Multi-agent collaborative management method and system for whole-process engineering consultation

Abstract

The invention provides a multi-agent collaborative management method and system for whole-process engineering consultation. The method comprises the following steps: S1, obtaining a whole-process engineering consultation task; s2, task decomposition is conducted on the obtained whole-process engineering consultation task based on an LLM large language model, task decomposition characteristics of the whole-process engineering consultation task are obtained, and the task decomposition characteristics comprise one or more subtasks; s3, task classification feature extraction is carried out on each subtask of the task decomposition features, and task classification labels corresponding to the subtasks are obtained; s4, according to the task classification label corresponding to each sub-task, matching a corresponding agent from the candidate agents to process the sub-task, and obtaining a sub-task solution output by the agent; and S5, performing optimization processing according to the obtained sub-task solution to obtain a whole-process engineering consultation task solution. According to the invention, the effect of providing the whole-process engineering consultation task solution can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of full-process engineering consulting management for construction projects, and in particular to a multi-agent collaborative management method and system for full-process engineering consulting. Background Art

[0002] With the continuous deepening of the digital transformation of the construction industry, the application of large language model technology has greatly promoted the development of artificial intelligence technology and promoted the deep integration of artificial intelligence technology and engineering management application systems. At present, the engineering management application system built on the large language model uses the trained large language model as the core component of the artificial intelligence system, combined with the intelligent agent to realize the reasoning of natural language, thereby outputting the reasoning conclusion. However, in the implementation of the full-process engineering consulting business, the task division and collaborative management of multiple agents of multi-target complex tasks cannot be realized, so that the standardized processing flow and accurate reasoning results required by the full-process engineering consulting business cannot be obtained, which has a significant impact on the accuracy of the output reasoning conclusions. Therefore, the present invention provides a large-model multi-agent collaborative management method and system technology solution for full-process engineering consulting. Summary of the Invention

[0003] In response to the above problems, the present invention aims to provide a multi-agent collaborative management method and system for full-process engineering consulting.

[0004] The purpose of the present invention is achieved by adopting the following technical solutions: In a first aspect, the present invention proposes a multi-agent collaborative management method for full-process engineering consulting, comprising: S1 obtains a full-process engineering consulting task, wherein the full-process engineering consulting task is a natural language input by a user; S2 decomposes the acquired full-process engineering consulting task based on the LLM large language model to obtain a task decomposition feature of the full-process engineering consulting task, wherein the task decomposition feature includes one or more subtasks; S3 extracts task classification features from each subtask of the task decomposition feature and obtains the task classification label corresponding to the subtask; S4 matches the corresponding agent from the candidate agents according to the task classification labels corresponding to each subtask to process the subtask and obtain the subtask solution output by the agent; S5 performs optimization processing based on the obtained subtask solutions to obtain a full-process engineering consulting task solution.

[0005] Preferably, the method further comprises: S6 conducts prediction verification based on the obtained full-process engineering consulting task solution and the digital twin model of the engineering project to obtain the solution verification results, and further trains and optimizes the relevant intelligent agents based on the solution verification results.

[0006] Preferably, step S2 includes: Based on the LLM large language model, the acquired full-process engineering consulting tasks are understood and divided, and the conditional characteristics and target tasks of the full-process engineering consulting tasks are extracted; and further enhanced retrieval is performed based on the acquired conditional characteristics and target tasks to obtain the corresponding potential knowledge needs, and based on the acquired potential knowledge needs, further understanding is performed based on the LLM large language model to obtain potential tasks; the target tasks and potential tasks are respectively used as subtasks to constitute the task decomposition characteristics of the full-process engineering consulting tasks; the conditional characteristics include engineering specifications and constraints.

[0007] Preferably, step S3 includes: Extract text feature vectors based on the description content of the subtask and the corresponding engineering project feature information , matching based on text feature vector Task classification labels corresponding to subtasks ,in , Represents the task classification label, Represents the set of task classification labels. Represents text feature vector The probability of belonging to task classification label j.

[0008] Preferably, step S4 includes: According to the task classification label corresponding to the subtask, the corresponding expert agent is matched to process the subtask and obtain the subtask solution; the expert agents include investment decision consulting agent, engineering approval and construction service agent, engineering survey management agent, engineering design management agent, engineering bidding and procurement management agent, construction project management agent, engineering supervision agent, engineering cost management agent, engineering operation and maintenance management agent, other consulting service agents, quality management agent, safety management agent, progress management agent, cost management agent, other target management agents, etc.

