Engineering investigation risk assessment method based on AI

Abstract

The invention discloses an engineering investigation risk assessment method based on AI, and relates to the technical field of engineering investigation risk assessment, and the method comprises the following steps: data collection: collecting data related to engineering investigation, including geological data and environmental data; the geological data collection method comprises the following steps: firstly, determining a prospecting purpose and required information, determining the boundary of a prospecting area so as to determine a prospecting target and range, and collecting existing geological reports, maps and research paper data in the prospecting area; according to the method, the engineering investigation risk is evaluated by adopting an artificial intelligence technology, and a large amount of geological and environmental data are deeply analyzed by utilizing a machine learning algorithm, so that the defect that a traditional method excessively depends on subjective experience judgment is overcome, and the accuracy and objectivity of risk evaluation are improved; through the automatic feature selection and data preprocessing process, it is ensured that the evaluation result is more accurate, and interference of human factors is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of engineering investigation risk assessment, and in particular to an engineering investigation risk assessment method based on AI. Background Art

[0002] As a preliminary step in engineering construction, engineering investigation plays a vital role in ensuring engineering safety, economy and environmental protection. Traditional engineering investigation risk assessment methods mainly rely on expert experience, qualitative analysis and limited field measurements; With the rapid development of artificial intelligence technology, especially the powerful capabilities of machine learning algorithms in data processing and pattern recognition, new technical means have been provided for engineering survey risk assessment. The AI-based engineering survey risk assessment method can overcome the above shortcomings and greatly improve the accuracy, efficiency and practicality of risk assessment through automated data collection, efficient data processing, accurate pattern recognition and prediction capabilities, as well as intuitive result visualization and real-time notification mechanisms.

[0003] However, there is no mature method in the existing technology that systematically applies AI technology to engineering survey risk assessment, especially in the comprehensive application of key links such as data preprocessing, feature selection, model construction, evaluation and result visualization. Therefore, developing an engineering survey risk assessment method based on AI can not only improve the quality of risk assessment.

[0004] In order to solve the above defects, a technical solution is now provided. Summary of the invention

[0005] The purpose of this invention is to solve the problems of strong subjectivity, low efficiency, insufficient data utilization and incomplete risk assessment in traditional engineering survey risk assessment methods, and to propose an engineering survey risk assessment method based on AI.

[0006] The purpose of the present invention can be achieved through the following technical solutions: The AI-based engineering survey risk assessment method includes the following steps: S1: Data collection: Collect data related to engineering survey, including geological data and environmental data; S2: Data preprocessing: clean and standardize the collected data and extract key features; First, the collected data is cleaned, including missing value processing, outlier detection, duplicate data processing and error correction; Then integrate the data. When the data comes from different sources, merge the data from different sources into a unified data set, and convert data of different formats or types into a unified format; Then unify the measurement units of all data and scale the data so that it falls within a specific range; Then, features with information content are selected from the original data for analysis. For feature selection, a comprehensive analysis is performed on the feature judgment indexes to obtain the feature selection value. The obtained selection value is compared with the preset selection threshold. If the selection value is greater than the preset selection threshold, the feature is used as a feature with information content for subsequent analysis. Otherwise, the feature is screened out. New features are constructed for the selected features or existing features are converted to better represent the data. The data is then smoothed to reduce noise, and the PCA method is used to reduce the dimension and extract the main information. The categorical data is then converted into a format that can be processed by the machine learning algorithm, and the categorical features are binary encoded. Then convert the continuous features into discrete features, convert the scaled data into a standard normal distribution with a mean of 0 and a standard deviation of 1, and then divide the data into data sets for model training and testing; S3: Model assessment: Risk assessment of engineering survey is carried out through the model based on the pre-processed geological data and environmental data; S4: Result visualization: Visualize the evaluation results of the model.

