Work order intelligent distribution method and device for work personnel scheduling in property scene
By identifying property work orders using a vector model and combining material storage and personnel location data, the optimal resources are selected, solving the inefficiency problem of work order allocation in the property management system and achieving efficient and intelligent resource scheduling.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- RUNYING PROPERTY TECHNOLOGY SERVICES CO LTD
- Filing Date
- 2026-04-24
- Publication Date
- 2026-07-21
AI Technical Summary
In existing property management systems, the information submitted by owners or tenants is usually conversational and unstructured. Current technology relies on human agents to read and understand the information, ignoring key dimensions such as the real-time location of the agent, current workload, tools held, and the inventory and location of required materials. This makes it impossible to allocate property service work orders in a way that minimizes costs and maximizes response time.
A vector model is used to identify the work order input information, generate a list of material requirements and personnel qualification requirements, and select the optimal designated warehouse and disposal personnel based on the material storage situation and personnel location to form a property service work order.
It achieves accurate semantic recognition of spoken work orders, improves information processing efficiency, and selects the optimal resources through multi-dimensional screening, solving the problems of resource waste and response delay in traditional scheduling, and realizing intelligent collaborative scheduling with the lowest overall cost and fastest response speed.
Smart Images
Figure CN122434162A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of property service technology, and in particular to a method and device for intelligent work order allocation for dispatching workers in property scenarios. Background Technology
[0002] Property management systems aim to achieve intelligent, standardized, and efficient property management, thereby reducing operating costs and improving service quality. Existing property management systems have the following technical problems: 1. The information submitted by owners or tenants through the property management system is usually in conversational and unstructured natural language. Current technology heavily relies on human agents to read and understand this information. 2. Simply considering dimensions like "task type" and "approximate personnel skills" ignores crucial factors such as personnel's real-time location, current workload, tools held, required material inventory and location, and service area priority. Furthermore, it fails to calculate the property service work order with the lowest overall cost and fastest response time among multiple available resources. For example, a worker from area A might be dispatched to area D, while a closer worker from area C is available. There's also the inability to coordinate personnel and material scheduling, leading to inefficient scenarios like "personnel arrive but materials don't" or "people running back and forth to retrieve materials." Summary of the Invention
[0003] To overcome the shortcomings of existing technical solutions, this invention provides a method and device for intelligent work order allocation in property management scenarios for scheduling workers.
[0004] The technical solution adopted by this invention to solve its technical problem is: In a first aspect, the present invention provides a method for intelligent work order allocation for dispatching workers in a property management scenario. The method is applied to a property management system, which is connected to a caller terminal via signal transmission. The method includes: Receive work order input information sent from the calling terminal; The work order input information is identified by a preset vector model to obtain a list of material requirements and personnel qualification requirements, and corresponding disposal step suggestions are generated based on the work order input information. Based on the material demand list, query the material storage status of each warehouse and select the optimal designated warehouse; Based on the aforementioned personnel qualification requirements, multiple qualified operators will be screened to obtain the optimal personnel for handling the situation. The proposed disposal steps, the designated warehouse, and the disposal personnel are integrated to form a property service work order; The property service work order is sent to the calling terminal.
[0005] Secondly, the present invention also provides a work order intelligent dispatching device for dispatching workers in a property management scenario. The dispatching device is installed in a property management system, which is connected to the caller terminal via signal transmission. The work order intelligent dispatching method includes: The receiving module is used to receive work order input information sent from the calling terminal; The identification module is used to identify the work order input information through a preset vector model to obtain a list of material requirements and personnel qualification requirements, and to generate corresponding handling step suggestions based on the work order input information. The query module is used to query the material storage status of each warehouse according to the material demand list and select the optimal designated warehouse. The screening module is used to screen multiple qualified operators based on the personnel qualification requirements in order to obtain the optimal personnel for handling the situation. The integration module is used to integrate the proposed disposal steps, the designated warehouse, and the disposal personnel to form a property service work order; The sending module is used to send the property service work order to the calling terminal.
[0006] Thirdly, the present invention provides a computer device, the computer device including a processor, a network interface, a memory and a communication bus, wherein the processor, the network interface and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When the processor executes the program stored in the memory, it implements the steps of the intelligent work order allocation method for scheduling workers in a property management scenario described in the first aspect above.
[0007] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the intelligent work order allocation method for scheduling workers in a property management scenario described in the first aspect above.
[0008] Compared with the prior art, the beneficial effects of the present invention are: By using vector models to perform semantic recognition and automatic classification of colloquial and unstructured work order descriptions, resources and personnel needs can be accurately extracted, replacing human agents' reading comprehension and significantly improving information processing efficiency.
[0009] The scheduling process comprehensively considers the real-time location of personnel, current load, and warehouse material storage status. Through multi-dimensional filtering and matching degree calculation, the optimal designated warehouse and disposal personnel are accurately selected, and the two are integrated collaboratively. This solves the inefficiencies of traditional scheduling, such as cross-regional order dispatch, resource waste, and "personnel and materials not being synchronized," which are caused by ignoring real-time status. It achieves intelligent collaborative scheduling with the lowest overall cost and the fastest response speed. Attached Figure Description
[0010] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0011] Figure 1 This is a flowchart of a work order intelligent allocation method for dispatching workers in a property management scenario, according to an embodiment of the present invention.
[0012] Figure 2 This is a schematic block diagram of an intelligent work order dispatching device for dispatching workers in a property management scenario, according to an embodiment of the present invention.
[0013] Figure 3 This is a schematic block diagram of a computer device according to an embodiment of the present invention. Detailed Implementation
[0014] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0015] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.
[0016] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0017] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.
