Service personnel scheduling optimization method and system based on intelligent terminal

Abstract

The invention relates to the technical field of service management, in particular to a service personnel scheduling optimization method and system based on an intelligent terminal. The method comprises the following steps: collecting user behavior track data and service historical data through an intelligent terminal, extracting regional service request features, and dividing service space partitions; and service personnel data are obtained, a service capability portrait is constructed, and task requirements are accurately matched. Based on regional service features and service space partition, service personnel data are subjected to priority division, and efficient task allocation is ensured. And furthermore, in combination with the behavior track and the ability portrait of the service personnel, cross-regional collaborative scheduling is optimized, and the resource utilization rate is improved. According to the invention, customer demands and service personnel can be accurately matched, dynamic optimization scheduling is realized, the operation efficiency of the foot bath city is improved, the scheduling cost is reduced, and the customer satisfaction is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of service management, and in particular to a method and system for optimizing service personnel scheduling based on an intelligent terminal. Background Art

[0002] With the rapid development of smart terminals, the Internet of Things (IoT), artificial intelligence (AI) and big data analysis technologies, the information and intelligent management of the service industry has become the key to improving operational efficiency and user experience. In places such as foot baths, massage clubs, and health care centers, the scheduling and management of service personnel has a vital impact on the company's operational efficiency, customer satisfaction, and service quality. However, the existing scheduling methods still have many problems in the actual operation process, which makes it difficult to meet the needs of the modern service industry, and more efficient and intelligent optimization solutions are urgently needed. At present, most foot baths and massage and health care industries still rely on traditional manual scheduling methods, such as service personnel manually arranged by front desk staff, and assigned based on customer needs, room availability, and technician availability, resulting in low utilization of service personnel and an imbalance in technicians being busy or idle for a long time. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a service personnel scheduling optimization method and system based on intelligent terminals to solve at least one of the above technical problems.

[0004] The present application provides a service personnel scheduling optimization method based on a smart terminal, the method comprising: Obtain user behavior trajectory data and user service history data through the smart terminal, and perform regional service request feature extraction and service space partition division based on the user behavior trajectory data and user service history data to obtain regional service request feature data and service space partition data respectively; Obtain service personnel data, and construct a service capability portrait based on the service personnel data to obtain service capability portrait data; According to the regional service request feature data and the service space partition data, the service personnel data is divided into personnel allocation priority groups to obtain personnel allocation priority data; Obtain service personnel behavior trajectory data, and conduct cross-regional collaborative scheduling of personnel allocation priority data based on service personnel behavior trajectory data and service capability portrait data to obtain service personnel scheduling optimization data.

[0005] In the present invention, user behavior trajectory data and user service history data are obtained in real time through intelligent terminals to achieve accurate identification of user service needs. By extracting regional service request features and dividing service space partitions, the service demand characteristics of different regions can be accurately portrayed, so that the scheduling system can optimize personnel allocation in a targeted manner and improve service response efficiency. A service capability portrait is constructed based on service personnel data, and service personnel are comprehensively evaluated from multiple dimensions such as skill level, work experience, and service quality, so that scheduling decisions can match the most suitable service personnel and improve the overall service quality. By dividing the personnel allocation priorities based on service personnel data, service personnel scheduling is made more reasonable, avoiding the waste of personnel resources and improving operational efficiency.

[0006] The present invention combines the behavior trajectory data of service personnel and uses a cross-regional collaborative scheduling mechanism to optimize personnel allocation, so that service personnel can provide services across regions within a reasonable range, alleviate the imbalance between service supply and demand in certain regions, and improve overall resource utilization. The present invention can effectively improve the intelligent level of service personnel scheduling, improve customer satisfaction, and reduce operating costs, providing related industries with more efficient and accurate scheduling solutions.

[0007] Optionally, the obtaining of user behavior trajectory data and user service history data through a smart terminal, and performing regional service request feature extraction and service space partition division according to the user behavior trajectory data and the user service history data, to obtain regional service request feature data and service space partition data respectively, includes: Obtain user behavior trajectory data and user service history data through smart terminals; Perform activity area density clustering according to user service history data to obtain user service history activity area clustering data; Perform trajectory classification based on user service historical activity area clustering data to obtain historical activity trajectory data; Fitting is performed based on historical activity trajectory data and user behavior trajectory data to obtain user behavior trajectory fitting data; Extract the stay duration feature and the activity period feature according to the historical activity trajectory data, and obtain the stay duration feature data and the activity period feature data respectively; Acquire service space data, and divide the service space data into service spaces according to the residence duration characteristic data and the activity period characteristic data to obtain service space partition data; Extract valley-peak features based on the dwell time feature data to obtain valley-peak feature data of the dwell time; According to the peak-valley characteristic data of the dwell time, regional service request characteristics are extracted from the service space partition data and the user service history data to obtain regional service request characteristic data.

[0008] The present invention obtains user behavior trajectory data and user service history data, and combines the activity area density clustering method to effectively identify the user's high-frequency activity area and accurately analyze the user's historical service demand distribution. Trajectory classification based on historical activity trajectory data is helpful to discover the user's service usage pattern and improve the accuracy of personalized service recommendation. Through the fitting of historical activity trajectory data and user behavior trajectory data, the user's potential behavior path can be dynamically predicted to improve the ability to perceive user needs in advance. The feature extraction of residence time and the feature extraction of activity period can identify the user's service demand preferences in different time periods, so that the system can reasonably divide the service space, optimize the distribution of service resources, and reduce resource waste or supply and demand imbalance caused by unreasonable regional division. Through the extraction of the valley peak feature of the residence time, the present invention can identify the peak and valley periods of service demand, making the extraction of regional service request features more targeted, which helps to reasonably allocate service personnel, alleviate the supply and demand pressure during peak periods, and can also optimize the scheduling strategy and improve service efficiency.

[0009] Optionally, performing trajectory classification according to user service historical activity area clustering data to obtain historical activity trajectory data includes: Perform regional classification based on user service historical activity regional clustering data to obtain historical activity regional classification data; Based on the historical activity area classification data, the residence time distribution calculation, the visit frequency calculation and the periodic behavior extraction are performed to obtain the historical residence time distribution data, the historical visit frequency data and the historical periodic behavior data respectively; Extract trajectory features based on historical stay time distribution data, historical visit frequency data, and historical periodic behavior data to obtain historical trajectory feature data; According to the historical trajectory feature data, the user service history data is clustered to obtain the historical activity trajectory data.

[0010] In the present invention, by performing regional classification on user service history activity area clustering data, the different activity areas of users are accurately divided, making data analysis more structured and helping to extract the service demand characteristics of different areas. By calculating the historical residence time distribution, access frequency, and periodic behavior extraction, the present invention can comprehensively characterize the behavioral characteristics of users in different areas. For example, residence time distribution data can identify the length of time a user stays in a specific area, access frequency data reflects the user's preference for certain services, and periodic behavior data can reveal the user's service usage patterns in a specific time period.

[0011] By extracting trajectory features from the above data, the user's historical trajectory pattern becomes clearer, thus achieving more accurate behavior analysis. Trajectory clustering based on historical trajectory feature data can effectively identify different types of user trajectory patterns, allowing the scheduling system to formulate more targeted scheduling strategies for different user groups.

[0012] Optionally, the acquiring of service space data and dividing the service space data into service space partitions according to the residence duration characteristic data and the activity period characteristic data to obtain service space partition data include: Get service space data; According to the characteristic data of the length of stay and the characteristic data of the activity period, the service space data is used to identify the stay points, and the service space stay point data is obtained; Calculate multi-scale dwell time based on the service space stay point data to obtain multi-scale dwell time data; Construct a stay map based on multi-scale stay duration data to obtain stay map data; Calculate the activity of each time period according to the stay graph data to obtain the activity data of the stay period; Perform time period clustering according to the activity data of the stay period to obtain time period clustering data; The service space data is labeled according to the time period clustering data to obtain the service space partition data.

