Method and device for predicting time for forming insoluble barrier by salt aggregation
By monitoring the process of salt aggregation biopack, the exchange cluster and neural network model are constructed to predict the time spent on salt aggregation to form insoluble barriers, which solves the problem of difficult prediction of salt aggregation time in high-salt soils, and improves the reliability of soil improvement and plant planting.
Patent Information
- Application Number
- CN202411538028.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-31
- Publication Date
- 2025-02-25
AI Technical Summary
During the ecological restoration process, the time when salt aggregation in high-salt soil forms insoluble barriers is difficult to accurately predict, affecting plant planting and soil improvement.
By monitoring the insoluble barrier process of the sample salt aggregation biological packet, the exchange group and neural network model are constructed, and the target neural network model with the smallest root mean square difference is screened to predict the time taken for salt aggregation to form insoluble barrier.
Accurate prediction of the insoluble barrier time of salt aggregation formation is achieved, and the reliability and efficiency of soil improvement and plant planting are improved.
Smart Images
Figure CN119514763A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method and device for predicting the time taken for salt to aggregate to form an insoluble barrier. Background Art
[0002] During ecological restoration, high-salt soils are often encountered. Plants with higher salt tolerance are usually selected for planting, but there are certain limits. Moreover, the salt tolerance of plant seedlings is relatively low. It is usually necessary to wash the salt with water to reduce the salinity of the soil to below the salt tolerance of the planted plants, so that the plants can be planted without being affected by salt. Phosphates can be used to form salt-tolerant biological bags with salt aggregation produced by salt in the soil, blocking further entry of salt. Among them, phosphates react with salt ions in multiple stages to form insoluble substances, and insoluble substances are formed in the outer layer of the matrix to form an outer layer with relatively poor water permeability, which reduces the infiltration of salt in the soil and slows down the penetration rate of salt in saline-alkali soil. The phosphate substances (monoammonium phosphate, diammonium phosphate, potassium dihydrogen phosphate) in the biological bag are themselves acid-base buffer systems, which stabilize the pH of the rhizosphere soil of plant seedlings.
[0003] The practical application problem to be solved is to use the salt in the soil to form a biological bag through the replacement process to block the entry of new salt. This is a complex multi-level bootstrap process. The formation time is nonlinear and the measurement is unstable, making it difficult to estimate directly. Summary of the invention
[0004] The purpose of the present application is to provide a method and device for predicting the time taken for salt to aggregate and form an insoluble barrier.
[0005] To achieve the above objectives, this application provides the following solutions:
[0006] In a first aspect, the present application provides a method for predicting the time taken for salt to aggregate and form an insoluble barrier, comprising:
[0007] The process of forming an insoluble barrier by the sample salt-aggregating biological package is monitored to obtain a sample data set, wherein the sample data set is composed of sample soil salt content data corresponding to the sample salt-aggregating biological package at different time points;
[0008] constructing an exchange group based on the sample data set;
[0009] Retrieving all non-applicable data sequences from the exchange group;
[0010] Constructing a neural network model corresponding to each of the non-applicable data sequences, comparing the root mean square deviations of the prediction results of each of the neural network models, and selecting a target neural network model with the smallest root mean square deviation;
[0011] The target soil salt content data of the target salt-aggregating biological package is input into the target neural network model to obtain the target time used by the target salt-aggregating biological package to form an insoluble barrier.
[0012] Optionally, the step of monitoring the process of sample salt aggregation biological bag forming an insoluble barrier to obtain a sample data set includes:
[0013] The process of forming an insoluble barrier by the sample salt-aggregating biological bag is monitored at preset time intervals to obtain the corresponding sample soil salt content data at different time points;
[0014] The sample soil salt content data are arranged according to time points, and the obtained matrix is used as a sample data set.
[0015] Optionally, the step of constructing an exchange group based on the sample data set includes:
[0016] A permutation group (G, *) is constructed based on the sample data set, where G is a set (S', Y) of column permutations of S, where X = (S'), Y = (T), T is a time point, and * is the column permutation k modulo n of the element a in G, and the column corresponding to the result of the modulo division is permuted with Y.
[0017] Optionally, the step of acquiring all non-applicable data sequences from the exchange group includes:
[0018] All non-applicable data columns are selected for all elements in the exchange group to obtain all non-applicable data sequences.