[0007] Furthermore, the specific process of comprehensively analyzing the judgment index of the feature in S2 to obtain the judgment value of the feature is as follows: The characteristics of the indicators include: Correlation coefficient: Calculates the linear correlation between the feature and the target variable. A high correlation coefficient indicates that the feature has a strong linear relationship with the target variable. Mutual information: measures the mutual dependence between features and target variables. The higher the mutual information, the more information the feature contains about the target variable. Feature importance: The contribution of the model output feature to the prediction, that is, the importance of the feature, which is quantified by percentage and recorded as the special weight ratio; Chi-square test: For classification problems, the chi-square test is used to evaluate the strength of the relationship between the feature value and the target category, and the chi-square test result is given a score of 1-100, recorded as the chi-square value. The higher the chi-square value, the stronger the relationship between the feature and the target category. Then, the obtained correlation coefficient, mutual information, special weight ratio and chi-square value are calibrated as xg, hx, tz and kf respectively, and after normalization, they are substituted into the following formula: To obtain the selected value QTZ, where They are the preset weight coefficients for correlation coefficient, mutual information, special weight ratio and chi-square value respectively.

[0008] Furthermore, the specific operation steps of S1 are as follows: The steps for collecting geological data are as follows: First, clarify the purpose of the survey and the information required, determine the boundaries of the survey area, thereby determining the survey target and scope, and collect existing geological reports, maps and research papers in the survey area; Then conduct field investigations to record geological phenomena such as geological outcrops, faults, joints and fissures, prepare geological maps, and use geophysical methods to detect underground results; Select key locations in the survey area for drilling and obtain rock and soil samples for laboratory testing; Then monitor groundwater levels, water quality and permeability; The steps for collecting environmental data are as follows: First, determine the environmental factors and impact range that need to be evaluated, collect environmental background information, and record the environmental conditions on site; Monitor pollutants in air, water and soil environments, use satellite or aerial remote sensing data to obtain large-scale environmental information, and assess ecosystems and biodiversity in the region; Then collect socioeconomic information on population, land use and economic development.

[0009] Furthermore, the key drilling position selection process in the geological data collection step is as follows: Based on the existing geological maps, literature and ground survey results, conduct preliminary geological analysis of the survey area, prepare or obtain geological maps of the survey area, and select locations that can represent different geological units, stratigraphic interfaces and tectonic zones; According to the project design and structural layout, select the area with the greatest impact on the project, avoiding known high-risk areas of landslides, debris flows and karst development; Based on the above analysis, determine the specific drilling locations, ensure that the locations cover all key geological features, and design the drilling plan, including the depth, diameter, and angle of the borehole; Carry out drilling construction according to the design, obtain rock and soil samples, conduct laboratory tests on the obtained rock and soil samples, analyze their physical and mechanical properties, integrate the drilling data with other geological data, and conduct comprehensive interpretation.

[0010] Furthermore, the geological data includes: topographic data, geological structure data, physical and mechanical properties of soil and rock, groundwater data, historical geological activity records and geophysical exploration data.

[0011] Furthermore, the environmental data includes: meteorological data, hydrological data, ecological data, pollution data, natural disaster data and remote sensing data.

[0012] Furthermore, the specific operation steps of S3 are as follows: Divide the preprocessed data into training sets, validation sets, and test sets, then select the most appropriate machine learning algorithm, including decision trees, random forests, support vector machines, and neural networks, use the training set to build a model, and define the model parameters and structure; Then use the training set data to train the model, adjust the model parameters to minimize the evaluation error, use the validation set data to tune the model, and find the optimal model parameters through cross-validation method; Test different models and use test set data to evaluate the performance of different models. Specifically, obtain the modularity value after the model test by comprehensive analysis of the evaluation indicators of the collected models. Then sort the modularity values ​​obtained after the different model tests by size, and select the model corresponding to the largest modularity value as the most suitable machine algorithm model; The model is trained and evaluated for multiple iterations to verify its stability and reliability, and the trained model is deployed in actual applications. The preprocessed geological data, environmental data and extracted related features are input into the model, and the engineering survey risks are output through the model and quantitatively analyzed to predict the potential impact of these risks on engineering projects. The model output includes risk probability, risk level or risk map, providing a scientific basis for engineering decision-making.

[0013] Furthermore, the process of obtaining the model value obtained after the model test by comprehensive analysis of the evaluation indicators of the acquisition model in S3 is as follows: The evaluation indicators include accuracy, recall, and F1 score-level mean square error. A comprehensive indicator function is designed to normalize different evaluation indicators and then enter them into the following formula: To get the modulus value MXZ, where ZQ, ZH, FF and MSE are the precision, recall and F1 score level mean square error quantization values ​​respectively. They are the weights assigned according to the importance of precision, recall, and F1 score-level mean square error.

