Property service dispatching method and system based on personnel digitization

Abstract

This invention provides a property service scheduling method and system based on personnel digitization, relating to the field of property service scheduling technology. The method includes: acquiring personnel information data streams, property service information data streams, and environmental information data streams; performing correlation feature analysis on historical multi-source data to obtain corresponding correlation feature vector sequences; based on the correlation feature vector sequences and historical service scheduling schemes, obtaining the predicted optimal matching benchmark for various work orders and maintenance personnel in the current scenario through a conditional temporal fusion model; constructing a coupled dual-graph structure based on the spatial division relationship of the property service area and the skill correlation relationship of maintenance personnel, and evaluating the real-time scheduling scheme of maintenance personnel based on the coupled dual-graph structure to obtain corresponding trajectory anomaly evaluation values; constructing a secondary rescheduling algorithm based on the trajectory anomaly evaluation values, and generating a secondary scheduling scheme based on the secondary rescheduling algorithm to improve the quality and efficiency of property services.

Description

Technical Field

[0001] This invention relates to the field of property service scheduling technology, specifically a property service scheduling method and system based on personnel digitization. Background Technology

[0002] With the advancement of digital transformation in the property management industry, the accuracy and efficiency of operation and maintenance scheduling have become core keys to improving the quality of property services. Traditional property service scheduling models rely heavily on manual experience to allocate work orders, lacking systematic integration and in-depth analysis of multi-source data such as operation and maintenance personnel, service work orders, and environmental conditions. This makes it difficult to achieve effective correlation of multi-dimensional data, resulting in a lack of scientific benchmarks for matching work orders with operation and maintenance personnel, and easily leading to problems such as skill mismatch, untimely response, and uneven personnel workload. Existing scheduling systems do not fully consider the spatial distribution characteristics of property service areas and the skill correlation of operation and maintenance personnel, making it impossible to accurately evaluate the rationality of real-time scheduling plans. Furthermore, they lack effective scheduling optimization mechanisms. When trajectory anomalies or unreasonable scheduling occur, it is difficult to quickly adjust to form the optimal scheduling plan, thereby affecting the continuity of property services and user satisfaction, and failing to meet the refined and efficient scheduling needs of property services in the current context of digital personnel.

[0003] Therefore, we now offer a property service scheduling method and system based on personnel digitization. Summary of the Invention

[0004] To address the aforementioned issues of inefficient and monotonous property service scheduling, the present invention aims to provide a property service scheduling method and system based on personnel digitization.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a property service dispatching system based on personnel digitization, the system comprising: a multi-dimensional data acquisition module, a data feature analysis module, a dual-graph structure matching module, and a dispatching efficiency evaluation module; The multi-dimensional data acquisition module is used to acquire personnel information data streams from operation and maintenance personnel, service information data streams from property services, and environmental information data streams, forming multi-source data; The data feature analysis module is used to perform correlation feature analysis on historically collected multi-source data to obtain the corresponding correlation feature vector sequence. Based on the correlation feature vector sequence and historical service scheduling scheme, the optimal matching benchmark for various work orders and maintenance personnel in the current scenario is obtained through a conditional time series fusion model. The dual-graph structure matching module is used to construct a coupled dual-graph structure based on the spatial division relationship of the property service area and the skill association relationship of the operation and maintenance personnel. The scheduling efficiency evaluation module evaluates the real-time scheduling scheme of operation and maintenance personnel based on the coupled dual-graph structure, and obtains the corresponding trajectory anomaly evaluation value; based on the trajectory anomaly evaluation value, a secondary rescheduling algorithm is constructed, and a secondary scheduling scheme is generated based on the secondary rescheduling algorithm to improve the quality and efficiency of property services.

[0006] Preferably, the process of obtaining the associated feature vector sequence includes: Data cleaning and standardization preprocessing were performed on the historically collected personnel information data stream, service information data stream, and environmental information data stream. The three types of preprocessed multi-source data were matched with the same time granularity, and personnel dimension features, work order dimension features, and environmental dimension features were extracted respectively. Using the historical property service work order ID as the core association key, an association mapping relationship was established between historical property service work orders and maintenance personnel, the spatial area to which the work order belongs, and the real-time environmental status, and a multi-source data association matrix was constructed. Based on the multi-source data association matrix, cross-correlation features of personnel dimension, work order dimension, and environmental dimension were mined, and personnel dimension features, work order dimension features, environmental dimension features and cross-correlation features were concatenated to generate a single time step association feature vector. The association feature vectors of all time steps of the same maintenance personnel were sorted in chronological order to form a corresponding association feature vector sequence.

[0007] Preferably, the process of obtaining the predicted optimal matching benchmark includes: Obtain historical service scheduling schemes for different operation and maintenance personnel and associated feature vector sequences related to the corresponding historical service scheduling schemes; use the service quality of the corresponding historical service scheduling schemes for operation and maintenance personnel as conditional labels and the associated feature vector sequences related to the corresponding historical service scheduling schemes as training data to construct a training dataset; complete the training of the conditional temporal fusion model based on the training dataset; input the associated feature vector sequences of different operation and maintenance personnel in the current scenario into the trained conditional temporal fusion model, and output the predicted skill matching degree threshold, predicted response time upper limit, and predicted personnel load coefficient of various work orders and operation and maintenance personnel in the current scenario, thereby forming the predicted optimal matching benchmark; Based on the associated feature vector sequence of different operation and maintenance personnel, the real-time matching degree between each operation and maintenance personnel and the service work order to be assigned is calculated in the current real-time state. The real-time matching degree is compared with the corresponding dimension index of the predicted optimal matching benchmark. The dimension-level deviation and the comprehensive deviation are calculated. The service work order to be assigned is assigned to the operation and maintenance personnel with the smallest comprehensive deviation. The corresponding dimension-level deviations are then used to form a primary matching deviation sequence.

[0008] Preferably, the process of constructing the coupled dual-map structure includes: A directed weighted spatial topology graph is formed based on the basic attributes of the property service area; an undirected weighted personnel skill association graph is formed based on the basic attributes of a single maintenance personnel; a mapping relationship is established between the area nodes of the spatial topology graph and the personnel skill association graphs of several maintenance personnel corresponding to the area nodes in the personnel skill association graph, thus constructing a coupled dual-graph structure. The primary matching deviation sequence is used as an additional attribute of the corresponding region node in the coupled dual-graph structure. The nodes of the coupled dual-graph structure are iteratively updated and graph-enhanced through a graph neural network. The enhanced iterative matching deviation coefficient of each node is output. The iterative matching deviation coefficients of the corresponding nodes are collected according to the corresponding maintenance personnel to form the iterative matching deviation coefficient sequence of the corresponding maintenance personnel.

[0009] Preferably, the process of obtaining the corresponding trajectory anomaly assessment value includes: Based on the spatial topology graph in the coupled dual-graph structure, the work order processing order, work order processing time, and node dwell time of each maintenance personnel in the real-time scheduling scheme corresponding to the service work order to be assigned are extracted, and a standardized planning trajectory sequence is generated according to the spatial node order. The iterative matching deviation coefficient sequence of the corresponding maintenance personnel is aligned one by one with the spatial nodes of the planning trajectory sequence, and the deviation weighting value of each trajectory node is calculated to obtain the trajectory score of a single node. Based on the preset trajectory status benchmark threshold, the trajectory anomaly evaluation value of the corresponding maintenance personnel at each trajectory node under the real-time scheduling scheme is calculated.

[0010] Preferably, a secondary rescheduling algorithm is constructed based on the trajectory anomaly evaluation value, and the process of generating a secondary scheduling scheme based on the secondary rescheduling algorithm includes: The trajectory anomaly assessment value is graded and processed according to the preset three-level trajectory anomaly threshold; based on the graded processing result, the next one or more trajectory nodes in the real-time scheduling plan of the operation and maintenance personnel are released and recorded as secondary release nodes. A secondary judgment is made on the comprehensive deviation of other maintenance personnel in the corresponding area node. If the comprehensive deviation of other maintenance personnel meets the allocation standard of the service work order to be assigned, the secondary release node is added to the planning trajectory sequence of other maintenance personnel who meet the allocation standard of the service work order to be assigned, thereby generating a secondary scheduling scheme and making a planning trajectory conflict judgment.

