Silt biomineralization dust suppression method based on experiment and machine learning

Abstract

The invention discloses a silt biomineralization dust suppression method based on experiment and machine learning, which comprises the following steps: S1, acquiring a soil sample, and testing and recording soil property parameters of the sample; s2, proportioning a plurality of parts of biomineralization dust suppressants; s3, mixing the soil sample with a biomineralization dust suppressant, carrying out engineering characteristic experiment on the tested soil sample after the mixing reaction is completed, and obtaining and recording engineering property parameters of the soil sample; s4, establishing a training subsystem, and constructing a prediction model; and S5, according to newly-collected silt sample data, the prediction model is called for real-time prediction of the reinforcement effect, and a prediction result is interpreted, analyzed and visually displayed. The system has the beneficial effects that experimental research and machine learning technologies are combined, and the system can predict the biomineralization dust suppression effect by comprehensively considering the influence of factors such as a biomineralization dust suppressant formula, a field soil condition and a field environment, so that a scientific basis is provided for biomineralization dust suppression of the silt foundation in the yellow flood area.

Description

Technical Field

[0001] The invention relates to the field of dust prevention and dust suppression, and in particular to a silt biomineralization dust suppression method based on experiments and machine learning. Background Art

[0002] Traditional physical dust suppression technologies, such as laying dust-proof nets, have long construction periods, aging and brittleness, secondary pollution, and easy blowing away in strong winds, posing threats to nearby roads, power lines and other facilities. Watering to suppress dust has problems such as wasting water resources and short duration. Traditional chemical dust suppressants have problems such as potential environmental threats, many raw materials, difficulty in preparation, and high costs. In recent years, biomineralization dust suppression technology has attracted attention as an emerging dust suppression treatment method. This technology uses enzymes to induce calcium carbonate precipitation, thereby cementing soil particles into a whole with a certain strength. However, the dust suppression effect of biomineralization is affected by many factors, such as urease activity, cementing liquid concentration and ratio. Among them, urease activity is affected by factors such as soybean type, soybean powder particle size, and the solid-liquid ratio of soybean powder to water. This makes it difficult to accurately predict and control the effect of biomineralization dust suppression in actual engineering.

[0003] In contrast, a Chinese patent with publication number CN113232863A discloses an intelligent dust suppression system and a dust suppression method, which relate to the technical field of dust suppression treatment. The method includes: obtaining environmental parameters of a preset area collected by an environmental monitoring device; determining a pollution index corresponding to the preset area according to the environmental parameters; generating a dust suppression strategy based on the pollution index, and sending the dust suppression strategy to the dust suppression system, so that the dust suppression system performs dust suppression treatment on the polluted dust source corresponding to the coordinate information according to the dust suppression strategy.

[0004] In this patent document, the main focus is on pollution monitoring of the on-site dust suppression system and spraying control of dust suppressants when pollution occurs. However, for different soil areas, there are large differences in the particle gradation and soil density of the Yellow River flooding area in different regions. The on-site environment such as temperature and pH value will also affect the strength, stiffness and permeability of biomineralized dust suppression. It is often necessary to select dust suppressants with appropriate effects, but in actual operation, the dust suppression effect is often not easy to predict. Usually, it can only be calculated and predicted through conventional physical models, which is very difficult in actual operation.

[0005] Therefore, providing a silt biomineralization dust suppression method based on experiments and machine learning has become a worthy research issue. Summary of the invention

[0006] In order to solve the deficiencies in the above-mentioned prior art, the purpose of the present invention is to combine experimental research and machine learning technology. By comprehensively considering the influence of factors such as the formula of biomineralization dust suppressant, on-site soil conditions and site environment, the system can predict the biomineralization dust suppression effect, and provide a scientific basis for biomineralization dust suppression of silt foundation in the Yellow River flooding area.

[0007] The object of the present invention is achieved in that: A silt biomineralization dust suppression method based on experiments and machine learning, comprising the following steps: S1: Obtain soil samples and test and record soil parameters of the samples; S2: Mix multiple portions of biomineralized dust suppressant; S3: Mix the soil sample with the biomineralized dust suppressant. After the mixing reaction is complete, conduct an engineering property experiment on the test soil sample to obtain and record its engineering property parameters; S4: Establish a training subsystem based on the soil parameters of the soil sample, the ratio of the biomineralized dust suppressant and the engineering property parameters of the mineralized soil sample, select a suitable machine learning model to cooperate with the training subsystem to build a prediction model, and assist in the training and optimization of the prediction model; S5: Based on the newly collected silt sample data, the prediction model is called to make a real-time prediction of the reinforcement effect, and the prediction results are interpreted, analyzed and visualized.

