Sand stabilization method based on water circulation regulation and control in sand desert photovoltaic field

Abstract

The invention discloses a sand stabilization method based on water circulation regulation and control in a sand desert photovoltaic field, and relates to the technical field of sand desert governing, the method comprises the following specific steps: S100, collecting real-time water resource monitoring data, S200, forming a comprehensive water resource information base, S300, constructing a water circulation prediction model and outputting a prediction result, and S400, determining a water circulation prediction model according to the prediction result. S500, the sand stabilization effect of the photovoltaic field area is evaluated regularly, and strategy adjustment and optimization are carried out according to the evaluation result, the water resource utilization efficiency can be remarkably improved, all-around and real-time accurate control over water resources is achieved through multi-water-source combined monitoring and big data platform integration, and the water resource utilization rate is improved. Water resource waste caused by information lag or inaccuracy is avoided, and irrigation water can be reasonably allocated according to real-time water demand conditions and water source supply capacities of vegetation in different areas on the basis of a water circulation prediction model and dynamic allocation.

Description

Technical Field

[0001] The present invention relates to the technical field of sandy desert control, and specifically to a sand fixation method based on water cycle regulation in a photovoltaic field in sandy deserts. Background Art

[0002] With the increasing global demand for clean energy, due to the rich solar energy resources, sandy desert areas have become important regions for the development of the photovoltaic industry. However, the ecological environment in these areas is fragile and sand and wind activities are frequent. This not only causes serious wear and erosion to photovoltaic equipment, reduces power generation efficiency, and shortens the service life of the equipment, but also exacerbates land desertification, seriously threatening the balance of the surrounding ecosystem. At the same time, water resources in sandy deserts are extremely scarce, limited precipitation is difficult to retain, and the extraction of groundwater is difficult. Traditional water resource management and sand fixation methods are difficult to effectively balance the dual needs of photovoltaic industry development and ecological protection in such a harsh environment.

[0003] Currently, there are deficiencies in the treatment of photovoltaic fields in sandy deserts in the prior art. In terms of water cycle regulation, for water resource monitoring, single-point and isolated methods are mostly used, and it is impossible to comprehensively and real-time grasp the dynamics of multiple water sources such as rainwater, groundwater, and surface water. It is difficult to accurately estimate the available water volume. The constructed water cycle model often only considers simple meteorological factors and ignores the complex effects of key factors such as soil texture and vegetation growth conditions on water migration and transformation. As a result, the water resource allocation is unreasonable, the irrigation plan lacks pertinence, resulting in water resource waste or the death of some vegetation due to water shortage; in terms of sand fixation, the sand barrier structure is single and fixed, and it cannot be flexibly adjusted according to the sand and wind intensity, and the wind prevention and sand fixation effect is limited; the vegetation planting does not fully combine the differences in light, temperature, and water distribution in the field area, the species selection is single, and the survival rate is low. It is difficult to form a stable and effective ecological protection system and cannot achieve the long-term sustainable sand fixation goal.

[0004] In summary, there is an urgent need for an innovative and comprehensive method in the current photovoltaic fields in sandy deserts, which can not only finely regulate the water cycle, achieve the efficient utilization of water resources, but also carry out scientific sand fixation in combination with the ecological characteristics of the field area to ensure the coordinated development of the photovoltaic industry and the ecological environment. Summary of the Invention

[0005] The purpose of the present invention is to make up for the deficiencies of the prior art, and provide a sand fixation method based on water cycle regulation in a photovoltaic field in sandy deserts. It can establish a water cycle prediction model based on big data, comprehensively consider various factors such as meteorological data, soil humidity, and vegetation growth conditions, accurately predict the water cycle in the photovoltaic field area, and according to the prediction results, adjust the irrigation plan and water resource allocation plan in advance to achieve the refined regulation of the water cycle, ensure the reasonable utilization of water resources and the healthy growth of vegetation.

[0006] To solve the above technical problems, the present invention provides the following technical solution: A sand fixation method based on water cycle regulation in a photovoltaic field in sandy desert areas. The specific steps of the method are as follows:

[0007] S100. Real-time monitor the water sources in the photovoltaic field area, and continuously obtain the water level L w (t), flow rate F w (t), and water quality data using a sensor network within time t, and generate real-time water resource monitoring data;

[0008] S200. Integrate the real-time water resource monitoring data into a big data platform, perform data preprocessing, and form a comprehensive water resource information database. Among them, the preprocessing includes:

[0009] For the water level L w (t), remove noise interference through to stabilize the water level data, where is the filtered water level value at time t, n is the size of the sliding window, i is the index variable used to traverse the data points within the sliding window, and L w (i) is the water level measurement value pointed to by the index variable i, and w i is the weighting coefficient and satisfies

