A planning method and computer readable storage medium

Abstract

The application provides a planning method and a computer readable storage medium. The planning method is used for a thermal power construction project, and the planning method comprises the following steps: obtaining data documents, wherein the data documents comprise a first data document of the thermal power construction project and a second data document of a construction contract; preprocessing the data documents to obtain analyzable text units; performing analysis and processing on the analyzable text units to obtain a construction knowledge graph and a contract constraint condition; simulating and deducing the thermal power construction project based on the construction knowledge graph and the contract constraint condition to obtain a multi-level progress plan of construction; obtaining abnormal information in the multi-level progress plan, analyzing and processing the abnormal information, and generating a processing opinion report; modifying the multi-level progress plan based on the processing opinion report to obtain a final construction plan; and wherein the analyzable text units comprise first sub-analyzable text units and second sub-analyzable text units.

Description

Technical Field

[0001] This invention relates to the field of planning, and more specifically, to a planning method and a computer-readable storage medium. Background Technology

[0002] Currently, there are still systemic defects in the processing of schedule planning for complex projects such as thermal power plants: First, the parsing of unstructured text relies on manual labor. Key information such as milestones, process logic, and special resource constraints in contracts and construction organization are buried deep in massive documents, which cannot be automatically extracted by general technologies, and are easily missed or misinterpreted by manual reading. Second, the generation of multi-level schedule plans lacks automation and standardization, relying on experience to build WBS and set logic. The hierarchical standards are inconsistent, the logical connections are chaotic, and it is difficult to form a unified benchmark. Third, the generation of plans, resource optimization, and deliverable output are isolated from each other. It is impossible to integrate hard constraints such as the exclusivity of special machinery and the supply cycle of materials into the scheduling. Moreover, the schedule plan, resource plan, and interface plan need to be pieced together manually, and it is impossible to output integrated deliverables. Fourth, there is a lack of linkage and verification mechanism with contracts and construction organization. After the plan is completed, it is impossible to automatically verify the compliance of the schedule and the feasibility of the process. When the contract or plan is changed, it needs to be adjusted manually, which is slow and prone to disconnection.

[0003] In summary, existing technologies suffer from gaps in each stage, including front-end parsing, mid-end constraint integration, and back-end integrated delivery, making it difficult to meet the high requirements of thermal power projects for accurate and intelligent schedule management. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art or related art.

[0005] Therefore, the first aspect of the present invention proposes a planning method.

[0006] A second aspect of the present invention provides a computer-readable storage medium.

[0007] In view of this, a first aspect of the present invention provides a planning method for thermal power plant construction projects, comprising: acquiring a data document, wherein the data document includes a first data document of the thermal power plant construction project and a second data document of the construction contract; preprocessing the data document to acquire parsable text units; parsing the parsable text units to acquire a construction knowledge graph and contractual constraints; simulating and extrapolating the thermal power plant construction project based on the construction knowledge graph and contractual constraints to acquire a multi-level construction schedule; acquiring abnormal information in the multi-level schedule, analyzing and processing the abnormal information, and generating a processing opinion report; modifying the multi-level schedule based on the processing opinion report to acquire a final construction plan; wherein the parsable text unit includes a first sub-parsable text unit and a second sub-parsable text unit.

[0008] In addition, the planning method in the above-mentioned technical solution provided by the present invention may also have the following additional technical features: In some technical solutions of the present invention, optionally, preprocessing the data document to obtain parsable text units includes: converting the data document into a format to obtain a first derived text; removing noise data from the first derived text to obtain a second derived text; dividing the second derived text into multiple text segments according to a preset title to obtain a third derived text; and annotating the content in the third derived text based on preset tags to obtain parsable text units.

[0009] In some technical solutions of the present invention, optionally, parsing processing is performed on parsable text units to obtain a construction knowledge graph and contractual constraints, including: training a text intelligent parsing model based on the annotation word set and semantic rules of thermal power construction projects to obtain a thermal power text intelligent parsing model; constructing a domain database based on domain knowledge of thermal power construction projects; identifying and extracting entities and logical relationships between different entities in the first sub-parsable text unit through the thermal power text intelligent parsing model; identifying and extracting contractual constraints in the second sub-parsable text unit through the thermal power text intelligent parsing model; instantiating entities, logical relationships, and contractual constraints into entity nodes and edges through the domain database; and constructing a construction knowledge graph based on entity nodes and edges.

[0010] In some technical solutions of the present invention, optionally, a text intelligent parsing model is trained based on the labeled word set and semantic rules of thermal power construction projects to obtain a thermal power text intelligent parsing model, including: obtaining a labeled word set, which includes vocabulary of thermal power construction projects, word weights, synonyms, and text range limitation rules; training the text intelligent parsing model based on the labeled word set and semantic rules, so that the trained text intelligent parsing model can identify entities in the first sub-parsable text unit, logical relationships between different entities, and contractual constraints in the second sub-parsable text unit, and using the trained text intelligent parsing model as the thermal power text intelligent parsing model.

[0011] In some technical solutions of the present invention, optionally, the domain database includes: an entity thesaurus for thermal power construction projects, a terminology database for construction procedures, and a database of schedule constraints and annotation examples for thermal power construction projects.

[0012] Optionally, in some technical solutions of the present invention, a multi-level construction schedule is obtained by simulating and extrapolating the thermal power plant construction project based on the construction knowledge graph and contractual constraints. This includes: generating a skeleton of the project-level work breakdown structure based on the construction knowledge graph; generating a time plan for the smallest task unit based on the process logic relationship of the construction knowledge graph and the time constraints in the contractual constraints; optimizing the time plan based on the multi-dimensional constraints of the construction knowledge graph to obtain an optimized time plan; and summarizing the optimized time plan step by step according to the hierarchical relationship of the skeleton to generate a multi-level schedule.

[0013] In some technical solutions of the present invention, optionally, the time plan is optimized based on the multi-dimensional constraints of the construction knowledge graph to obtain an optimized time plan, including: identifying conflicting items in the time plan that violate the multi-dimensional constraints; adjusting the time plan of the smallest task unit corresponding to the conflicting item according to a preset adjustment strategy until all multi-dimensional constraints are satisfied, and using the adjusted time plan as the optimized time plan.

[0014] In some technical solutions of the present invention, optionally, abnormal information in the multi-level schedule is obtained, the abnormal information is analyzed and processed, and a processing opinion report is generated, including: verifying the multi-level schedule based on contractual constraints and resource allocation, obtaining abnormal information; analyzing the abnormal information, and generating a processing opinion report.

[0015] In some technical solutions of the present invention, optionally, the multi-level schedule is verified based on contractual constraints and resource allocation to obtain abnormal information, including: verifying whether the completion time of each milestone node in the multi-level schedule meets the requirements according to the milestone nodes and the schedule requirements in the contractual constraints, and generating schedule abnormal information; judging whether there are resource conflicts or resource shortages in the multi-level schedule based on the number of equipment, personnel, and material supply cycles in the resource allocation, and obtaining the judgment result; generating resource abnormal information based on the judgment result; and merging the schedule abnormal information and resource abnormal information to obtain abnormal information.

[0016] Compared with the prior art, the planning method provided by the present invention has the following outstanding advantages: By constructing a domain database that integrates thermal power terminology, process logic, and multi-dimensional constraint models, and training a domain-specific intelligent text parsing model, computers can deeply "understand" the professional semantics in contracts and construction organization designs. This completely changes the core planning model that relies on manual interpretation and translation, enabling automated and high-precision extraction and instantiation of milestone events, complex process chains, and key constraint information related to "people, machines, materials, methods, and environment," thus ensuring the quality and efficiency of planning input data from the source.