[0018] To address the shortcomings of existing technologies, which rely heavily on human agents to read and understand service requests submitted by property owners or tenants through property management systems (which are typically in conversational, unstructured natural language), and which only consider one dimension (task type and approximate personnel skills) while neglecting other crucial factors such as personnel's real-time location, current workload, tools held, required material inventory and location, and service area priority, this invention provides an intelligent work order allocation method for scheduling personnel in property management scenarios.
[0019] The following details the specific steps of a work order intelligent allocation method for dispatching workers in a property management scenario, provided by an embodiment of the present invention. (See attached figure.) Figure 1 As shown, the specific steps of this intelligent work order allocation method include the following: In step S110, work order input information sent from the calling terminal is received.
[0020] Specifically, a predefined communication channel is maintained between the property management system and the calling terminals, such as resident apps, mini-programs, property staff handheld terminals, smart voice devices, or IoT sensors. For example, when a user clicks the "Submit" repair request button on the calling terminal, or when a sensor automatically triggers an alarm signal, the calling terminal sends a network request to the server.
[0021] Signal transmission is not just about transmitting a single sentence, but rather a data packet containing a wealth of information. Therefore, the system needs to break down and parse this data packet. For example, when exchanging data using JSON or XML format, the system parses the data packet content into two types of key information: user ID, contact information, room number, project ID, specific building location, GPS coordinates, time of occurrence, work order urgency level, appointment time, etc. Finally, it also needs to consider the specific problem description entered by the user.
[0022] To meet the requirements of subsequent vector models, the received raw data needs to be cleaned and standardized. That is, operations such as removing special characters, converting traditional and simplified Chinese, and preliminarily correcting spelling mistakes are performed on the input text. Moreover, the possible data format differences from different sources (iOS, Android, Web) need to be unified. For example, some terminals transmit "building", while others transmit "block", and the system maps them to a unified field standard in this step; if the calling terminal only transmits the "location ID", the system may automatically query the database in this step to supplement the detailed text address corresponding to this ID to facilitate the understanding of the subsequent large model.
[0023] In this way, through the standard communication protocol, multi-dimensional heterogeneous data including "text, pictures, locations, identities" are collected safely and completely, and are converted into standard digital objects that can be recognized, processed, and circulated by computer systems, laying a data foundation for subsequent AI semantic understanding.
[0024] In step S120, the work order input information is identified through a preset vector model to obtain a material requirements list and personnel qualification requirements, and corresponding disposal step suggestions are generated according to the work order input information.
[0025] Specifically, the system receives the work order input information transmitted from the previous step and extracts the core text, such as the repair description: "The sewer in the washroom on the second floor of Building 3 is blocked and overflowing". Using a specific splicing strategy, the key structured information (such as the location "Building 3" and the category name "facility repair") is inserted as keywords into the description text to form a complete context string, and the processed text is input into a preset text embedding model. This vector model can be understood as an Embedding model such as BGE-M3. Through such models, the text is mapped to a high-dimensional vector space, such as a 1024-dimensional vector embevent. In this space, words or sentences with similar semantics will have very close vectors in terms of distance. For example, the distance between the vectors of "sewer blockage" and "pipe dredging" is much smaller than the distance from the vector of "light bulb damage".
[0026] The system uses vector operations to "search" for templates in the SOP (Standard Operating Procedure) database and directly extract pre-defined requirements. Specifically, the system calculates the cosine similarity between the generated work order vector and all pre-stored standard work order type vectors in the SOP database. This cosine similarity measures the directional consistency between two vectors; the closer the value is to 1, the better the semantic match. The system selects the SOP entry with the highest similarity score as the current work order category. For example, if the input is identified as a pipe cleaning work order, once a SOP category is matched, the system automatically reads the pre-configured structured fields in that SOP entry, thus directly obtaining: 1. A list of material requirements, i.e., the tools and materials (such as pipe cleaners, anti-slip gloves, and pipe cleaners) required for this type of work order from the SOP; 2. Personnel qualification requirements, i.e., the skill tags or certificates required for this type of work order from the SOP.
[0027] By leveraging the logic of classification as attributes, there is no need to regenerate the material list from a large model. Instead, through precise vector classification and the attachment of standard business rules, the accuracy and standardization of extraction are guaranteed.
[0028] Finally, leveraging the capabilities of large-scale models, the system generates personalized operational guidance for specific scenarios. Using the original work order description text as query terms, the system performs a semantic search within the property work order handling process knowledge base to identify the most relevant knowledge fragments for the current work order. For example, it retrieves emergency handling documents for flooded bathroom drains, combines the original work order description with the retrieved relevant knowledge fragments to form prompts, and sends them to the large-scale language model. The large-scale language model then uses the learned knowledge and the retrieved context to perform reasoning and text generation, such as instructing the user to go to the site to shut off the main water valve for that floor, use a drain cleaning machine for initial unblocking, and if unblocking fails, notify the engineering department for pipe dismantling and inspection.
[0029] It should be noted that in certain complex scenarios or those not covered by the Standard Operating Procedure (SOP), the system will activate alternative paths, operating on a slightly different principle. For example, instead of relying on the pre-defined material list in the SOP, the system directly sends the work order description and classification results to the large model via prompts, requesting the model to generate a suggested list of tools and materials based on the retrieved knowledge. This approach is more flexible and suitable for non-standardized, unexpected work orders.
[0030] Therefore, it can be seen that by using the efficient semantic retrieval of the vector model, the work order category can be accurately identified, thereby quickly and standardly extracting the hard requirements for materials and personnel. Through the generation capability of the large model, combined with the specific context, the disposal steps are generated, realizing the automation of resource scheduling and providing intelligent guidance for on-site operations.