[0019] Optionally, the step of constructing a neural network model corresponding to each of the non-applicable data sequences, comparing the root mean square deviation of the prediction results of each of the neural network models, and selecting a target neural network model with the smallest root mean square deviation comprises:
[0020] Starting a plurality of threads, each of the threads competes for and modifies the non-applicable data sequence, and constructing a neural network model using the permuted matrix;
[0021] The root mean square differences of the prediction results of each of the neural network models are compared to screen out the target neural network model with the smallest root mean square difference, and the target neural network model satisfies the suboptimal learning result of the fixed point F(Xi, Y) = F'(Xi, Y) = Xi, where X = (S') and Y = (T).
[0022] Optionally, the step of inputting the target soil salt content data of the target salt-aggregating biological package into the target neural network model to obtain the target time taken by the target salt-aggregating biological package to form an insoluble barrier comprises:
[0023] Obtain target soil salinity data for target salt-aggregating bio-packages;
[0024] A data sequence obtained by screening out all non-applicable columns in the target soil salinity data;
[0025] The target soil salt content data after screening is input into the target neural network model to obtain the target time used by the target salt-aggregating biological bag to form an insoluble barrier.
[0026] In a second aspect, the present application provides a device for predicting the time taken for salt to aggregate and form an insoluble barrier, comprising:
[0027] An acquisition module, used to monitor the process of the sample salt aggregation biological package forming an insoluble barrier to obtain a sample data set, wherein the sample data set is composed of sample soil salt content data corresponding to the sample salt aggregation biological package at different time points;
[0028] A processing module, configured to construct an exchange group based on the sample data set;
[0029] Retrieving all non-applicable data sequences from the exchange group;
[0030] Constructing a neural network model corresponding to each of the non-applicable data sequences, comparing the root mean square deviations of the prediction results of each of the neural network models, and selecting a target neural network model with the smallest root mean square deviation;
[0031] The prediction module is used to input the target soil salt content data of the target salt-aggregating biological package into the target neural network model to obtain the target time used by the target salt-aggregating biological package to form an insoluble barrier.
[0032] Optionally, the acquisition module is further used to:
[0033] The process of forming an insoluble barrier by the sample salt-aggregating biological bag is monitored at preset time intervals to obtain the corresponding sample soil salt content data at different time points;
[0034] The sample soil salt content data are arranged according to time points, and the obtained matrix is used as a sample data set.
[0035] Optionally, the processing module is further configured to:
[0036] A permutation group (G, *) is constructed based on the sample data set, where G is a set (S', Y) of column permutations of S, where X = (S'), Y = (T), T is a time point, and * is the column permutation k modulo n of the element a in G, and the column corresponding to the result of the modulo division is permuted with Y.
[0037] Optionally, the processing module is further configured to:
[0038] All non-applicable data columns are selected for all elements in the exchange group to obtain all non-applicable data sequences.
[0039] Optionally, the processing module is further configured to:
[0040] Starting a plurality of threads, each of the threads competes for and modifies the non-applicable data sequence, and constructing a neural network model using the permuted matrix;
[0041] The root mean square differences of the prediction results of each of the neural network models are compared to screen out the target neural network model with the smallest root mean square difference, and the target neural network model satisfies the suboptimal learning result of the fixed point F(Xi, Y) = F'(Xi, Y) = Xi, where X = (S') and Y = (T).
[0042] Optionally, the prediction module is further used to:
[0043] Obtain target soil salinity data for target salt-aggregating bio-packages;
[0044] A data sequence obtained by screening out all non-applicable columns in the target soil salinity data;
[0045] The target soil salt content data after screening is input into the target neural network model to obtain the target time used by the target salt-aggregating biological bag to form an insoluble barrier.
[0046] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for predicting the time taken for salt aggregation to form an insoluble barrier as described in any one of the above.
[0047] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for predicting the time taken for salt aggregation to form an insoluble barrier as described in any one of the above.
[0048] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the method for predicting the time taken for salt aggregation to form an insoluble barrier as described in any one of the above.
[0049] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0050] The present application provides a method and device for predicting the time taken for salt aggregation to form an insoluble barrier. The method uses a sample data set collected from a salt aggregation biological package to construct a permutation of an exchange group. The regularity and traversability of the exchange group are then used to simulate the process of the salt aggregation biological package forming an insoluble barrier using a neural network model. The neural network models are then allowed to compete in search and solution to obtain an accurate estimate of the time taken for salt aggregation. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0052] Figure 1 A schematic flow chart of a method for predicting the time taken for salt to aggregate and form an insoluble barrier provided in one embodiment of the present application;
[0053] Figure 2 A schematic diagram of functional modules of a device for predicting the time taken for salt to aggregate and form an insoluble barrier provided in one embodiment of the present application;
[0054] Figure 3 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION
[0055] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0056] like Figure 1 As shown, some embodiments of the present application provide a method for predicting the time taken for salt to aggregate and form an insoluble barrier, and in the embodiments of the present application, the method includes the following steps 101 to 105. Among them:
[0057] Step 101, monitoring the process of the sample salt-aggregating biological package forming an insoluble barrier to obtain a sample data set, wherein the sample data set is composed of sample soil salt content data corresponding to the sample salt-aggregating biological package at different time points.