[0017] Abandoning the traditional approach of relying on generic templates and manual scheduling, this system constructs a project-specific construction knowledge graph, unifying the modeling of multi-dimensional hard constraints such as contractual constraints, technological logic relationships, exclusive constraints on special machinery, key material supply constraints, and environmentally sensitive process window constraints. Based on this construction knowledge graph, the system employs constraint solving and optimization algorithms for intelligent deduction. The generated multi-level schedule naturally incorporates resource feasibility and technological compliance, fundamentally preventing the disconnect between the plan and on-site execution conditions, and significantly improving the reliability and executability of the plan.

[0018] It can intelligently diagnose logical contradictions, resource conflicts, and feasibility defects in source documents based on the results of project planning and generate structured processing reports. This enables project teams to proactively identify and correct potential problems in early planning before construction begins, optimizing not only the multi-level schedule itself but also driving improvements in the quality of upstream documents such as contracts and construction plans, forming an intelligent closed loop of "plan generation → anomaly identification → report generation → plan iteration." It outputs integrated, executable core deliverables, laying a precise benchmark for digital collaborative management and control throughout the project lifecycle.

[0019] This invention also outputs a final construction plan that can be directly used to guide on-site construction, resource procurement, and professional collaboration. It can be integrated with project management, BIM, procurement, and other systems through standardized interfaces, serving as an authoritative progress benchmark for the digital twin operation of the project and providing a solid data foundation for the lean management and dynamic control of the project.

[0020] A second aspect of the present invention provides a computer-readable storage medium on which a program is stored, and when the program is executed by a processor, it implements the planning method in any of the above-described technical solutions.

[0021] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0022] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a first flowchart of a planning method according to an embodiment of the present invention; Figure 2 This is a second flowchart of a planning method according to an embodiment of the present invention; Figure 3 This is a third flowchart of a planning method according to an embodiment of the present invention; Figure 4This is a fourth flowchart of a planning method according to an embodiment of the present invention; Figure 5 This is a fifth flowchart of a planning method according to an embodiment of the present invention; Figure 6 This is a sixth flowchart of a planning method according to an embodiment of the present invention; Figure 7 This is a seventh flowchart of a planning method according to an embodiment of the present invention; Figure 8 This is the eighth flowchart of a planning method according to an embodiment of the present invention. Detailed Implementation

[0023] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0024] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0025] In the construction preparation phase of thermal power plant construction projects, existing technologies typically rely on project managers to manually read construction documents such as construction organization designs and special construction plans, as well as project contract documents, to manually extract process information, resource arrangements, and contract period constraints, and then use general progress management software (such as Primavera P6 and Microsoft Project) to enter and build the progress plan item by item.

[0026] This approach has significant technical drawbacks: Firstly, construction and contract documents are voluminous and fragmented, making manual extraction of key information inefficient and prone to overlooking milestones, schedule constraints, or technological logic, leading to incomplete plans or logical inconsistencies. Secondly, manually generated schedules are difficult to automatically verify during generation to ensure they meet mandatory contractual time requirements and the feasibility of construction techniques. They also cannot proactively identify potential anomalies such as schedule delays and resource conflicts during the planning phase, often requiring repeated manual verification during later construction phases, resulting in frequent rework and high adjustment costs. Therefore, there is an urgent need for a planning method that can automatically extract key information from construction and contract documents, intelligently generate compliant and feasible construction plans, and proactively identify anomalies in the plans, thereby reducing manual intervention and improving the efficiency and quality of plan development. The following is a reference... Figure 1A planning method and a computer-readable storage medium are described according to some embodiments of the present invention.

[0027] like Figure 1 As shown, a first aspect of the present invention provides a planning method for thermal power plant construction projects, comprising: Step 102: Obtain data documents, wherein the data documents include a first data document of the thermal power construction project and a second data document of the construction contract; Step 104: Preprocess the data document to obtain parsable text units; Step 106: Parse the parsable text units to obtain the construction knowledge graph and contract constraints; Step 108: Based on the construction knowledge graph and contractual constraints, simulate and extrapolate the thermal power construction project to obtain a multi-level construction schedule. Step 110: Obtain abnormal information from the multi-level schedule, analyze and process the abnormal information, and generate a processing opinion report; Step 112: Based on the processing feedback report, modify the multi-level schedule to obtain the final construction plan; The parsable text unit includes a first sub-parsable text unit and a second sub-parsable text unit.

[0028] This invention provides a planning method for thermal power plant construction projects. In practice, the user first needs to obtain data documents, including a first data document and a second data document. The first data document refers to technical documents related to the construction of the thermal power plant project, such as construction organization design and specialized construction plans; the second data document refers to the construction contract, which includes legally binding clauses such as the construction period, milestones, and delivery conditions. The user can select a system capable of executing the planning method proposed in this invention. The system inputs the first and second data documents into a processing platform. After obtaining the first and second data documents, the system preprocesses them to obtain parsable text units. Here, "preprocessing" refers to the initial organization and transformation of the original documents into a standardized text form suitable for subsequent parsing by the processing platform, but it does not limit the specific technical means used.

[0029] The parsable text unit includes a first sub-parsable text unit and a second sub-parsable text unit. The first sub-parsable text unit originates from a first data document, and the second sub-parsable text unit originates from a second data document. Next, the system parses the parsable text units (i.e., it parses both the first and second sub-parsable text units) to obtain the construction knowledge graph and contract constraints.

[0030] It should be noted that the construction knowledge graph is a knowledge model that uses a graph structure (nodes and edges) to represent the logical relationships and dependencies between various construction tasks, equipment, and resources in a thermal power plant construction project; contract constraints are mandatory time requirements extracted from the construction contract, such as the latest completion date of each milestone node and the total project duration.

[0031] After obtaining the construction knowledge graph and contractual constraints, the system performs computer simulations of thermal power plant construction projects based on these constraints, automatically generating multi-level schedules. It should be noted that a multi-level schedule refers to a construction timetable organized according to the project hierarchy (such as system level, sub-item level, and work process level), reflecting the planned start and end times of tasks at different levels.

[0032] Building upon this foundation, the system further acquires anomaly information within the multi-level schedule. This anomaly information can refer to issues within the multi-level schedule that violate contractual constraints or construction logic, such as milestone delays, resource conflicts, or logical inconsistencies. The system analyzes and processes this anomaly information, automatically generating a processing report. This report typically includes a detailed description of the anomaly, problem location, root cause analysis, and suggested modifications.

[0033] Finally, the user or system modifies the multi-level schedule based on the suggestions in the processing feedback report to obtain the final construction plan.

[0034] Specifically, based on the processing feedback report, the multi-level schedule is modified to obtain the final construction plan, including: user-made modifications based on judgment and / or system-based modifications to the multi-level schedule based on the suggestions in the processing feedback report.

[0035] The final construction plan is a revised formal plan document that meets the requirements of contractual constraints and construction feasibility. It can be directly used to guide on-site construction and progress control of thermal power plant construction projects.

[0036] Through the above steps, this invention achieves automated generation from original construction documents and contract documents to an executable construction plan, effectively overcoming many shortcomings of existing technologies that rely on manual operation. Specifically, users only need to input the first data document and the second data document, and the system can automatically complete the entire process of preprocessing, parsing, deduction, anomaly detection, and modification. There is no need for manual reading and input of information such as the construction period, milestone nodes, and process logic in the documents, thereby completely avoiding the problems of low efficiency and easy omission or misinterpretation of key constraints by manual extraction.

[0037] Meanwhile, since the process logic represented by the construction knowledge graph and the mandatory time requirements in the contract constraints are integrated at the planning generation stage, the output multi-level schedule plan is naturally compliant and feasible. There is no need to separately verify whether it meets the contract period and construction logic after generation, which solves the defects of existing technology where planning and constraint verification are separated and cannot be automatically verified.

[0038] Furthermore, the system proactively acquires anomaly information from multi-level schedule plans and generates handling opinion reports. This allows for the early identification of potential problems such as project delays and resource conflicts during the planning stage, guiding users or the system to make targeted modifications. This avoids the passive situation of having to wait until subsequent construction or repeated manual verification to discover problems, as is the case with traditional methods. The final construction plan is a formal plan document that has been corrected for anomalies and meets contractual constraints and construction feasibility requirements, significantly improving the efficiency, compliance, and executability of thermal power plant construction schedule planning.