[0013] In the present invention, through the identification of stay points, the system can accurately capture the user's stay behavior in different service areas, ensure that the spatial division can reflect the real service demand distribution, and avoid resource waste or supply and demand imbalance caused by static partitioning. By adopting multi-scale stay duration calculation, the stay rules of users at various stay points can be analyzed at different time granularities, so that the spatial partition can take into account short-term hot spots and long-term trends, ensuring the adaptability and accuracy of spatial division. Based on the construction of the stay graph, a structured regional stay pattern can be formed to further improve the interpretability of service demand. The time period activity calculation combines the user's activity level in different time periods, so that the system can identify high-demand time periods and improve the timeliness of spatial division. By optimizing the time series characteristics through time period clustering, the system can more accurately extract service hotspots in different time periods and provide data support for dynamic scheduling. By marking the service space data, intelligent and adaptive spatial division is realized, so that the scheduling system can dynamically adjust the service strategy, improve service response efficiency, optimize resource allocation, enhance user experience, and ensure the efficiency and flexibility of the overall operation.

[0014] Optionally, performing multi-scale residence time calculation according to the service space residence point data to obtain multi-scale residence time data includes: Calculate the continuous residence time and the stationary residence time according to the service space stay point data, and obtain the continuous residence time data and the stationary residence time data respectively, wherein the stationary residence time data is the total time that the user stays at the corresponding stay point in the business premises; Perform time granularity aggregation based on the continuous residence time data and the dwelling point residence time data to obtain the duration granularity data and the dwelling point time granularity data; Generate multi-scale residence time data based on duration granularity data and dwell time granularity data to obtain preliminary multi-scale residence time data; Perform position entropy calculation and stay stability calculation on the preliminary multi-scale stay duration data to obtain position entropy data and stay stability data respectively; According to the position entropy data and the stay stability data, the preliminary multi-scale stay duration data is weightedly fused to obtain the multi-scale stay duration data.

[0015] In the present invention, through the calculation of continuous residence time and stationary residence time, the user's stay characteristics in different environments can be identified respectively, especially the stationary residence time data can reflect the service demand intensity of the user in the business premises, and provide strong support for optimizing service resource scheduling. Through time granularity aggregation, the present invention can analyze the user's stay behavior at different time scales, making the service demand forecast more accurate, while avoiding information loss caused by too large or too small time scale. In addition, using position entropy calculation and stay stability calculation, the stability of the user's stay behavior can be quantified, so that the scheduling system can identify high-value user behavior patterns and improve the rationality of resource allocation. By optimizing multi-scale stay duration data through attention weighted fusion, the credibility and applicability of the data are improved, thereby providing a more accurate decision-making basis for intelligent service scheduling, improving service matching efficiency and user satisfaction.

[0016] Optionally, the dwell time valley peak feature data includes first local peak valley data and second local peak valley data, and the valley peak feature extraction is performed according to the dwell time feature data to obtain the dwell time valley peak feature data, including: Perform local extreme value detection based on the residence time characteristic data to obtain the first local peak and valley data; The residence time threshold is sorted and screened according to the residence time feature data to obtain the second local peak and valley data.

[0017] In the present invention, the first local peak-valley data is obtained through local extreme value detection, which can effectively identify the short-term change trend in the dwell time data, so that the system can dynamically perceive the instantaneous fluctuation of service demand, thereby optimizing the real-time scheduling strategy. By using the dwell time threshold sorting and screening to extract the second local peak-valley data, abnormal data can be effectively filtered, and the robustness of peak-valley identification can be improved, thereby ensuring that the scheduling system accurately allocates resources during peak periods, reasonably adjusts the service personnel configuration during trough periods, and optimizes the overall resource utilization.

[0018] Optionally, the acquiring of service personnel data and constructing a service capability portrait based on the service personnel data to obtain the service capability portrait data includes: Get service personnel data; Perform multi-dimensional capability numerical mapping based on service personnel data to obtain multi-dimensional capability numerical data; A capability profile is constructed based on multi-dimensional capability numerical data to obtain service capability profile data.

[0019] The present invention can quantify the ability characteristics of different service personnel by acquiring service personnel data and mapping multi-dimensional ability values, including skill level, service experience, work efficiency and other dimensions, so that personnel scheduling can be optimized based on objective data. Constructing service capability portraits based on multi-dimensional ability value data can intuitively display the core capabilities of each service personnel, so that the scheduling system can match the most suitable service personnel according to different task requirements, and improve the rationality of task allocation and service quality.

[0020] Optionally, the personnel allocation priority data is divided according to the regional service request feature data and the service space partition data to obtain the personnel allocation priority data, including: Perform regional service demand matching according to regional service request feature data, service space partition data, and service personnel data to obtain regional service demand matching data; Optimize service response time according to regional service demand matching data to obtain service response time data; Optimize staff utilization based on service response time data to obtain staff utilization data; Extract the current regional deployment level based on the service response time data and personnel utilization data to obtain the current regional deployment level data; Extract the adjacent area dispatching level based on the service response time data and the personnel utilization rate data to obtain the adjacent area dispatching level data; Extract the cross-regional dispatch candidate layer based on the service response time data and personnel utilization data to obtain the cross-regional deployment layer data; Priority levels are generated based on the current area allocation layer data, adjacent area allocation layer data, and cross-area allocation layer data to obtain personnel allocation priority data.

[0021] In the present invention, through regional service demand matching, the service demand intensity of different regions can be accurately identified, and intelligent matching can be performed in combination with service personnel data to ensure the reasonable allocation of service resources and improve scheduling accuracy. By optimizing service response time and personnel utilization, the work efficiency of service personnel can be maximized while ensuring rapid response, avoiding waste of personnel resources and making overall operations more efficient. Through multi-level deployment schemes, including current area, adjacent area and cross-regional scheduling level extraction, the present invention can flexibly adjust personnel scheduling strategies locally and globally, so that the system can dynamically deploy service personnel according to actual needs, alleviate local service resource shortages, and improve overall service stability.

[0022] Optionally, the acquiring of service personnel behavior trajectory data, and performing cross-regional collaborative scheduling of personnel allocation priority data according to the service personnel behavior trajectory data and service capability portrait data to obtain service personnel scheduling optimization data, includes: Obtain service personnel behavior trajectory data; According to the service personnel behavior trajectory data and service capability portrait data, the task redistribution benefit is calculated based on the personnel allocation priority data to obtain the task redistribution benefit data; Calculate regional service satisfaction according to current regional deployment layer data and regional service request feature data to obtain regional service satisfaction data; When it is determined that the regional service satisfaction data is less than or equal to the preset service satisfaction threshold data, cross-regional collaborative scheduling is performed according to the task redistribution income data to obtain service personnel scheduling optimization data; When it is determined that the regional service satisfaction data is greater than the preset service satisfaction threshold data, the current regional deployment layer data is determined as the service personnel scheduling optimization data.

[0023] In the present invention, by obtaining the behavior trajectory data of service personnel and combining the service capability portrait data to calculate the task redistribution benefit, it is possible to accurately evaluate the adaptability of different service personnel in cross-regional scheduling, ensure that the scheduling decision complies with the principle of overall benefit optimization, and improve resource utilization. The regional service satisfaction calculation is based on the current regional deployment layer data and regional service request feature data, and can monitor the service supply and demand balance in the region in real time to ensure that when the service satisfaction in the region is lower than the threshold, the system can automatically trigger cross-regional scheduling to optimize personnel allocation and improve the overall service quality. By setting the service satisfaction threshold, the system can flexibly adjust the scheduling strategy. When the regional service satisfaction is higher than the threshold, the scheduling plan of the region is used first to reduce unnecessary personnel flow and improve scheduling stability; when the satisfaction is lower than the threshold, the system performs cross-regional collaborative scheduling based on the task redistribution benefit data to ensure efficient matching of service needs and reduce user waiting time.

[0024] Optionally, the present application further provides a service personnel scheduling optimization system based on a smart terminal, which is used to execute the service personnel scheduling optimization method based on a smart terminal as described above, and the service personnel scheduling optimization system based on a smart terminal includes: The service demand perception module is used to obtain user behavior trajectory data and user service history data through the intelligent terminal, and extract regional service request features and divide the service space into zones according to the user behavior trajectory data and user service history data, and obtain regional service request feature data and service space partition data respectively; A service capability portrait construction module is used to obtain service personnel data and construct a service capability portrait based on the service personnel data to obtain service capability portrait data; A service personnel allocation priority division module is used to divide the service personnel data into personnel allocation priorities according to the regional service request feature data and the service space partition data to obtain personnel allocation priority data; The service personnel scheduling optimization module is used to obtain service personnel behavior trajectory data, and to perform cross-regional collaborative scheduling of personnel allocation priority data based on the service personnel behavior trajectory data and service capability portrait data to obtain service personnel scheduling optimization data.

