A method for dynamically matching supply and demand of on-site service, an electronic device and a storage medium

Abstract

This invention relates to the field of on-demand services, specifically to a method, electronic device, and storage medium for dynamically matching supply and demand in on-demand services. The method includes: acquiring data on multiple service items within a target area; constructing a gridded order distribution model based on the service item data to obtain a monthly order volume forecast for each grid; calculating the supply-demand gap within the target area based on the monthly order volume forecast for each grid; determining the optimal location for service stations based on the supply-demand gap and a preset location selection model; constructing a multi-factor order prediction model based on historical order data, weather data, and activity data to generate early warning information; calculating a deviation factor based on the deviation between actual order data and predicted order data, and adaptively correcting the parameters of the order prediction model. This invention achieves dynamic matching of supply and demand in on-demand services through gridded order aggregation, multi-dimensional supply-demand gap calculation, intelligent service station location selection, dynamic early warning, and a self-evolving model closed loop, thereby improving service resource allocation efficiency and user satisfaction.

Description

Technical Field

[0001] This invention relates to the field of on-site services, specifically to a method for dynamically matching supply and demand for on-site services, an electronic device, and a storage medium. Background Technology

[0002] With the development of mobile internet and the sharing economy, the on-demand service industry has flourished, encompassing various service types such as on-demand car washing, housekeeping services, and repair and installation. Taking on-demand car washing services as an example, service companies need to deploy service personnel and equipment in specific areas to provide car owners with on-demand car washing services. However, in actual operation, service companies face many technical challenges.

[0003] First, existing order forecasting methods are typically based on simple historical data statistics, which cannot accurately predict capacity demand after a new project goes live. Specifically, before a new project goes live, the lack of historical order data makes it difficult for service companies to predict order volume, easily leading to blindly opening up orders without sufficient manpower to handle them. Second, existing service station location schemes rely mainly on experience-based judgment, lacking scientific planning methods based on order density, and cannot differentiate locations according to the order distribution characteristics of different regions. Third, existing manpower allocation schemes are usually static, unable to be dynamically adjusted according to factors such as weather changes and promotional activities, resulting in lagging technician allocation and frequent last-minute emergency response situations. Finally, existing supply and demand matching mechanisms are static, unable to adaptively optimize based on actual operational data, leading to inefficient service resource allocation.

[0004] Chinese patent CN108171452A discloses a method and apparatus for selecting express delivery points. This method determines the location of express delivery points based on cluster analysis of order data. However, this method only focuses on order aggregation within the delivery radius coverage area and does not involve the calculation of supply and demand gaps or the dynamic adjustment of service station scale. Chinese patent CN121936790A discloses an intelligent order dispatch method for housekeeping services. This method allocates tasks based on dynamic bipartite graph matching, but similarly does not involve demand forecasting or service station location planning. Chinese patent CN110276666A discloses a real-time service transaction method based on both supply and demand. This method matches services through keyword matching, but does not establish an order forecasting model or a supply and demand matching mechanism.

[0005] In summary, the existing technology has the following technical shortcomings: it cannot accurately predict the order volume of new projects; it cannot scientifically plan service station locations based on order density; it cannot dynamically adjust the supply and demand matching scheme according to external factors; and it cannot continuously optimize the prediction model through actual operational data. Therefore, this application is submitted. Summary of the Invention

[0006] The primary objective of this invention is to provide a dynamic matching method for supply and demand of on-site services. This method achieves dynamic matching of supply and demand for on-site services through gridded order aggregation, multi-dimensional supply and demand gap calculation, intelligent service station site selection, dynamic early warning, and model self-evolution closed loop, thereby solving technical problems in existing technologies such as inaccurate order forecasting, reliance on experience for site selection, and static supply and demand matching.

[0007] A second objective of the present invention is to provide an electronic device, including a processor, a memory, and a communication interface, for performing the steps of the supply and demand dynamic matching method.

[0008] A third objective of this invention is to provide a readable, non-transitory storage medium storing a computer program or instructions for implementing the aforementioned dynamic supply and demand matching method.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A method for dynamically matching supply and demand for on-site services includes the following steps:

[0011] Step S1: Obtain data on multiple service items within the target area. The service item data includes item location information, total number of households, parking space ratio, and parking space saturation.

[0012] Step S2: Construct a gridded order distribution model based on the service item data to obtain the monthly order volume prediction value for each grid.

[0013] Step S3: Calculate the supply and demand gap in the target area based on the monthly order volume forecast for each grid.

[0014] Step S4: Determine the optimal location scheme for the service station based on the supply and demand gap and the preset location model.

[0015] Step S5: Construct a multi-factor order prediction model based on historical order data, weather data, and event data to generate early warning information.

[0016] Step S6: Calculate the deviation factor based on the deviation between the actual order data and the predicted order data, and adaptively correct the parameters of the order prediction model.

[0017] Preferably, the construction of the gridded order distribution model in step S2 specifically includes:

[0018] The target area is divided into multiple grid units according to a preset grid division rule. The area of ​​each grid unit is a preset value, preferably 1 km × 1 km or 500 m × 500 m.