[0039] Furthermore, such as Figure 2 As shown, in some embodiments of the present invention, preprocessing of the data document to obtain parsable text units includes: Step 202: Convert the format of the data document to obtain the first derived text; Step 204: Remove noisy data from the first derived text and obtain the second derived text; Step 206: Divide the second derived text into multiple text segments according to the preset title to obtain the third derived text; Step 208: Annotate the content in the third derived text based on preset tags to obtain parsable text units.

[0040] In the above embodiments, firstly, the system performs format conversion on the acquired data documents (including the first data document and the second data document), uniformly converting source files of different formats (such as PDF, Word, etc.) into plain text, thereby obtaining the first derived text. The first derived text refers to the original text content after format unification, which eliminates the structural differences caused by different document formats and provides a unified input for subsequent processing.

[0041] It should be noted that because preprocessing only standardizes the document format and other issues, and is unrelated to the specific content of the file, the system operates the same way on the first data document and the second data document.

[0042] Specifically, data documents may include documents such as the "EPC General Contracting Contract" and the "Construction Organization Design" (including construction plans for each specialty).

[0043] It should be noted that the "EPC General Contracting Contract" refers to a contracting agreement signed between the owner and the general contractor that covers the entire process of design, procurement, and construction. It clearly stipulates legally binding clauses such as the project's construction period, milestones, delivery standards, and liability for breach of contract, and serves as a mandatory basis for formulating the construction schedule.

[0044] The "Overall Construction Organization Design" (including construction plans for each specialty) is a comprehensive technical document prepared by the construction team. It systematically plans the construction deployment, main construction methods, resource allocation (such as personnel, machinery, and materials), overall construction schedule, and specific construction plans for each specialty (such as boiler, steam turbine, electrical, and thermal control). It serves as the fundamental source for extracting core information such as process logic, resource requirements, and technological requirements. This invention uses these two types of documents as input and automatically generates a refined construction schedule through subsequent processing.

[0045] Furthermore, the system removes noisy data from the first derived text to obtain the second derived text. The second derived text refers to the clean text obtained after removing this noise from the first derived text. This noisy data includes, but is not limited to, redundant information such as headers, footers, watermarks, irrelevant blank lines, and non-textual comments. Removing this noise ensures that the text content used for subsequent parsing is pure and interference-free.

[0046] Based on this, the system segments the second derived text into multiple text segments according to preset titles, thus obtaining the third derived text. The third derived text refers to the multiple text segments with independent semantics obtained after segmenting the second derived text according to preset titles. Preset titles refer to chapter titles with structural characteristics in thermal power plant construction engineering documents, such as "Construction Period and Milestone Nodes" in the contract or "Boiler Professional Construction Plan" in the construction organization design. The system automatically identifies the text boundaries based on these titles, segmenting the second derived text into several text segments with independent semantics. Each text segment corresponds to a relatively complete document sub-block, thus obtaining the third derived text.

[0047] Finally, the system annotates the content in the third-generation derived text based on preset tags to obtain parsable text units. The preset tags include categories such as "milestone nodes," "construction logic," "mechanical configuration," "material requirements," and "process requirements." The system automatically assigns corresponding tags to key content within each text segment according to predefined rules, forming structured text with preliminary semantic markings, which can then be parsed. These tags provide a contextual framework for subsequent parsing of the parsable text units.

[0048] It should be noted that "milestone nodes" refer to key time nodes clearly stipulated in the construction contract, such as "completion of boiler hydrostatic test" and "first unit connected to grid for power generation", which are used to mark major events that the project must achieve on schedule.

[0049] "Construction logic" refers to the sequential dependence or causal relationship between procedures during construction, such as "the installation of the heating surface can only begin after the boiler steel frame hoisting is completed."

[0050] "Mechanical configuration" refers to the information on the main mechanical equipment required for construction, including equipment name, model, quantity, planned arrival time, and service scope, such as "ZSC800 tower crane, 1 unit, to arrive in August 2025, for boiler large plate girder hoisting".

[0051] "Material requirements" refers to a list of key materials or equipment required during construction, including the name, specifications, quantity, and delivery time of the materials / equipment, such as "boiler heating surface tube bank, delivery in March 2026".

[0052] "Process requirements" refers to the technical specifications, construction methods, or special process conditions that must be followed during construction, such as "concrete pouring in winter requires the use of a comprehensive heat storage method for curing, extending the curing period by 3 days."

[0053] Through the four steps of format conversion, noise removal, title segmentation, and tag annotation, the original messy and unstructured engineering documents are gradually transformed into parsable text units with clear structure and preliminary semantic annotations. This lays a high-quality data foundation for subsequent entity recognition, relationship extraction, and the construction of construction knowledge graphs, significantly improving the accuracy and efficiency of parsing and processing.

[0054] Furthermore, in some embodiments of the present invention, such as Figure 3 As shown, parsable text units are parsed to obtain a construction knowledge graph and contract constraints, including: Step 302: Train a text intelligent parsing model based on the labeled word set and semantic rules of thermal power construction projects to obtain the thermal power text intelligent parsing model; Step 304: Construct a domain database based on domain knowledge of thermal power plant construction projects; Step 306: Using the thermal power text intelligent parsing model, identify and extract the entities in the first sub-parsable text unit and the logical relationships between different entities; Step 308: Using the thermal power text intelligent parsing model, identify and extract the contract constraints in the second sub-parsable text unit; Step 310: Instantiate entities, logical relationships, and contractual constraints into entity nodes and edges using the domain database; Step 312: Construct a construction knowledge graph based on entity nodes and edges.

[0055] In the above embodiments, firstly, the system trains a text intelligent parsing model based on the labeled word set and semantic rules of thermal power construction projects, thereby obtaining the thermal power text intelligent parsing model.

[0056] The term set refers to a collection of keywords in the thermal power field that are manually compiled in advance, such as words like "steam turbine," "boiler," "large plate beam," and "cover," along with their corresponding weights, synonym pairs (e.g., "cover" and "cylinder" are synonyms), and text range limitation rules (e.g., extracting the content after the word "to"); semantic rules refer to the matching logic used to understand the meaning of the text, such as bidirectional matching of synonyms and keyword quantity statistics.

[0057] Specifically, the intelligent text parsing model can be implemented using an NLP (Natural Language Processing) model. NLP is an abbreviation for Natural Language Processing, referring to a computational model trained using machine learning techniques that can understand and process human language. Specifically, pre-trained language models based on the Transformer architecture (such as BERT, RoBERTa, etc.) can be used. By pre-training on a large amount of general text, these models acquire basic grammatical and semantic knowledge. Based on this, fine-tuning is performed using the thermal power construction engineering annotation vocabulary and semantic rules provided in this invention. The model can quickly adapt to the specialized language characteristics of the thermal power field, learning to identify entities in the text (such as processes, equipment, milestone nodes) and the constraints between entities (such as temporal dependencies, resource bindings). Compared to traditional rule-based methods, NLP models can better handle complex situations such as synonyms and contextual dependencies, significantly improving parsing accuracy and generalization ability. The intelligent parsing model for thermal power plant text obtained by adopting the NLP model (i.e., a fine-tuned NLP model) is a domain-adapted natural language processing model specifically designed to automatically extract structured information from contracts and construction documents of thermal power plant construction projects.

[0058] Based on these labeled word sets and semantic rules, the basic language model is trained to recognize professional entities and constraint relationships in thermal power texts. The model obtained after training is the intelligent parsing model for thermal power texts.

