Decoration engineering scheduling system and method based on worker digital portrait and intelligent dispatching

Abstract

The application discloses a decoration engineering scheduling system and method based on a worker digital portrait and intelligent dispatching, and belongs to the technical field of computer data processing, and comprises the following: a worker digital portrait construction module; an NLP analysis module; an intelligent matching module, which is used for converting the worker multi-dimensional digital portrait and a structured process demand label into a feature vector, calculating a matching degree score based on a cosine similarity algorithm, and obtaining a worker recommendation list; and a load balancing dispatching module, which is based on an integer programming model, takes workload equalization, construction commuting cost minimization and project duration guarantee maximization as optimization targets, generates an optimal dispatching scheme in combination with a real-time geographic position and a schedule state of the worker, pushes a dispatching task to the corresponding worker, and receives feedback. The application can solve the technical problems of low matching accuracy, unbalanced load and low scheduling efficiency in the prior art, and realizes accurate skill matching, global load balancing, improved scheduling automation level and data closed loop.

Description

Technical Field

[0001] This invention relates to the field of computer data processing technology, specifically to a decoration project scheduling system and method based on worker digital profiling and intelligent order dispatching. Background Technology

[0002] In the process of the decoration industry's transformation towards standardized management, many decoration companies have built a central dispatch system. However, due to limitations in management experience and technical capabilities, many core pain points have been exposed in actual implementation, and the dispatch efficiency and effectiveness have fallen far short of expectations.

[0003] Existing renovation project scheduling technology mainly relies on manual experience for order dispatching and management, and still has the following technical shortcomings. (1) The information dimension is singular, only including basic identity attributes and failing to construct a multi-dimensional digital profile of workers that integrates skills, performance and spatiotemporal status; (2) The project requirements are stored in unstructured text form, lacking an automatic extraction and tagging mechanism for construction process difficulties, making it difficult to quantify and analyze the core project requirements; (3) The scheduling logic only performs one-way matching based on job type and does not establish a quantitative adaptation model of worker skill characteristics and project process requirements, resulting in skill mismatch; (4) The lack of a global algorithm for the real-time geographical location, schedule saturation and workload of workers makes it difficult to achieve optimal resource allocation and load balancing; (5) The overall scheduling process relies on manual offline communication and lacks an automated order dispatch mechanism based on multi-objective optimization algorithms, resulting in low scheduling efficiency and limited rationality of resource allocation.

[0004] Therefore, there is an urgent need for a decoration project scheduling system and method based on multi-dimensional digital profiling of workers and intelligent algorithms to solve the technical problems of low matching accuracy, unbalanced load, and low scheduling efficiency in the existing technology. Summary of the Invention

[0005] In view of the above-mentioned shortcomings of the existing technology, the present invention provides a decoration project scheduling system and method based on worker digital profiles and intelligent order dispatch.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A renovation project scheduling system based on worker digital profiling and intelligent work order dispatching includes: The worker digital profile construction module (1) is used to collect and process workers' skill, performance, spatiotemporal and certification feature data to construct a multi-dimensional digital profile of workers that includes standardized process labels and quantitative scores. The NLP parsing module (2) extracts and assigns weights to the core construction difficulty keywords in the project requirement text based on the BERT pre-trained model, and generates structured process requirement tags. The worker digital profile construction module (1) and NLP parsing module (2) are both connected to the intelligent matching module (3); The intelligent matching module (3) is used to convert the multidimensional digital profile of the workers and the structured process requirement tags into feature vectors, calculate the matching score based on the cosine similarity algorithm, and perform secondary screening and sorting to obtain a list of recommended workers. The load balancing dispatch module (4) is based on an integer programming model. With the optimization goals of workload balancing, minimizing construction commuting costs, and maximizing project schedule guarantee, it generates the optimal dispatch plan by combining the real-time geographical location and schedule status of workers, and pushes the dispatch task to the corresponding worker's APP and receives feedback. The objective function is:

[0007] in, For workers i To the project j Commuting distance As a variable for order dispatching decisions, For workers i Current slot saturation For the average load of all workers, For the project j The expected completion time This is the project deadline. These are weighting coefficients, representing the priority of commuting costs, load balancing, and project schedule assurance, respectively.

[0008] Furthermore, the worker digital profile construction module (1) includes a data acquisition and preprocessing unit, a tagging processing unit, and a profile update unit.