[0014] Furthermore, the specific visualization method in S4 is as follows: Heat map: For spatial data, the risk level of different parts of the survey area is displayed through heat maps; Geographic Information Systems: Use GIS tools to spatially visualize geological and environmental data and model results; Risk Map: Create a risk map that uses different colors or patterns to represent different levels of risk; 3D visualization: For complex geological structures, 3D visualization technology is used to display the underground conditions predicted by the model; Animation: When data changes over time, use animation to show the risk trend; Dashboard: Develop interactive dashboards that integrate different views and tools to allow users to customize the information they want to view; Data dashboards: Use data dashboards to present summaries of key performance indicators and model outputs; Parallel coordinates plots, for multivariate data, identify the relationships between different risk factors through parallel coordinates plots; While the results are being visualized, relevant personnel will be notified of any emergency situations that arise as follows: Email Notification: Automatically send emails to relevant personnel with summaries of model results or links to detailed visualizations; Mobile App Push: Send real-time push notifications via mobile apps for risk situations that require a quick response; SMS and Instant Messaging: Use SMS or instant messaging services to quickly communicate critical information; Automated report generation: Generate automated reports, including visualizations and key metrics, and distribute them; Meetings and workshops: Organize meetings or workshops to discuss model results and follow-up action plans; Alarm system: For high-risk situations, set up an alarm system to ensure that relevant personnel can be notified immediately; Integration into workflow: Integrate risk assessment results into existing workflow management systems to ensure relevant information is visible to relevant personnel when performing tasks; Social Media: For situations that require public engagement or awareness raising, use social media platforms to share results; Customized notifications: Customize notification content and methods based on user roles and preferences; API Integration: Provides an API interface that allows other systems or applications to trigger specific actions based on model results.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention uses artificial intelligence technology to assess engineering survey risks and uses machine learning algorithms to conduct in-depth analysis of a large amount of geological and environmental data, thereby overcoming the shortcomings of traditional methods that rely too much on subjective experience and judgment, improving the accuracy and objectivity of risk assessment, and ensuring more accurate assessment results and reducing interference from human factors through automated feature selection and data preprocessing processes. The present invention realizes the automated collection and preprocessing of geological and environmental data, including data cleaning, standardization, feature extraction and dimensionality reduction, which significantly improves the efficiency of data processing. It uses PCA and other methods to reduce the dimension and extract the main information. At the same time, it adopts an automated feature selection process, which greatly reduces the time for data preparation, speeds up the risk assessment, and improves the overall work efficiency. The present invention provides a variety of result visualization methods, including charts, heat maps, GIS spatial visualization, risk maps, three-dimensional visualization and animation, making the risk assessment results more intuitive and easy to understand. In addition, through the integrated alarm system and automated report generation, relevant personnel can be notified of potential risks in a timely manner, providing strong support for engineering decision-making and ensuring the timeliness and effectiveness of risk management measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings; Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0017] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] It should be understood that the terms "include" and "comprising" used in the specification and claims of the present disclosure indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0019] It should also be understood that the terms used in this disclosure are only for the purpose of describing specific embodiments and are not intended to limit the disclosure. As used in this disclosure and claims, the singular forms of "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should also be further understood that the term "and / or" used in this disclosure and claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0020] like Figure 1 As shown in the figure, the AI-based engineering survey risk assessment method includes the following steps: Data collection: Collect data related to engineering survey, including geological data and environmental data; The steps for collecting geological data are as follows: First, clarify the purpose of the survey and the information required, determine the boundaries of the survey area, and thus determine the survey target and scope, collect existing geological reports, maps and research papers in the survey area; then conduct field surveys, record geological phenomena such as geological outcrops, faults, joints and fissures, and make geological maps, including information such as strata, structure, and lithology, and use geophysical methods to detect underground results, including seismic, electromagnetic, gravity, and magnetic methods; select key locations in the survey area for drilling, and obtain rock and soil samples for laboratory testing. The process of selecting key locations for drilling is as follows: According to the existing geological maps, literature and ground survey results, conduct preliminary geological analysis of the survey area, prepare or obtain geological maps of the survey area, including stratigraphic distribution, structural characteristics, rock types, etc., select locations that can represent different geological units, stratigraphic interfaces, structural belts, etc., select areas with the greatest impact on the project according to the project design and structural layout, and avoid known high-risk areas such as landslides, debris flows, karst development, etc. Based on the above analysis, determine specific drilling points to ensure that the points can cover all key geological features, design drilling plans, including the depth, diameter, angle, etc. of the borehole, carry out drilling construction according to the design, obtain rock and soil samples, conduct laboratory tests on the obtained rock and soil samples, analyze their physical and mechanical properties, integrate the drilling data with other geological data, and conduct comprehensive interpretation; Then monitor groundwater levels, water quality and permeability; The geological data include: Topographic data: including contour lines, slopes, slope directions, etc.; geological structure data: such as stratum distribution, rock types, faults, joints, fissures and other geological structural information; physical and mechanical properties of soil and rock: including soil type, density, particle size distribution, shear strength, compressibility, etc.; groundwater data: groundwater level, water quality, permeability, flow rate and flow direction; historical geological activity records: such as historical records of disasters such as earthquakes, landslides, and mud-rock flows; geophysical exploration data: underground structure and property information obtained through geophysical methods; The steps for collecting environmental data are as follows: First, determine the environmental factors and impact ranges that need to be assessed, collect environmental background information, including natural environmental data such as climate, hydrology, vegetation, and wildlife, and record the environmental conditions on site, including pollution sources and sensitive areas; monitor pollutants in environmental media such as air, water, and soil, use satellite or aerial remote sensing data to obtain a wide range of environmental information, and assess the ecosystem and biodiversity in the region; then collect socio-economic information such as population, land use, and economic development; Environmental data includes: Meteorological data: including temperature, precipitation, wind speed, humidity, etc.; hydrological data: hydrological characteristics of rivers, lakes, and groundwater, such as water level, flow rate, and flow; ecological data: vegetation type, coverage, biodiversity, etc.; pollution data: pollutant content in soil, water bodies, and air; natural disaster data: frequency and intensity of disasters such as floods, droughts, storms, and earthquakes; remote sensing data: information on land use, vegetation changes, etc. obtained through satellite or aerial photography.

