Driver and passenger matching guiding method and system based on dynamic supply and demand prediction

Abstract

The invention discloses a driver and passenger matching guiding method and system based on dynamic supply and demand prediction, and the method comprises the steps: enabling a thermal service to initiate a request for obtaining predicted order issuing data to an algorithm service based on the predicted supply and demand data after receiving a predicted supply and demand data request, carrying out real-time statistics on current idle driver data in each geographic grid according to the driver position data and the driver departure data; the algorithm service receives a request for predicting order-issuing data, calls a prediction model to obtain the predicted order-issuing data of each geographic grid, and feeds back the predicted order-issuing data to the thermal service; and according to the current idle driver data and the predicted order issuing data, obtaining predicted supply-demand ratio data in each geographic grid, according to the predicted supply-demand ratio data in each geographic grid and the supply-demand ratio reference data, matching the heating power gear data, and feeding back the heating power gear data to the client. The traditional historical heating power is upgraded to a dynamic probability field, and through minute-level order density prediction, the driver order receiving prediction accuracy is greatly improved.

Description

Technical Field

[0001] This invention relates to the field of ride-hailing technology, and more specifically, to a driver-passenger matching guidance method and system based on dynamic supply and demand forecasting. Background Technology

[0002] In ride-hailing platforms, efficiently guiding idle drivers to areas with demand is crucial for improving overall operational efficiency, reducing driver vacancy rates, and shortening passenger waiting times. Current technologies commonly use heatmaps as an auxiliary tool to guide drivers, typically generated based on historical order data. However, this method suffers from significant lag, failing to reflect future supply and demand trends. Drivers often find that peak demand has passed after arriving in a certain area based on the outdated heatmap, resulting in a "weak heatmap," which negatively impacts driver trust and user experience, ultimately reducing the heatmap's guidance efficiency.

[0003] See Figure 1 , Figure 1 This is a client interface diagram of a method for generating heatmaps based on recent historical order data provided by existing technology. Figure 1 The text displays the color depth of different grids corresponding to areas A through B, calculated using order data from the past 5 minutes. A darker grid color indicates more order data from the past 5 minutes. Existing technology provides a method for generating heatmaps based on recent historical order data. This method typically collects all passenger order data from the past 5 minutes, performs simple grid aggregation, and maps it to different color depths based on order density compared to the current city's median order data, thus forming heatmap areas on the map. This solution has the following technical problems: First, data lag and distortion: relying on historical order data, without... First, the data is unpredictable in predicting short-term demand changes, and driver guidance is not timely. Second, there is a lack of consideration for the dynamic balance of supply and demand: it only considers the demand side (order volume) and ignores the distribution of the supply side (current idle driver data). Even if there are many orders in a certain area, if there are more idle drivers in that area, the probability of individual drivers accepting orders will decrease, failing to reflect this crucial information. Third, the driver guidance strategy is inefficient: due to data lag and distortion, as well as the lack of consideration for the dynamic balance of supply and demand, the decisions made by drivers based on the existing heat map are often ineffective or inefficient, resulting in drivers driving empty, waiting time being too long, and order acceptance rate decreasing.

[0004] Therefore, providing a solution to the problems in existing technologies, such as the inability to predict short-term demand changes due to data lag and distortion, the lack of dynamic supply and demand balance considerations, and the inefficiency of driver-side guidance strategies, is an urgent technical issue to be addressed in this field. Summary of the Invention

[0005] In view of this, the present invention provides a driver-passenger matching guidance method based on dynamic supply and demand forecasting, in order to solve the problems in the prior art that the inability to predict short-term demand changes in the future due to data lag and distortion, the lack of consideration for dynamic supply and demand balance, and the inefficiency of driver-side guidance strategies.

[0006] Firstly, this application provides a driver-passenger matching guidance method based on dynamic supply and demand forecasting, including: After receiving the predicted supply-demand ratio data request sent by the client, the heating service initiates a request to the algorithm service to obtain the predicted order data based on the predicted supply-demand ratio data. At the same time, it calculates the current idle driver data in each geographic grid in real time based on driver location data and driver dispatch data. The algorithm service receives a request for predicted order data, calls the prediction model, obtains the predicted order data for each geographic grid, and feeds it back to the heat service. Based on the current available driver data and the predicted order data, the predicted supply-demand ratio data for each geographic grid is obtained. Based on the predicted supply-demand ratio data and the supply-demand ratio benchmark data for each geographic grid, the hot spot data is matched and the hot spot data is fed back to the client.

[0007] Optionally, the expression of the prediction model is: , In the formula, Represents the predictive model. This represents the total data for all geographic networks. This represents the volume-weighted value. Indicates the current number Predicted order placement data for each geographic grid. Indicates the current number TTOP represents the actual order data for each geographic grid.

[0008] Optionally, the transaction volume-weighted value is determined by the average revenue of historical orders across the grid and the average revenue of historical orders across the entire grid, and the transaction volume-weighted value is expressed as:

[0009] In the formula, This represents the average revenue from historical orders within the grid. This represents the average revenue from all historical orders across the entire grid.

[0010] Optionally, the expression for the predicted supply-demand ratio data is: , In the formula, This indicates the predicted supply-demand ratio data. This represents the predicted order data for a specific future time period. This indicates the current number of idle drivers.