[0025] The purpose of the present invention is to obtain user behavior trajectory data and user service history data in real time through an intelligent terminal, and accurately identify the service demand patterns of different regions based on regional service request feature extraction and service space partitioning. By constructing a service capability portrait, the system can quantitatively analyze multidimensional data such as the skill level, work experience, and historical service performance of service personnel, and form a visualized capability label, so that the scheduling decision is not only based on geographical location matching, but also combined with personnel capability characteristics to achieve optimal personnel allocation. Personnel allocation priority division can be carried out according to the demand pressure, resource supply and personnel load status of different regions, and the personnel scheduling strategy can be reasonably adjusted to ensure that high-demand areas are met first, reduce personnel vacancy rate, and optimize overall operational efficiency. Combined with the service personnel behavior trajectory data, historical trajectory pattern analysis and real-time trajectory monitoring are used to achieve cross-regional collaborative scheduling. When the regional service satisfaction is lower than the preset threshold, the system can dynamically adjust the personnel distribution based on the task redistribution benefit calculation to achieve the optimal configuration of service resources. It effectively alleviates the problem of imbalance between local service supply and demand, improves the adaptability and flexibility of the system, and ensures that efficient service support can still be provided in a complex business environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings: Figure 1 A flowchart showing a method for optimizing service personnel scheduling based on a smart terminal according to an embodiment of the present invention is provided; Figure 2 A flowchart showing the steps of a service demand perception method according to an embodiment is shown; Figure 3 A flowchart showing a method for constructing a service capability profile according to an embodiment of the present invention is shown; Figure 4 A flowchart showing the steps of a method for assigning priorities to service personnel according to an embodiment of the present invention is shown; Figure 5 A flowchart showing the steps of a service personnel scheduling optimization method according to an embodiment is shown; The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0027] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0028] In addition, the drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0029] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0030] See also Figures 1 to 5 The present application provides a service personnel scheduling optimization method based on an intelligent terminal, the method comprising: S1. Obtain user behavior trajectory data and user service history data through a smart terminal, and perform regional service request feature extraction and service space partition division based on the user behavior trajectory data and user service history data to obtain regional service request feature data and service space partition data respectively; In one embodiment, the user's location data is collected through a smart terminal (such as a mobile phone, tablet or wearable device), and the location is tracked by GPS, WiFi signals, Bluetooth beacons, etc. Combined with the user's time dimension, the length of stay, movement path, and visit frequency in different areas are analyzed. The user trajectory data is processed using a sliding window method to obtain user behavior trajectory data. The user's historical service consumption is recorded, including service type, frequency, duration, evaluation and other information, to build a user consumption portrait, extract the user's preference characteristics, and calculate the formula: ,in For users For service type The preference weight of For users Past consumer services The frequency of is the total number of users, The number of users who have this type of consumption records.

[0031] Partition user behavior trajectory data and service history data: ,in is the cost function of clustering, which describes the sum of the distances between all data points in the partition and the center point of each region. For the A collection of user data for a region. is the number of preset partitions, is the feature vector of specific user trajectory or service data, For the A collection of user data for a region. For the The feature vector mean or cluster center of each area. The user's stay time and consumption preference vector (for example, a combination of foot therapy, massage, and rest preferences) are used as clustering features. Through repeated iterative calculations, the user data is automatically divided into multiple functional space types such as foot therapy space, massage space, tea break space, and transfer area (such as front desk, changing room). For example, cluster analysis shows that the average stay time of users in the foot therapy area (area 1) is significantly longer than that in other areas; the stay time in the lounge (area 2) is moderate but the frequency is higher; the front desk area (area 3) is characterized by a shorter user stay time but a very high visit frequency, reflecting that this area is a typical transitional functional space.

[0032] Identify high-frequency service demand areas based on user service history data: , For space point is the center and the radius is The set of data points contained in the neighborhood of is the user trajectory data point, is another user trajectory data point, It is a collection of spatial data points of all functional areas within the Foot Bath City. User trajectory data points and The Euclidean distance between is a given neighborhood radius (spatial distance threshold). If more than 20 user resident data points are gathered within a radius of 5 meters near the foot massage service area, the system automatically determines that this is a high-frequency service demand hotspot. The foot massage city manager / system will analyze the demand saturation level of the area based on this and reasonably allocate service personnel resources.

[0033] S2. Obtain service personnel data, and construct a service capability profile based on the service personnel data to obtain service capability profile data; In one embodiment, the location, working hours, order status, skill level, customer rating and other information of the service personnel are collected through the intelligent terminal. The behavior pattern of the service personnel is modeled: , is the hidden state of the current time step, indicating the current capability characteristics of the service personnel, To activate the function, ensure that the data is mapped to the appropriate range. is the historical state weight matrix, is the hidden state of the previous time step, i.e., the historical capability characteristics of the service personnel, is the input data weight matrix, is the input feature of the current time step (including location, order status, user rating, etc.), is a bias term that adjusts the balance of network output. Construct a service capability profile, such as setting a service skill vector: ,in Representative service skills, such as massage skills, communication skills, customer satisfaction, etc. , for The weight of Rate the overall capabilities of the service.

[0034] In one embodiment, various data of service personnel are collected in real time through smart terminals (such as mobile phones, tablets, smart bracelets, etc.) to build their ability portraits. The collected information includes but is not limited to location information, which collects the distribution of service personnel's work locations and active areas based on GPS or WiFi. Working hours, record the service personnel's daily working hours and idle time to evaluate their workload. Order situation, statistics on the service personnel's historical order quantity, order type and completion rate, to measure their service capabilities. Skill level, combined with training records, calculates the technician's mastery of different skills, such as massage techniques, foot therapy skills, etc. Customer rating, analyzes the quality of its service based on user feedback (such as user reviews, return visit data). Analyze the behavior patterns of service personnel through time series modeling to predict their future service capability development trends.

[0035] ,in Representative Service skill ratings, including Training duration is calculated by counting the course duration (unit: hours) that service personnel complete on the skill training platform to measure their skill learning; Calculate the service personnel's cumulative working years (unit: year) to assess their accumulated experience. The order completion rate is calculated by calculating the proportion of completed orders of the service personnel to the total number of orders received, which measures their fulfillment ability. To score user satisfaction, you can use a camera to capture the user's expression after the service and analyze their satisfaction; or analyze the emotional tendency in the user's evaluation (such as "very satisfied", "average", "bad") and convert it into a quantitative score; In order to cope with the peak period, the technician's order acceptance rate during peak hours is counted to evaluate his work efficiency when busy. In a large foot massage shop, technicians A and B need to optimize service matching to ensure that highly skilled technicians provide services in high-end rooms, while novice technicians are responsible for ordinary services. The training time, order completion rate, and user satisfaction score of technicians A and B are monitored in real time through smart terminals to calculate the service capability scores of both. Technician A's data: training time: 50 hours, working time: 3 years, order completion rate: 98%, user satisfaction score (based on image recognition and text analysis): 4.8 / 5, peak period coping ability: 85%; set weights =(0.2, 0.2, 0.2, 0.2, 0.2). Technician B's data: training time: 30 hours, working time: 1 year, order completion rate: 92%, customer satisfaction score: 4.5 / 5, peak response capability: 75%, set the same weight. Technician A's score =48.76Higher than Technician B's score =40.3, so technician A is more suitable for high-end foot massage services, while technician B is more suitable for ordinary massage services.

