Site pollution treatment scheme generation method and device

Abstract

The application discloses a site pollution treatment scheme generation method and device, the method comprises the following steps: collecting pollution data of multiple dimensions of a target area, the pollution data at least comprising pollutant characteristic data, site characteristic data and environmental factor data; performing data processing on pollution data of each dimension respectively, and fusing pollution data of each dimension after data processing to obtain fused multi-modal data; inputting the fused multi-modal data into a pre-constructed scheme generation model to obtain a pollution treatment scheme output by the scheme generation model; wherein the scheme generation model is constructed by using historical site pollution data after data processing and a historical treatment scheme corresponding to the historical site pollution data. Through model construction and calling, a pollution treatment scheme that can be selected and referred to is generated, and the technical problem that the generation of a site pollution treatment scheme completely depends on artificial experience and on-site investigation is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a site pollution treatment scheme generation method and device. BACKGROUND

[0002] Site pollution problems are usually related to industrial production, chemical waste leakage or historical pollution, especially in the process of industrial site, abandoned chemical plant and urban construction expansion, the demand for pollution treatment increases significantly. Traditional site pollution treatment methods rely on expert experience and on-site investigation, which is not only time-consuming, laborious and subjective, but also has problems such as difficulty in accurate identification, insufficient monitoring data and inflexible treatment scheme when dealing with complex pollutants.

[0003] Therefore, a site pollution treatment scheme generation method and device are provided to solve the technical problem that the generation of site pollution treatment scheme completely depends on artificial experience and on-site investigation. SUMMARY

[0004] To this end, the embodiments of the present application provide a site pollution treatment scheme generation method and device, which generates pollution treatment schemes for selection and reference through model construction and retrieval, so as to solve the technical problem that the generation of site pollution treatment scheme completely depends on artificial experience and on-site investigation.

[0005] In order to achieve the above purpose, the embodiments of the present application provide the following technical solutions:

[0006] The present application provides a site pollution treatment scheme generation method, which comprises:

[0007] Collecting pollution data of multiple dimensions of a target area, the pollution data at least including pollutant characteristic data, site characteristic data and environmental factor data;

[0008] Respectively processing the pollution data of each dimension, and fusing the pollution data of each dimension after data processing to obtain fused multi-modal data;

[0009] Inputting the fused multi-modal data into a pre-constructed scheme generation model to obtain a pollution treatment scheme output by the scheme generation model;

[0010] The scheme generation model is constructed by using various types of historical site pollution data after data processing and historical treatment schemes corresponding to the historical site pollution data.

[0011] In some embodiments, constructing the scheme generation model specifically comprises:

[0012] Obtaining a large amount of various types of historical site pollution data;

[0013] data cleaning is performed on the historical site pollution data, and target information is labeled;

[0014] Using natural language processing algorithms, target features are extracted from the labeled historical site pollution data, and a site pollution remediation technology large model is constructed based on retrieval enhancement generation by combining semantic understanding and keyword matching.

[0015] In some embodiments, the pollutant characteristic data at least includes the type, concentration and spatial distribution of the pollutant; and / or,

[0016] The site characteristic data at least includes soil type, pH value, organic matter content and groundwater distribution; and / or,

[0017] The environmental factor data at least includes temperature, humidity and rainfall data of the contaminated site.

[0018] In some embodiments, the data processing of the pollutant characteristic data further includes:

[0019] According to the widely collected soil and water samples in the target area, the concentration level of the target pollutant in the soil and water samples is determined;

[0020] Using sampling data and spatial interpolation method, the concentration distribution map of the target pollutant is drawn, and the spatial distribution characteristics and migration trend of the target pollutant are determined based on the concentration distribution map.

[0021] In some embodiments, the data type of the pollution data at least includes text data, image data and sensor reading data.

[0022] In some embodiments, the pollution data of each dimension after data processing is fused, specifically including:

[0023] The text data is encoded using a word embedding algorithm to obtain an embedding vector corresponding to the historical governance record text data of the site to be treated;

[0024] Image features are extracted from the image data;

[0025] The pollutant and environmental numerical features are extracted;

[0026] The embedding vector and the image feature are fused by a data fusion algorithm to obtain a fusion feature output by the deep learning model.

[0027] In some embodiments, the document feature extraction adopts TF-IDF (Term Frequency-Inverse Document Frequency):

[0028]

[0029] wherein, TF(t, d) is the frequency of word t in document d, N is the total number of documents, DF(t) is the number of documents containing word t;Term Frequency-Inverse Document Frequency is the Chinese meaning of "term frequency-inverse document frequency", usually abbreviated as TF-IDF, which is a widely used weight calculation method in information retrieval and text mining, used to evaluate the importance of a word to a file set or a file in a corpus;Term Frequency (TF) represents the frequency of a word appearing in a document, and Inverse Document Frequency (IDF) is used to measure the common degree of the word in the entire document set, and a high TF-IDF value of a word in a document means that the word is common in the document, but not common in the entire document set, so the word is likely to be a keyword of the document.

[0030] The image features in the image data are extracted using a preset expression, and the preset expression is:

[0031]

[0032] wherein, S(i,j) represents the pixel value of the convolution operation output feature map at position (i,j), I(i+m,j+n) is the pixel value of the input image at position (i+m,j+n), K(m,n) is the element of the convolution kernel, and m,n represent the index of the convolution kernel;

[0033] The numerical features are numerical features of pollutants, site characteristics, environmental factors, etc. obtained through a sensing array, spatial interpolation, on-site measurement, etc.