[0031] In step S130, the material storage status of each warehouse is queried according to the material demand list, and the optimal designated warehouse is selected.
[0032] Specifically, the system obtains the material demand list generated in the previous step and the project ID to which the work order belongs. Instead of conducting a random search in all warehouses of the entire property management company, the system first isolates them geographically or administratively based on the "project ID". The system then queries the database for all warehouses associated with the project (such as: project center main warehouse, A area gate warehouse, B area basement warehouse, etc.) to form a candidate warehouse list. Based on the principle of geographical proximity, the system prioritizes inventory resources within the project to reduce logistics costs and time for cross-project transfers.
[0033] The system retrieves real-time material storage information for each candidate warehouse via API or database queries. The retrieved data includes material name, specifications, inventory quantity, and storage location. The system then constructs a double loop: the outer loop iterates through each item in the demand material list, while the inner loop iterates through each item in the current warehouse inventory list. For each demand item and inventory item combination, the system calculates the text similarity between its name and key description. The calculation method can use edit distance, Jaccard similarity coefficient, or string semantic similarity algorithms. By setting a threshold, if the similarity calculation result is ≥ the threshold%, the system considers the material in the inventory to meet the demand. If a match is successful, the warehouse's match count is incremented by 1. If the demand list has 5 items and a warehouse matches 4, then the warehouse's match count is 4.
[0034] After obtaining the matching counts of all candidate warehouses, the system needs to decide which one to recommend. Specifically, the system directly compares the warehouses and selects the warehouse with the highest value. By prioritizing the recommendation of the warehouse that can provide the most types of required materials, the repairman only needs to go to one warehouse to collect most of the items, reducing the number of errands.
[0035] Finally, the system locks the warehouse with the highest overall score as the designated warehouse and passes the ID, location, and specific material list matched within that warehouse to the next stage.
[0036] In step S140, the optimal disposal personnel are selected from multiple qualified operators based on the personnel qualification requirements.
[0037] Specifically, the system obtains the personnel qualification requirements for the work order, such as holding a high-voltage electrician's certificate and the project ID to which the work order belongs. The system queries the personnel database, first filtering out employees belonging to that project, comparing their skill tags or certificate fields in their files, and eliminating employees who do not possess the required qualifications. For example, if the work order requires electrical work, employees without the relevant certificates are directly filtered out, thus forming a preliminary list of candidates. The system uses a backend service interface to capture the current dynamic status data of the candidates in real time, such as whether they are online / idle, busy, or off-duty / on vacation, and also counts the number of work orders that the employee has completed or is currently handling that day. Subsequently, the system obtains the current GPS geographic coordinates of the candidates and calculates the straight-line distance or path distance between each person's current location and the location where the work order occurred. For example, the system sets a service radius threshold, such as 3 kilometers or a 15-minute journey, and only retains candidates within this distance range to avoid dispatching work orders that are too far away, resulting in excessive response delays. Finally, the optimal personnel are selected for handling the work.
[0038] Specifically, the system compares the number of completed work orders for the remaining candidates today. For example, if employee A has completed 8 work orders today, while employee B has only completed 2, the system will prioritize employee B. Thus, the system locks the employee ID that meets all conditions and has the least workload as the optimal dispatcher. This employee's ID, name, contact information, and real-time location will be integrated into the final dispatch instruction, thereby achieving efficient utilization of human resources.
[0039] In steps S150 and S160, the suggested disposal steps, designated warehouse, and disposal personnel are integrated to form a property service work order. The property service work order is then sent to the requesting terminal.
[0040] Specifically, the system aggregates the three previous outputs into a single context: suggested action steps, designated warehouse ID, and action personnel ID. The system calculates and constructs the action path logic: the action personnel must first go to the designated warehouse to retrieve items from the material list, and then proceed to the work order location to perform the action according to the suggested steps. The system encapsulates this information into a standard data format, converts the encapsulated data into a network transmission format, and packages the data into HTTP requests or WebSocket message frames according to the interface protocols supported by the calling end (APP, mini-program, Web). The data packets travel through the internet, carrier networks, and firewalls, ultimately pinpointing the target calling end device. By integrating these disparate algorithm results into action plans with business guidance, the optimal strategy can be delivered to the executor seamlessly and in real-time, truly realizing the implementation from intelligent analysis to on-site operations.
[0041] In some specific embodiments, the steps involve identifying the work order input information using a preset vector model to obtain a list of material requirements and personnel qualification requirements, specifically including the following steps: The work order input information is fed into the bge-m3 embedding model to generate the work order information embedding vector.
[0042] Specifically, a simple descriptive text, such as "the corridor light is broken," may lack context regarding which project or type of light it refers to. The system uses a keyword concatenation method, inserting structured metadata, such as project ID, device name, and repair location, as strong feature keywords into the original work order description text. For example, the original input "the light is not working" becomes "Project ID: Xingfuli," the device is "public corridor light," and the description is "the light is not working." This ensures that the generated vector not only contains what happened but also where it happened and which device is involved, allowing for more precise differentiation of similar issues in different scenarios within the vector space.
[0043] In text embedding models, since Chinese characters cannot be directly processed, the input text string is first segmented into the smallest semantic units, or tokens. For example, "water pipe burst" is segmented into sub-words such as water, pipe, burst, and crack, or even finer-grained sub-words. Each token is converted into an initial numerical vector. At this point, each word is just an independent numerical representation and does not yet have contextual relationships. This model then analyzes the degree of association between each word in the sentence and all other words. For example, in "water pipe burst due to aging," the model uses an attention mechanism to capture the strong association between "burst," "water pipe," and "aging," while ignoring irrelevant function words. It performs non-linear transformations and mappings on the associated features to extract higher-level semantic features. After processing by the Transformer layer, the model outputs a series of vector sequences for each token in the text. Specifically, the token sequence is compressed through pooling strategies, such as using average pooling to calculate the average of all token vectors in the sentence and merge them into a single, fixed-dimensional vector. Finally, this vector is usually L2 normalized to scale its length to a unit length, making it more convenient and accurate for subsequent calculation of cosine similarity.