[0021] Data preprocessing: clean and standardize the collected data and extract key features; First, the collected data is cleaned, including: Missing value processing: Identify missing values ​​in the data and handle missing values ​​by filling, deleting or retaining them; Outlier detection: Find outliers or outliers in the data; Duplicate data processing: Check and handle duplicate records in the data set; Error correction: Correct errors in the data entry or recording process; Then integrate the data. When the data comes from different sources, merge the data from different sources into a unified data set, and convert the data of different formats or types into a unified format; then unify the measurement units of all data, and scale the data proportionally so that the data falls into a specific interval, such as [0, 1]; then select the features with information content from the original data for analysis. For the selection of features, conduct a comprehensive analysis of the judgment indicators of the features to determine whether they meet the selection requirements. The judgment indicators of the features include: Correlation coefficient: calculates the linear correlation between the feature and the target variable. A high correlation coefficient indicates that the feature has a strong linear relationship with the target variable. Mutual information: measures the mutual dependence between the feature and the target variable. The higher the mutual information, the more information the feature contains about the target variable. Feature importance: in some algorithms (such as random forests), the model can output the contribution of the feature to the prediction, that is, the importance of the feature, and quantify it by percentage, recorded as the special weight ratio. Chi-square test: for classification problems, the chi-square test is used to evaluate the strength of the relationship between the feature value and the target category, and the chi-square test result is assigned a score of 1-100, recorded as the chi-square value. The higher the chi-square value, the stronger the relationship between the feature and the target category. Then, the obtained correlation coefficient, mutual information, special weight ratio and chi-square value are calibrated as xg, hx, tz and kf respectively, and after normalization, they are substituted into the following formula: To obtain the selected value QTZ, where are the preset weight coefficients for correlation coefficient, mutual information, special weight ratio and chi-square value, and their values ​​are 0.93, 1.12, 1.31 and 0.96 respectively; Then the obtained selection value QTZ is compared with the preset selection threshold. If the selection value QTZ is greater than the preset selection threshold, the feature is used as an informative feature for subsequent analysis. Otherwise, the feature is screened out. New features are constructed for the selected features or existing features are converted to better represent the data. The data is then smoothed to reduce noise, and the PCA method is used to reduce the dimension and extract the main information. The categorical data is then converted into a format that can be processed by the machine learning algorithm, and the categorical features are binary encoded. The continuous features are then converted into discrete features, and the scaled data is converted into a standard normal distribution with a mean of 0 and a standard deviation of 1. The data is then divided into data sets for model training and testing.