[0019] Calculate the predicted monthly order volume for each service item within each grid cell;

[0020] The monthly order volume forecasts of each service item within the same grid cell are aggregated to obtain the monthly order volume of the grid cell.

[0021] Preferably, the calculation of the predicted monthly order volume for each service item within each grid cell includes:

[0022] Obtain the monthly car wash frequency and penetration discount factor for the service item;

[0023] The monthly order volume forecast is calculated using the formula: Monthly order volume forecast = Total number of vehicles in the project × Monthly car wash frequency × Penetration rate discount coefficient.

[0024] The total number of vehicles in the project is calculated as follows: total number of households × parking space ratio × parking space saturation.

[0025] Preferably, the method further includes: generating a heat map based on the monthly order volume of each grid cell; identifying hotspot areas in the heat map where the order density is higher than a preset density threshold, and marking the hotspot areas as priority site construction areas.

[0026] Preferably, the calculation of the supply-demand gap in step S3 specifically includes:

[0027] Obtain the preset coverage radius and preset daily average service capacity of each service station;

[0028] Calculate the total monthly order volume of each grid within the preset coverage radius;

[0029] The supply-demand gap is obtained by comparing the total monthly order volume with the preset daily average service capacity.

[0030] Preferably, the step S4 of determining the optimal location scheme for the service station specifically includes:

[0031] The coordinates and monthly order volume predictions of each service item are used as input parameters and input into the preset clustering algorithm model;

[0032] The clustering algorithm model is used to calculate the locations of multiple candidate sites;

[0033] The optimal number of stations is determined based on the location of the candidate stations and the supply-demand gap in the surrounding grids;

[0034] The number of service personnel to be configured at each service station is determined based on the optimal number of stations and the order coverage of each candidate station location.

[0035] Preferably, the preset clustering algorithm model is a K-means clustering algorithm model or a P-median clustering algorithm model.

[0036] Preferably, the construction of the multi-factor order prediction model in step S5 includes:

[0037] Obtain historical order data for the same period, recent fluctuation coefficients, weather forecast data, and activity plan data for the preset area;

[0038] The predicted order value is calculated using the formula: Future Preset Order Value for Specified Days = Historical Order Trend for the Same Period × Recent Volatility Coefficient × Weather Correction Factor × Activity Correction Factor.

[0039] The weather correction factor is calculated by weighting the precipitation probability and temperature data in the weather forecast data, and the activity correction factor is calculated by weighting the activity type and the number of participants in the activity plan data.

[0040] Preferably, the generation of early warning information includes:

[0041] If the predicted order value is greater than the preset warning threshold, a warning message is generated;

[0042] The warning information includes at least one of the following: warning of additional technicians, warning of new sites, and warning of site overload.

[0043] Preferably, the trigger condition for the technician warning is: the predicted order volume for the next preset number of days is greater than a preset multiple of the service station's average daily capacity;

[0044] The triggering conditions for the new site warning are: there are no service stations within the preset coverage radius of the service station and the predicted monthly order volume is greater than the preset order threshold;

[0045] The trigger condition for the site overload warning is: the number of consecutive preset days of actual orders exceeds the preset ratio of the service station's capacity.

[0046] Preferably, the adaptive parameter correction in step S6 includes:

[0047] Obtain actual monthly order data for each service item according to a preset cycle;

[0048] The actual monthly order data is compared with the corresponding predicted monthly order data to calculate the deviation factor; wherein, the deviation factor = actual monthly order data / predicted monthly order data;

[0049] The initial value of the penetration rate discount coefficient and the empirical value of the car wash frequency are adjusted according to the deviation factor.

[0050] Recalculate the predicted order value based on the adjusted parameters and verify the prediction accuracy.

[0051] Preferably, it also includes: adjusting the recommended threshold for each service station based on the actual operating data after the service station is established;

[0052] Specifically, if the actual order volume within a preset number of days after website launch is less than the preset percentage of the expected order volume, the website launch recommendation threshold will be lowered; if the actual order volume exceeds the expected order volume, the website launch recommendation threshold will be raised or a new website warning will be triggered.

[0053] In summary, this application has the following beneficial effects:

[0054] By constructing a gridded order distribution model, the target area is divided into multiple grid units, and the order volume in each grid is aggregated and analyzed. This clearly presents the spatial distribution characteristics of orders within the target area, providing data support for service station site selection.

[0055] By calculating the supply-demand gap, regions with supply-demand imbalances can be accurately identified, providing a quantitative basis for the allocation of service resources.

[0056] By using clustering algorithm models for service station site selection, factors such as order density and coverage can be comprehensively considered to achieve the optimal layout of service stations.

[0057] By using a multi-factor order forecasting model, which comprehensively considers information from multiple dimensions such as historical trends, weather factors, and activity factors, the accuracy of order forecasting can be improved.

[0058] By calculating the deviation factor and using the parameter adaptive correction mechanism, the prediction model can be continuously optimized, and the prediction accuracy can be continuously improved as actual operational data accumulates.