[0009] Preferably, step S5 includes: Extract key terms from subtask solutions obtained by a single expert agent; Based on the corresponding engineering specifications, the compliance analysis of the key clause information in the task solution is performed, and the variable content in the key clause is compared with the corresponding standard value range in the engineering specifications. When the key clause information exceeds the standard range, the content corresponding to the key clause information in the subtask solution is adaptively corrected, and the corresponding variables are adjusted to the value closest to the current value and within the standard range, so that the subtask solution complies with the engineering specifications.

[0010] Preferably, step S5 includes: Subtask solutions obtained for multiple expert agents: Extract key terms from subtask solutions for the target task ; Extract key terms from subtask solutions for each potential task ;in is a variable, , represents the corresponding key terms information in the subtask solution of the nth potential task, represents the total number of potential tasks; Based on the target task, calculate the impact change parameters of each subtask on the target task :

[0011] Where, Indicates the parameter affecting the target task, k is a variable, , corresponding to different potential tasks, is the total number of potential tasks, represents the expert agent corresponding to subtask k; The task classification label representing the target task; Represents an expert agent Task classification The processing capacity of the agent is analyzed by statistically analyzing the agent's processing power in task classification. The success rate statistics are obtained; Represents the key terms information corresponding to subtask k Effect on target tasks The partial derivative of is used to characterize the influence of key clause information on the target task effect; Will affect the change parameters As the target value, adjust the key terms of the current subtask solution until the change parameter is affected Less than the set standard threshold , where Z represents the set standard threshold, or the maximum number of adjustments; obtain the optimized subtask solution corresponding to each potential task; The subtask solutions of the target task are integrated with the optimized subtask solutions of the potential tasks to obtain the full-process engineering consulting task solution.

[0012] Preferably, step S6 includes: According to the obtained full-process engineering consulting task solution, input it into the established BIM-based full-process engineering consulting management model to implement performance analysis and risk analysis, and obtain the corresponding performance analysis results and risk analysis results. Based on the obtained performance analysis results and risk analysis results, record the adaptability and adoption degree of the task solution to the target task.

[0013] In the second aspect, the present invention proposes a multi-agent collaborative management system for full-process engineering consulting, including a receiving module, a processing module and an output module, wherein: The receiving module is used to receive the whole process engineering consulting task transmitted by the user; The processing module is used to execute the multi-agent collaborative management method for full-process engineering consulting as shown in any embodiment of the first aspect to obtain a full-process engineering consulting task solution; The output module is used to output the whole process engineering consulting task solution.

[0014] The beneficial effects of the present invention are as follows: a multi-agent collaborative management method and system for full-process engineering consulting is proposed, wherein the full-process engineering consulting task is first obtained, the full-process engineering consulting task is understood and decomposed based on the LLM large language model, and based on the subtasks obtained by decomposition, the corresponding task classification labels are first extracted based on the subtasks to mark the type of the subtask, and according to the task classification labels, the corresponding agents are further matched to process the subtasks, thereby obtaining targeted subtask solutions; further optimization processing is performed based on the obtained subtask solutions, and finally a solution for the meritorious consulting task is generated. By combining task decomposition, classification label extraction and agent matching, the corresponding task type can be targeted based on a dedicated agent, which helps to improve the professionalism of the full-process engineering consulting task processing. The subtask solutions obtained by multiple agents are then targeted and jointly optimized, thereby improving the compatibility of the quality inspections of different subtask solutions, thereby improving the adaptability and comprehensive performance of the final full-process engineering consulting task solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention is further described with reference to the accompanying drawings. However, the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative effort.

[0016] Figure 1 This is a method flow chart of a multi-agent collaborative management method for full-process engineering consulting according to an embodiment of the present invention; Figure 2 This is a framework diagram of a multi-agent collaborative management system for full-process engineering consulting as shown in an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The present invention is further described in conjunction with the following application scenarios.

