical resource scheduling scheme are generated by comprehensively considering the vegetation basic water requirement, the resource scheduling weight and the allocation optimization scheme, and irrigation is performed.

[0045] The soil data includes soil conductivity, soil permeability and soil moisture content; the vegetation data includes vegetation species data, growth stage data and vegetation distribution; the underground data includes underground water level change data and geological feature change data; and the environmental data includes tide time data, environmental temperature data, environmental humidity data and rainfall data.

[0046] In the embodiment, the optimization of irrigation management is realized by collecting and standardizing the soil, vegetation, underground and environmental monitoring data of the seawall area. The vegetation water requirement is dynamically corrected combined with multi-source data, the influence of salt is effectively alleviated, the vegetation health is improved, and the irrigation efficiency is improved.

[0047] The salt stress factor model comprises a data preprocessing layer, a feature extraction layer and a stress calculation layer, which are specifically referred to Figure 3 :

[0048] The data preprocessing layer normalizes the soil conductivity, uses a piecewise nonlinear regression method to establish the conversion relationship between soil conductivity and soil salt concentration in different concentration intervals, and corrects according to the soil type and temperature to obtain standardized salt concentration data; the feature extraction layer queries the preset vegetation salt tolerance database according to the vegetation species data, extracts the salt sensitivity coefficient of the corresponding vegetation, calculates the growth stage correction factor combined with the growth stage data, establishes a coupling response model of temperature and humidity and salt stress, and calculates the environmental influence factor; the stress calculation layer adopts a nonlinear fusion algorithm, considers the interaction and threshold effect between factors, and performs multi-dimensional weighted calculation on the standardized salt concentration data, the salt sensitivity coefficient, the growth stage correction factor and the environmental influence factor to obtain a comprehensive salt stress coefficient; based on the comprehensive salt stress coefficient and the preset vegetation theoretical water requirement, a dynamic correction algorithm is used to calculate the vegetation basic water requirement;

[0049] The vegetation salt tolerance database contains the salt lethal concentration, the optimal growth salt range, the salt sensitivity parameters at different growth stages and the temperature and humidity coupling influence coefficients of common seawall vegetation, and the dynamic correction algorithm uses a nonlinear mapping function to segmentally correct the theoretical water requirement according to the grade range of the comprehensive salt stress coefficient.

[0050] The salt stress factor model further comprises an adaptive learning layer and a real-time correction layer: the adaptive learning layer collects the deviation data of the actual growth condition of the vegetation and the prediction result, uses an incremental learning algorithm to dynamically update the vegetation salt tolerance database, and when the prediction deviation of the same vegetation type exceeds 15% within 7 consecutive days, automatically triggers the parameter learning mechanism, re-trains the salt sensitivity coefficient using the monitoring data of the last 30 days, and the update frequency is adaptively adjusted to every week, every month or every quarter according to the deviation degree;

[0051] The real-time correction layer sets up a multi-source data cross-validation mechanism. Through triple verification of leaf salt content spectrum detection, vegetation growth index remote sensing monitoring and soil salt sensor, when the difference between any two detection results exceeds 20%, data quality evaluation is started. The Bayesian fusion algorithm is used to allocate weights and correct the results of multi-source data, so as to ensure the accuracy and reliability of the calculation of salt stress coefficient. The adaptive learning and multi-source data cross-validation mechanism are introduced, which solves the problem that the traditional static parameter model cannot adapt to environmental changes, and improves the accuracy and reliability of salt stress evaluation.

[0052] In this embodiment, the salt stress factor model is used to optimize irrigation management in coastal areas. The irrigation accuracy is improved, the water resource waste is reduced, the vegetation healthy growth and yield increase are promoted, and efficient and sustainable water resource management is realized.