[0034] The present application also provides a site pollution treatment scheme generation device, the device comprising:

[0035] The data acquisition unit is configured to acquire pollution data of multiple dimensions of a target area, wherein the pollution data at least includes pollutant characteristic data, site characteristic data and environmental factor data.

[0036] The data fusion unit is configured to perform data processing on the pollution data of each dimension respectively, and fuse the pollution data of each dimension after data processing to obtain fused multi-modal data.

[0037] The result generation unit is configured to input the fused multi-modal data into a pre-constructed scheme generation model to obtain a pollution treatment scheme output by the scheme generation model.

[0038] The scheme generation model is constructed by using historical site pollution data after data processing and historical treatment schemes corresponding to the historical site pollution data.

[0039] The application further provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the method when executing the program.

[0040] The application further provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executable on a processor to implement the steps of the method.

[0041] The site pollution treatment scheme generation method and device provided by the application collect pollution data of multiple dimensions of a target area, the pollution data at least including pollutant characteristic data, site characteristic data and environmental factor data, perform data processing on the pollution data of each dimension respectively, fuse the pollution data of each dimension after data processing to obtain fused multi-modal data, input the fused multi-modal data into a pre-constructed scheme generation model, and the pollution treatment scheme output by the scheme generation model can be obtained, wherein the scheme generation model is constructed by using historical site pollution data after data processing and historical treatment schemes corresponding to the historical site pollution data.

[0042] In this way, the method and device provided by the application can effectively convert massive unstructured data such as typical site pollution treatment engineering projects into structured data through deep learning coding, so as to make an artificial intelligence large model learn existing massive engineering cases, repair treatment technologies, and migrate and apply site pollution multi-source multi-modal data to form a site soil and groundwater pollution treatment multi-modal professional intelligent agent, realize intelligent recommendation and iterative optimization of a technical scheme, and ensure sensitive data security, data timeliness and decision accuracy. Therefore, a pollution treatment scheme for selection and reference can be generated through model construction and calling, and the technical problem that site pollution treatment scheme generation completely depends on artificial experience and on-site investigation is solved. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only exemplary, and for those skilled in the art, other drawings can be obtained from the provided drawings without creative labor.

[0044] The structures, proportions, sizes, etc. shown in the specification are only used to cooperate with the content disclosed in the specification for understanding and reading by those skilled in the art, and do not have technical substantive significance, and any modification of the structure, change of the proportion relationship or adjustment of the size, without affecting the effects and purposes that can be achieved by the present application, should still fall within the scope of the technical content disclosed by the present application.

[0045] Figure 1 One of the flowcharts of the site pollution treatment scheme generation method provided by the present application;

[0046] Figure 2 The second flowchart of the site pollution treatment scheme generation method provided by the present application;

[0047] Figure 3 The third flowchart of the site pollution treatment scheme generation method provided by the present application;

[0048] Figure 4 The structural schematic diagram of the site pollution treatment scheme generation device provided by the present application;

[0049] Figure 5 The structural block diagram of the computer device provided by the present application. DETAILED DESCRIPTION

[0050] The embodiments of the present application are described below by specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in the specification. Obviously, the described examples are part of the examples of the present application, not all. Based on the examples in the present application, all other examples obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0051] In one specific embodiment, as shown in Figure 1 The site pollution treatment scheme generation method provided by the present application includes the following steps:

[0052] S110: Collecting pollution data of multiple dimensions of the target area, the pollution data at least including pollutant characteristic data, site characteristic data and environmental factor data; wherein the pollutant characteristic data at least including the type, concentration and spatial distribution of the pollutant; the site characteristic data at least including soil type, pH value, organic matter content and groundwater distribution; the environmental factor data at least including temperature, humidity and rainfall data of the pollution site, which can be collected by various instruments arranged at the collection nodes.

[0053] S120: data processing is performed on the pollution data of each dimension respectively, and the pollution data of each dimension after data processing is fused to obtain fused multi-modal data; since the collected data involves multiple formats, in order to ensure the uniformity of the data format and facilitate the input of the subsequent scheme generation model, the data of each dimension needs to be processed and fused, and the fused data is used as the input data of the subsequent scheme generation model.

[0054] S130: inputting the fused multi-modal data into a pre-constructed scheme generation model to obtain a pollution control scheme output by the scheme generation model; in theory, the input pollution control scheme can be one or more, and is displayed in the form of a table, a document or other visual forms, so as to facilitate the reference and screening of the staff.

[0055] The scheme generation model is constructed by using various types of historical site pollution data after data processing and historical treatment schemes corresponding to the historical site pollution data; during the model construction process, domestic and foreign literatures, case studies, engineering project reports and the like in the field of site pollution control can be collected to form a rich set of unstructured and structured data as an external knowledge base of the large model; the collected data is cleaned and labeled to extract key features and information, so as to ensure the quality and consistency of the data; a site pollution repair technology large model based on retrieval enhanced generation is constructed by using natural language processing (NLP) technology combined with semantic understanding and keyword matching, and the large model is the above-mentioned scheme generation model.