[0059] Specifically, such as Figure 4 As shown, a text intelligent parsing model is trained based on the labeled word set and semantic rules of thermal power construction projects to obtain a thermal power text intelligent parsing model, including: Step 402: Obtain the tagging term set for thermal power construction projects. The tagging term set includes the vocabulary of thermal power construction projects, the weight of the vocabulary, synonyms, and the rules for limiting the text scope. Step 404: Based on the labeled word set and semantic rules, train the text intelligent parsing model so that the trained text intelligent parsing model can identify entities in the first sub-parseable text unit, logical relationships between different entities, and contractual constraints in the second sub-parseable text unit. Use the trained text intelligent parsing model as the thermal power text intelligent parsing model.

[0060] In a further embodiment of the present invention, firstly, the system acquires a set of labeled terms for thermal power plant construction projects. The labeled term set refers to a set of structured data pre-organized by humans to guide model learning, specifically including four parts: first, vocabulary related to thermal power plant construction projects, such as common terms like "steam turbine," "boiler," "large plate beam," and "covering"; second, the weight corresponding to each term, used to indicate the importance of the term in the recognition process, for example, "blowing pipe" has a weight of 6, and "steel frame" has a weight of 4; third, synonymous terms, i.e., different expressions of the same concept, for example, "covering" and "cylinder" are synonyms, as are "acid washing" and "chemical cleaning"; fourth, rules for limiting the scope of text, for example, in the expression "from the start of construction to the pouring of the first batch of concrete in the main plant," the content after the word "to" is limited to be extracted as key information. The above-mentioned labeled term set provides domain-specific supervisory signals for model training.

[0061] Secondly, based on the aforementioned annotated word set and semantic rules, the system trains the initial intelligent text parsing model. Semantic rules refer to the matching logic used to understand the meaning of text, such as bidirectional matching of synonyms, keyword count, and text position determination. The training goal is to enable the model to accurately identify various entities (such as process names, equipment names, and milestone nodes) from the first parsable text unit, as well as the logical relationships between entities (such as temporal dependencies and resource bindings), and the contractual constraints in the second parsable text unit. After training, the resulting model is used as the intelligent text parsing model for thermal power plants. This model possesses professional parsing capabilities specific to the thermal power construction field and can automatically extract key information from newly input documents.

[0062] It should be noted that "bidirectional synonym matching" means that when the system processes text, it correlates a group of synonymous words with each other. No matter which word in the group is used in the document, it can be recognized as the same concept. For example, after setting "cylinder covering" and "cylinder buckling" as bidirectional matching, when "cylinder covering" or "cylinder buckling" appears in the document, the system will understand them as events corresponding to the same milestone node (turbine cylinder covering / cylinder buckling), thus avoiding missing key information due to different word usages. Bidirectional matching means that the matching relationship is reciprocal, that is, A matches B and at the same time B matches A, ensuring that no matter which word is used in the text, it can be correctly recognized. "Keyword quantity statistics" refers to counting the number of occurrences of words belonging to a preset labeled word set in a sentence or paragraph. If a segment contains more thermal power keywords, then this segment is more likely to belong to key progress information, and the system can increase the weight of this segment or process it preferentially based on this. "Text position determination" refers to extracting effective content based on the relative position of words in the text or their relationship with specific marker words (such as "to", "from", "until"). For example, for "from the start to the pouring of the first batch of concrete in the main plant building", by determining the position after the word "to", "the pouring of the first batch of concrete in the main plant building" is extracted as the event corresponding to the milestone node. These three rules are used to assist the model in more accurately locating key information.

[0063] Through the above steps, the present invention uses a small-scale but high-quality labeled word set and semantic rules to achieve the adaptation of the general language model to the thermal power field at a low cost. Compared with the method of completely relying on training with a large amount of manually labeled data, this method significantly reduces the workload of data preparation while maintaining a high recognition accuracy. The obtained intelligent parsing model for thermal power texts provides a dedicated parsing tool for subsequent entity extraction, relationship recognition, and knowledge graph construction, thereby improving the automation degree and domain adaptability of the entire plan-making process.

[0064] After "training an intelligent parsing model for texts based on a labeled word set and semantic rules for thermal power construction projects to obtain an intelligent parsing model for thermal power texts", the system pre-constructs a domain database based on the domain knowledge of thermal power construction projects.

[0065] Specifically, the domain database includes: an entity word library for thermal power construction projects, a construction process term library, progress restriction conditions for thermal power construction projects, and a labeled example database.

[0066] Among them, the entity word library for thermal power construction projects is used to cover the entity names and attributes of equipment, systems, processes, and partial commissioning in thermal power construction projects, providing a standardized vocabulary basis for text parsing.

[0067] The construction procedure terminology library defines universally applicable and deterministic logical relationships in thermal power plant construction. Specifically, it includes sequence dependency rules and state delivery rules. Sequence dependency rules are used to define the temporal logical relationships between procedures within the same discipline, such as "Finish-to-Start" (FS) and "Start-to-Start" (SS). State delivery rules are used to define the triggering conditions for work between different disciplines, such as "Completion of the crane track installation is a prerequisite for starting the turbine platform placement work."

[0068] Schedule constraints for thermal power plant construction projects refer to various rules or parameters that impose rigid constraints on the construction schedule, including but not limited to the sequential dependencies between work processes, the availability window of critical resources, the supply cycle of materials, and the impact of environmental factors on operations. These conditions are not optional optimization objectives, but rather hard boundaries that the schedule plan must adhere to; any violation will render the plan infeasible or impossible to execute on site.

[0069] The annotation example database refers to a database that stores a large number of historical document fragments pre-annotated manually by domain experts. Each annotation fragment is marked with information such as entity type (e.g., process, equipment, milestone node), logical relationship (e.g., sequence dependency, state delivery), and constraint attributes (e.g., resource exclusivity, schedule constraints). The annotation example database serves two purposes: firstly, it is used to fine-tune and train the text intelligent parsing model, enabling it to recognize technical terms and complex logic in the thermal power field; secondly, during automatic parsing, it can serve as a reference template for rule matching, improving the system's annotation accuracy and generalization ability for new data documents.

[0070] Specifically, the schedule constraints used in thermal power plant construction projects include: a schedule constraint feature model library, a special machinery exclusivity constraint model, a key material timing constraint model, and an environmentally sensitive process window constraint model.

[0071] Among them, the schedule constraint feature model library is used to clearly define the key constraint types and their quantitative attributes that directly determine the scheduling of processes. It is a bridge to transform resource conditions into schedule constraints, specifically including special machinery exclusive constraint model, key material timing constraint model and environmentally sensitive process window constraint model.

[0072] The special machinery exclusivity constraint model is used to define the unique identifier, key performance parameters, and planned service time window of irreplaceable large machinery that has a decisive impact on project schedule. Its core features are uniqueness and time exclusivity, which are used to force the avoidance of time conflicts in the schedule.

[0073] The critical material timing constraint model is used to define the supply cycle, warehousing requirements, and other attributes of equipment and materials that have long cycles, are customized, or need to be delivered synchronously. Its core is to establish a hard link between material demand and process start time.

[0074] The environmentally sensitive process window constraint model first defines the process categories and their trigger thresholds that are sensitive to specific environmental factors (such as low temperature, rainfall, and strong winds). Second, it associates the historical meteorological statistics of the project location with the probabilistic operation window. Finally, it generates a rule library of measures-schedule conversion to quantify the impact of common environmental response measures on schedule or efficiency.

[0075] It should be noted that the "Measures-Schedule Conversion Rule Library" is a component of the schedule constraint feature model library. It quantifies different construction measures taken due to varying project conditions (such as geological conditions, construction environment, and technical solutions) into specific schedule adjustment values ​​or efficiency impact coefficients. For example, in foundation treatment engineering, different pile types such as precast piles, steel pipe piles, or cast-in-place piles can be selected based on geological conditions. Each pile type has different construction processes and required schedules, and the rule library will output the corresponding schedule increment based on the selected measure. Another example is when using the integrated heat storage method to cure concrete during winter construction; the "Measures-Schedule Conversion Rule Library" outputs "equivalent extension of curing period by 3 days." When rainy season construction uses measures such as rain shelters, it may output "efficiency coefficient 0.85." Through the "Measures-Schedule Conversion Rule Library," the system can automatically convert various construction measures described in the text into calculable schedule parameters or efficiency coefficients in the schedule plan, thereby ensuring the feasibility and adaptability of the plan under different project conditions.