[0053] The multi-dimensional time series model includes a time series decomposition layer, a correlation analysis layer and a weight calculation layer:

[0054] The time series decomposition layer uses the Fourier transform method to periodically decompose the tidal time data, extracts the main tidal period characteristics, uses the moving average algorithm to analyze the trend of the groundwater level change data, and obtains the groundwater level response data; the correlation analysis layer establishes a permeability classification matrix based on soil permeability data and groundwater level response data, uses correlation analysis method to calculate the correlation degree of tidal period and soil moisture content change, and combines rainfall data to analyze the comprehensive influence of rainfall, evapotranspiration, runoff and deep seepage on soil moisture by using complete water balance model, to obtain multi-factor correlation relationship data; the weight calculation layer is based on multi-factor correlation relationship data, uses the analytic hierarchy process to allocate weights to tidal influence, groundwater recharge, soil permeability and rainfall contribution, and combines geological feature change data for regional differentiation adjustment, to calculate resource scheduling weight under different tidal stages and environmental conditions. The complete water balance model is a quantitative framework based on the law of conservation of mass, which is used to describe the input, output and storage change relationship of water in a specific region within a given period.

[0055] The multi-dimensional time series model also includes an anomaly detection layer and a multi-scale analysis layer: the anomaly detection layer uses a dual anomaly detection mechanism based on statistics and machine learning, uses the 3-sigma criterion to identify outliers in the tidal time data, groundwater level data and soil moisture content data, and marks statistical anomalies when the data deviates from the historical mean value by more than 3 standard deviations; at the same time, the isolated forest algorithm is used to detect abnormal patterns in the multi-dimensional data space, and the abnormal score threshold is set to 0.6. When abnormal data is detected, the data repair program is automatically started, and the time series interpolation and adjacent monitoring point data fusion method are used for abnormal value replacement;

[0056] The multi-scale analysis layer establishes a hierarchical analysis framework for four time scales: hourly, daily, weekly, and monthly. The hourly analysis focuses on the immediate impact of short-term tidal fluctuations on soil moisture, the daily analysis focuses on the coupling effects of diurnal temperature differences and tidal cycles, the weekly analysis focuses on the cumulative impact of tidal size period changes, and the monthly analysis focuses on seasonal tides and long-term trends in groundwater levels. The results of each scale analysis are integrated through weighted fusion to form a comprehensive time series feature. The weight distribution is 0.2 for hourly, 0.3 for daily, 0.3 for weekly, and 0.2 for monthly. The multi-scale time analysis framework and intelligent anomaly detection mechanism are established, which can capture the complex dynamic characteristics of the tidal-soil moisture relationship at different time scales, and improve the robustness of the time series model.

[0057] In this embodiment, the multi-dimensional time series model is used to optimize coastal irrigation management, integrating tidal, groundwater, soil, and weather data. Its three-layer architecture of time series decomposition, correlation analysis, and weight calculation simulates environmental factors affecting soil moisture. Techniques such as Fourier transform and analytic hierarchy process are used to improve model accuracy and adaptability. Considering the coastal environment, such as high salinity and water level fluctuations, the irrigation strategy optimization is ensured, and the crop resilience and yield are improved.

[0058] The method for obtaining the optimization scheme is as follows: according to the vegetation distribution data, the seawall area is divided into several sub-regions by using a clustering algorithm, and the geological stability level of each sub-region is assigned based on the geological feature change data in the underground data; a water resource allocation network is constructed according to the geographical location relationship between the sub-regions and the pipe network layout conditions, and the flow-pressure relationship and transport capacity constraint parameters of each allocation channel are set based on the principles of hydraulics; the minimum total allocation cost and the maximum resource utilization balance are taken as the objective functions, the minimum water demand guarantee of each sub-region, the pipe network transport capacity, and the hydraulic balance are taken as the constraint conditions, and dynamic optimization calculation is performed combined with environmental data to generate an allocation optimization scheme based on regional characteristic parameters.

[0059] In this embodiment, the clustering algorithm is used to optimize the division of seawall sub-regions combined with vegetation distribution data, and the stability level is assigned based on the geological feature change data, which improves the scientificity of water resource allocation. The allocation network is constructed based on the principles of hydraulics, and the flow-pressure relationship and transport capacity constraints are set to achieve efficient allocation of water resources. Taking the minimum total allocation cost and the maximum resource utilization balance as the target, combined with dynamic optimization calculation, the resource utilization efficiency is improved under the constraints of guaranteeing the minimum water demand, the pipe network transport capacity, and the hydraulic balance.

