Thinking chain-driven dangerous chemical analysis and pretreatment method and system
Through the fusion of multi-perspective analysis driven by thinking chain and multiple analysis modes, the analysis and pre-processing of hazardous chemicals is solved by using large language model LLMs, which solves the problems of slow updates of existing systems and the singleness of technology stacks, and improves the reliability of analysis and flexibility, enhancing hazardous chemical management and accident response capabilities.
Patent Information
- Application Number
- CN202411034531.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-22
AI Technical Summary
The existing hazardous chemical analysis system based on machine learning and deep learning has problems such as slow technology updates, single technology stack and inconsistent training data sets, making it difficult to build a complete hazardous chemical consulting program, affecting hazardous chemical management and Accident response ability.
Using thinking chain-driven multi-perspective analysis and multiple analysis modes, using large language model LLMs for hazardous chemical analysis and pre-treatment, through work splitting, multiple analysis, cross-scoring and fusion analysis, a one-time analysis of the entire process is generated. Reporting and preprocessing solutions.
It improves the reliability, flexibility and objectivity of hazardous chemical analysis, ensures the timeliness and accuracy of analysis results, and enhances the management and accident response capabilities of hazardous chemicals.
Smart Images

Figure CN119006247A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a thought chain driven hazardous chemical analysis and pretreatment method and system. Background Art
[0002] In the warehouses of large manufacturing companies, there are often many chemical raw materials, which are potential explosion risks. In order to effectively prevent the occurrence of hazardous chemical explosion accidents, we must take a series of preventive measures. This includes classifying and storing different hazardous chemicals, strictly controlling the temperature and humidity of the warehouse, ensuring the safe loading and unloading of hazardous chemicals, keeping the equipment in good condition, completely eliminating possible fire sources, and making adequate firefighting preparations.
[0003] The production safety authorities are well aware of the seriousness of hazardous chemical explosion accidents and will conduct in-depth investigations on such accidents to form comprehensive and accurate explosion accident investigation reports. In order to analyze the explosion accident investigation reports more deeply and propose corresponding prevention and rescue measures, we usually use a variety of intelligent analysis systems. These systems include but are not limited to technical expert group consultations, machine learning-based hazardous chemical analysis systems, and deep learning-based hazardous chemical analysis systems. Through these systems, we can obtain the types of hazardous chemicals, detailed physical and chemical properties, and obtain targeted prevention and rescue suggestions.
[0004] However, these systems also have certain challenges. For example, technical expert group consultations may face difficulties in coordinating online and offline meetings, which may affect the timeliness of analysis and consultation. Hazardous chemical analysis systems based on machine learning are difficult to build a complete hazardous chemical consulting solution because the algorithms they use are gradually losing popularity, technology updates are becoming slow, and the singleness of the technology stack makes it difficult to effectively disassemble the problem. Similarly, hazardous chemical analysis systems based on deep learning also face similar problems. Due to the inconsistency of training data sets and model structures, as well as the limitations of the technology stack, it is difficult to build a complete consulting solution. In order to solve these problems, we need to continue to explore and innovate in order to improve our ability to manage hazardous chemicals and respond to accidents. Summary of the invention
[0005] The purpose of the present invention is to solve the above-mentioned problems and to provide a thought chain driven hazardous chemical analysis and pretreatment method and system. The thought chain driven method ensures that the whole process of hazardous chemical analysis and pretreatment tasks, including work splitting, multiple analysis, cross-scoring, fusion analysis, and report construction, is completed in one go, thereby technically ensuring the reliability, flexibility, and objectivity of hazardous chemical analysis.
[0006] To achieve the above-mentioned purpose, the technical solution of the present invention is: a thinking chain driven hazardous chemical analysis and pretreatment method, which utilizes the thinking chain driven multi-perspective analysis of hazardous chemicals and the thinking chain driven fusion analysis of hazardous chemicals using multiple analysis modes to complete the hazardous chemical analysis business and generate a hazardous chemical analysis report; utilizes the thinking chain driven calculation of hazardous chemical pretreatment solutions to complete the hazardous chemical pretreatment business and generate pretreatment solutions.
