A salt-resistant and synergistic fertilizer and a preparation method thereof

Abstract

The present application discloses a salt-resistant synergistic fertilizer and a preparation method thereof, which relates to the field of intelligent preparation. It collects hot air temperature data in real time through a temperature sensor, and introduces a data processing and analysis algorithm based on artificial intelligence and deep learning at the back end to perform time series analysis on the hot air real-time temperature data, so as to learn and capture the local time series characteristics of the hot air real-time temperature and propagate the aggregated information, thereby automatically performing the corresponding atomization drying control. In this way, the injection valve opening can be adaptively adjusted based on the dynamic change of the real-time temperature of the hot air, so as to optimize the atomization drying process in a more intelligent monitoring manner, ensure the adaptability of temperature control during the atomization drying process and the stability of the preparation quality of the salt-resistant synergistic fertilizer.

Description

Technical Field

[0001] The present application relates to the field of intelligent preparation, and more specifically, to a salt-resistant and synergistic fertilizer and a preparation method thereof. Background Art

[0002] In modern agricultural production, soil salinization has become an important factor restricting crop growth and yield improvement. The existence of saline-alkali land not only affects the growth environment of crops, but also causes the loss of soil nutrients and the evaporation of water. Salt-resistant and synergistic fertilizers can improve the tolerance of plants to salt through various mechanisms, including enhancing the absorption capacity of plant roots, improving the physiological metabolic process of plants, and promoting the growth and development of plants. Therefore, it is crucial to use salt-resistant and synergistic fertilizers to help plants better absorb nutrients, improve soil quality, enhance crop salt resistance, and reduce salt-alkali hazards.

[0003] The invention with publication number CN103044120A discloses a soil-improving, synergistic, salt- and alkali-resistant water-soluble fertilizer and its preparation method, which can use the waste liquid generated in the amino acid fermentation production process as a carrier, and add active humic acid made from weathered coal, lignite or peat rich in humic acid, as well as other water-soluble fertilizers and trace elements. After mixing in a specific proportion, the powdered material is made by spray drying technology. The advantages of this solution are low cost, energy saving, and environmental protection, and it can effectively utilize waste and realize the recycling of resources. In addition, this fertilizer has good water solubility, does not leave harmful residues, is rich in nutrients, and is highly safe, which helps to improve soil structure and reduce salinity hazards.

[0004] However, in the preparation process of the above-mentioned soil-improving and salt-alkali-resistant water-soluble fertilizer, when spray drying is performed to prepare powdered materials, the hot air temperature is usually adjusted by fixed temperature control rules or manual monitoring. The intelligence level of this method is low, which will affect the accuracy and adaptability of temperature control. And temperature fluctuations will cause changes in the drying speed of the material. Too high a temperature may cause the surface of the material to harden rapidly, and the internal moisture cannot evaporate in time, resulting in excessive internal humidity. The final product may not be dry enough or even agglomerate; while too low a temperature will slow down the drying process, extend the production cycle, and increase energy consumption. In addition, temperature fluctuations will lead to inconsistent quality of different batches of water-soluble fertilizer products. This is because temperature changes may cause chemical changes in certain components in the fertilizer. For example, some organic substances or trace elements may decompose or denature at high temperatures, thereby affecting the content of the effective ingredients in the fertilizer.

[0005] Therefore, a preparation scheme of an optimized salt-resistant and enhanced fertilizer is desired. Summary of the invention

[0006] In order to solve the above technical problems, this application is proposed.

[0007] According to one aspect of the present application, a method for preparing a salt-resistant and synergistic fertilizer is provided, comprising:

[0008] The weathered coal, lignite, peat or peat is crushed and then added with nitric acid for mixed oxidation treatment to obtain a mixed oxidized material;

[0009] Adding ammonia water to the mixed oxidizing material, mixing, stirring, extracting and separating the mixed oxidizing material to obtain active humic acid;

[0010] The active humic acid, amino acid fermentation waste liquid, urea, water-soluble phosphate fertilizer, water-soluble potash fertilizer and trace elements are mixed and stirred to obtain a mixed solution;

[0011] The mixed liquid is transported to a spray drying tower and sprayed out through a high-speed centrifugal atomizer, and then falls in parallel with hot air to obtain a salt-resistant and synergistic fertilizer;

[0012] The mixed liquid is transported to a spray drying tower and sprayed out through a high-speed centrifugal atomizer, and then falls in parallel with hot air to obtain a salt-resistant and synergistic fertilizer, including:

[0013] A time queue for obtaining the real-time temperature of hot air collected by a temperature sensor;

[0014] Performing local time series correlation feature extraction on the time queue of the hot air real-time temperature to obtain a sequence of hot air local time series correlation feature vectors;

[0015] Inputting the sequence of the local temporal correlation feature vectors of hot air into a feature message propagation network based on the characteristic spatiotemporal transfer characteristics to obtain a hot air temperature spatiotemporal transfer aggregation representation vector;

[0016] Based on the hot air temperature spatiotemporal transfer aggregation representation vector, a control instruction is determined, wherein the control instruction includes a decoded value of a recommended value of an injection valve opening.

[0017] Further, extracting local time series correlation features from the time queue of the hot air real-time temperature to obtain a sequence of hot air local time series correlation feature vectors includes:

[0018] Sequentially segmenting and data structuring the time queue of the hot air real-time temperature to obtain a sequence of local time series input vectors of the hot air real-time temperature;

[0019] Each hot air real-time temperature local time series input vector in the sequence of the hot air real-time temperature local time series input vector is respectively input into the hot air temperature time series associated feature extractor based on the RNN model to obtain the sequence of the hot air local time series associated feature vectors.