[0009] Furthermore, the data acquisition and preprocessing unit is used to acquire raw data in real time and to perform data cleaning on the raw data; The tagging processing unit is used to convert skill characteristics into standardized process tags, and to convert certification characteristics, performance characteristics and spatiotemporal characteristics into quantitative scores through score calculation; and to integrate the standardized process tags and quantitative score results to form a multi-dimensional digital profile of the worker. The portrait update unit is used to perform real-time updates of portrait data.

[0010] Furthermore, the NLP parsing module (2) includes a text normalization unit, a BERT parsing unit, a tag conversion unit, and a manual review interface; Text standardization unit, used to standardize unstructured original project requirements text; The BERT parsing unit is used to extract core construction difficulty keywords and assign weights to the standardized text using the BERT model (natural language processing deep learning model), as well as to extract project attributes and obtain grade requirements and performance requirements; The tag conversion unit is used to generate structured tags containing process requirement tags, weight assignments, project attributes, level requirements, and performance requirements.

[0011] Furthermore, the NLP parsing module (2) also includes a manual review interface for manual review, and the structured labels after manual review are output to the project requirement model.

[0012] Furthermore, the intelligent matching module (3) performs the following steps: Transform the process tags, accumulated craftsmanship levels, and quantitative scores in the multidimensional digital profile of workers into high-dimensional worker profile feature vectors; Transform the process requirement labels and weights, project attributes, level requirements and performance requirements in the project requirement model into a project feature vector; The cosine similarity between the worker profile feature vector and the project feature vector is calculated to obtain a preliminary matching list. Then, based on the accumulated craftsmanship level and performance score, the workers are further screened and sorted to generate a recommended list of screened workers.

[0013] Furthermore, the formula for calculating the cosine similarity is:

[0014] in, A Feature vectors for creating worker profiles B For project feature vectors, n The number of feature dimensions.

[0015] Furthermore, the load balancing dispatch module (4) includes a constraint setting unit, an integer programming calculation unit, an automatic push unit, and a dynamic adjustment unit; the automatic push unit communicates with the worker's APP. The constraint setting unit is used to configure the maximum number of orders that a single worker can dispatch during the same period based on the cumulative craftsman level. The integer programming model adopts a multi-objective optimization model with the objective functions of minimizing commuting distance, balancing workload, and meeting project schedule requirements. It receives a list of recommended workers, uses a load balancing algorithm as the core calculation unit, and generates the optimal dispatching scheme in combination with constraints. The automatic push unit is used to send the dispatch tasks to the worker's terminal APP corresponding to the optimal dispatch plan according to the optimal dispatch plan. The dynamic adjustment unit is used to receive confirmation / rejection feedback from the worker's app. If the worker rejects the request, the information is fed back to the integer programming model, the worker who rejected the request is removed from the worker list, and the optimal dispatch plan is regenerated until the worker's app provides confirmation feedback.

[0016] Furthermore, it also includes a data linkage module (5) for realizing data interaction with external management systems and visual monitoring of the scheduling process; the data linkage module (5) includes a data interface unit and a visualization dashboard; the data interface unit is used to perform bidirectional data synchronization with the HR system, payroll system, quality inspection system and engineering management system through API.

[0017] To better achieve the objectives of this invention, this invention also provides a method for scheduling decoration projects based on worker digital profiles and intelligent work order dispatching, comprising the following steps: S1. Construction of worker digital profiles: Collect and process worker skill, performance, spatiotemporal and certification characteristic data to construct multi-dimensional digital profiles of workers that include standardized process labels and quantitative scores; S2. Project Requirements NLP Analysis: Based on the BERT pre-trained model, the core construction difficulty keywords of the project requirements text are extracted and weighted to generate structured process requirement tags. S3. Intelligent matching calculation: The multi-dimensional digital profile of workers and the structured process requirement tags are transformed into feature vectors. The matching score is calculated based on the cosine similarity algorithm, and a secondary screening and sorting is performed based on the cumulative craftsman level and performance score to generate a worker recommendation list. S4. Load Balancing Dispatch: Based on an integer programming model, with the optimization goals of workload balancing, minimizing construction commuting costs, and maximizing project schedule assurance, the system generates the optimal dispatch plan by combining the real-time geographical location and availability of workers; the dispatched tasks are pushed to the worker's mobile app, feedback is received, and data from other related systems is updated synchronously via API.