[0060] The generation mode of the vegetation water requirement prediction result and the hierarchical resource scheduling scheme is: the vegetation basic water requirement, the resource scheduling weight and the allocation optimization scheme are comprehensively calculated by using a data fusion algorithm, a multivariate prediction model is established by using a machine learning method, the vegetation water requirement change trend in a future period is predicted in combination with historical data and real-time monitoring data; a prediction confidence interval is established based on error analysis of the prediction model and parameter uncertainty propagation theory, a three-level early warning mechanism is established according to the confidence interval width and the prediction deviation, and the scheduling response of the corresponding level is automatically triggered when the prediction water requirement deviation exceeds the set threshold; a decision tree algorithm is used in combination with an expert knowledge base to generate a hierarchical resource scheduling scheme including regular scheduling, early warning scheduling and emergency scheduling according to historical scheduling data and current prediction results, and a scheduling time table and a resource allocation scheme are output.

[0061] The vegetation water requirement prediction adopts a multi-model integration and dynamic feedback optimization mechanism: the multi-model integration adopts three prediction models of support vector machine regression, random forest regression and long short-term memory neural network to run in parallel, each model is trained based on different feature combinations, the support vector machine model focuses on soil and environmental factors, the random forest model focuses on vegetation and growth stage factors, and the neural network model focuses on time sequence and tidal factors; a dynamic weight integration algorithm is used to adjust the integration weight in real time according to the recent prediction accuracy of each model, when the prediction error of a single model is less than 5% for three consecutive days, the weight is increased to 0.5, when the error is more than 15%, the weight is reduced to 0.2, and the remaining weight is evenly distributed among the other models;

[0062] The dynamic feedback optimization mechanism establishes a prediction effect evaluation and model parameter automatic optimization system, compares the predicted value with the measured value every 24 hours to calculate the mean absolute error and the root mean square error, automatically triggers the model retraining program when the prediction error continuously exceeds the set threshold, uses a hyperparameter optimization method combining grid search and genetic algorithm to automatically adjust the model parameters to improve the prediction accuracy, and simultaneously establishes a prediction confidence dynamic adjustment mechanism to dynamically adjust the confidence interval width according to the similarity of historical prediction accuracy and current environmental conditions, thereby improving the credibility of the prediction result. The multi-model integration and dynamic feedback optimization strategy significantly improves the prediction accuracy through model complementation and adaptive optimization.

[0063] In the embodiment, the vegetation water requirement change trend is predicted, a multivariate prediction model is established in combination with historical and real-time data, and the intelligibility and response speed of irrigation management are effectively improved. The decision tree algorithm and the expert knowledge base are used to generate a hierarchical resource scheduling scheme, the resource allocation is optimized, the water resource waste is significantly reduced, and the timeliness and accuracy of irrigation are ensured.

[0064] The present application significantly optimizes irrigation management by collecting and standardizing soil, vegetation, underground, and environmental monitoring data in the seawall area, combining salt stress factor models, multi-dimensional time series models, and water resource allocation networks. Through a three-layer architecture of data preprocessing, feature extraction, and stress calculation, the vegetation water demand is dynamically corrected, effectively alleviating the impact of salt and improving vegetation health and yield. The multi-dimensional time series model integrates the tidal cycle, geological structure differences, and soil water relationship to optimize irrigation timing and water quantity, saving water resources. The water resource allocation network realizes multi-region collaboration based on regional characteristic parameters to ensure fair and efficient use of resources. The comprehensive vegetation water demand prediction and hierarchical resource scheduling scheme support forward-looking irrigation planning, reducing costs and waste. The present application improves irrigation efficiency and promotes sustainable agricultural development in saline coastal areas, with significant economic and social benefits; the weights appearing in this embodiment are obtained by machine learning analysis of historical data.

[0065] Embodiment two:

[0066] Collect environmental monitoring data in the seawall area, including soil data, vegetation data, underground data, and environmental data.

[0067] The soil data includes soil conductivity, soil permeability, and soil moisture content;

[0068] The vegetation data includes vegetation species data, growth stage data, and vegetation distribution; each vegetation corresponds to different salt tolerance parameters; the growth stage data is divided into four stages: germination, growth, maturity, and senescence, and each stage is set with a corresponding water demand correction coefficient; vegetation distribution is obtained using GPS positioning and remote sensing image recognition technology.

[0069] The underground data includes underground water level change data and geological feature change data; the underground water level change data is recorded by an automatic water level monitor every hour; the geological feature change data includes soil thickness, geological structure type, and underground water permeability, where the geological structure is divided into clay layer, sandy soil layer, and sandy gravel layer, each corresponding to different water transmission capacity parameters.