[0020] Furthermore, the sequence of the hot air local time series associated feature vectors is input into a feature message propagation network based on the feature spatiotemporal transfer characteristics to obtain a hot air temperature spatiotemporal transfer aggregation representation vector, including:

[0021] Extracting a current hot air local time series association feature vector from the sequence of hot air local time series association feature vectors, and defining other hot air local time series association feature vectors in the sequence of hot air local time series association feature vectors as historical hot air local time series association feature vectors to obtain a sequence of historical hot air local time series association feature vectors;

[0022] Calculate the hot air local time series static energy value of each historical hot air local time series associated characteristic vector in the sequence of the historical hot air local time series associated characteristic vectors;

[0023] Determine the hot air local time series energy spatial attenuation factor based on the spatial span coefficient between each of the historical hot air local time series associated characteristic vectors and the current hot air local time series associated characteristic vector;

[0024] Determine the hot air local time series energy time attenuation factor based on the time span coefficient between each of the historical hot air local time series associated characteristic vectors and the current hot air local time series associated characteristic vector;

[0025] Based on the hot air local time series energy spatial attenuation factor and the hot air local time series energy temporal attenuation factor, the hot air local time series static energy value of each of the historical hot air local time series associated characteristic vectors is temporally modulated to obtain the hot air local time series dynamic energy spatiotemporal transfer value of each of the historical hot air local time series associated characteristic vectors relative to the current hot air local time series associated characteristic vector;

[0026] Inputting the hot air local time series dynamic energy spatiotemporal transfer values ​​of the historical hot air local time series associated feature vectors into the energy transfer gating module based on the mask function to obtain a sequence of hot air local time series dynamic energy spatiotemporal transfer weights;

[0027] Based on the sequence of the spatiotemporal transfer weights of the local time series dynamic energy of the hot air, a weighted sum of the sequence of the historical local time series associated feature vectors of the hot air is calculated to obtain a historical local time series node feature transfer aggregation modulation vector;

[0028] The position-based sum of the historical hot air local time series node feature transfer aggregation modulation vector and the current hot air local time series associated feature vector is calculated to obtain the hot air temperature spatiotemporal transfer aggregation representation vector.

[0029] Further, calculating the hot air local time series static energy value of each historical hot air local time series associated characteristic vector in the sequence of the historical hot air local time series associated characteristic vectors includes:

[0030] Calculating the mean and standard deviation of the historical hot air local time series associated characteristic vectors respectively to obtain the historical hot air local time series characteristic mean and the historical hot air local time series standard deviation;

[0031] After calculating the fourth power of the position difference between the historical hot air local time series associated feature vector and the historical hot air local time series feature mean, the calculated fourth power modulates the expected value of the historical hot air local time series offset vector to obtain the hot air local time series feature offset expected factor;

[0032] After calculating the division between the expected factor of the local hot air time series characteristic offset and the fourth power of the historical local hot air time series standard deviation, the result is added to the maximum eigenvalue in the historical hot air local time series associated eigenvector to obtain the static energy value of the local hot air time series.

[0033] Further, based on the spatial span coefficient between each of the historical hot air local time series associated characteristic vectors and the current hot air local time series associated characteristic vector, determining the hot air local time series energy spatial attenuation factor includes:

[0034] Calculating the spatial distance span between the historical hot air local time series associated characteristic vector and the current hot air local time series associated characteristic vector to obtain the historical hot air local time series spatial span coefficient;

[0035] The division between the hot air local time series static energy value corresponding to the historical hot air local time series associated characteristic vector and the historical hot air local time series spatial span coefficient is calculated to obtain the hot air local time series energy spatial attenuation factor.

[0036] Further, based on the time span coefficient between each of the historical hot air local time series associated characteristic vectors and the current hot air local time series associated characteristic vector, determining the hot air local time series energy time attenuation factor includes:

[0037] After calculating the subtraction between the timestamp of the current hot air local time series associated feature vector and the timestamp of the historical hot air local time series associated feature vector, the obtained timestamp value is rounded down to obtain the hot air local time series time span coefficient;

[0038] After dividing the hot air local time series time span coefficient by the inverse scale parameter controlling the time decay period, the obtained time span decay representation coefficient is used as an exponential power, and the natural exponential function value with the natural constant e as the base is calculated to obtain the time span decay class support representation coefficient;

[0039] The division between the local time series static energy value of the hot air and the time span attenuation class support representation coefficient is calculated to obtain the local time series energy time attenuation factor of the hot air.

[0040] Furthermore, based on the hot air local time series energy spatial attenuation factor and the hot air local time series energy temporal attenuation factor, the hot air local time series static energy values ​​of the respective historical hot air local time series associated characteristic vectors are temporally and spatially modulated to obtain the hot air local time series dynamic energy spatiotemporal transfer values ​​of the respective historical hot air local time series associated characteristic vectors relative to the current hot air local time series associated characteristic vector, including: calculating the position-weighted sum of the hot air local time series energy spatial attenuation factor and the hot air local time series energy temporal attenuation factor to obtain the hot air local time series dynamic energy spatiotemporal transfer value.

[0041] Further, the hot air local time series dynamic energy spatiotemporal transfer values ​​of the respective historical hot air local time series associated feature vectors are input into the energy transfer gating module based on the mask function to obtain a sequence of hot air local time series dynamic energy spatiotemporal transfer weights, including: in the energy transfer gating module based on the mask function, in response to the hot air local time series dynamic energy spatiotemporal transfer value being greater than a threshold hyperparameter, The function normalizes the hot air local temporal dynamic energy spatiotemporal transfer value that is greater than the threshold hyperparameter to obtain the hot air local temporal dynamic energy spatiotemporal transfer weight.

[0042] Further, based on the hot air temperature time-space transfer aggregate representation vector, a control instruction is determined, the control instruction includes a decoded value of the injection valve opening recommended value, including: inputting the hot air temperature time-space transfer aggregate representation vector into a decoder-based injection valve opening controller to obtain a control instruction.

[0043] According to another aspect of the present application, a salt-resistant and synergistic fertilizer is provided, and the salt-resistant and synergistic fertilizer is prepared by the preparation method of the salt-resistant and synergistic fertilizer as described above.