[0018] Compared with the prior art, the beneficial effects of this invention are as follows: (1) This invention can achieve precise skill matching: by constructing a digital profile of workers with four dimensions and analyzing project requirements based on BERT, the cosine similarity algorithm is used to upgrade from "job matching" to "ability matching", which solves the problem of skill mismatch from the root. It can increase the first-time acceptance rate of construction by more than 25% and reduce the rework rate by more than 30%.

[0019] (2) This invention can achieve global load balancing: by using a multi-objective optimization algorithm to coordinate the spatiotemporal characteristics of workers with the overall employment situation, it solves the industry pain point of uneven workload of workers, ensures reasonable employment of core craftsmen and saturation of workload of ordinary workers, and can reduce the turnover rate of core craftsmen by more than 40%.

[0020] (3) This invention can improve the level of scheduling automation: it transforms the traditional offline scheduling that relies on human experience into a quantifiable and automated intelligent algorithm, realizing a fully automated process from project requirement analysis to task assignment confirmation, and improving scheduling efficiency by more than 90%.

[0021] (4) This invention can form a data closed loop: realize the full data linkage between the scheduling system and the existing management system of the decoration enterprise, construct a full-process closed loop of "profile construction - demand analysis - intelligent matching - load dispatching - construction assessment - profile update", and form a scheduling management system that can be monitored, traced and iterated. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0023] Figure 1 This is a schematic diagram of the architecture of the decoration project scheduling system based on worker digital profiling and intelligent order dispatching of the present invention; Figure 2 This is a flowchart illustrating the construction process of the worker digital profile in the decoration project scheduling system based on worker digital profiles and intelligent dispatching, as described in this invention. Figure 3 This is a flowchart of the project requirements NLP analysis for the decoration project scheduling system based on worker digital profiling and intelligent order dispatching, as presented in this invention. Figure 4 This is a flowchart illustrating the intelligent matching and dispatching process of the decoration project scheduling system based on worker digital profiles and intelligent dispatching, as presented in this invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0025] Example 1: Please refer to the accompanying drawings in the instruction manual. Figures 1-4 A renovation project scheduling system based on worker digital profiles and intelligent order dispatching includes: Worker Digital Profile Construction Module 1 is used to collect and process workers' skill, performance, spatiotemporal and certification feature data to construct a multi-dimensional digital profile of workers that includes standardized process labels and quantitative scores. The worker digital profile construction module 1 includes a data acquisition and preprocessing unit, a tagging processing unit, and a profile updating unit.

[0026] The data acquisition and preprocessing unit is used to connect with the engineering management system, quality inspection system, HR (human resources) system, payroll system, customer feedback system, construction assessment system and worker-side APP to acquire raw data in real time; and to perform data cleaning on the raw data; data cleaning specifically includes deduplication, error correction and formatting of the raw data.

[0027] Skills characteristic data includes job type, process expertise, number of successful processes and total number of process assignments, project experience, etc. Performance characteristic data include first-time acceptance rate, rework rate, schedule compliance rate, and customer satisfaction rate; Spatiotemporal characteristic data includes real-time geographic location, availability of slots, and available construction periods; Certification feature data includes artisan level and professional qualification certificate; The tagging processing unit is used to convert skill characteristics into standardized process tags, and to convert certification characteristics, performance characteristics and spatiotemporal characteristics into quantitative scores through score calculation; and to integrate the standardized process tags and quantitative score results to form a multi-dimensional digital profile (dynamic profile) of the worker. The quantification method for the certification features is as follows: craftsman level is mandatory, while certificate level is optional. If both craftsman level and certificate level are available, a weighted arithmetic average is used to calculate the comprehensive certification score based on a mapping table between craftsman level and certificate level to the [0,1] interval. The cumulative craftsman level and its corresponding mapping value are then obtained based on the comprehensive certification score. If only craftsman level is available, the craftsman level and its corresponding mapping value are used as the cumulative craftsman level and its corresponding mapping value. All mapping relationships are configurable parameters of the system, allowing enterprises to adjust them according to their own job level system.