[0070] The environmental data includes tidal time data, environmental temperature data, environmental humidity data, and rainfall data; the tidal time data records the specific time and tidal height of each daily high tide and low tide.

[0071] Based on the soil data, a salt stress factor model is established, combined with vegetation data and environmental data, to calculate the comprehensive salt stress coefficient of the monitoring point, and to dynamically correct the vegetation water demand based on the comprehensive salt stress coefficient to obtain the basic water demand of the vegetation.

[0072] The salt stress factor model includes a data preprocessing layer, a feature extraction layer, and a stress calculation layer:

[0073] The data preprocessing layer adopts the minimum-maximum standardization algorithm to normalize the soil conductivity. The specific calculation method is to subtract the minimum conductivity value in the region from the measured conductivity value, and then divide by the difference between the maximum conductivity value and the minimum conductivity value. The piecewise linear regression method is used to establish the conversion relationship between soil conductivity and soil salt concentration in different concentration intervals. When the conductivity is in the range of 0-2 millisiemens per centimeter, the soil salt concentration is equal to the conductivity multiplied by a coefficient of 0.64. When the conductivity is in the range of 2-8 millisiemens per centimeter, the soil salt concentration is equal to the conductivity multiplied by 0.64 plus the conductivity squared multiplied by a coefficient of 0.02. When the conductivity is greater than 8 millisiemens per centimeter, the soil salt concentration is equal to the conductivity multiplied by a coefficient of 0.8. According to the soil type and temperature, the correction coefficient of clay soil is 1.1, and the correction coefficient of sandy soil is 0.9. When the environmental temperature increases by 10°C, the salt concentration calculation result is multiplied by a temperature correction coefficient of 1.05, and the standardized salt concentration data is obtained.

[0074] The feature extraction layer queries the preset vegetation salt tolerance database according to the vegetation species data. The database includes the salt sensitivity coefficient, the optimal growth salt range, and the salt lethal concentration of reed. Combined with the growth stage data, the growth stage correction factor is calculated. The correction factor for the germination stage is 1.3, the correction factor for the growth stage is 1.0, the correction factor for the mature stage is 0.8, and the correction factor for the senescence stage is 1.2. The coupling response model of temperature and humidity and salt stress is established. When the environmental temperature exceeds 30°C and the humidity is less than 60%, the salt stress enhancement coefficient is 1.2. When the environmental temperature is lower than 10°C and the humidity is higher than 85%, the salt stress attenuation coefficient is 0.8. Under other conditions, the coefficient is 1.0. The environmental influence factor is calculated.

[0075] The stress calculation layer adopts a weighted average fusion algorithm, considering the interaction and threshold effect between factors. The standardized salt concentration data, salt sensitivity coefficient, growth stage correction factor, and environmental influence factor are calculated in multiple dimensions. The weight distribution is 0.4 for salt concentration, 0.3 for sensitivity coefficient, 0.2 for growth stage, and 0.1 for environmental factor. When any factor exceeds the set threshold, the threshold effect processing is started. When the salt concentration exceeds the vegetation lethal concentration, the stress coefficient is directly set to 1.0. The comprehensive salt stress coefficient is obtained. Based on the comprehensive salt stress coefficient and the preset vegetation theoretical water requirement, the segmented correction algorithm is used to calculate the vegetation basic water requirement. When the stress coefficient is less than 0.2, the water requirement correction coefficient is 1.0. When the stress coefficient is in the range of 0.2-0.5, the correction coefficient is 1.2. When the stress coefficient is in the range of 0.5-0.8, the correction coefficient is 1.5. When the stress coefficient is greater than 0.8, the correction coefficient is 2.0.

[0076] The vegetation theoretical water requirement database includes the daily water requirement of various types of vegetation under standard environmental conditions.

[0077] Based on soil data, underground data and environmental data, a multi-dimensional time series model considering the difference of geological structure is established to obtain the relationship between tidal cycle and soil moisture replenishment, and the resource scheduling weight under different tidal stages and environmental conditions is obtained.