[0076] By coordinating the above four types of information, the domain database can significantly improve the accuracy and efficiency of extracting domain knowledge from unstructured documents, making the constructed construction knowledge graph more in line with the actual logic and constraints of thermal power engineering, thereby providing a reliable data foundation for automatically generating compliant and feasible construction schedule plans.

[0077] After “the system pre-builds a domain database based on the domain knowledge of thermal power construction projects”, the system calls the thermal power text intelligent parsing model trained above to identify and extract the first sub-parsable text unit (from construction organization design and other construction documents), and extracts the entity and the logical relationship between different entities.

[0078] Among them, an entity refers to an object that can be independently named during the construction process, such as a specific process ("boiler steel frame hoisting"), equipment ("ZSC800 tower crane"), resources ("concrete team"), etc.; a logical relationship refers to the sequence, dependency or binding relationship between these entities, such as "completion of boiler steel frame acceptance" is a prerequisite for "start of heating surface hoisting".

[0079] Specifically, "identifying and extracting entities and logical relationships between different entities in the first sub-parseable text unit through the thermal power text intelligent parsing model" includes: the system parses the first sub-parseable text unit, transforming it into computable processes, resource configurations, and logical relationships between different entities, and binding it with various models in the domain database. In this embodiment, the process and logic are extracted first, the process sequence in the construction plan is identified, and based on the process logic rule base in the domain database, the sequential dependencies between processes (such as FS and SS relationships within the same discipline) and cross-discipline state delivery relationships (such as "completing the installation of the crane track" triggering "the turbine platform is in place") are automatically labeled or inferred. Secondly, key resources are bound to logical relationships between different entities. Based on the schedule constraint feature model library, resource information is accurately extracted and specific instances are created. For major construction machinery, attributes such as model, quantity, main purpose, and planned arrival / departure time in the construction machinery equipment table are identified, and special machinery exclusive constraint instances are created and strongly bound to the processes that depend on the machinery. For key materials, component names and delivery times in the equipment demand plan are identified, and requirements are established with relevant processes. Supply chain linkage. Finally, environmental constraints are identified and instantiated: When parsing sections on "winter construction" and "rainy season construction" in the construction organization design, the system executes an intelligent conversion process—first, it identifies specific work procedures in the text that specify applicable environmental measures, tags these procedures with environmental sensitivity labels, then queries the historical meteorological database based on the project's geographical location, automatically calculates unsuitable work windows for the area (such as "low-temperature periods"), and then invokes the measures. The schedule conversion rule library quantifies environmental response measures (such as "integrated heat storage method maintenance") into schedule adjustment values ​​(such as "equivalent extension of maintenance period by 3 days") or efficiency coefficients (such as 0.85). Finally, it creates an environmentally sensitive process window constraint instance, explicitly binds it to the relevant process, and includes core attributes such as constraint type, recommended work window, schedule adjustment or efficiency coefficient in non-recommended windows.

[0080] Simultaneously, the system also identifies and extracts the second sub-parseable text unit (derived from the construction contract) using a thermal power text intelligent parsing model, extracting contractual constraints from it. These contractual constraints refer to mandatory time requirements explicitly stipulated in the contract, such as the latest completion date for each milestone and the total project duration.

[0081] Specifically, "identifying and extracting contractual constraints in the second sub-parseable text unit through the thermal power text intelligent parsing model" includes: the system combines thermal power professional terms and entity library in the domain database to accurately identify and extract mandatory constraints directly related to the time plan, and transform them into instantiable information. The core content mainly includes four categories: First, milestone nodes, extracting events explicitly defined in the contract (such as "completion of boiler hydrostatic test") and their mandatory completion dates, and assigning a unique name to each node; Second, the critical schedule framework, extracting the total project duration and key segment durations (such as "from the zero-meter completion of the main plant foundation to the boiler hydrostatic test"), and using them as the top-level global time constraint for the project; Third, schedule-bound deliverables or work interfaces, extracting intermediate deliverables or work interface descriptions explicitly agreed upon in the contract as prerequisites for work processes (such as "completion of crane track installation" as a prerequisite for "turbine platform placement"), used for subsequent definition of task interfaces; Fourth, bill of quantities information, parsing the bill of quantities in the contract appendix, extracting the sub-items and their corresponding quantities (such as cubic meters of concrete, tons of steel reinforcement, meters of pipe, number of equipment, etc.), and establishing the association between the quantity entities and professional, system, and location information.

[0082] Next, the system invokes the constructed domain database to instantiate the extracted entities, logical relationships between entities, and contractual constraints into entity nodes and edges. "Instantiation" refers to transforming abstract information identified from text into concrete data objects that can be processed by a computer; "entity nodes" are the basic units in a knowledge graph, each node representing an entity (such as a process or a resource); "edges" are directed lines connecting two nodes, used to represent the logical relationships between them (such as sequence, resource binding, etc.). Through the terminology specifications and constraint models in the domain database, the system can accurately map entities and relationships into nodes and edges in the graph structure.

[0083] Finally, based on these entity nodes and edges, the system constructs a complete construction knowledge graph. This construction knowledge graph provides a structured data foundation for subsequent schedule simulation and deduction, enabling the system to perform logical retrieval, constraint verification, and path analysis directly on the graph.

[0084] Specifically, the system stores all entities, attributes, and relationships extracted and instantiated above using a graph data model, thereby constructing a structured and computable construction knowledge graph for thermal power projects. Specifically, in this construction knowledge graph, entities (such as processes, machinery, milestone nodes, and quantities) are stored as nodes, each containing its corresponding attribute information (such as the duration of a process, the model and service time window of machinery, and the mandatory completion date of milestone nodes); relationships between entities (such as the logical order between processes, the resource occupancy relationship of processes, and the binding relationship between resources and constraints) are stored as edges directly connecting nodes, with a clearly defined relationship type for each edge (such as "prerequisite," "bound," and "occupied"). This graph structure storage method naturally fits the complex constraint network in thermal power engineering, directly and efficiently supporting the retrieval, traversal, and inference calculation of multi-dimensional relationships, such as resource conflict detection and critical path analysis. Through the construction knowledge graph, a core data foundation is provided for the subsequent intelligent generation and optimization of schedule plans.

[0085] The above method enables automated conversion from parsable text units to construction knowledge graphs and contractual constraints, laying a solid data foundation for subsequent knowledge graph-based schedule generation and optimization.

[0086] Furthermore, in some embodiments of the present invention, such as Figure 5 As shown, a simulation of a thermal power plant construction project is conducted based on a construction knowledge graph and contractual constraints to obtain a multi-level construction schedule, including: Step 502: Generate the skeleton of the project-level work breakdown structure based on the construction knowledge graph; Step 504: Based on the technological logic relationships in the construction knowledge graph and the time constraints in the contractual constraints, generate a time plan for the smallest task unit. Step 506: Based on the multi-dimensional constraints of the construction knowledge graph, optimize the time plan and obtain the optimized time plan; Step 508: Summarize the optimized time plan according to the hierarchical relationship of the skeleton to generate a multi-level schedule plan.

[0087] In the above embodiments, during the generation of multi-level schedules based on the construction knowledge graph, the system no longer relies on a general WBS template, but directly traverses the constructed construction knowledge graph. First, based on entity nodes such as "system" and "profession" in the graph and their mutual "inclusion" relationships, the system automatically extracts a project-level work breakdown structure skeleton that conforms to the standards for thermal power engineering. For example, a hierarchical framework from "thermal system" to "boiler installation" and then to "steel frame hoisting" defines the basic hierarchical relationship of the plan, but does not yet contain specific time information.