[0010] For the flow rate data, use an adaptive threshold correction algorithm based on the flow rate change rate to correct the abnormal flow rate values caused by equipment failures and sudden rainfall. The flow rate change rate is Then the corrected flow rate value is And Among them, is the corrected flow rate value at time t, α is the threshold coefficient, σ r (t-k:t) is the standard deviation of the flow rate change rate in the past k moments, β is the correction coefficient, and sgn(r F (t)) is the sign function. When r F (t)>0, sgn(r F (t)) = 1; when r F (t) < 0, sgn(r F (t)) = -1;

[0011] The water quality data is represented by a vector as where pH w (t) is the acidity and alkalinity, DO w (t) is the dissolved oxygen content, and SS w (t) is the dissolved salt concentration;

[0012] S300. Based on the data sources in the water resource information database, combined with multi-source information of meteorology, soil, and vegetation, construct a water cycle prediction model, and obtain accurate water cycle prediction results through the water cycle prediction model. The water cycle prediction model includes an input layer, a hidden layer, and an output layer;

[0013] S400. According to the prediction results, combined with the water demand priority of vegetation and the available water volume of each water source, use the dynamic water resource allocation algorithm and adjust the irrigation plan in real time;

[0014] S500. Regularly evaluate the sand fixation effect of the photovoltaic field area, and feedback the evaluation results to S300 and S400. Adjust the weights of the water cycle prediction model in S300, and adjust the water demand priority of the vegetation and the water resource irrigation volume in this area in S400 to carry out strategy adjustment and optimization.

[0015] Further, the meteorological information in S300 includes air temperature T a (t), precipitation P(t), and wind speed V w (t);

[0016] The soil information includes soil humidity SH(t) and soil texture parameter ST;

[0017] The vegetation information includes vegetation coverage VC(t) and vegetation growth stage GS(t). When the vegetation is in the germination stage, GS(t)=1, the seedling stage is 2, the growth stage is 3, and the mature stage is 4.

[0018] Furthermore, the input layer of the water cycle prediction model in S300 includes the water level flow rate water quality parameters and the data information of meteorology, soil, and vegetation T a (t), P(t), V w (t), SH(t), ST, VC(t), GS(t), and convert the data information in the water cycle into comprehensive digital features X.

[0019] Furthermore, the hidden layer of the water cycle prediction model in S300 adopts a multi-layer perceptron structure. The multi-layer perceptron consists of m neuron nodes. The nodes are connected by different weights and can automatically learn the non-linear relationship between the input data. For the jth neuron node, 1≤j≤m, it receives the inputs from all nodes in the input layer, performs a weighted summation operation, and calculates the linear combination of the input digital features where n is the number of nodes in the input layer, w ij is the weight connecting the ith node in the input layer and the jth neuron in the hidden layer, x i is the value of the ith node in the input layer digital feature X, and netj is the weighted input sum of the j-th neuron, and for the weighted input sum net j is non-linearly transformed through an activation function, and the activation function a j = max(0, net j ), where a j is the output of the j-th neuron in the hidden layer, and this output will be used as the input for the next layer and the output layer;

[0020] The neurons in each layer perform weighted summation on the output of the previous layer and are non-linearly transformed through the activation function to uncover the potential influence patterns of different factors on each link of the water cycle.

[0021] Furthermore, the output layer of the water cycle prediction model in S300 outputs the predicted value E(t + Δt) of the moisture evaporation rate and the predicted value D(t + Δt) of the soil moisture penetration depth within the time t, where Δt is the prediction step size, and:

[0022] For the output node, weighted summation calculation is performed. Two neurons are set, corresponding to the evaporation rate prediction and the penetration depth prediction respectively. For the neuron of evaporation rate prediction, k = 1, where m is the number of neurons in the hidden layer, and w jk is the weight connecting the j-th neuron in the hidden layer and the k-th neuron in the output layer, a j is the output of the j-th neuron in the hidden layer, and net E is the weighted input sum of the output layer neuron before the activation function processing, and the evaporation rate prediction E(t + Δt) = net E ;

[0023] For the neuron of penetration depth prediction, k = 2, and the penetration depth prediction D(t + Δh) = net D ;

[0024] Similarly, the predicted value SHC(t + Δt) of the soil moisture content in different regions is obtained, and the weights w ij and w jk are adjusted by backpropagation of the error between the predicted value and the true value using gradient descent with an adaptive learning rate to accurately output the prediction results of the moisture evaporation rate and the soil moisture penetration depth.

[0025] Furthermore, the water cycle prediction model optimizes the model weights through gradient descent, and its gradient descent is where and are the weights connecting the i-th input node and the j-th hidden node before and after the update respectively, η(t) is the adaptive learning rate and where η 0 is the initial learning rate, Loss is the loss function and where N is the number of samples, y k is the actual monitored value of the k-th sample, is the model prediction value of the k-th sample. By continuously adjusting the weights, the model can accurately predict the water cycle dynamics based on the input multi-source information, and make the mean square error Loss continuously decrease, that is, the model prediction value is equal to the actual observed value.

