Method for improving saline-alkali land based on fly ash

Through deep learning model and artificial intelligence technology, the correlation relationship between soil improvement ratio scheme and effect image was analyzed, and the best improvement scheme was inferred, which solved the problem of uncertain improvement effect in traditional methods in saline-alkali land, and improved the intelligence of soil improvement and automation level.

CN118504793BActive Publication Date: 2024-11-15RESOURCE & ENVIRONMENT ENG BEIJING
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410963484.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-15
Estimated Expiration
2044-07-18

AI Technical Summary

Technical Problem

Traditional soil improvement methods are difficult to ensure the universality and reproducibility of the improvement effect in saline-alkali land, and rely on experience and small-scale experiments, which consumes time and effort, making it difficult to effectively utilize industrial waste such as fly ash.

Method used

Using deep learning model combined with artificial intelligence technology, by obtaining multiple soil improvement ratio schemes and corresponding soil structure improvement effect images, correlative feature extraction and multi-scale perception enhancement processing are performed, and the features of information excitation-pressing module and prior distribution are used. Dynamic constraint interactive module, infer the best improvement solution.

Benefits of technology

The manual analysis cycle is reduced, the degree of intelligence and automation of soil improvement methods is improved, the sustainable use of saline-alkali land is achieved, and the reliability and universality of the improvement effect is ensured.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118504793B_ABST
    Figure CN118504793B_ABST
Patent Text Reader

Abstract

The present application relates to a method for improving saline-alkali land based on fly ash. The method includes: using an intelligent algorithm to explore the responsive correlation between the soil structure improvement effect image corresponding to different soil improvement ratio schemes and the scheme itself, as well as the dynamic change law between each soil structure improvement effect image, so as to analyze and deduce the best soil improvement ratio scheme. In this way, the best ratio scheme is inferred based on limited experimental data, the cycle of manual analysis is reduced, the intelligence and automation of soil improvement methods are improved, and technical support is provided for the sustainable utilization of saline-alkali land.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of soil improvement, and in particular, to a method for improving saline-alkali land based on fly ash. Background Art

[0002] With the acceleration of industrialization, a large amount of industrial by-products are discharged into the environment, which has a serious impact on soil structure and ecological environment. Salinization is one of the main problems affecting land quality and agricultural production, especially in arid and semi-arid areas. Fly ash, as an abundant industrial waste, is considered to have the potential to improve saline-alkali soil because it contains a variety of beneficial elements and minerals. However, how to effectively use fly ash and other materials to improve saline-alkali soil is a technical challenge.

[0003] The key to soil improvement is to find the right material ratio to promote the optimization of soil structure and improve soil fertility. Traditional soil improvement methods rely on experience and small-scale experiments, which is not only time-consuming and labor-intensive, but also difficult to ensure the universality and replicability of the improvement effect. With the development of artificial intelligence and machine learning technology, the use of deep learning models to analyze and infer the optimal ratio of soil improvement provides a new solution to the above technical problems. Summary of the invention

[0004] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0005] In a first aspect, the present application provides a method for improving saline-alkali land based on fly ash, the method comprising:

[0006] Acquire multiple soil improvement ratio schemes and multiple soil structure improvement effect images corresponding to the multiple soil improvement ratio schemes, wherein each of the soil improvement ratio schemes includes fly ash, desulfurized gypsum and organic fertilizer, and the fly ash includes silicon dioxide, aluminum oxide, calcium oxide, iron oxide, titanium oxide, sulfur oxide, potassium oxide, phosphorus oxide, magnesium oxide and sodium oxide;

[0007] Extracting correlation features from the plurality of soil improvement ratio schemes and the plurality of soil structure improvement effect images to obtain an improvement ratio component correlation feature map and a formula improvement effect correlation feature map;

[0008] Inputting the improved ratio component association feature map and the improved effect association feature map between recipes into an information excitation-squeezing module including a feature multi-scale perception enhancement module and a multi-scale semantic feature extraction module to obtain an improved ratio component association feature vector and an improved effect association feature vector between recipes;

[0009] Inputting the improved ratio component association feature vector and the improved effect association feature vector between the formulas into a feature dynamic constraint interaction module based on prior distribution to obtain an improved formula-improved effect interaction feature vector;

[0010] The best improvement scheme is determined based on the improvement formula-improvement effect interaction feature vector.

[0011] Optionally, correlation features are extracted from the multiple soil improvement ratio schemes and the multiple soil structure improvement effect images to obtain an improvement ratio component correlation feature map and a formula improvement effect correlation feature map, including: arranging the multiple soil improvement ratio schemes into a full-sample soil improvement ratio input matrix according to the formula sample dimension; arranging the multiple soil structure improvement effect images into a soil structure improvement effect input tensor according to the image sample dimension; inputting the full-sample soil improvement ratio input matrix into an improvement ratio component correlation feature extractor based on a void convolutional neural network model to obtain the improvement ratio component correlation feature map; inputting the soil structure improvement effect input tensor into an improvement effect correlation feature extractor based on a three-dimensional convolutional neural network model to obtain the improvement effect correlation feature map.

[0012] Optionally, the improved ratio component association feature map and the improved effect association feature map between recipes are input into an information excitation-squeezing module including a feature multi-scale perception enhancement module and a multi-scale semantic feature extraction module to obtain an improved ratio component association feature vector and an improved effect association feature vector between recipes, including: inputting the improved ratio component association feature map and the improved effect association feature map between recipes into the feature multi-scale perception enhancement module of the information excitation-squeezing module to obtain a multi-scale perception-enhanced improved ratio component association feature map and a multi-scale perception-enhanced improved effect association feature map between recipes; inputting the multi-scale perception-enhanced improved ratio component association feature map and the multi-scale perception-enhanced improved effect association feature map between recipes into the multi-scale semantic feature extraction module of the information excitation-squeezing module to obtain the improved ratio component association feature vector and the improved effect association feature vector between recipes.

[0013] Optionally, the improved ratio component association feature map and the improved effect association feature map between recipes are input into the feature multi-scale perception enhancement module of the information excitation-squeezing module to obtain a multi-scale perception enhanced improved ratio component association feature map and a multi-scale perception enhanced improved effect association feature map between recipes, including: inputting the improved ratio component association feature map into the low-frequency detection branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association local low-frequency activation feature vector; inputting the improved ratio component association feature map into the simplified compression branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association compression feature map; inputting the improved ratio component association feature map into the global space perception branch of the feature multi-scale perception enhancement module to obtain an improved The improved ratio component association receptive field expansion activation feature matrix is ​​obtained; the improved ratio component association receptive field expansion activation feature matrix is ​​point-multiplied by position with each feature matrix of the improved ratio component association compression feature map along the channel dimension to obtain an improved ratio component association compression-expansion activation feature map; each eigenvalue in the improved ratio component association local low-frequency activation feature vector is used as a weight, and each feature matrix of the improved ratio component association compression feature map along the channel dimension is point-multiplied to obtain an improved ratio component association compression-low-frequency fusion feature map; the improved ratio component association compression-expansion activation feature map and the improved ratio component association compression-low-frequency fusion feature map are added by position to obtain an improved ratio component association multi-scale fusion activation feature map; the improved ratio component association multi-scale fusion activation feature map is subjected to hole convolution coding to obtain the multi-scale perception enhancement improved ratio component association feature map.