[0056] During use, the scheme generation model can be called to quickly retrieve relevant technologies and cases in the knowledge base according to user needs. Specifically, according to the needs of site ecological restoration, environmental risk reduction or green low-carbon technology, the scheme generation model cross-modal analyzes and identifies the complex pollution characteristics in the site by using the coded information obtained after data processing and fusion, judges the types, sources and potential diffusion paths of pollutants, automatically generates the optimal or better site repair recommendation scheme, intelligently recommends the pollution control scheme such as chemical repair, physical isolation or biological repair technology, and adjusts the repair strategy through a continuously optimized algorithm to ensure the maximization of the site repair effect. That is, the method first constructs a multi-modal large model with enhanced retrieval (various types of site treatment materials as a knowledge base, and a domestic large model such as Wenxin Yiyang as a large model), collects various data of the site to be treated, fuses the multi-modal data, lets the large model analyze, and finally gives the treatment scheme.

[0057] In some embodiments, as shown in Figure 2 the process of constructing the scheme generation model specifically includes the following steps:

[0058] S210: Obtain a large amount of historical site pollution data of various types; in a specific use scenario, when constructing a scheme generation model, first collect structured and unstructured data from a wide range of sources, such as typical site pollution remediation projects, engineering cases, academic literature, professional reports, and other diversified sources accumulated worldwide, especially domestically and internationally; structured data usually refers to information stored in the form of tables, databases, etc., with clear fields and formats, such as specific parameters of pollution remediation technology, detailed data of successful cases, and specific concentration values of site pollutants; unstructured data covers a wider range of content, such as project descriptions, case analyses, expert opinions, and technical documents; these data collectively form the basis for analyzing site pollution characteristics and developing remediation strategies, and form the structured knowledge base required for large models.

[0059] S220: Data cleaning and labeling of target information on historical site pollution data; specifically, data cleaning and preprocessing includes deduplication and standardization processing, text segmentation, and stop word removal, etc.; among them, deduplication and standardization can use hash functions or string comparison algorithms (such as Jaccard similarity) for literature deduplication; text segmentation can use segmentation tools (such as Jieba segmentation) to decompose text into words or phrases; stop word removal refers to removing common stop words to improve information retrieval accuracy. The cleaned data can be represented as: Cleaned_Data = {xi∈Raw_Data∣f(xi)}

[0060] Threshold}, where f(xi) is used to evaluate the relevance of the data.

[0061] S230: Use natural language processing algorithms to extract target features from labeled historical site pollution data, combine semantic understanding and keyword matching to form a site pollution remediation knowledge base, and construct a large site pollution repair technology model based on retrieval enhancement generation.

[0062] Specifically, in step S230, feature extraction and data normalization processing are performed. Through the use of NLP (Natural Language Processing) techniques such as word segmentation, part-of-speech tagging, and named entity recognition, key features in the text are accurately extracted to ensure data quality and consistency. Feature extraction is the process of extracting useful information from raw data for analysis and prediction, reducing dimensionality and improving model efficiency. Data normalization converts data of different scales to a uniform standard, eliminating model performance problems caused by scale differences. Through feature extraction and data normalization, data becomes easier to process, improving model accuracy and convergence speed. Keyword matching quickly locates key information by directly searching for predefined terms or phrases, making it suitable for handling explicit and structured content. Suitable scenarios include technical names (e.g., "chemical immobilization") and pollutant names (e.g., "lead," "benzene"). Implementation methods include regular expression or string matching techniques. Semantic understanding uses large models to deeply analyze unstructured text, extracting implicit information, making it suitable for handling ambiguous and indirectly described content. Suitable scenarios include descriptive pollution site information (e.g., "high concentration of heavy metals in groundwater") and technical application conditions (e.g., "suitable for high-pH soil"). Implementation methods include combining named entity recognition (NER) and dependency syntax analysis to obtain contextual semantics through large models.

[0063] In specific use scenarios, feature extraction can use TF-IDF (Term Frequency-Inverse Document Frequency):

[0064]

[0065] where TF(t, d) is the frequency of word t in document d, N is the total number of documents, and DF(t) is the number of documents containing word t. Term Frequency-Inverse Document Frequency is the Chinese meaning of "Term Frequency-Inverse Document Frequency," commonly abbreviated as TF-IDF. It is a widely used weight calculation method in information retrieval and text mining, used to evaluate the importance of a word to a file set or a file in a corpus. Term Frequency (TF) represents the frequency of a word appearing in a document, while Inverse Document Frequency (IDF) measures the commonality of the word in the entire document set. A high TF-IDF value of a word in a document means that the word is common in the document but rare in the entire document set, so the word is likely to be a key word in the document.

[0066] Word embedding can use Word2Vec, which maps each word to a fixed-dimensional vector. The formula is as follows:

[0067]

[0068] where w t is the center word, w i is the context word, n is the context window size, is to predict the context word (context word) according to the current word (center word), given a sentence, the model wants to maximize the probability of predicting the context word when given the center word. Word2Vec is a computational model for generating word vectors, i.e. the representation of words in a continuous vector space, which was introduced by Google in 2013 and is an efficient distributed word representation method that can capture semantic relationships between words. Word2Vec has two architectures, Continuous Bag-of-Words (CBOW) and Skip-Gram, which are not described in detail here.

[0069] Specifically, data normalization can be achieved using the following formula:

[0070]

[0071] x ′ represents the normalized data, x represents the data before normalization, and this formula is used to normalize the data x to the range [0, 1], where min(X) and max(X) are the minimum and maximum values of the data set x, respectively. This process ensures that data from different sources and dimensions are compared and analyzed on a unified scale.