[0011] Preferably, the process of determining conflicts in the planned trajectory includes: Obtain the planned trajectory sequence from the original scheduling scheme of other maintenance personnel and the corresponding secondary release node in the secondary scheduling scheme; and extract the work order processing time and node dwell time corresponding to the secondary release node; Based on the extracted work order processing time, the work order processing time and node dwell time of adjacent trajectory nodes in the timeline of other maintenance personnel's planned trajectory sequences are matched. The transfer time between the adjacent trajectory node and the secondary release node is obtained and recorded as the previous transfer time and the next transfer time. If the sum of the node dwell time and the previous transfer time of the previous trajectory node overlaps with the work order processing time corresponding to the secondary release node, it is marked as a previous planned trajectory conflict. If the sum of the node dwell time and the next transfer time of the secondary release node overlaps with the work order processing time corresponding to the next trajectory node, it is marked as a next planned trajectory conflict. If neither the previous planned trajectory conflict nor the next planned trajectory conflict is satisfied, it is considered that the planned trajectory does not conflict.

[0012] A second aspect of the present invention also provides a property service scheduling method based on personnel digitization, which implements the aforementioned property service scheduling system based on personnel digitization, comprising: Acquire personnel information data streams for operation and maintenance personnel, service information data streams for property services, and environmental information data streams; The system performs correlation feature analysis on historical multi-source data to obtain the corresponding correlation feature vector sequence. Based on the correlation feature vector sequence and historical service scheduling schemes, it obtains the optimal matching benchmark for various work orders and maintenance personnel in the current scenario through a conditional time series fusion model. A coupled dual-graph structure is constructed based on the spatial division of property service areas and the skill relationships of operation and maintenance personnel. The real-time scheduling scheme for operation and maintenance personnel is evaluated based on the coupled dual-graph structure to obtain the corresponding trajectory anomaly evaluation value. Based on the trajectory anomaly evaluation value, a secondary rescheduling algorithm is constructed, and a secondary scheduling scheme is generated based on the secondary rescheduling algorithm to improve the quality and efficiency of property services.

[0013] Compared with the prior art, the beneficial effects of the present invention are: by constructing a complete property service scheduling system and method based on personnel digitalization, the present invention realizes the systematic integration and in-depth application of multi-source data, effectively solving the problems of reliance on human experience and insufficient scheduling optimization in the traditional scheduling mode.

[0014] Through multi-dimensional data collection and correlation feature analysis, a scientific benchmark for matching work orders with maintenance personnel was established, improving the accuracy of work order allocation and achieving reasonable matching of personnel and work orders. The construction of a coupled dual-graph structure integrates spatial topology and personnel skill correlation characteristics, providing comprehensive structural support for scheduling evaluation. The application of trajectory anomaly assessment and secondary rescheduling algorithms can promptly identify unreasonable aspects in the scheduling process and perform dynamic optimization, ensuring the feasibility and efficiency of the scheduling scheme. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0016] Figure 1 This is a schematic diagram of a module for a property service dispatch system based on personnel digitization.

[0017] Figure 2 This is a schematic diagram illustrating the steps of a property service scheduling method based on personnel digitization.

[0018] Figure 3 A flowchart for constructing a coupled dual-map structure.

[0019] Figure 4 This is a flowchart for trajectory evaluation and judgment. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] like Figure 1 As shown, the property service dispatch system based on personnel digitization is characterized by comprising: a multi-dimensional data acquisition module, a data feature analysis module, a dual-graph structure matching module, and a dispatch efficiency evaluation module. The multi-dimensional data acquisition module is used to acquire personnel information data streams from operation and maintenance personnel, service information data streams from property services, and environmental information data streams, forming multi-source data; The data feature analysis module is used to perform correlation feature analysis on historically collected multi-source data to obtain the corresponding correlation feature vector sequence. Based on the correlation feature vector sequence and historical service scheduling scheme, the optimal matching benchmark for various work orders and maintenance personnel in the current scenario is obtained through a conditional time series fusion model.

[0023] In this embodiment, the process of obtaining the associated feature vector sequence includes: Step S10: Historical multi-source data cleaning and standardization preprocessing; It should be noted that for the historically collected personnel information data stream, service information data stream, and environmental information data stream, it is necessary to take into account the special characteristics of property service data, such as missing personnel attendance records, incomplete work order information, and large fluctuations in environmental data, and perform cleaning and standardization processing respectively to remove abnormal data, fill in missing values, and unify the data format, so as to lay the foundation for subsequent feature extraction.

[0024] In this embodiment, the personnel information data stream is preprocessed. This data stream mainly includes basic information of maintenance personnel, such as age, gender, skill level, and work group; attendance data, such as arrival time, on-duty hours, and number of leave days; skill data, such as proficiency in work order types, work order handling skills, and skill assessment scores; and trajectory data, such as historical service areas and movement speed. Samples with duplicate personnel information, abnormal skill data, or more than 30% missing trajectory data are removed. For missing attendance data, the mean of personnel in the same work group and position is used to impute the missing data, as shown in the following formula: ;in, For the first The first maintenance personnel Fill-in values ​​for attendance indicators, such as on-duty hours, The number of people in the same shift and position. For the first The first person in the same work group and position The actual values ​​of the attendance indicators are used. Min-max standardization is used to map the various characteristics of personnel to the [0,1] interval to eliminate the influence of dimensions.

[0025] In this embodiment, the service information data stream is preprocessed. This data stream mainly includes basic information about property service work orders, such as work order ID, work order type, work order level, initiation time, required completion time, actual completion time, work order region, work order content, and processing results, such as completion quality score and resident satisfaction. Work orders with duplicate IDs, logical contradictions between initiation and completion times (e.g., completion time earlier than initiation time), or ambiguous region are removed. For missing completion quality scores, the median of similar work orders of the same type and level is used for imputation, as shown in the following formula: ;in, This is the imputation value for the t-th missing score report. Scoring the actual completion quality of work orders of the same type and level. This indicates that the median is used. For work order levels (1-5, with level 1 being the lowest and level 5 being the highest), completion quality scores, and processing times (i.e., the difference between initiation time and completion time), z-score standardization is used.

[0026] In this embodiment, the environmental information data stream is preprocessed. This data stream mainly includes real-time environmental parameters of the property service area, such as temperature, humidity, rainfall, wind speed, and light intensity, as well as area status, such as pedestrian traffic, vehicle traffic, and equipment operating status. The data originates from park sensors and monitoring equipment. Abnormal data caused by sensor malfunctions and missing data with a data collection interval exceeding one hour are removed. For missing environmental parameters, linear interpolation is used to fill in the gaps, as shown in the following formula: ;in, For the first Missing environmental parameter values ​​at each point in time. , The first , The actual environmental parameter values ​​at each point in time. , , These represent the acquisition times at corresponding time points. Min-max normalization is used to map various environmental parameters to... The interval formula is the same as the min-max standardized formula for personnel information data flow, ensuring that the dimensions of environmental characteristics are consistent with those of personnel and work order characteristics.

[0027] Step S11: Multi-source data temporal granularity matching and multi-dimensional feature extraction; It should be noted that, in combination with the actual scenario of property service scheduling, the hour is selected as the unified time granularity to match the time cycle of work order processing. This avoids data redundancy caused by overly fine time granularity and feature loss caused by overly coarse time granularity. The three types of preprocessed multi-source data are matched to the same hour-level time granularity, and personnel dimension features, work order dimension features, and environmental dimension features are extracted respectively.