[0026] Furthermore, the S400 combines the predicted results, that is, the predicted value E(t+Δt) of the water evaporation rate in the photovoltaic field area and the predicted value D(t+Δt) of the soil water infiltration depth, with the vegetation water demand priority PR i (t) and the available water volume of each water source. The available water volume of the water source is directly reflected by the flow rate F w (t), and uses dynamic water resource allocation to adjust the irrigation plan in real time, and predicts the additional water volume required for each vegetation area due to evaporation and soil water changes: for the i-th vegetation area, the additional water demand SHC(t+Δt) is the predicted value of the change in soil water content in different regions, A i is the area of the i-th vegetation area, α is the evaporation correlation coefficient, β is the soil water change coefficient, m is the number of soil layers, and then combines with the basic water demand of the vegetation to calculate the total water demand

[0027] Furthermore, the S400 calculates the water shortage urgency of each vegetation area where is the irrigation water volume obtained by this area at the previous moment, and the irrigation water volume allocated to the i-th vegetation area γ i is the distribution ratio coefficient, is the total flow rate of all water sources at time t. The irrigation water volume is determined by the water shortage urgency to accurately irrigate the water to each vegetation area in real time, ensuring the vegetation growth demand and the sand fixation effect.

[0028] Compared with the prior art, the sand-fixing method based on water cycle regulation in a sandy desert photovoltaic field area has the following beneficial effects:

[0029] I. The present invention can significantly improve the efficiency of water resource utilization. On the one hand, through the combined monitoring of multiple water sources and the integration of big data platforms, it realizes the all-round, real-time and accurate control of water resources, avoiding water resource waste caused by lagging or inaccurate information. And based on the water cycle prediction model and dynamic allocation, it can rationally allocate irrigation water according to the real-time water demand of vegetation in different regions and the water supply capacity of water sources, ensuring that every drop of water is used effectively. On the other hand, precise irrigation promotes the healthy growth of vegetation, increases the vegetation coverage rate, enhances the water and fertilizer retention capacity of the soil, reduces soil erosion, and thus achieves the purpose of sand fixation.

[0030] II. The water cycle regulation of the present invention based on big data provides suitable water conditions for vegetation at different growth stages and with different tolerances, improves the survival rate of vegetation and species diversity, forms a multi-level and stable vegetation community. The roots of the vegetation penetrate deep into the sand layer, fixing sand and preventing wind, reducing the wind speed of sandstorms, and decreasing the occurrence frequency of sand and dust weather, effectively curbing the process of land desertification. At the same time, the growth of vegetation improves the local microclimate, increases air humidity, regulates soil temperature, creates a good environment for the reproduction of microorganisms, further promotes soil improvement, forms a virtuous ecological cycle, gradually restores the ecological vitality of sandy desert areas, assists the growth of sand-fixing vegetation, and reduces the cost of manual soil loosening.

[0031] Other advantages, objectives and features of the present invention will be described to some extent in the subsequent specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0033] Figure 1 It is an operation diagram of a sand fixation method based on water cycle regulation in a sandy desert photovoltaic field area;

[0034] Figure 2 It is the implementation flowchart of Embodiment II. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention objective, the following will, in combination with the drawings and preferred embodiments, detail the specific implementation manners, structures, features and their effects of the present invention as follows.

[0036] Embodiment I

[0037] As Figure 1 shown, this embodiment focuses on the photovoltaic power generation area in sandy desert, and comprehensively demonstrates the actual operation process of a sand fixation method based on water cycle regulation in the photovoltaic power generation area in sandy desert. This method relies on technical links such as multi-source data collection, big data platform analysis, water cycle model construction, and dynamic resource allocation, aiming to achieve the efficient utilization of water resources and the effective control of wind and sand, and promote the harmonious coexistence of the photovoltaic industry and the ecological environment.

[0038] First, in the rainwater collection pool in the photovoltaic power generation area, high-precision water level sensors are installed at different depths and key positions. These sensors can measure the water level data in real time, denoted as L w (t), where t represents time. The water level data reflects the water storage situation in the rainwater collection pool at different times and is an important basic data for subsequent water resource management. In the groundwater well, a flow velocity-water level integrated monitor is installed vertically to obtain the groundwater level change and flow rate data F w (t). The flow rate data intuitively reflects the recharge or consumption rate of groundwater and is crucial for mastering the water source dynamics. A non-contact ultrasonic flowmeter and a multi-parameter water quality monitor are set at the key cross-section of the river. To ensure the reliability of the water level data, a weighted sliding average filtering algorithm is used for processing, where n is the size of the sliding window, and its value depends on the analysis of the fluctuation characteristics of the water level data and the data smoothing requirements, and is determined by observing and statistically analyzing the historical water level data; i is an index variable used to traverse the data points within the sliding window; L w (i) is the water level measurement value pointed to by the index variable i; v i is the weighting coefficient and satisfies This weighting method makes the data closer to the current moment have a higher weight, aiming to highlight the importance of recent data for the current water level assessment, effectively reduce the fluctuation of the water level data caused by environmental interference, and ensure that the water level data can truly reflect the actual water level state of the rainwater collection pool. For the flow rate data, an adaptive threshold correction algorithm based on the flow rate change rate is used. Let the flow rate change rate be which reflects the relative change of the flow rate at adjacent times and can be used to judge whether there are abnormal fluctuations in the flow rate data. The corrected flow rate value is