[0072] The scheme generation model can be understood as a large model based on a retrieval model. To ensure security, a domestic large model is used. The enhanced retrieval model can be implemented using BM25 (Best Matching 25), and the specific expression is:

[0073]

[0074] where BM25(q, d) represents the relevance score of document d and query q, q represents the query (composed of one or more query terms), d represents the document (candidate document in the retrieval system), t i represents the i-th term in the query, f(t i ,d) is the frequency of term t i in document d, |d| is the length of the document, is the average length of the document, IDF(t i ) represents the inverse document frequency of term t i (measuring the importance of the term), k and b are adjustment parameters, with value ranges of [1.2, 2.0] and [0, 1] respectively.

[0075] The retrieval module retrieves relevant documents from the knowledge base using similarity calculation (such as cosine similarity), and the expression of the speed of similarity is:

[0076]

[0077] Where CosineSimilarity(A,B) represents the cosine similarity of AB, A represents the content of the knowledge base, and B represents the content of the query, i.e. site data, ||A|| and ||B|| represent the length of the vector.

[0078] The generation module is used to generate responses according to the retrieval results, using beam search (heuristic search algorithm) or top-k sampling to optimize the diversity and accuracy of the generated.

[0079] Based on the above modular retrieval process, the prompt words required by the large model are automatically generated, and the repair and remediation technology for the specific contaminated site is output, and the expression of the prompt word engineering is:

[0080] Prompt=Template(Variables)

[0081] Where Prompt represents the instructions or questions that the generative model needs to input, which are used to guide the model output, Template(Variables) represents the format or template of the prompt word (for example "provide repair suggestions for the following site: {site conditions}"), Variables represents dynamic variables, which represent the input data provided by the user (such as site conditions, pollutant characteristics).

[0082] Further, the data of the pollutant characteristics of the site to be treated is processed, and then includes:

[0083] According to the soil and water samples collected in the target area, the concentration level of the target pollutant in the soil and water samples is determined;

[0084] Using the sampling data and spatial interpolation method, the concentration distribution map of the target pollutant is drawn, and the spatial distribution characteristics and migration trend of the target pollutant are determined based on the concentration distribution map.

[0085] In one specific use scenario, during the process of collecting and processing multi-modal pollutant sample data of contaminated sites, multi-dimensional pollution data of the site is collected through Internet of Things devices, such as pollutant characteristics, site characteristics, and environmental factors. The pollutant characteristics include the types, concentrations, and spatial distribution of pollutants. The site characteristics include the type of soil, pH value, organic matter content, and groundwater distribution. The environmental factors include temperature, humidity, and rainfall data of the contaminated site. During the site multi-modal data collection and transmission process, multi-dimensional pollution data of the site is collected and transmitted in real time through on-site collection, sensor array, and Internet of Things (I oT) devices, including pollutant characteristics, site characteristics, and environmental factors. The data types of the pollution data at least include text data, image data, and sensor reading data.

[0086] Specifically, there are various types of pollutants, including heavy metals, organic pollutants, and petroleum pollutants, etc., and their concentrations and spatial distribution are different. In order to accurately grasp these information, in this embodiment, the sampling method adopted includes grid point method and random point method, etc., and soil and water samples are widely collected in the contaminated site; these samples are then sent to the laboratory for in-depth analysis to determine the specific concentration level of the pollutants; combined with geographic information system (G I S), the sampling data and spatial interpolation method are used to draw a pollutant concentration distribution map, so as to clearly show the spatial distribution characteristics and potential migration trend of the pollutants.

[0087] In terms of site characteristics, this embodiment mainly focuses on key elements such as soil type, pH value, organic matter content, and groundwater distribution; through careful observation on site and professional laboratory analysis, the soil type and texture of the contaminated site are accurately determined; at the same time, representative soil samples are collected, and the pH value and organic matter content of the soil samples are accurately measured to fully understand the basic chemical properties of the soil. In addition, this embodiment makes full use of geological exploration data and hydrogeological survey data to deeply analyze the distribution range, flow direction and flow rate of groundwater, and further scientifically assesses the migration risk of pollutants in groundwater.

[0088] Environmental factors also have important influence on the diffusion of pollutants and the effect of remediation. In this embodiment, advanced meteorological monitoring equipment is equipped to regularly record the temperature and humidity data of the contaminated site, and to deeply analyze the potential influence of these data on the migration and transformation of pollutants. At the same time, historical rainfall data is widely collected to deeply analyze the specific effect of rainfall on the erosion, migration and diffusion of pollutants, so as to predict the change trend of pollutants under future rainfall conditions. In addition, according to the specific circumstances, this embodiment also fully considers other environmental factors that may affect the remediation effect, such as wind direction, wind speed, and solar radiation, etc.

[0089] In this embodiment, the historical remediation record generally includes remediation scheme and technology, pollutant characteristics, remediation effect evaluation and monitoring data; wherein the remediation scheme and technology are specific to past remediation methods (such as bioremediation, chemical remediation, physical isolation, etc.) and the technical equipment and materials used; the pollutant characteristics include data such as the type, concentration, diffusion path of pollutants at the historical remediation site; the remediation effect evaluation includes the change of pollutant concentration after remediation, remediation time, cost, environmental impact, etc.; the monitoring data refers to the monitoring data during and after the remediation process, including pollutant concentration, environmental conditions, changes in site characteristics, etc.; through data cleaning, labeling and structured processing, it is converted into a usable data set.