[0059] Through multiple early warning mechanisms, manpower shortages and site overload risks can be identified in advance, enabling dynamic adjustments to supply and demand. Attached Figure Description

[0060] Figure 1. Schematic diagram of the dynamic matching method for supply and demand of on-site services. It shows the complete process from obtaining static attribute data of the project (S101), calculating the monthly estimated order volume (S102), generating the global order heat distribution (S103), calculating the supply and demand gap (S104), to entering the service station site selection (S105) or dynamic scheduling early warning (S106) according to whether there is a gap, and finally forming a self-evolving closed loop through actual operation data collection and parameter adaptive correction (S107).

[0061] Figure 2 is a schematic diagram of the gridded order distribution model. It shows the complete process from obtaining the total number of households, parking space ratio, and parking space saturation of the project (S201), to calculating the total number of vehicles in the project (S202) and the monthly order volume of the project (S203-S204), then dividing the area according to the preset grid (S205), aggregating the monthly order volume of each grid (S206), and finally generating a global order heat map and identifying peak areas (S207). The core calculation formulas are explained at the bottom.

[0062] Figure 3 is a flowchart illustrating the method for determining the optimal site location for service stations. It shows that after obtaining the overall order heatmap and existing station locations (S301), the process is divided into two paths based on whether existing stations exist in the vicinity: if no existing stations exist, it determines whether the monthly order volume has reached the station construction threshold to identify essential locations for new stations or areas to be observed; if existing stations exist, it calculates the capacity gap to determine whether expansion is needed or the status quo should be maintained; finally, the optimal station location, the list of covered projects, the expected monthly order volume, and the station size level are output through a clustering algorithm (S307) (S308).

[0063] Figure 4. Schematic diagram of the multi-factor order forecasting model. It shows the three-layer architecture of the model: the input layer contains four types of inputs: historical weekly trend data, recent volatility coefficient, weather correction factor, and event / promotion correction factor; the calculation layer obtains the order forecast value for the next N days through multiplication of multiple factors; the output layer generates three types of results: personnel addition warning, new site warning, and site overload warning.

[0064] Figure 5 shows a flowchart of the parameter adaptive correction method. It illustrates that after collecting actual operational data (S501), the deviation factor (the ratio of actual monthly order volume to predicted monthly order volume) is calculated (S502), and it is determined whether the deviation exceeds the preset range: if it does, the model parameters, including the penetration rate coefficient, service frequency coefficient, and volatility coefficient, are revised (S504); if not, the current parameters are maintained (S505). Subsequently, the site recommendation threshold is adaptively adjusted (S506). Based on the rule that if orders after site construction are lower than expected, the threshold is lowered; if orders are overloaded, the threshold is raised or expansion is triggered, and the optimized model parameters are output (S507) and the process is repeated until the next data feedback, forming a self-evolving closed loop.

[0065] Figure 6. Schematic diagram of the five-layer operation mechanism. It shows the five-layer architecture from top to bottom: Input layer (static data of property projects, historical benchmark data of the industry, and existing resource data) → Algorithm layer (computing power breakdown of single projects, global aggregation modeling, and supply and demand gap calculation) → Output layer (site construction decision, manpower allocation, material and equipment matching, and timely early warning) → Execution layer (site construction / personnel on-duty / equipment deployment, and collection of real operation data) → Optimization layer (data verification, parameter correction, target achievement judgment, and iterative loop), where data from the optimization layer flows back to the algorithm layer to form a closed-loop iteration.

[0066] Figure 7. Schematic diagram of the electronic device. It shows the core components of the electronic device that implements the above method and their interconnections, including a processor (executes program instructions), a memory (stores programs and instructions), an input / output interface (interacts with external devices), a communication module (enables communication between devices), and a bus connecting all the above components. Detailed Implementation

[0067] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0068] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms "a," "the," and "the" as used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

[0069] To better understand the above technical solution, the following will provide a detailed explanation of the above technical solution in conjunction with specific implementation methods.

[0070] In some embodiments of this application, the step S1 of obtaining multiple service item data within the target area includes: obtaining administrative division information of the target area from the data management system, and identifying multiple property projects or communities to be served as service items within the target area; the service item data includes at least a project identifier, the latitude and longitude coordinates of the project address, the total number of households, the parking space ratio, and the parking space saturation.

[0071] In the specific implementation process, the target area can be divided into street level or community level according to administrative divisions, or it can be customized according to geographical boundaries. When acquiring service project data, it is first necessary to identify all property projects or communities to be served within the target area, and then collect static attribute data for each project. The project location information is used to determine the coordinate position of the project on the map, providing a basis for subsequent grid division and distance calculation; the total number of households refers to the total number of residential households in the property project, reflecting the potential user scale of the project; the parking space ratio refers to the ratio of the total number of parking spaces to the total number of households, for example, 1:1 means that each household corresponds to one parking space; the parking space saturation refers to the proportion of currently sold or rented parking spaces to the total number of parking spaces, reflecting the vehicle ownership rate and potential car wash demand of the project.

[0072] By adopting the above technical solution, complete static attribute data of service items is collected, providing a data foundation for subsequent order volume prediction and supply-demand matching calculations.

[0073] In some preferred embodiments of this application, step S2, which involves constructing a gridded order distribution model, includes the following sub-steps:

[0074] Sub-step S21: Mesh division: Divide the target area into multiple mesh units according to the preset mesh division rules.