[0078] The multi-dimensional time series model includes a time series decomposition layer, a correlation analysis layer and a weight calculation layer:

[0079] The time series decomposition layer uses the fast Fourier transform method to periodically decompose the tidal time data, identifies the main tidal cycle characteristics including the semidiurnal tide cycle of 12 hours and the diurnal tide cycle of 24 hours, extracts the tidal amplitude and phase information, and uses the moving average algorithm to analyze the trend of the underground water level change data, eliminating the influence of short-term fluctuations, calculating the response time delay of the underground water level to the tidal change, and obtaining the underground water level response data.

[0080] The correlation analysis layer establishes a three-level permeability classification matrix based on soil permeability data, and the permeability coefficient of high permeability soil is greater than 10 -4 cm / s, corresponding to a weight coefficient of 0.7, the permeability coefficient of medium permeability soil is between 10 -6 cm / s and 10 -4 cm / s, corresponding to a weight coefficient of 0.5, and the permeability coefficient of low permeability soil is less than 10 -6 cm / s, corresponding to a weight coefficient of 0.3; the Pearson correlation analysis method is used to calculate the correlation between the tidal cycle and the change of soil moisture content, and the correlation coefficient is greater than 0.7, indicating strong correlation, 0.3-0.7 indicating moderate correlation, and less than 0.3 indicating weak correlation; combined with rainfall data, the comprehensive influence of rainfall, evapotranspiration, runoff and deep seepage on soil moisture is analyzed by water balance analysis method, the evapotranspiration is estimated by Penman formula, the runoff coefficient is set to 0.1-0.4 according to soil type, and the deep seepage is calculated according to soil permeability, to obtain multi-factor correlation data.

[0081] The weight calculation layer is based on the multi-factor correlation data, and the analytic hierarchy process is used to allocate weights to the contributions of tidal influence, groundwater recharge, soil permeability and rainfall, with tidal influence weight 0.35, groundwater recharge weight 0.25, soil permeability weight 0.25, and rainfall contribution weight 0.15; combined with geological feature change data for regional differentiation adjustment, all weights in clay layer region are multiplied by an adjustment coefficient of 0.8, weights in sand layer region remain unchanged, and weights in sand and gravel layer region are multiplied by an adjustment coefficient of 1.2; according to the four tidal stages of rising tide period, high tide period, falling tide period and low tide period, the resource scheduling weight is calculated respectively.

[0082] According to the vegetation data, underground data and environmental data of different seawall regions, based on the multi-region collaborative water resource allocation network, the optimization scheme based on regional characteristic parameters is obtained.

[0083] According to the vegetation distribution data, the seawall area is divided into several sub-regions; based on the geological feature change data in the underground data, the geological stability of each sub-region is assigned, and the geological stability is divided into three levels of A level stability, B level medium, and C level instability; according to the change range of underground water level and the type of geological structure, the region with underground water level change less than 0.5 meters and clay layer structure is evaluated as A level, the region with underground water level change of 0.5-1.5 meters and sand layer structure is evaluated as B level, and the region with underground water level change greater than 1.5 meters and gravel layer structure is evaluated as C level.

[0084] According to the geographical location relationship between the sub-regions and the pipe network layout conditions, a water resource allocation network is constructed, and the network topology structure adopts a tree distribution mode; based on the principle of hydraulics, the flow-pressure relationship and the transmission capacity constraint parameters of each allocation channel are set.

[0085] Taking the minimization of total allocation cost and the maximization of resource utilization balance as the objective function, the total allocation cost includes water source cost, transmission cost and management cost; the resource utilization balance is measured by calculating the variance of the water demand satisfaction rate of each sub-region, and the smaller the variance, the better the balance; the constraint conditions include that the minimum water demand guarantee of each sub-region is not less than 80% of the theoretical water demand, the pipe network transmission capacity is not more than 90% of the design flow, and the hydraulic balance error of each node is not more than 5%; combined with the rainfall and evaporation prediction in the environmental data, dynamic optimization calculation is carried out, genetic algorithm is used for solution, and an allocation optimization scheme based on regional characteristic parameters is generated.

[0086] Integrating the vegetation basic water demand, resource scheduling weight and allocation optimization scheme, the vegetation water demand prediction result and hierarchical resource scheduling scheme are generated, and irrigation is carried out.

[0087] The vegetation basic water demand, resource scheduling weight and allocation optimization scheme are integrated by using a weighted fusion algorithm; a multivariate prediction model is established by using a support vector machine regression method, the input variables include historical vegetation water demand, environmental temperature and humidity, tide data and soil moisture content, and the output variable is the predicted value of future vegetation water demand; combined with the historical monitoring data and real-time monitoring data, the future vegetation water demand change trend is predicted.