[0044] The final output embedvent is a high-dimensional array. This vector is a coordinate point in a high-dimensional space. If two work orders are semantically very similar, such as "water pipe leak" and "pipe seepage", the vectors generated by the two will be very close in this space. If they are work orders of completely different types, such as "water pipe leak" and "elevator malfunction", the vectors of the two will be far apart in the space.
[0045] By leveraging the powerful attention mechanism of text embedding models, we can gain a deep understanding of the complex relationships and contextual semantics between words in the concatenated work order text, and map unstructured text information into a mathematical coordinate point in a high-dimensional semantic space.
[0046] The cosine similarity between the embedded vector of the work order information and the embedded vectors corresponding to various work order handling types in the SOP database is calculated.
[0047] Specifically, this refers to the embedding generated by the text embedding model in the previous step. It represents the semantic coordinates of the new work order that needs to be processed, such as a burst water pipe in the bathroom on the 3rd floor. It also corresponds to the workflow embeddings pre-existing in the SOP database for various work order handling types. Each vector represents a standard work order category, such as standard-water pipe repair, standard-electrical fault, etc. In the high-dimensional space generated by the text embedding model, each word and each sentence is a point, and the arrow pointing from the origin to this point is a vector. If two texts are semantically very similar, such as "water pipe burst" and "pipe leak", their vectors will point in almost the same direction in space. If two texts are semantically unrelated, such as "water pipe burst" and "elevator is broken", their vectors will form a large angle. Cosine similarity measures this similarity by calculating the cosine value of the angle between two vectors. When the angle is 0 degrees, that is, the directions are completely consistent, Cosθ=1. When the angle is 90 degrees, Cosθ=0. When the angle is 180 degrees, Cosθ=-1.
[0048] In actual computer calculations, it's not necessary to actually measure angles with a protractor. Instead, the vector dot product formula is used to quickly obtain the result. Since the vectors generated by text embedding models are usually normalized (i.e., the vector length is scaled to 1 unit), the formula for calculating cosine similarity simplifies to the vector dot product, as follows: The system multiplies the first value of the work order vector (embevent) by the first value of the SOP vector, the second by the second, and so on, up to the Nth dimension. The sum of all the products is then obtained as the cosine similarity score. If the score is high, such as 0.85 or 0.92, it indicates that the two vectors have a high degree of semantic overlap.
[0049] In this way, the embedding vector of the current work order is used to perform the dot product calculation mentioned above with each workflow embedding vector in the database to generate a score list. For example, the score calculated with "Pipe Repair SOP" is 0.92, the score calculated with "Electricity Emergency Repair SOP" is 0.15, and the score calculated with "Cleaning Service SOP" is 0.08. After the calculation is completed, the system will sort all the scores and find the option with the highest score.
[0050] This step does not rely on hard keyword matching, but rather on semantic understanding. Therefore, even if the user inputs "the water room is full of water" or "the pipes have burst," as long as their direction in the vector space is close to the standard SOP "leaking pipes," the calculated cosine similarity will be high, thus achieving accurate classification.
[0051] The SOP database with the highest similarity is selected as the matching and classification result. Based on the matching and classification result, a list of material requirements and personnel qualification requirements are obtained.
[0052] Specifically, the system has a candidate list containing all entries in the SOP database and their corresponding similarity scores (e.g., SOP A - 0.85, SOP B - 0.42, SOP C - 0.15). The system sorts the scores in the list in descending order and selects the entry that ranks first. The semantics of the current work order have the highest semantic overlap with the top-1 SOP entry. Therefore, the category represented by the SOP entry, such as "indoor water pipe repair", is officially confirmed as the classification result of the current work order.
[0053] Once the SOPID is locked, the system no longer needs to perform complex reasoning. Instead, it directly retrieves the pre-stored standard configuration under that ID, just like looking up a dictionary. This ensures the standardization and compliance of business execution. Specifically, the system uses the SOPID as the query key, and the database returns a list of materials corresponding to that ID, where the list is a predefined standard configuration.
[0054] For example, if the SOPID is matched to WaterPipe Leak (005), the system will directly read its attached material list, such as pipe wrenches, PTFE tape, replacement pipe fittings, waterproof tape, etc.
[0055] Similarly, each SOP entry also has a personnel field. The system queries the personnel skill tags or certificate requirements corresponding to that ID. For example, for 005WaterPipeLeak, the system reads that the qualification requirement is a basic plumber certificate or pipe repair skills, which will be directly used as the screening condition for the subsequent personnel dispatch model. Only employees with this tag will be considered for order assignment by the system.
[0056] Therefore, by using the highest score of cosine similarity to determine the type of work order, and using the qualitative results to extract all the pre-set resource attributes of that type of work order, absolute standards and compliance can be guaranteed for subsequent material and personnel needs.
[0057] In a further embodiment, the step involves generating corresponding handling step suggestions based on the work order input information, specifically including the following steps: The work order information embedding vector is input into the RAG knowledge base of the property work order handling process; knowledge fragments related to the work order information embedding vector are retrieved from the RAG knowledge base; both the knowledge fragments and the work order information embedding vector are input into the artificial intelligence language model, and the artificial intelligence language model generates handling step suggestions.