[0007] In one embodiment of the present invention, the thought chain driven multi-perspective analysis of hazardous chemicals uses large language models (LLMs) as analysis tools, specifically using multiple large language models (LLMs) to perform multi-perspective analysis of hazardous chemicals.
[0008] In one embodiment of the present invention, the thought chain driven multi-perspective analysis of hazardous chemicals is specifically implemented in the following steps: 1) the user starts the thought chain driven hazardous chemical analysis task; 2) the thought chain drives the acquisition of a list of hazardous chemical analysis models, and the hazardous chemical analysis model adopts a general analysis model based on a large language model (LLMs) which is retrained after optimization of the hazardous chemical field database; 3) the thought chain drives the distribution of the explosion accident investigation report and analysis prompt words to each hazardous chemical analysis model; 4) each hazardous chemical analysis model outputs the hazardous chemical analysis results, including information on physical and chemical properties and preventive measures.
[0009] In one embodiment of the present invention, the thought chain drives the use of multiple analysis modes to fuse and analyze hazardous chemicals, uses multiple fusion analysis modes to fuse and analyze hazardous chemicals, and then regresses and evaluates the fusion analysis results, and uses the optimal analysis result as the result of the fusion analysis.
[0010] In one embodiment of the present invention, the thought chain-driven fusion analysis of hazardous chemicals using multiple analysis modes is specifically implemented as follows: 1) obtaining a list of analysis results of multi-perspective analysis of hazardous chemicals; 2) obtaining fusion analysis results of the fusion analysis mode driven by the thought chain; 3) scoring the fusion analysis results by comparing with historical explosion accident investigation reports; 4) using the fusion analysis result with the highest score as the final hazardous chemical analysis result, including information on physical and chemical properties and preventive measures.
[0011] In one embodiment of the present invention, in step 2), the fusion analysis results of the fusion analysis model driven by the thinking chain are obtained, specifically: a fusion analysis model driven by the thinking chain is obtained, including linear fusion, nonlinear fusion, mesh fusion, and tree fusion, and hazardous chemical data and hazardous chemical explosion data are extracted from the analysis results of hazardous chemicals, and analysis prompt words of the large language model LLMs are generated to obtain analysis results, and linear fusion, nonlinear fusion, mesh fusion, and tree fusion are used to fuse the analysis results.
[0012] In one embodiment of the present invention, in step 3), by comparing with historical explosion accident investigation reports, multiple dimensions of indicators are preferably used to score the fusion analysis results. For example, an evaluation matrix x, i.e., the fusion analysis results, is formed from the correlation of hazardous chemicals, independence of hazardous chemicals, completeness of physical / chemical analysis, accuracy of physical / chemical analysis, and rationality of analysis reports. Then, the Mahalanobis distance is used to calculate and score the fusion analysis results. The Mahalanobis distance formula is as follows:
[0013]
[0014] Where S is the data covariance matrix, S -1 is the inverse matrix of the covariance matrix, y is the mean vector, and x is the new fusion analysis result. Unlike the Euclidean distance, the Mahalanobis distance takes into account the shape and direction of the data. In a multidimensional space with different variances and correlations, the Mahalanobis distance can more accurately reflect the actual distance between points. Based on the value of the Mahalanobis distance, the fusion analysis result can be scored. For example, a threshold can be set, and the smaller the distance, the higher the score, indicating that the consistency of the fusion analysis result with the historical data is better.
[0015] In one embodiment of the present invention, the thought chain driven calculation of hazardous chemical pretreatment scheme has specific implementation steps: 1) obtaining the fusion analysis results of hazardous chemicals driven by the thought chain using multiple analysis modes for fusion analysis; 2) obtaining physical property information of hazardous chemicals; 3) obtaining chemical property information of hazardous chemicals; 4) calculating the degree of explosion hazard of hazardous chemicals based on the physical and chemical properties of hazardous chemicals; 5) calculating protective measures and rescue measures based on the calculated degree of explosion hazard of hazardous chemicals.