[0075] Sub-step S22, Single-item order volume calculation: Calculate the predicted monthly order volume for each service item within each grid cell.

[0076] Sub-step S23, Grid Order Aggregation: Aggregate the predicted monthly order volume values ​​of each service item within the same grid unit to obtain the monthly order volume of the grid unit.

[0077] In the specific implementation process, the preset grid unit area during grid division can be set according to actual needs, preferably 1 km × 1 km or 500 m × 500 m. Larger grid areas are suitable for areas with a wider service range, while smaller grid areas are suitable for core areas with higher service density. The specific implementation method of grid division can adopt the standard grid division method in geographic information systems, dividing the latitude and longitude range of the target area into equal parts according to preset latitude and longitude intervals.

[0078] When calculating order volume for a single project, the monthly car wash frequency and penetration rate discount factor for that project are first obtained. Monthly car wash frequency refers to the average number of times each car needs to be washed within a specific time period. This can be set based on regional experience; for example, a high-end residential area could have a monthly car wash frequency of 2.0 times, while a typical residential area could have 1.2 times. The penetration rate discount factor is the percentage discount applied to potential order volume, considering factors such as the observation period and trial operation period. Initially, it is typically set at 3% to 8%, and can be adjusted to 10% to 20% after the operation matures. The formula for calculating the predicted monthly order volume is:

[0079] Monthly order volume forecast = Total number of vehicles in the project × Monthly car wash frequency × Penetration rate discount factor

[0080] The formula for calculating the total number of vehicles in the project is as follows:

[0081] Total number of vehicles in the project = Total number of households × Parking space ratio × Parking space saturation

[0082] When aggregating grid orders, the monthly order volume forecasts of all service items falling within the same grid cell are summed to obtain the aggregated monthly order volume for that grid cell. The aggregation formula is:

[0083] Grid monthly order volume = Σ(Forecast monthly order volume for all projects within this grid)

[0084] By adopting the above technical solution and using gridded order distribution modeling, the dispersed service project order volume can be spatially aggregated to form a visualized order density distribution map, providing accurate data support for service station site selection.

[0085] In some preferred embodiments of this application, the method further includes the step of generating a heat map based on the monthly order volume of each grid cell and identifying hotspot areas.

[0086] In practice, heatmaps visualize the monthly order volume of each grid cell, with color intensity representing order density. Orders can be categorized by percentile of monthly order volume; for example, the 30% with the highest order density can be marked as hotspot areas. Identifying hotspot areas helps operations personnel quickly locate high-demand regions and prioritize deploying service resources there.

[0087] By adopting the above technical solution and visually displaying the heat map, operators can intuitively understand the degree of order concentration in the target area, providing a basis for decision-making on service resource allocation.

[0088] In some embodiments of this application, the calculation of the supply-demand gap in step S3 includes the following sub-steps:

[0089] Sub-step S31: Obtain service station parameters: Obtain the preset coverage radius and preset daily average service capacity of each service station.

[0090] Sub-step S32, Coverage Calculation: Calculate the total monthly order volume of each grid within the preset coverage radius.

[0091] Sub-step S33, Gap Calculation: Compare the total monthly order volume with the preset daily average service capacity to obtain the supply-demand gap.

[0092] In practical implementation, the preset coverage radius refers to the maximum geographical distance that a service station can effectively serve, typically 3 kilometers, and can be adjusted according to the actual service type. The preset daily average service capacity refers to the maximum number of orders a service station can complete per day under normal working conditions, determined by factors such as the number of service personnel at the station and the number of service orders each person can complete per day. For example, if a service station has 4 technicians, and each technician can complete 15 service orders per day, then the service station's daily average service capacity is 60 orders.

[0093] When calculating coverage, first determine the candidate or existing locations of service stations. Then, draw a service coverage circle centered on that location with a preset coverage radius. Calculate the total monthly order volume of all grid cells within this coverage circle. For existing service stations, the formula for calculating the supply-demand gap is:

[0094] Supply-demand gap = Total monthly order volume - Monthly service station capacity

[0095] The monthly service capacity of a service station = the average daily service capacity × 30.

[0096] For candidate site locations, by calculating the supply and demand gap within the coverage area, it is possible to assess whether the location is suitable for building a site and the required construction scale.

[0097] By adopting the above technical solution, the supply and demand gap can be accurately calculated to quantify the degree of supply and demand imbalance in each region, providing a quantitative basis for the allocation of service resources and avoiding resource waste or insufficient service capacity.

[0098] In some preferred embodiments of this application, step S4, determining the optimal location of the service station, includes the following sub-steps:

[0099] Sub-step S41, Candidate site generation: The coordinates and monthly order volume prediction values ​​of each service item are used as input parameters and input into a preset clustering algorithm model. Multiple candidate site locations are calculated through the clustering algorithm model.

[0100] Sub-step S42, Determining the number of stations: Determine the optimal number of stations based on the locations of the candidate stations and the supply and demand gap of the surrounding grids.

[0101] Sub-step S43, personnel configuration determination: Determine the number of service personnel for each service station based on the optimal number of stations and the order coverage of each candidate station location.