[0058] Specifically, this step aims to transform unstructured work order requirements into mathematical query requests. The system embeds the work order information generated in the previous step into a vector as the query vector and loads it into the RAG system's vector retrieval engine. The property work order handling process knowledge base is not a traditional keyword database, but a preprocessed high-dimensional vector database. It stores massive amounts of historical work order processing records, equipment maintenance manuals, SOPs (Standard Operating Procedures), and emergency plan texts. All of these texts have been converted into corresponding embedded vectors and indexed, such as the HNSW index. Using a vector search engine, in a high-dimensional space of hundreds of millions, the system finds the nearest neighbor point in semantic distance to the current work order vector, thus achieving a leap from "keyword matching" to "semantic understanding matching." Even if the work order description and the knowledge base document do not use completely identical terms, as long as they are semantically related, the search can be triggered.
[0059] The vector search engine calculates the similarity score between the query vector and all candidate options in the knowledge base. The system sorts them according to the score and extracts the top K (e.g., Top-5) knowledge fragments with the highest similarity. These fragments contain historical solutions, operating procedures, or expert experience for the current work order type. This process is equivalent to cutting out the most relevant contextual information from the huge knowledge base as the factual basis for generating the answer. This effectively solves the "illusion" problem that may occur in general large language models and ensures that the generated suggestions comply with the specific norms of property management.
[0060] Finally, the system constructs a structured input instruction, defining the model's role (e.g., a senior property engineering expert) and output format requirements, such as listing operation guidelines step-by-step. The Top-K knowledge fragments retrieved in the previous step are filled into the instruction's context window, serving as an external knowledge base for model reasoning. The original description information of the work order or its semantic features are input to clarify the specific problem to be solved. Subsequently, the AI language model utilizes its internal attention mechanism to analyze the correlation between the work order information and the retrieved knowledge fragments. This model automatically focuses on the key steps in the knowledge base to solve the current problem, ignoring irrelevant information. Based on a deep understanding of the instruction and context, the model generates a coherent and logically rigorous text by predicting tokens one by one. The final output is formatted as specific handling steps, for example: 1. Immediately shut off the main water valve for the floor; 2. Use a water extractor to remove accumulated water; 3. Replace the damaged sealing ring.
[0061] By using vector retrieval to pinpoint the scope of professional knowledge, and then using a large language model to logically reorganize the knowledge and generate text, the transformation from data to intelligent decision-making is achieved.
[0062] In some specific embodiments, the step of querying the material storage status of each warehouse based on the material demand list and selecting the optimal designated warehouse includes the following steps: Based on the material demand list, obtain a warehouse list of all candidate warehouses and material storage information for each candidate warehouse. The material storage information includes various types of material information.
[0063] Specifically, the system first defines the service area based on the project ID or geographic coordinates of the work order. It then performs spatial or relational queries in the database to filter out all available warehouse entities within this service radius. This reduces the massive global inventory data into a local set of candidate warehouses, lowering computational complexity. For each candidate warehouse, the system retrieves its current material storage information through database relational queries; this information forms the input features for the matching algorithm. Various material information exists in the database as structured records, including unique material identifiers, material names, specifications, inventory quantities, and storage locations. The system loads this data into memory to prepare for subsequent comparison operations.
[0064] The material storage information stored in each candidate warehouse is matched with the material storage information required by the material demand list, and the candidate warehouse with the highest matching value is selected as the optimal designated warehouse.
[0065] Specifically, the system employs nested loop logic for comparison. The outer loop iterates through the candidate warehouse list, while the inner loop iterates through each requirement in the material demand list. For each specific warehouse, the system checks its inventory records to see if there are materials that can meet the current requirement. The system compares the names of the materials in the inventory with those in the demand list. In practice, fuzzy matching algorithms can be used, specifically edit distance algorithms, Jaccard similarity coefficients, or semantic vector-based similarity calculations. A similarity threshold is then set, such as ≥80%. If the similarity between the inventory material name and the demand material name exceeds the threshold, and the inventory quantity meets the minimum requirement, the material is considered a successful match. If the material is not present in the inventory, or the name difference is too large, causing the similarity to fall below the threshold, the match is considered a failure.
[0066] The system accumulates and calculates the matching results for each warehouse, determining its matching value based on the number of material types that meet the requirements or their satisfaction weight. For example, if the requirement list contains 5 types of materials, and warehouse A successfully matches 4, its matching value is 4; if warehouse B matches 3, its matching value is 3. Furthermore, a weighting factor can be introduced into the matching value; for instance, critical materials (such as water pumps) have a higher weight than general materials (such as tape), resulting in a weighted total matching score.
[0067] Finally, the system sorts the matching values calculated for all candidate warehouses, runs the maximum value parameter function, and locates the warehouse node with the largest matching value. By selecting the warehouse with the highest matching value, it means that the warehouse can meet the maximum material demand in the fewest number of times, thereby maximizing the material collection efficiency of maintenance personnel and reducing the additional time cost and logistics consumption caused by collecting materials from multiple warehouses.
[0068] In a further embodiment, the step involves matching the material storage information stored in each candidate warehouse with the material storage information required by the material demand list, and selecting the candidate warehouse with the highest matching value as the optimal designated warehouse. This specifically includes the following steps: Iterate through each type of material information required by the material demand list, and iterate through each type of material information stored in each candidate warehouse; compare each type of material information stored in each candidate warehouse with the corresponding type of material information in the material demand list to obtain the matching value of each candidate warehouse.
[0069] Specifically, a nested iterative structure is constructed. First, the algorithm locks a candidate warehouse (outer loop). Then, a comparison channel is established between the inventory materials in the warehouse (inner loop 1) and the materials in the material demand list (inner loop 2). For each material in the demand list, such as "DN20PVC pipe", the system performs a full scan of the inventory records in the current candidate warehouse. By using fuzzy string matching or semantic similarity calculation, the system calculates the similarity score between the inventory material name and the demand material name, for example, by using edit distance or cosine similarity. If the similarity score reaches a preset standard, such as greater than or equal to 80%, it is determined that the inventory material covers the demand item.