[0028] In this embodiment, for personnel-dimensional feature extraction, the core features of a single operations and maintenance personnel within each hour are extracted using an hourly time step to form a personnel-dimensional feature vector. ,in For time steps (hours). Number the maintenance personnel. For personnel characteristics, the core features and extraction methods include: Skill matching Within this time step, the degree of matching between the types of work orders that maintenance personnel are proficient in and the types of work orders to be processed within this time period, with a value range of [0,1], and the formula is as follows. ;in This indicates the number of work order types the user is proficient in. Indicates the number of work order types to be processed; Indicates the total number of pending work order types. On-call status. : 1 indicates the user is on duty during that time step, 0 indicates the user is off duty during that time step. Processing proficiency. Calculated based on the average processing time and standard deviation of similar work orders in the past by this maintenance personnel. ,in This indicates the historical average processing time; This represents the average processing time for the same position, ranging from [0,1]. Higher values ​​indicate higher proficiency. (Movement efficiency) The ratio of the average movement speed of maintenance personnel within this time step to the average movement speed of maintenance personnel within the area. The value ranges from [0, 2], with values ​​exceeding 1 indicating higher-than-average mobility. Fatigue coefficient. : Calculated based on the continuous on-duty time prior to this time step, ,in The continuous on-duty time is measured in hours and ranges from 0 to 1. A higher value indicates a higher level of fatigue.

[0029] In this embodiment, for work order-dimensional feature extraction, the core features of all work orders to be processed within each hour are extracted, forming a work order-dimensional feature vector, with each hour as the time step. ,in The number of pending work orders within this time step is as follows: Work order urgency Based on work order level conversion, level 1 is set to 0.2, level 2 to 0.4, level 3 to 0.6, level 4 to 0.8, and level 5 to 1.0; work order density. The ratio of the number of work orders within this time step to the average number of work orders per hour in the region. The value ranges from [0,3], with values ​​exceeding 1 indicating a high density of work orders during that period; average processing difficulty Calculated based on the standard processing time corresponding to the work order type. ,in Indicates the standard duration of the corresponding work order; This represents the average standard duration of the corresponding work order, with a value range of [0,2]. A higher value indicates greater difficulty. (Regional concentration) The concentration of all work orders within this time step within the specified time step is calculated using the reciprocal of the variance. ,in For the first The area coordinates of each work order are in the range of [0,1]. The higher the value, the more concentrated the area.

[0030] In this embodiment, environmental dimension features are extracted by taking each hour as a time step, extracting the core environmental features of the property service area within that time step, and forming an environmental dimension feature vector. ,in The core environmental characteristics of the property service area during this time step are as follows: Environmental comfort Calculations based on temperature and humidity. , Indicates standardized temperature; Indicates standardized humidity; where , , The value range is [0,1]. A higher number indicates better comfort. , Weighting coefficient; Weather impact coefficient Calculations based on rainfall and wind speed , Indicates standardized rainfall; This represents standardized wind data; where , , The value ranges from [0,1], with higher values ​​indicating less impact of weather on services; , Weighting coefficients; regional activity Calculation based on pedestrian and vehicle traffic , Indicates standardized pedestrian flow; Indicates standardized traffic flow; , , The value range is [0,1]. A higher value indicates a more active region; , These are the weighting coefficients.

[0031] Step S12: Construct a multi-source data association matrix; It should be noted that, using the historical property service work order ID as the core association key, and combining the strong association relationships of work orders, personnel, areas, and environment in the property service scenario, an association mapping relationship is established between historical property service work orders and maintenance personnel, the spatial area to which the work order belongs, and the real-time environmental status. This constructs a multi-source data association matrix, enabling deep association of the three types of data.

[0032] In this embodiment, all historical work orders are represented as rows, with each row corresponding to a unique work order ID, denoted as the row index of the matrix. ;in The total number of historical work orders; with personnel characteristics, work order characteristics, and environmental characteristics as columns, denoted as the column index of the matrix. ;in The total dimension of the three types of features; Matrix elements Indicates the first The first work order corresponds to the Feature value, specific mapping rule: If the eigenvalue of the term, If the feature is a personnel feature, the value is the feature value of the maintenance personnel who processed the work order at the corresponding time step; if it is a work order feature, the value is the feature value of the work order itself; if it is an environmental feature, the value is the environmental feature value corresponding to the time step in which the work order was initiated; the multi-source data association matrix is ​​normalized.

[0033] Step S13: Obtain the cross-correlation features and the single-time-step correlation feature vector; It should be noted that, based on the constructed multi-source data association matrix and combined with the actual needs of property service scheduling, such as the matching of personnel skills with work order difficulty and the impact of environmental conditions on personnel processing efficiency, cross-correlation features of personnel, work order and environmental dimensions are mined, and personnel dimension features, work order dimension features, environmental dimension features and cross-correlation features are concatenated to generate a single time step association feature vector.

[0034] In this embodiment, cross-correlation features are obtained. These cross-correlation features mainly reflect the mutual influence relationships between the three types of basic features. Based on the actual scenario, four core cross features are mined, as follows: Personnel-work order matching degree This reflects the overall matching degree between maintenance personnel and work orders within that time step, calculated based on the correlation between personnel skill matching degree and work order processing difficulty. ,in Represents the dot product. , This represents the modulus of the corresponding vector, with a value range of [0,1]. A higher value indicates a better match; Person-Environment Fit. This reflects the work adaptability of maintenance personnel in the current environment, calculated based on the negative correlation between personnel fatigue coefficient and environmental comfort. The value ranges from [0,1], with higher values ​​indicating better compatibility; Work Order - Environmental Impact. This reflects the degree to which environmental conditions affect work order processing, calculated based on the product of the weather impact coefficient and the work order urgency. The value ranges from [0,1], with higher values ​​indicating a greater impact of the environment on work order processing; 3D Collaboration Degree This reflects the degree of coordination among three types of features: personnel, work orders, and environment. It is calculated based on the comprehensive correlation of these three basic feature vectors. The value ranges from [0,1], and the higher the value, the better the synergy.

[0035] In this embodiment, the personnel-dimensional feature vector, work order-dimensional feature vector, environmental-dimensional feature vector, and four types of cross-correlation features extracted within a single time step are sequentially concatenated to generate the correlation feature vector for a single time step. The splicing formula is as follows: in, .

[0036] It should be noted that the data fully includes the basic characteristics and interrelationships of maintenance personnel, work orders, and environment within a single time step, providing data support for subsequent prediction of the optimal matching benchmark.

[0037] Step S14: Obtain the associated feature vector sequence; In chronological order, the associated feature vectors of the same operations and maintenance personnel across all time steps (hourly granularity). Sort the data to form a corresponding sequence of associated feature vectors. ,in This is the total number of historical working hours for the operations and maintenance personnel, and .

[0038] Based on the property service scenario, the sorting of the associated feature vector sequence must meet the following requirements: the vector sequence of the same maintenance personnel is arranged continuously in time. If the maintenance personnel has no work orders to process at a certain time step, such as when they are resting or on leave, then in the associated feature vector of the corresponding time step, the personnel's on-duty status is set to 0, the work order-related features are set to 0, and the environmental features are filled in according to the actual collected values ​​to ensure the continuity and integrity of the sequence; the associated feature vector sequence is deredundant by removing vectors with completely identical feature values ​​for multiple consecutive time steps, such as when the maintenance personnel have not processed work orders for a long time and the environment has not changed, thereby reducing data redundancy and improving the efficiency of subsequent model processing.

[0039] It should be noted that the associated feature vector sequence It can fully reflect the characteristic change patterns of the operation and maintenance personnel at different time steps, different work orders, and different environments, providing high-quality input for the training of the conditional time series fusion model in the data feature analysis module, and ensuring that the model can accurately predict the optimal matching benchmark between various work orders and operation and maintenance personnel in the current scenario.

[0040] In this embodiment, the specific process of obtaining the predicted optimal matching benchmark includes: It should be noted that the predicted optimal matching benchmark is based on a conditional time series fusion model, which is used to guide the accurate matching of work orders and maintenance personnel in the current scenario. It is adapted to the actual scenarios of property services such as residential parks and commercial complexes, ensuring the practicality and accuracy of the benchmark.