[0088] Based on the error analysis of the prediction model and the parameter uncertainty propagation theory, the prediction confidence interval is established, the Monte Carlo method is used for 1000 times of random sampling, the 95% confidence interval of the prediction result is calculated, and the confidence interval width is used as the measurement index of the prediction uncertainty; According to the confidence interval width and the prediction deviation, a three-level early warning mechanism is established, the first level warning: the prediction deviation is less than 10% and the confidence interval width is less than 15%, the conventional scheduling mode is adopted; The second level warning: the prediction deviation is in the range of 10%-25% or the confidence interval width is in the range of 15%-30%, the early warning scheduling mode is adopted, the monitoring frequency is increased and the standby water source is prepared; The third level warning: the prediction deviation is greater than 25% or the confidence interval width is greater than 30%, the emergency scheduling mode is adopted, the emergency water source is started and the artificial intervention is implemented.

[0089] The decision tree algorithm is combined with the expert knowledge base to generate a hierarchical resource scheduling scheme including conventional scheduling, early warning scheduling and emergency scheduling according to historical scheduling data and current prediction results; The expert knowledge base contains 200 scheduling rules, covering the optimal scheduling strategy under different vegetation types, seasonal characteristics and weather conditions; The root node of the decision tree determines the prediction deviation level, the second layer node determines the vegetation type and growth stage, the third layer node determines the environmental condition, and the leaf node outputs the specific scheduling scheme; The scheduling scheme includes irrigation schedule, water distribution scheme and equipment operation parameter, and outputs detailed scheduling schedule and resource allocation scheme.

[0090] Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method for predicting vegetation water demand for seawall ecological protection, characterized in that: include: Collect environmental monitoring data of the seawall area, including soil data, vegetation data, underground data and environmental data; A salt stress factor model was established based on soil data. Combined with vegetation data and environmental data, the comprehensive salt stress coefficient of the monitoring point was calculated. Based on the comprehensive salt stress coefficient, a multi-factor dynamic correction was performed on the vegetation water requirement to obtain the basic vegetation water requirement. The salt stress factor model includes a data preprocessing layer, a feature extraction layer, and a stress calculation layer: The data preprocessing layer normalizes soil electrical conductivity and uses a piecewise nonlinear regression method to establish a conversion relationship between soil electrical conductivity and soil salt concentration in different concentration ranges. It then corrects the data based on soil type and temperature to obtain standardized salt concentration data. The feature extraction layer queries a preset vegetation salt tolerance database based on vegetation species data, extracts the salt sensitivity coefficient of the corresponding vegetation, calculates the growth stage correction factor based on the growth stage data, establishes a coupled response model of temperature, humidity and salt stress, and calculates the environmental impact factor. The stress calculation layer uses a nonlinear fusion algorithm, taking into account the interaction and threshold effect between various factors, and performs a multi-dimensional weighted calculation on the standardized salt concentration data, salt sensitivity coefficient, growth stage correction factor and environmental impact factor to obtain a comprehensive salt stress coefficient. Based on the comprehensive salt stress coefficient and the preset theoretical water requirement of vegetation, a dynamic correction algorithm is used to calculate the basic water requirement of vegetation; The vegetation salt tolerance database contains the lethal salt concentration of common seawall vegetation, the optimal growth salt range, the salt sensitivity parameters of different growth stages, and the temperature and humidity coupling influence coefficient. The dynamic correction algorithm uses a nonlinear mapping function to perform segmented correction on the theoretical water demand according to the level range of the comprehensive salt stress coefficient. Based on soil, underground, and environmental data, a multidimensional time series model of the relationship between tidal cycles and soil water replenishment was established, taking into account differences in geological structure, to obtain resource scheduling weights under different tidal stages and environmental conditions. Based on the vegetation data, underground data and environmental data of different seawall areas, and based on the multi-regional coordinated water resource allocation network, an allocation optimization plan based on regional characteristic parameters is obtained; The basic water demand of vegetation, resource scheduling weight and allocation optimization plan are comprehensively considered to generate vegetation water demand prediction results and hierarchical resource scheduling plans for irrigation; The vegetation water demand prediction results and hierarchical resource scheduling scheme are generated as follows: a data fusion algorithm is used to comprehensively calculate the basic vegetation water demand, resource scheduling weight and allocation optimization scheme, a multivariate prediction model is established through a machine learning method, and historical data and real-time monitoring data are combined to predict the trend of vegetation water demand changes in future periods; a prediction confidence interval is established based on the error analysis of the prediction model and the parameter uncertainty propagation theory, and a three-level early warning mechanism is established according to the confidence interval width and the prediction deviation. When the predicted water demand deviation exceeds the set threshold, the corresponding level of scheduling response is automatically triggered; a decision tree algorithm is used in combination with an expert knowledge base to generate a hierarchical resource scheduling scheme including conventional scheduling, early warning scheduling and emergency scheduling based on historical scheduling data and current prediction results, and a scheduling schedule and resource allocation plan are output.