[0028] I. Common artisan skill levels are: Junior, Intermediate, Senior, Technician, and Senior Technician. The mapping table from artisan skill level to the [0,1] interval is as follows:

[0029] II. Professional qualification certificates (such as the "Professional Qualification Certificate" issued by the Human Resources and Social Security Department) are also divided into five levels: Level 5 / Elementary, Level 4 / Intermediate, Level 3 / Advanced, Level 2 / Technician, and Level 1 / Senior Technician. The mapping table from certificate level to the [0,1] interval is as follows:

[0030] III. The overall certification score is calculated using a weighted average strategy:

[0031] Usually taken That is, equal-weighted average. The numerical values ​​are mapped to the artisan level. This is a numerical value mapped to the certificate level.

[0032] A mapping between cumulative craftsman levels and their corresponding values ​​is established based on the comprehensive certification score. The mapping table from the comprehensive certification score to the [0,1] interval is as follows:

[0033] The image update unit is used to update the image data in real time according to construction nodes or daily based on an incremental update mechanism.

[0034] The standardized process labels include process labels for different types of decoration work such as complex diagonal tiling, smart home wiring, underfloor heating leveling, and diamond-cut craftsmanship, as well as process skill scores calculated by the number of successful processes / the total number of process orders; the quantitative scores include cumulative craftsman level mapping values, performance scores, spatiotemporal feature scores, etc.; among them, the performance feature data (such as first-time acceptance rate, customer satisfaction rate) are weighted and normalized to be converted into quantitative scores (performance scores) in the range of [0, 1], which are used for subsequent similarity calculations.

[0035] NLP parsing module 2 extracts and assigns weights to key construction difficulty keywords from the project requirement text based on the BERT pre-trained model, generating structured process requirement tags. The NLP parsing module 2 includes a text standardization unit, a BERT parsing unit, a tag conversion unit, and a manual review interface. The BERT parsing unit is used to extract core construction difficulty keywords through the BERT model and assign weights to them. The weight of core construction difficulty is set to 80%, and the weight of conventional process is set to 20%. The weights are configurable.

[0036] The text standardization unit is used to standardize unstructured original project requirement text (referred to as unstructured text, such as "tile laying throughout the house, floor heating leveling in the bathroom"), specifically by preprocessing the input text such as cleaning, word segmentation, and removal of stop words. The BERT parsing unit is used to extract core construction difficulty keywords and assign weights to the standardized text using the BERT model, as well as to extract project attributes and obtain grade requirements and performance requirements. The BERT model uses a pre-trained model fine-tuned based on industry corpus to extract core construction difficulty keywords and assign weights, with the weight of core construction difficulty set at 80% and the weight of conventional processes set at 20%, and the weights are configurable.

[0037] The BERT model is trained through the following steps: 1) Corpus Construction: Collect texts such as decoration project requirement documents, technical briefing records, and process standards, and form an industry corpus after anonymization. 2) Pre-trained Model Loading: Load a general Chinese BERT pre-trained model (such as bert-base-chinese) as initial parameters. 3) Domain-Adaptive Fine-tuning: Continue self-supervised training on the decoration corpus using the Masked Language Model (MLM) task, enabling the model to learn the vocabulary distribution and semantic features of the decoration domain. 4) Task Adaptation and Weight Assignment: Add a difficulty classification header to the top of the fine-tuned model, and conduct supervised training using manually labeled process keywords and difficulty levels (core construction difficulties / conventional processes). While the model outputs keywords, it automatically assigns preset weights based on their difficulty category (e.g., 80% for core construction difficulties, 20% for conventional processes), which are configurable parameters. 5) Evaluation and Deployment: After verifying that the extraction accuracy meets the requirements on the test set, it is packaged as an API service for NLP parsing units to call. Through the above training, the BERT model can accurately extract core construction difficulty keywords from project requirement texts and assign weights accordingly.

[0038] In addition, the NLP parsing module also includes an auxiliary extraction unit that combines rules with NER to independently extract the following structured project attributes from the same requirement text: Construction location: converted to latitude and longitude or administrative division code through the address parsing interface; Schedule requirement: extracted in days through regular expressions and calculated as the deadline based on the start date; Project priority: identified keywords such as "urgent" and "top priority" and mapped to priority scores; Other constraints: such as "workdays only" and "noise-free periods", used for constraint filtering in subsequent order assignments; The above attributes and process requirement tags are stored together in the project requirement model and used for worker skill matching, commuting cost calculation, schedule guarantee optimization and constraint judgment, as well as project budget, customer level, project type, etc.