[0041] Step S20: Construct the training dataset; Obtain historical service scheduling schemes for different operations and maintenance personnel and associated feature vector sequences related to the corresponding historical service scheduling schemes; and construct a training dataset by using the service quality of the corresponding historical service scheduling schemes for the operations and maintenance personnel as conditional labels and the associated feature vector sequences that match them as training data. In this embodiment, historical service scheduling schemes and corresponding associated feature vector sequences of maintenance personnel were selected over the past 12 months, covering typical property scenarios such as seasonal environmental changes, peak work orders during holidays, and staff shifts. Invalid data, such as data from periods with incomplete temporary staff replacements, canceled but unexecuted work orders, and periods with abnormal environmental data, were removed. Complete data of the entire process from work order initiation, staff allocation, to service completion were then selected, ensuring the completeness of the time steps in the corresponding associated feature vector sequences. Using the service quality of the corresponding maintenance personnel's historical service scheduling plans as a condition label, and based on the core evaluation indicators of service quality in the property scenario, a weighted summation method is used to calculate the label value. The value ranges from [0,1]. A higher value indicates better service quality. The formula is as follows: in, For the first The maintenance personnel handled the first Service quality label for each work order The completion quality score for this work order is standardized and taken as [0,1]. The response timeliness rate, which is the ratio of the actual response time to the required response time, is standardized and takes the value [0,1]. For homeowner satisfaction, after standardization, the value is [0,1]. , , The weighting coefficients are set based on the needs of the property management scenario. For example, the weight of homeowner satisfaction in residential communities is higher. , , Commercial complexes have a higher weighting for response efficiency. , , And satisfy The sequence of associated feature vectors for each operations and maintenance personnel. The corresponding service quality labels are used as training input data. As conditional labels, a sequence is matched with a set of corresponding labels to form the training dataset. ,in Total number of maintenance personnel For the first The historical work orders processed by each maintenance personnel; the training set, validation set, and test set are divided in a 7:2:1 ratio for model training, parameter tuning, and performance verification.

[0042] Step S21: Train the conditional temporal fusion model; It should be noted that the conditional temporal fusion model adopts an architecture that combines temporal feature extraction with conditional constraint fusion, adapting to the temporal characteristics of associated feature vector sequences and the conditional constraints of property service quality.

[0043] In this embodiment, the core parameters of the conditional temporal fusion model are set to be adjusted to fit the property management scenario. Specifically, the temporal feature extraction layer uses an LSTM structure with a hidden layer dimension of 64; the conditional fusion layer uses a fully connected layer to fuse temporal features with service quality labels; and the output layer has three neurons, corresponding to the predicted skill matching threshold, the predicted response time upper limit, and the predicted personnel load coefficient, respectively. The learning rate is initialized. Number of iterations Batch size The weighted mean squared error loss function is adopted, and the weights are set according to the importance of different output indicators in the property scenario. The formula is as follows: ; in, , , The first The threshold for the actual skill matching degree of each maintenance personnel, the upper limit of the actual response time, and the actual personnel load coefficient are obtained from historical scheduling data statistics. , , These are the predicted values ​​corresponding to the model; , , These are the weighting coefficients. Greater than , Greater than This is to prioritize the accuracy of skill matching threshold predictions, aligning with the core requirement of matching personnel to job roles in the property management industry. The training set is input into the model, and the Adam optimizer is used to minimize the loss function, with iterations of 10... Validate the model performance once, using the loss value of the validation set as the evaluation metric. If five consecutive validation sets show a loss value of 5, the model's performance will be evaluated. If the loss value on the validation set does not decrease, an early stopping strategy is adopted to stop training and avoid overfitting. A grid search method is then used to fine-tune the model parameters (learning rate, hidden layer dimension, batch size) to ensure that the model's prediction accuracy on the test set is ≥90%, meeting the actual accuracy requirements of property management scheduling. The specific optimization process will not be detailed here. After training is complete, the optimal model parameters are saved, resulting in the trained conditional time-series fusion model.

[0044] Step S22: Obtain the predicted skill matching threshold, predicted response time limit, and predicted personnel load coefficient for various work orders and maintenance personnel in the current scenario; In this embodiment, the associated feature vector sequence of each maintenance personnel in the current scenario (current hour time step) is obtained. The data is preprocessed to match the training data format. The preprocessed sequences are then input into the model, which outputs three initial prediction metrics: Initial prediction skill matching threshold The value range is [0,1], representing the first... Minimum skill level required for operations and maintenance personnel to efficiently handle work orders; initial predicted response time limit. : indicates the first The maximum reasonable response time for maintenance personnel in the current scenario; exceeding this time will affect service quality; initial predicted personnel load factor. The value range is [0,1], representing the first... The maximum load capacity that an operations and maintenance personnel can currently handle for a work order. 0 indicates no load, and 1 indicates full load.

[0045] Based on the real-time status of the current property scenario, the initial prediction indicators are calibrated to ensure that the baseline matches the actual scheduling needs. The calibration formula is as follows: Post-calibration skill matching threshold Adjustments based on work order urgency: ;in, The urgency of the work orders to be assigned in the current scenario is calculated as described above. The higher the urgency, the higher the skill matching threshold requirement, ensuring that urgent work orders are assigned to personnel with more suitable skills.

[0046] Upper limit of response time after calibration Adjustments will be made based on weather conditions. ;in, This is the weather impact coefficient for the current scenario, which is the calculation process of the weather impact coefficient mentioned above. The greater the weather impact, the higher the upper limit of the response time should be appropriately relaxed to match the efficiency of personnel movement in the actual environment.

[0047] Calibrated personnel load factor Adjustments will be made based on the real-time status of personnel. ;in, For the current scenario, the first The fatigue coefficient of maintenance personnel, as described in the above fatigue coefficient calculation process, indicates that the higher the fatigue level, the lower the load capacity, thus avoiding personnel overload that could lead to a decline in service quality.

[0048] The three calibrated core indicators are combined to form the first [indicator] in the current scenario. Predicted optimal matching benchmark between maintenance personnel and various work orders For different types of work orders, such as maintenance work orders, cleaning work orders, and security work orders, corresponding matching benchmarks for maintenance personnel are generated to ensure the benchmarks are relevant. For example, maintenance work orders focus on skill matching, while cleaning work orders focus on load factor.

[0049] Based on the associated feature vector sequences of different maintenance personnel, the real-time matching degree between each maintenance personnel and the service work orders to be assigned is calculated in the current real-time state. The real-time matching degree includes the real-time skill matching degree, the real-time response time, and the real-time personnel load coefficient. Based on the personnel-work order matching degree... The calculation process calculates the real-time skill matching degree between each operations and maintenance personnel and the service work orders to be assigned in the current real-time state. The real-time response time is calculated based on a comprehensive consideration of personnel mobility efficiency and environmental weather impact. ,in This indicates the average standard duration of the corresponding work order; This indicates the corresponding weather impact coefficient; Real-time mobility efficiency for operations and maintenance. The real-time personnel load factor is determined by the number of currently assigned work orders, continuous on-duty time, and work order density. ;in Indicates maintenance personnel Number of work orders accepted so far; Indicates maintenance personnel Maximum number of work orders that can be accepted per unit of time; Indicates maintenance personnel Real-time fatigue coefficient; This indicates the current work order density.

[0050] Compare real-time matching accuracy with predicted optimal matching baseline The corresponding dimensional indicators are used to calculate dimensional deviations and comprehensive deviations. The dimensional deviations include skill matching dimensional deviations, response time dimensional deviations, and personnel workload dimensional deviations. Based on the real-time matching degree and the predicted optimal matching benchmark... The absolute value of the difference is taken as the corresponding dimensional deviation, and the comprehensive deviation is calculated by weighted summation. The service work order to be assigned is assigned to the maintenance personnel with the smallest comprehensive deviation, and the corresponding dimensional deviations are used to form a primary matching deviation sequence.