2. The method for predicting vegetation water demand for seawall ecological protection according to claim 1, characterized in that: The soil data includes soil electrical conductivity, soil permeability and soil moisture content; the vegetation data includes vegetation type data, growth stage data and vegetation distribution; the underground data includes groundwater level change data and geological characteristic change data; the environmental data includes tidal time data, ambient temperature data, ambient humidity data and rainfall data.

3. The method for predicting vegetation water demand for seawall ecological protection according to claim 1, characterized in that: The multidimensional time series model includes a time series decomposition layer, a correlation analysis layer, and a weight calculation layer: The time series decomposition layer uses the Fourier transform method to periodically decompose the tidal moment data, extract the main tidal cycle characteristics, and use the sliding average algorithm to perform trend analysis on the groundwater level change data to obtain the groundwater level response data; the correlation analysis layer establishes a permeability classification matrix and groundwater level response data based on soil permeability data, and uses the correlation analysis method to calculate the correlation between the tidal cycle and soil moisture content changes. Combined with rainfall data, the complete water balance model is used to analyze the comprehensive impact of rainfall, evapotranspiration, runoff and deep infiltration on soil moisture to obtain multi-factor correlation relationship data; the weight calculation layer uses the hierarchical analysis method based on the multi-factor correlation relationship data to assign weights to tidal influence, groundwater recharge, soil infiltration and rainfall contribution, and combines the geological characteristics change data to make regional differentiated adjustments to calculate the resource scheduling weights under different tidal stages and environmental conditions.

4. The method for predicting vegetation water demand for seawall ecological protection according to claim 1, characterized in that: The method for obtaining the allocation optimization plan is as follows: based on the vegetation distribution data, a clustering algorithm is used to divide the seawall area into several sub-areas, and a geological stability level is assigned to each sub-area based on the geological characteristic change data in the underground data; a water resource allocation network is constructed based on the geographical relationship between the sub-areas and the pipe network layout conditions, and the flow-pressure relationship and transmission capacity constraint parameters of each allocation channel are set based on the principles of hydraulics; Taking minimization of total allocation cost and maximization of resource utilization balance as objective functions, and minimum water demand guarantee of each sub-region, pipeline network transportation capacity and hydraulic balance as constraints, dynamic optimization calculation is performed in combination with environmental data to generate an allocation optimization plan based on regional characteristic parameters.

5. A vegetation water demand prediction system for seawall ecological protection, characterized in that: The method for predicting vegetation water demand for seawall ecological protection according to claim 1 comprises: The data acquisition unit collects environmental monitoring data of the seawall area; the vegetation basic water demand acquisition unit establishes a salt stress factor model based on soil data, calculates the comprehensive salt stress coefficient of the monitoring point, and performs multi-factor dynamic correction on the vegetation water demand to obtain the vegetation basic water demand; the tide-environment joint unit establishes a multi-dimensional time series model based on soil data, underground data and environmental data to obtain resource scheduling weights under different tidal stages and environmental conditions; the allocation optimization plan acquisition unit obtains an allocation optimization plan based on regional characteristic parameters based on the vegetation data, underground data and environmental data of different seawall areas and a multi-region collaborative water resource allocation network; the scheduling plan generation unit comprehensively considers the vegetation basic water demand, resource scheduling weight and allocation optimization plan to generate vegetation water demand prediction results and a hierarchical resource scheduling plan for irrigation.

Images