[0036] S3, dividing the service personnel data into personnel allocation priorities according to the regional service request feature data and the service space partition data, and obtaining personnel allocation priority data; In one embodiment, the influence weight is determined, including the service demand matching degree ( ), service personnel skill level ( ), agent idle state ( ), expected return ( ). Calculate the priority score: ,in Assign priority data to people, is the matching weight, For the matching degree, is the ability score weight, Score your abilities. is the idle status weight data of the service personnel, The idle status of service personnel. In a large foot massage city, technicians need to be intelligently scheduled during peak order periods to maximize service quality and revenue. There are multiple service technicians to choose from, but the scheduling priority needs to be determined based on the service demand matching degree, skill score, and idle status. The technician with the highest priority score is assigned to the order.

[0037] Service matching degree (calculated based on user preferences and service capabilities): , For users Demand preferences for services (calculated based on historical consumption records and user ratings), For service personnel A competency score for a skill (based on training, experience, user feedback, etc.).

[0038] S4. Obtain service personnel behavior trajectory data, and perform cross-regional collaborative scheduling of personnel allocation priority data based on the service personnel behavior trajectory data and service capability portrait data to obtain service personnel scheduling optimization data.

[0039] In one embodiment, cross-region collaborative scheduling optimization: ,in For service personnel For customers Proceeds from providing services, Indicates whether to assign the person (1 for assigned, 0 for unassigned). When order demand changes, adjust the allocation strategy: , Status Next action The Q value, The current distribution of orders on each floor (e.g., 1F has dense orders, 2F has fewer orders), is an optional action, i.e., the technician cross-floor scheduling strategy. For immediate rewards, if technician scheduling optimizes order completion rate, Take a positive value; if the dispatch causes the technician to wait for a long time or the dispatch between floors is excessive, then Take negative value, is the discount factor, which controls the influence of long-term returns and takes a value between 0 and 1. For the next state The best choice among all possible actions.

[0040] In the same foot massage shop, different floors have different service function spaces. For example, the first floor mainly receives short-term foot massage customers with high turnover; the second floor mainly provides advanced massage, and customers stay longer; the third floor and above have VIP rooms or boxes, with longer service time but lower order frequency. Due to the different order demand characteristics on different floors, how to efficiently dispatch service personnel, optimize technician allocation strategies, and improve overall operational efficiency has become a key challenge.

[0041] Inside the foot spa, each technician is equipped with a smart terminal (such as RFID, Bluetooth beacon, WiFi positioning, etc. on the work badge) to record their work trajectory. The collected data includes the current floor (such as 1F, 2F, 3F); the type of service currently being performed (foot therapy, massage, VIP exclusive service, etc.); the time when the last service was completed; and the availability status (whether it is idle). In addition, combined with the technician's historical service records, a service capability portrait is constructed, including the service categories they are good at (such as massage, foot therapy, essential oil massage, etc.); the average service time; and customer evaluation feedback.

[0042] Determine the optimal matching solution between technicians and orders, and calculate the revenue value of each technician i when serving customer j: Cross-regional collaborative scheduling optimization: ,in For service personnel For customers The profit from providing services, the profit calculation includes the matching degree between technicians and orders (based on the service capability profile); the dispatch cost of technicians to the floor where the order is located (moving across floors takes time); and the urgency of the customer's needs (whether it is a VIP appointment). Indicates whether the person is assigned (1 for assigned, 0 for unassigned). , Status Next, execute the action (Dispatching a technician to a certain floor) value (such as increased task completion rate, reduced customer waiting time, etc.), The current distribution of orders on each floor (e.g., 1F has dense orders, 2F has fewer orders), is an optional action, i.e., the technician cross-floor scheduling strategy. For instant rewards, calculate order completion rate and technician utilization rate, is the discount factor used to balance short-term and long-term benefits (a preset parameter, set to 0.5), is the optional action in the next state, is the maximum value in the next state, For the next state value, The next state is the new floor state formed after the technician is dispatched, such as: "1F technicians have been reinforced, orders have decreased; 2F technicians have decreased but orders remain unchanged", etc., including new order distribution and technician distribution. The floor order distribution and technician idle state are used as system states ; Choose whether to dispatch a technician to a different floor, i.e. The value is {stay on this floor, dispatch to other floors}; the reward mechanism is to successfully complete the order: reward +1; technicians are idle for a long time: reward −0.5; technicians cause order timeouts due to cross-floor dispatch: reward −1. Input the current floor order situation and output the recommended dispatch strategy; simulate training in historical data, adjust the Q value, and gradually optimize the dispatch strategy; continuously update the Q value in actual operation to achieve intelligent dispatch.

[0043] Currently, there are many orders on 1F, but the number of technicians is insufficient; there are fewer orders on 3F, and the technician idle rate is high; some technicians need to be deployed from 3F to 1F. The execution process is to obtain the technician's behavior trajectory; record the technician's current floor and idle status; calculate the order allocation income; calculate the current technician's income under different order allocations. If cross-floor scheduling, the income needs to be subtracted from the inter-floor movement cost. Run DQN prediction to determine whether to dispatch technicians from 3F to 1F; execute technician scheduling; if the scheduling is beneficial to the overall order fulfillment rate, assign the technician to the new floor.

[0044] Optionally, the obtaining of user behavior trajectory data and user service history data through a smart terminal, and performing regional service request feature extraction and service space partition division according to the user behavior trajectory data and the user service history data, to obtain regional service request feature data and service space partition data respectively, includes: S11, obtaining user behavior trajectory data and user service history data through smart terminals; In one embodiment, GPS, WiFi, and Bluetooth beacons are used to obtain the user's location trajectory and collect user service history data, including past consumption records, visit frequency, consumption amount, service duration, etc.

[0045] S12, performing activity area density clustering according to the user service history data to obtain user service history activity area clustering data; In one embodiment, , For point In the neighborhood radius The point set inside For database For any data point in A collection of all data points of historical activities for user services, For point With point The spatial distance between is the neighborhood radius, calculate the point In the neighborhood If the point set in satisfies a certain density (at least MinPts points), then Belongs to clustering. Generate user service history activity area clustering data, that is, the set of long-term stay points of users in a specific area. The foot bath city includes foot therapy space and massage space. User A visited the foot therapy space 15 times in the past month, and each stay time exceeded 30 minutes. The location was identified as a long-term service demand point for the user by the clustering algorithm. User B only visited the massage space twice in the past month, which did not meet the minimum density requirement, so it was not identified as a long-term stay point.

[0046] S13, performing trajectory classification according to user service historical activity area clustering data to obtain historical activity trajectory data; In one embodiment, the similarity of each trajectory is calculated: , For the trajectory and The similarity value between is the discrete sampling point index on the trajectory, For the trajectory In the The horizontal coordinate of the sampling point, For the trajectory In the The horizontal coordinate of the sampling point, For the trajectory In the The vertical coordinate of the sampling point, For the trajectory In the The vertical coordinate of the sampling point, is the total number of trajectory sampling points. User A moves between foot therapy space X and foot therapy space Y, forming a trajectory User B moves between foot therapy space Y and massage space Z, forming a trajectory . Calculate the similarity S(T1, T2) between the two. If the value is small, they belong to the same travel mode.

[0047] Perform trajectory clustering to classify user trajectories: , is the clustering objective function, which indicates the clustering quality, For the Cluster indexes, is the set number of clusters, that is, the number of trajectory categories for classification, is the user trajectory data point, For the Clusters, is the center trajectory of the cluster.

[0048] Table 1: Typical trajectory categories S14, performing fitting according to the historical activity trajectory data and the user behavior trajectory data to obtain user behavior trajectory fitting data; In one embodiment, trajectory curve fitting is performed: , is the fitted trajectory curve, which represents the mathematical model of the user behavior trajectory. is a constant term, indicating the initial offset of the trajectory, is the first polynomial coefficient, indicating the slope trend change of the trajectory, is the time or space coordinate data, is the second polynomial coefficient, which indicates the trend change of the rate of change of the trajectory. are polynomial coefficients, representing the trend change of the trajectory, and the fitting error is calculated: , is the trajectory fitting error, which indicates the deviation between the fitting curve and the actual trajectory. is the trajectory point index, is the total number of trajectory sampling points, For the The coordinates of the user's real trajectory points, For the fitting curve The predicted value at is used to obtain the fitting data of the user behavior trajectory.