[0014] Optionally, the improved ratio component association feature map is input into the low-frequency detection branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association local low-frequency activation feature vector, including: performing point convolution processing on the improved ratio component association feature map to obtain an improved ratio component association channel compression feature map; performing global mean pooling on the improved ratio component association channel compression feature map to obtain an improved ratio component association local low-frequency feature vector; performing nonlinear activation on the improved ratio component association local low-frequency feature vector to obtain the improved ratio component association local low-frequency activation feature vector.

[0015] Optionally, the improved ratio component association feature map is input into the simplified compression branch of the feature multi-scale perception enhancement module to obtain the improved ratio component association compression feature map, including: performing point convolution processing on the improved ratio component association feature map to obtain the improved ratio component association compression feature map.

[0016] Optionally, the improved ratio component association feature map is input into the global space perception branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association receptive field expansion activation feature matrix, including: performing hole convolution encoding on the improved ratio component association feature map to obtain an improved ratio component association receptive field expansion feature map; performing point convolution processing on the improved ratio component association receptive field expansion feature map to obtain an improved ratio component association receptive field expansion global feature matrix; performing nonlinear activation on the improved ratio component association receptive field expansion global feature matrix to obtain the improved ratio component association receptive field expansion activation feature matrix.

[0017] Optionally, the multi-scale perceptual enhancement improved ratio component association feature map and the multi-scale perceptual enhancement improved effect association feature map between recipes are input into the multi-scale semantic feature extraction module of the information excitation-squeezing module to obtain the improved ratio component association feature vector and the improved effect association feature vector between recipes, including: upsampling the multi-scale perceptual enhancement improved ratio component association feature map to obtain an upsampled multi-scale perceptual enhancement improved ratio component association feature map; using the improved ratio component association global semantic feature extraction module to process the upsampled multi-scale perceptual enhancement improved ratio component association feature map to obtain an improved ratio component association global semantic extraction feature vector; dimensionally adjusting the improved ratio component association global semantic extraction feature vector to obtain an improved ratio component association global semantic feature vector; using the improved ratio component association local semantic feature extraction module to process the upsampled multi-scale perceptual enhancement improved ratio component association feature map to obtain an improved ratio component association global semantic extraction feature vector. The good ratio component association feature map is processed to obtain an improved ratio component association local semantic extraction feature vector; the improved ratio component association local semantic extraction feature vector is dimensionally adjusted to obtain an improved ratio component association local semantic feature vector; the improved ratio component association global semantic feature vector and the improved ratio component association local semantic feature vector are added element by element by position to obtain an improved ratio component association multi-scale semantic fusion feature vector; the improved ratio component association multi-scale semantic fusion feature vector is activated based on the sigmoid function to obtain the improved ratio component association feature vector; wherein the improved ratio component association global semantic feature extraction module includes a global average pooling layer, a first point convolution layer, a first batch normalization layer and a first activation layer, and the improved ratio component association local semantic feature extraction module includes a second point convolution layer, a second batch normalization layer and a second activation layer.

[0018] Optionally, the improved ratio component association feature vector and the improved effect association feature vector between formulas are input into a feature dynamic constraint interaction module based on prior distribution to obtain an improved formula-improved effect interaction feature vector, including: calculating an exponential function value with a natural constant as the base and a first weight coefficient as the dot product value of each eigenvalue in the improved ratio component association feature vector as the exponent to obtain a nonlinear improved ratio component association feature vector; calculating an exponential function value with a natural constant as the base and a second weight coefficient as the dot product value of each eigenvalue in the improved effect association feature vector between formulas as the exponent to obtain a nonlinear improved effect association feature vector between formulas; calculating the dot product of the first Gaussian distribution function coefficient and the nonlinear improved ratio component association feature vector and the dot product of the second Gaussian distribution function coefficient and the nonlinear improved effect association feature vector between formulas, and adding them element by element positionally to obtain the improved formula-improved effect interaction feature vector.

[0019] Optionally, determining the inference best improvement plan based on the improvement formula-improvement effect interaction feature vector includes: inputting the improvement formula-improvement effect interaction feature vector into a decoder-based best improvement plan reasoner to obtain the inference best improvement plan.

[0020] By adopting the above technical scheme, the responsiveness relationship between the soil structure improvement effect images corresponding to different soil improvement ratio schemes and the schemes themselves is explored through the use of intelligent algorithms, as well as the dynamic change rules between the various soil structure improvement effect images, so as to analyze and deduce the best soil improvement ratio scheme. In this way, the best ratio scheme is inferred based on limited experimental data, the cycle of manual analysis is reduced, the intelligence and automation of soil improvement methods are improved, and technical support is provided for the sustainable utilization of saline-alkali land.

[0021] Other features and advantages of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and other features, advantages and aspects of the embodiments of the present application will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale. In the drawings:

[0023] Figure 1 The present invention is a flow chart showing a method for improving saline-alkali land based on fly ash according to an exemplary embodiment.

[0024] Figure 2 It is a block diagram of a system for improving saline-alkali land based on fly ash according to an exemplary embodiment.

[0025] Figure 3 It is a block diagram of an electronic device according to an exemplary embodiment.

[0026] Figure 4 It is an application scenario diagram of a method for improving saline-alkali land based on fly ash according to an exemplary embodiment. DETAILED DESCRIPTION

[0027] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not intended to limit the scope of protection of the present application.

[0028] It should be understood that the various steps described in the method implementation of the present application can be performed in different orders and / or performed in parallel. In addition, the method implementation may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.

[0029] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0030] It should be noted that the concepts such as "first" and "second" mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0031] It should be noted that the modifications of "one" and "plurality" mentioned in the present application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0032] The names of the messages or information exchanged between multiple devices in the embodiments of the present application are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0033] Fly ash, a major solid waste from coal-fired power plants, has good permeability and adsorption, which can promote the uniform distribution of water and nutrients in the soil and improve the permeability of the soil. Fly ash contains a large amount of active ingredients such as silicates and aluminates, which can promote the formation of aggregate structure and the improvement of soil structure, and improve the water retention capacity and nutrient content of the soil. Using new technologies to use fly ash for the improvement of saline-alkali land and turning waste into treasure can realize the resource utilization of fly ash, the improvement of saline-alkali land and the development of land resources, and synergistically solve the outstanding problems that restrict the high-quality development of Yancheng. It is of great significance to Yancheng's construction of a green and low-carbon development demonstration zone and the increase of land and reduction of emissions.