[0039] and

[0040] where α is the threshold coefficient used to define whether the flow rate change exceeds the normal range, and σ r(t-k: The standard deviation of the flow rate change rate in the past k moments, where the value of k depends on the fluctuation period and data characteristics of the flow rate data, and is determined by analyzing historical flow rate data. This standard deviation is used to measure the stability of the flow rate change. β is the correction coefficient, and sgn(r F (t)) is the sign function. When r F (t)>0, sgn(r F (t)) = 1, indicating an increase in the flow rate. When r F (t)<0, sgn(r F (t)) = -1, indicating a decrease in the flow rate. Through this algorithm, it is possible to effectively identify and correct abnormal flow rate values caused by equipment failures, sudden rainfall, or other abnormal situations, ensuring that the flow rate data can accurately reflect the continuous replenishment or consumption trend of the water source. In terms of water quality monitoring, a multi-parameter on-line water quality analyzer automatically extracts water samples at certain time intervals and synchronously detects key water quality indicators such as pH w (t), dissolved oxygen (DO w (t)), conductivity, turbidity, etc. These indicators constitute the water quality vector where SS w (t) is the dissolved salt concentration, and the pH w (t) reflects the acidity or alkalinity of the water and directly affects the chemical environment where the roots of vegetation are located; the dissolved oxygen DO w (t) is the dissolved oxygen content, and its content affects the activity of microorganisms in the water, which is then related to the health of the soil ecosystem; the dissolved salt concentration SS w (t) is related to whether vegetation can normally absorb water and avoid salt damage. These water quality parameters together determine the suitability of the water source for irrigation and maintaining the ecosystem. Through continuous monitoring and data processing of the above sensor network, real-time water resource monitoring data is generated, and the data is sent to the data aggregation node at the edge of the field area by means of wireless transmission technology. After receiving the real-time water resource monitoring data from each sensor, the data aggregation node transmits it to the big data platform through the 4G / 5G network.

[0041] On the above-mentioned big data platform, in-depth processing is carried out on the water level data. For the flow rate data, fine processing is carried out according to the adaptive threshold correction algorithm based on the flow rate change rate. When the flow rate change rate r F (t) is calculated, it is necessary to accurately obtain the standard deviation σ r (t-k:t) of the flow rate change rate in the past k moments, and effectively store and statistically analyze the historical flow rate data. The flow rate value is judged and corrected according to the set threshold coefficient α and correction coefficient β to ensure the accuracy and reliability of the flow rate data and provide a solid data basis for subsequent water resource analysis. At the same time, strict quality assessment and outlier processing are carried out on the water quality data. If the dissolved oxygen content DO w(t) deviates from the normal range, and it is necessary to immediately check the calibration of the water quality analyzer and whether the water sample collection process is standardized. If it is determined that the data is abnormal, the historical data is reasonably estimated and corrected to ensure the validity of the water quality data, so as to accurately evaluate the impact of the water source on vegetation growth and the ecological environment. After the above data preprocessing operations, the water level, flow rate, and original water quality data are stored in the big data platform according to the time series, and a comprehensive water resource information database is constructed to provide high-quality data support for the subsequent water cycle prediction model.