[0090] To ensure the real-time and integrity of the data, this embodiment uses a cloud platform for efficient transmission and storage of multi-modal data; the collected data covers soil, groundwater, air quality and other aspects, as well as sensor electrical signal data, image monitoring data and historical remediation records, etc. Through the powerful functions of the cloud platform, we can realize centralized management and rapid access to these data, providing strong support for subsequent pollution analysis and remediation work.

[0091] In some embodiments, the pollution data in each dimension after data processing is fused, specifically including:

[0092] The text data is encoded using a word embedding algorithm to obtain an embedding vector corresponding to the historical remediation record text data of the site to be remediated;

[0093] Extracting image features from the image data;

[0094] Extracting pollutant and environmental numerical features;

[0095] The embedding vector and the image feature are fused through a data fusion algorithm to obtain a fusion feature output by the deep learning model; specifically, through deep learning encoding (word encoding, image feature extraction, numerical normalization, etc.) technology, data fusion algorithm, fusion of multi-modal data such as soil, groundwater, air quality, etc. sensor electrical signal data, image monitoring and historical pollution remediation record text data of the site to be remediated, multi-modal data is transmitted and processed through the cloud platform, forming a unified data model to support subsequent pollution analysis and decision-making.

[0096] Based on the pre-processed multi-modal data, this embodiment constructs a cross-modal pollution site data fusion model to make full use of the advantages of deep learning technology to accurately identify the types, sources and diffusion paths of pollutants in complex pollution sites. Through cross-modal analysis, the model can comprehensively consider various factors such as soil characteristics (including soil type, pH value, organic matter content, etc.), environmental conditions (such as temperature, humidity, rainfall, etc.) and treatment technology characteristics, etc., for comprehensive and in-depth analysis.

[0097] The text data encoding can use Word2Vec, which maps each word to a fixed-dimensional vector, and the formula is as follows:

[0098]

[0099] where w t is the center word, w i is the context word, n is the context window size, and is to predict the context word (context word) according to the current word (center word). Given a sentence, the model wants to maximize the probability of predicting the context word given the center word. Word2Vec is a computational model for generating word vectors, i.e. the representation of a word in a continuous vector space. This computational model is a high-efficiency distributed word representation method that can capture the semantic relationship between words. Word2Vec has two architectures, Continuous Bag-of-Words (CBOW) and Skip-Gram, which are not described in detail here.

[0100] The image features are extracted using a convolutional neural network, and the expression is as follows:

[0101]

[0102] where S(i,j) represents the pixel value of the output feature map at position (i,j) of the convolution operation, I(i+m,j+n) is the pixel value of the input image at position i+m,j+n, K(m,n) is the element of the convolution kernel, and m,n represent the index of the convolution kernel.

[0103] The expression of cross-modal data fusion is as follows:

[0104] Y=g(X1,X2,…,X n ;θ)

[0105] where Y is the model output (multi-modal data fusion vector), X i is the i-th modal data (such as field image, historical text, etc.), i ranges from 1 to n, g is a deep learning model, and θ is a model parameter. This formula shows how to integrate the features of multiple modal data.

[0106] Further, in the process of intelligent decision-making and pollution control scheme recommendation, according to the needs such as site ecological restoration, environmental risk reduction, green low-carbon technology, relying on the multi-modal large model of site repair and treatment technology, our system has shown excellent intelligence in the field of site pollution repair. The workflow of the large model begins with accurate grasp of user needs and in-depth analysis of on-site collected pollutant data. In the scheme generation model based on the construction, according to the actual multi-modal data of the site, through the prompt word engineering, the system intelligently selects the optimal treatment scheme from the vast treatment technology library, and outputs the selectable treatment scheme to assist in scheme generation. The generated scheme covers various technical means such as physical repair, chemical repair, biological repair, and integrated repair system integrating multiple technologies. Physical repair may include soil excavation, leaching, solidification / stabilization, etc.; chemical repair may involve oxidation, reduction, extraction, etc. chemical reactions; biological repair uses the metabolic activity of microorganisms or plants to degrade pollutants, and an automatically derived integrated scheme system. The treatment scheme recommended by the model not only considers the current pollution situation, but also fully considers the future trend of change. Through the continuous optimization algorithm, the repair strategy can be continuously iterated and improved to adapt to the changes of pollution conditions, and the ability of dynamic adjustment ensures the maximization of repair effect and avoids the poor treatment effect caused by the rigidity of the scheme.

[0107] In actual operation, the treatment scheme generation method can be implemented by the system as a carrier. The system can provide users with detailed treatment scheme reports, including specific steps of recommended technologies, required materials, expected effects, and possible risks and countermeasures. At the same time, the system also supports users to adjust and optimize the scheme according to the actual situation to ensure that the finally implemented treatment scheme is both scientific and feasible.