[0034] The specific implementation of the present application is described in detail below with reference to the accompanying drawings.

[0035] Figure 1 is a flow chart of a method for improving saline-alkali land based on fly ash according to an exemplary embodiment. Figure 1 As shown, the method includes:

[0036] Step 101, obtaining a plurality of soil improvement ratio schemes and a plurality of soil structure improvement effect images corresponding to the plurality of soil improvement ratio schemes, wherein each of the soil improvement ratio schemes includes fly ash, desulfurized gypsum and organic fertilizer, and the fly ash includes silicon dioxide, aluminum oxide, calcium oxide, iron oxide, titanium oxide, sulfur oxide, potassium oxide, phosphorus oxide, magnesium oxide and sodium oxide;

[0037] Step 102: extracting correlation features from the plurality of soil improvement ratio schemes and the plurality of soil structure improvement effect images to obtain an improvement ratio component correlation feature graph and a formula improvement effect correlation feature graph;

[0038] Step 103, inputting the improved ratio component association feature map and the improved effect association feature map between recipes into an information excitation-squeezing module including a feature multi-scale perception enhancement module and a multi-scale semantic feature extraction module to obtain an improved ratio component association feature vector and an improved effect association feature vector between recipes;

[0039] Step 104, inputting the improved ratio component association feature vector and the improved effect association feature vector between the formulas into a feature dynamic constraint interaction module based on prior distribution to obtain an improved formula-improvement effect interaction feature vector;

[0040] Step 105: Determine and infer the best improvement plan based on the improvement formula-improvement effect interaction feature vector.

[0041] The characteristics of saline-alkali soil are: the salt composition is mainly chloride, but it varies with the degree of desalination. Light saline soil is mainly bicarbonate and chloride, and medium and heavy saline soil is mainly chloride and sulfate. The salt distribution in the soil profile is "T" shaped, and the highest content is in the upper 2-20 cm.

[0042] Since organic matter is not easy to accumulate in saline-alkali soil, the content of soluble salt (such as sodium chloride, magnesium chloride, and calcium chloride) is high. On the one hand, it affects the survival and growth of crops; on the other hand, salt has a direct destructive effect on the physical and chemical properties of the soil. Salt ions are easily hydrated when they come into contact with water, highly disperse soil particles, destroy soil structure, cause soil compaction and nutrient loss, affect soil water penetration, aggravate salt return, and are not conducive to root development.

[0043] Coastal saline-alkali land is mainly distributed in coastal tidal flats and river deltas. It is chloride saline soil with the following characteristics:

[0044] (1) Due to the influence of monsoons and tides and the constraints of soil formation, the salt content of soil changes regularly with the distance from the sea. The farther away from the sea, the lighter the salt content, and the closer to the sea, the heavier the salt content.

[0045] (2) The terrain is low and flat, the groundwater level and mineralization are high, and the receiving area is affected by the sea tide, so the groundwater outflow is often difficult.

[0046] (3) The terrain is low-lying and the drainage area is easily affected by sea tides or floods. It is difficult to drain the area completely by gravity. It is necessary to build flood control, moisture prevention and mechanical drainage projects to avoid flood disasters.

[0047] (4) The soil is poor and low in organic matter and phosphorus, and needs to be fertilized and improved for crop production.

[0048] In one embodiment of the present application, saline-alkali soils from different plots and depths in a designated area of ​​Yancheng were sampled, and the pH value and total salt content data of the soils are shown in Table 1.

[0049] Table 1 Results of pH value and total salt content of saline-alkali soil

[0050]

[0051] As can be seen from Table 1, the total salt content of the surface soil of Plot 2 is as high as 8.86 g / kg, which is a severe saline-alkali soil, and its organic matter content is at the normal farmland level, at 1.39%. The total salt content and pH of other soil layers are both moderate saline-alkali land.

[0052] As an abundant industrial waste, fly ash is considered to have the potential to improve saline-alkali soil because it contains a variety of beneficial elements and minerals. The key to soil improvement is to find a suitable material ratio to promote the optimization of soil structure and improve soil fertility. Traditional soil improvement methods rely on experience and small-scale experiments, which is not only time-consuming and labor-intensive, but also difficult to ensure the universality and replicability of the improvement effect. With the development of artificial intelligence and machine learning technology, the use of deep learning models to analyze and infer the optimal ratio of soil improvement provides a new solution to the above technical problems.

[0053] Fly ash is used to improve saline-alkali land because of its physical properties. After fly ash is applied to the soil, it can significantly improve the soil structure, reduce bulk density, increase porosity, increase ground temperature, and reduce expansion rate, especially for the physical properties of clay soil. In addition, it can also keep warm and protect soil, promote nutrient conversion, and make water, fertilizer, air, and heat tend to be coordinated. The amount of fly ash applied should be controlled within a certain range.

[0054] Among them, 2%, 3% (fly ash) + 0.5% (desulfurized gypsum) + 1% (organic fertilizer) are the best improved ratios, which can effectively reduce the salt content by 70%.

[0055] In one embodiment of the present application, an X-ray fluorescence spectrometer (XRF) was used to analyze the chemical composition of two fly ash samples, and the results are shown in Table 2. The chemical compositions of the two fly ash samples are similar, and their main components are SiO2 and Al2O3, and the aluminum-silicon ratio is 0.68. In addition, they also contain a certain amount of CaO, Fe2O3, TiO2, SO3, etc.

[0056] Table 2 Main chemical components of fly ash

[0057] composition SiO2 Al2O3 CaO Fe2O3 TiO2 SO3 K2O P2O5 MgO Na2O No. 1 ash (wt%) 50.79 34.31 5.47 3.85 2.16 1.38 0.65 0.35 0.27 0.25 No. 2 ash (wt%) 49.83 33.85 6.30 4.33 2.38 1.41 0.71 0.35 0.29 0.25

[0058] The treated fly ash samples were digested by microwave digestion instrument, and the contents of 8 typical heavy metal elements (cadmium, mercury, arsenic, lead, chromium, copper, nickel, and zinc) were tested by inductively coupled plasma mass spectrometry (ICP-Ms). The results are shown in Table 3, which meet the requirements of the national standard GB15618-2018 "Soil Environmental Quality Agricultural Land Soil Pollution Risk Control Standard (Trial)" for agricultural land soil pollution risk screening values. By comparing the data of pH>7.5 in Table 3, it can be seen that the contents of cadmium, mercury, arsenic, lead, chromium, copper, nickel, and zinc in the two treated fly ash samples are all lower than the lowest value of the agricultural land risk screening value in the national standard GB15618-2018 "Soil Environmental Quality Agricultural Land Soil Pollution Risk Control Standard (Trial)", and can be used to prepare new materials for saline-alkali land improvement.