[0042] Based on the data sources in the water resource information database described above, a water cycle prediction model is constructed by combining multi-source information of meteorology, soil, and vegetation. The meteorological information is obtained through the automatic weather stations installed in the field area, including the air temperature T a (t), which affects the water evaporation rate; the precipitation P(t), which directly replenishes the water source; the wind speed V w (t), which accelerates water evaporation and sand dust migration and is interrelated with the transpiration of vegetation. The soil information is obtained through soil moisture sensors and soil texture detection equipment. The soil moisture SH(t) is expressed in terms of volumetric water content, with a value range of 0-1, reflecting the actual water content state of the soil and determining the ease of water infiltration and root water absorption of vegetation; the soil texture parameter ST, different textures affect the water retention, air permeability performance of the soil and the root penetration resistance; the vegetation information is obtained by combining UAV aerial photography and ground surveys. The vegetation coverage VC(t) has a value range of 0-1, intuitively showing the coverage ratio of vegetation on the ground surface and being closely related to water interception and transpiration; the vegetation growth stage GS(t), which is divided into different stages such as the germination period, seedling period, growth period, and maturity period. Different stages have significant differences in water demand, from a small amount of water required in the germination period to maintain germination to a large amount of water required in the maturity period to ensure metabolism and reproduction. The water level after the above S200 treatment flow rate water quality parameters and the parameters related to meteorology, soil, and vegetation T a (t), P(t), V w (t), SH(t), ST, VC(t), GS(t), etc. are used as the input layer nodes to construct a water cycle prediction model based on neural network. The input layer nodes convert these data into comprehensive digital features X; the hidden layer adopts a multi-layer perceptron structure, which consists of m neuron nodes. The value of m needs to be determined comprehensively according to factors such as the complexity of the input data, the learning ability of the model, and the computing resources. For the jth neuron node (1≤j≤m), it receives the inputs from all nodes in the input layer and performs an addition and summation operation where n is the number of input layer nodes, w ij is the weight connecting the ith node in the input layer and the jth neuron in the hidden layer, x iis the value of the i-th node in the digital feature X of the input layer, net j is the weighted input sum of the j-th neuron. For the weighted input sum net j through the activation function a j = max(0, net j ) for non-linear transformation, where a j is the output of the j-th neuron in the hidden layer. This output will be used as the input for the next layer and the output layer. Each layer of neurons performs weighted summation on the output of the previous layer and conducts non-linear transformation through the activation function to explore the potential influence patterns of different factors on each link of the water cycle; corresponding neurons are set in the output layer to respectively output the predicted value E(t + Δt) of the moisture evaporation rate, the predicted value D(t + Δt) of the soil moisture penetration depth, and the predicted value SHC(t + Δt) of the soil moisture content in different regions within a certain future time (assuming the prediction step size is Δt) in the photovoltaic field area. For the neurons predicting the evaporation rate, weighted summation calculation is performed where m is the number of neurons in the hidden layer, w jk is the weight connecting the j-th neuron in the hidden layer and the corresponding evaporation rate prediction neuron in the output layer, a j is the output of the j-th neuron in the hidden layer, net E is the weighted input sum of the neuron in the output layer before the activation function processing, and the predicted evaporation rate E(t + Δt) = net E is obtained. For the neurons predicting the penetration depth, similarly and then the predicted penetration depth D(t + Δt) = net D is obtained. During the model training process, the gradient descent algorithm with an adaptive learning rate is used to optimize the model weights. The formula is where and are the weights connecting the i-th input node and the j-th hidden node before and after the update respectively. η(t) is the adaptive learning rate and where η 0 is the initial learning rate, and its value needs to be adjusted according to the convergence situation and data characteristics in the initial stage of model training; γ is the evaluation constant used to stabilize the change process of the learning rate. The loss function where N is the number of samples, y k is the actual monitored value of the k-th sample, is the model prediction value for the k-th sample. It is trained with a large amount of historical data (such as meteorological data, vegetation growing season data, and water source monitoring data over the past few years), and the weights are continuously adjusted to enable the model to accurately predict the water cycle dynamics based on the input multi-source information, and to continuously reduce the mean square error Loss until the model converges to a better prediction performance.

[0043] According to the obtained water cycle prediction results, namely the predicted evaporation rate E(t + Δt) of the photovoltaic field area, the predicted soil water infiltration depth D(t + Δt), and the predicted change in soil water content SHC(t + Δt) in different regions, combined with the vegetation water demand priority PR i (t) and the available water volume of each water source, use the dynamic water resource allocation algorithm to adjust the irrigation plan in real time. Predict the additional water volume required for each vegetation area due to evaporation and soil water changes. For the i-th vegetation area, its area is denoted as A i The evaporation correlation coefficient is α, the soil water change coefficient is β, and the number of soil layers is m. It is divided according to the vertical structure and water movement characteristics of the soil. According to the formula Calculate the additional water demand. In this formula, E(t + Δt) is the predicted evaporation rate, and SHC ij (t + Δt) is the predicted change in the water content of the j-th layer of soil in the i-th vegetation area in the future Δt time, and SH ij (t) is the actual water content of the j-th layer of soil in the i-th vegetation area at present. By calculating the sum of the absolute values of the difference between the two, multiplying by the soil water change coefficient and the area of the region, and adding the additional water demand caused by evaporation (calculated from the evaporation rate, evaporation correlation coefficient, and area of the region), the additional water volume required for this vegetation area due to environmental factor changes is obtained. Then, combined with the basic water demand of the vegetation Calculate the total water demand Calculate the water shortage urgency of each vegetation area where is the irrigation water volume in the previous moment in this region. The water shortage urgency is used to intuitively measure the current water shortage state of the vegetation, and its value range is between 0 and 1. The closer it is to 1, the more serious the water shortage is, and the more urgently irrigation is needed to supplement water. Let the distribution ratio coefficient be γ i , which is determined according to the vegetation water demand priority PR i (t). The vegetation area with a higher water demand priority has a larger distribution ratio coefficient to ensure its water supply. The irrigation water volume allocated to the i-th vegetation area where $Q_t$ is the total flow rate of all water sources at time $t$. By comprehensively considering the water shortage urgency and the distribution ratio coefficient, the total flow rate is reasonably distributed to achieve precise control of the irrigation water volume in each vegetation area, ensuring the efficient use of water resources and the healthy growth of vegetation.