[0049] S15, extracting the stay duration feature and the activity period feature according to the historical activity trajectory data, and obtaining the stay duration feature data and the activity period feature data respectively; In one embodiment, the user's residence time in different areas is calculated: , is the total time a user spends in a certain area. is the index of the number of visits to the area, is the total number of visits by the user in this area. The time when the user leaves the area. The time when the user enters the area. Extract the activity time characteristics, specifically calculate the number of visits in different time periods: , For users in the time period The percentage of visits within For the period The number of visits within is the total number of visits.

[0050] S16, obtaining service space data, and dividing the service space data into service space according to the residence duration characteristic data and the activity period characteristic data to obtain service space partition data; In one embodiment, the service demand density between regions is calculated: , is the regional service demand density (service demand per unit area), is the user's index in this area, is the total number of users in the region, For the The length of stay of users, is the area of ​​the region. Regional division: , For point The distance to the nearest service point, For existing service points, is the set of all seed points (initial service points), is the regional point that needs to be allocated. Set the seed point set of Foot Spa City A, lobby service point (technician A, technician B); VIP room service point (technician C); rest area service point (technician D). Calculate the distance from each user in the lobby to the nearest technician: if the distance from user U1 to technician A is 2 meters, and the distance to technician B is 3 meters, then the nearest technician is A. If the distance from user U2 to technician A is 5 meters, and the distance to technician B is 2 meters, then the nearest technician is B. According to the nearest neighbor principle, user U1 is assigned to technician A, and U2 is assigned to technician B to ensure balanced service.

[0051] S17, extracting valley-peak features according to the dwell time characteristic data to obtain valley-peak feature data of the dwell time; In one embodiment, the peak value of the dwell time in the time series is calculated: , is the highest value (peak value) of the user's stay time in the time series, The length of time a user stays in a certain period of time. is the time index, To observe the time range, calculate the valley value of the residence time in the time series: , is the lowest value (valley value) of the user's stay time in the time series. The length of time a user stays in a certain period of time. is the time index, The observation time range.

[0052] S18. Extract regional service request features from the service space partition data and the user service history data according to the dwell time valley-peak feature data to obtain regional service request feature data.

[0053] In one embodiment, , calculate the time period The service request density within Represents the request density during this time period. Representative period The number of requests within Indicates the total duration of user stay in this time period.

[0054] Extract regional service request feature data and vectorize the above features, or, , A comprehensive service request characteristic for the region (used to measure the overall demand in the region), is the time period index, is the total number of all statistical time periods, For time period weights (adjusted according to business needs, such as higher weights during peak hours), For time period service request density.

[0055] Optionally, performing trajectory classification according to user service historical activity area clustering data to obtain historical activity trajectory data includes: Perform regional classification based on user service historical activity regional clustering data to obtain historical activity regional classification data; In one embodiment, the service activity areas of different users are classified to identify high-frequency service areas. , is the clustering objective function, which measures the compactness of the clustering results. is the cluster index, is the set number of clusters (i.e. the number of regional classifications), The user's historical activity area data points (i.e. the user's geographic location), For the clusters (i.e., a classified service area), is the center point of the cluster, indicating the center position of the area. Record the user's access data in foot therapy space X, massage space Y, and SPA center Z. Each user's access data point (longitude, latitude, access frequency) is input as a data set. Set the number of clusters =3, divided into high-frequency area, medium-frequency area, and low-frequency area. Calculate the distance from each data point to the cluster center and update the cluster center. Distinguish the areas according to the frequency of visits to the cluster center. High-frequency area (C1): Areas frequently visited by users (such as near foot massage space X). Medium-frequency area (C2): Areas occasionally visited by users (such as massage space Y). Low-frequency area (C3): Areas rarely visited by users (such as SPA center Z).

[0056] Based on the historical activity area classification data, the residence time distribution calculation, the visit frequency calculation and the periodic behavior extraction are performed to obtain the historical residence time distribution data, the historical visit frequency data and the historical periodic behavior data respectively; In one embodiment, the user is calculated in the area Total stay time , For users in the area The total stay time in The access index for users entering this area. is the total number of visits by users in this area. For user The time you leave the area. For user The time when the user enters the area. Internal access area The frequency, , For the period inner area The frequency of visits, For users in the time period Visit area The number of times The total number of all visits by the user. Analyze the periodicity of user access patterns: , For user access behavior in the cycle The strength of is the time index, is the total time span of the data, For time The number of visits within is an imaginary unit, is the circumference constant, is the periodic frequency for analysis. User C visits SPA center Z 5 times on Friday, 4 times on Saturday, and 3 times on Sunday. Calculate the periodic behavior characteristics. If the visit data shows a fixed peak every Friday, it means that the user has a periodic pattern.

[0057] Extract trajectory features based on historical stay time distribution data, historical visit frequency data, and historical periodic behavior data to obtain historical trajectory feature data; In one embodiment, a trajectory feature vector is constructed: , is the trajectory feature vector, is the residence time characteristic, To access the frequency feature, It is characterized by periodic behavior.

[0058] According to the historical trajectory feature data, the user service history data is clustered to obtain the historical activity trajectory data.

[0059] In one embodiment, the optimal alignment distance between two trajectories T1 and T2 is calculated , For the trajectory and The optimal alignment distance between is one of the two trajectories, is the other of the two trajectories, To minimize the function, is the index of the trajectory point, Another one for the trajectory point index, is the total number of sampling points on the trajectory, For the trajectory On Track points, For the trajectory On Track points, For trajectory points and Calculate the spatial distance between two trajectory clusters and Minimum distance between , For two trajectory clusters and The distance between Trajectory cluster The trajectory points inside is one of two different trajectory clusters (i.e., a collection of multiple trajectories), Trajectory cluster The trajectory points inside is another of two different trajectory clusters (i.e., a collection of multiple trajectories), For trajectory points and The distance between.

[0060] Optionally, the acquiring of service space data and dividing the service space data into service space partitions according to the residence duration characteristic data and the activity period characteristic data to obtain service space partition data include: Get service space data; In one embodiment, the user's activity data at different locations is collected, including geographic coordinate data, timestamp, and visit duration. The user's spatial activity data is collected through GPS, WiFi positioning, and Bluetooth beacons.

[0061] According to the characteristic data of the length of stay and the characteristic data of the activity period, the service space data is used to identify the stay points, and the service space stay point data is obtained; In one embodiment, the user's stay points in different areas, that is, the spatial locations where the user stays for a long time, are identified, specifically, the stay point identification is performed as follows: ,in It is the data of the service space stop point. is the trajectory point, is the set of all trajectory points, for The distance between is the neighborhood radius (set to 5-10m).

[0062] Calculate multi-scale dwell time based on the service space stay point data to obtain multi-scale dwell time data; In one embodiment, the residence time distribution at different time granularities (minutes, hours, days) is calculated. , For time scale The average length of stay in is the time scale (e.g. 5 minutes, 30 minutes, 1 hour), The index item for the user entering and exiting the area, indicating the user's Entering the area for the first time, The time when the user leaves the area. The time when the user enters the area. is the number of visits in this time scale.

[0063] Construct a stay map based on multi-scale stay duration data to obtain stay map data; In one embodiment, a space-time stay network is constructed by connecting different stay points, and the distance threshold between the stay points is set. : , For stopover and The distance between For stopover The horizontal axis of For stopover The horizontal axis of For stopover The vertical coordinate of For stopover The vertical coordinate of . , then establish the edge: , For the stop map, For the stop point collection, is the set of edges between the stop points.

[0064] Calculate the activity of each time period according to the stay graph data to obtain the activity data of the stay period; In one embodiment, the activity level of the stay point in different time periods is calculated, and the frequency of visiting a stay point in time period t is calculated: , For stopover In the period The activity within For the period access The number of times The total number of all visits.

[0065] Perform time period clustering according to the activity data of the stay period to obtain time period clustering data; In one embodiment, user activity patterns in different time periods, such as peak periods and trough periods, are identified by clustering the time periods. , is the time period clustering objective function, which measures the tightness of clustering. is the cluster index, is the number of clusters, that is, the number of categories for time period classification, is the activity data point of the time period, For the clusters (i.e., a type of user active time period), is the center point of the cluster, indicating the average activity of this time period. Users' service needs have different patterns at different times of the day. For example, midday is a high-demand period for light massage, while night is a high-demand period for deep therapy. Cluster analysis is performed on activity data of different time periods to identify peak periods, trough periods, and regular active periods.