[0059] Table 3 Contents of 8 typical heavy metal elements in fly ash

[0060] Heavy Metal Elements cadmium mercury arsenic lead chromium copper nickel Zinc No. 1 ash content (ppm) 0.23 0.18 2.62 97.45 151.97 42.11 56.14 76.82 No. 2 ash content (ppm) 0.21 0.09 11.26 58.17 132.47 31.52 54.50 27.47

[0061] Desulfurized gypsum improves saline-alkali land mainly based on its chemical improvement mechanism. In the soil, clay with smaller particle size forms soil colloids with humus. When soil colloids come into contact with Na2CO3, NaHCO3, NaCl, etc. in saline-alkali soil, soil colloids containing Na+ are formed. Soil colloids containing Na+ have good dispersibility in soil solution and can be dispersed in soil pores to form an impermeable soil layer. Therefore, the permeability of saline-alkali soil is poor, and it is easy to form cracks in the soil layer during water loss, which is not conducive to plant growth. It is difficult to remove Na+ in the soil by traditional irrigation. After desulfurized gypsum is applied to saline-alkali land, it can form Ca2+ after being dissolved in water in the soil. Ca2+ can replace Na+ on soil colloids, is not easy to adsorb water molecules, and can form micro-clusters between colloidal particles. When water molecules penetrate between particles, the micro-clusters will swell, and when water is lost, the micro-clusters will crack. The repetition of this process makes the soil loose and more permeable, which is beneficial to the expansion of plant roots and the ability to absorb water and fertilizer, thereby achieving the purpose of improving saline-alkali soil. The most effective components in fly ash are SiO2 and Al2O3, and it also contains FeO, Fe2O3, CaO, TiO2, MgO, K2O, Na2O, SO3, MnO2, and P2O5.

[0062] Based on this, the technical concept of this application is: using intelligent algorithms to explore the responsive correlation between the soil structure improvement effect images corresponding to different soil improvement ratio schemes and the schemes themselves, as well as the dynamic change rules between the various soil structure improvement effect images, so as to analyze and deduce the best soil improvement ratio scheme. In this way, the best ratio scheme is inferred based on limited experimental data, the cycle of manual analysis is reduced, the intelligence and automation of soil improvement methods are improved, and technical support is provided for the sustainable utilization of saline-alkali land.

[0063] Specifically, in the technical solution of the present application, first, a plurality of soil improvement ratio schemes and a plurality of soil structure improvement effect images corresponding to the plurality of soil improvement ratio schemes are obtained, wherein each of the soil improvement ratio schemes includes fly ash, desulfurized gypsum and organic fertilizer, and the fly ash includes silicon dioxide, aluminum oxide, calcium oxide, iron oxide, titanium oxide, sulfur oxide, potassium oxide, phosphorus oxide, magnesium oxide and sodium oxide. It should be understood that different soil improvement ratio schemes may have different effects on soil structure and fertility. By collecting a variety of ratio schemes, the effects of combinations of different components and proportions on soil improvement effects can be analyzed. That is, by comparing the relationship between different ratio schemes and soil improvement effects, the correlation between them can be studied, thereby inferring a more effective improvement formula. More specifically, in the soil improvement ratio scheme, fly ash provides a rich source of minerals for the soil, helps to increase the porosity of the soil and improve the soil structure; desulfurized gypsum has the ability to neutralize alkaline soil, helps to lower the pH value of saline-alkali soil and improve the acid-base balance of the soil; and organic fertilizers such as compost, green manure, seaweed fertilizer, animal manure and cake fertilizer can provide the organic matter needed by the soil, promote microbial activity and increase the biological activity of the soil.

[0064] In one embodiment of the present application, correlation feature extraction is performed on the multiple soil improvement ratio schemes and the multiple soil structure improvement effect images to obtain an improvement ratio component correlation feature map and a formula improvement effect correlation feature map, including: arranging the multiple soil improvement ratio schemes into a full-sample soil improvement ratio input matrix according to the formula sample dimension; arranging the multiple soil structure improvement effect images into a soil structure improvement effect input tensor according to the image sample dimension; inputting the full-sample soil improvement ratio input matrix into an improvement ratio component correlation feature extractor based on a void convolutional neural network model to obtain the improvement ratio component correlation feature map; inputting the soil structure improvement effect input tensor into an improvement effect correlation feature extractor between formulas based on a three-dimensional convolutional neural network model to obtain the improvement effect correlation feature map between formulas.

[0065] Then, the multiple soil improvement ratio schemes are arranged according to the formula sample dimension into a full sample soil improvement ratio input matrix; at the same time, the multiple soil structure improvement effect images are arranged according to the image sample dimension into a soil structure improvement effect input tensor. That is, the discretely distributed soil improvement ratio schemes and soil structure improvement effect images are integrated into structured data by arrangement, so as to facilitate input into the model for analysis.

[0066] Next, the full sample soil improvement ratio input matrix is ​​input into the improvement ratio component association feature extractor based on the hole convolution neural network model to obtain the improvement ratio component association feature map; at the same time, the soil structure improvement effect input tensor is input into the formula improvement effect association feature extractor based on the three-dimensional convolution neural network model to obtain the formula improvement effect association feature map. Among them, compared with the traditional convolution processing, the hole convolution (also called dilated convolution) can capture a wider range of dependencies in the data, which is very useful when processing data with complex structures. In the soil improvement ratio scheme, the proportion of different components may have a complex effect on the soil structure, and the hole convolution can help the model learn these complex interactions. The three-dimensional convolution neural network is suitable for processing data with three dimensions. Generally speaking, it can be used to process data with a time dimension (the third dimension). In the technical solution of the present application, the third dimension is the sample dimension, that is, each soil improvement ratio scheme. In this way, the change law and mutual correlation between the soil structure improvement effect images under different samples can be captured.

[0067] In one embodiment of the present application, the improved ratio component association feature map and the improved effect association feature map between recipes are input into an information excitation-squeezing module including a feature multi-scale perception enhancement module and a multi-scale semantic feature extraction module to obtain an improved ratio component association feature vector and an improved effect association feature vector between recipes, including: inputting the improved ratio component association feature map and the improved effect association feature map between recipes into the feature multi-scale perception enhancement module of the information excitation-squeezing module to obtain a multi-scale perception-enhanced improved ratio component association feature map and a multi-scale perception-enhanced improved effect association feature map between recipes; inputting the multi-scale perception-enhanced improved ratio component association feature map and the multi-scale perception-enhanced improved effect association feature map between recipes into the multi-scale semantic feature extraction module of the information excitation-squeezing module to obtain the improved ratio component association feature vector and the improved effect association feature vector between recipes.