[0102] In practical implementation, the clustering algorithm model can employ either K-means clustering or P-median clustering. The input to the K-means clustering algorithm is the coordinates of each service item and the predicted monthly order volume. The output is K cluster centers as candidate site locations. Determining the value of K requires considering order distribution characteristics and service site construction costs. Optimization can be achieved by enumerating different K values ​​and calculating the total coverage distance or total site construction cost for each scheme. The P-median clustering algorithm aims to minimize the weighted total distance from each service item to the nearest site, thus obtaining a more optimal site layout scheme.

[0103] Once the number of sites is determined, the supply-demand gap within the coverage area of ​​each candidate site is first calculated. If the supply-demand gap is positive and exceeds a preset threshold, then the location is suitable for site construction. The formula for calculating the preset threshold is:

[0104] Website construction threshold = Total monthly order volume within the coverage radius - Monthly service capacity of the service station

[0105] If the site creation threshold is greater than zero, it means that the existing service capacity is insufficient to cover the order demand in the region, and a new site needs to be built or the capacity expanded.

[0106] When staffing is determined, the required number of service personnel is calculated based on the monthly order volume forecast for each candidate site. The calculation formula is as follows:

[0107] Number of service personnel = Monthly order volume forecast / (Average daily service capacity × 30 × Number of working days per person per month)

[0108] Specifically, if the monthly order volume is between 1,800 and 3,000, it is recommended to have 4 service personnel; if the monthly order volume is between 3,000 and 5,000, it is recommended to have 6 to 7 service personnel; if the monthly order volume exceeds 5,000, it is recommended to split the service into two service stations.

[0109] By adopting the above technical solution and using a clustering algorithm model for service station site selection, it is possible to comprehensively consider multiple factors such as order density, coverage, and supply-demand gap, thereby achieving global optimization of service station layout, reducing total service costs while improving service quality.

[0110] In some preferred embodiments of this application, the multi-factor order prediction model uses the following calculation formula:

[0111] Forecast of orders in the next few days = Historical order trend for the same period × Recent volatility coefficient × Weather correction factor × Event correction factor

[0112] In practical implementation, historical order trend refers to using historical data from the same period as a forecast benchmark, such as using the average daily order volume of the past four weeks as the benchmark value. The recent volatility coefficient is an adjustment factor used to reflect recent order change trends, and its calculation formula is:

[0113] Recent volatility coefficient = Average daily order volume over the last 3 days / Average daily order volume over the same period in history (3 days)

[0114] The weather correction factor is calculated based on weather forecast data. Precipitation reduces the demand for car washes, and excessively high or low temperatures may also affect the willingness to wash a car. The formula for calculating the weather correction factor is:

[0115] Weather correction factor = β1 × (1 - probability of precipitation) + β2 × temperature coefficient

[0116] Among them, β1 and β2 are preset weighting coefficients, and the temperature coefficient is calculated based on the deviation of the daily average temperature from the comfortable temperature range.

[0117] The activity adjustment factor is calculated based on the activity plan within the preset area. Promotional activities, community events, etc., may lead to an increase in order volume. The formula for calculating the activity adjustment factor is:

[0118] Activity correction factor = γ1 × Activity type coefficient + γ2 × Expected participation rate

[0119] Among them, γ1 and γ2 are preset weighting coefficients, and the activity type coefficient is set according to the degree of influence of the activity on car wash demand.

[0120] By adopting the above technical solution and introducing a multi-factor correction mechanism, the impact of external factors such as weather changes and promotional activities on order volume can be fully considered, thereby improving the accuracy of order forecasting and providing a reliable basis for the advance allocation of service resources.

[0121] In some preferred embodiments of this application, the early warning information includes three types: technician addition warning, new site addition warning, and site overload warning. The triggering conditions and processing methods for each type of warning are as follows:

[0122] Additional technician alert: This alert is triggered when the predicted order volume for the next 7 days exceeds the service station's average daily capacity multiplied by 1.2. The specific formula for judgment is:

[0123] Adding technicians requires: Forecast weekly order volume > (Daily average service capacity × 1.2 × 7)

[0124] The procedure for handling the additional technician warning is as follows: send an additional technician notification to the operations system, suggesting that the dispatch center complete the allocation of service personnel before the predicted peak order period.

[0125] New Site Alert: This alert is triggered when there are no service stations within the preset coverage radius and the predicted monthly order volume exceeds the preset order threshold. The specific judgment formula is as follows:

[0126] New site selection criteria = (No service stations within the coverage area) AND (Forecasted monthly order volume > 1800 orders)

[0127] The 1800 orders is a preset monthly order threshold, set based on the minimum monthly order volume required for a service station to break even. The handling method for new site alerts is as follows: a suggestion for a new site is sent to the planning system, providing candidate site locations and recommended size information.

[0128] Site overload warning: This warning is triggered when actual orders exceed 90% of the service station's capacity for three consecutive days. The specific judgment formula is as follows:

[0129] Overload condition = (Day 1 order volume > average daily service capacity × 0.9) AND (Day 2 order volume > average daily service capacity × 0.9) AND (Day 3 order volume > average daily service capacity × 0.9)

[0130] The handling method for site overload warnings is to send diversion suggestions to the dispatch system, suggesting that some orders be diverted to nearby service stations or delayed for processing the next day.