[0039] The system obtains level and performance requirements based on project attributes. Accumulated craftsman level and performance requirements are obtained as follows: The system automatically generates default requirement values ​​based on a preset rule mapping table, using attributes such as project budget, client level, and project type. These values ​​can be adjusted via a manual review interface. All requirement values ​​are normalized to the [0,1] range. The mapping relationship is as follows:

[0040] The tag conversion unit is used to generate structured tags containing process requirement tags, weight assignments, project attributes, level requirements, and performance requirements, such as {large brick paving: 0.8, underfloor heating leveling: 0.8, construction location: XX Community, Chaoyang District, Beijing, construction period requirement: 15 days, project priority: urgent, ..., cumulative craftsman level requirement: 0.8, performance requirement: 0.8}.

[0041] An optional manual review interface is provided for manual review, including correcting or supplementing the automatically parsed results. The manually reviewed structured tags are then output to the project requirement model.

[0042] The project requirements model is a structured data model of project requirements. It is a standardized data structure (JSON format) that includes process requirement tags, weight assignments, project attributes, level requirements, and performance requirements.

[0043] The worker digital profile construction module 1 and the NLP parsing module 2 are both connected to the intelligent matching module 3; The intelligent matching module 3 is used to convert the multi-dimensional digital profile of the worker and the structured process requirement tags into feature vectors, and to make the two dimensions correspond one-to-one. The matching score is calculated based on the cosine similarity algorithm, and the cumulative craftsman level mapping value and performance score are combined for secondary screening and sorting to obtain a worker recommendation list. The intelligent matching module 3 performs the following steps: The process tags and quantitative scores in the multidimensional digital profile of workers are transformed into high-dimensional worker profile feature vectors; specifically, these include: process skills and process skill scores corresponding to project requirements, cumulative craftsman level mapping values, and performance scores.

[0044] For example: Worker feature vector = [0.8 (large brick laying), 0.8 (underfloor heating leveling), 0.9 (comprehensive certification score), 0.95 (performance score)]; Transform the process requirement labels and weights, project attributes, level requirements, and performance requirements in the project requirement model into a project feature vector; including: process requirement labels and weights, level requirements, and performance requirements.

[0045] For example: Project feature vector = [0.8 (large brick paving requirements), 0.8 (underfloor heating leveling requirements), 0.8 (grade requirements), 0.8 (performance requirements)].

[0046] The cosine similarity between the worker profile feature vector and the project feature vector is calculated to obtain a preliminary matching list. For example, workers with a matching score of ≥85 are included in the preliminary matching list. The formula for calculating cosine similarity is:

[0047] in, A Feature vectors for creating worker profiles B For project feature vectors, n is the number of feature dimensions, and i is the dimension index.

[0048] Then, based on the accumulated craftsman level mapping value and performance score, the workers are further filtered and sorted to generate a recommended list of filtered workers.

[0049] The secondary screening and sorting process employs a hierarchical sorting method: First, all workers meeting the matching criteria are grouped from highest to lowest based on their cumulative craftsmanship level mapping value (Senior Technician, Technician, Senior Worker, Intermediate Worker, Junior Worker); then, within each level group, they are sorted from highest to lowest based on their comprehensive performance score; finally, the groups are concatenated in level order to generate a worker recommendation list. For new workers with n < 3 construction attempts, their performance score is multiplied by [the factor needed for the next step]. This reduces the initial recommendation priority. This recommendation list serves as the input candidate set for the load balancing dispatch module.

[0050] This invention uses cosine similarity as an initial screening method, and then refines the recommended list through hierarchical sorting based on cumulative craftsmanship level and performance score.

[0051] The load balancing dispatch module 4 receives multi-dimensional digital profiles of workers and structured process requirement tags. Based on an integer programming model, it generates the optimal dispatch plan by combining the real-time geographical location and availability of workers with the optimization goals of workload balancing, minimizing construction commuting costs, and maximizing project schedule assurance. The dispatch tasks are then pushed to the corresponding worker's mobile app. The module also receives feedback and updates data from other related systems via API. The load balancing dispatch module 4 includes a constraint setting unit, an integer programming calculation unit, an automatic push unit, and a dynamic adjustment unit; the automatic push unit communicates with the worker's APP. The constraint setting unit is used to configure the maximum number of orders that a single worker can dispatch during the same period based on the cumulative craftsman level. The integer programming model adopts a multi-objective optimization model with the objective functions of minimizing commuting distance, balancing workload, and meeting project schedule requirements. It receives a list of recommended workers, uses a load balancing algorithm as the core calculation unit, and generates the optimal dispatching scheme in combination with constraints. The objective function of the integer programming model is:

[0052] in, For workers i To the project j Commuting distance As a variable for order dispatching decisions, For workers i Current slot saturation For the average load of all workers, For the project j The expected completion time This is the project deadline. These are weighting coefficients, representing the priority of commuting costs, load balancing, and project schedule assurance, respectively, and are all positive real numbers.