[0058] In an embodiment of the present application, the process of sample salt aggregation bio-bags forming an insoluble barrier is monitored in real time to obtain soil salinity data at different time points. This ensures that enough data points can be obtained to build an accurate sample data set. The sample data set consists of these soil salinity data collected at different time points. The soil salinity data corresponding to each time point represents an observation point, and these observation points together constitute a complete data set for the entire monitoring process.
[0059] Step 102: construct an exchange group based on the sample data set.
[0060] It should be noted that the commutative group is a group that consists of its own set G and the binary operation *<G,*> In addition to satisfying the general group axioms, namely the associative law of operations, G has an identity element, and all elements of G have inverses, it also satisfies the commutative law axiom.<G,*> , you only need to swap Y with the column with ordinal number i in x0, x1, x2, ..., Y to become x0, x1, ..., xi-1, Y, xi+1..., xi. G is the set formed by swapping the columns of x0, x1, x2, ..., Y, and * is the result of learning. g1 is the result of swapping the column of Y with ordinal number t1, g2 is the result of swapping the column of Y with ordinal number t2, and g1*g2 is the result of swapping the sum of the two columns with ordinal numbers t1 and t2 (t1+t2) modulo the total number of columns with ordinal number and Y.<G,*> The rules ensure its regularity and ergodicity.
[0061] Since the exchange group reflects the distribution of the data under permutation (the exchange group permutation will multiply the amount of data several times), it will look more similar to their population. Now by simulating the actual salt permutation process, the multi-level competition process triggered, filtering out those irregular, unstable, and noisy data, we will get a relatively accurate distribution, enough to find the rules.
[0062] All data (x0, x1, x2, ...) constitute a low-dimensional continuous sequence. After proving that all elements of G are permuted to meet the learning results, they constitute a high-dimensional segmented continuous sequence. According to the proof process of Brouwer's fixed point theorem, if the existence of a fixed point is confirmed, the fusion of high-dimensional and low-dimensional data is achieved, bringing suboptimal conditions and increasing the search space. In the figure below, the horizontal line represents the bottom straight line or the line segment Li after the low-dimensional data is broken, which is the previous broken line segment plus 1, that is, Li+1=Li+1 (starting from L0, non-zero), and the vertical line represents the mapping of the trapezoidal matrix sequence to the bottom straight line. In this application, we need to confirm that the mapping of the horizontal line segment in the trapezoidal matrix to the bottom straight line satisfies the fixed point F'(xi)=xi.
[0063] In the embodiment of the present application, a mathematical commutative group (G, *) is constructed based on the sample data set. This commutative group is for further processing and analysis of the data set, so as to extract useful information from complex data. G is a set (S', Y) consisting of column permutations of S, where X = (S'), Y = (T), T is a time point, and * is the modulo division of the column permutation k of the element a in G by n.
[0064] Step 103, obtaining all non-applicable data sequences from the exchange group.
[0065] In the embodiment of the present application, all inapplicable data sequences are selected from the constructed exchange group, thereby excluding those data sequences that are useless for subsequent model construction, thereby improving the accuracy and efficiency of the model. Specifically, all elements in the exchange group can be traversed, and the inapplicable data columns therein can be selected respectively to form an inapplicable data sequence.
[0066] Step 104, construct a neural network model corresponding to each of the non-applicable data sequences, compare the root mean square deviations of the prediction results of each of the neural network models, and screen out the target neural network model with the smallest root mean square deviation.
[0067] In the embodiment of the present application, a corresponding neural network model is constructed for each non-applicable data sequence. Due to the diversity of non-applicable data sequences, each model is optimized and trained for the specific situation of a specific data sequence. Multiple threads are started to compete and modify the non-applicable data sequence, and the neural network model is constructed using the permuted matrix.
[0068] Compare the prediction results of each neural network model, using the root mean square deviation (RMSD) as the evaluation indicator. The smaller the RMSD, the closer the model's prediction results are to the actual situation, and the more accurate the model is. Filter out the target neural network model with the smallest RMSD. This model will be used in the subsequent prediction process. Make sure that the target neural network model meets specific learning conditions, such as the suboptimal result of the fixed point F(Xi, Y) = F'(Xi, Y) = Xi, to ensure the stability and reliability of the model.