[0131] By adopting the above technical solutions and through multiple early warning mechanisms, it is possible to cover various scenarios such as manpower allocation, site construction, and operation scheduling, so as to achieve early detection, early warning, and early handling of supply and demand risks.

[0132] In some preferred embodiments of this application, the parameter adaptive correction described in step S6 includes the following sub-steps:

[0133] Sub-step S61: Actual data collection: Obtain the actual monthly order data for each service item according to the preset cycle.

[0134] Sub-step S62, Deviation factor calculation: Compare the actual monthly order data with the corresponding predicted monthly order data to calculate the deviation factor.

[0135] Sub-step S63, parameter adjustment: Adjust the initial value of the penetration rate discount coefficient and the empirical value of the car wash frequency according to the deviation factor.

[0136] Sub-step S64, Accuracy Verification: Recalculate the predicted order value based on the adjusted parameters and verify the prediction accuracy.

[0137] In practice, the preferred data collection cycle is once a month, with the actual completed order volume for each service item tallied at the end of each month. The formula for calculating the deviation factor is:

[0138] Deviation factor = Actual monthly order volume / Forecasted monthly order volume

[0139] If the deviation factor is greater than 1, it indicates that the predicted value is too low, and the relevant parameters need to be adjusted upwards; if the deviation factor is less than 1, it indicates that the predicted value is too high, and the relevant parameters need to be adjusted downwards. When adjusting the parameters, the penetration rate discount coefficient and monthly car wash frequency are adjusted according to the direction and magnitude of the deviation factor. The adjustment formula is as follows:

[0140] New penetration rate discount factor = Original penetration rate discount factor × Deviation factor

[0141] New car wash frequency = Original car wash frequency × Deviation factor

[0142] During accuracy verification, the predicted order volume for each historical month is recalculated using the adjusted parameters, and the prediction error is calculated and statistically analyzed. If the prediction error for several consecutive months is within a preset range (e.g., ±10%), it indicates that the model parameters have converged and the model can enter the regular operation phase.

[0143] By adopting the above technical solution, through the calculation of the deviation factor and the parameter adaptive correction mechanism, the prediction model can continuously learn and optimize, adapt to the order change patterns in different regions and periods, and improve the accuracy and robustness of the prediction.

[0144] In some preferred embodiments of this application, an adaptive adjustment mechanism for the service station recommendation threshold is also included.

[0145] In practice, the service station recommendation threshold refers to the minimum order volume required to trigger the establishment of a new service station. If the actual order volume within 7 days of service station establishment is less than 80% of the expected order volume, it indicates that the order demand in the region is lower than expected, and the service station establishment recommendation threshold needs to be lowered to avoid over-construction. The formula for lowering the threshold is:

[0146] New recommendation threshold = Original recommendation threshold × 0.8

[0147] If, after a service station is established, the order volume exceeds 90% of its capacity for several consecutive days, it indicates that the order demand in that area is higher than expected, requiring an increase in the service station recommendation threshold or triggering a new site suggestion. The formula for increasing the threshold is:

[0148] New recommendation threshold = Original recommendation threshold × 1.2

[0149] Alternatively, a new site suggestion can be triggered directly to divert some orders.

[0150] By adopting the above technical solution and adaptively adjusting the recommended threshold, the service station layout plan can be dynamically optimized according to actual operation data, avoiding problems such as resource waste or insufficient service capacity.

[0151] In some preferred embodiments of this application, the dynamic matching method for supply and demand of on-site services also includes a five-layer operating mechanism:

[0152] The first layer is the input layer: input data includes static data of property projects, historical benchmark data of the industry, and existing stock resource data. Static data of property projects includes information such as project identification, address, coordinates, total number of households, parking space ratio, and parking space saturation; historical benchmark data of the industry includes empirical values ​​such as monthly car wash frequency and penetration rate discount coefficient for different grades of communities; existing stock resource data includes information such as the location, service capacity, and staffing of existing service stations.

[0153] The second layer is the algorithm layer: the algorithm layer includes a single-project computing power decomposition module, a global aggregation modeling module, and a supply-demand gap calculation module. The single-project computing power decomposition module is responsible for calculating the monthly order volume forecast for each service project; the global aggregation modeling module is responsible for spatially aggregating the order volume of each project into a grid and modeling its heat distribution; the supply-demand gap calculation module is responsible for calculating the degree of supply-demand imbalance in each region.

[0154] The third layer is the output layer: The output includes site construction decision suggestions, manpower allocation plans, material and equipment matching plans, and timeliness warning information. Site construction decision suggestions include candidate site locations, recommended construction scale, and expected return on investment; manpower allocation plans include the number of personnel at each service station and shift schedules; material and equipment matching plans include the quantity of equipment needed at each service station and the amount of consumables required; and timeliness warning information includes warnings for adding technicians, adding new sites, and site overload.

[0155] The fourth layer is the execution layer: The execution layer carries out actual operations according to the recommended plans output by the algorithm layer, including building service stations as planned, allocating service personnel, and configuring materials and equipment. During execution, it captures real operational data, including actual order volume, actual service personnel workload, and actual material consumption.