[0053] In actual calculations, to avoid the influence of different dimensions on the optimization results, the system first performs dimensionless normalization on the three components of the objective function: Commuting distance: Calculates the maximum commuting distance among the current candidate dispatching schemes. ,make This indicates the relative commuting distance; Load balancing item: Calculates the maximum value of the sum of squared load deviations among the current candidate dispatch schemes. (Right now (the maximum value that can be obtained), let This indicates the degree of relative load imbalance. Project Delay Item: Calculates the maximum number of days of project delay among the current candidate work order allocation schemes. ,make This indicates a relative delay in the construction period.

[0054] The maximum values ​​mentioned above can all be obtained by traversing the current candidate dispatch schemes or by statistical analysis of historical data. After normalization, all three components become dimensionless relative values ​​(their values ​​all range from...). (between), in this case, weighted summation is possible. The objective function is equivalent to:

[0055] in These are dimensionless weighting coefficients, representing the degree of importance the company attaches to the three optimization objectives, satisfying... .

[0056] Order dispatch decision variables Let be a 0-1 variable, representing whether worker i is assigned to project j (1 indicates assignment, 0 indicates no assignment). The solver finds the optimal combination of values ​​that minimizes the objective function under the following constraints: each project is assigned to exactly one worker of that type; the number of concurrent tasks for each worker does not exceed a preset threshold; and worker skills must match project requirements.

[0057] For example, the system detected that bricklayer Zhang's current schedule saturation is 60% (i.e., (), lower than the current average workload of all workers Its current geographical location is 8 kilometers away from the project site (i.e., This indicates that Mr. Zhang has sufficient availability to ensure the project is completed on schedule; therefore, the project schedule guarantee item is valid. .

[0058] The automatic push unit is used to push the dispatch tasks to the worker's end APP corresponding to the optimal dispatch plan through the APP, based on the optimal dispatch plan. The dynamic adjustment unit is used to receive confirmation / rejection feedback from the worker's app. If the worker rejects the work, the information is fed back to the integer programming model, the worker who rejected the work is removed from the worker list, and the optimal dispatch plan is regenerated until the worker's app provides confirmation feedback. Data linkage module 5 is used to realize data interaction with external management systems and visualize the scheduling process monitoring.

[0059] The data linkage module 5 includes a data interface unit and a visualization dashboard; the data interface unit is used to perform bidirectional data synchronization with HR human resources system, payroll system, quality inspection system, engineering management system, etc. via API.

[0060] A visual dashboard is a monitoring interface for centrally displaying scheduling data, and it consists of four core components: Regional employment heat map: Shows the distribution of workers and employment density in each region.

[0061] Worker availability Gantt chart: Visualizes the worker availability saturation and task timeline.

[0062] Project progress dashboard: Track the construction progress and key milestones of each project.

[0063] Real-time matching quality monitoring: Monitors the matching quality indicators of the current order.

[0064] The HR (Human Resources) system synchronizes information such as worker files, craftsman levels, professional qualification certificates, accumulated craftsman levels, technical expertise, project experience, customer satisfaction rates, real-time geographical location, availability, and available construction periods. The payroll system synchronizes working hours and piece-rate data for salary calculation. The quality inspection system manages quality acceptance, records and manages the acceptance results (pass / fail) and rework status of each construction node, and calculates workers' first-time acceptance pass rate, rework rate, and customer evaluation scores. The project management system manages project progress and tasks, records basic project information (project name, address, duration, etc.), tracks the actual start and finish times of each construction node, calculates schedule compliance rate, calculates material loss rate (ratio of requisitioned materials to actual usage), generates on-site safety and civilized construction scores (submitted by inspectors), and manages the task association between workers and projects (dispatch records, attendance status). The customer feedback system receives customer evaluation scores (1-5 stars converted to 0-100 points).