[0069] Step 105, inputting the target soil salt content data of the target salt-aggregating biological package into the target neural network model, and obtaining the target time used by the target salt-aggregating biological package to form an insoluble barrier.
[0070] In the embodiment of the present application, the target soil salt content data of the target salt-aggregating biological package is input into the selected target neural network model. Before input, the data may need to be screened to remove those data columns that are not suitable for the model. Through model calculation, the target time used by the target salt-aggregating biological package to form an insoluble barrier is obtained.
[0071] The embodiment of the present application constructs a permutation of an exchange group by using a sample data set collected for a salt-aggregating biological package, and then utilizes the regularity and traversability of the exchange group to simulate the process of the salt-aggregating biological package forming an insoluble barrier using a neural network model, and allows the neural network models to conduct a competitive search and solution to obtain an accurate estimate of the time it takes for salt aggregation.
[0072] Optionally, the step 101 includes:
[0073] Step 1011, the process of the sample salt agglomeration biological bag forming an insoluble barrier is monitored at preset time intervals to obtain corresponding sample soil salt content data at different time points.
[0074] Step 1012: Arrange the sample soil salt content data according to time points, and use the obtained matrix as a sample data set.
[0075] In the embodiment of the present application, the whole process of the sample salt aggregation bio-bag forming an insoluble barrier can be regularly monitored. The preset time interval can be a few minutes, a few hours, a few days or longer, depending on the needs of monitoring and the accuracy requirements of the research design. At each preset time interval, the soil salinity of the sample is measured once, and the salt content data at that time point is recorded. In this way, over time, a series of soil salinity data points at different time points will be accumulated.
[0076] Through the above regular monitoring, a series of soil salinity data that changes over time can be obtained. These data points reflect the impact of salt-aggregating biological packages on soil salinity at different stages and are the basis for subsequent analysis. The soil salinity data at each time point are independent, but they are connected in series through the timeline to form a complete data sequence.
[0077] After obtaining the soil salinity data at all time points, the next step is to arrange the data in chronological order. This is done to ensure the consistency and continuity of the data in the time dimension, which is convenient for subsequent data analysis and model building. The arranged data will form a time series, and each time point corresponds to a specific salt content value.
[0078] The soil salinity data arranged in chronological order are organized into a matrix, where each row represents a time point and each column represents a different observation indicator (in this scenario, there may be only one column, i.e., soil salinity). This matrix is the sample data set for subsequent data analysis. With this data set, researchers can conduct more in-depth analysis and modeling of the process of salt-aggregating biota forming insoluble barriers.
[0079] Optionally, step 102 includes: constructing a permutation group (G, *) based on the sample data set, wherein G is a set (S', Y) consisting of column permutations of S, wherein X = (S'), Y = (T), T is a time point, and * is the column permutation k modulo n of the element a in G, and the column corresponding to the result of the modulo division is permuted with Y.
[0080] It should be noted that the commutative group (G, *): is an algebraic structure consisting of a set of elements and a binary operation, which satisfies the associative law, has an identity element, and each element has an inverse. The commutative group here refers to the group whose operations between elements satisfy the commutative law. G is the group, and * is the binary operation on the interaction group. Column permutation refers to the operation of rearranging the columns of a matrix. In algorithms and data structures, column permutation is often used for rearrangement or randomization of data. Modulo division is a mathematical operation, equivalent to finding the remainder. Given two integers a and n (n is not equal to 0), the result of modulo division of a by n is the remainder after dividing the two numbers.
[0081] G is a set consisting of column permutations of the sample data set S. Here, S' can be regarded as a set of columns obtained after S has undergone some transformation (possibly rearrangement or other mathematical transformations). Y is a set of time points, usually expressed as a vector or list, which contains all observed time points T. X = (S'), where X is expressed as another notation of S', probably for the sake of being more concise or in line with a specific context in subsequent mathematical or algorithmic representations. Y = (T), where Y directly represents the set of time points T, that is, all observed time points.
[0082] Specifically, all possible column permutations are generated from the sample data set S, and these column permutations constitute the set G. For each element in G (i.e., each column permutation method), a modulo operation is performed, and the result determines which columns will be permuted with the time point set Y. The modulo operation here is used to generate an index for selecting a specific column from the permuted column set. Through the binary operation *, the elements in G can operate on each other to generate new elements or results, satisfying the properties of the commutative group.
[0083] Optionally, the step 103 includes: selecting all non-applicable data columns for all elements in the exchange group respectively to obtain all non-applicable data sequences.