[0016] In one embodiment of the present invention, in step 4), the explosion hazard degree of the hazardous chemicals is calculated according to the physical and chemical properties of the hazardous chemicals, specifically:
[0017] First, the hazard index is calculated based on the physical and chemical properties of hazardous chemicals:
[0018] Danger index (DI) = a·density + b·boiling point + c·flash point + d·acidity and alkalinity + e·flammability + f·explosiveness;
[0019] Among them, a, b, c, d, e, f are weight coefficients;
[0020] Secondly, calculate the degree of hazard of hazardous chemical explosion:
[0021]
[0022] Among them, w i refers to the weight of the ith physical or chemical property, P irefers to the probability of the i-th physical or chemical property.
[0023] The present invention also provides a thought chain driven hazardous chemical analysis and pretreatment system / server, which executes the above method, including:
[0024] The hazardous chemicals analysis business module uses the multi-perspective analysis of hazardous chemicals driven by the thinking chain and the integration of multiple analysis modes driven by the thinking chain to analyze hazardous chemicals, complete the hazardous chemicals analysis business, and generate hazardous chemicals analysis reports;
[0025] The hazardous chemical pretreatment business module uses the thinking chain-driven calculation of hazardous chemical pretreatment solutions to complete hazardous chemical pretreatment business and generate pretreatment solutions.
[0026] Compared with the prior art, the present invention has the following beneficial effects: the present invention adopts large language models (LLMs) as analysis tools, and uses the idea of thinking chain to disassemble the hazardous chemical analysis business and hazardous chemical pretreatment business, and splits them into: multi-perspective analysis of hazardous chemicals driven by thinking chain, fusion analysis of hazardous chemicals driven by thinking chain, and hazardous chemical pretreatment solutions driven by thinking chain; through two-stage hazardous chemical analysis, using a variety of analysis models with different characteristics and fusion analysis of multiple modes, the reliability, flexibility and objectivity of the analysis are relatively guaranteed to the greatest extent. The whole process of work splitting, multiple analysis, cross-scoring, fusion analysis, and report construction in the present invention is completed in one go, which technically guarantees the reliability, flexibility and objectivity of hazardous chemical analysis. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] Figure 1 Schematic diagram of the hazardous chemical analysis and pretreatment method driven by the thought chain.
[0028] Figure 2 Analyze hazardous chemicals from multiple perspectives driven by thought chain.
[0029] Figure 3 Driven by the thought chain, multiple analysis modes are integrated to analyze hazardous chemicals.
[0030] Figure 4 A variety of fusion analysis models driven by thought chain.
[0031] Figure 5 A computational hazardous chemical pretreatment solution driven by thought chain.
[0032] Figure 6 A hazardous chemical analysis and pretreatment system / server driven by ThinkChain. DETAILED DESCRIPTION
[0033] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings.
[0034] The present invention provides a hazardous chemical analysis and pretreatment method driven by a thinking chain, which uses a thinking chain-driven multi-perspective analysis of hazardous chemicals and a thinking chain-driven fusion analysis of hazardous chemicals using multiple analysis modes to complete hazardous chemical analysis services and generate hazardous chemical analysis reports; and uses a thinking chain-driven calculation of hazardous chemical pretreatment solutions to complete hazardous chemical pretreatment services and generate pretreatment solutions.
[0035] The present invention also provides a thought chain driven hazardous chemical analysis and pretreatment system / server, which executes the above method, including:
[0036] The hazardous chemicals analysis business module uses the multi-perspective analysis of hazardous chemicals driven by the thinking chain and the integration of multiple analysis modes driven by the thinking chain to analyze hazardous chemicals, complete the hazardous chemicals analysis business, and generate hazardous chemicals analysis reports;
[0037] The hazardous chemical pretreatment business module uses the thinking chain-driven calculation of hazardous chemical pretreatment solutions to complete hazardous chemical pretreatment business and generate pretreatment solutions.
[0038] The following is the specific implementation process of the present invention.