[0065] The construction assessment system is used to assess projects and automatically obtains the following assessment indicators: first-time acceptance rate (obtained from the quality inspection system); schedule compliance rate, material loss rate, and on-site safety and civilized construction score (obtained from the project management system); customer evaluation score (obtained from the customer feedback system). After normalizing the above indicators to the [0,1] range, they are weighted and summed according to preset weights (e.g., first-time acceptance rate 40%, (1-schedule compliance rate) 20%, (1-material loss rate) 10%, safety and civilized construction score 10%, and customer evaluation score 20%) to obtain the comprehensive performance score (out of 100) of this construction project.

[0066] The system automatically compares the process tag required for the project when assigning the work order (such as "large brick laying") with the process type actually completed by the worker during construction. If the acceptance record or construction log confirms that the worker has successfully completed the process, the system adds one "successful practice" count to the worker's corresponding skill tag and updates their confidence score for that process (e.g., number of successful processes / total number of work orders assigned for that process). If rework or substandard processes are caused by insufficient skills, the confidence score is lowered accordingly.

[0067] Example 2: A renovation project scheduling method based on worker digital profiling and intelligent work order dispatching, comprising the following steps: S1. Construction of worker digital profiles: Collect and process worker skill, performance, spatiotemporal and certification characteristic data to construct multi-dimensional digital profiles of workers that include standardized process labels and quantitative scores; S2. Project Requirements NLP Analysis: Based on the BERT pre-trained model, the core construction difficulty keywords of the project requirements text are extracted and weighted to generate structured process requirement tags. S3. Intelligent matching calculation: The multi-dimensional digital profile of workers and the structured process requirement tags are transformed into feature vectors. The matching score is calculated based on the cosine similarity algorithm, and a secondary screening and sorting is performed based on the cumulative craftsman level and performance score to generate a worker recommendation list. S4. Load Balancing Dispatch: Based on an integer programming model, with the optimization goals of workload balancing, minimizing construction commuting costs, and maximizing project schedule assurance, the system generates the optimal dispatch plan by combining the real-time geographical location and availability of workers; the dispatched tasks are pushed to the worker's mobile app, feedback is received, and data from other related systems is updated synchronously via API.

[0068] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A renovation project scheduling system based on worker digital profiling and intelligent task dispatching, characterized in that: include: The worker digital profile construction module (1) is used to collect and process workers' skill, performance, spatiotemporal and certification feature data to construct a multi-dimensional digital profile of workers that includes standardized process labels and quantitative scores. The NLP parsing module (2) extracts and assigns weights to the core construction difficulty keywords in the project requirement text based on the BERT pre-trained model, and generates structured process requirement tags. The worker digital profile construction module (1) and NLP parsing module (2) are both connected to the intelligent matching module (3); The intelligent matching module (3) is used to convert the multidimensional digital profile of the workers and the structured process requirement tags into feature vectors, calculate the matching score based on the cosine similarity algorithm, and perform secondary screening and sorting to obtain a list of recommended workers. The load balancing dispatch module (4) is based on an integer programming model. With the optimization goals of workload balancing, minimizing construction commuting costs, and maximizing project schedule guarantee, it generates the optimal dispatch plan by combining the real-time geographical location and schedule status of workers, and pushes the dispatch task to the corresponding worker's APP and receives feedback. The objective function is: ； in, Let i be the commuting distance from worker i to project j. As a variable for order dispatching decisions, Let i be the current availability saturation level. For the average load of all workers, Let be the expected completion time for project j. This is the project deadline. These are weighting coefficients, representing the priority of commuting costs, load balancing, and project schedule assurance, respectively.

2. The decoration project scheduling system based on worker digital profiling and intelligent dispatching as described in claim 1, characterized in that, The worker digital profile construction module (1) includes a data acquisition and preprocessing unit, a tagging processing unit, and a profile update unit.

3. The decoration project scheduling system based on worker digital profiling and intelligent dispatching as described in claim 2, characterized in that, The data acquisition and preprocessing unit is used to acquire raw data in real time and to perform data cleaning on the raw data. The tagging processing unit is used to convert skill characteristics into standardized process tags, and to convert certification characteristics, performance characteristics and spatiotemporal characteristics into quantitative scores through score calculation; and to integrate the standardized process tags and quantitative score results to form a multi-dimensional digital profile of the worker. The portrait update unit is used to perform real-time updates of portrait data.