[0108] Further, to achieve continuous monitoring and adaptive optimization, during the implementation phase of site pollution remediation, the system on which the embodiment relies exhibits strong real-time monitoring and optimization capabilities. Through Internet of Things devices such as sensor networks, unmanned aerial vehicle inspection, remote monitoring cameras, etc., the system can continuously collect the latest data of the site, including soil pollution concentration, groundwater quality, air quality, temperature, humidity, and other key indicators. These data are transmitted to the system background in real time, providing strong support for the monitoring of remediation effects. The system uses advanced algorithms and models to analyze the real-time monitoring data in depth to evaluate the effectiveness of the remediation measures. By comparing the data changes before and after remediation, the system can intuitively show the remediation effects, such as the degree of reduction of pollutant concentration, the improvement of water quality and air quality, etc. This real-time feedback mechanism ensures that the remediation process is always under control, facilitating the timely discovery and resolution of problems. More importantly, the system does not stop at the monitoring level but uses new monitoring data as input to continuously iterate and optimize the remediation scheme. Through data-driven decision support, the system can automatically adjust the remediation strategy, such as changing the parameters of physical remediation, adjusting the amount of chemical reagents, optimizing the conditions of biological remediation, etc., to adapt to changes in pollution conditions and the needs of remediation effects. This dynamic optimization process ensures that the remediation scheme is always in the best state, improving remediation efficiency and reducing secondary environmental impact.

[0109] During the remediation implementation process, the system also considers various factors such as weather conditions, soil types, groundwater flow direction, etc. to develop more accurate remediation strategies. By considering these factors comprehensively, the system can predict and respond to potential challenges and risks, ensuring the safety and reliability of the remediation process. Ultimately, through continuous monitoring, real-time feedback, and data iteration, the system can ensure that the efficiency and effectiveness of site remediation reach the optimal level; at the same time, through dynamic optimization of the pollution control process, we minimize the secondary impact on the environment, achieving green and sustainable remediation goals. To ensure data security and privacy, the system uses strict data encryption techniques such as data desensitization and permission control mechanisms, especially for the processing of sensitive data, to ensure the timeliness and confidentiality of the data.

[0110] For ease of understanding, the following takes a specific use scenario as an example to briefly describe the implementation process of the method provided by the present application.

[0111] For example, as shown in FIG. 1, a site contaminated by heavy metals is to be remediated. The system first conducts a comprehensive survey of the site, including soil sampling, groundwater sampling, and air sampling, to determine the extent and nature of the pollution. Based on the survey results, the system develops a remediation plan tailored to the specific conditions of the site. The plan includes the selection of appropriate remediation technologies, such as physical remediation, chemical remediation, and biological remediation, as well as the determination of the optimal parameters for each technology. Figure 3As shown, due to long-term industrial production activities, the soil and groundwater in a certain chemical park are severely polluted, with main pollutants including heavy metals (such as chromium and copper) and organic chemicals (such as benzene series and ketone solvents). The presence of these pollutants not only affects soil quality, but also seeps into groundwater, posing potential threats to the surrounding ecological environment and residents' health. Therefore, it is necessary to implement contaminated site remediation and use an intelligent decision-making method based on a multi-modal large model to guide the development of remediation plans.

[0112] Pollution monitoring and data collection: By combining unmanned aerial vehicle remote sensing, ground sampling, and sensor arrays, comprehensive pollutant sample data of the contaminated site in the park are obtained; the data include heavy metal content, chemical substance concentration, soil pH value, temperature and humidity, rainfall, and other environmental factors in the soil and groundwater, as well as the first-stage treatment records, to comprehensively understand the types, concentrations, and spatial distribution of pollutants.

[0113] Pollutant characteristics and remediation target analysis: The analysis results show that the concentrations of heavy metals such as chromium and copper in the park are seriously exceeded, and there is a large range of benzene series and ketone chemical solvent pollution; by integrating data such as soil types and groundwater flow conditions, the model determines the mobility and influence range of pollutants, and further defines the areas to be remediated.

[0114] Intelligent decision-making and remediation plan development: After inputting the pollutant characteristics, site characteristics, and environmental data into the multi-modal large model with knowledge base enhanced retrieval (this example uses the domestic reasoning large model Tian Gong 4.0 (o1) as an example), the large model quickly analyzes and proposes remediation targets and methods; according to the recommended scheme generated by the model, the remediation team develops the following specific remediation schemes and gives related parameters, such as pollutant concentration threshold, analyzes possible pollution diffusion path, time and space dosing parameters, treatment cycle, etc.

[0115] Remediation of heavy metal contaminated areas;

[0116] Soil leaching method: high-efficiency soil leaching technology is used to remove heavy metal ions such as chromium and copper in the soil through water solvent;

[0117] Solidification and stabilization technology: using solidification agent to fix the residual heavy metals in the soil to prevent their migration and diffusion again;

[0118] Remediation of organic pollutant areas;

[0119] Biodegradation technology: by introducing specific microorganisms to promote the degradation of benzene series and ketone solvents, the harm of these substances to the environment and human body is reduced;

[0120] Chemical oxidation: Adding oxidants to facilitate the transformation of organic pollutants and enhance degradation. Multi-pollutant remediation: A combination of various techniques is used to address different pollutants, aiming to minimize their impact on soil and groundwater and achieve effective removal and ecological restoration.

[0121] Remediation of areas with coexistence of heavy metals and organic pollutants, physical and chemical remediation: Soil leaching combined with chemical oxidation: While addressing heavy metal pollution, soil leaching is used to remove soluble heavy metals, and chemical oxidation is used to degrade accompanying organic pollutants. This combination can effectively reduce pollutant concentrations in a short time and enhance the universality of remediation; modified soil solidification: For residual heavy metals, solidification and stabilization technology can be used. Modified soil solidification fixes harmful substances in the soil by adding specific chemicals to prevent the migration of pollutants. Organic substances released during solidification are further degraded by chemical oxidation technology.