[0088] It's important to note that WBS is an abbreviation for Work Breakdown Structure. In project management, WBS is a hierarchical structure that breaks down a large project into smaller, more manageable components according to systems, specialties, sub-items, and processes. For example, in a thermal power plant construction project, the first-level WBS could be the entire power plant project, the second-level WBS could be broken down into boiler systems, turbine systems, electrical systems, etc., and the third-level WBS could be further broken down into specific tasks such as steel frame installation, heating surface installation, and pipe welding. The WBS skeleton refers to this hierarchical framework that has not yet been filled with time information; it defines the parent-child relationships between tasks and the overall structure.

[0089] Then, based on the technological logic relationships in the construction knowledge graph and the time constraints in the contractual constraints, the system generates a time plan for the smallest task unit. The technological logic relationships refer to the sequential dependencies between work processes (e.g., "heating surface installation can only begin after the boiler steel frame hoisting is completed"), and the time constraints refer to the mandatory time boundaries such as the completion dates of milestone nodes and the total construction period specified in the contract. Using the time corresponding to the milestone node as the time anchor, the system works backward from the technological logic relationships to calculate the planned start and end times for each smallest task unit (i.e., a specific, indivisible operation, such as "hoisting the first steel column"), thus forming a preliminary time plan.

[0090] Specifically, the initial time plan can be estimated using net work duration: the system calculates the duration required for continuous construction of a work package based on the quantity of work, construction quotas, and planned resource input. The formula is: Net Work Duration = Quantity of Work ÷ (Quota Unit Output × Resource Input). Here, quota unit output refers to the amount of work completed per unit of resource per unit of time (e.g., the amount of concrete poured per man-day), and resource input refers to the amount of resources allocated to the work package (e.g., the number of workers in the concrete team). Using this formula, the system can reasonably estimate the theoretical construction time of each work package, providing a basis for subsequent scheduling.

[0091] Then, based on the multi-dimensional constraints in the construction knowledge graph, the system optimizes the above time plan to obtain an optimized time plan. The multi-dimensional constraints include, but are not limited to, temporal logic constraints (i.e., the sequential dependencies between processes in the construction knowledge graph), exclusive constraints of special machinery (e.g., a tower crane can only be used on one work face at a time; if a work package must use a specific machine, its scheduling must be strictly arranged within the available time window of that machine, and ensure that the machine is not occupied by other work packages at the same time), supply cycle constraints of critical materials (i.e., from the temporal constraint example of critical materials, the planned start time of a work package must not be earlier than the arrival time of its necessary materials, such as installation can only begin after a certain equipment arrives), work window constraints of environmentally sensitive processes (i.e., from the window constraint example of environmentally sensitive processes and process requirements, work packages must be arranged within the allowed environmental work window; if process breaks are involved, a mandatory break time must be automatically added after the net work period before driving subsequent processes, such as extending the work period for winter concrete curing), and deadline constraints of milestone nodes (i.e., the mandatory completion date corresponding to the milestone node in the contract, serving as the mandatory target of the entire scheduling network), etc. The system detects whether there are any conflicts in the time plan that violate these constraints (e.g., the same tower crane is assigned to two different processes at the same time), and adjusts the time plan of the relevant smallest task unit according to the preset adjustment strategy (e.g., adjusting the start time of non-critical processes, increasing resource buffers, etc.) until all constraints are met, resulting in an optimized time plan.

[0092] Finally, the system optimizes the time schedule by aggregating it hierarchically according to the work breakdown structure framework, generating a multi-level schedule. Specifically, the system merges the parameters (start and end times) from the time schedule of the smallest task unit at the bottom level with the parent tasks at the upper level, calculating the start and end times of each sub-item, each unit project, and even the entire project, thus forming a schedule table with multiple granularities, including Level 1 (overall project), Level 2 (unit project), and Level 3 (sub-item project). This multi-level schedule can meet the management's need for overall project duration control and also guide the specific construction arrangements at the on-site operational level.

[0093] Through the above steps, this invention achieves automated generation and optimization of multi-level schedules from construction knowledge graphs. Compared with manual scheduling, this method automatically incorporates process logic, contract deadlines, and multi-dimensional resource constraints during the generation process. The resulting schedule not only meets contractual period requirements but also demonstrates high feasibility and executability in terms of resource utilization and environmental impact, significantly improving the efficiency and quality of schedule preparation for thermal power plant construction projects.

[0094] Furthermore, in some embodiments of the present invention, such as Figure 6As shown, based on the multi-dimensional constraints of the construction knowledge graph, the time plan is optimized to obtain the optimized time plan, including: Step 602: Identify conflicting items in the time plan that violate multi-dimensional constraints; Step 604: Adjust the time plan of the smallest task unit corresponding to the conflict item according to the preset adjustment strategy until all multi-dimensional constraints are met, and use the adjusted time plan as the optimized time plan.

[0095] In the above embodiments, during the optimization of the time plan based on the multi-dimensional constraints of the construction knowledge graph, the system first identifies conflicting items in the time plan that violate the multi-dimensional constraints. A conflicting item refers to a situation where the start or end time of a minimum task unit (i.e., an indivisible work package) in the time plan contradicts any of the constraints in the multi-dimensional constraints. For example, the same tower crane may be assigned to two different work packages at the same time, or the planned start time of a work package may be earlier than the arrival time of its required critical materials.

[0096] After identifying the conflict, the system adjusts the time schedule of the smallest task unit corresponding to the conflict according to the preset adjustment strategy.

[0097] Specifically, the preset adjustment strategies can be rules such as delaying work packages on non-critical paths, increasing resource buffers, adjusting process sequences, or reallocating resources without affecting the critical path.

[0098] The system repeatedly performs conflict identification and time adjustment until all multi-dimensional constraints are met. At this point, the adjusted time plan is used as the optimized time plan.

[0099] Through the above optimization process, the system can automatically eliminate resource conflicts, timing contradictions, and constraint violations in the time plan, so that the optimized time plan is feasible on site while meeting the contract period and construction logic, providing a reliable foundation for the subsequent aggregation and generation of multi-level schedule plans.

[0100] Furthermore, in some embodiments of the present invention, such as Figure 7 As shown, the system retrieves anomaly information from multi-level schedules, analyzes and processes this anomaly information, and generates a processing report, including: Step 702: Verify the multi-level schedule based on contractual constraints and resource allocation, and obtain abnormal information; Step 704: Analyze the abnormal information and generate a processing opinion report.

[0101] In the above embodiments, the system verifies the generated multi-level schedule based on contractual constraints and resource allocation to obtain abnormal information. Resource allocation refers to the arrangement of construction resources extracted from the construction organization design, including the models and quantities of major equipment, the staffing scale of various professional personnel, and the supply cycle of key materials.

[0102] Specifically, resource allocation can be identified and extracted from the first sub-parseable text unit through the thermal power text intelligent parsing model.

[0103] The system compares the timelines of each task in the multi-level schedule with the contractual constraints to check for any overdue milestones or exceeding the total project duration. Simultaneously, it compares the resource requirements of each task within the same time period in the plan with the available resources in the resource allocation to check for resource conflicts (e.g., the same tower crane being assigned to two different tasks simultaneously) or resource shortages (e.g., the required number of personnel exceeding the team size). Any violations of contractual constraints or resource allocation issues discovered during the above verification process are recorded as anomalies. Anomaly information includes at least the anomaly type (e.g., schedule anomaly or resource anomaly), the location of occurrence (the relevant task node or time period), and specific details (e.g., number of days of delay, name of conflicting resource).

[0104] The system then analyzes the acquired anomaly information and generates a processing report. Specifically, the analysis process includes, but is not limited to: identifying the possible causes of the anomaly based on its anomaly type, such as milestone delays possibly due to insufficient estimation of the duration of preceding processes, or resource conflicts possibly due to multiple tasks sharing the same critical equipment without staggering their schedules; the system can also assess the impact of the anomaly on the overall project duration or subsequent processes by combining contract terms and construction logic.