[0084] In the present embodiment, non-applicable data columns refer to data columns that are not suitable for direct use in subsequent analysis during data analysis or model building due to missing data, outliers, failure to meet specific conditions, etc. These data columns may have a negative impact on the accuracy of the prediction results, and therefore need to be identified and excluded.
[0085] It is necessary to traverse all elements in the exchange group G. Each element represents an arrangement of data columns. For each arrangement, identify the inapplicable data columns according to certain rules or conditions (such as data integrity check, outlier detection, etc.). This process may involve statistical tests, threshold judgments, or other data analysis techniques. Collect the inapplicable data columns identified in each arrangement to prepare for subsequent steps. For each arrangement, combine the inapplicable data columns in a certain order (such as the order in the original data set) to form an inapplicable data sequence. Repeat the above process until all elements in the exchange group G have been traversed, thereby obtaining all possible inapplicable data sequences.
[0086] Optionally, the step 104 includes:
[0087] Step 1041, start multiple threads, each of which competes and modifies the non-applicable data sequence, and uses the permuted matrix to build a neural network model.
[0088] Step 1042, compare the root mean square differences of the prediction results of each of the neural network models, screen out the target neural network model with the smallest root mean square difference, and the target neural network model satisfies the suboptimal learning result of the fixed point F(Xi, Y) = F'(Xi, Y) = Xi, where X = (S'), Y = (T).
[0089] In an embodiment of the present application, each thread attempts to modify the non-applicable data sequence. Each thread may modify the non-applicable data sequence in a different way, such as by filling missing values, smoothing or feature transformation, in an attempt to improve the data quality and make it more suitable for model training. The modified data (i.e., the permuted matrix) is used as the input of the neural network model. Based on the modified data, each thread independently builds a neural network model. These models have the same structure, but due to the different input data, they learn different feature representations during the training process.
[0090] After training, each neural network model will make predictions on the same set of validation data and calculate the root mean square difference between the predictions and the true values. RMSD is a common indicator for measuring the prediction accuracy of a model. The smaller the value, the more accurate the model prediction. The RMSD of all models is compared, and the model with the smallest RMSD is selected as the target neural network model.
[0091] Optionally, the step 105 includes:
[0092] Step 1051, obtaining target soil salt content data of the target salt-aggregating biological package.
[0093] Step 1052: Screen out all inapplicable columns in the target soil salinity data to obtain a data sequence.
[0094] Step 1053, input the screened target soil salt content data into the target neural network model to obtain the target time taken by the target salt-aggregating biological package to form an insoluble barrier.
[0095] In an embodiment of the present application, it is necessary to obtain the target soil salinity data of the target salt-aggregating biological package from the relevant data source. These data may include soil salinity values measured at different time points, which are key input information for predicting the time to form an insoluble barrier. After obtaining the target soil salinity data, it is next necessary to screen out all non-applicable columns in the data. After screening out non-applicable columns, the obtained data sequence needs to be further preprocessed to ensure that it meets the input requirements of the target neural network model. The preprocessing steps may include normalization, standardization, missing value filling, etc., to eliminate the influence of different dimensions and improve the training efficiency and prediction performance of the model.
[0096] The pre-processed target soil salinity data is input into the target neural network model determined through multiple rounds of screening and evaluation. This model is trained based on a large amount of sample data and can capture the complex relationship between soil salinity changes and the time to form an insoluble barrier. The target neural network model outputs the target time for the target salt-aggregating biomass to form an insoluble barrier based on the input target soil salinity data. This time prediction result is of great significance for guiding actual production activities and optimizing agricultural irrigation plans.
[0097] It is worth noting that the formation of an insoluble barrier by the target salt-aggregating bio-package mentioned in this application is a salt-to-salt replacement process, which in turn blocks the penetration of salt, similar to a bootstrap process, which is complex and cannot be analyzed using conventional divide-and-conquer analysis. Conventional divide-and-conquer analysis methods require the assumption of a staged process, which leads to a large amount of human error.
[0098] The solution provided by the present application is to use the suboptimal conditions under the fixed point to filter out unstable and noisy data, and obtain enough data sequences in the competition process through the exchange group permutation to meet the randomness, convergence and uniformity and bring convergence. At the same time, when judging the fixed point, the present application uses the method of filtering out the unstable data in the early stage, leaving the data with stable distribution, and then estimating its own stopping probability, which brings about a cyclic process similar to the exchange group. This is also an information bottleneck process. The fixed point is the information bottleneck that must be met. Under this bottleneck, the filtered information is actually sent back for processing. The continuous cycle brings dissipative structure, forming dimensionality and self-organization. It can be realized by test-driven development process. The necessary condition for using the test-driven development process is that there is an overfitting relationship between the output and the input, that is, the output changes a little bit, and the input changes on a large scale. The output data here is very simple, that is, the reaction time, plus the complex reaction process, which just meets the conditions of the test-driven development process.