[0066] The service space data is labeled according to the time period clustering data to obtain the service space partition data.

[0067] In one embodiment, the service space is labeled with different functional areas according to the time period clustering data. The service space is labeled as high activity area, medium activity area, and low activity area: , For Region Active categories, For Region In the period The activity within is the first activity threshold, such as 0.25, is the second activity threshold, , such as 0.15.

[0068] Optionally, performing multi-scale residence time calculation according to the service space residence point data to obtain multi-scale residence time data includes: Calculate the continuous residence time and the stationary residence time according to the service space stay point data, and obtain the continuous residence time data and the stationary residence time data respectively, wherein the stationary residence time data is the total time that the user stays at the corresponding stay point in the business premises; In one embodiment, the continuous residence time is calculated by calculating the continuous residence time of the user in a certain area. , The continuous time a user stays in a certain area. The time when the user leaves the area. The time when the user enters the area. The dwell time is calculated by calculating the cumulative dwell time of the user at different stopover points in the business premises. , For users at the station The total residence time, The number of times the user visits the site. for, For user The time of leaving the station, For user The time to enter the station for the first time.

[0069] Perform time granularity aggregation based on the continuous residence time data and the dwelling point residence time data to obtain the duration granularity data and the dwelling point time granularity data; In one embodiment, the residence time is aggregated at different time granularities (eg, 5 minutes, 30 minutes, 1 hour). , For time scale The polymerization residence time under is the index of a single stay, is the total number of stay data points at this time scale, is the single stay time, For time scale Analyze user behavior patterns in different time ranges and perform calculations on different time scales. User A’s stay data: 15 minutes (weight 0.2), 30 minutes (weight 0.1), 60 minutes (weight 0.05), calculate 5-minute aggregated stay time, suitable for short-term user analysis, such as quick massage and foot therapy services. 5-minute data reflects the average service time of short-term customers and can be used to optimize quick service items. User B’s stay data: 15 minutes (weight 0.1), 45 minutes (weight 0.2), 90 minutes (weight 0.15), calculate 30-minute aggregated stay time, 30-minute data reflects the main service time of users, which can be used to optimize the time design of standard packages. User C’s stay data: 60 minutes (weight 0.2), 90 minutes (weight 0.3), 120 minutes (weight 0.5), calculate 1-hour aggregated stay time, 1-hour data reflects the average service time of long-term users, which can be used to optimize premium packages and membership services.

[0070] Generate multi-scale residence time data based on duration granularity data and dwell time granularity data to obtain preliminary multi-scale residence time data; In one embodiment, the dwell time data at different time scales are calculated , is the average residence time under multiple scales, is the number of time scales selected, is the time scale index, For time scale The length of stay below.

[0071] Perform position entropy calculation and stay stability calculation on the preliminary multi-scale stay duration data to obtain position entropy data and stay stability data respectively; In one embodiment, location entropy (H) is used to measure the user's visit distribution at different stop points, that is, the degree of dispersion of user behavior. The larger the location entropy, the more dispersed the user's stop locations are, and they prefer multi-region activities; the smaller the location entropy, the more concentrated the user's stop behavior is. The user's visit distribution entropy at different locations is calculated to measure the dispersion of user behavior. , is the position entropy, is the stop point index, is the number of different stopover points, For users to access The stay stability (S) is used to measure the time stability of users at different stay points, that is, whether they frequently switch their stay locations or whether they have a fixed stay time pattern. , To maintain stability, is the number of stop points visited by users, is the stop point index, For the The residence time, is the mean of all residence times.

[0072] According to the position entropy data and the stay stability data, the preliminary multi-scale stay duration data is weightedly fused to obtain the multi-scale stay duration data.

[0073] In one embodiment, the location entropy data, the stay stability data and the preliminary multi-scale stay duration data are combined to calculate the multi-scale stay duration data through weighted fusion to optimize the service resource allocation. , is the multi-scale residence time data, is the position entropy weighting coefficient, is the position entropy, is the weighted coefficient of stay stability, To maintain stability, is the initial multi-scale residence time weighting coefficient, This is preliminary multi-scale dwell time data. Add technicians in areas with longer fused multi-scale dwell time to improve service efficiency. Provide a more efficient technician shift mechanism in areas with many short-stay users to reduce waiting time. If user A stays for 18.93 minutes on the fused multi-scale data, a 20-minute quick treatment package can be recommended. If user C's dwell time is as high as 26.67 minutes, a 30-minute deep SPA plan can be recommended. If the fused multi-scale dwell time in some areas increases significantly, the technician schedule can be adjusted to improve the utilization of service resources.

[0074] Optionally, the dwell time valley peak feature data includes first local peak valley data and second local peak valley data, and the valley peak feature extraction is performed according to the dwell time feature data to obtain the dwell time valley peak feature data, including: Perform local extreme value detection based on the residence time characteristic data to obtain the first local peak and valley data; In one embodiment, local extreme values ​​in the user stay time data, namely local peak values ​​(peaks) and local valley values ​​(valleys), are calculated. The local extreme values ​​are calculated, and the local peak values ​​satisfy: , For time point Length of stay at is the residence time at the previous time point, and the local valley value satisfies: .

[0075] In one embodiment, a first-order derivative detection is used: , For time point The first derivative (rate of change) of the dwell time at , For time point The second derivative of the residence time at (trend of change), , generate the first local peak-valley data (preliminary screened dwell time peak and valley values). If the first-order derivative is zero and the second-order derivative is less than zero, it is a peak. If the first-order derivative is zero and the second-order derivative is greater than zero, it is a valley.

[0076] The residence time threshold is sorted and screened according to the residence time feature data to obtain the second local peak and valley data.

[0077] In one embodiment, the stable peak and valley values ​​are further screened by sorting by the dwell time threshold. : , The length of stay is Axuan data. For the Duration of stay, is the threshold of significant peaks, is the threshold of significant valley value. Sort all peak and valley values ​​by duration from large to small: , is the extreme value data sorted by the length of stay. is a sorting function that arranges the stay time from large to small and selects the top k most significant peaks and valleys: .

[0078] Optionally, the acquiring of service personnel data and constructing a service capability portrait based on the service personnel data to obtain the service capability portrait data includes: S21. Obtain service personnel data; In one embodiment, various information of service personnel is collected, including basic information (age, gender, work experience); historical performance data (customer ratings, order completion rate, complaint rate); skill data (service category, professional certification, training records); working hours (average daily working hours, peak workload). Data is collected through smart terminals, work management systems, and customer evaluation systems.

[0079] S22, performing multi-dimensional capability numerical mapping according to the service personnel data to obtain multi-dimensional capability numerical data; In one embodiment, the service personnel's ability characteristics are converted into quantifiable multi-dimensional numerical data. For example, work experience (unit: year), service time (unit: hour), training time (unit: hour), number of orders (unit: time), customer feedback score (unit: 0-100), average service time (unit: minute), specific skill certification (unit: level, such as level 1-5). (1) Experience value mapping, setting the mapping relationship between work experience and ability score: 0-1 year: low experience (value range: 10-20), 1-3 years: medium experience (value range: 20-40), 3-5 years: high experience (value range: 40-60), more than 5 years: rich experience (value range: 60-100). (2) Training time mapping, setting the mapping relationship between training time and ability score: 0-10 hours: basic training (value range: 10-30), 10-50 hours: advanced training (value range: 30-60), more than 50 hours: professional training (value range: 60-100). (3) Customer feedback score mapping, set the customer feedback score to 0-100 points, and directly use it as the ability score. (4) Number of orders mapping, set the impact of the number of orders: 0-100 orders: novice (value range: 10-30), 100-500 orders: skilled (value range: 30-60), 500 orders and above: senior (value range: 60-100). (5) Skill level mapping, set the standard of skill level: Level 1 (beginner): value 20, Level 2 (intermediate): value 40, Level 3 (advanced): value 60, Level 4 (senior): value 80, Level 5 (expert): value 100.

[0080] S23. Construct a capability profile based on the multi-dimensional capability numerical data to obtain service capability profile data.