[0044] Regularly evaluate the sand fixation effect in the photovoltaic power generation area, and adjust and optimize the strategy according to the evaluation results. For example, when it is found that the reduction ratio of the sand flow rate in a certain area does not meet the expectation, trace back to check whether the water source monitoring data in this area is accurate, or re-examine the parameter weights related to wind speed in the water cycle prediction model, and adjust the water demand priority or water resource allocation strategy of the vegetation in this area to continuously improve the sand fixation effect.

[0045] In summary, this embodiment details the implementation process of the sand fixation method based on water cycle regulation in the photovoltaic power generation area of sandy desert. The whole process is closely linked, and each link complements each other, effectively solving the water resource management and wind and sand prevention problems faced by the photovoltaic development and ecological protection in sandy desert areas.

[0046] Embodiment 2

[0047] As Figure 2 shown, on the basis of Embodiment 1, this embodiment details the specific implementation process of a sand fixation method based on water cycle regulation in the photovoltaic power generation area of sandy desert:

[0048] First, enter the multi-source data collection initialization stage (S100). Install water level sensors at different heights and corner positions at the bottom and side walls of the rainwater collection pool to ensure accurate monitoring of water level changes. At the same time, set flow sensors near the inlet and outlet to accurately measure the water inflow and outflow. For groundwater wells, install integrated water level and flow sensors at different depths to capture the dynamics of groundwater. In the area where the river or seasonal stream flows through the power generation area, set up comprehensive monitoring stations for flow velocity, water level, and water quality at certain intervals. In the soil, implant soil moisture sensors in layers according to the soil stratification and the distribution of vegetation roots to comprehensively master the vertical distribution of soil moisture content. In addition, monitor meteorological elements such as air temperature, humidity, wind speed, wind direction, and precipitation around the power generation area in real time. Through the collaborative work of these sensors, a comprehensive and multi-level data collection system is constructed to provide rich and accurate data support for subsequent water cycle regulation and sand fixation work.

[0049] Then, enter the database construction stage (S200). Building a powerful, stable, and reliable big data platform is a key link in achieving efficient water cycle regulation. This platform needs to have a high data storage capacity and fast data processing capabilities, and be able to accommodate a large amount of sensor acquisition data, historical meteorological data, soil analysis reports, and vegetation monitoring data, etc. In terms of data storage, an architecture combining a distributed file system and a relational database is adopted to ensure the secure storage and convenient retrieval of data. During the data integration and processing process, data from different sensors and data sources are unified in format and standardized conversion. For example, data such as water level, flow rate, and soil humidity collected by sensors of different brands and models are converted into a unified unit and data format for subsequent analysis. For abnormal data, data cleaning algorithms are used for identification and processing. For example, by setting a reasonable threshold range, it is judged whether the water level data exceeds the normal fluctuation range. If it exceeds, further analysis is carried out to determine whether it is caused by sensor failure or special hydrological events. For flow rate data, combined with meteorological information and historical data, abnormal peaks or troughs caused by heavy rain or equipment failure are identified. In terms of improving data quality, data interpolation and completion techniques are adopted. When some sensors have short-term failures or data transmission interruptions, the missing data are reasonably estimated and supplemented using the historical data trend. Ensure that the data quality in the big data platform is reliable and can truly reflect the ecological environment and water resource status of the photovoltaic field area.

[0050] Subsequently, enter the water cycle feature modeling and prediction stage (S300). Build a water cycle prediction model, design the structure and the number of nodes of the input layer, hidden layer, and output layer. During the model training process, cross-validation and parameter tuning techniques are adopted. The data is divided into a training set, a validation set, and a test set. The training set is used to train the model, and the hyperparameters of the model, such as the learning rate, are adjusted through the validation set to prevent the model from overfitting. For example, during the training of a neural network model, by continuously adjusting the learning rate and observing the change of the loss function on the validation set, the optimal learning rate value is found, so that the model can converge quickly during the training process and maintain good generalization ability. In the model verification stage, the test set data is used to strictly evaluate the model to comprehensively measure the accuracy of the model prediction.

[0051] Next, enter the dynamic water resource optimization and allocation stage (S400). According to the output results of the water cycle prediction model, combined with the water demand characteristics of vegetation and the available water volume of water sources, formulate a dynamic water resource allocation strategy. Classify different vegetation areas in detail, determine the water demand priority according to the vegetation type, growth stage and ecological function. In the aspect of irrigation plan formulation, comprehensively consider the soil moisture condition, meteorological forecast and vegetation water demand curve, use the data of soil moisture sensors to judge the water shortage degree of the soil in different areas, and adjust the irrigation time and irrigation volume in combination with the precipitation probability and evaporation prediction of the meteorological forecast. For example, in the period with a relatively high precipitation probability, appropriately reduce the irrigation volume; in the case of high temperature and drought and low soil moisture content, increase the irrigation frequency and water volume. Ensure the efficient use of water resources and the healthy growth of vegetation, and effectively enhance the sand fixation effect.