[0070] The system maintains a matching counter with an initial value of 0 for each candidate warehouse. Whenever a material that meets the requirements list is found in the warehouse (i.e., the comparison is successful), the matching counter for that warehouse is incremented by 1. After iterating through all materials in the requirements list, the final value of the counter is the matching value for that warehouse. This matching value visually reflects how many types of materials the warehouse can meet for a work order. For example, if there are 5 types of requirements and warehouse A has a matching value of 4, it means that warehouse A can provide 4 of them.
[0071] Retrieve all candidate warehouses whose matching values meet the matching threshold.
[0072] Specifically, after obtaining the matching values of all candidate warehouses, the system introduces a matching threshold. This is a preset qualification line (e.g., a matching value ≥ 2). The system compares the matching values of all warehouses with the threshold. Warehouses with matching values ≥ the threshold are considered valid warehouses and retained for the next round; conversely, warehouses with matching values < the threshold, such as warehouses that can only provide 0 or 1 type of material, are considered invalid warehouses and directly removed from the candidate list. This avoids assigning work orders to warehouses with almost no stock available, ensuring a lower limit for scheduling and reducing invalid work order assignments.
[0073] The candidate warehouse with the highest matching value is selected as the optimal designated warehouse.
[0074] Specifically, in the set of "valid warehouses" remaining after threshold filtering, the system executes a maximum value search algorithm. By comparing the matching values of these valid warehouses, the system locks the warehouse with the largest value, which is then defined as the optimal designated warehouse.
[0075] By traversing through all areas to achieve full coverage checks, comparing and counting to quantify the degree of material satisfaction, filtering out unqualified options through threshold filtering, and finally using an optimization algorithm to lock in the warehouse with the highest resource coverage, the material scheduling becomes more precise and efficient.
[0076] It should be noted that in this embodiment of the invention, if there is a tie where multiple warehouses have the same highest matching value, a secondary weight can be introduced to break the tie. However, the core principle described in this step is to maximize the matching value.
[0077] In a further embodiment, after obtaining all candidate warehouses whose matching values satisfy the matching threshold, the step further includes the following steps: Check whether the matching values of each candidate repository that meets the matching threshold are consistent. Specifically, the system performs statistical analysis on the matching values of all valid warehouses that have passed the initial screening. If the matching values of each warehouse form an equal sequence, such as all being 4 or all being 5, they are considered to be consistent. In this case, there is no difference in the breadth of material coverage among these warehouses. If the values are different, such as 4, 3, 2, they are considered to be inconsistent. Based on the judgment result, the system will execute different branch logic paths.
[0078] If the matching values of all candidate warehouses that meet the matching threshold are consistent, then the candidate warehouse that is closest to the target geographical location and / or the candidate warehouse that meets the overall inventory quantity of the material information corresponding to the material demand list is selected.
[0079] Specifically, when the types of materials cannot be distinguished, the system activates secondary optimization indicators for multi-dimensional comprehensive evaluation. Specifically, assuming equal material types, closer proximity results in lower transportation costs and faster response times. The system calculates the path distance or straight-line distance between each candidate warehouse coordinate and the target geographical location (which can be understood as the work order location or the current location of the maintenance worker), prioritizing the warehouse with the smallest distance value. This reflects the principle of prioritizing timeliness in logistics scheduling. The matching value only reflects the type coverage, not the quantity. Selecting a warehouse with a larger inventory can reduce the risk of secondary material requisition due to inventory depletion and can cope with sudden large-scale consumption. The system accumulates or weighted sums the inventory quantities of the matched materials to obtain the total inventory, prioritizing the warehouse with the largest total inventory.
[0080] It should be noted that distance and inventory can be weighted according to certain factors, such as 60% distance and 40% inventory, to calculate the overall score, and the one with the highest score can be selected.
[0081] If the matching values of the candidate warehouses that meet the matching threshold are inconsistent, the next step is to select the candidate warehouse with the highest matching value as the optimal designated warehouse.
[0082] Specifically, if warehouse A is close but can only provide 3 types of resources, the matching value is 3, while warehouse B is slightly farther away but can provide 5 types of resources, the matching value is 5. The system must select warehouse B as the optimal designated warehouse.
[0083] In this way, the accuracy of material allocation is ensured, and the optimal balance between efficiency and safety is achieved when capabilities are comparable.
[0084] In some specific embodiments, the step involves screening from multiple qualified operators based on personnel qualification requirements to obtain the optimal handling personnel, specifically including the following steps: Based on the personnel qualification requirements, all qualified operators are selected from the project personnel list, and a candidate operator list is generated.
[0085] Specifically, the personnel qualification requirements, derived from the structured data extracted from the SOP database in the previous step, can be understood as a set of skill tags or a list of certificate codes. For example, a work order might require a high-voltage electrician certificate and a high-altitude work permit, and the worker must meet all of these conditions simultaneously. The project personnel list, on the other hand, consists of pre-stored files of all on-duty employees for that project in the database. Each employee record includes an employee ID, name, and associated qualification / skill field. This field stores all the certificates and skill tags held by that employee.
[0086] The system iterates through the project personnel list and compares the qualification fields in each employee's file. Only when an employee's skill tags fully cover the qualification requirements of the work order will the employee be selected to generate a candidate list of workers. This ensures that only those who can do the work are excluded, thus achieving compliance in scheduling.
[0087] The system allows users to query the real-time status of each candidate in the candidate task list. The real-time status includes the candidate's current geographical location and current workload.