[0099] This application can be interpreted as first using the local assumption method to confirm the fixed point, and then screening out individuals with stable correlations, so that "good" data can be continuously selected to increase its influence. Other conventional methods are to continuously expand the sample range, increase representative samples, achieve a larger search range, and gain advantages (relying on pairing between samples with more distant relationships to obtain better offspring). This application is the other way around, using a filtering process to screen out some "bad" data and increase the influence of "good" data.
[0100] The main reasons mentioned in this application may be the following four points: the data has unstable parts; the parameters are too sensitive; the model is too complicated; the influence of various factors in the process of training the model is uneven.
[0101] Generally speaking, there are several common solutions: Get more representative data: Get more data from the data source; Use a suitable model: Reduce the number of network layers, the number of neurons, etc., which can limit the network's fitting ability; Dropout; Data cleaning: Correct incorrect annotations or delete incorrect data; Combine multiple models.
[0102] This application mainly relies on the Markov process composed of stable random data remaining after filtering out the previous unstable data. Since the Markov process is a process that uses a relatively simple time series as input to generate random values, it obtains the required results through simple and feasible strategies, which is exactly the same as neural network learning, which is a method that can learn with the help of simple strategies to achieve optimization goals. Markov processes bring a wide range of applications. However, the possible shortcomings of these methods still cannot solve the problem of parameter sensitivity. For example, all data cannot be obtained before knowing the learning results, but are obtained gradually and progressively. Another problem is that the output result is a random distribution, and it is impossible to quickly make the average value or the value after simple calculation and conversion converge to a predicted expected result. Finally, nonlinearity increases the complexity of calculation, and some factors cannot achieve the maximum influence (i.e., the maximum output or growth rate of each factor in the calculation), which is easily ignored in the calculation process, and constantly brings about the phenomenon of parameter oversensitivity.
[0103] The phenomenon of parameter oversensitivity is that the dimension that the predetermined model can express exceeds the dimension of the currently available data. A conventional solution to overfitting is to "reduce the dimension" of the model, optimize the model parameters, and produce a suitable low-dimensional model. However, if you want to build an excellent global processing model, excessive dimensionality reduction will eventually lose the model's adaptability.
[0104] The solution of this application is to first use the local assumption method to confirm the fixed point and then filter out the individuals that are not related. This application filters out a lot of data in a large range because these data may be unstable, and uses adjacent data or the local sample data remaining after the local data model is carefully selected to perform local dimensionality reduction. Specifically, the measured local data sequence is continuously fused with itself (or the sample data with the closest kinship) to produce a local dimensionality reduction or simplified sequence, and then they are used for fitting (including obtaining a reasonable estimate of the distribution without bias through an adaptive process).
[0105] From the local data set, after confirming the fixed point, we can get the output that covers the whole world. It can also bring about the invariant projection transformation, for example, rotating an image will not change the distance from a point to the curve. From the local data set, according to the invariant projection transformation process, we can also get the output that covers the whole world. Fixed point is a kind of invariance. For example, in the dynamic system, invariance can be expressed by the conservation of kinetic energy and momentum corresponding to the motion of rigid objects. The local data set is the corresponding locality assumption, which is the principle of locality. In physics, the principle of locality, also known as the principle of locality or the regional principle, holds that a specific object can only be affected by the forces around it. A physics theory that includes the principle of locality is called a local theory. According to the field theory of classical physics, the action of a certain point affects another point, and the space in between, such as the field, will become the medium of motion. To affect another point, a wave or a particle must first pass through the space between the two points before it can cause an impact. According to the principle of locality, from the state of a certain local area in the past, the events that occurred in the local area can be used to calculate the current state of the local area using the model of the local area. That is to say, due to locality, local data and local models are the same as global data and global models. This is because the fixed point brings the invariance of global correlation relations. In this physics example, it is the law of conservation of kinetic energy and momentum. The invariance brought by this fixed point can bring enough related equations. In other words, under these invariance relations, the input and output of the system satisfy a one-to-one relationship, that is, the relationship between the states. So it becomes A→A, or the self-mapping of the set A, which brings a long enough self-mapping data sequence, satisfying the principle of asymptotic equidivision. Therefore, when each input X in the local area is long enough (the number of self-mappings is large enough), it will produce an independent and equally probable output distribution, which is a global distribution relationship. These relationships can be checked and confirmed from the data changes of the system by the process of discovering fixed points.