[0044] Compared with the prior art, the salt-resistant synergistic fertilizer and its preparation method provided by the present application monitor and collect hot air temperature data in real time through a temperature sensor, and introduce a data processing and analysis algorithm based on artificial intelligence and deep learning at the back end to perform time series analysis on the hot air real-time temperature data, so as to learn and capture the local time series characteristics of the hot air real-time temperature and propagate the aggregated information, thereby automatically performing the corresponding atomization drying control. In this way, the injection valve opening can be adaptively adjusted based on the dynamic changes in the real-time temperature of the hot air, thereby optimizing the atomization drying process in a more intelligent monitoring manner, ensuring the adaptability of temperature control during the atomization drying process and the stability of the preparation quality of the salt-resistant synergistic fertilizer. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0046] Figure 1 The present invention is a flow chart of a method for preparing a salt-resistant and synergistic fertilizer according to an embodiment of the present application;

[0047] Figure 2 A data flow diagram of a method for preparing a salt-resistant and synergistic fertilizer according to an embodiment of the present application;

[0048] Figure 3 It is a flow chart of sub-step S4 of the method for preparing the salt-resistant and synergistic fertilizer according to an embodiment of the present application;

[0049] Figure 4 This is a flow chart of sub-step S42 of the method for preparing salt-resistant and synergistic fertilizer according to an embodiment of the present application. DETAILED DESCRIPTION

[0050] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0051] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0052] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0053] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0054] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0055] In the technical solution of the present application, a method for preparing a salt-resistant and synergistic fertilizer is proposed. Figure 1 The present invention is a flow chart of a method for preparing a salt-resistant and synergistic fertilizer according to an embodiment of the present application. Figure 2 The data flow diagram of the method for preparing the salt-resistant and synergistic fertilizer according to the embodiment of the present application is shown in FIG. Figure 1 and Figure 2 As shown, the preparation method of the salt-resistant synergistic fertilizer according to the embodiment of the present application includes the following steps: S1, crushing weathered coal, lignite, peat or peat, adding nitric acid for mixed oxidation treatment to obtain a mixed oxidized material; S2, adding ammonia water to the mixed oxidized material and performing mixed stirring and extraction separation treatment to obtain active humic acid; S3, mixing the active humic acid, amino acid fermentation waste liquid, urea, water-soluble phosphate fertilizer, water-soluble potassium fertilizer and trace elements to obtain a mixed solution; S4, conveying the mixed solution to a spray drying tower, spraying it through a high-speed centrifugal atomizer, and then descending co-currently with the hot air to obtain a salt-resistant synergistic fertilizer.

[0056] In particular, in said S1, weathered coal, lignite, peat or peat are crushed and then mixed with nitric acid for oxidation treatment to obtain a mixed oxidized material. It should be understood that nitric acid is a strong oxidant that can oxidize the organic matter in these materials. The oxidation reaction can destroy the structure of organic matter and produce new functional groups. The oxidized materials have a higher nutrient utilization rate, and the oxidized materials can improve the soil structure. They can increase the organic matter content of the soil, improve the soil's ability to retain water and fertilizer, and promote the activity of soil microorganisms.

[0057] In particular, in S2, ammonia water is added to the mixed oxidized material and mixed, stirred, extracted and separated to obtain active humic acid. It should be understood that the mixed oxidized material is acidic after being oxidized by nitric acid, and humic acid is an organic acid that is insoluble in water under acidic conditions. After adding ammonia water, humic acid reacts with ammonia water to form soluble ammonium salts. The humic acid ammonium salt solution obtained by separation and extraction is concentrated, dried and other steps to obtain active humic acid. Active humic acid can promote the development of plant roots and improve the absorption and utilization rate of nutrients by plants.

[0058] In particular, in S3, the active humic acid, amino acid fermentation waste liquid, urea, water-soluble phosphate fertilizer, water-soluble potash fertilizer and trace elements are mixed and stirred to obtain a mixed solution. Among them, there is a synergistic effect between the various nutrients in the mixed solution. For example, humic acid can promote the absorption and utilization of nutrients such as nitrogen, phosphorus and potassium by plants, while amino acids can promote the growth and development of plants.

[0059] In particular, in S4, the mixed solution is transported to a spray drying tower and sprayed out through a high-speed centrifugal atomizer, and then descends in parallel with the hot air to obtain a salt-resistant and synergistic fertilizer. The fertilizer particles obtained by spray drying have high stability, are not easy to absorb moisture and agglomerate, and spray drying can be carried out at a lower temperature, which helps to retain the active ingredients in the mixed solution, such as active humic acid, amino acids and trace elements. However, in the preparation process of soil-improving and synergistic salt-alkali resistant water-soluble fertilizer, when spray drying is performed to prepare powdered materials, the hot air temperature is usually adjusted by a fixed temperature control rule or manual monitoring method, which has a low level of intelligence and affects the accuracy and adaptability of temperature control. And temperature fluctuations can cause changes in the drying speed of the material. Too high a temperature may cause the surface of the material to harden rapidly, and the internal moisture cannot evaporate in time, resulting in excessive internal humidity, and the final product may not be dry enough, or even agglomeration; while too low a temperature will slow down the drying process, extend the production cycle, and increase energy consumption. In addition, temperature fluctuations can lead to inconsistent quality of different batches of water-soluble fertilizer products, because temperature changes may cause chemical changes in certain components of the fertilizer. For example, some organic substances or trace elements may decompose or denature at high temperatures, thereby affecting the content of the active ingredients in the fertilizer. In a specific example of the present application, Figure 3As shown, the S4 includes: S41, obtaining a time queue of the real-time temperature of the hot air collected by the temperature sensor; S42, performing local temporal correlation feature extraction on the time queue of the real-time temperature of the hot air to obtain a sequence of local temporal correlation feature vectors of the hot air; S43, inputting the sequence of local temporal correlation feature vectors of the hot air into a characteristic message propagation network based on the characteristic spatiotemporal transfer characteristics to obtain a hot air temperature spatiotemporal transfer aggregation representation vector; S44, determining a control instruction based on the hot air temperature spatiotemporal transfer aggregation representation vector, the control instruction including a decoded value of a recommended value of the injection valve opening.