[0081] In one embodiment, the capability profile of service personnel is generated by radar chart, cluster analysis and other methods. Assume that the capability dimension set is: , is work experience (years), is the training duration (hours), is the number of orders received (times), Rate the customer (0-100), is the skill level (1-5), and the capability vector of each service staff is: , is the ability value of the service personnel in the work experience dimension, is the ability value of service personnel in the dimension of training time, is the service personnel's ability value in terms of the number of orders received. is the ability value of the service personnel in the customer rating dimension, It is the ability value of the service personnel in the skill level dimension.

[0082] Optionally, the personnel allocation priority data is divided according to the regional service request feature data and the service space partition data to obtain the personnel allocation priority data, including: S31, performing regional service demand matching according to regional service request feature data, service space partition data, and service personnel data to obtain regional service demand matching data; In one embodiment, the matching degree of the needs in the region is calculated to ensure that the skills and availability of the service personnel are adapted to the needs of the current region, and the matching score is calculated: , To match regional service needs, is the service category index, is the total number of service categories, For the The demand intensity of the service request, Rate the skills of service personnel (such as massage, pedicure, etc.).

[0083] S32, optimizing the service response time according to the regional service demand matching data to obtain service response time data; In one embodiment, the service response time is calculated: , for, is the service personnel’s current distance from the customer, is the average moving speed of the service personnel (a preset value), The average task allocation time of service personnel. Service response time refers to the total time from when a customer makes a service request to when the service personnel arrives at the customer's location. Select available technicians from those with shorter response times to improve dispatch efficiency. If a technician's response time is too long, adjust the task allocation method or re-dispatch the order.

[0084] S33, optimizing personnel utilization rate according to the service response time data to obtain personnel utilization rate data; In one embodiment, the personnel utilization rate is calculated: , is the staff utilization rate, The actual working time of the service personnel. The utilization rate is used to measure the proportion of the service personnel's actual working time to their available working time. If U is close to 1, it means that the technician is busy and needs to reduce dispatch or increase rest time. If U is too low, it means that the technician has a long idle period, and the shift can be optimized to improve the utilization rate. If the utilization rate is close to 1 (such as above 0.93), reduce the technician's shift to avoid excessive fatigue. If the utilization rate is low (such as below 0.5), adjust the technician's work schedule to reduce idle time.

[0085] S34, extracting the current regional deployment level according to the service response time data and the personnel utilization rate data to obtain the current regional deployment level data; In one embodiment, the deployment level within the current area is calculated to ensure that personnel are preferentially dispatched within the current area, and the priority dispatch index within the current area is calculated: , Deploy layer data for the current region, is the weight of the regional service demand matching degree, The matching degree of regional service needs (measure the matching degree between technicians’ skills and the needs of the region), is the weight of service response time, The service response time of the technician (seconds). The maximum response time (in seconds) allowed in the current region, used to standardize the response time. is the weight of staff utilization, It is the staff utilization rate (measurement of the technicians' work saturation level).

[0086] S35, extracting the adjacent area allocation layer according to the service response time data and the personnel utilization rate data to obtain the adjacent area allocation layer data; In one embodiment, the dispatching priority of service personnel in adjacent areas is calculated, and when the current area cannot meet the demand, they are deployed from nearby areas. The dispatching priority of adjacent areas is calculated: , Allocation of layer data for neighboring areas, Deploy layer data for the current area (measure the adaptability of technicians in the current area), is the distance from the service personnel’s neighboring area to the current area, The maximum allowed cross-region dispatch distance.

[0087] S36, extracting the cross-regional scheduling candidate layer according to the service response time data and the personnel utilization rate data to obtain the cross-regional deployment layer data; In one embodiment, the priority of cross-regional deployment is calculated. When the adjacent area cannot meet the demand, personnel are dispatched from a more distant area. The cross-regional dispatch priority is calculated: , To deploy layer data across regions, The priority index allocated to the neighboring area is the scheduling priority of the neighboring area calculated previously. is the natural exponential term, is the distance for cross-region dispatching, This is the cross-regional scheduling attenuation factor (to control the impact of long-distance scheduling), for example, 20-50m.

[0088] S37. Generate a priority level based on the current area allocation layer data, the adjacent area allocation layer data, and the cross-area allocation layer data to obtain personnel allocation priority data.

[0089] In one embodiment, the current area, adjacent areas and cross-regional dispatching levels are integrated to construct a data group or data structure for storage and generate personnel allocation priorities. Alternatively, when the current area cannot meet the demand, priority is given to dispatching from adjacent areas. If the demand still cannot be met, further cross-regional dispatching is performed from further areas (taking the maximum value of the current area dispatching layer data, adjacent area dispatching layer data and cross-regional dispatching layer data).

[0090] Optionally, the acquiring of service personnel behavior trajectory data, and performing cross-regional collaborative scheduling of personnel allocation priority data according to the service personnel behavior trajectory data and service capability portrait data to obtain service personnel scheduling optimization data, includes: S41, obtaining service personnel behavior trajectory data; In one embodiment, trajectory data is collected through GPS, WiFi positioning, and Bluetooth beacons to form a spatiotemporal trajectory set; S42, calculating the task redistribution benefit of the personnel allocation priority data according to the service personnel behavior trajectory data and the service capability portrait data, and obtaining the task redistribution benefit data; In one embodiment, the benefits of cross-regional dispatch of service personnel are calculated to ensure that the overall benefits are maximized after dispatch optimization. , is the total benefit of task redistribution, is the task sequence item, is the total number of tasks, For the task The service value (measured by business indicators after the task is completed, such as user retention rate, repurchase rate or system evaluation indicators after the service is completed), The time required for the task, It is the cross-regional dispatching cost (such as time loss, dispatching management cost, etc.).

[0091] S43, calculating the regional service satisfaction according to the current regional deployment layer data and the regional service request characteristic data to obtain the regional service satisfaction data; In one embodiment, the current personnel dispatching scheme in the area is calculated to meet the proportion of user needs, to ensure that the needs in the area are met first, and to calculate the service satisfaction in the area: , For regional service satisfaction, is the number of orders that have been matched under the current scheduling plan, is the total order quantity in this area.

[0092] S44, when it is determined that the regional service satisfaction data is less than or equal to the preset service satisfaction threshold data, cross-regional collaborative scheduling is performed according to the task redistribution income data to obtain service personnel scheduling optimization data; In one embodiment, when the regional service satisfaction is lower than the set threshold, cross-regional collaborative scheduling is performed, and the service satisfaction threshold is set. , , then cross-region collaborative scheduling is performed based on the task redistribution revenue data. Calculate the cross-region scheduling priority: , It is the priority of cross-region dispatch. The higher the value, the more suitable the technician is for cross-region dispatch. Redistribute the benefits of the task, that is, the benefits that can be brought by dispatching the technician, It is the distance between the technician's area and the current area (unit: meter), reflecting the cost of cross-area scheduling. , Redistribute revenue for tasks, is the task sequence item, is the total number of tasks, For the task The service value (the duration of the order task is multiplied by the task weight, and the task weight is obtained by mapping the task content through the system), For the task Total service time, including the technician's actual service time and travel time, For the task The dispatching cost (including the time cost of technicians dispatching across regions and the time cost of customers waiting)

[0093] S45. When it is determined that the regional service satisfaction data is greater than the preset service satisfaction threshold data, the current regional deployment layer data is determined as service personnel scheduling optimization data.

[0094] In one embodiment, it is determined whether the service satisfaction of the current area has reached a set threshold. If the satisfaction is high enough, , there is no need to perform cross-region scheduling, and the scheduling plan of the current area can be directly used as the optimization plan.

[0095] Optionally, the present application further provides a service personnel scheduling optimization system based on a smart terminal, which is used to execute the service personnel scheduling optimization method based on a smart terminal as described above, and the service personnel scheduling optimization system based on a smart terminal includes: The service demand perception module is used to obtain user behavior trajectory data and user service history data through the intelligent terminal, and extract regional service request features and divide the service space into zones according to the user behavior trajectory data and user service history data, and obtain regional service request feature data and service space partition data respectively; A service capability portrait construction module is used to obtain service personnel data and construct a service capability portrait based on the service personnel data to obtain service capability portrait data; A service personnel allocation priority division module is used to divide the service personnel data into personnel allocation priorities according to the regional service request feature data and the service space partition data to obtain personnel allocation priority data; The service personnel scheduling optimization module is used to obtain service personnel behavior trajectory data, and to perform cross-regional collaborative scheduling of personnel allocation priority data based on the service personnel behavior trajectory data and service capability portrait data to obtain service personnel scheduling optimization data.