[0106] This approach is equivalent to using a group G that represents geometric rotation transformations.
[0107] According to the definition of the commutative group, there must be a pattern in the commutative group G, such as the mapping from two dimensions to one dimension. First, because the group G is formed by simple and consistent operations, according to Gödel's theorem, there must be incompleteness, that is, there are elements that cannot be generated by simple operations, which contradicts the definition of the commutative group G. The two form a sequence of ∽A, A, ∽A, A, ∽A, A, ∽A, A, where A is complete and ∽A is incomplete. In the continuous cycle, dissipative structures are brought about, self-organization is carried out, and the result of dimensionality increase is obtained.
[0108] Since the commutative group G has regularities, the Harsanyi transformation can be constructed through a competition process. Assuming that the commutative group G brings a "natural" distribution, there is a cycle underneath this distribution. The process of continuous learning and optimization can eliminate the uncertainty of the parameters and obtain the model actually needed.
[0109] Based on the same inventive concept, the embodiment of the present application also provides a device for predicting the time taken by salt aggregation to form an insoluble barrier, which is used to implement the method for predicting the time taken by salt aggregation to form an insoluble barrier mentioned above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of one or more devices for predicting the time taken by salt aggregation to form an insoluble barrier provided below can be referred to the limitations of the method for predicting the time taken by salt aggregation to form an insoluble barrier above, and will not be repeated here.
[0110] In an exemplary embodiment, Figure 2 As shown, a device 20 for predicting the time taken for salt to aggregate and form an insoluble barrier is provided, comprising:
[0111] An acquisition module 201 is used to monitor the process of the sample salt aggregation biological package forming an insoluble barrier to obtain a sample data set, wherein the sample data set is composed of sample soil salt content data corresponding to the sample salt aggregation biological package at different time points;
[0112] A processing module 202 is used to construct an exchange group based on the sample data set;
[0113] Retrieving all non-applicable data sequences from the exchange group;
[0114] Constructing a neural network model corresponding to each of the non-applicable data sequences, comparing the root mean square deviations of the prediction results of each of the neural network models, and selecting a target neural network model with the smallest root mean square deviation;
[0115] The prediction module 203 is used to input the target soil salt content data of the target salt-aggregating biological package into the target neural network model to obtain the target time used by the target salt-aggregating biological package to form an insoluble barrier.
[0116] Optionally, the acquisition module 201 is further used to:
[0117] The process of forming an insoluble barrier by the sample salt-aggregating biological bag is monitored at preset time intervals to obtain the corresponding sample soil salt content data at different time points;
[0118] The sample soil salt content data are arranged according to time points, and the obtained matrix is used as a sample data set.
[0119] Optionally, the processing module 202 is further configured to:
[0120] A permutation group (G, *) is constructed based on the sample data set, where G is a set (S', Y) of column permutations of S, where X = (S'), Y = (T), T is a time point, and * is the column permutation k modulo n of the element a in G, and the column corresponding to the result of the modulo division is permuted with Y.
[0121] Optionally, the processing module 202 is further configured to:
[0122] All non-applicable data columns are selected for all elements in the exchange group to obtain all non-applicable data sequences.
[0123] Optionally, the processing module 202 is further configured to:
[0124] Starting a plurality of threads, each of the threads competes for and modifies the non-applicable data sequence, and constructing a neural network model using the permuted matrix;
[0125] The root mean square differences of the prediction results of each of the neural network models are compared to screen out the target neural network model with the smallest root mean square difference, and the target neural network model satisfies the suboptimal learning result of the fixed point F(Xi, Y) = F'(Xi, Y) = Xi, where X = (S') and Y = (T).
[0126] Optionally, the prediction module 203 is further used for:
[0127] Obtain target soil salinity data for target salt-aggregating bio-packages;
[0128] A data sequence obtained by screening out all non-applicable columns in the target soil salinity data;
[0129] The target soil salt content data after screening is input into the target neural network model to obtain the target time used by the target salt-aggregating biological bag to form an insoluble barrier.
[0130] The embodiment of the present application constructs a permutation of the exchange group by using a sample data set collected for the salt-aggregating biological package, and then utilizes the regularity and traversability of the exchange group. The model uses a neural network model to simulate the process of the salt-aggregating biological package forming an insoluble barrier, and allows the neural network models to conduct competitive search and solution to obtain an accurate estimate of the time taken for salt aggregation.
[0131] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 3As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store predicted data on the time taken for salt to aggregate to form an insoluble barrier. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for predicting the time taken for salt to aggregate to form an insoluble barrier is implemented.