[0018] See also Figure 1 , which presents a multi-agent collaborative management method for full-process engineering consulting, including: S1 obtains a full-process engineering consulting task, wherein the full-process engineering consulting task is a natural language input by a user; S2 decomposes the acquired full-process engineering consulting task based on the LLM large language model to obtain a task decomposition feature of the full-process engineering consulting task, wherein the task decomposition feature includes one or more subtasks; S3 extracts task classification features from each subtask of the task decomposition feature and obtains the task classification label corresponding to the subtask; S4 matches the corresponding agent from the candidate agents according to the task classification labels corresponding to each subtask to process the subtask and obtain the subtask solution output by the agent; S5 performs optimization processing based on the obtained subtask solutions to obtain a full-process engineering consulting task solution.

[0019] The above-mentioned embodiment of the present invention proposes a multi-agent collaborative management method for full-process engineering consulting, wherein the full-process engineering consulting task is first obtained, and the full-process engineering consulting task is understood and decomposed based on the LLM large language model. Based on the subtasks obtained by decomposition, the corresponding task classification labels are first extracted based on the subtasks to mark the type of the subtask. According to the task classification labels, the corresponding agents are further matched to process the subtasks, thereby obtaining targeted subtask solutions; the obtained subtask solutions are further optimized and finally generated to solve the meritorious consulting tasks. By combining task decomposition, classification label extraction and agent matching, the corresponding task types can be processed based on dedicated agents, which helps to improve the professionalism of the full-process engineering consulting task processing. The subtask solutions obtained by multiple agents are then targeted and jointly optimized, thereby improving the compatibility of the quality inspection of different subtask solutions, thereby improving the adaptability and comprehensive performance of the final full-process engineering consulting task solution.

[0020] The multi-agent collaborative management method for full-process engineering consulting shown in the above implementation method can be executed and implemented based on servers, smart terminals, etc., and ultimately feedback outputs a full-process engineering consulting task solution.

[0021] Preferably, the method further comprises: S6 conducts prediction verification based on the obtained full-process engineering consulting task solution and the digital twin model of the engineering project to obtain the solution verification results, and further trains and optimizes the relevant intelligent agents based on the solution verification results.

[0022] Based on the full-process engineering consulting task solution output by the system, performance testing and risk analysis are further carried out based on digital twin technology and predictive analysis technology to further verify the feasibility and execution effect of the full-process engineering consulting task solution, and then use the prediction results as a reference for further training and setting of intelligent agents, so as to further improve the performance and effect of the system and intelligent agents.

[0023] Among them, the intelligent agent shown in the present invention generally refers to an expert intelligent agent set up for different knowledge classifications in the full-process engineering consulting scenario, wherein the expert intelligent agent can be trained based on the knowledge and historical data of one or more fields, thereby improving the ability and effect of the expert intelligent agent in handling consulting tasks in specialized fields.

[0024] Preferably, step S2 includes: Based on the LLM large language model, the acquired full-process engineering consulting tasks are understood and divided, and the conditional characteristics and target tasks of the full-process engineering consulting tasks are extracted; and further enhanced retrieval is performed based on the acquired conditional characteristics and target tasks to obtain the corresponding potential knowledge needs, and based on the acquired potential knowledge needs, further understanding is performed based on the LLM large language model to obtain potential tasks; the target tasks and potential tasks are respectively used as subtasks to constitute the task decomposition characteristics of the full-process engineering consulting tasks; the conditional characteristics include engineering specifications and constraints.

[0025] Considering that in the full-process engineering consulting scenario of a construction project, a large number of engineering specifications and standards will be involved, it is usually necessary to simultaneously consider the specifications of different fields when analyzing and processing tasks in a single field. Therefore, in the above-mentioned implementation method, when understanding and decomposing the acquired full-process engineering consulting tasks, the engineering specification information corresponding to the full-process engineering consulting tasks is first extracted, and enhanced retrieval is performed based on the engineering specification information and task conditions, task objectives, etc., so as to determine the potential conditions and potential tasks that need to be met simultaneously to achieve the target task under the corresponding engineering specifications, and to expand and decompose the target task as the corresponding subtask, so that when generating solutions in the future, in addition to considering the target task itself, the potential tasks can be further used as consideration criteria, thereby improving the overall quality of the subsequent solutions.

[0026] In one scenario, the current full-process engineering consulting task is understood based on the LLM large language model, wherein engineering project feature information is extracted according to the engineering project corresponding to the full-process engineering consulting task (if there is no corresponding engineering project, for general full-process engineering consulting tasks, the preset general engineering project feature information is used), and the corresponding engineering specifications (such as GB-numbered standards) are matched according to the engineering project feature information. Further enhanced retrieval is performed based on the matched engineering specifications, constraints and target tasks to obtain the potential knowledge requirements that need to be met simultaneously to complete the target task under the constraints. Further understanding and task extraction are performed based on the potential knowledge requirements on the standard knowledge to obtain potential task information.