[0008] In a preferred embodiment, the training subsystem includes: Data cleaning and standardization unit, which removes outliers and noise from experimental data and performs standardization processing; A feature extraction unit, used to extract key features related to dust suppression effect; Model selection and construction unit, used to select appropriate machine learning algorithms to build prediction models based on the characteristics of the problem and the nature of the data; The model training and optimization unit is used to train and optimize the model using historical experimental data.

[0009] In a preferred embodiment, the prediction model comprises: A data receiving and processing unit, used to receive newly collected silt sample data and perform necessary preprocessing; The prediction unit is used to call the trained machine learning model to make real-time predictions on the dust suppression effect of new samples; The result analysis and visualization unit is used to interpret, analyze and visualize the prediction results.

[0010] In a preferred embodiment, the machine learning algorithm included in the machine learning model is random forest, support vector machine, gradient boosting, neural network.

[0011] In a preferred embodiment, the soil parameters of the sample include water content, soil density, particle grading, liquid limit, plastic limit, pH value, and soil temperature.

[0012] The engineering property parameters of the test soil sample are compressive strength, tensile strength, stiffness and permeability.

[0013] In a preferred embodiment, the collected soil samples are sealed and stored, and are marked with detailed geographical location information.

[0014] In a preferred embodiment, the biomineralization dust suppressant comprises soybean urease, urea and calcium chloride bio-grouting liquid materials, and the soybean urease, urea and calcium chloride bio-grouting liquid materials need to be recorded in detail when configuring the biomineralization dust suppressant, and the soybean variety, soybean powder particle size, solid-liquid ratio, urea and calcium chloride solution concentrations and the ratio between each solution must be recorded in detail.

[0015] In a preferred embodiment, the number of dust suppression treatments and the amount of biomineralized dust suppressant used in a single treatment need to be recorded in detail, and the data is regularly uploaded to the machine learning model for learning and recording.

[0016] Positive beneficial effects: Combining experimental research and machine learning technology, this experimental method aims to more comprehensively reveal the mechanism and influencing factors of biomineralization dust suppression by comprehensively considering the influence of factors such as biomineralization dust suppression formulation, on-site soil conditions and site environment, and provide a solid data foundation for subsequent intelligent prediction and analysis; The system can predict the engineering properties of soil after biomineralization dust suppression, such as strength, stiffness and permeability, and provide a scientific basis for biomineralization dust suppression of silt foundation in the Yellow River flooding area. It can automatically optimize the formula of biomineralization dust suppressants and dust suppression process parameters, reduce engineering risks, and improve engineering quality and sustainability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the construction method of the present invention; Figure 2 The experimental design and data collection flow chart of the technology of the present invention; Figure 3 This is a flow chart of the machine learning model and training subsystem of the technology of the present invention. DETAILED DESCRIPTION

[0018] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0019] See also Figure 1-Figure 3 As shown, the present application provides a method for dust suppression by biomineralization of silt based on experiments and machine learning, comprising the following steps: S1: Obtain soil samples and test and record soil parameters of the samples; Specifically, select representative locations from the silt distribution area to collect silt samples. After the samples are collected, they should be sealed and stored, marked with detailed geographical location information, and the sample soil should be tested.

[0020] When collecting samples, molds of uniform size should be used as much as possible. For example, a 10cm*10cm square mold, a 15cm*15cm square mold, a 10cm diameter round mold, or a 15cm diameter round mold can be selected. There is no limit on the size of the mold. The mold can be used to assist in the collection of silt samples, and multiple silt samples can be collected at the same location.

[0021] The soil parameters of the collected silt samples are measured and recorded in the laboratory. Specifically, the soil parameters of the samples include but are not limited to water content, soil density, liquid limit, plastic limit, and particle grading.

[0022] S2: Mix multiple portions of biomineralized dust suppressant according to demand; When preparing the biomineralization dust suppressant, the main materials used are: water, soybeans, urea and calcium chloride, etc., among which soybean urease can be extracted from soybeans. Urease can promote the hydrolysis of urea to produce carbonate ions, which combine with calcium ions in calcium chloride to form calcium carbonate crystals. Urease is combined with a certain concentration of urea and calcium chloride solution to prepare biomineralization liquid. By spraying the biomineralization liquid on the soil surface, the surface soil particles are cemented into one, thereby achieving the effect of dust suppression; When mixing biomineralized dust suppressants, it is necessary to record in detail information such as soybean variety, soybean powder particle size, solid-liquid ratio, urea and calcium chloride solution concentrations, and the ratio between each solution.