[0039] like Figure 1 As shown, the present invention proposes a thought chain driven hazardous chemical analysis and pretreatment method, including: 1) multi-perspective analysis of hazardous chemicals driven by thought chain, see Figure 2 ; 2) Thinking chain driven fusion analysis of hazardous chemicals, see Figure 3 ; 3) Thinking chain driven hazardous chemical pretreatment plan, see Figure 5 .
[0040] Among them, step 1) and step 2) involve the concept of thought chain. Thought chain can decompose complex business into a series of reasoning and problem solving, emphasizing clear logical decomposition and building logically rigorous business solutions. Specifically, the present invention uses large language models (LLMs) as analysis tools, and uses the thought chain idea to decompose hazardous chemical analysis business and hazardous chemical pretreatment business. Use thought chain driven multi-perspective analysis of hazardous chemicals (see Figure 2 ) and thought chain drive to use multiple analysis modes to integrate and analyze hazardous chemicals ( Figure 3 ), 2-stage analysis, complete hazardous chemical analysis business, generate hazardous chemical analysis report. Use the thinking chain driven calculation hazardous chemical pretreatment plan to complete the hazardous chemical pretreatment business and generate pretreatment solutions.
[0041] 1. The multi-perspective analysis of hazardous chemicals driven by the thinking chain uses the large language model LLMs as an analysis tool. Here, multiple LLMs models are used to analyze hazardous chemicals from multiple perspectives. For details, see Figure 2 .
[0042] Thinking Chain Drive uses multiple hazardous chemical analysis models to complete the multi-perspective analysis of hazardous chemicals. The detailed steps are as follows: 1) The user starts the hazardous chemical analysis task driven by Thinking Chain. 2) Thinking Chain Drive obtains the list of hazardous chemical analysis models. 3) Thinking Chain Drive distributes the explosion accident investigation report and analysis prompts to the hazardous chemical analysis model. 4) The hazardous chemical analysis model outputs the analysis results. 5) The hazardous chemical analysis results include information such as physical and chemical properties and preventive measures.
[0043] Step 2) The list of hazardous chemical analysis models is obtained by thinking chain drive. The general analysis model based on LLMs is preferred, and the LLMs model is optimized and retrained by the hazardous chemical field database. LLMs can be large language models such as GPT, LLaMa and QWen. The hazardous chemical field database includes explosion accident investigation reports, journals and papers, etc.
[0044] Step 3) The thought chain drives the explosion accident investigation report and analysis prompts to be distributed to the hazardous chemical analysis model. Construct LLMs prompts for multi-perspective analysis around physical properties, chemical characteristics, physical hazards, chemical hazards, physical preventive measures, chemical preventive measures, typical disposal cases, hazardous chemical standards, hazardous chemical regulations, etc.
[0045] Steps 4 and 5) The hazardous chemical analysis results include information such as physical and chemical properties and preventive measures. Generate multi-perspective hazardous chemical analysis results based on the physical characteristics, chemical characteristics, and preventive measures of hazardous chemicals, and evaluate and filter the multi-perspective hazardous chemical analysis results to filter out low-quality hazardous chemical analysis results. Prevent low-quality analysis results from interfering with the subsequent fusion analysis process.
[0046] 2. Fusion analysis of hazardous chemicals driven by the thinking chain. Use multiple fusion analysis modes to analyze hazardous chemicals. Then regress and evaluate the fusion analysis results. Take the best analysis result as the result of fusion analysis. For details, see Figure 3 .
[0047] The specific steps are as follows: 1) Obtain a list of analysis results of multi-perspective analysis of hazardous chemicals. 2) Obtain the fusion analysis mode driven by the thinking chain, linear fusion, nonlinear fusion, mesh fusion, and tree fusion of the above analysis results. 3) Score the fusion analysis results by comparing with historical explosion accident investigation reports. 4) Take the fusion analysis result with the highest score as the final hazardous chemical analysis result. 5) The hazardous chemical analysis results include information such as physical and chemical properties and preventive measures.