[0060] Specifically, the S41 and S42 obtain the time queue of the hot air real-time temperature collected by the temperature sensor; and extract the local time series correlation feature of the hot air real-time temperature time queue to obtain a sequence of hot air local time series correlation feature vectors. In a specific example of the present application, Figure 4 As shown, the S42 includes: S421, sequence segmentation and data structuring of the time queue of the hot air real-time temperature to obtain a sequence of local time series input vectors of the hot air real-time temperature; S422, inputting each hot air real-time temperature local time series input vector in the sequence of the hot air real-time temperature local time series input vector into a hot air temperature time series association feature extractor based on the RNN model to obtain a sequence of the hot air local time series association feature vectors.

[0061] More specifically, in S421, the time queue of the hot air real-time temperature is segmented and data structured to obtain a sequence of local time series input vectors of the hot air real-time temperature. Considering that the hot air real-time temperature has the dynamic characteristics of the time series and the time series change law in the time dimension, and these time series information are more obvious in different local time periods, it is helpful to more accurately control the opening of the injection valve. Based on this, in the technical solution of the present application, the time queue of the hot air real-time temperature is segmented and data structured to obtain a sequence of local time series input vectors of the hot air real-time temperature.

[0062] More specifically, in S422, each hot air real-time temperature local time series input vector in the sequence of the hot air real-time temperature local time series input vector is input into the hot air temperature time series associated feature extractor based on the RNN model to obtain the sequence of the hot air local time series associated feature vector. It should be understood that each hot air real-time temperature local time series input vector in the sequence of the hot air real-time temperature local time series input vector is input into the hot air temperature time series associated feature extractor based on the RNN model for feature mining, so as to extract the local time series associated feature information of the hot air real-time temperature in different local time periods in the time dimension, thereby obtaining a sequence of hot air local time series associated feature vectors. It is worth mentioning that RNN is a neural network used to process sequential data, such as text, time series, and audio. The uniqueness of RNN is that they have a "memory" ability to remember previous inputs and use them to process current inputs. At each time step, RNN combines the current input with the hidden state and uses a nonlinear activation function (such as tanh or ReLU) to calculate a new hidden state and output.

[0063] It is worth mentioning that in other examples of the present application, the local time series correlation feature extraction can be performed on the time queue of the hot air real-time temperature to obtain a sequence of hot air local time series correlation feature vectors through the following steps, for example: define the size of the local window; use the defined local window size to slide a window on the time series of the hot air real-time temperature. At each time step, the window moves one step to extract the local time series correlation features within the window; in each window, the local time series correlation features are extracted; the extracted local time series correlation features are combined into a local time series correlation feature vector. This vector represents the local time series correlation features of the time series of the hot air real-time temperature within the window to obtain the sequence of the hot air local time series correlation feature vectors.

[0064] Specifically, in S43, the sequence of the hot air local time series association feature vectors is input into a feature message propagation network based on the feature time-space transfer characteristics to obtain a hot air temperature time-space transfer aggregation representation vector. It should be understood that since each hot air local time series association feature vector in the sequence of the hot air local time series association feature vectors respectively contains the local time series association characteristics of the hot air temperature changing with time, these local time series features have mutual correlation and interaction. Therefore, in order to capture and express the dynamic change law of the hot air temperature in time and space, so as to achieve more precise injection valve opening control, in the technical solution of the present application, the sequence of the hot air local time series association feature vectors is further input into a feature message propagation network based on the feature time-space transfer characteristics to obtain a hot air temperature time-space transfer aggregation representation vector.

[0065] In an embodiment of the present application, the sequence of the hot air local time series association feature vectors is input into a feature message propagation network based on the feature time-space transfer characteristics to obtain a hot air temperature time-space transfer aggregation representation vector, including: first, extracting the current hot air local time series association feature vector from the sequence of the hot air local time series association feature vectors, and defining other hot air local time series association feature vectors in the sequence of the hot air local time series association feature vectors as historical hot air local time series association feature vectors to obtain a sequence of historical hot air local time series association feature vectors; that is, the current hot air local time series association feature vector is used as the feature representation of the current node, and at the same time, the hot air local time series association feature vector is used as the feature representation of the current node. The other feature vectors in the sequence of historical hot air local time series correlation feature vectors are regarded as historical hot air local time series correlation feature vectors, so as to construct a feature network on the time series; then, the hot air local time series static energy value of each historical hot air local time series correlation feature vector in the sequence of historical hot air local time series correlation feature vectors is calculated; here, the hot air local time series static energy value is calculated according to the maximum eigenvalue, mean and variance of the historical hot air local time series correlation feature vector, which helps to quantify the spatial inherent properties of the feature vector and provide a benchmark for the subsequent potential energy spatiotemporal decay, which helps the model understand the static importance of the hot air local time series characteristics of each node in the spatial dimension and the time dimension; then, based on Based on the spatial span coefficient between each historical hot air local time series associated characteristic vector and the current hot air local time series associated characteristic vector, the hot air local time series energy spatial attenuation factor is determined; and based on the time span coefficient between each historical hot air local time series associated characteristic vector and the current hot air local time series associated characteristic vector, the hot air local time series energy time attenuation factor is determined; wherein, these two factors are used to adjust the weight of information transmission between nodes, so that the model can adjust the intensity of information transmission according to the distance between nodes, enhance the sensitivity of the model to locality and timeliness, and make the transmission of hot air local time series characteristic information in the node more consistent with the actual physical process; and then, Based on the hot air local time series energy spatial attenuation factor and the hot air local time series energy time attenuation factor, the hot air local time series static energy value of each historical hot air local time series associated characteristic vector is subjected to spatiotemporal modulation to obtain the hot air local time series dynamic energy spatiotemporal transfer value of each historical hot air local time series associated characteristic vector relative to the current hot air local time series associated characteristic vector; it should be understood that the effect of the spatiotemporal modulation is to make the propagation of node features more consistent with the actual physical process, strengthen the model's ability to capture the temporal dynamic changes of the hot air local time series associated features, and improve the model's processing effect on time-sensitive tasks, such as subsequent injection valve opening control tasks;Then, the hot air local time series dynamic energy spatiotemporal transfer values ​​of each historical hot air local time series associated characteristic vector are input into the energy transfer gating module based on the mask function to obtain a sequence of hot air local time series dynamic energy spatiotemporal transfer weights; here, the energy transfer gating module based on the mask function can screen and strengthen the spatiotemporal transfer information of important hot air local time series characteristics through the gating mechanism, improve the accuracy and effectiveness of information transfer, and make the model pay more attention to the features that contribute to the hot air temperature time series monitoring task and the subsequent injection valve opening control task; further, based on the sequence of hot air local time series dynamic energy spatiotemporal transfer weights, the weighted sum of the sequence of historical hot air local time series associated characteristic vectors is calculated to obtain the historical hot air local time series node feature transfer aggregation modulation vector; that is, the hot air temperature local time series feature information from different historical nodes is integrated by weighted summing up the dynamic energy spatiotemporal transfer values ​​of the historical hot air local time series associated characteristic vectors. The historical hot air local time series node feature transfer aggregation modulation vector integrates the hot air temperature local time series feature information on the historical node, so that the model can use the temperature local time series context information in the time series, and improve the richness and accuracy of the hot air local time series node information transfer aggregation feature representation of the historical node; finally, the positional sum of the historical hot air local time series node feature transfer aggregation modulation vector and the current hot air local time series associated feature vector is calculated to obtain the hot air temperature spatiotemporal transfer aggregation representation vector; in the technical solution of the present application, the effect of the hot air temperature spatiotemporal transfer aggregation representation vector is to provide a hot air temperature local time series feature information transfer aggregation representation that integrates the current node and historical node information, provides rich context information for the downstream injection valve opening control task, and enhances the model's processing ability for complex tasks. ;