4. The decoration project scheduling system based on worker digital profiling and intelligent dispatching as described in claim 3, characterized in that, The NLP parsing module (2) includes a text normalization unit, a BERT parsing unit, a tag conversion unit, and a manual review interface; Text standardization unit, used to standardize unstructured original project requirements text; The BERT parsing unit is used to extract core construction difficulty keywords and assign weights to the standardized text using the BERT model, as well as extract project attributes and obtain grade requirements and performance requirements. The tag conversion unit is used to generate structured tags containing process requirement tags, weight assignments, project attributes, level requirements, and performance requirements.

5. The decoration project scheduling system based on worker digital profiling and intelligent dispatching as described in claim 4, characterized in that, The NLP parsing module (2) also includes a manual review interface for manual review, and the structured labels after manual review are output to the project requirement model.

6. The decoration project scheduling system based on worker digital profiling and intelligent dispatching as described in claim 5, characterized in that, The intelligent matching module (3) performs the following steps: Transform the process tags, accumulated craftsmanship levels, and quantitative scores in the multidimensional digital profile of workers into high-dimensional worker profile feature vectors; Transform the process requirement labels and weights, project attributes, level requirements and performance requirements in the project requirement model into a project feature vector; The cosine similarity between the worker profile feature vector and the project feature vector is calculated to obtain a preliminary matching list. Then, based on the accumulated craftsmanship level and performance score, the workers are further screened and sorted to generate a recommended list of screened workers.

7. The decoration project scheduling system based on worker digital profiling and intelligent dispatching as described in claim 6, characterized in that, The formula for calculating cosine similarity is: ； Where A is the worker profile feature vector, B is the project feature vector, and n is the number of feature dimensions.

8. The decoration project scheduling system based on worker digital profiling and intelligent dispatching as described in claim 7, characterized in that, The load balancing dispatch module (4) includes a constraint setting unit, an integer programming calculation unit, an automatic push unit, and a dynamic adjustment unit; the automatic push unit communicates with the worker's APP. The constraint setting unit is used to configure the maximum number of orders that a single worker can dispatch during the same period based on the cumulative craftsman level. The integer programming model adopts a multi-objective optimization model with the objective functions of minimizing commuting distance, balancing workload, and meeting project schedule requirements. It receives a list of recommended workers, uses a load balancing algorithm as the core calculation unit, and generates the optimal dispatching scheme in combination with constraints. The automatic push unit is used to send the dispatch tasks to the worker's APP corresponding to the optimal dispatch plan according to the optimal dispatch plan. The dynamic adjustment unit is used to receive confirmation / rejection feedback from the worker's app. If the worker rejects the request, the information is fed back to the integer programming model, the worker who rejected the request is removed from the worker list, and the optimal dispatch plan is regenerated until the worker's app provides confirmation feedback.

9. The decoration project scheduling system based on worker digital profiling and intelligent dispatching as described in claim 8, characterized in that, It also includes a data linkage module (5) for realizing data interaction with external management systems and visual monitoring of the scheduling process; the data linkage module (5) includes a data interface unit and a visualization dashboard; the data interface unit is used to perform bidirectional data synchronization with the HR system, payroll system, quality inspection system and engineering management system through API; 10. A method for scheduling decoration projects based on worker digital profiling and intelligent work order dispatching, utilizing the decoration project scheduling system based on worker digital profiling and intelligent work order dispatching as described in claim 9, characterized in that, Includes the following steps: S1. Construction of worker digital profiles: Collect and process worker skill, performance, spatiotemporal and certification characteristic data to construct multi-dimensional digital profiles of workers that include standardized process labels and quantitative scores; S2. Project Requirements NLP Analysis: Based on the BERT pre-trained model, the core construction difficulty keywords of the project requirements text are extracted and weighted to generate structured process requirement tags. S3. Intelligent matching calculation: The multi-dimensional digital profile of workers and the structured process requirement tags are transformed into feature vectors. The matching score is calculated based on the cosine similarity algorithm, and a secondary screening and sorting is performed based on the cumulative craftsman level and performance score to generate a worker recommendation list. S4 Load Balancing Dispatch: Based on an integer programming model, with the optimization goals of workload balancing, minimizing construction commuting costs, and maximizing project schedule assurance, the optimal dispatch plan is generated by combining the real-time geographical location and availability of workers. The task assignment is pushed to the worker's mobile app, feedback is received, and data from other related systems is updated synchronously via API.

Images