[0051] The dual-graph structure matching module is used to construct a coupled dual-graph structure based on the spatial division relationship of the property service area and the skill association relationship of the operation and maintenance personnel. In this embodiment, the specific process of constructing the coupled dual-map structure includes: Step S30: Construct a directed weighted spatial topology graph; Based on the fundamental attributes of the property service area, and according to the characteristics of area division, area status, and work order area in the environmental information data stream, a directed weighted spatial topology map is constructed. ; Among them, nodes Based on the smallest management unit of the property service area as the basic node, and combined with the actual scenario of work order scheduling, the area is divided into several functional zones, such as building zones in residential parks and floor zones in commercial complexes. Each zone node corresponds to a unique identifier ID, denoted as... , , This represents the total number of nodes in the region. The basic attributes of each node include: region coordinates, region area, device density within the region, average number of work orders per hour, and real-time environmental characteristics.

[0052] side : A directed edge, used to represent the spatial connectivity between nodes in a region and the reachability of work order scheduling. If the region With the area If rapid movement of maintenance personnel is possible and there is a need for cross-regional work order scheduling, then a directed edge should be established. The direction is from the area where the maintenance personnel are currently located to the area where the work order to be dispatched is located.

[0053] Weight The edge weights are calculated using a weighted summation method, based on environmental characteristics, personnel movement efficiency, and work order characteristics. The formula is as follows: ;in For the area With the area The standardized distance between them is mapped to [0,1] using min-max normalization, with smaller weights for closer distances. This is the average weather impact coefficient between the two areas. The higher the value, the smaller the impact of weather on movement, and the greater the weight. This represents the average work order density between the two regions. A higher value indicates a more urgent need for cross-regional scheduling and a greater weight. The weight ranges from [0,1] and is used to quantify the convenience and priority of scheduling between regions.

[0054] It should be noted that constructing a spatial topology map provides a foundation for spatial matching of subsequent operation and maintenance personnel scheduling, ensuring that the map can reflect the scheduling scenarios of the actual property service area.

[0055] Step S31: Construct an undirected weighted personnel skill association graph; Taking the basic attributes of individual operations and maintenance personnel as the core, and based on the preprocessing results of personnel information data stream, personnel dimensional characteristics, and skill matching deviations in the primary matching deviation sequence, an undirected weighted personnel skill association graph is constructed. ; Among them, nodes Based on a single operations and maintenance (O&M) personnel as the basic node, each node corresponds to a unique O&M personnel ID. , recorded as , , This represents the total number of operations and maintenance personnel. The basic attributes of each node include: basic information about the operations and maintenance personnel, such as skill level and work group; core indicators in the personnel-dimensional feature vector; the predicted optimal matching benchmark; and the skill matching deviation in the primary matching deviation sequence.

[0056] side : Undirected edge, used to represent the skill association between maintenance personnel. If two maintenance personnel belong to the same shift, have overlapping expertise in work order types, or can substitute for each other, an undirected edge is established based on the skill assessment scores in the skill data. If the score difference is ≤10 points, an undirected edge is established. This reflects the feasibility of collaborative scheduling among operation and maintenance personnel and conforms to the characteristics of team and skill data in the personnel information data flow.

[0057] Weight Edge weights are used to quantify the skill similarity and collaborative compatibility between two operations and maintenance personnel. Based on skill-related indicators in the personnel dimension features, the formula is as follows: ,in The similarity of skill matching between two operations and maintenance personnel, i.e. ; This represents the average proficiency level of the two maintenance personnel. , , The proficiency levels of the two maintenance personnel are respectively. The fatigue matching coefficient for the two maintenance personnel, i.e. , These are the fatigue coefficients for two maintenance personnel. The weights range from [0,1], with higher values ​​indicating a higher similarity in skills between the two personnel and easier collaborative scheduling.

[0058] It should be noted that building a personnel skill association graph is to explore the skill associations and collaboration potential among operations and maintenance personnel, providing personnel-level support for subsequent scheduling optimization and secondary rescheduling.

[0059] Step S32: Construct a coupled dual-map structure; Based on the spatial topology map constructed above Personnel Skills Correlation Map By combining the personnel-area binding relationship in property service scheduling, i.e., the maintenance personnel's team corresponds to a fixed service area, and based on the team in the personnel information data stream and the historical service area in the trajectory data, a mapping relationship between the two is established, and finally a coupled dual-graph structure is constructed. ,in The specific process for mapping the incidence matrix is ​​as follows: Mapping and association rules need to be pre-defined to obtain a spatial topology map. Each region node With this as the core, a skill association map of related personnel. All operation and maintenance personnel nodes within the service area of ​​this region This means that the historical service area of ​​the operation and maintenance personnel includes the area, or the team to which they belong is responsible for the area, forming a one-to-many mapping relationship between area nodes and several personnel nodes.

[0060] Among them, the mapping correlation matrix for The binary mapping matrix, where Represents personnel nodes With area nodes There is a relationship between them; that is, the maintenance personnel are responsible for the service in this area. This indicates no correlation. The rows of the matrix correspond to the area nodes in the spatial topology map, and the columns correspond to the personnel nodes in the personnel skill association map, ensuring clear and traceable mapping relationships based on historical service area characteristics in the operation and maintenance personnel trajectory data.

[0061] Set coupling constraints: Add coupling constraints to ensure the rationality of the mapping association. 1. Each personnel node At least one region node is associated. This means that all maintenance personnel have a fixed service area; 2. Each area node The number of associated personnel nodes shall not be less than the minimum personnel configuration corresponding to the average hourly work orders in this area; 3. The average skill matching degree between the associated personnel nodes and the area nodes is greater than or equal to the average skill matching degree threshold in the previously predicted optimal matching benchmark, so as to avoid mismatch between personnel skills and area work order requirements.

[0062] To further explain, through the above mapping and association, the regional scheduling needs of the spatial topology map and the personnel service capabilities of the personnel skill association map are deeply coupled, forming a coupled dual-map structure that can simultaneously take into account the scheduling convenience of the spatial dimension and the skill adaptability of the personnel dimension.

[0063] Step S33: Iteratively update nodes, perform graph augmentation, and obtain the iterative matching deviation coefficient sequence;

[0064] The primary matching deviation sequence is used as the corresponding patch node in the coupled dual-graph structure. The additional attribute is the primary matching deviation corresponding to the work orders to be assigned within the area. If there are multiple work orders to be assigned within the area, the average of the primary matching deviations of each work order is taken as the additional attribute of the area node. The nodes of the coupled dual-graph structure are iteratively updated and graph-enhanced through a graph neural network (GNN). Finally, the iterative matching deviation coefficients are output and aggregated to form a sequence. The specific process is as follows: For spatial topological maps Each region node The average value of the primary matching deviation sequence of the currently unassigned work orders in this area is used as the additional attribute of this node. ,in This represents the average skill matching deviation of all work orders within the designated area. The average value of the response time deviation, This represents the average value of the personnel load deviation. This represents the average of the overall matching deviations.

[0065] Based on a graph neural network (GNN) as the core model, the number of iterations is set to 5-8 to avoid overfitting due to excessive iteration. Each iteration process includes: node feature aggregation, that is, for the spatial topological graph... Each area node Aggregate its neighboring nodes through directed edges The basic attributes, additional attributes, and edge weights of associated nodes are used to update node features through convolution operations; for personnel skill association graphs... Each personnel node Aggregate its neighboring nodes through undirected edges The basic attributes of the associated nodes and edge weights are used to update node features. This is achieved by mapping the association matrix. ,accomplish and The feature interactions between them will connect the area nodes. The updated characteristics are passed to all associated personnel nodes. This serves as a constraint for updating the characteristics of personnel nodes; [the personnel node...] The updated features are passed to its associated shard nodes. This is used to adjust the additional attributes of area nodes, achieving coordinated updates of spatial and personnel dimensions. After each iteration, based on the updated node features, the iterative matching deviation coefficient for each node is calculated. Area node matching deviation coefficient ; This represents the number of iterations. , , ; This represents the average weight of the edges adjacent to the node. This represents the average edge weight of the personnel nodes associated with this node; for personnel nodes... Personnel node matching deviation coefficient ; , , The first After the next iteration, the deviations in skills, response time, and load dimensions corresponding to this personnel node are as follows: , , .