[0088] Specifically, since static files cannot reflect current availability, the system must introduce dynamic data in the time dimension. Therefore, the current latitude and longitude coordinates of the candidate are captured through the location service interface to determine their spatial location. In addition, the current "number of pending work orders" or "number of work orders completed today" of the employee must be queried through the business system interface to quantify their busyness and provide necessary input parameters for subsequent spatiotemporal distance calculation and load balancing decisions.
[0089] Filter the list of candidate workers to select all candidates whose current geographical location meets the target geographical location range threshold and whose current workload is below the workload threshold.
[0090] Specifically, among the candidates, further elimination is performed on those who, while technically qualified, may be unable to arrive in time or lack the time to process orders. Therefore, the straight-line distance or path distance between the candidate's location and the target geographical location (where the work order occurred) is calculated. If this distance exceeds a preset threshold, the candidate is deemed too far away and eliminated, ensuring service timeliness. The current workload of each candidate is compared to a preset workload threshold, such as processing no more than five orders simultaneously. If the workload has reached its limit, the candidate is deemed overloaded and eliminated, thus ensuring service quality and response speed.
[0091] The candidate operator whose current geographical location is closest to the target geographical location and whose current workload is the lowest is selected as the optimal disposal operator.
[0092] Specifically, among the remaining available personnel, a binary selection process is no longer employed. Instead, the best candidates are chosen based on comprehensive indicators, primarily involving a trade-off between two core objectives: proximity and low workload. The remaining personnel are ranked according to their distance from the target location; the closer the distance, the lower the space cost, and the higher the priority. They are also ranked according to their current workload; the lower the workload, the higher the availability and the higher the priority.
[0093] Based on the above, the final selection can be made using the following two indicators: 1. Sequence selection; 2. Weighted scoring.
[0094] Specifically: 1. Sequence filtering prioritizes selecting the closest person; if multiple people are close, the person with the lowest workload is selected. 2. Weighted scoring involves constructing a scoring function Score=w1*(1 / distance)+w2*(1 / workload) and calculating the highest score.
[0095] Ultimately, identifying the optimal personnel ensures that work orders are assigned to the employee who can arrive on-site the fastest and has the most dedicated service, thereby improving overall operational efficiency.
[0096] This invention also provides an intelligent work order allocation device 100 for scheduling workers in a property management scenario. The intelligent work order allocation device 100 is installed in the property management system, and the property management system is connected to the caller for signal transmission. The intelligent work order allocation device 100 is used to execute any of the aforementioned intelligent work order allocation methods for scheduling workers in a property management scenario.
[0097] Specifically, the following describes in detail the specific structure of a work order intelligent dispatching device 100 for dispatching workers in a property management scenario, as provided in the embodiments of the present invention. Figure 2 As shown, the intelligent work order sorting device 100 specifically includes the following components: The receiving module 110 is used to receive work order input information sent from the calling terminal; The identification module 120 is used to identify the work order input information through a preset vector model to obtain a list of material requirements and personnel qualification requirements, and to generate corresponding disposal step suggestions based on the work order input information. The query module 130 is used to query the material storage status of each warehouse according to the material demand list and select the optimal designated warehouse. The screening module 140 is used to screen multiple qualified operators according to the personnel qualification requirements in order to obtain the optimal disposal personnel. The integration module 150 is used to integrate the disposal step suggestions, the designated warehouse, and the disposal personnel to form a property service work order; The sending module 160 is used to send the property service work order to the calling terminal.
[0098] This invention further provides a computer device, which can be used to execute the intelligent work order allocation method for scheduling workers in a property setting as described in any of the above embodiments.
[0099] See Figure 3 The computer device 500 includes a processor 502, a memory, and a network interface 505 connected via a communication bus 501. The memory may include a storage medium 503 and internal memory 504.
[0100] The storage medium 503 may store an operating system 5031 and a computer program 5032. When the computer program 5032 is executed, it enables the processor 502 to execute a work order intelligent allocation method for scheduling workers in a property setting. The storage medium 503 may be a volatile storage medium or a non-volatile storage medium.
[0101] The processor 502 provides computing and control capabilities to support the operation of the entire computer device 500.
[0102] The internal memory 504 provides an environment for the operation of the computer program 5032 in the storage medium 503. When the computer program 5032 is executed by the processor 502, the processor 502 can execute a work order intelligent allocation method for the scheduling of workers in a property scenario.
[0103] This network interface 505 is used for network communication, such as providing data transmission. Those skilled in the art will understand that... Figure 3The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the computer device 500 to which the present invention is applied. The specific computer device 500 may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.
[0104] The processor 502 is used to run the computer program 5032 stored in the memory to implement the corresponding functions in the above-mentioned intelligent work order allocation method for scheduling workers in a property scenario.
[0105] Those skilled in the art will understand that Figure 3 The embodiments of the computer device shown do not constitute a limitation on the specific configuration of the computer device. In other embodiments, the computer device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. For example, in some embodiments, the computer device may include only memory and a processor. In such embodiments, the structure and function of the memory and processor are different from those shown. Figure 3 The embodiments shown are consistent and will not be described again here.
[0106] It should be understood that, in this embodiment of the invention, the processor 502 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.
[0107] In another embodiment of the invention, a computer-readable storage medium is provided. This computer-readable storage medium may be volatile or non-volatile. The computer-readable storage medium stores a computer program, wherein when executed by a processor, the computer program implements the steps included in the above-described intelligent work order allocation method for dispatching workers in a property management scenario.
[0108] Those skilled in the art will readily understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.
[0109] In the embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Units with the same function may be grouped into one unit. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, or it may be an electrical, mechanical, or other form of connection.