[0023] S3: Mix the soil sample with the biomineralized dust suppressant, and after the mixing reaction is complete, conduct an engineering property experiment on the test soil sample to obtain and record its engineering property parameters; During the experiment, prepare experimental equipment for geotechnical tests, such as unconfined compression testers, consolidation testers, and permeameters, and test and record key parameters obtained from the experiment. Ensure the accuracy and reliability of the experimental equipment. The experiment covers a variety of variables and conditions. Spray a certain amount of biomineralized dust suppressant on the silt sample, and measure and record the parameters of the soil sample after sufficient reaction. The engineering property parameters of the tested soil samples include but are not limited to: compressive strength, tensile strength, stiffness, and permeability; Record all key data during the experiment, including but not limited to changes in soil properties before and after mineralization, chemical reaction data during biomineralization, etc. The data should be regularly organized and stored in a safe and reliable database to facilitate the establishment of subsequent training subsystems.

[0024] S4: Establish a training subsystem based on the soil parameters of the soil sample, the ratio of the biomineralized dust suppressant and the engineering property parameters of the mineralized soil sample, select a suitable machine learning model to cooperate with the training subsystem to build a prediction model, and assist the prediction model in training and optimization; After multiple groups of silt mineralization experiments are carried out through S1-S4 above, the experimental parameters are uniformly uploaded to the database as learning parameters of the training subsystem. The learning parameters are the soil parameters of the soil samples, the engineering property parameters of the biomineralization dust suppressant and the mineralized soil samples, the number of dust suppression treatments, and the amount of biomineralization dust suppressant used in a single time.

[0025] The training subsystem includes a data cleaning and standardization unit, a feature extraction unit, a model selection and construction unit, and a model training and optimization unit, which are used to process experimental data, select appropriate machine learning algorithms to build prediction models, and use historical experimental data to train and optimize the models; Specifically, since the raw data usually contains noise, errors, and inconsistencies, the experimental data and the physical and chemical properties data of the samples are cleaned through the data cleaning and standardization unit to remove outliers and noise and perform standardization. The cleaning process for experimental parameter data includes: (1) Remove noise: standardize text data; (2) Handling missing values ​​and duplicate data: For missing samples, you can choose to fill, delete, or replace them with default values.

[0026] (3) Identify and pre-edit data that is obviously erroneous or abnormal.

[0027] This can make the learning parameters more standardized and reduce the errors in the experimental parameters during learning.

[0028] Through cross-validation, grid search and parameter optimization of the machine learning model, when specifically selected, the machine learning algorithm included in the machine learning model can be random forest, support vector machine, gradient boosting, neural network. At the same time, the model is trained and optimized using historical experimental data, and the key features related to dust suppression effect are extracted using feature engineering methods to improve the final prediction accuracy.

[0029] Select a suitable machine learning model to cooperate with the training subsystem for training and optimization, and build a prediction model for predicting the dust suppression effect of silt biomineralization dust suppressant to ensure that the model has high prediction accuracy and strong generalization ability.

[0030] S5: Based on the newly collected silt sample data, the prediction model is called to make a real-time prediction of the reinforcement effect, and the prediction results are interpreted, analyzed and visualized.

[0031] The prediction model includes a data receiving and processing unit, a prediction unit, and a result analysis and visualization unit.

[0032] Specifically, in the data receiving and processing unit, it is mainly used to receive the newly collected silt sample data. The newly collected silt sample can be uploaded to the prediction model, and the data can be processed to ensure that the data format is consistent with the data format used when training the model. The data is analyzed through the prediction model to determine the appropriate machine learning model.

[0033] The prediction unit uses the trained machine learning model to analyze and predict the biomineralized dust suppressant ratio and dust suppression effect of the new sample; The result analysis and visualization unit includes a display screen that can display the prediction results. It can visualize the dust suppression result analysis between the biomineralized dust suppressant and the new soil sample, assist operators in construction simulation, and select the appropriate biomineralized dust suppressant and the appropriate biomineralized dust suppressant construction method according to the dust suppression effect of silt, thereby greatly improving construction efficiency.

[0034] Specifically, the prediction model is helpful for quickly obtaining the proportion of biomineralization dust suppressants that meet the mineralization needs of the soil samples based on the basic information of the soil samples after collecting and testing the soil samples. After the corresponding biomineralization dust suppressants are proportioned and used, the soil samples are tested again for verification, and the test results are uploaded to the training subsystem to assist the machine learning model in learning and reference. The prediction model can be corrected at any time to assist the prediction model in obtaining a more optimal method solution.