[0066] During the iterative update process, graph enhancement operations are performed simultaneously to improve the robustness and adaptability of the graph: 1. Spatial topological map Based on real-time environmental characteristics, the edge weights are dynamically adjusted. If a certain area encounters severe weather, the corresponding edge weights are appropriately reduced to decrease the priority of cross-area scheduling. 2. Personnel skill correlation map Based on the real-time fatigue coefficient of maintenance personnel, the edge weights are dynamically adjusted. If the fatigue coefficient of a certain maintenance personnel is too high, the edge weight of the personnel node associated with it is appropriately increased to enhance the possibility of collaborative shift replacement. 3. For the mapping association matrix, the association relationship is dynamically adjusted according to the work order density. If the work orders in a certain area surge, idle maintenance personnel in other areas can be temporarily associated to meet the skill matching threshold requirements and supplement service capacity.

[0067] It should be noted that when the change in the iteration matching deviation coefficient of all nodes is ≤0.01 in two consecutive iterations and the number of iterations reaches the preset upper limit, the iteration stops and the final iteration matching deviation coefficient is output.

[0068] Organize by maintenance personnel number, and group each maintenance personnel... All associated area nodes The area node matching deviation coefficient The personnel node matching deviation coefficient of the individual. Sort by time to form an iterative matching deviation coefficient sequence for the corresponding maintenance personnel. ,in For the number of iterations, For this maintenance personnel The coefficients of the next iteration For the maintenance personnel, the associated area node number The coefficients of the next iteration.

[0069] It should be noted that this iterative matching deviation coefficient sequence fully reflects the changing pattern of matching deviations of maintenance personnel in different iterations and under different area associations. It integrates spatial topology features, personnel skill features, and primary matching deviation information, providing accurate and comprehensive input data for the trajectory anomaly assessment and secondary rescheduling algorithm construction of the subsequent scheduling efficiency evaluation module. This ensures that the secondary scheduling scheme can specifically solve the matching deviation problem in primary scheduling and improve the quality and efficiency of property service scheduling.

[0070] The scheduling efficiency evaluation module evaluates the real-time scheduling scheme of operation and maintenance personnel based on the coupled dual-graph structure, and obtains the corresponding trajectory anomaly evaluation value; based on the trajectory anomaly evaluation value, a secondary rescheduling algorithm is constructed, and a secondary scheduling scheme is generated based on the secondary rescheduling algorithm to improve the quality and efficiency of property services.

[0071] In this embodiment, the specific process of obtaining the corresponding trajectory anomaly evaluation value includes: Step S40: Extract real-time scheduling scheme trajectory-related information; In this embodiment, the spatial topology map is based on the coupled dual-map structure. Extract the core trajectory information of each operations and maintenance personnel from the real-time scheduling plan corresponding to the service work orders to be assigned, specifically including: 1. Work order processing order: Record the nodes of the spatial topology map area to which the corresponding work order belongs, forming a node sequence, according to the order in which the maintenance personnel receive the work orders; 2. Work order processing time: Record the specific timestamp when the maintenance personnel begin processing the work order corresponding to each node; 3. Node dwell time: Records the actual time spent by maintenance personnel processing work orders at each region node.

[0072] Step S41: Generate a standardized planning trajectory sequence; In this embodiment, a spatial topological map is used. Based on the node ID, and according to the extracted work order processing order, processing time, and dwell time, a standardized planning trajectory sequence is generated. Each element corresponds to a trajectory node and contains three core pieces of information: node identifier, processing time, and dwell time. Standardization processing must meet the following requirements: 1. Node ID and Spatial Topology Map The node IDs are completely identical to ensure spatial matching; 2. The processing times are arranged in ascending order of time, which conforms to the sequential logic of work order processing; 3. Standardize the dwell time to the [0,1] range to eliminate the dimensional influence of different work order processing times.

[0073] Step S42: Align iteratively match the deviation coefficient sequence with the standardized planned trajectory sequence; In this embodiment, the corresponding maintenance personnel will be... Iterative matching deviation coefficient sequence Align each node with the spatial nodes of the standardized planning trajectory sequence, following the alignment rules: 1. Each area node in the standardized planning trajectory sequence Based on this, the iterative matching deviation coefficient is used to match the nodes associated with the maintenance personnel in this area. 2. If a node in the planned trajectory sequence For multiple iteration deviation coefficients, if the node is associated with multiple iterations, the average of all corresponding coefficients is taken as the matching deviation coefficient of the node; 3. Each trajectory node has a unique corresponding iterative matching deviation coefficient.

[0074] Step S43: Calculate the trajectory score for a single node; In this embodiment, based on the aligned iterative matching deviation coefficient, and according to the dwell time of the trajectory node and the edge weights of the spatial topology graph, the deviation weighted value of each trajectory node is calculated, i.e., the single-node trajectory score. ,,in The iterative matching deviation coefficient after aligning the trajectory nodes is set to [0,1]. The smaller the coefficient, the higher the score. This represents the average weight of the edges adjacent to the node. The standardized dwell time for this node, with a value in the range [0,1], is used; a shorter dwell time results in a higher score. Single-node trajectory score. The value range is [0,1]. The higher the score, the more reasonable the trajectory planning of the node and the smaller the matching deviation.

[0075] Step S44: Calculate the trajectory anomaly assessment value; In this embodiment, a preset trajectory state reference threshold is used. Based on the actual scenario of property service dispatching, the value range is [0.5, 0.7], which can be adjusted according to the type of park, such as commercial complexes with dense work orders. Take 0.6, residential area. Set the value to 0.55 and assign a single-node trajectory score to each trajectory node. With trajectory state reference threshold By comparison, the trajectory anomaly assessment value of the corresponding operation and maintenance personnel at each trajectory node under this real-time scheduling scheme is calculated. The formula is as follows: Trajectory Anomaly Assessment Value The value range is [0,1]. This indicates that the trajectory of this node is normal. The larger the value, the more severe the trajectory anomaly. When this occurs, the node is marked as a key abnormal node and requires secondary rescheduling.

[0076] It should be noted that the trajectory anomaly evaluation value fully reflects the planning rationality of each trajectory node, integrating multi-dimensional information such as iterative matching deviation, spatial scheduling convenience, and work order processing timeliness. This provides an accurate basis for anomaly judgment for the subsequent construction of the secondary rescheduling algorithm, ensuring that the secondary scheduling can specifically solve the trajectory anomaly problem.

[0077] In this embodiment, a secondary rescheduling algorithm is constructed based on the trajectory anomaly evaluation value. The specific process of generating a secondary scheduling scheme based on the secondary rescheduling algorithm includes: Step S50: Graded processing of trajectory anomaly assessment values; In this embodiment, two levels of trajectory anomaly thresholds are preset, combined with trajectory anomaly evaluation values. The value range is used to classify each trajectory node, and the classification criteria are aligned with the priority of property service scheduling, as follows: 1. Level 1 Normal: The node's trajectory is basically reasonable, so there is no need to release the node. Only the abnormal information is recorded for subsequent scheduling optimization. 2. Level 2 anomaly: The trajectory of this node is severely unreasonable, requiring the release of this trajectory node or two or three subsequent consecutive trajectory nodes for comprehensive scheduling adjustments. The two-level trajectory anomaly thresholds can be adjusted based on actual scheduling accuracy requirements.

[0078] Step S51: Determine the secondary release node; In this embodiment, based on the above-described hierarchical processing results, for trajectory nodes exhibiting secondary anomalies, a node release operation is performed, and these are designated as secondary release nodes. Specific rules: 1. Secondary release node: Releases the current trajectory node and the next 2 or 3 consecutive trajectory nodes; 2. Secondary release nodes must be associated with corresponding work order information, including work order ID, work order type, urgency, etc., to ensure that they can be reassigned after release. Record the region node ID, work order processing time, and dwell time of the secondary release node to provide a basis for subsequent reassignment and conflict judgment.