[0022] Model evaluation: used to conduct risk assessment of engineering surveys through models based on pre-processed geological data and environmental data; The preprocessed data is divided into a training set, a validation set, and a test set. The most appropriate machine learning algorithm is selected, including decision trees, random forests, support vector machines, and neural networks. The training set is used to build a model and define the model parameters and structure. The model is then trained using the training set data, and the model parameters are adjusted to minimize the evaluation error. The model is tuned using the validation set data, and the optimal model parameters are found through the cross-validation method. Different models are tested, and the performance of different models is evaluated using the test set data. Specifically, the evaluation indicators of the model are collected, including accuracy, recall rate, and F1 score-level mean square error. A comprehensive indicator function is designed, and different evaluation indicators are normalized and inserted into the following formula: To get the modulus value MXZ, where ZQ, ZH, FF and MSE are the precision, recall and F1 score level mean square error quantization values ​​respectively. The weights are assigned according to the importance of accuracy, recall, and F1 score-level mean square error. The modularity values ​​obtained after testing different models are sorted by size, and the model corresponding to the largest modularity value is selected as the most suitable machine algorithm model. The model is trained and evaluated for multiple iterations to verify its stability and reliability, and the trained model is deployed in actual applications. The preprocessed geological data, environmental data and extracted related features are input into the model, and the engineering survey risks are output through the model and quantitatively analyzed to predict the potential impact of these risks on engineering projects. The model output includes risk probability, risk level or risk map, providing a scientific basis for engineering decision-making.

[0023] Result visualization: Visualize the evaluation results of the model; The specific visualization method is as follows: Charts and graphs: Use bar charts, line charts, and scatter plots to display model outputs; Heat maps: For spatial data, use heat maps to display risk levels in different parts of the survey area; Geographic Information Systems: Use GIS tools to spatially visualize geological and environmental data and model results; Risk maps: Create risk maps that use different colors or patterns to represent different levels of risk; 3D visualization: For complex geological structures, use 3D visualization technology to display the underground conditions predicted by the model; Animation: When data changes over time, use animation to display risk trends; Dashboards: Develop interactive dashboards that integrate different views and tools to allow users to customize the information they want to view; Data dashboards: Use data dashboards to display key performance indicators and summaries of model outputs; Parallel coordinate plots: For multivariate data, use parallel coordinate plots to identify the relationship between different risk factors; While the results are being visualized, relevant personnel will be notified of any emergency situations that arise as follows: Email notifications: automatically send emails to relevant personnel, containing a summary of model results or links to detailed visualization results; Mobile application push: send real-time push notifications through mobile applications, especially for risk situations that require a quick response; SMS and instant messaging: use SMS or instant messaging services (such as Slack, WeChat, etc.) to quickly convey key information; Automated report generation: generate automated reports, including visualizations and key indicators, and distribute them through the system; Meetings and workshops: organize meetings or workshops to discuss model results and subsequent action plans; Alert system: for high-risk situations, set up an alert system to ensure that relevant personnel can be notified immediately; Integration into workflow: integrate risk assessment results into existing workflow management systems to ensure that relevant personnel can see relevant information when performing tasks; Social media: for situations that require public participation or awareness raising, use social media platforms to share results; Customized notifications: customize notification content and methods based on the user's role and preferences; API integration: provide an API interface to allow other systems or applications to trigger specific actions based on model results.