[0027] In one scenario, for example, for the full-process engineering consulting task of "providing a plan to complete the construction of an office building within the budget", the constraint obtained after understanding according to the LLM large language model is "within the budget", and the target task is "complete the construction of the office building within the budget". Further, according to the corresponding engineering project, appropriate engineering specification information is matched from the knowledge base, and enhanced retrieval is performed in the engineering specification information based on the constraint conditions and target tasks to obtain potential knowledge needs, such as "according to Article z of GB505xx-yyyy, to complete the construction of the office building within the budget, the change visa rate must be controlled to <5%"; based on the potential knowledge needs, further understanding is performed based on the LLM large language model (understanding is performed based on the relevant engineering specifications for controlling the change visa rate) and task division, and the potential tasks for controlling the change visa rate are "reasonable schedule plan formulation to reduce the change visa rate" and "reliable quality control strategy to reduce the change visa rate", which are further spliced ​​with the target task to obtain the task decomposition characteristics of the three corresponding subtasks.

[0028] The extraction of potential tasks achieved through the above-mentioned method can complement the large language model's expansion of its relevant understanding of "budget" to the collaborative understanding of "quality" and "planning", thereby improving the accuracy and rationality of the understanding and confirmation of the entire process of engineering consulting tasks.

[0029] Preferably, step S3 includes: Extract text feature vectors based on the description content of the subtask and the corresponding engineering project feature information , matching based on text feature vector Task classification labels corresponding to subtasks ,in , Represents the task classification label, Represents the set of task classification labels. Represents text feature vector The probability of belonging to task classification label j.

[0030] Among them, the text feature vector can be used to extract the description content of the subtask and the corresponding engineering project feature information through text neural network models such as Word2Vec, or through deep learning models such as BERT‌ to extract text feature vectors, or through other existing text feature models to complete the text feature extraction task. The present invention does not make specific limitations here.

[0031] Among them, the task classification label includes task classification information related to the whole process engineering management, including investment decision consulting, engineering approval and construction services, engineering survey management, engineering design management, engineering bidding and procurement management, construction project management, engineering supervision, engineering cost management, engineering operation and maintenance management, other consulting services, quality management, safety management, progress management, cost management, other target management, etc.

[0032] Among them, the text feature vector The probability of belonging to the task classification label j can be processed based on a trained semantic feature classifier, where the semantic feature classifier can be implemented based on classification models such as SVM support vector machine, or based on the dedicated classification Transformer architecture in the LMM large language model to complete the task of text feature vector classification.

[0033] Based on the obtained subtask content, text feature vectors are extracted as a basis, and then the corresponding task classification labels are matched based on the text feature vectors. This lays the foundation for the subsequent assignment of corresponding agents to subtasks. Categorizing subtasks through task classification allows for the division of agents based on their specific domains within the entire process of engineering consulting management, thereby improving the generation of subtask solutions.

[0034] Preferably, step S4 includes: According to the task classification label corresponding to the subtask, the corresponding expert agent is matched to process the subtask and obtain the subtask solution; the expert agents include investment decision consulting agent, engineering approval and construction service agent, engineering survey management agent, engineering design management agent, engineering bidding and procurement management agent, construction project management agent, engineering supervision agent, engineering cost management agent, engineering operation and maintenance management agent, other consulting service agents, quality management agent, safety management agent, progress management agent, cost management agent, other target management agents, etc.

[0035] Based on the specific knowledge areas involved in the full-process engineering consulting scenario, corresponding expert agents can be set up for specialized fields to solve corresponding tasks and provide more targeted solutions.

[0036] In another scenario, the setting method of expert agents can also be classified according to fields, and set as cost agents, safety agents, progress agents, quality agents, etc. to meet the needs of specialized task solutions.

[0037] Preferably, step S5 includes: Extract key terms from subtask solutions obtained by a single expert agent; Based on the corresponding engineering specifications, the compliance analysis of the key clause information in the task solution is performed, and the variable content in the key clause is compared with the corresponding standard value range in the engineering specifications. When the key clause information exceeds the standard range, the content corresponding to the key clause information in the subtask solution is adaptively corrected, and the corresponding variables are adjusted to the value closest to the current value and within the standard range, so that the subtask solution complies with the engineering specifications.