[0079] Step S52: Secondary judgment of overall deviation by other maintenance personnel; In this embodiment, for the service work order to be assigned corresponding to the secondary release node, other maintenance personnel in the area to which the node belongs are screened in the coupled dual-graph structure. A secondary judgment is made based on the comprehensive deviation calculated by the other maintenance personnel. Judgment criteria: the comprehensive deviation of the other maintenance personnel is less than or equal to the preset comprehensive deviation threshold, which is generally taken as 0.2, and the predicted optimal matching benchmark of the maintenance personnel meets the following conditions: the skill matching degree threshold is less than or equal to the real-time skill matching degree of the work order, the upper limit of the response time is greater than or equal to the predicted real-time response time of the work order, and the personnel load coefficient is greater than or equal to the real-time personnel load coefficient of the work order. In this case, the maintenance personnel is considered to meet the allocation criteria for the service work order to be assigned.

[0080] Step S53: Generate a secondary scheduling scheme; In this embodiment, the secondary release node is added to the planned trajectory sequence of other maintenance personnel that meet the allocation criteria, based on the principle of minimizing overall deviation. The work order processing order, processing time, and node dwell time of this maintenance personnel are adjusted to generate a secondary scheduling scheme. Adjustment rules: 1. Work order processing sequence: Insert the secondary release node into the planned trajectory sequence of the maintenance personnel to ensure that the processing time increases in chronological order, which is consistent with the movement efficiency of the maintenance personnel and the spatial connectivity of the nodes in the area; 2. Processing Time: Determine the processing time of the secondary release node based on the processing completion time of the current node and the transfer time between nodes for the maintenance personnel; 3. Duration of stay: Based on the original duration of the work order, if the maintenance personnel are more proficient in handling the task, the duration of stay can be appropriately shortened to ensure that the work order is completed on time.

[0081] Step S54: Detect conflict in the planned trajectory; In this embodiment, after the secondary scheduling scheme is generated, the planning trajectory conflict judgment process is immediately triggered to check for time and space conflicts between the secondary release node and other trajectory nodes in the scheme, so as to ensure the feasibility of the secondary scheduling scheme. If there is a conflict, it is returned for adjustment; if there is no conflict, it is confirmed to be effective.

[0082] In this embodiment, the specific process of determining planned trajectory conflicts includes: Step S60: Extract the information required for conflict determination of the planned trajectory; In this embodiment, the planned trajectory sequence in the original scheduling scheme of other maintenance personnel receiving the secondary release node is obtained; the secondary release nodes corresponding to other maintenance personnel in the secondary scheduling scheme are extracted, and the work order processing time, node dwell time, and spatial topology map of the secondary release node are obtained. The parameters required for calculating the transfer time between adjacent nodes include the distance between nodes and the efficiency of personnel movement.

[0083] Step S61: Match adjacent trajectory nodes and calculate the transition time; In this embodiment, based on the planned processing time of the secondary release node, two adjacent trajectory nodes on the timeline are matched in the original planned trajectory sequence received by the maintenance personnel, and denoted as the previous trajectory node and the subsequent trajectory node. The transfer time between the secondary release node and the previous and subsequent trajectory nodes is calculated and denoted as the previous transfer time. and subsequent transfer time The forward transfer time is calculated based on the ratio of the actual distance between the previous trajectory node and the secondary release node to the real-time movement efficiency of the maintenance personnel. The forward transfer time is calculated based on the ratio of the actual distance between the subsequent trajectory node and the secondary release node to the real-time movement efficiency of the maintenance personnel. The unit of transfer time is consistent with the duration of stay.

[0084] Step S62: Determine conflict in the pre-planned trajectory; In this embodiment, the processing end time of the previous trajectory node is calculated. If the sum of the node dwell time of the previous trajectory node and the previous transfer time is greater than or equal to the work order processing time corresponding to the secondary release node, it is marked as a previous planned trajectory conflict.

[0085] It should be noted that a pre-planned trajectory conflict refers to a situation where the maintenance personnel, after completing the work order at the pre-trajectory node, are unable to arrive before the planned processing time at the secondary release node, resulting in a time conflict and requiring adjustments to the scheduling plan.

[0086] Step S63: Post-planning trajectory conflict judgment; In this embodiment, the processing end time of the secondary release node is calculated. If the sum of the node dwell time and the subsequent transfer time of the secondary release node is greater than or equal to the work order processing time corresponding to the subsequent trajectory node, it is marked as a subsequent planned trajectory conflict.

[0087] It should be noted that a post-planning trajectory conflict refers to a situation where the maintenance personnel, after completing the work order at the secondary release node, are unable to arrive before the planned processing time of the post-trajectory node, resulting in a time conflict and requiring adjustments to the scheduling plan.

[0088] Step S64: Conflict determination result assessment; In this embodiment, if both the pre-planned trajectory conflict and the post-planned trajectory conflict are met simultaneously, it is considered a serious time conflict, and other qualified maintenance personnel need to be selected to allocate the secondary release node. If only one type of conflict is met, it is considered a minor time conflict, and the processing time of the secondary release node is adjusted, such as delaying the processing time corresponding to the pre-conflict or advancing the processing time corresponding to the post-conflict, and the transfer time is recalculated and judged again. If neither the pre-planned trajectory conflict nor the post-planned trajectory conflict is met, it is considered that the planned trajectory does not conflict, and the secondary scheduling scheme can be taken effect directly.

[0089] Step S65: Supplementary conflict resolution; In this embodiment, if a planned trajectory conflict exists and the adjustment process cannot resolve it, and the work order completion time is still exceeded after adjustment, the process returns to the secondary release node allocation stage to re-select other qualified maintenance personnel. If there are no other qualified maintenance personnel in the area, a mapping correlation matrix based on the coupled dual-graph structure can be used. Temporarily associate idle maintenance personnel from other areas to meet the skill matching threshold requirements, ensuring that secondary release nodes can be allocated in a timely manner and avoiding work order delays.

[0090] It should be noted that by judging the planned trajectory conflicts as described above, the feasibility and rationality of the secondary scheduling plan are ensured, and delays in work order processing and a decline in service quality due to trajectory conflicts are avoided. Ultimately, the scheduling plan is optimized, the efficiency and accuracy of property service scheduling are improved, and a complete closed loop is formed with the system architecture and functions of each module mentioned above.

[0091] like Figure 2 As shown, a property service scheduling method based on personnel digitization includes: Acquire personnel information data streams for operation and maintenance personnel, service information data streams for property services, and environmental information data streams; The system performs correlation feature analysis on historical multi-source data to obtain the corresponding correlation feature vector sequence. Based on the correlation feature vector sequence and historical service scheduling schemes, it obtains the optimal matching benchmark for various work orders and maintenance personnel in the current scenario through a conditional time series fusion model. A coupled dual-graph structure is constructed based on the spatial division of property service areas and the skill relationships of operation and maintenance personnel. The real-time scheduling scheme for operation and maintenance personnel is evaluated based on the coupled dual-graph structure to obtain the corresponding trajectory anomaly evaluation value. Based on the trajectory anomaly evaluation value, a secondary rescheduling algorithm is constructed, and a secondary scheduling scheme is generated based on the secondary rescheduling algorithm to improve the quality and efficiency of property services.

[0092] Optionally, in this embodiment, those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0093] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0094] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, 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 storage medium and includes several instructions to cause one or more electronic devices to execute all or part of the steps of the methods described in the various embodiments of this application.

[0095] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0096] In the several embodiments provided in this application, it should be understood that the disclosed application can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of units or modules may be electrical or other forms.

[0097] 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 this embodiment according to actual needs.

[0098] Furthermore, the functional units in the various embodiments of this application 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.