[0068] Considering that there may be complex and diverse feature distribution patterns in the full-sample soil improvement ratio input matrix (aggregate form of each soil improvement ratio scheme) and the soil structure improvement effect input tensor (aggregate form of each soil structure improvement effect image). Specifically, soil improvement materials such as fly ash, desulfurized gypsum and organic fertilizer each contain a variety of chemical components, which interact with each other in different proportions and may produce nonlinear and complex chemical and physical reactions. In the soil structure improvement effect image, since the soil structure may be significantly different in different spatial positions, this causes the improvement effect to show different characteristics in different image regions. Although the void convolution can expand the receptive field of feature extraction, and the three-dimensional convolution can capture the correlation and mutual influence of the soil improvement effects between different samples, the single perception ability is obviously insufficient for learning and capturing complex and diverse feature distribution patterns. Therefore, in the technical solution of the present application, it is expected that the improved ratio component association feature map and the improved effect association feature map between formulas will be input into an information excitation-squeezing module including a feature multi-scale perception enhancement module and a multi-scale semantic feature extraction module to more comprehensively understand the complex associations between various ratio schemes and various soil improvement effects based on a dual multi-scale feature perception mechanism, thereby obtaining an improved ratio component association feature vector and an improved effect association feature vector between formulas.

[0069] In an embodiment of the present application, the specific process of inputting the improved ratio component association feature map into the feature multi-scale perception enhancement module includes: first inputting the improved ratio component association feature map into the low-frequency detection branch of the feature multi-scale perception enhancement module to obtain the improved ratio component association local low-frequency activation feature vector; and inputting the improved ratio component association feature map into the simplified compression branch of the feature multi-scale perception enhancement module to obtain the improved ratio component association compression feature map; at the same time, inputting the improved ratio component association feature map into the global space perception branch of the feature multi-scale perception enhancement module to obtain the improved ratio component association receptive field expansion activation feature matrix. Here, the feature multi-scale perception enhancement module uses the low-frequency detection branch to process and learn the local detail information in the improved ratio component association feature map; and uses the simplified compression branch to reduce the number of channels of feature distribution to reduce the overhead of computing resources; at the same time, the global space perception branch is used to capture the global overall distribution pattern of the improved ratio component association feature. More specifically, the global space perception branch uses the hole convolution coding method to expand the perception range of the original feature. Subsequently, the improved ratio component association receptive field expansion activation feature matrix is ​​point-multiplied by position with each feature matrix of the improved ratio component association compression feature map along the channel dimension to obtain an improved ratio component association compression-expansion activation feature map; after that, each eigenvalue in the improved ratio component association local low-frequency activation feature vector is used as a weight to perform point multiplication on each feature matrix of the improved ratio component association compression feature map along the channel dimension to obtain an improved ratio component association compression-low-frequency fusion feature map; then, the improved ratio component association compression-expansion activation feature map and the improved ratio component association compression-low-frequency fusion feature map are added by position to obtain an improved ratio component association multi-scale fusion activation feature map; finally, the improved ratio component association multi-scale fusion activation feature map is subjected to hole convolution coding to obtain the multi-scale perception-enhanced improved ratio component association feature map. That is, after the feature distribution of the simplified and compressed branch is superimposed with the features of the other two branches, the perception range is expanded again by the hole convolution set at the tail, and finally the integrated features of the local details and the overall distribution features are enhanced to form a multi-scale feature representation of the association relationship of the ratio scheme. Among them, the process of inputting the inter-formula improvement effect association feature map into the feature multi-scale perception enhancement module of the information excitation-squeezing module to obtain the multi-scale perception enhancement inter-formula improvement effect association feature map is similar to the process of obtaining the multi-scale perception enhancement improved ratio component association feature map mentioned above, which will not be repeated here.

[0070] In one embodiment of the present application, the improved ratio component association feature map is input into the low-frequency detection branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association local low-frequency activation feature vector, including: performing point convolution processing on the improved ratio component association feature map to obtain an improved ratio component association channel compression feature map; performing global mean pooling on the improved ratio component association channel compression feature map to obtain an improved ratio component association local low-frequency feature vector; performing nonlinear activation on the improved ratio component association local low-frequency feature vector to obtain the improved ratio component association local low-frequency activation feature vector.

[0071] Furthermore, in one embodiment of the present application, the improved ratio component association feature map is input into the simplified compression branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association compression feature map, including: performing point convolution processing on the improved ratio component association feature map to obtain the improved ratio component association compression feature map.

[0072] Furthermore, in one embodiment of the present application, the improved ratio component association feature map is input into the global space perception branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association receptive field expansion activation feature matrix, including: performing hole convolution encoding on the improved ratio component association feature map to obtain an improved ratio component association receptive field expansion feature map; performing point convolution processing on the improved ratio component association receptive field expansion feature map to obtain an improved ratio component association receptive field expansion global feature matrix; performing nonlinear activation on the improved ratio component association receptive field expansion global feature matrix to obtain the improved ratio component association receptive field expansion activation feature matrix.

[0073] Furthermore, the multi-scale perceptual enhancement improved ratio component association feature map and the multi-scale perceptual enhancement improved effect association feature map between recipes are input into the multi-scale semantic feature extraction module of the information excitation-squeezing module to obtain the improved ratio component association feature vector and the improved effect association feature vector between recipes, including: upsampling the multi-scale perceptual enhancement improved ratio component association feature map to obtain an upsampled multi-scale perceptual enhancement improved ratio component association feature map; using the improved ratio component association global semantic feature extraction module to process the upsampled multi-scale perceptual enhancement improved ratio component association feature map to obtain an improved ratio component association global semantic extraction feature vector; dimensionally adjusting the improved ratio component association global semantic extraction feature vector to obtain an improved ratio component association global semantic feature vector; using the improved ratio component association local semantic feature extraction module to process the upsampled multi-scale perceptual enhancement improved ratio component association feature map to obtain an improved ratio component association global semantic extraction feature vector. The improved ratio component association feature map is processed to obtain an improved ratio component association local semantic extraction feature vector; the improved ratio component association local semantic extraction feature vector is dimensionally adjusted to obtain an improved ratio component association local semantic feature vector; the improved ratio component association global semantic feature vector and the improved ratio component association local semantic feature vector are added element by element by position to obtain an improved ratio component association multi-scale semantic fusion feature vector; the improved ratio component association multi-scale semantic fusion feature vector is activated based on the sigmoid function to obtain the improved ratio component association feature vector; wherein the improved ratio component association global semantic feature extraction module includes a global average pooling layer, a first point convolution layer, a first batch normalization layer and a first activation layer, and the improved ratio component association local semantic feature extraction module includes a second point convolution layer, a second batch normalization layer and a second activation layer.

[0074] Specifically, the improved ratio component association feature map and the improved effect association feature map between recipes are input into the feature multi-scale perception enhancement module of the information excitation-squeezing module using the following multi-scale formula to obtain a multi-scale perception enhanced improved ratio component association feature map and a multi-scale perception enhanced improved effect association feature map between recipes; wherein the multi-scale formula is:

[0075]

[0076] in, It is a modified ratio component correlation characteristic diagram. represents 1×1 point convolution, and They represent 3×3 dilated convolutions with dilation rates of 2 and 3, respectively. represents the sigmoid activation function, represents global mean pooling, and denote dot product and positional addition respectively, It is a multi-scale perceptual enhancement improved ratio component association feature map.