[0066] Among them, the process of calculating the hot air local time series static energy value of each historical hot air local time series associated feature vector in the sequence of the historical hot air local time series associated feature vectors includes: respectively calculating the mean and standard deviation of the historical hot air local time series associated feature vectors to obtain the historical hot air local time series feature mean and the historical hot air local time series standard deviation; after calculating the fourth power of the position difference between the historical hot air local time series associated feature vector and the historical hot air local time series feature mean, the calculated fourth power modulates the expected value of the historical hot air local time series offset vector to obtain the hot air local time series feature offset expectation factor; after calculating the division between the hot air local time series feature offset expectation factor and the fourth power of the historical hot air local time series standard deviation, adding it to the maximum eigenvalue in the historical hot air local time series associated feature vector to obtain the hot air local time series static energy value.

[0067] More specifically, based on the spatial span coefficient between each historical hot air local time series association characteristic vector and the current hot air local time series association characteristic vector, the process of determining the hot air local time series energy spatial attenuation factor includes: calculating the spatial distance span between the historical hot air local time series association characteristic vector and the current hot air local time series association characteristic vector to obtain the historical hot air local time series spatial span coefficient; calculating the division between the hot air local time series static energy value corresponding to the historical hot air local time series association characteristic vector and the square of the historical hot air local time series spatial span coefficient to obtain the hot air local time series energy spatial attenuation factor.

[0068] More specifically, based on the time span coefficient between each historical hot air local time series associated characteristic vector and the current hot air local time series associated characteristic vector, the process of determining the hot air local time series energy time decay factor includes: after calculating the subtraction between the timestamp of the current hot air local time series associated characteristic vector and the timestamp of the historical hot air local time series associated characteristic vector, rounding down the obtained timestamp value to obtain the hot air local time series time span coefficient; after dividing the hot air local time series time span coefficient by the inverse scale parameter that controls the time decay period, using the obtained time span decay representation coefficient as an exponential power, calculating the natural exponential function value with the natural constant e as the base to obtain the time span decay class support representation coefficient; calculating the division between the hot air local time series static energy value and the time span decay class support representation coefficient to obtain the hot air local time series energy time decay factor.

[0069] More specifically, based on the hot air local time series energy spatial attenuation factor and the hot air local time series energy temporal attenuation factor, the hot air local time series static energy value of each historical hot air local time series associated characteristic vector is subjected to spatiotemporal modulation to obtain the hot air local time series dynamic energy spatiotemporal transfer value of each historical hot air local time series associated characteristic vector relative to the current hot air local time series associated characteristic vector, and the process includes: calculating the position-weighted sum of the hot air local time series energy spatial attenuation factor and the hot air local time series energy temporal attenuation factor to obtain the hot air local time series dynamic energy spatiotemporal transfer value.

[0070] More specifically, the process of inputting the hot air local time series dynamic energy spatiotemporal transfer values ​​of the respective historical hot air local time series associated feature vectors into the energy transfer gating module based on the mask function to obtain a sequence of hot air local time series dynamic energy spatiotemporal transfer weights includes: in the energy transfer gating module based on the mask function, in response to the hot air local time series dynamic energy spatiotemporal transfer value being greater than a threshold hyperparameter, The function normalizes the hot air local temporal dynamic energy spatiotemporal transfer value that is greater than the threshold hyperparameter to obtain the hot air local temporal dynamic energy spatiotemporal transfer weight.