[0038] Among them, the key terms in the subtask solution can be extracted through keyword retrieval matching, entity recognition combined with knowledge base, etc.

[0039] In one scenario, for example, in the sub-task solutions for project quality acceptance, the key clauses extracted include "Clause 1: Concrete strength test ≥ 3 times / layer; Clause 2: Steel bar spacing error ≤ 5mm"; for example, for budget-related solutions, the key clauses extracted include "Prepare alternative plans in case market price fluctuations exceed 5%" and "Design change rate is controlled within 2%."

[0040] In one scenario, the variable content in the key term information may be a value (such as a specific value) or a trend (such as increase, decrease, maintenance, etc.).

[0041] After obtaining a task solution from a single expert agent, key terms are extracted from the solution. This allows for summarizing and generalizing subtask solutions based on these key terms, thereby improving the accuracy of information extraction. Based on the extracted key terms, these terms are then compared against the standards in the corresponding engineering specifications to ensure that the key terms meet the relevant engineering specifications, thereby improving the feasibility and compliance of the task solution.

[0042] Preferably, step S5 includes: Subtask solutions obtained for multiple expert agents: Extract key terms from subtask solutions for the target task ; Extract key terms from subtask solutions for each potential task ;in is a variable, , represents the corresponding key terms information in the subtask solution of the nth potential task, represents the total number of potential tasks; Based on the target task, calculate the impact change parameters of each subtask on the target task :

[0043] Where, Indicates the parameter affecting the target task, k is a variable, , corresponding to different potential tasks, is the total number of potential tasks, represents the expert agent corresponding to subtask k; The task classification label representing the target task; Represents an expert agent Task classification The processing capacity of the agent is analyzed by statistically analyzing the agent's processing power in task classification. The success rate statistics are obtained; Represents the key terms information corresponding to subtask k Effect on target tasks The partial derivative of is used to characterize the influence of key clause information on the target task effect; Will affect the change parameters As the target value, adjust the key terms of the current subtask solution until the change parameter is affected Less than the set standard threshold , where Z represents the set standard threshold, or the maximum number of adjustments; obtain the optimized subtask solution corresponding to each potential task; The subtask solutions of the target task are integrated with the optimized subtask solutions of the potential tasks to obtain the full-process engineering consulting task solution.

[0044] Among them, when statistically analyzing the success rate of an intelligent agent in processing a certain type of task classification, the final acceptance rate (probability, such as the final accepted statistical number / total processing number) of the task solution given by the intelligent agent when processing a collaborative task involving a certain type of task classification can be obtained.

[0045] Among them, the ways to adjust key terms include changing the plan, deleting terms, adding conditions, etc.

[0046] Among them, the methods of fusing the subtask solutions of the target task with the optimized subtask solutions of the potential tasks include splicing, superposition, weighted superposition, fusion based on feature similarity, etc., which are not specifically limited in the present invention.

[0047] As in the aforementioned solution, by extracting potential tasks and obtaining corresponding subtask solutions, the breadth and comprehensiveness of the solution can be improved. However, multiple subtask solutions exist independently and need to be integrated.

[0048] Since the task solutions corresponding to different subtasks are obtained by different expert agents, and different expert agents have different basic knowledge and consideration logic when considering tasks in different fields, the task solutions obtained for potential tasks may cause resistance and conflict with the target task (for example, contradictory solutions), resulting in the final solution to the full-process engineering consulting task being unsatisfactory.

[0049] During the actual execution of the plan, the smaller the impact of other potential tasks on the target task results, the lower the contradiction caused by the potential plan to the target task. The primary task of the potential task solution is to improve the stability of the target task (to avoid inconsiderate situations), but the effect of the target task can be improved by the solution of the target task itself (a solution specifically for improving the target effect). Based on the above ideas, the integration of sub-task solutions can improve the stability of the solution while ensuring the effect of the target task.