[0122] Remediation of areas with coexistence of organic pollutants and nutrient pollution, biological remediation and nutrient regulation: biodegradation and nutrient supplementation: For the degradation of organic pollutants, specific microbial populations are used for biological remediation, and the growth and degradation efficiency of microorganisms are promoted by adjusting the nutrients in the soil (such as nitrogen, phosphorus, etc.). By optimizing environmental conditions, the remediation effect is further improved. Microbial-induced precipitation: In certain cases, microorganisms can promote the degradation of organic pollutants in the soil, and the byproducts produced during the degradation process can help solidify heavy metals, forming a more stable complex pollution environment.

[0123] Multi-pollutant alternative remediation technology, phased remediation:

[0124] Phase one: Pretreatment of pollutants: For complex pollution, physical and chemical methods are first used for preliminary removal of pollutants. For example, soil leaching is used to reduce heavy metal concentration, or surfactant extraction is used to remove organic pollutants. The purpose of this stage is to reduce the concentration of pollutants through simple and efficient methods to create conditions for subsequent biological remediation.

[0125] Phase two: Combination of biological remediation and stabilization: After pretreatment, biological remediation technology is used to address microbial degradation of organic pollutants, and stabilization technology is used to fix heavy metals in the soil. This stage focuses on improving the durability of remediation and reducing the possibility of pollutant migration.

[0126] Phase three: Synergistic effect of ecological restoration and complex remediation system: Through plant remediation, soil improvement, and other measures, the ecological function of the site is restored, and intelligent monitoring and adjustment are used to ensure long-term remediation.

[0127] Repair process and monitoring: During the repair process, monitoring technology is used to track the repair effect in real time, including key parameters such as pollutant concentration, soil and groundwater quality; based on the monitoring results, the repair team can dynamically adjust the repair scheme to ensure that the repair effect reaches the expected target.

[0128] Repair effectiveness: By using the multi-modal based intelligent decision-making method, the pollution site remediation work in the chemical industrial park has achieved remarkable results; the repair team can quickly and accurately determine the repair target and scheme, avoiding the inefficiency and blindness that may occur in traditional repair methods; due to the full integration of multi-source data from remote sensing, ground sampling, and environmental factors of sensor arrays, and based on the retrieval and reasoning of large models in existing technology library, the decision-making result is more accurate, effectively improving the repair effect and reducing the repair cost.

[0129] The successful application of this project verifies the effectiveness of the multi-modal based site pollution remediation intelligent decision-making method, through comprehensive data analysis and model prediction, the repair work is more scientific and efficient, providing a reference for future pollution site remediation.

[0130] In the above specific embodiments, the method provided by the present application utilizes massive unstructured data of typical site pollution treatment engineering projects, etc., which can effectively convert these unstructured data into structured data through deep learning coding, so as to enable artificial intelligence large models to learn existing massive engineering cases, remediation and treatment technologies, and site multi-source and multi-modal pollution data through retrieval augmentation generation (RAG), prompt engineering, etc., and migrate to form a multi-modal professional intelligent agent for site soil and groundwater pollution treatment, realize intelligent recommendation and iterative optimization of technical schemes, and ensure sensitive data security, data timeliness and decision accuracy. Through model construction and retrieval, pollution treatment schemes for selection and reference are generated, solving the technical problem that site pollution treatment scheme generation completely depends on artificial experience and field investigation.

[0131] In addition to the above method, the present application also provides a site pollution treatment scheme generation device, as shown in Figure 4 The device comprises:

[0132] The data acquisition unit 410 is configured to acquire pollution data of multiple dimensions of a target area, wherein the pollution data at least includes pollutant characteristic data, site characteristic data and environmental factor data;

[0133] The data fusion unit 420 is configured to perform data processing on the pollution data of each dimension respectively, and fuse the pollution data of each dimension after data processing to obtain fused multi-modal data;

[0134] The result generation unit 430 is configured to input the fused multi-modal data into a pre-constructed scheme generation model to obtain a pollution treatment scheme output by the scheme generation model.

[0135] The scheme generation model is constructed by using historical site pollution data after data processing and historical treatment schemes corresponding to the historical site pollution data.

[0136] In one embodiment, a computer device, which can be a server, has an internal structure as shown in Figure 5 The computer device includes a processor, a memory and a network interface connected by a system bus. The processor of the computer device is configured 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 model prediction. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The model prediction of the computer device is configured to store static information and dynamic information data. The network interface of the computer device is configured to communicate with an external terminal through a network connection. The computer program is executed by the processor to implement the steps in the above method embodiments.

[0137] Those skilled in the art can understand that Figure 5 The structure shown in the above embodiment is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0138] Corresponding to the above-mentioned embodiments, the present application also provides a computer storage medium containing one or more program instructions. The one or more program instructions are configured to be executed to perform the above-mentioned method.

[0139] The present application also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer readable storage medium. When the computer program is executed by a processor, the computer can execute the above-mentioned method.

[0140] In an embodiment of the present application, the processor can be an integrated circuit chip with the processing capability. The processor can be a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components.