[0071] In summary, in the above embodiment, the sequence of the local temporal correlation feature vectors of hot air is input into a characteristic message propagation network based on the characteristic spatiotemporal transfer characteristics to obtain the hot air temperature spatiotemporal transfer aggregation representation vector, including: inputting the sequence of the local temporal correlation feature vectors of hot air into a characteristic message propagation network based on the characteristic spatiotemporal transfer characteristics to process the sequence with the following characteristic message propagation formula to obtain the hot air temperature spatiotemporal transfer aggregation representation vector; wherein the characteristic message propagation formula is:

[0072] ;

[0073] ;

[0074] ;

[0075] ;

[0076] ;

[0077] ;

[0078] ;

[0079] in, is the sequence of local temporal correlation feature vectors of hot air, are the first, second, and third in the sequence of the local temporal correlation feature vectors of the hot air. and The local temporal correlation feature vector of hot air, For the The first of the local time series correlation feature vectors of hot air The eigenvalues ​​at the positions, and They are respectively The characteristic mean and standard deviation of the local time series correlation feature vector of hot air, is to calculate the expected value, It means to take the maximum value in the vector. is the local time series static energy value of hot air, express and The spatial span coefficient between for and The energy space attenuation factor between and express and timestamp, Indicates a round-down operation. is the descaling parameter that controls the time decay period, is the value of the natural exponential function, for and The energy time decay factor between and is a trainable hyperparameter, for Relative to The dynamic energy space-time transfer value of For mask operation, is the number of vectors in the sequence of the local temporal correlation feature vectors of hot air, is the threshold hyperparameter, for function, A vector representing the spatiotemporal transfer aggregation of the hot air temperature.

[0080] More specifically, the S44 determines the control instruction based on the hot air temperature spatiotemporal transfer aggregation representation vector, and the control instruction includes the decoded value of the injection valve opening recommended value. In a specific example of the present application, the hot air temperature spatiotemporal transfer aggregation representation vector is input into the injection valve opening controller based on the decoder to obtain the control instruction, and the control instruction includes the decoded value of the injection valve opening recommended value. That is, the spatiotemporal transfer aggregation representation characteristics of the hot air temperature are used to perform decoding regression, so as to perform corresponding atomization drying control. In this way, the injection valve opening can be adaptively adjusted based on the real-time dynamic temperature changes of the hot air, so as to optimize the atomization drying process in a more intelligent monitoring manner, ensuring the adaptability of temperature control during the atomization drying process and the stability of the preparation quality of salt-resistant and synergistic fertilizers.

[0081] That is, considering that the sequence of the local temporal correlation feature vectors of the hot air represents the cyclic temporal correlation characteristics of the real-time temperature of the hot air in the local time domain, when it is input into the feature message propagation network based on the characteristic spatiotemporal transfer characteristics, based on the temporal and spatial transfer differences in the potential energy of the timing nodes between the local time domains, the hot air temperature spatiotemporal transfer aggregation representation vector will also have a global time domain propagation-type aggregation difference, that is, the characteristic manifold of the hot air temperature spatiotemporal transfer aggregation representation vector has expression sparsity and uneven distribution in the high-dimensional feature space, which in turn affects the accuracy of the control instructions obtained through the decoder-based injection valve opening controller.

[0082] Based on this, in a preferred embodiment, the hot air temperature spatiotemporal transfer aggregation representation vector is input into a decoder-based injection valve opening controller to obtain a control instruction, including:

[0083] Performing L random sampling on the hot air temperature spatiotemporal transfer aggregation representation vector to obtain L hot air temperature spatiotemporal transfer aggregation random perturbation eigenvalues;

[0084] Arrange the L hot air temperature space-time transfer aggregate random disturbance eigenvalues ​​in order from large to small to obtain a hot air temperature space-time transfer aggregate random disturbance ordered vector;

[0085] Calculating the inner product of the heat air temperature space-time transfer aggregation representation vector to obtain the heat air temperature space-time transfer aggregation energy modulation characteristic value;

[0086] Based on the hot air temperature spatiotemporal transfer aggregation energy modulation characteristic value, the hot air temperature spatiotemporal transfer aggregation random perturbation ordered vector is semantically scaled to obtain a semantically enhanced hot air temperature spatiotemporal transfer aggregation random perturbation ordered vector;

[0087] Calculate the product of the semantically enhanced hot air temperature spatiotemporal transfer aggregation random perturbation ordered vector and the transposed vector of the hot air temperature spatiotemporal transfer aggregation representation vector to obtain the hot air temperature spatiotemporal transfer aggregation best example prompt orthogonal modulation matrix;

[0088] Calculate the matrix product between the best example prompt orthogonal modulation matrix of the hot air temperature space-time transfer aggregation and the hot air temperature space-time transfer aggregation representation vector to obtain the optimized hot air temperature space-time transfer aggregation representation vector;

[0089] The optimized hot air temperature spatiotemporal transfer aggregation representation vector is input into the decoder-based injection valve opening controller to obtain the control instruction.

[0090] The above optimization process is expressed as follows:

[0091] ;

[0092] ;

[0093] ;

[0094] ;

[0095] ;

[0096] ;

[0097] ;

[0098] ;

[0099] in, represents a normal distribution, represents the Gaussian density probability function, and denote the mean and variance of the hot air temperature spatiotemporal transfer aggregation representation vector, respectively, represents a random variable that follows a normal distribution, represents the number of random samplings, Indicates The random disturbance eigenvalues ​​of the spatiotemporal transmission of hot air temperature are aggregated, represents the thermal air temperature spatiotemporal transfer aggregation representation vector, The first vector representing the thermal air temperature spatiotemporal transfer aggregation representation vector eigenvalues, It means to calculate the average value of the vector. represents the length of the thermal air temperature spatiotemporal transfer aggregation representation vector, It indicates that the random disturbance eigenvalues ​​of the spatiotemporal transfer of hot air temperature are arranged in an orderly manner from large to small. Represents the ordered vector of random disturbances of thermal air temperature spatiotemporal transmission aggregation, It represents the modulation characteristic value of the aggregated energy of the hot air temperature space-time transfer, represents the square of the vector's two-norm, It means point multiplication by position. represents matrix multiplication, represents the transpose of a vector, The best example of hot air temperature spatiotemporal transfer aggregation prompts the orthogonal modulation matrix, A representation vector representing the optimized spatiotemporal transfer aggregation of hot air temperature.