[0110] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention, depending on actual needs.
[0111] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0112] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a computer-readable storage medium and includes several instructions to cause a computer device (which may be a personal computer, a data transmitter, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned computer-readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks.
[0113] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. A method for intelligent work order allocation for dispatching workers in a property management setting, characterized in that, The intelligent work order allocation method is applied to a property management system, which is connected to the caller terminal via signal transmission. The intelligent work order allocation method includes: Receive work order input information sent from the calling terminal; The work order input information is identified by a preset vector model to obtain a list of material requirements and personnel qualification requirements, and corresponding disposal step suggestions are generated based on the work order input information. Based on the material demand list, query the material storage status of each warehouse and select the optimal designated warehouse; Based on the aforementioned personnel qualification requirements, multiple qualified operators will be screened to obtain the optimal personnel for handling the situation. The proposed disposal steps, the designated warehouse, and the disposal personnel are integrated to form a property service work order; The property service work order is sent to the calling terminal.
2. The intelligent work order allocation method for dispatching workers in a property management scenario according to claim 1, characterized in that, The step of identifying the work order input information using a preset vector model to obtain the material demand list and personnel qualification requirements includes: The work order input information is input into the text embedding model to generate a work order information embedding vector; The cosine similarity between the embedded vector of the work order information and the embedded vectors corresponding to various work order processing types in the SOP database is calculated. The SOP database with the highest similarity was selected as the matching and classification result; The material demand list and personnel qualification requirements are obtained based on the matching and classification results.
3. The intelligent work order allocation method for dispatching workers in a property management scenario according to claim 2, characterized in that, The step of generating corresponding processing step suggestions based on the work order input information includes: The work order information is embedded into a vector and input into the RAG knowledge base of the property work order processing flow; Retrieve knowledge fragments related to the work order information embedding vector from the RAG knowledge base; The knowledge fragments and the work order information embedding vectors are both input into the artificial intelligence language model, and the proposed handling steps are generated through the artificial intelligence language model.
4. The intelligent work order allocation method for dispatching workers in a property management scenario according to claim 1, characterized in that, The step of querying the material storage status of each warehouse according to the material demand list and selecting the optimal designated warehouse includes: Based on the material demand list, obtain a warehouse list of all candidate warehouses and material storage information of each candidate warehouse in the warehouse list. The material storage information includes various material information. The material storage information stored in each of the candidate warehouses is matched with the material storage information required by the material demand list, and the candidate warehouse with the highest matching value is selected as the optimal designated warehouse.
5. The intelligent work order allocation method for dispatching workers in a property management scenario according to claim 4, characterized in that, The process of matching the material storage information stored in each of the candidate warehouses with the material storage information required by the material demand list, and selecting the candidate warehouse with the highest matching value as the optimal designated warehouse, includes: Iterate through each type of material information required by the material demand list, and iterate through each type of material information stored in each of the candidate warehouses; The information on each type of material stored in each of the candidate warehouses is compared with the information on the material demand list to obtain the matching value of each candidate warehouse. Obtain all candidate warehouses whose matching values satisfy the matching threshold; The candidate warehouse with the highest matching value is selected as the optimal designated warehouse.
6. The intelligent work order allocation method for dispatching workers in a property management scenario according to claim 4, characterized in that, After obtaining all candidate warehouses whose matching values satisfy the matching threshold, the process further includes: Detect whether the matching values of each candidate warehouse that meets the matching threshold are consistent; If the matching values of all candidate warehouses that meet the matching threshold are consistent, then the candidate warehouse that is closest to the target geographical location and / or the candidate warehouse that meets the total inventory of the material information corresponding to the material demand list is selected as the optimal designated warehouse. If the matching values of the candidate warehouses that meet the matching threshold are inconsistent, the step is to select the candidate warehouse with the highest matching value as the optimal designated warehouse.
7. The intelligent work order allocation method for dispatching workers in a property management scenario according to claim 1, characterized in that, The process of selecting the optimal personnel for handling a case from multiple qualified workers based on the personnel qualification requirements includes: Based on the personnel qualification requirements, all qualified operators are selected from the project personnel list, and a candidate operator list is generated. The real-time status of each candidate in the candidate task list is queried in real time, and the real-time status is the current geographical location and current workload of each candidate. Filter the list of candidate workers to select all candidates whose current geographical location meets the target geographical location range threshold and whose current workload is lower than the workload threshold. The candidate whose current geographical location is closest to the target geographical location and whose current workload is the lowest is selected as the optimal person to handle the situation.
8. A work order intelligent dispatching device for dispatching workers in property management scenarios, characterized in that, The intelligent work order dispatching device is installed in the property management system, which is connected to the caller terminal via signal transmission. The intelligent work order dispatching device includes: The receiving module is used to receive work order input information sent from the calling terminal; The identification module is used to identify the work order input information through a preset vector model to obtain a list of material requirements and personnel qualification requirements, and to generate corresponding handling step suggestions based on the work order input information. The query module is used to query the material storage status of each warehouse according to the material demand list and select the optimal designated warehouse. The screening module is used to screen multiple qualified operators based on the personnel qualification requirements in order to obtain the optimal personnel for handling the situation. The integration module is used to integrate the proposed disposal steps, the designated warehouse, and the disposal personnel to form a property service work order; The sending module is used to send the property service work order to the calling terminal.
9. A computer device, characterized in that, The computer device includes a processor, a network interface, a memory, and a communication bus, wherein the processor, network interface, and memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements the steps of the work order intelligent allocation method for scheduling workers in a property setting as described in any one of claims 1-7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the work order intelligent allocation method for scheduling workers in a property scenario as described in any one of claims 1-7.