[0132] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0133] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.
[0134] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0135] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0136] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0137] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0138] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.
[0139] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0140] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A method for predicting the time taken for salt to aggregate and form an insoluble barrier, characterized in that: The method for predicting the time taken for the salt to aggregate and form an insoluble barrier includes: The process of forming an insoluble barrier by the sample salt-aggregating biological package is monitored to obtain a sample data set, wherein the sample data set is composed of sample soil salt content data corresponding to the sample salt-aggregating biological package at different time points; constructing an exchange group based on the sample data set; Retrieving all non-applicable data sequences from the exchange group; Constructing a neural network model corresponding to each of the non-applicable data sequences, comparing the root mean square deviations of the prediction results of each of the neural network models, and selecting a target neural network model with the smallest root mean square deviation; The target soil salt content data of the target salt-aggregating biological package is input into the target neural network model to obtain the target time used by the target salt-aggregating biological package to form an insoluble barrier.
2. The method for predicting the time taken for salt to aggregate and form an insoluble barrier according to claim 1, characterized in that: The step of monitoring the process of sample salt aggregation biological bag forming an insoluble barrier to obtain a sample data set includes: The process of forming an insoluble barrier by the sample salt-aggregating biological bag is monitored at preset time intervals to obtain the corresponding sample soil salt content data at different time points; The sample soil salt content data are arranged according to time points, and the obtained matrix is used as a sample data set.
3. The method for predicting the time taken for salt to aggregate and form an insoluble barrier according to claim 1, characterized in that: The step of constructing an exchange group based on the sample data set includes: A permutation group (G, *) is constructed based on the sample data set, where G is a set (S', Y) of column permutations of S, where X = (S'), Y = (T), T is a time point, and * is the column permutation k modulo n of the element a in G, and the column corresponding to the result of the modulo division is permuted with Y.
4. The method for predicting the time taken for salt to aggregate and form an insoluble barrier according to claim 1, characterized in that: The step of obtaining all non-applicable data sequences from the exchange group comprises: All non-applicable data columns are selected for all elements in the exchange group to obtain all non-applicable data sequences.
5. The method for predicting the time taken for salt to aggregate and form an insoluble barrier according to claim 1, characterized in that: The step of constructing a neural network model corresponding to each of the non-applicable data sequences, comparing the root mean square deviation of the prediction results of each of the neural network models, and selecting a target neural network model with the smallest root mean square deviation comprises: Starting a plurality of threads, each of the threads competes for and modifies the non-applicable data sequence, and constructing a neural network model using the permuted matrix; The root mean square differences of the prediction results of each of the neural network models are compared to screen out the target neural network model with the smallest root mean square difference, and the target neural network model satisfies the suboptimal learning result of the fixed point F(Xi, Y) = F'(Xi, Y) = Xi, where X = (S') and Y = (T).
6. The method for predicting the time taken for salt to aggregate and form an insoluble barrier according to claim 1, characterized in that: The step of inputting the target soil salt content data of the target salt-aggregating biological package into the target neural network model to obtain the target time used by the target salt-aggregating biological package to form an insoluble barrier comprises: Obtain target soil salinity data for target salt-aggregating bio-packages; A data sequence obtained by screening out all non-applicable columns in the target soil salinity data; The target soil salt content data after screening is input into the target neural network model to obtain the target time used by the target salt-aggregating biological bag to form an insoluble barrier.
7. A device for predicting the time taken for salt to aggregate and form an insoluble barrier, characterized in that: The device for predicting the time taken for the salt to aggregate and form an insoluble barrier comprises: An acquisition module, used to monitor the process of the sample salt aggregation biological package forming an insoluble barrier to obtain a sample data set, wherein the sample data set is composed of sample soil salt content data corresponding to the sample salt aggregation biological package at different time points; A processing module, configured to construct an exchange group based on the sample data set; Retrieving all non-applicable data sequences from the exchange group; Constructing a neural network model corresponding to each of the non-applicable data sequences, comparing the root mean square deviations of the prediction results of each of the neural network models, and selecting a target neural network model with the smallest root mean square deviation; The prediction module is used to input the target soil salt content data of the target salt-aggregating biological package into the target neural network model to obtain the target time used by the target salt-aggregating biological package to form an insoluble barrier.
8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for predicting the time taken for salt aggregation to form an insoluble barrier as described in any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting the time taken for salt to aggregate and form an insoluble barrier as described in any one of claims 1 to 6 are implemented.
10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method for predicting the time taken for salt to aggregate and form an insoluble barrier as described in any one of claims 1 to 6 are implemented.