[0024] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. The AI-based engineering survey risk assessment method is characterized by: The following steps are involved: S1: Data collection: Collect data related to engineering survey, including geological data and environmental data; S2: Data preprocessing: clean and standardize the collected data and extract key features; First, the collected data is cleaned, including missing value processing, outlier detection, duplicate data processing and error correction; Then integrate the data. When the data comes from different sources, merge the data from different sources into a unified data set, and convert data of different formats or types into a unified format; Then unify the measurement units of all data and scale the data so that it falls within a specific range; Then, features with information content are selected from the original data for analysis. For feature selection, a comprehensive analysis is performed on the feature judgment indexes to obtain the feature selection value. The obtained selection value is compared with the preset selection threshold. If the selection value is greater than the preset selection threshold, the feature is used as a feature with information content for subsequent analysis. Otherwise, the feature is screened out. New features are constructed for the selected features or existing features are converted to better represent the data. The data is then smoothed to reduce noise, and the PCA method is used to reduce dimensionality and extract information. The categorical data is then converted into a format that can be processed by the machine learning algorithm, and the categorical features are binary encoded; Then convert the continuous features into discrete features, convert the scaled data into a standard normal distribution with a mean of 0 and a standard deviation of 1, and then divide the data into data sets for model training and testing; S3: Model assessment: Risk assessment of engineering survey is carried out through the model based on the pre-processed geological data and environmental data; S4: Result visualization: Visualize the evaluation results of the model.

2. The AI-based engineering survey risk assessment method according to claim 1, characterized in that: The specific process of comprehensively analyzing the feature judgment index in S2 to obtain the feature selection value is as follows: The characteristics of the indicators include: Correlation coefficient: Calculates the linear correlation between the feature and the target variable. A high correlation coefficient indicates that the feature has a strong linear relationship with the target variable. Mutual information: measures the mutual dependence between features and target variables. The higher the mutual information, the more information the feature contains about the target variable. Feature importance: The contribution of the model output feature to the prediction, that is, the importance of the feature, which is quantified by percentage and recorded as the special weight ratio; Chi-square test: For classification problems, the chi-square test is used to evaluate the strength of the relationship between the feature value and the target category, and the chi-square test result is given a score of 1-100, recorded as the chi-square value. The higher the chi-square value, the stronger the relationship between the feature and the target category. Then, the obtained correlation coefficient, mutual information, special weight ratio and chi-square value are calibrated as xg, hx, tz and kf respectively, and after normalization, they are substituted into the following formula: To obtain the selected value QTZ, where They are the preset weight coefficients for correlation coefficient, mutual information, special weight ratio and chi-square value respectively.

3. The AI-based engineering survey risk assessment method according to claim 1, characterized in that: The specific operation steps of S1 are as follows: The steps for collecting geological data are as follows: First, clarify the purpose of the survey and the information required, determine the boundaries of the survey area, thereby determining the survey target and scope, and collect existing geological reports, maps and research papers in the survey area; Then conduct field investigations to record geological phenomena such as geological outcrops, faults, joints and fissures, prepare geological maps, and use geophysical methods to detect underground results; Select key locations in the survey area for drilling and obtain rock and soil samples for laboratory testing; Then monitor groundwater levels, water quality and permeability; The steps for collecting environmental data are as follows: First, determine the environmental factors and impact range that need to be evaluated, collect environmental background information, and record the environmental conditions on site; Monitor pollutants in air, water and soil environments, use satellite or aerial remote sensing data to obtain large-scale environmental information, and assess ecosystems and biodiversity in the region; Then collect socioeconomic information on population, land use and economic development.

4. The AI-based engineering investigation risk assessment method according to claim 3 is characterized in that: The process of selecting the key locations to be drilled in the geological data collection step is as follows: Conduct preliminary geological analysis of the survey area based on existing geological maps, literature and ground survey results, prepare or obtain geological maps of the survey area, and select locations that can represent different geological units, stratigraphic interfaces and tectonic zones; According to the project design and structural layout, select the area with the greatest impact on the project, avoiding known high-risk areas of landslides, debris flows and karst development; Based on the above analysis, determine the specific drilling locations, ensure that the locations cover all key geological features, and design the drilling plan, including the depth, diameter and angle of the borehole; Carry out drilling construction according to the design, obtain rock and soil samples, conduct laboratory tests on the obtained rock and soil samples, analyze their physical and mechanical properties, integrate the drilling data with geological data, and conduct comprehensive interpretation.