[0048] Step 2) Obtain the fusion analysis mode driven by the thinking chain, linear fusion, nonlinear fusion, mesh fusion, and tree fusion of the above analysis results. Extract hazardous chemical data and hazardous chemical explosion data from the analysis results of hazardous chemicals. This step is equivalent to the hazardous chemical analysis subtask. Then use the above data to generate the LLMs analysis prompt words for the analysis task and obtain the results of the hazardous chemical analysis subtask (LLMs is an optional tool, and the working principle of LLMs is not expanded). The results of the hazardous chemical analysis subtask require a fusion analysis driven by the thinking chain. The optional fusion analysis modes include: linear fusion, nonlinear fusion, mesh fusion, and tree fusion, see Figure 4 The purpose of fusion analysis is to obtain a logically rigorous comprehensive analysis report.
[0049] Step 3) By comparing with historical explosion accident investigation reports, multiple dimensions of indicators are selected to score the fusion analysis results. For example, the evaluation matrix x, i.e., the fusion analysis results, is formed from the correlation of hazardous chemicals, independence of hazardous chemicals, completeness of physical / chemical analysis, accuracy of physical / chemical analysis, and rationality of analysis reports. The Mahalanobis distance is then used to calculate and score the fusion analysis results. The Mahalanobis distance formula is as follows:
[0050]
[0051] Where S is the data covariance matrix, S -1 is the inverse matrix of the covariance matrix, y is the mean vector, and x is the new fusion analysis result. Unlike the Euclidean distance, the Mahalanobis distance takes into account the shape and direction of the data. In a multidimensional space with different variances and correlations, the Mahalanobis distance can more accurately reflect the actual distance between points. Based on the value of the Mahalanobis distance, the fusion analysis result can be scored. For example, a threshold can be set, and the smaller the distance, the higher the score, indicating that the consistency of the fusion analysis result with the historical data is better.
[0052] According to another aspect of the present invention, Figure 4It shows a variety of fusion analysis models driven by thought chains. The results of the hazardous chemical analysis subtasks require fusion analysis driven by thought chains. The optional fusion analysis modes include: linear fusion, nonlinear fusion, mesh fusion, and tree fusion. The fusion analysis model driven by thought chains can split large analysis tasks into sub-analysis tasks with logical levels according to typical analysis patterns, and then process the sub-analysis tasks independently and in parallel, and then fuse the analysis results into a logically rigorous comprehensive analysis report.
[0053] Among them, linear fusion means that there is a parallel relationship between sub-analysis tasks, and the analysis tasks are processed according to the logic of the parallel relationship, and the analysis results are integrated into a logically rigorous comprehensive analysis report. Non-linear fusion means that there is a progressive relationship between sub-analysis tasks, and the analysis tasks are processed according to the logic of the non-linear relationship, and the analysis results are integrated into a logically rigorous comprehensive analysis report. Network fusion means that there is a mesh relationship between sub-analysis tasks, and the mesh relationship needs to be split into a multi-branch tree relationship. The analysis tasks are processed according to the logic of the multi-branch tree relationship, and the analysis results are integrated into a logically rigorous comprehensive analysis report. Tree fusion refers to the analysis tasks of the tree relationship with multiple subjects and multiple branches, and the analysis tasks are processed according to the logic of the tree relationship, and the analysis results are integrated into a logically rigorous comprehensive analysis report.
[0054] 3. Calculation of hazardous chemicals pretreatment scheme driven by thinking chain, see Figure 5 , the specific steps are as follows:
[0055] 1) Obtain the results of the fusion analysis of hazardous chemicals driven by the above thought chain. 2) Obtain the physical property information of hazardous chemicals, which may include color, odor, density, boiling point, melting point, flash point, etc. 3) Obtain the chemical property information of hazardous chemicals, which may include acidity, alkalinity, corrosiveness, flammability, explosiveness, biological hazard, etc. 4) Calculate the degree of explosion hazard of hazardous chemicals based on the physical and chemical properties of hazardous chemicals. 5) Calculate the protective measures and rescue measures based on the calculated degree of explosion hazard of hazardous chemicals.