[0077] In an embodiment of the present application, further, the multi-scale perceptual enhancement improved ratio component association feature map is input into the multi-scale semantic feature extraction module of the information excitation-squeezing module to focus on the multi-scale features, thereby obtaining the core mechanism of the improved ratio component association feature vector: the multi-scale semantic feature extraction module designs an improved ratio component association global semantic feature extraction module and an improved ratio component association local semantic feature extraction module to respectively capture the global perceptual association distribution pattern and the local perceptual association distribution pattern in the multi-scale perceptual enhancement improved ratio component association feature map. In particular, the improved ratio component association global semantic feature extraction module has a global average pooling layer, which can ignore the channel specificity in the feature map to a certain extent, thereby obtaining the global distribution feature pattern. Similarly, the process and core mechanism of obtaining the improved effect association feature vector between the recipes are also the same, which will not be elaborated here.

[0078] Furthermore, the improved ratio component association feature vector and the improved effect association feature vector between the formulas are input into the feature dynamic constraint interaction module based on the prior distribution to optimize the interactive fusion process using the prior distribution as guidance information, thereby obtaining the improved formula-improvement effect interactive feature vector. It should be understood that there may be complex interactive effects between the soil improvement ratio components and the improvement effects. Through the guidance of the prior distribution, the feature selection and integration process can be optimized to identify and quantify the interactive effects and complex uncertainties between the two.

[0079] In an embodiment of the present application, the specific process of inputting the improved ratio component association feature vector and the improved effect association feature vector between the formulas into the feature dynamic constraint interaction module based on the prior distribution to obtain the improved formula-improved effect interaction feature vector includes: first, calculating the exponential function value with the natural constant as the base and the first weight coefficient and the point multiplication value of each eigenvalue in the improved ratio component association feature vector as the exponent to obtain the nonlinear improved ratio component association feature vector; at the same time, calculating the exponential function value with the natural constant as the base and the second weight coefficient and the point multiplication value of each eigenvalue in the improved effect association feature vector between the formulas as the exponent to obtain the nonlinear improved effect association feature vector between the formulas. Here, by calculating the exponential function value, the nonlinear response relationship is introduced into the original vector representation to reveal the complex pattern in the data. Then, the point multiplication of the first Gaussian distribution function coefficient and the nonlinear improved ratio component association feature vector and the point multiplication of the second Gaussian distribution function coefficient and the nonlinear improved effect association feature vector between the formulas are calculated and added element by element to obtain the improved formula-improved effect interaction feature vector. Among them, the introduction of the first Gaussian distribution function coefficient and the second Gaussian distribution function coefficient actually combines the prior distribution information with the characteristic pattern associated with the improved ratio components and the characteristic pattern associated with the improvement effects between formulas, so as to achieve a more reasonable prediction of the soil improvement effect; at the same time, the introduction of Gaussian distribution provides the model with a natural way to deal with the uncertainty and changes in the data, enhancing the robustness of the model.

[0080] Specifically, the improved ratio component association feature vector and the improved effect association feature vector between the formulas are input into the feature dynamic constraint interaction module based on prior distribution to obtain the improved formula-improvement effect interaction feature vector according to the following constraint formula; wherein the constraint formula is:

[0081]

[0082] in, and are the improved ratio component association feature vector and the improved effect association feature vector between the formulas, and are the first weight coefficient and the second weight coefficient, respectively. and denote the first Gaussian distribution function coefficient and the second Gaussian distribution function coefficient, respectively. represents an exponential function with a natural constant as base, It is the improved formula-improvement effect interaction feature vector.

[0083] Then, the improved formula-improvement effect interaction feature vector is input into the best improvement solution reasoner based on the decoder to obtain the inferred best improvement solution. That is, the mechanism of the decoder is used to predict and infer the best improvement solution according to the responsive complex correlation distribution pattern between the formula and the effect expressed by the improved formula-improvement effect interaction feature vector, so as to provide effective technical support for decision makers.

[0084] In one embodiment of the present application, determining the best improvement plan for inference based on the improvement formula-improvement effect interaction feature vector includes: inputting the improvement formula-improvement effect interaction feature vector into a decoder-based best improvement plan inference device to obtain the best improvement plan for inference.

[0085] In the technical solution of the present application, the improved ratio component association feature map is used to represent the semantic association features between the components of multiple soil improvement ratio schemes, and the improved effect association feature map between the formulas represents the semantic association features of the local improved image effects between the multiple soil structure improvement effect images corresponding to the multiple soil improvement ratio schemes. Considering that the multiple soil improvement ratio schemes and the multiple soil structure improvement effect images have differences in data modality and source domain content representation, this makes the improved ratio component association feature map and the improved effect association feature map between the formulas have a more significant feature distribution inconsistency. After information excitation-squeezing processing, the obtained improved ratio component association feature vector and the improved effect association feature vector between the formulas will have a more significant feature distribution inconsistency. In this way, when the improved ratio component association feature vector and the improved effect association feature vector between the formulas are input into the feature dynamic constraint interaction module based on the prior distribution, the overall feature distribution of the improved formula-improvement effect interaction feature vector obtained will have the problem of insufficient aggregation, thereby affecting its decoding convergence efficiency through the decoder, that is, affecting the efficiency of decoding training and the accuracy of decoding results.

[0086] In a preferred embodiment, the improved formula-improvement effect interaction feature vector is passed through a decoder-based optimal improvement solution reasoner to obtain the inference optimal improvement solution, specifically comprising:

[0087] After multiplying the improved formula-improved effect interaction feature vector by the length of the improved formula-improved effect interaction feature vector, the improved formula-improved effect interaction feature vector is then subtracted from the norm of the improved formula-improved effect interaction feature vector to obtain the improved formula-improved effect interaction full semantic representation vector;

[0088] After multiplying the improved formula-improved effect interaction feature vector by the square root of the length of the improved formula-improved effect interaction feature vector, the improved formula-improved effect interaction feature vector is subtracted from the second norm of the improved formula-improved effect interaction feature vector to obtain an improved formula-improved effect interaction semi-semantic representation vector;

[0089] Respectively calculating the square root of the absolute value of each eigenvalue of the improved formula-improved effect interaction full semantic representation vector and the improved formula-improved effect interaction semi-semantic representation vector to obtain the improved formula-improved effect interaction full semantic modulation vector and the improved formula-improved effect interaction semi-semantic modulation vector;

[0090] Respectively calculating the base-two logarithmic value of each eigenvalue of the improved formula-improved effect interaction full semantic modulation vector and the improved formula-improved effect interaction semi-semantic modulation vector to obtain the improved formula-improved effect interaction full semantic information vector and the improved formula-improved effect interaction semi-semantic information vector;

[0091] Calculating the weighted sum of the improved formula-improved effect interaction full semantic information vector and the improved formula-improved effect interaction semi-semantic information vector using a weight coefficient as a hyperparameter to obtain an optimized improved formula-improved effect interaction feature vector;

[0092] The optimized improvement formula-improvement effect interaction feature vector is passed through the decoder-based best improvement solution inference device to obtain the inferred best improvement solution.