[0105] Based on the analysis results, the system automatically generates a handling opinion report. This report lists each anomaly and its corresponding handling suggestions in a structured format. Specifically, the handling opinion report includes: the issue item, i.e., the specific inconsistency, contradiction, or infeasibility point; the issue location, i.e., the source document name, chapter, and original clause reference corresponding to the anomaly; the impact analysis, i.e., the quantitative or qualitative impact of the anomaly on the project schedule, critical path, or cost; and the modification suggestions, i.e., the specific adjustment direction for the source document. Through the above structured information, the handling opinion report can clearly and completely present the source, consequences, and resolution path of each anomaly, providing a direct basis for subsequent manual or automatic modifications to multi-level schedules. The handling opinion report provides a clear and traceable decision-making basis for subsequent manual or automatic modifications to the schedule.

[0106] Through the above steps, this invention can systematically identify contract violations and resource conflicts in multi-level schedules and automatically generate a processing opinion report containing root cause analysis and modification suggestions. This automates the schedule verification and problem diagnosis process, significantly improves the comprehensiveness and efficiency of anomaly detection, and provides strong support for rapid iterative optimization of the schedule.

[0107] Furthermore, in some embodiments of the present invention, such as Figure 8 As shown, the multi-level schedule is validated based on contractual constraints and resource allocation to obtain anomaly information, including: Step 802: Based on the milestone nodes and schedule requirements in the contract constraints, verify whether the completion time of each milestone node in the multi-level schedule meets the requirements, and generate schedule exception information. Step 804: Based on the number of equipment, personnel, and material supply cycle in the resource allocation, determine whether there are resource conflicts or insufficient resources in the multi-level schedule and obtain the judgment result; Step 806: Based on the judgment result, generate resource anomaly information; Step 808: Merge the schedule exception information and resource exception information to obtain the exception information.

[0108] In the above embodiments, during the process of verifying multi-level schedules based on contractual constraints and resource allocation to obtain anomaly information, the system first verifies the planned completion time of each milestone node in the multi-level schedule one by one according to the milestone nodes and schedule requirements in the contractual constraints. Milestone nodes in the contractual constraints refer to key events explicitly stipulated in the contract (e.g., "completion of boiler hydrostatic test"), and each node has a mandatory completion date; schedule requirements include the total project duration and the duration of each segment. The system compares the planned completion time of each milestone node in the multi-level schedule with the mandatory contract date. If the planned time is later than the contract requirement, it is determined to be a schedule anomaly, and the deviation value (e.g., the number of days overdue) is recorded. All schedule anomaly information is summarized to form a schedule anomaly information set.

[0109] Secondly, the system determines whether there are resource conflicts or shortages in the multi-level schedule based on the number of equipment, personnel, and material supply cycles in the resource allocation. The resource allocation originates from the construction organization design in the first parsable text unit. The equipment quantity includes the number of machines of each model and their available time windows; the personnel quantity includes the size of each professional work team; and the material supply cycle includes the arrival time of critical materials. The system overlays resource requirements for all tasks within the same time period in the multi-level schedule. If a piece of equipment is assigned to multiple tasks at the same time, it is considered a resource conflict; if the total number of personnel required in a certain time period exceeds the number of available work teams, or if the material demand time is earlier than the supply cycle, it is considered a resource shortage. The above-mentioned resource anomalies are recorded one by one, and the judgment results are obtained, including the anomaly type, the tasks involved, and the conflict period, and resource anomaly information is generated based on the judgment results.

[0110] Finally, the system merges the schedule exception information with the resource exception information to obtain complete exception information, which is used for subsequent analysis and report generation.

[0111] Through the above verification, the system can automatically identify schedule violations and resource issues in multi-level schedules, providing a comprehensive and accurate abnormal data foundation for the optimization and adjustment of the schedule.

[0112] For ease of understanding, the method of the present invention will be explained through one embodiment of the present invention.

[0113] In one embodiment of the present invention, the present invention describes the process of generating and optimizing a multi-stage schedule for a 2×1000MW ultra-supercritical thermal power plant.

[0114] First, project management personnel log into the system and upload the project's EPC general contracting contract and overall construction organization design (including major construction plans for architecture, boiler, steam turbine, electrical, thermal control, welding, etc.) to the system. The system converts all documents into plain text and filters out irrelevant information such as headers and footers.

[0115] The system automatically identifies the structure of sections such as "Construction Period and Milestones" in the contract, "Main Construction Machinery Configuration" and "Winter Construction Measures" in the construction organization design. Based on the document's section structure (e.g., "Article 3 Construction Period and Milestones" in the contract, and "4.2 Main Construction Schemes" in the construction organization design), the system intelligently segments the text and assigns preliminary tags such as "Contract Clauses," "Milestones," "Construction Technology," and "Resource Configuration" to each paragraph, forming pre-processed text to be parsed. The system further segments the text and pre-labels it with tags such as "Contract Milestones," "Construction Logic," "Machinery Resources," and "Environmental Constraints," establishing context for in-depth parsing.

[0116] The system calls a pre-built domain database to perform deep parsing and instantiation of the preprocessed text. The domain database contains thermal power terminology (such as "boiler hydrostatic test"), process logic rules (such as "completion of boiler steel frame acceptance" is a prerequisite for "start of heating surface hoisting"), and schedule constraint feature models (such as "ZSC800 luffing tower crane" is defined as "special machinery exclusive constraint").

[0117] The system identifies the milestone "Completion date of the 'Boiler Hydraulic Test' shall not be later than May 30, 2026" from the contract text and instantiates it as a milestone node in the construction knowledge graph, with attributes including name and mandatory date. The system then parses the contract appendix, the "Bill of Quantities," and extracts "Foundation Structure..." Steam turbine foundation concrete C40 4400m 3 Information such as "engineering quantity entity" is created and associated with the "steam turbine base" part.

[0118] From the "Boiler Steel Frame Hoisting Plan," the system identifies the process sequence: "Column Hoisting → Beam Hoisting → Vertical Support Installation," and automatically establishes its sequential dependency relationship based on the rule base. From the "Main Construction Machinery Table," it identifies "No. 1 ZSC800 Tower Crane (Number: T1), scheduled to enter the site on August 1, 2025, to serve the hoisting of the boiler steel frame and heating surfaces." The system creates a special machinery exclusive constraint instance "T1" and strongly binds it to processes such as "Boiler Large Plate Beam Hoisting." The system analyzes the "Main Plant Frame Concrete Pouring Requires Comprehensive Heat Storage Method Curing" in the "Winter Construction Plan" and executes an intelligent conversion process: ① Marks "Main Plant Frame Concrete Pouring" as a low-temperature sensitive process; ② Queries historical meteorological data of the project site to determine the "low-temperature period" as November 15th to March 15th of the following year; ③ Calls the rule base to quantify "Comprehensive Heat Storage Method" as "Equivalent Extension of Curing Period: 3 days." Finally, an environmentally sensitive process window constraint instance is generated and bound to this process.

[0119] All the instantiated entities (tasks, machines, milestones, constraints) and their relationships (logic, binding, constraints) mentioned above are stored in the construction knowledge graph in the form of nodes and edges, forming a visualized and reasonable construction knowledge graph for thermal power projects, which serves as the sole data source for all intelligent operations.

[0120] The system uses a construction knowledge graph for simulation and deduction to generate a preliminary multi-level schedule. It traverses entities such as "boiler system" and "steam turbine system" and their relationships within the knowledge graph, automatically generating a multi-level work breakdown structure (WBS) framework that conforms to the power plant structure. The system uses contract milestones as mandatory anchor points and leverages the logical relationships in the graph for multi-point reverse scheduling. Work process logic, the exclusive time window of tower crane T1, the delivery date of critical equipment, and the allowable window for winter concrete pouring are simultaneously input into the scheduling engine as multi-dimensional constraints. For the smallest task unit, "boiler large plate girder hoisting," its net work period is calculated based on the bound workload, quotas, and team resources; the planned schedule must include it within the available time window of tower crane T1.