[0056] The present invention uses the thinking chain driven calculation of hazardous chemical pretreatment scheme to complete the hazardous chemical pretreatment business and generate a pretreatment solution. Thinking chain disassembles the hazardous chemical pretreatment business into the above-mentioned several hazardous chemical pretreatment subtasks (i.e. the above-mentioned 5 steps). The hazardous chemical pretreatment business can be completed through a simple chain mode. Among them, step 2) and step 3) obtain physical properties and chemical properties from chemical databases, literature, supplier information, etc. Note that this part of the data should not use LLMs data as much as possible, and the level of specialization of Internet data is not high. Step 4) Calculate the hazard index based on physical and chemical properties, the typical formula: hazard index (DI) = a·density+b·boiling point+c·flash point+d·acidity+e·flammability+f·explosiveness. Among them, a, b, c, d, e, f are weight coefficients, which are adjusted according to specific circumstances. Step 5) Use weighting method or other statistical methods to combine all characteristic data into a comprehensive score. According to the calculated degree of explosion hazard, formulate protective measures and rescue measures.
[0057]
[0058] Among them, w i refers to the weight of the ith physical or chemical property, P i refers to the probability of the i-th physical or chemical property.
[0059] Step 5) Protective measures include: Determine the required level of protection based on the EHD score. For example, a high EHD value corresponds to more stringent protective measures, such as explosion-proof clothing, respirators, etc. Develop specific protective measures, including the selection and use of personal protective equipment (PPE). Rescue measures include: Rescue measures: Develop an emergency rescue plan based on the EHD score. For example, a high EHD value corresponds to a more detailed and rapid emergency response plan. Develop on-site first aid measures, including evacuation routes, preparation of first aid supplies, etc.
[0060] According to another aspect of the present invention, Figure 6301 is a hazardous chemical analysis and pretreatment system / server driven by the thinking chain. The system / server includes a central processing unit 301, a memory 302, a memory 303, a neural network processor 304, other peripheral devices, and a computer program stored on the memory 302 and can be run on the central processing unit 301 and the neural network processor 304. When the processor 301 executes the computer program, each step of the hazardous chemical analysis and pretreatment driven by the thinking chain as described above is implemented. The central processing unit 301 refers to a typical high-performance CPU chip of Intel or AMD. The neural network processor 304 refers to an IP, a submodule, an independent NPU chip, and a GPU chip containing a neural network unit (NPU). The neural network processor hardware is suitable for accelerating the computing load of deep learning network types, that is, it can hardware accelerate deep learning models and matrix convolution operations.
[0061] According to another aspect of the present invention, a computer readable medium is provided, wherein a computer program is stored on the computer readable medium, and the computer program is executed by a processor to implement the thought chain driven hazardous chemical analysis and pre-processing calculation method as described above.
[0062] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0063] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0064] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0065] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0066] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.
Claims
1. A thought chain driven method for hazardous chemicals analysis and pretreatment, It is characterized in that Analyze hazardous chemicals from multiple perspectives driven by the thinking chain and analyze hazardous chemicals by integrating multiple analysis modes driven by the thinking chain, complete the hazardous chemical analysis business, and generate hazardous chemical analysis reports; Use the thinking chain driven calculation of hazardous chemical pretreatment solutions to complete hazardous chemical pretreatment business and generate pretreatment solutions.
2. A method for analyzing and preprocessing hazardous chemicals driven by a thought chain according to claim 1, It is characterized in that The thought chain-driven multi-perspective analysis of hazardous chemicals uses large language models (LLMs) as analysis tools, specifically using a variety of large language models (LLMs) to perform multi-perspective analysis of hazardous chemicals.
3. A method for analyzing and preprocessing hazardous chemicals driven by a thought chain according to claim 1 or 2, It is characterized in that The thinking chain driven multi-perspective analysis of hazardous chemicals is specifically implemented in the following steps: 1) the user starts the thinking chain driven hazardous chemical analysis task; 2) the thinking chain drives the acquisition of a list of hazardous chemical analysis models, and the hazardous chemical analysis model adopts a general analysis model based on a large language model (LLMs) that is optimized and retrained with a database in the hazardous chemical field; 3) the thinking chain drives the distribution of the explosion accident investigation report and analysis prompt words to each hazardous chemical analysis model; 4) each hazardous chemical analysis model outputs the hazardous chemical analysis results, including information on physical and chemical properties and preventive measures.