[0093] Accordingly, in this preferred embodiment, the applicant of the present application takes the improved formula-improved effect interaction feature vector as a feature set, wherein the change semantics represented in units of the eigenvalues ​​of the improved formula-improved effect interaction feature vector, in order to dynamically aggregate the semantic set as a whole composed of different change semantics of the improved formula-improved effect interaction feature vector without ignoring the individual semantic change information, the individual features of the improved formula-improved effect interaction feature vector and the collective expression of the vector scale of the improved formula-improved effect interaction feature vector are used as full-width and half-width feature representations, and the low-rank norms of different dimensions of the overall semantics of the feature set of the improved formula-improved effect interaction feature vector are used as phases through negative correlation calculation and amplitude-phase information representation modulation is performed to dynamically adjust the semantic information change relationship of the improved formula-improved effect interaction feature vector, thereby improving the overall semantic information expression aggregation of the feature set of the improved formula-improved effect interaction feature vector, and improving the decoding convergence efficiency of the improved formula-improved effect interaction feature vector when it is decoded by the best improvement scheme inference based on the decoder.

[0094] In summary, the above scheme is adopted to mine the responsiveness relationship between the soil structure improvement effect images corresponding to different soil improvement ratio schemes and the scheme itself by using intelligent algorithms, as well as the dynamic change law between the various soil structure improvement effect images, so as to analyze and deduce the best soil improvement ratio scheme. In this way, the best ratio scheme is inferred based on limited experimental data, the cycle of manual analysis is reduced, the intelligence and automation of soil improvement methods are improved, and technical support is provided for the sustainable utilization of saline-alkali land.

[0095] Figure 2 FIG. 1 is a block diagram of a system for improving saline-alkali land based on fly ash according to an exemplary embodiment. Figure 2 As shown, the system 200 includes:

[0096] The soil structure improvement effect image acquisition module 201 is used to acquire multiple soil improvement ratio schemes and multiple soil structure improvement effect images corresponding to the multiple soil improvement ratio schemes, wherein each of the soil improvement ratio schemes includes fly ash, desulfurized gypsum and organic fertilizer, and the fly ash includes silicon dioxide, aluminum oxide, calcium oxide, iron oxide, titanium oxide, sulfur oxide, potassium oxide, phosphorus oxide, magnesium oxide and sodium oxide;

[0097] A correlation feature extraction module 202 is used to extract correlation features from the plurality of soil improvement ratio schemes and the plurality of soil structure improvement effect images to obtain an improvement ratio component correlation feature map and a formula improvement effect correlation feature map;

[0098] A multi-scale perception enhancement and information excitation-squeezing module 203, which is used to input the improved ratio component association feature map and the improved effect association feature map between recipes into an information excitation-squeezing module including a feature multi-scale perception enhancement module and a multi-scale semantic feature extraction module to obtain an improved ratio component association feature vector and an improved effect association feature vector between recipes;

[0099] A feature dynamic constraint interaction module 204, used for inputting the improved ratio component association feature vector and the improved effect association feature vector between the formulas into a feature dynamic constraint interaction module based on prior distribution to obtain an improved formula-improved effect interaction feature vector;

[0100] The best improvement solution determination module 205 is used to determine the best improvement solution based on the improvement formula-improvement effect interaction feature vector.

[0101] Reference below Figure 3, which shows a schematic diagram of the structure of an electronic device 600 suitable for implementing the embodiment of the present application. The terminal device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0102] like Figure 3 As shown, the electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the electronic device 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0103] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 3 The electronic device 600 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0104] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 609, or installed from the storage device 608, or installed from the ROM 602. When the computer program is executed by the processing device 601, the above-mentioned functions defined in the method of the embodiment of the present application are executed.

[0105] It should be noted that the computer-readable medium mentioned above in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0106] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0107] The computer-readable medium may be included in the electronic device, or may exist independently without being installed in the electronic device.

[0108] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0109] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0110] The modules involved in the embodiments described in this application can be implemented by software or hardware. The name of the module does not limit the module itself in some cases. For example, the test parameter acquisition module can also be described as a "module for acquiring device test parameters corresponding to the target device".

[0111] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0112] In the context of the present application, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0113] Figure 4 : is an application scenario diagram of a method for improving saline-alkali land based on fly ash according to an exemplary embodiment. Figure 4 As shown, in this application scenario, first, a plurality of soil improvement ratio schemes and a plurality of soil structure improvement effect images corresponding to the plurality of soil improvement ratio schemes are obtained (for example, Figure 4 Then, the obtained multiple soil structure improvement effect images are input to a server (for example, Figure 4 In S) shown in the figure, the server is capable of processing the multiple soil structure improvement effect images based on the fly ash improvement algorithm for saline-alkali land to determine the best improvement plan.

[0114] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

[0115] In addition, although each operation is described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or to be performed in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the application. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0116] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely example forms of implementing the claims. Regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Claims