[0096] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0097] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A service personnel scheduling optimization method based on intelligent terminals, characterized in that: The method comprises: Obtain user behavior trajectory data and user service history data through the smart terminal, and perform regional service request feature extraction and service space partition division based on the user behavior trajectory data and user service history data to obtain regional service request feature data and service space partition data respectively; Obtain service personnel data, and construct a service capability portrait based on the service personnel data to obtain service capability portrait data; According to the regional service request feature data and the service space partition data, the service personnel data is divided into personnel allocation priority groups to obtain personnel allocation priority data; Obtain service personnel behavior trajectory data, and conduct cross-regional collaborative scheduling of personnel allocation priority data based on service personnel behavior trajectory data and service capability portrait data to obtain service personnel scheduling optimization data.

2. The method according to claim 1, characterized in that The method of obtaining user behavior trajectory data and user service history data through the smart terminal, and performing regional service request feature extraction and service space partition division according to the user behavior trajectory data and the user service history data, to obtain regional service request feature data and service space partition data respectively, includes: Obtain user behavior trajectory data and user service history data through smart terminals; Perform activity area density clustering according to user service history data to obtain user service history activity area clustering data; Perform trajectory classification based on user service historical activity area clustering data to obtain historical activity trajectory data; Fitting is performed based on historical activity trajectory data and user behavior trajectory data to obtain user behavior trajectory fitting data; Extract the stay duration feature and the activity period feature according to the historical activity trajectory data, and obtain the stay duration feature data and the activity period feature data respectively; Acquire service space data, and divide the service space data into service spaces according to the residence duration characteristic data and the activity period characteristic data to obtain service space partition data; Extract valley-peak features based on the dwell time feature data to obtain valley-peak feature data of the dwell time; According to the peak-valley characteristic data of the dwell time, regional service request characteristics are extracted from the service space partition data and the user service history data to obtain regional service request characteristic data.

3. The method according to claim 2, characterized in that The process of performing trajectory classification according to the user service historical activity area clustering data to obtain historical activity trajectory data includes: Perform regional classification based on user service historical activity regional clustering data to obtain historical activity regional classification data; Based on the historical activity area classification data, the residence time distribution calculation, the visit frequency calculation and the periodic behavior extraction are performed to obtain the historical residence time distribution data, the historical visit frequency data and the historical periodic behavior data respectively; Extract trajectory features based on historical stay time distribution data, historical visit frequency data, and historical periodic behavior data to obtain historical trajectory feature data; According to the historical trajectory feature data, the user service history data is clustered to obtain the historical activity trajectory data.

4. The method according to claim 2, characterized in that: The obtaining of the service space data and dividing the service space data into service space according to the residence time characteristic data and the activity time characteristic data to obtain the service space partition data includes: Get service space data; According to the characteristic data of the length of stay and the characteristic data of the activity period, the service space data is used to identify the stay points, and the service space stay point data is obtained; Calculate multi-scale dwell time based on the service space stay point data to obtain multi-scale dwell time data; Construct a stay map based on multi-scale stay duration data to obtain stay map data; Calculate the activity of each time period according to the stay graph data to obtain the activity data of the stay period; Perform time period clustering according to the activity data of the stay period to obtain time period clustering data; The service space data is labeled according to the time period clustering data to obtain the service space partition data.

5. The method according to claim 4, characterized in that The multi-scale dwell time calculation is performed according to the service space dwell point data to obtain the multi-scale dwell time data, including: Calculate the continuous residence time and the stationary residence time according to the service space stay point data, and obtain the continuous residence time data and the stationary residence time data respectively, wherein the stationary residence time data is the total time that the user stays at the corresponding stay point in the business premises; Perform time granularity aggregation based on the continuous residence time data and the dwelling point residence time data to obtain the duration granularity data and the dwelling point time granularity data; Generate multi-scale residence time data based on duration granularity data and dwell time granularity data to obtain preliminary multi-scale residence time data; Perform position entropy calculation and stay stability calculation on the preliminary multi-scale stay duration data to obtain position entropy data and stay stability data respectively; According to the position entropy data and the stay stability data, the preliminary multi-scale stay duration data is weightedly fused to obtain the multi-scale stay duration data.

6. The method according to claim 2, characterized in that The dwell time valley peak feature data includes first local peak valley data and second local peak valley data, and the valley peak feature extraction is performed according to the dwell time feature data to obtain the dwell time valley peak feature data, including: Perform local extreme value detection based on the residence time characteristic data to obtain the first local peak and valley data; The residence time threshold is sorted and screened according to the residence time feature data to obtain the second local peak and valley data.

7. The method according to claim 1, characterized in that The acquiring of service personnel data and constructing a service capability portrait based on the service personnel data to obtain service capability portrait data includes: Get service personnel data; Perform multi-dimensional capability numerical mapping based on service personnel data to obtain multi-dimensional capability numerical data; A capability profile is constructed based on multi-dimensional capability numerical data to obtain service capability profile data.

8. The method according to claim 1, characterized in that The personnel allocation priority data is divided according to the regional service request feature data and the service space partition data to obtain the personnel allocation priority data, including: Perform regional service demand matching according to regional service request feature data, service space partition data, and service personnel data to obtain regional service demand matching data; Optimize service response time according to regional service demand matching data to obtain service response time data; Optimize staff utilization based on service response time data to obtain staff utilization data; Extract the current regional deployment level based on the service response time data and personnel utilization data to obtain the current regional deployment level data; Extract the adjacent area dispatching level based on the service response time data and the personnel utilization rate data to obtain the adjacent area dispatching level data; Extract the candidate layer of cross-regional dispatch based on service response time data and personnel utilization data to obtain cross-regional dispatch layer data; Priority levels are generated based on the current area allocation layer data, the adjacent area allocation layer data, and the cross-area allocation layer data to obtain personnel allocation priority data.

9. The method according to claim 8, characterized in that The service personnel behavior trajectory data is obtained, and the personnel allocation priority data is coordinated and dispatched across regions according to the service personnel behavior trajectory data and the service capability portrait data to obtain the service personnel dispatch optimization data, including: Obtain service personnel behavior trajectory data; According to the service personnel behavior trajectory data and service capability portrait data, the task redistribution benefit is calculated based on the personnel allocation priority data to obtain the task redistribution benefit data; Calculate regional service satisfaction according to current regional deployment layer data and regional service request feature data to obtain regional service satisfaction data; When it is determined that the regional service satisfaction data is less than or equal to the preset service satisfaction threshold data, cross-regional collaborative scheduling is performed according to the task redistribution income data to obtain service personnel scheduling optimization data; When it is determined that the regional service satisfaction data is greater than the preset service satisfaction threshold data, the current regional deployment layer data is determined as the service personnel scheduling optimization data.

10. A service personnel scheduling optimization system based on intelligent terminals, characterized in that: Used to execute the service personnel scheduling optimization method based on the intelligent terminal according to claim 1, the service personnel scheduling optimization system based on the intelligent terminal comprises: A user behavior service preprocessing module is used to obtain user behavior trajectory data and user service history data through an intelligent terminal, and extract regional service request features and divide service space into zones according to the user behavior trajectory data and user service history data, and obtain regional service request feature data and service space partition data respectively; A service capability portrait construction module is used to obtain service personnel data and construct a service capability portrait based on the service personnel data to obtain service capability portrait data; A service personnel allocation priority division module is used to divide the service personnel data into personnel allocation priorities according to the regional service request feature data and the service space partition data to obtain personnel allocation priority data; The service personnel scheduling optimization module is used to obtain service personnel behavior trajectory data, and to perform cross-regional collaborative scheduling of personnel allocation priority data based on the service personnel behavior trajectory data and service capability portrait data to obtain service personnel scheduling optimization data.

Images