[0100] Accordingly, by performing L random sampling on the thermal air temperature spatiotemporal transfer aggregate representation vector to obtain a set of new feature values ​​based on the original features but containing slight changes, randomness is introduced to explore different aspects of the feature space, and by simulating the noise or variation in the actual data, the model is better generalized to unseen data. Then, these thermal air temperature spatiotemporal transfer aggregate random perturbation feature values ​​are ordered from large to small to form a thermal air temperature spatiotemporal transfer aggregate random perturbation ordered vector to identify which perturbation directions may be the most important for the current problem. Such an ordering not only helps the algorithm focus on the feature differences that can best distinguish different categories, but also helps to reveal hidden data structures or patterns. Subsequently, the self-inner product of the hot air temperature spatiotemporal transfer aggregation representation vector is calculated to obtain the hot air temperature spatiotemporal transfer aggregation energy modulation eigenvalue, which reflects the overall strength or "energy" of the original feature vector and provides a benchmark for measuring the importance of the feature. Using this hot air temperature spatiotemporal transfer aggregation energy modulation eigenvalue as a reference point, the previously formed hot air temperature spatiotemporal transfer aggregation random perturbation ordered vector is semantically scaled to construct a semantically enhanced hot air temperature spatiotemporal transfer aggregation random perturbation ordered vector. In this way, those disturbances that appear to be more important in the original feature space are amplified, thereby enhancing the model's sensitivity to changes in key features, while also retaining relatively small but potentially useful disturbance information.

[0101] Then, the product of the semantically enhanced hot air temperature spatiotemporal transfer aggregation random perturbation ordered vector and the transpose of the hot air temperature spatiotemporal transfer aggregation representation vector is calculated to obtain the hot air temperature spatiotemporal transfer aggregation best sample prompt orthogonal modulation matrix, the purpose of which is to find the way in which the features interact from two different perspectives. The formed matrix can be regarded as a conversion tool for capturing and representing the complex linear relationship between features. Finally, the final feature structure modulation is completed by applying the hot air temperature spatiotemporal transfer aggregation best sample prompt orthogonal modulation matrix constructed above to the original hot air temperature spatiotemporal transfer aggregation representation vector, so that the features are more in line with the requirements of the objective function, reducing redundant information, while maximizing the retention of useful information, so as to improve learning efficiency and prediction performance.

[0102] In summary, the feature manifold optimization process of the hot air temperature spatiotemporal transfer aggregate representation vector is explained, which extracts the best prompt sample based on the fine-grained target retrieval of random perturbations, and then guides adversarial robustness learning based on the best prompt sample to complete the structure of the high-dimensional manifold of the feature vector in the feature space, so as to improve the detail and structural completeness of the feature expression. In this way, the expression effect of the hot air temperature spatiotemporal transfer aggregate representation vector is improved, and the accuracy of the control instructions obtained by the decoder-based injection valve opening controller is improved. In this way, the injection valve opening can be adjusted more accurately based on the real-time dynamic temperature changes of the hot air, so as to optimize the atomization drying process in a more intelligent monitoring manner, ensuring the adaptability of temperature control in the atomization drying process and the stability of the preparation quality of salt-resistant and synergistic fertilizers.

[0103] In summary, the method for preparing the salt-resistant synergistic fertilizer according to the embodiment of the present application is explained, which collects hot air temperature data in real time through a temperature sensor, and introduces a data processing and analysis algorithm based on artificial intelligence and deep learning at the back end to perform time series analysis on the hot air real-time temperature data, so as to learn and capture the local time series characteristics of the hot air real-time temperature and propagate the aggregated information, thereby automatically performing the corresponding atomization drying control. In this way, the injection valve opening can be adaptively adjusted based on the dynamic changes in the real-time temperature of the hot air, thereby optimizing the atomization drying process in a more intelligent monitoring manner, ensuring the adaptability of temperature control during the atomization drying process and the stability of the preparation quality of the salt-resistant synergistic fertilizer.