1. A method for improving saline-alkali land based on fly ash, It is characterized in that include: Acquire multiple soil improvement ratio schemes and multiple soil structure improvement effect images corresponding to the multiple soil improvement ratio schemes, wherein each of the soil improvement ratio schemes includes fly ash, desulfurized gypsum and organic fertilizer, and the fly ash includes silicon dioxide, aluminum oxide, calcium oxide, iron oxide, titanium oxide, sulfur oxide, potassium oxide, phosphorus oxide, magnesium oxide and sodium oxide; Extracting correlation features from the plurality of soil improvement ratio schemes and the plurality of soil structure improvement effect images to obtain an improvement ratio component correlation feature map and a formula improvement effect correlation feature map; Inputting the improved ratio component association feature map and the improved effect association feature map between recipes into an information excitation-squeezing module including a feature multi-scale perception enhancement module and a multi-scale semantic feature extraction module to obtain an improved ratio component association feature vector and an improved effect association feature vector between recipes; Inputting the improved ratio component association feature vector and the improved effect association feature vector between the formulas into a feature dynamic constraint interaction module based on prior distribution to obtain an improved formula-improved effect interaction feature vector; Determine the best improvement plan based on the improvement formula-improvement effect interaction feature vector; Extracting correlation features from the plurality of soil improvement ratio schemes and the plurality of soil structure improvement effect images to obtain an improvement ratio component correlation feature map and a formula improvement effect correlation feature map, including: Arranging the plurality of soil improvement ratio schemes into a full-sample soil improvement ratio input matrix according to the formula sample dimension; Arranging the plurality of soil structure improvement effect images into a soil structure improvement effect input tensor according to image sample dimensions; Inputting the soil improvement ratio input matrix of all samples into the improvement ratio component association feature extractor based on the hole convolutional neural network model to obtain the improvement ratio component association feature map; Inputting the soil structure improvement effect input tensor into a formula improvement effect correlation feature extractor based on a three-dimensional convolutional neural network model to obtain a formula improvement effect correlation feature map; The improved ratio component association feature map and the improved effect association feature map between recipes are input into an information excitation-squeezing module including a feature multi-scale perception enhancement module and a multi-scale semantic feature extraction module to obtain an improved ratio component association feature vector and an improved effect association feature vector between recipes, including: Input the improved ratio component association characteristic graph and the improved effect association characteristic graph between recipes into the feature multi-scale perception enhancement module of the information excitation-squeezing module to obtain a multi-scale perception enhanced improved ratio component association characteristic graph and a multi-scale perception enhanced improved effect association characteristic graph between recipes; Input the multi-scale perception enhancement improved ratio component association feature map and the multi-scale perception enhancement improved effect association feature map between recipes into the multi-scale semantic feature extraction module of the information excitation-squeezing module to obtain the improved ratio component association feature vector and the improved effect association feature vector between recipes; Inputting the improved ratio component association feature map and the improved effect association feature map between recipes into the feature multi-scale perception enhancement module of the information excitation-squeezing module to obtain a multi-scale perception enhanced improved ratio component association feature map and a multi-scale perception enhanced improved effect association feature map between recipes, including: Inputting the improved ratio component association feature map into the low-frequency detection branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association local low-frequency activation feature vector; Inputting the improved ratio component association feature map into the simplification and compression branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association compression feature map; Inputting the improved ratio component association feature map into the global space perception branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association receptive field expansion activation feature matrix; Multiplying the improved ratio component associated receptive field expansion activation feature matrix by each feature matrix along the channel dimension of the improved ratio component associated compression feature map by position point to obtain an improved ratio component associated compression-expansion activation feature map; Taking each eigenvalue in the improved ratio component associated local low-frequency activation feature vector as a weight, performing point multiplication on each feature matrix along the channel dimension of the improved ratio component associated compression feature map to obtain an improved ratio component associated compression-low-frequency fusion feature map; Adding the improved ratio component associated compression-expansion activation feature map and the improved ratio component associated compression-low frequency fusion feature map according to position to obtain an improved ratio component associated multi-scale fusion activation feature map; Performing dilated convolution coding on the improved ratio component association multi-scale fusion activation feature map to obtain the multi-scale perceptual enhancement improved ratio component association feature map; Inputting the improved ratio component association feature map into the low-frequency detection branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association local low-frequency activation feature vector, including: Performing point convolution processing on the improved ratio component association characteristic map to obtain an improved ratio component association channel compression characteristic map; Performing global mean pooling on the improved ratio component association channel compression feature map to obtain an improved ratio component association local low-frequency feature vector; Performing nonlinear activation on the improved ratio component associated local low-frequency feature vector to obtain the improved ratio component associated local low-frequency activation feature vector; Inputting the improved ratio component association feature map into the simplified compression branch of the feature multi-scale perception enhancement module to obtain the improved ratio component association compression feature map, including: Performing point convolution processing on the improved ratio component association characteristic map to obtain the improved ratio component association compression characteristic map; The improved ratio component association feature map is input into the global space perception branch of the feature multi-scale perception enhancement module to obtain an improved ratio component association receptive field expansion activation feature matrix, including: Performing dilated convolution coding on the improved ratio component association feature map to obtain an improved ratio component association receptive field expansion feature map; Performing point convolution processing on the improved ratio component association receptive field expansion feature map to obtain an improved ratio component association receptive field expansion global feature matrix; Nonlinearly activating the improved ratio component association receptive field expansion global feature matrix to obtain the improved ratio component association receptive field expansion activation feature matrix; Inputting the multi-scale perception enhancement improved ratio component association feature map and the multi-scale perception enhancement improved effect association feature map between recipes into the multi-scale semantic feature extraction module of the information excitation-squeezing module to obtain the improved ratio component association feature vector and the improved effect association feature vector between recipes, including: Upsampling the multi-scale perceptual enhancement improved ratio component association feature map to obtain an upsampled multi-scale perceptual enhancement improved ratio component association feature map; Using an improved proportion component association global semantic feature extraction module to process the upsampled multi-scale perceptual enhancement improved proportion component association feature map to obtain an improved proportion component association global semantic extraction feature vector; Dimensionally adjusting the improved proportion component association global semantic extraction feature vector to obtain an improved proportion component association global semantic feature vector; Using an improved ratio component association local semantic feature extraction module to process the upsampled multi-scale perceptual enhancement improved ratio component association feature map to obtain an improved ratio component association local semantic extraction feature vector; Dimensionally adjusting the improved proportion component associated local semantic extraction feature vector to obtain an improved proportion component associated local semantic feature vector; Calculate the improved proportion component association global semantic feature vector and the improved proportion component association local semantic feature vector, and add them element by element according to position to obtain an improved proportion component association multi-scale semantic fusion feature vector; Performing activation processing based on a sigmoid function on the improved proportion component association multi-scale semantic fusion feature vector to obtain the improved proportion component association feature vector; The improved ratio component association global semantic feature extraction module includes a global average pooling layer, a first point convolution layer, a first batch normalization layer and a first activation layer, and the improved ratio component association local semantic feature extraction module includes a second point convolution layer, a second batch normalization layer and a second activation layer; Inputting the improved ratio component association feature vector and the improved effect association feature vector between the formulas into a feature dynamic constraint interaction module based on prior distribution to obtain an improved formula-improvement effect interaction feature vector, including: Calculate an exponential function value with a natural constant as the base and a first weight coefficient as the dot product value of each eigenvalue in the improved proportion component association eigenvector as the exponent to obtain a nonlinear improved proportion component association eigenvector; Calculate an exponential function value with a natural constant as the base and a second weight coefficient as the dot product value of each eigenvalue in the inter-formula improvement effect association eigenvector as the exponent to obtain a nonlinear inter-formula improvement effect association eigenvector; The improved formula-improvement effect interaction feature vector is obtained by calculating the dot product of the first Gaussian distribution function coefficient and the nonlinear improved ratio component associated feature vector and the dot product of the second Gaussian distribution function coefficient and the nonlinear improved effect associated feature vector.

2. The method for improving saline-alkali land based on fly ash according to claim 1, It is characterized in that Determining the best improvement scheme based on the improvement formula-improvement effect interaction feature vector includes: The improved recipe-improvement effect interaction feature vector is input into the decoder-based optimal improvement solution inference device to obtain the inferred optimal improvement solution.