[0052] Finally, enter the real-time execution and adaptive feedback stage (S500). Regularly conduct a comprehensive evaluation of the sand fixation effect in the photovoltaic power generation area. The sand fixation effect evaluation indicators cover multiple aspects, including the monitoring of the intensity of sand and wind activities, the analysis of the change in vegetation coverage, and the determination and evaluation of the degree of soil erosion. The intensity of sand and wind activities is monitored by sand and wind monitoring stations installed around and inside the power generation area, and parameters such as the wind speed, wind direction, sediment transport volume and dust concentration of sand and wind are recorded. Compare the data before and after treatment to judge the inhibitory effect of the sand fixation measures on sand and wind activities; the vegetation coverage is regularly monitored by using unmanned aerial vehicle (UAV) aerial photography, calculate the coverage area and the change in coverage rate of different vegetation types, and intuitively reflect the growth and expansion of vegetation; the degree of soil erosion adopts the method of combining soil sampling and topographic survey, analyze the soil particle composition, nutrient loss and topographic and geomorphic changes, and evaluate the role of sand fixation in soil conservation. According to the evaluation results, timely adjust the sand fixation strategy. If it is found that the sand and wind activities in a certain area are still strong, the reason may be that the sand barrier setting is unreasonable or the vegetation coverage is insufficient, and then adjust the sand barrier layout, increase the vegetation planting density or replace the vegetation variety that is more adaptable to the environment. For the aspect of soil moisture management, if the situation that too much or too little soil moisture in some areas affects the growth of vegetation occurs, re-examine the water cycle regulation strategy, optimize the irrigation plan and the water resource allocation plan. Through continuous evaluation and adjustment, continuously improve the sand fixation method based on water cycle regulation, and realize the continuous improvement and stable development of the ecological environment in the photovoltaic power generation area on sandy land and desert.

[0053] In summary, this embodiment aims to provide a sand fixation method based on water cycle regulation. Through the joint monitoring of multiple water sources and big data analysis, realize the scientific allocation of water resources in the photovoltaic power generation area and the refined regulation of the water cycle, improve the water resource utilization efficiency, promote the growth of vegetation, and enhance the sand fixation effect.

[0054] The above are only the preferred embodiments of the present invention, and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above with the preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments of equivalent changes by using the above-disclosed technical content within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A sand fixation method based on water cycle regulation in a photovoltaic field in a sandy desert, characterized in that: The specific steps of this method are: S100: Monitor the water source in the photovoltaic field in real time and use the sensor network to continuously obtain the water level L within time t. w (t), flow rate F w (t) Water quality Data, generate real-time water resources monitoring data; S200, integrating the real-time water resource monitoring data into a big data platform, performing data preprocessing, and forming a comprehensive water resource information database, wherein the preprocessing includes: For the water level L w (t), through Remove noise interference and stabilize the water level data. is the filtered water level value at time t, n is the sliding window size, i is the index variable used to traverse the data points in the sliding window, L w (i) is the water level measurement pointed to by the index variable i, w i is the weighting coefficient and satisfies For flow data, an adaptive threshold correction algorithm based on flow change rate is used to correct flow anomalies caused by equipment failure and sudden rainfall. The flow change rate is The corrected flow value is and in, is the corrected flow value at time t, α is the threshold coefficient, σ r (tk:t) is the standard deviation of the flow rate change rate in the past k moments, β is the correction coefficient, sgn(r F (t)) is a sign function, when r F When (t)>0, sgn(r F (t))=1, when r F (t)<0,sgn(r F (t)) = -1; The water quality data is represented by a vector as pH w (t) is pH, DO w (t) is the dissolved oxygen content, SS w (t) is the dissolved salt concentration; S300, based on the data source in the water resources information database, combined with the multi-source information of meteorology, soil, and vegetation, a water cycle prediction model is constructed, and accurate water cycle prediction results are obtained through the water cycle prediction model, wherein the water cycle prediction model includes an input layer, a hidden layer, and an output layer; S400, according to the prediction results, combined with the vegetation water demand priority and the available water volume of each water source, using a dynamic water resource allocation algorithm, and adjusting the irrigation plan in real time; S500: Regularly evaluate the sand fixation effect of the photovoltaic field area, and feed back the evaluation results to S300 and S400. In S300, adjust the weight of the water cycle prediction model, and in S400, adjust the water demand priority and water resource irrigation amount of the vegetation in the area to adjust and optimize the strategy.

2. The sand fixation method based on water cycle regulation in a photovoltaic field in a sandy desert according to claim 1 is characterized in that: The weather information in S300 includes the temperature T a (t), precipitation P(t) wind speed V w (t); Soil information includes soil moisture SH(t) and soil texture parameters ST; Vegetation information includes vegetation coverage VC(t) and vegetation growth stage GS(t). When the vegetation is in the budding stage, GS(t)=1, in the seedling stage it is 2, in the growth stage it is 3, and in the mature stage it is 4.