[0156] The fifth layer is the optimization layer: This layer verifies the actual data from the execution layer, calculates the deviation factor, and adjusts the model parameters of the algorithm layer based on the deviation factor. The optimization layer also includes a target achievement determination function; if a certain indicator fails to reach the preset target for several consecutive periods, the model tuning process is triggered. The optimization results output by the optimization layer are fed back to the algorithm layer, forming a closed-loop iteration.

[0157] By adopting the above technical solution and through the organizational structure of the five-layer operation mechanism, a complete closed loop can be achieved from data input to algorithm calculation, from solution output to actual execution, and from result feedback to model optimization, ensuring the continuous operation and continuous optimization of the supply and demand dynamic matching system.

[0158] In some specific embodiments of this application, taking District A of a certain city as an example, the actual application effect of the present invention is illustrated.

[0159] There are 50 property projects in Zone A. First, static data for each project was collected, including project coordinates, total number of households, parking space ratio, and parking space saturation. Zone A was divided into 200 grid units using a 1 km × 1 km grid.

[0160] Calculate the projected monthly order volume for each project. Assuming Project A1 has 1000 units, a parking ratio of 1:1, a parking saturation of 80%, a monthly car wash frequency of 1.5 times / month, and a penetration rate discount factor of 10%, then the projected monthly order volume for Project A1 is:

[0161] Project A1's monthly order volume = 1000 × 1 × 0.8 × 1.5 × 0.1 = 120 orders / month

[0162] Aggregate all projects within Area A by grid to obtain the monthly order volume for each grid. Assuming grid G1 contains 3 projects with monthly order volumes of 120, 80, and 100 respectively, then the monthly order volume for grid G1 is 300.

[0163] A heat map was generated based on the monthly order volume of each grid, and 30 hot grids were identified as priority areas for service station construction. Within these hot grids, K-means clustering was used to select service station locations, with the cluster centers serving as candidate site locations.

[0164] Calculate the supply-demand gap for each candidate site. Assuming candidate site C1 has a monthly order volume of 2000 orders within its coverage area and a service station's monthly service capacity of 1800 orders, the supply-demand gap is 200 orders, indicating a need to add a new site or expand capacity. Based on the supply-demand gap and order coverage, it is recommended to build a medium-sized service station with 6 service personnel at location C1.

[0165] In actual operation, the system calls a multi-factor order forecast model daily to calculate the order forecast for the next 7 days. Assuming a predicted weekly order volume of 500 orders, a service station's average daily service capacity of 60 orders, and a warning threshold of 1.2, the warning trigger condition is 500 > 60 × 1.2 × 7 = 504 orders, which is close to the warning threshold. The system then sends an alert to add two technicians, suggesting temporary deployment of two service personnel for support.

[0166] At the end of each month, the system collects actual monthly order data from each service station, calculates the deviation factor, and adjusts the model parameters. Assuming that the actual monthly order volume for project A1 is 130 orders and the predicted value is 120 orders, with a deviation factor of 1.083, the system adjusts the penetration rate discount coefficient from 10% to 10.83% to improve the accuracy of subsequent predictions.

[0167] As can be seen from the above embodiments, the technical solution of the present invention can effectively solve the supply and demand matching problem in the on-demand service industry, realize accurate prediction of order volume, scientific site selection of service stations, dynamic allocation of human resources and continuous optimization of prediction models, and has significant technical effects and practical value.

[0168] The present invention also provides an electronic device, including a processor, a memory, and a communication interface. The memory stores a computer program, which includes instructions executable by the processor. The processor executes the computer program to implement the steps of the on-site service supply and demand dynamic matching method described in any of the above embodiments. The communication interface communicates with external data sources and user terminals, receives input data such as service item data, historical order data, weather data, and activity data, and pushes early warning information and decision suggestions to the user terminal.

[0169] The present invention also provides a readable non-transitory storage medium storing a computer program or instructions thereon, wherein the computer program or instructions, when executed by a processor, implement the steps of the on-site service supply and demand dynamic matching method described in any of the above embodiments. The readable non-transitory storage medium may include, but is not limited to, a read-only memory, a programmable read-only memory, flash memory, random access memory, etc.

[0170] This invention achieves dynamic matching and continuous optimization of on-site service supply and demand through gridded order distribution modeling, multi-dimensional supply-demand gap calculation, service station site selection optimization based on clustering algorithms, multi-factor order prediction models, and adaptive correction mechanisms for deviation factors. Its five-layer operational mechanism ensures a complete closed-loop operation of the system, from data input to algorithm calculation, from solution output to actual execution, and from result feedback to model optimization. The technical solution of this invention effectively solves the technical problems of inaccurate order prediction, reliance on experience for site selection, and static supply-demand matching in existing technologies, improving service resource allocation efficiency and user satisfaction.