5. The AI-based engineering investigation risk assessment method according to claim 3 is characterized in that: The geological data include: topographic data, geological structure data, physical and mechanical properties of soil and rock, groundwater data, historical geological activity records and geophysical exploration data.

6. The AI-based engineering investigation risk assessment method according to claim 3 is characterized in that: The environmental data include: meteorological data, hydrological data, ecological data, pollution data, natural disaster data and remote sensing data.

7. The AI-based engineering investigation risk assessment method according to claim 1, characterized in that: The specific operation steps of S3 are as follows: Divide the preprocessed data into training sets, validation sets, and test sets, then select the most appropriate machine learning algorithm, including decision trees, random forests, support vector machines, and neural networks, use the training set to build a model, and define the model parameters and structure; Then use the training set data to train the model, adjust the model parameters to minimize the evaluation error, use the validation set data to tune the model, and find the optimal model parameters through cross-validation method; Test different models and use test set data to evaluate the performance of different models. Specifically, obtain the modularity value after the model test by comprehensive analysis of the evaluation indicators of the collected models. Then sort the modularity values ​​obtained after the different model tests by size, and select the model corresponding to the largest modularity value as the most suitable machine algorithm model; The model is trained and evaluated for multiple iterations to verify its stability and reliability, and the trained model is deployed in actual applications. The preprocessed geological data, environmental data and extracted related features are input into the model, and the engineering survey risks are output through the model and quantitatively analyzed to predict the potential impact of these risks on engineering projects. The model output includes risk probability, risk level or risk map, providing a scientific basis for engineering decision-making.

8. The AI-based engineering investigation risk assessment method according to claim 7 is characterized in that: The process of obtaining the model value obtained after the model test by comprehensive analysis of the evaluation indicators of the acquisition model in S3 is as follows: The evaluation indicators include accuracy, recall, and F1 score-level mean square error. A comprehensive indicator function is designed to normalize different evaluation indicators and then enter them into the following formula: To get the modulus value MXZ, where ZQ, ZH, FF and MSE are the precision, recall and F1 score level mean square error quantization values ​​respectively. They are the weights assigned according to the importance of precision, recall, and F1 score-level mean square error.

9. The AI-based engineering investigation risk assessment method according to claim 1, characterized in that: The specific visualization method in S4 is as follows: Charts and Graphs: Use bar charts, line charts, and scatter plots to present model output; Heat map: For spatial data, the risk level of different parts of the survey area is displayed through heat maps; Geographic Information Systems: Use GIS tools to spatially visualize geological and environmental data and model results; Risk Map: Create a risk map that uses different colors or patterns to represent different levels of risk; 3D visualization: For complex geological structures, 3D visualization technology is used to display the underground conditions predicted by the model; Animation: When data changes over time, use animation to show the risk trend; Dashboard: Develop interactive dashboards that integrate different views and tools to allow users to customize the information they want to view; Data dashboards: Use data dashboards to present summaries of key performance indicators and model outputs; Parallel coordinates plots, for multivariate data, identify the relationships between different risk factors through parallel coordinates plots; While the results are being visualized, relevant personnel will be notified of any emergency situations that arise as follows: Email Notification: Automatically send emails to relevant personnel with summaries of model results or links to detailed visualizations; Mobile App Push: Send real-time push notifications via mobile apps for risk situations that require a quick response; SMS and Instant Messaging: Use SMS or instant messaging services to quickly communicate critical information; Automated report generation: Generate automated reports, including visualizations and key metrics, and distribute them; Meetings and workshops: Organize meetings or workshops to discuss model results and follow-up action plans; Alarm system: For high-risk situations, set up an alarm system to ensure that relevant personnel can be notified immediately; Integration into workflow: Integrate risk assessment results into existing workflow management systems to ensure relevant information is visible to relevant personnel when performing tasks; Social Media: For situations that require public engagement or awareness raising, use social media platforms to share results; Customized notifications: Customize notification content and methods based on user roles and preferences; API Integration: Provides an API interface that allows other systems or applications to trigger specific actions based on model results.

Images