[0050] Therefore, the above-mentioned embodiment of the present invention further proposes a collaborative optimization solution based on subtask solutions obtained by different expert agents, thereby improving the overall performance of the full-process engineering consulting task solution. First, subtasks typically include target tasks and potential tasks. Therefore, the target task is set as the highest priority for final consideration, and the potential task solutions must ultimately serve the target task. Based on this approach, key terms information of each potential task solution is first extracted. Based on the proposed impact change parameter, the degree of impact of the potential task solution on the current target task is comprehensively reflected. This is used as the target parameter to reduce the impact change parameter, prevent the solutions provided by non-expert agents from interfering with the target task, and help improve the stability of the solution. Finally, the obtained target task solution and the potential task solutions are integrated to obtain a more comprehensive solution (while ensuring the effectiveness of the target task, the stability and adaptability of the provided solution are improved), thereby improving the quality and effectiveness of the solution provided.

[0051] In one scenario, when the target task is a solution for "budget optimization", the solution provides relevant technical solutions for budget optimization. As for the technical solutions provided for the potential tasks of "plan" and "quality", if the solutions provided will cause huge fluctuations in the "budget" in exchange for the corresponding "plan improvement" or "quality improvement", then the corresponding solutions should be weakened or abandoned to ensure that "budget is given top priority" and supplement solutions related to "plan" and "quality" to improve the stability of the overall solution (reduce the volatility of the impact of plan and quality on the budget).

[0052] Preferably, step S6 includes: According to the obtained full-process engineering consulting task solution, input it into the established BIM-based full-process engineering consulting management model to implement performance analysis and risk analysis, and obtain the corresponding performance analysis results and risk analysis results. Based on the obtained performance analysis results and risk analysis results, record the adaptability and adoption degree of the task solution to the target task.

[0053] By analyzing and recording the degree of adaptability and adoption, it is helpful to count the successful and failed cases of the intelligent agent for task solutions, thereby counting the relevant success rate characteristic data, which is convenient for subsequent further adjustment and optimization of the intelligent agent's performance.

[0054] See also Figure 2 , which shows a multi-agent collaborative management system for full-process engineering consulting, including a receiving module, a processing module and an output module, wherein, The receiving module is used to receive the whole process engineering consulting task transmitted by the user; The processing module is used to execute the above Figure 1 The multi-agent collaborative management method for full-process engineering consulting is shown, and a solution to the full-process engineering consulting task is obtained; The output module is used to output the whole process engineering consulting task solution.

[0055] It should be noted that the processing module is also used to execute Figure 1 The specific implementation methods corresponding to each step in the present invention will not be repeated here.

[0056] It should be noted that the functional units / modules in the various embodiments of the present invention may be integrated into a single processing unit / module, each unit / module may exist physically separately, or two or more units / modules may be integrated into a single unit / module. The aforementioned integrated units / modules may be implemented in the form of hardware or software functional units / modules.

[0057] From the above description of the embodiments, those skilled in the art will clearly understand that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code, or any appropriate combination thereof. For hardware implementation, the processor may be implemented in one or more of the following: an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field-programmable gate array (FPGA), a processor, a controller, a microcontroller, a microprocessor, or other electronic units designed to implement the functionality described herein, or any combination thereof. For software implementation, some or all of the processes of the embodiments may be performed by a computer program instructing the relevant hardware. During implementation, the program may be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media includes any medium that facilitates the transfer of a computer program from one location to another. Storage media can be any available medium that can be accessed by a computer. Computer-readable media may include, but are not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer.

[0058] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the protection scope of the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should analyze that the technical solutions of the present application can be modified or equivalently replaced without departing from the essence and scope of the technical solutions of the present application.

Claims

1. A multi-agent collaborative management method for full-process engineering consulting, characterized by: include: S1 obtains a full-process engineering consulting task, wherein the full-process engineering consulting task is a natural language input by a user; S2 decomposes the acquired full-process engineering consulting task based on the LLM large language model to obtain a task decomposition feature of the full-process engineering consulting task, wherein the task decomposition feature includes one or more subtasks; S3 extracts task classification features from each subtask of the task decomposition feature and obtains the task classification label corresponding to the subtask; S4 matches the corresponding agent from the candidate agents according to the task classification labels corresponding to each subtask to process the subtask and obtain the subtask solution output by the agent; S5 performs optimization processing based on the obtained subtask solutions to obtain a full-process engineering consulting task solution.

2. The multi-agent collaborative management method for full-process engineering consulting according to claim 1 is characterized in that: The method further includes: S6 conducts prediction verification based on the obtained full-process engineering consulting task solution and the digital twin model of the engineering project to obtain the solution verification results, and further trains and optimizes the relevant intelligent agents based on the solution verification results.