[0099] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A property service dispatch system based on personnel digitization, characterized in that, The system includes: a multi-dimensional data acquisition module, a data feature analysis module, a dual-graph structure matching module, and a scheduling efficiency evaluation module; The multi-dimensional data acquisition module is used to acquire personnel information data streams from operation and maintenance personnel, service information data streams from property services, and environmental information data streams, forming multi-source data; The data feature analysis module is used to perform correlation feature analysis on historically collected multi-source data to obtain the corresponding correlation feature vector sequence. Based on the correlation feature vector sequence and the pre-acquired historical service scheduling scheme, the optimal matching benchmark for various work orders and maintenance personnel in the current scenario is obtained through the conditional time series fusion model. The dual-graph structure matching module is used to construct a coupled dual-graph structure based on the spatial division relationship of the property service area and the skill association relationship of the operation and maintenance personnel. The scheduling efficiency evaluation module evaluates the real-time scheduling scheme of operation and maintenance personnel based on the coupled dual-graph structure, and obtains the corresponding trajectory anomaly evaluation value; based on the trajectory anomaly evaluation value, a secondary rescheduling algorithm is constructed, and a secondary scheduling scheme is generated based on the secondary rescheduling algorithm to improve the quality and efficiency of property services.

2. The property service dispatch system based on personnel digitization according to claim 1, characterized in that, The process of obtaining the associated feature vector sequence includes: Data cleaning and standardization preprocessing were performed on the historically collected personnel information data stream, service information data stream, and environmental information data stream. The three types of preprocessed multi-source data were matched with the same time granularity, and personnel dimension features, work order dimension features, and environmental dimension features were extracted respectively. Using the historical property service work order ID as the core association key, an association mapping relationship was established between historical property service work orders and maintenance personnel, the spatial area to which the work order belongs, and the real-time environmental status, and a multi-source data association matrix was constructed. Based on the multi-source data association matrix, cross-correlation features of personnel dimension, work order dimension, and environmental dimension were mined, and personnel dimension features, work order dimension features, environmental dimension features and cross-correlation features were concatenated to generate a single time step association feature vector. The association feature vectors of all time steps of the same maintenance personnel were sorted in chronological order to form a corresponding association feature vector sequence.

3. The property service dispatch system based on personnel digitization according to claim 2, characterized in that, The process of obtaining the predicted optimal matching baseline includes: Obtain historical service scheduling schemes for different operation and maintenance personnel and associated feature vector sequences related to the corresponding historical service scheduling schemes; use the service quality of the corresponding historical service scheduling schemes for operation and maintenance personnel as conditional labels and the associated feature vector sequences related to the corresponding historical service scheduling schemes as training data to construct a training dataset; complete the training of the conditional temporal fusion model based on the training dataset; input the associated feature vector sequences of different operation and maintenance personnel in the current scenario into the trained conditional temporal fusion model, and output the predicted skill matching degree threshold, predicted response time upper limit, and predicted personnel load coefficient of various work orders and operation and maintenance personnel in the current scenario, thereby forming the predicted optimal matching benchmark; Based on the associated feature vector sequence of different operation and maintenance personnel, the real-time matching degree between each operation and maintenance personnel and the service work order to be assigned is calculated in the current real-time state. The real-time matching degree is compared with the corresponding dimension index of the predicted optimal matching benchmark. The dimension-level deviation and the comprehensive deviation are calculated. The service work order to be assigned is assigned to the operation and maintenance personnel with the smallest comprehensive deviation. The corresponding dimension-level deviations are then used to form a primary matching deviation sequence.

4. The property service dispatching system based on personnel digitization according to claim 3, characterized in that, The process of constructing a coupled dual-map structure includes: A directed weighted spatial topology graph is formed based on the basic attributes of the property service area; an undirected weighted personnel skill association graph is formed based on the basic attributes of a single maintenance personnel; a mapping relationship is established between the area nodes of the spatial topology graph and the personnel skill association graphs of several maintenance personnel corresponding to the area nodes in the personnel skill association graph, thus constructing a coupled dual-graph structure. The primary matching deviation sequence is used as an additional attribute of the corresponding region node in the coupled dual-graph structure. The nodes of the coupled dual-graph structure are iteratively updated and graph-enhanced through a graph neural network. The enhanced iterative matching deviation coefficient of each node is output. The iterative matching deviation coefficients of the corresponding nodes are collected according to the corresponding maintenance personnel to form the iterative matching deviation coefficient sequence of the corresponding maintenance personnel.

5. The property service dispatch system based on personnel digitization according to claim 4, characterized in that, The process of obtaining the corresponding trajectory anomaly assessment value includes: Based on the spatial topology graph in the coupled dual-graph structure, the work order processing order, work order processing time, and node dwell time of each maintenance personnel in the real-time scheduling scheme corresponding to the service work order to be assigned are extracted, and a standardized planning trajectory sequence is generated according to the spatial node order. The iterative matching deviation coefficient sequence of the corresponding maintenance personnel is aligned one by one with the spatial nodes of the planning trajectory sequence, and the deviation weighting value of each trajectory node is calculated to obtain the trajectory score of a single node. Based on the preset trajectory status benchmark threshold, the trajectory anomaly evaluation value of the corresponding maintenance personnel at each trajectory node under the real-time scheduling scheme is calculated.

6. The property service dispatch system based on personnel digitization according to claim 5, characterized in that, Based on the trajectory anomaly assessment value, a secondary rescheduling algorithm is constructed. The process of generating a secondary scheduling scheme based on the secondary rescheduling algorithm includes: The trajectory anomaly assessment value is graded and processed according to the preset three-level trajectory anomaly threshold; based on the graded processing result, the next one or more trajectory nodes in the real-time scheduling plan of the operation and maintenance personnel are released and recorded as secondary release nodes. A secondary judgment is made on the comprehensive deviation of other maintenance personnel in the corresponding area node. If the comprehensive deviation of other maintenance personnel meets the allocation standard of the service work order to be assigned, the secondary release node is added to the planning trajectory sequence of other maintenance personnel who meet the allocation standard of the service work order to be assigned, thereby generating a secondary scheduling scheme and making a planning trajectory conflict judgment.

7. The property service dispatch system based on personnel digitization according to claim 3, characterized in that, The process of determining conflicts in the planned trajectory includes: Obtain the planned trajectory sequence from the original scheduling scheme of other maintenance personnel and the corresponding secondary release node in the secondary scheduling scheme; and extract the work order processing time and node dwell time corresponding to the secondary release node; Based on the extracted work order processing time, the work order processing time and node dwell time of adjacent trajectory nodes in the timeline of other maintenance personnel's planned trajectory sequences are matched. The transfer time between the adjacent trajectory node and the secondary release node is obtained and recorded as the previous transfer time and the next transfer time. If the sum of the node dwell time and the previous transfer time of the previous trajectory node overlaps with the work order processing time corresponding to the secondary release node, it is marked as a previous planned trajectory conflict. If the sum of the node dwell time and the next transfer time of the secondary release node overlaps with the work order processing time corresponding to the next trajectory node, it is marked as a next planned trajectory conflict. If neither the previous planned trajectory conflict nor the next planned trajectory conflict is satisfied, it is considered that the planned trajectory does not conflict.

8. A property service scheduling method based on personnel digitization, implementing the property service scheduling system based on personnel digitization as described in any one of claims 1 to 7, characterized in that, include: Acquire personnel information data streams for operation and maintenance personnel, service information data streams for property services, and environmental information data streams; Perform correlation feature analysis on historical multi-source data to obtain the corresponding correlation feature vector sequence; Based on the associated feature vector sequence and historical service scheduling scheme, the optimal matching benchmark for various work orders and maintenance personnel in the current scenario is obtained through a conditional time series fusion model. A coupled dual-graph structure is constructed based on the spatial division of property service areas and the skill relationships of operation and maintenance personnel. The real-time scheduling scheme for operation and maintenance personnel is evaluated based on the coupled dual-graph structure to obtain the corresponding trajectory anomaly evaluation value. Based on the trajectory anomaly evaluation value, a secondary rescheduling algorithm is constructed, and a secondary scheduling scheme is generated based on the secondary rescheduling algorithm to improve the quality and efficiency of property services.

Images