4. A thought chain driven hazardous chemical analysis and pretreatment method according to claim 1, It is characterized in that The thought chain drives the use of multiple analysis modes to integrate and analyze hazardous chemicals, uses multiple fusion analysis modes to integrate and analyze hazardous chemicals, and then regresses and evaluates the fusion analysis results, and takes the optimal analysis result as the result of the fusion analysis.
5. A thought chain driven hazardous chemical analysis and pretreatment method according to claim 1 or 4, It is characterized in that The thought chain-driven fusion analysis of hazardous chemicals using multiple analysis modes is specifically implemented in the following steps: 1) obtaining a list of analysis results of multi-perspective analysis of hazardous chemicals; 2) obtaining fusion analysis results of the fusion analysis mode driven by the thought chain; 3) scoring the fusion analysis results by comparing with historical explosion accident investigation reports; 4) using the fusion analysis result with the highest score as the final hazardous chemical analysis result, including information on physical and chemical properties and preventive measures.
6. A thought chain driven hazardous chemical analysis and pretreatment method according to claim 5, It is characterized in that In step 2), the fusion analysis results of the fusion analysis mode driven by the thinking chain are obtained, specifically: the fusion analysis mode driven by the thinking chain is obtained, including linear fusion, nonlinear fusion, mesh fusion, and tree fusion. From the analysis results of hazardous chemicals, hazardous chemical data and hazardous chemical explosion data are extracted, and analysis prompt words of the large language model LLMs are generated to obtain the analysis results, and the linear fusion, nonlinear fusion, mesh fusion, and tree fusion are used to fuse the analysis results.
7. A thought chain driven hazardous chemical analysis and pretreatment method according to claim 5, It is characterized in that In step 3), the fusion analysis results are compared with the historical explosion accident investigation reports, and then the Mahalanobis distance is used to calculate and score the results. The Mahalanobis distance formula is as follows: Where S is the data covariance matrix, S -1 is the inverse matrix of the covariance matrix, y is the mean vector, and x is the fusion analysis result.
8. A thought chain driven hazardous chemical analysis and pretreatment method according to claim 1, It is characterized in that The specific implementation steps of the thought chain driven calculation hazardous chemical pretreatment scheme are as follows: 1) obtaining the fusion analysis results of hazardous chemicals by using multiple analysis modes driven by the thought chain; 2) obtaining the physical property information of hazardous chemicals; 3) obtaining the chemical property information of hazardous chemicals; 4) calculating the explosion hazard degree of hazardous chemicals according to the physical and chemical properties of hazardous chemicals; 5) Calculate the protective and rescue measures based on the degree of hazard posed by the explosion of hazardous chemicals.
9. A thought chain driven hazardous chemical analysis and pretreatment method according to claim 8, It is characterized in that In step 4), the degree of explosion hazard of hazardous chemicals is calculated based on the physical and chemical properties of the hazardous chemicals, specifically: First, the hazard index is calculated based on the physical and chemical properties of hazardous chemicals: Danger index (DI) = a·density+b·boiling point+c·flash point+d·acidity+e·flammability+f·explosiveness, where a, b, c, d, e, and f are weight coefficients; Secondly, calculate the degree of hazard of hazardous chemical explosion: Among them, w i refers to the weight of the ith physical or chemical property, P i refers to the probability of the i-th physical or chemical property.
10. A thought chain driven hazardous chemicals analysis and pretreatment system, It is characterized in that Executing the method according to any one of claims 1 to 9, comprising: The hazardous chemicals analysis business module uses the multi-perspective analysis of hazardous chemicals driven by the thinking chain and the integration of multiple analysis modes driven by the thinking chain to analyze hazardous chemicals, complete the hazardous chemicals analysis business, and generate hazardous chemicals analysis reports; The hazardous chemical pretreatment business module uses the thinking chain-driven calculation of hazardous chemical pretreatment solutions to complete hazardous chemical pretreatment business and generate pretreatment solutions.