[0035] The present invention organically combines experiments with machine learning methods, which not only improves the prediction accuracy of the dust suppression effect of silt biomineralization in the Yellow River flooding area, but also provides strong support for practical engineering applications.

[0036] The intelligent prediction and analysis subsystem mainly includes data receiving and processing module, prediction module, result analysis and visualization module; In a preferred embodiment, the prediction model receives the newly collected silt sample data through a data receiving interface, and the interface performs necessary preprocessing on the received data to ensure that the data format is consistent with the data format used when training the model. When working, the prediction model calls the trained machine learning model to make real-time predictions of the reinforcement effect of the newly collected silt sample data, and the prediction results are output in the form of probability or specific numerical values; the prediction results are interpreted and analyzed graphically, so as to display the result interface to the user.

[0037] In addition, this module also provides interfaces with other data analysis tools to facilitate users to conduct more in-depth data analysis and mining.

[0038] In order to verify the beneficial effects of the present invention, a comparative test was conducted with the traditional method. The following is a partial comparative data table: It can be seen from the comparative data that the intelligent prediction and analysis system of the present invention has shown significant advantages in prediction accuracy, construction cost, engineering efficiency, environmental impact, etc. In particular, in terms of prediction accuracy, the prediction accuracy of the present invention is improved by more than 20% compared with the traditional method.

[0039] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for dust suppression by biomineralization of silt based on experiments and machine learning, characterized in that: The steps include: S1: Obtain soil samples and test and record soil parameters of the samples; S2: Mix multiple portions of biomineralized dust suppressant; S3: Mix the soil sample with the biomineralized dust suppressant. After the mixing reaction is completed, conduct an engineering property experiment on the test soil sample to obtain and record its engineering property parameters; S4: Establish a training subsystem based on the soil parameters of the soil sample, the ratio of the biomineralized dust suppressant and the engineering property parameters of the mineralized soil sample, select a suitable machine learning model to cooperate with the training subsystem to build a prediction model, and assist in the training and optimization of the prediction model; S5: Based on the newly collected silt sample data, the prediction model is called to make a real-time prediction of the reinforcement effect, and the prediction results are interpreted, analyzed and visualized.

2. The method for dust suppression by biomineralization of silt based on experiments and machine learning according to claim 1 is characterized in that: The training subsystem comprises: Data cleaning and standardization unit, which removes outliers and noise from experimental data and performs standardization processing; A feature extraction unit, used to extract key features related to dust suppression effect; Model selection and construction unit, used to select appropriate machine learning algorithms to build prediction models based on the characteristics of the problem and the nature of the data; The model training and optimization unit is used to train and optimize the model using historical experimental data.

3. The method for dust suppression by biomineralization of silt based on experiments and machine learning according to claim 2 is characterized in that: The prediction model includes: A data receiving and processing unit, used to receive newly collected silt sample data and perform necessary preprocessing; The prediction unit is used to call the trained machine learning model to make real-time predictions on the dust suppression effect of new samples; The result analysis and visualization unit is used to interpret, analyze and visualize the prediction results.

4. The method for dust suppression by biomineralization of silt based on experiments and machine learning according to claim 3 is characterized in that: The machine learning algorithms included in the machine learning model are random forest, support vector machine, gradient boosting, and neural network.

5. A method for dust suppression by biomineralization of silt based on experiments and machine learning according to any one of claims 1 to 4, characterized in that: The soil parameters of the sample include water content, soil density, particle gradation, liquid limit, pH value, and soil temperature; The engineering property parameters of the test soil sample are compressive strength, tensile strength, stiffness and permeability.

6. The method for dust suppression by biomineralization of silt based on experiments and machine learning according to claim 5 is characterized in that: The collected soil samples are sealed and marked with detailed geographical location information.

7. The method for dust suppression by biomineralization of silt based on experiments and machine learning according to claim 6 is characterized in that: The biomineralization dust suppressant comprises soybean urease, urea and calcium chloride biological grouting liquid materials, and the soybean urease, urea and calcium chloride biological grouting liquid materials need to be recorded in detail when configuring the biomineralization dust suppressant, and the soybean variety, soybean powder particle size, solid-liquid ratio, urea and calcium chloride solution concentrations and the ratio between each solution agent are recorded in detail.

8. The method for dust suppression by biomineralization of silt based on experiments and machine learning according to claim 7 is characterized in that: The number of dust suppression treatments and the amount of biomineralized dust suppressant used in a single treatment need to be recorded in detail, and the data should be regularly uploaded to the machine learning model for learning and recording.

Images