[0104] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for preparing a salt-resistant and synergistic fertilizer, characterized in that: include: The weathered coal, lignite, peat or peat is crushed and then added with nitric acid for mixed oxidation treatment to obtain a mixed oxidized material; Adding ammonia water to the mixed oxidizing material, mixing, stirring, extracting and separating the mixed oxidizing material to obtain active humic acid; The active humic acid, amino acid fermentation waste liquid, urea, water-soluble phosphate fertilizer, water-soluble potash fertilizer and trace elements are mixed and stirred to obtain a mixed solution; The mixed liquid is transported to a spray drying tower and sprayed out through a high-speed centrifugal atomizer, and then falls in parallel with hot air to obtain a salt-resistant and synergistic fertilizer; The mixed liquid is transported to a spray drying tower and sprayed out through a high-speed centrifugal atomizer, and then falls in parallel with hot air to obtain a salt-resistant and synergistic fertilizer, including: A time queue for obtaining the real-time temperature of hot air collected by a temperature sensor; Performing local time series correlation feature extraction on the time queue of the hot air real-time temperature to obtain a sequence of hot air local time series correlation feature vectors; Inputting the sequence of the local temporal correlation feature vectors of hot air into a feature message propagation network based on the characteristic spatiotemporal transfer characteristics to obtain a hot air temperature spatiotemporal transfer aggregation representation vector; determining a control instruction based on the hot air temperature spatiotemporal transfer aggregation representation vector, the control instruction comprising a decoded value of a recommended value of an injection valve opening; Extracting local time series correlation features from the time queue of the hot air real-time temperature to obtain a sequence of hot air local time series correlation feature vectors includes: Sequentially segmenting and data structuring the time queue of the hot air real-time temperature to obtain a sequence of local time series input vectors of the hot air real-time temperature; Inputting each hot air real-time temperature local time series input vector in the sequence of the hot air real-time temperature local time series input vector into the hot air temperature time series correlation feature extractor based on the RNN model to obtain the sequence of the hot air local time series correlation feature vector; Inputting the sequence of the local temporal correlation feature vectors of the hot air into a feature message propagation network based on the characteristic spatiotemporal transfer characteristics to obtain a hot air temperature spatiotemporal transfer aggregation representation vector, including: Extracting a current hot air local time series association feature vector from the sequence of hot air local time series association feature vectors, and defining other hot air local time series association feature vectors in the sequence of hot air local time series association feature vectors as historical hot air local time series association feature vectors to obtain a sequence of historical hot air local time series association feature vectors; Calculate the hot air local time series static energy value of each historical hot air local time series associated characteristic vector in the sequence of the historical hot air local time series associated characteristic vectors; Determine the hot air local time series energy spatial attenuation factor based on the spatial span coefficient between each of the historical hot air local time series associated characteristic vectors and the current hot air local time series associated characteristic vector; Determine the hot air local time series energy time attenuation factor based on the time span coefficient between each of the historical hot air local time series associated characteristic vectors and the current hot air local time series associated characteristic vector; Based on the hot air local time series energy spatial attenuation factor and the hot air local time series energy temporal attenuation factor, the hot air local time series static energy value of each of the historical hot air local time series associated characteristic vectors is temporally modulated to obtain the hot air local time series dynamic energy spatiotemporal transfer value of each of the historical hot air local time series associated characteristic vectors relative to the current hot air local time series associated characteristic vector; Inputting the hot air local time series dynamic energy spatiotemporal transfer values ​​of the historical hot air local time series associated feature vectors into the energy transfer gating module based on the mask function to obtain a sequence of hot air local time series dynamic energy spatiotemporal transfer weights; Based on the sequence of the spatiotemporal transfer weights of the local time series dynamic energy of the hot air, a weighted sum of the sequence of the historical local time series associated feature vectors of the hot air is calculated to obtain a historical local time series node feature transfer aggregation modulation vector; Calculate the positional sum of the historical hot air local time series node feature transfer aggregation modulation vector and the current hot air local time series associated feature vector to obtain the hot air temperature spatiotemporal transfer aggregation representation vector; Calculating the hot air local time series static energy value of each historical hot air local time series associated characteristic vector in the sequence of the historical hot air local time series associated characteristic vectors comprises: Calculating the mean and standard deviation of the historical hot air local time series associated characteristic vectors respectively to obtain the historical hot air local time series characteristic mean and the historical hot air local time series standard deviation; After calculating the fourth power of the position difference between the historical hot air local time series associated feature vector and the historical hot air local time series feature mean, the calculated fourth power modulates the expected value of the historical hot air local time series offset vector to obtain the hot air local time series feature offset expected factor; After calculating the division between the expected factor of the local time series characteristic deviation of the hot air and the fourth power of the standard deviation of the historical local time series of the hot air, the result is added to the maximum eigenvalue in the associated eigenvector of the historical local time series of the hot air to obtain the static energy value of the local time series of the hot air; Determining the hot air local time series energy spatial attenuation factor based on the spatial span coefficient between each of the historical hot air local time series associated characteristic vectors and the current hot air local time series associated characteristic vector includes: Calculating the spatial distance span between the historical hot air local time series associated characteristic vector and the current hot air local time series associated characteristic vector to obtain the historical hot air local time series spatial span coefficient; Calculate the division between the hot air local time series static energy value corresponding to the historical hot air local time series associated characteristic vector and the historical hot air local time series spatial span coefficient to obtain the hot air local time series energy spatial attenuation factor; Determining the hot air local time series energy time attenuation factor based on the time span coefficient between each of the historical hot air local time series associated characteristic vectors and the current hot air local time series associated characteristic vector includes: After calculating the subtraction between the timestamp of the current hot air local time series associated feature vector and the timestamp of the historical hot air local time series associated feature vector, the obtained timestamp value is rounded down to obtain the hot air local time series time span coefficient; After dividing the hot air local time series time span coefficient by the inverse scale parameter controlling the time decay period, the obtained time span decay representation coefficient is used as an exponential power, and the natural exponential function value with the natural constant e as the base is calculated to obtain the time span decay class support representation coefficient; Calculate the division between the hot air local time series static energy value and the time span attenuation class support representation coefficient to obtain the hot air local time series energy time attenuation factor; Based on the hot air local time series energy spatial attenuation factor and the hot air local time series energy time attenuation factor, the hot air local time series static energy value of each of the historical hot air local time series associated characteristic vectors is temporally modulated to obtain the hot air local time series dynamic energy spatiotemporal transfer value of each of the historical hot air local time series associated characteristic vectors relative to the current hot air local time series associated characteristic vector, including: calculating the position-weighted sum between the hot air local time series energy spatial attenuation factor and the hot air local time series energy time attenuation factor to obtain the hot air local time series dynamic energy spatiotemporal transfer value; The hot air local time series dynamic energy spatiotemporal transfer value of each historical hot air local time series associated feature vector is input into the energy transfer gating module based on the mask function to obtain a sequence of hot air local time series dynamic energy spatiotemporal transfer weights, including: in the energy transfer gating module based on the mask function, in response to the hot air local time series dynamic energy spatiotemporal transfer value being greater than a threshold hyperparameter, The function normalizes the hot air local temporal dynamic energy spatiotemporal transfer value greater than the threshold hyperparameter to obtain the hot air local temporal dynamic energy spatiotemporal transfer weight; Based on the hot air temperature time-space transfer aggregation representation vector, a control instruction is determined, and the control instruction includes a decoded value of the injection valve opening recommendation value, including: inputting the hot air temperature time-space transfer aggregation representation vector into a decoder-based injection valve opening controller to obtain a control instruction.

2. A salt-resistant and synergistic fertilizer, characterized in that: The salt-resistant and synergistic fertilizer is prepared by the preparation method of the salt-resistant and synergistic fertilizer according to claim 1.

Images