3. The multi-agent collaborative management method for full-process engineering consulting according to claim 1 is characterized in that: Step S2 includes: Based on the LLM large language model, the acquired full-process engineering consulting tasks are understood and divided, and the conditional characteristics and target tasks of the full-process engineering consulting tasks are extracted; and further enhanced retrieval is performed based on the acquired conditional characteristics and target tasks to obtain the corresponding potential knowledge needs, and based on the acquired potential knowledge needs, further understanding is performed based on the LLM large language model to obtain potential tasks; the target tasks and potential tasks are respectively used as subtasks to constitute the task decomposition characteristics of the full-process engineering consulting tasks; the conditional characteristics include engineering specifications and constraints.

4. The multi-agent collaborative management method for full-process engineering consulting according to claim 3 is characterized in that: Step S3 includes: Extract text feature vectors based on the description content of the subtask and the corresponding engineering project feature information , matching based on text feature vector Task classification labels corresponding to subtasks ,in , Represents the task classification label, Represents the set of task classification labels. Represents text feature vector The probability of belonging to task classification label j.

5. The multi-agent collaborative management method for full-process engineering consulting according to claim 4 is characterized in that: Step S4 includes: According to the task classification label corresponding to the subtask, the corresponding expert agent is matched to process the subtask and obtain the subtask solution; the expert agents include investment decision consulting agent, engineering approval and construction service agent, engineering survey management agent, engineering design management agent, engineering bidding and procurement management agent, construction project management agent, engineering supervision agent, engineering cost management agent, engineering operation and maintenance management agent, other consulting service agents, quality management agent, safety management agent, progress management agent, cost management agent, and other target management agents.

6. The multi-agent collaborative management method for full-process engineering consulting according to claim 5 is characterized in that: Step S5 includes: Extract key terms from subtask solutions obtained by a single expert agent; Based on the corresponding engineering specifications, the compliance analysis of the key clause information in the task solution is performed, and the variable content in the key clause is compared with the corresponding standard value range in the engineering specifications. When the key clause information exceeds the standard range, the content corresponding to the key clause information in the subtask solution is adaptively corrected, and the corresponding variables are adjusted to the value closest to the current value and within the standard range, so that the subtask solution complies with the engineering specifications.

7. The multi-agent collaborative management method for full-process engineering consulting according to claim 5 is characterized in that: Step S5 includes: Subtask solutions obtained for multiple expert agents: Extract key terms from subtask solutions for the target task ; Extract key terms from subtask solutions for each potential task ;in is a variable, , represents the corresponding key terms information in the subtask solution of the nth potential task, represents the total number of potential tasks; Based on the target task, calculate the impact change parameters of each subtask on the target task : Where, Indicates the parameter affecting the target task, k is a variable, , corresponding to different potential tasks, is the total number of potential tasks, represents the expert agent corresponding to subtask k; The task classification label representing the target task; Represents an expert agent Task classification The processing capacity of the agent is analyzed by statistically analyzing the agent's processing power in task classification. The success rate statistics are obtained; Represents the key terms information corresponding to subtask k Effect on target tasks The partial derivative of is used to characterize the influence of key clause information on the target task effect; Will affect the change parameters As the target value, adjust the key terms of the current subtask solution until the change parameter is affected Less than the set standard threshold , where Z represents the set standard threshold, or the maximum number of adjustments; obtain the optimized subtask solution corresponding to each potential task; The subtask solutions of the target task are integrated with the optimized subtask solutions of the potential tasks to obtain the full-process engineering consulting task solution.

8. The multi-agent collaborative management method for full-process engineering consulting according to claim 5 is characterized in that: Step S6 includes: According to the obtained full-process engineering consulting task solution, input it into the established BIM-based full-process engineering consulting management model to implement performance analysis and risk analysis, and obtain the corresponding performance analysis results and risk analysis results. Based on the obtained performance analysis results and risk analysis results, record the adaptability and adoption degree of the task solution to the target task.

9. A multi-agent collaborative management system for full-process engineering consulting, characterized by: include: Receiving module, processing module and output module, wherein, The receiving module is used to receive the whole process engineering consulting task transmitted by the user; The processing module is used to execute the multi-agent collaborative management method for full-process engineering consulting as shown in any one of claims 1 to 8 above, to obtain a full-process engineering consulting task solution; The output module is used to output the whole process engineering consulting task solution.

Images