[0121] The system uses a constraint solving algorithm to calculate a set of feasible start and end dates that satisfy all constraints for all the smallest task units, automatically forming a three-level schedule network, and then summarizing them to generate a two-level schedule (the two together constitute the multi-level schedule of this invention).

[0122] Following the planning simulation, the system initiated the acquisition and analysis of anomaly information. The simulation revealed that in October 2025, the planned times for the "boiler large plate girder hoisting" and "steam drum hoisting" processes, both bound to the same tower crane T1, overlapped, triggering a resource conflict anomaly. The system conducted root cause analysis, not simply adjusting the plan, but tracing the source back to the construction organization design. The system found that the "service scope description" for tower crane T1 in the construction organization covered both of these critical processes, but failed to consider their time exclusivity, thus diagnosing a "conflict between the special machinery resource allocation plan and the critical process route in the construction organization," and pinpointing the specific chapter.

[0123] The system automatically generated a processing report, pointing out the aforementioned issues and suggesting "refining the detailed usage plan for tower crane T1 or assessing the need to add backup machinery." Based on the report, the engineer revised the construction organization document, clarifying the detailed lifting sequence for T1. The system incrementally updated the constraints of T1 in the construction knowledge graph and automatically re-analyzed the plans for the affected area, quickly obtaining a coordinated and optimized schedule.

[0124] The system integrates all results to generate a final construction plan that can be directly applied. The system outputs a final construction plan package containing: an optimized multi-level schedule (level two and level three schedules); a daily-precise plan for the hoisting, transfer, and maintenance of tower crane T1; derived time windows for the demand of key equipment such as boiler heating surfaces and large-diameter pipelines; clearly defined milestones and standards for civil engineering delivery and installation, and installation delivery and commissioning; and the aforementioned issues and optimization records (i.e., the content of the handling opinion report). The final construction plan is published to the project collaborative management platform as the sole benchmark for collaboration and dispute resolution among all parties. Simultaneously, its data is synchronized to the project BIM management platform via an interface for 4D construction simulation and to drive the procurement system to generate accurate purchase orders. A second aspect of this invention provides a computer-readable storage medium storing a program that, when executed by a processor, implements the planning method described in any of the above embodiments.

[0125] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0126] In the claims, description, and accompanying drawings of this invention, the term "plural" refers to two or more. Unless otherwise explicitly defined, the terms "upper," "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and simplifying the descriptive process, and are not intended to indicate or imply that the device or element referred to must have the described specific orientation, or be constructed and operated in a specific orientation. Therefore, these descriptions should not be construed as limiting the invention. The terms "connected," "installed," "fixed," etc., should be interpreted broadly. For example, "connected" can be a fixed connection between multiple objects, a detachable connection between multiple objects, or an integral connection; it can be a direct connection between multiple objects or an indirect connection between multiple objects through an intermediate medium. For those skilled in the art, the specific meaning of the above terms in this invention can be understood based on the specific circumstances described above.

[0127] In the claims, description, and accompanying drawings of this invention, the terms "one embodiment," "some embodiments," "specific embodiment," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of the invention. In the claims, description, and accompanying drawings of this invention, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0128] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A planning method for thermal power plant construction projects, characterized in that, include: Obtain data documents, wherein the data documents include a first data document of the thermal power construction project and a second data document of the construction contract; The data document is preprocessed to obtain parsable text units; The parsable text units are parsed to obtain the construction knowledge graph and contract constraints. Based on the construction knowledge graph and the contractual constraints, a simulation and deduction of the thermal power construction project is carried out to obtain a multi-level construction schedule. Obtain abnormal information from the multi-level schedule, analyze and process the abnormal information, and generate a processing opinion report; Based on the processing feedback report, the multi-level schedule is modified to obtain the final construction plan; The parsable text unit includes a first sub-parsable text unit and a second sub-parsable text unit; The parsing process of the parsable text units to obtain the construction knowledge graph and contract constraints includes: A text intelligent parsing model is trained based on the annotation word set and semantic rules of thermal power construction projects to obtain a thermal power text intelligent parsing model; Based on domain knowledge of thermal power plant construction projects, a domain database is constructed; The thermal power text intelligent parsing model is used to identify and extract the logical relationships between entities and different entities in the first sub-parsable text unit. The contractual constraints in the second sub-parseable text unit are identified and extracted using the thermal power text intelligent parsing model. The entities, logical relationships, and contractual constraints are instantiated as entity nodes and edges using the domain database. Based on the entity nodes and the edges, the construction knowledge graph is constructed.

2. The planning method according to claim 1, characterized in that, The preprocessing of the data document to obtain parsable text units includes: The data document is formatted to obtain the first derived text; Remove noisy data from the first derived text to obtain the second derived text; The second derived text is segmented into multiple text segments based on the preset title to obtain the third derived text; The content in the third derived text is labeled based on preset tags to obtain the parsable text unit.

3. The planning method according to claim 1, characterized in that, The text intelligent parsing model is trained based on the labeled word set and semantic rules of thermal power construction projects to obtain the thermal power text intelligent parsing model, including: Obtain the labeled term set, which includes vocabulary related to thermal power construction projects, the weight of the vocabulary, synonyms of the vocabulary, and rules for limiting the text range; Based on the labeled word set and the semantic rules, the text intelligent parsing model is trained so that the trained text intelligent parsing model can identify the entities in the first sub-parsable text unit, the logical relationships between the different entities, and the contractual constraints in the second sub-parsable text unit. The trained text intelligent parsing model is then used as the thermal power text intelligent parsing model.

4. The planning method according to claim 1, characterized in that, The domain database includes: A database of entity terms for thermal power plant construction projects, a database of construction procedure terms, and a database of schedule constraints and annotation examples for thermal power plant construction projects.

5. The planning method according to claim 1, characterized in that, The simulation and deduction of thermal power plant construction projects based on the construction knowledge graph and the contractual constraints to obtain multi-level construction schedules includes: Based on the construction knowledge graph, generate the skeleton of the project-level work breakdown structure; Based on the technological logic relationships of the construction knowledge graph and the time constraints in the contractual constraints, a time plan for the smallest task unit is generated. Based on the multi-dimensional constraints of the construction knowledge graph, the time plan is optimized to obtain an optimized time plan; The optimized time plan is summarized level by level according to the hierarchical relationship of the skeleton to generate the multi-level schedule plan.

6. The planning method according to claim 5, characterized in that, The optimization of the time plan based on the multi-dimensional constraints of the construction knowledge graph to obtain an optimized time plan includes: Identify conflicting items in the time plan that violate the multi-dimensional constraints; The time plan of the smallest task unit corresponding to the conflict item is adjusted according to the preset adjustment strategy until all the multi-dimensional constraints are met, and the adjusted time plan is used as the optimized time plan.

7. The planning method according to claim 1, characterized in that, The process of acquiring abnormal information from the multi-level schedule, analyzing and processing the abnormal information, and generating a processing opinion report includes: The multi-level schedule is validated based on the contract constraints and resource allocation to obtain the abnormal information. The abnormal information is analyzed to generate the processing opinion report.

8. The planning method according to claim 7, characterized in that, The step of validating the multi-level schedule based on the contractual constraints and resource allocation, and obtaining the anomaly information, includes: Based on the milestone nodes and the schedule requirements in the contract constraints, verify whether the completion time of each milestone node in the multi-level schedule meets the requirements, and generate schedule exception information. Based on the number of equipment, personnel, and material supply cycle in the resource configuration, determine whether there are resource conflicts or insufficient resources in the multi-level schedule and obtain the judgment result. Based on the judgment result, resource anomaly information is generated; The project schedule anomaly information and the resource anomaly information are merged to obtain the anomaly information.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the planning method according to any one of claims 1 to 8.

Images