3. The sand fixation method based on water cycle regulation in a photovoltaic field in a sandy desert according to claim 2 is characterized in that: The input layer of the water cycle prediction model in S300 includes the pre-processed water level flow Water quality parameters And weather, soil, and vegetation data information a (t), P(t), V w (t), SH(t), ST, VC(t), GS(t), convert the data information in the water cycle into a comprehensive digital feature X.

4. The sand fixation method based on water cycle regulation in a photovoltaic field in a sandy desert according to claim 3 is characterized in that: The hidden layer of the water cycle prediction model in S300 adopts a multi-layer perceptron structure. The multi-layer perceptron consists of m neuron nodes. The nodes are connected by different weights and can automatically learn the nonlinear relationship between input data. For the jth neuron node, 1≤j≤m, it receives input from all nodes in the input layer, performs weighted summation operations, and calculates the linear combination of the input digital features. Where n is the number of input layer nodes, w ij is the weight connecting the i-th node in the input layer to the j-th neuron in the hidden layer, x i is the value of the i-th node in the digital feature X of the input layer, net j is the weighted input sum of the jth neuron, and the weighted input sum net j A nonlinear transformation is performed by an activation function, wherein the activation function a j =max(0,net j ), where a j is the output of the jth neuron in the hidden layer, which will serve as the input of the next layer and the output layer; Each layer of neurons performs a weighted summation of the outputs of the previous layer and performs nonlinear transformation through an activation function to uncover the potential impact patterns of different factors on each link of the water cycle.

5. The sand fixation method based on water cycle regulation in a photovoltaic field in a sandy desert according to claim 4, characterized in that: The output layer of the water cycle prediction model in S300 outputs the predicted value of water evaporation rate E(t+Δt) and predicted value of soil water penetration depth D(t+Δt) of the photovoltaic field within time t, where Δt is the prediction step length, wherein: For the output node, a weighted sum calculation is performed and two neurons are set, corresponding to the evaporation rate prediction and the penetration depth prediction respectively. For the evaporation rate prediction neuron k = 1, Where m is the number of neurons in the hidden layer, w jk is the weight connecting the jth neuron in the hidden layer to the kth neuron in the output layer, a j is the output of the jth neuron in the hidden layer, net E is the sum of the weighted inputs of the output layer neurons before the activation function is processed, and the evaporation rate prediction E(t+Δt)=net E ; For penetration depth prediction, k=2 neurons are used. Get the penetration depth prediction D(t+Δh)=net D ; Similarly, the predicted value SHC(t+Δt) of soil moisture content in different regions is obtained. The weight w is adjusted according to the error back propagation between the predicted value and the true value by using the gradient descent of the adaptive learning rate. ij 、w jk , accurately output the prediction results of water evaporation rate and soil moisture penetration depth.

6. The sand fixation method based on water cycle regulation in a photovoltaic field in a sandy desert according to claim 5, characterized in that: The water cycle prediction model optimizes the model weights by gradient descent, and its gradient descent is in, and are the weights connecting the i-th input node and the j-th hidden node before and after the update, η(t) is the adaptive learning rate and Where η0 is the initial learning rate, Loss is the loss function and Where N is the number of samples, y k is the actual monitoring value of the kth sample, is the model prediction value of the kth sample. By continuously adjusting the weights, the model can accurately predict the water cycle dynamics based on the input multi-source information, and the mean square error Loss is continuously reduced, that is, the model prediction value is equal to the actual observation value.

7. The sand fixation method based on water cycle regulation in a photovoltaic field in a sandy desert according to claim 5, characterized in that: The step S400 is based on the prediction results, i.e., the predicted value of the evaporation rate of the photovoltaic field E(t+Δt), the predicted value of the soil moisture penetration depth D(t+Δt), and the vegetation water demand priority PR. i (t) and the available water volume of each water source, the available water volume of the water source is determined by the flow rate F w (t) Direct reflection, using dynamic water resource allocation to adjust irrigation plans in real time, predicting the additional water required for each vegetation area due to evaporation and soil moisture changes: For the i-th vegetation area, the additional water requirement is the predicted value of soil moisture content change in different regions, A i is the area of ​​the ith vegetation region, α is the evaporation correlation coefficient, β is the soil moisture variation coefficient, m is the number of soil layers, and combined with the basic water demand of vegetation Calculate total water demand 8. The sand fixation method based on water cycle regulation in a photovoltaic field in a sandy desert according to claim 7, characterized in that: The S400 calculates the water shortage urgency of each vegetation area in is the amount of irrigation water that the area has received at the previous moment, and the amount of irrigation water allocated to the i-th vegetation area γ i is the allocation ratio coefficient, The total flow of all water sources at time t is used to determine the irrigation water volume using the water shortage urgency, so that water can be accurately irrigated to each vegetation area in real time to ensure vegetation growth needs and sand fixation effects.

Images