[0171] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for dynamically matching supply and demand for on-site services, characterized in that, include: Acquire data on multiple service items within the target area, including item location information, total number of households, parking space ratio, and parking space saturation. A gridded order distribution model is constructed based on the service item data to obtain the monthly order volume forecast for each grid; the supply and demand gap in the target area is calculated based on the monthly order volume forecast for each grid. The optimal location for the service station is determined based on the supply-demand gap and the preset location model; a multi-factor order prediction model is constructed based on historical order data, weather data, and activity data to generate early warning information; The deviation factor is calculated based on the discrepancy between actual order data and predicted order data, and the parameters of the order prediction model are adaptively corrected.

2. The method according to claim 1, characterized in that, The step of constructing a gridded order distribution model based on the service item data includes: dividing the target area into multiple grid units according to a preset grid division rule, with the area of ​​each grid unit being a preset value; calculating the predicted monthly order volume of each service item within each grid unit; and aggregating the predicted monthly order volume of each service item within the same grid unit to obtain the monthly order volume of the grid unit.

3. The method according to claim 2, characterized in that, The calculation of the monthly order volume forecast for each service item within each grid cell includes: obtaining the monthly car wash frequency and penetration rate discount coefficient of the service item; and calculating the monthly order volume forecast according to the formula: monthly order volume forecast = total number of vehicles in the project × monthly car wash frequency × penetration rate discount coefficient; wherein, the total number of vehicles in the project = total number of households × parking space ratio × parking space saturation.

4. The method according to claim 2, characterized in that, Also includes: A heat map is generated based on the monthly order volume of each grid cell; hotspot areas with order density higher than a preset density threshold are identified in the heat map, and these hotspot areas are marked as priority site construction areas.

5. The method according to claim 1, characterized in that, The step of calculating the supply-demand gap in the target area based on the predicted monthly order volume of each grid includes: obtaining the preset coverage radius and preset daily average service capacity of each service station; calculating the total monthly order volume of each grid within the preset coverage radius; and comparing the total monthly order volume with the preset daily average service capacity to obtain the supply-demand gap.

6. The method according to claim 1, characterized in that, The step of determining the optimal location scheme for service stations based on the supply-demand gap and a preset location model includes: inputting the coordinates and monthly order volume forecasts of each service item into a preset clustering algorithm model as input parameters; calculating multiple candidate site locations through the clustering algorithm model; determining the optimal number of stations based on the supply-demand gap between the candidate site locations and the surrounding grids; and determining the number of service personnel for each service station based on the optimal number of stations and the order coverage of each candidate site location.

7. The method according to claim 6, characterized in that, The preset clustering algorithm model is either the K-means clustering algorithm model or the P-median clustering algorithm model.

8. The method according to claim 1, characterized in that, The construction of a multi-factor order prediction model based on historical order data, weather data, and activity data includes: acquiring historical order data for the same period, recent fluctuation coefficients, weather forecast data, and activity plan data within a preset area; calculating the predicted order value according to the formula: Predicted order value for the preset number of days in the future = Historical order trend for the same period × Recent fluctuation coefficient × Weather correction factor × Activity correction factor; wherein, the weather correction factor is obtained by weighting the precipitation probability and temperature data in the weather forecast data, and the activity correction factor is obtained by weighting the activity type and number of participants in the activity plan data.

9. The method according to claim 8, characterized in that, The generation of early warning information includes: generating early warning information if the predicted order value is greater than a preset early warning threshold; wherein the early warning information includes at least one of the following: an early warning for adding technicians, an early warning for adding new sites, and an early warning for site overload.

10. The method according to claim 9, characterized in that, The trigger condition for the technician addition warning is: the predicted order volume for the next preset number of days is greater than a preset multiple of the service station's average daily capacity; the trigger condition for the new site warning is: there are no service stations within the preset coverage radius of the service station and the predicted monthly order volume is greater than a preset order threshold. The trigger condition for the site overload warning is: the number of consecutive preset days of actual orders exceeds the preset ratio of the service station's capacity.

11. The method according to claim 1, characterized in that, The process of calculating a deviation factor based on the discrepancy between actual order data and predicted order data, and adaptively correcting the parameters of the order prediction model, includes: acquiring actual monthly order data for each service item at a preset period; comparing the actual monthly order data with the corresponding predicted monthly order data to calculate the deviation factor; wherein, the deviation factor = actual monthly order data / predicted monthly order data; adjusting the initial value of the penetration rate discount coefficient and the empirical value of the car wash frequency according to the deviation factor; and recalculating the predicted order value based on the adjusted parameters and verifying the prediction accuracy.

12. The method according to claim 11, characterized in that, Also includes: The recommendation threshold for each service station will be adjusted based on the actual operational data after its establishment. Specifically, if the actual order volume within a preset number of days after the station's establishment is less than the preset proportion of the expected order volume, the recommendation threshold for the station will be lowered. If the actual order volume exceeds the expected order volume, the website building recommendation threshold will be increased or a new site warning will be triggered.

13. An electronic device, characterized in that, It includes a processor, a memory, and a communication interface; the memory is used to store a computer program; the processor is used to execute the computer program to implement the steps of the method as described in any one of claims 1 to 12.

14. A readable non-transitory storage medium, characterized in that, The storage medium stores a computer program or instructions that, when executed by a processor, implement the steps of the method as described in any one of claims 1 to 12.

Images