[0141] The disclosed methods, steps, and logic block diagrams in the embodiments of the present application can be implemented or executed by using a general purpose processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof. The steps of the methods disclosed in the embodiments of the present application can be directly embodied as one or more hardware coding processes executed by the processor, or embodied as software modules executed by the coding processor. The software modules can be located in the memory, the flash memory, the Read Only Memory (ROM), the Programmable Read Only Memory (PROM), the Electrically Programmable Read Only Memory (EPROM), the Electrically Erasable Programmable Read Only Memory (EEPROM), the registers, or other forms of the storage medium in the art. The processor reads information in the storage medium and combines the hardware to implement the steps of the above methods.

[0142] The storage medium can be a memory, for example, can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories.

[0143] The non-volatile memory can be a Read Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically EPROM (EEPROM), or a flash memory.

[0144] The volatile memory can be a Random Access Memory (RAM), which is used as an external cache. By way of example, and not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The below-described subject matter can be implemented with computers using one or more of the above or any other suitable memory.

[0145] The storage media described in the embodiments of the present application is intended to include, but not be limited to, these and any other suitable types of memory.

[0146] Those skilled in the art should be aware that the functions described in the embodiments of the present application can be implemented in combination of hardware and software in one or more of the above examples. When the software is applied, the corresponding functions can be stored in a computer readable medium or transmitted as one or more instructions or codes on the computer readable medium. The computer readable medium includes a computer storage medium and a communication medium, wherein the communication medium includes any medium that facilitates the transfer of computer programs from one place to another. The storage medium can be any available medium that can be accessed by a general or special purpose computer.

[0147] The above detailed description is further intended to serve as a purpose, technical solutions and beneficial effects of the present application. It should be understood that the above is only a specific embodiment of the present application, and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made on the basis of the technical solutions of the present application shall be included in the protection scope of the present application.

Claims

1. A method for generating a site contamination remediation plan, characterized in that, The method includes: Collect pollution data from multiple dimensions of the target area, including at least pollutant characteristic data, site characteristic data, and environmental factor data; The pollution data of each dimension is processed separately, and the processed pollution data of each dimension is then fused to obtain fused multimodal data. The fused multimodal data is input into a pre-built scheme generation model to obtain the pollution control scheme output by the scheme generation model; The scheme generation model is constructed using historical site pollution data after data processing, as well as historical remediation schemes corresponding to the historical site pollution data. The types of pollution data include at least text data, image data, and sensor reading data; This involves fusing pollution data from various dimensions after data processing, specifically including: The text data is encoded using a word embedding algorithm to obtain the embedding vector corresponding to the historical governance record text data of the site to be governed; Extract image features from the image data; Extract pollutant and environmental numerical features; combine the embedded vectors and image features using a data fusion algorithm to obtain the fused features output by the deep learning model; Among them, document feature extraction adopts - : in, It is a word In the document The frequency in, where N is the total number of documents, It is a word containing The number of documents; Image features are extracted from the image data using a preset expression, whereby the preset expression is: in, This indicates that the output feature map of the convolution operation is at position ( , The pixel value of ) The input image is at position pixel values, These are elements of the convolution kernel. This represents the index of the convolution kernel; Numerical characteristics are the numerical features of pollutants, site characteristics, and environmental factors obtained through operations such as sensor arrays, spatial interpolation, and on-site measurements. The expression for cross-modal data fusion is: in, It is the model output. This represents the data for the i-th mode, where the value of i ranges from 1 to n. It is a deep learning model. These are model parameters.

2. The method for generating a site pollution remediation plan according to claim 1, characterized in that, The specific components of the solution generation model include: Acquire a large amount of historical site contamination data of various types; The historical site contamination data was cleaned, and target information was marked. By using natural language processing algorithms, target features are extracted from labeled historical site pollution data. Combined with semantic understanding and keyword matching, a large-scale site pollution remediation technology model based on retrieval enhancement is constructed.

3. The method for generating a site pollution remediation plan according to claim 1, characterized in that, The pollutant characteristic data includes at least the type, concentration, and spatial distribution of the pollutant; and / or, The site characteristic data shall include at least soil type, pH value, organic matter content, and groundwater distribution; and / or, The environmental factor data includes at least the temperature, humidity, and rainfall data of the contaminated site.

4. The method for generating a site pollution remediation plan according to claim 1, characterized in that, Data processing of pollutant characteristic data, followed by: Based on extensively collected soil and water samples from the target area, determine the concentration levels of the target pollutants in the soil and water samples. Using sampling data and spatial interpolation methods, a concentration distribution map of the target pollutant is drawn, and the spatial distribution characteristics and migration trend of the target pollutant are determined based on the concentration distribution map.

5. A site pollution remediation scheme generation device, based on the site pollution remediation scheme generation method according to any one of claims 1-4, characterized in that, The device includes: The data acquisition unit is used to collect pollution data from multiple dimensions of the target area, including at least pollutant characteristic data, site characteristic data, and environmental factor data. The data fusion unit is used to process the contamination data of each dimension separately, and then fuse the processed contamination data of each dimension to obtain fused multimodal data. The result generation unit is used to input the fused multimodal data into a pre-built scheme generation model to obtain the pollution control scheme output by the scheme generation model. The proposed solution generation model is constructed using historical site pollution data after data processing, as well as historical remediation solutions corresponding to the historical site pollution data.

6. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in any one of claims 1-4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1-4.

Images