[0011] Optionally, the supply-demand ratio benchmark data is the median of the supply-demand ratio of all non-zero thermal grids at the benchmark time within a specific future time period, based on the driver's capacity level.

[0012] Optionally, a benchmark supply-demand percentage is obtained based on the predicted supply-demand ratio data and the benchmark supply-demand ratio data, and the benchmark supply-demand percentage is expressed as: , In the formula, Indicates the baseline supply and demand percentage. This represents the baseline data for the supply-demand ratio.

[0013] Optionally, the heat level data includes a first heat level data, a second heat level data, a third heat level data, and a fourth heat level data, wherein the first heat level data is greater than the second heat level data, the second heat level data is greater than the third heat level data, and the third heat level data is greater than the fourth heat level data. When the baseline supply-demand percentage is (250%, ∞), the heat level data is the first heat level data; When the baseline supply and demand percentage is (100%, 250%), the heat level data is the second heat level data; When the baseline supply and demand percentage is (80%, 100%), the heat level data is the third heat level data; When the baseline supply-demand percentage is (0, 80%), the heat level data is the fourth heat level data.

[0014] Optionally, the heat service receives the predicted supply-demand ratio data, sends a request for predicted order placement data to the algorithm service based on the predicted supply-demand ratio data, and before real-time statistical analysis of the current available driver data in each geographic grid based on driver location data and driver dispatch data, it also includes: Historical order data, driver location data, and driver dispatch data are acquired. A deep learning algorithm is used to train the historical order data, driver location data, and driver dispatch data to obtain the prediction model. The prediction model is used to predict the order dispatch data in each geographic grid within a specific future time period.

[0015] Optionally, it also includes: The client initiates a request to the heating service to obtain the predicted supply-demand ratio data; The client receives heat level data fed back by the heat service, renders each geographic grid based on the heat level data, and generates a dynamic predicted heat map. The heat level data is determined by the predicted supply-demand ratio data and the supply-demand ratio benchmark data within each geographic grid.

[0016] Secondly, this application provides a driver-passenger matching guidance system based on dynamic supply and demand forecasting, comprising: The first data request module is used to receive the predicted supply-demand ratio data request sent by the client, and then, based on the predicted supply-demand ratio data, initiate a request to the algorithm service to obtain the predicted order data. At the same time, it uses driver location data and driver dispatch data to statistically analyze the current idle driver data in each geographic grid in real time. The model invocation module is coupled to the first data request module and the data matching module respectively. It is used to receive the request to obtain the predicted order data, invoke the prediction model, obtain the predicted order data of the geographic grid, and feed it back to the heat service. The data matching module, coupled to the model calling module, is used to receive the predicted order placement data, obtain the predicted supply-demand ratio data in each geographic grid based on the current idle driver data and the predicted order placement data, match the hot spot data based on the predicted supply-demand ratio data and the supply-demand ratio benchmark data in each geographic grid, and feed the hot spot data back to the client.

[0017] Compared with existing technologies, the driver-passenger matching guidance method based on dynamic supply and demand forecasting provided by this invention achieves at least the following beneficial effects: The present invention provides a driver-passenger matching guidance method and system based on dynamic supply and demand forecasting. The driver-passenger matching guidance method includes: after receiving a request for predicted supply-demand ratio data from a client, the heat service initiates a request to the algorithm service to obtain predicted order data based on the predicted supply-demand ratio data, and simultaneously calculates the current available driver data in each geographic grid in real time based on driver location data and driver dispatch data; the algorithm service receives the request for predicted order data, calls the prediction model to obtain the predicted order data for each geographic grid, and feeds it back to the heat service; based on the current available driver data and the predicted order data, the predicted supply-demand ratio data in each geographic grid is obtained; based on the predicted supply-demand ratio data in each geographic grid and the supply-demand ratio benchmark data, heat service level data is matched, and the heat service level data is fed back to the client. The above solution achieves the following technical effects: First, through the synergy between heat map service b and algorithm service c, and intelligent scheduling guidance based on predictive models, the traditional "historical heat map" is upgraded to a "dynamic probability field (calculating the order acceptance probability of each area in the city within a specific future time period and guiding drivers to locations with higher probabilities)." By predicting order density at the minute level, the accuracy of driver order acceptance prediction is significantly improved, effectively alleviating the information bias of "hot areas not being hot," and greatly enhancing driver trust and satisfaction. Second, through the synergy of predicted supply-demand ratio data, supply-demand ratio benchmark data, and heat map level data, drivers are guided from areas with "high driver concentration" to areas with "demand exceeding supply," thereby achieving a dynamic and balanced allocation of the ride-hailing platform's overall capacity, reducing the system's empty-running rate, improving overall order acceptance efficiency, and thus optimizing the overall allocation of transportation resources. Third, the ride-hailing platform significantly shortens the average order acceptance time for drivers. This optimization directly improves the driver's operational efficiency per unit time, manifested in an increase in average response volume and average order completion volume. Higher efficiency leads to more stable income, thereby enhancing driver stickiness and satisfaction with the platform. Ultimately, a stable and sufficient supply of transportation capacity, coupled with a better passenger travel experience, jointly drove the continuous growth of core indicators such as total order volume, market share, and overall financial performance of the platform. Therefore, this invention addresses the technical problems in existing technologies, such as the inability to predict short-term demand changes due to data lag and distortion, the lack of dynamic supply-demand balance considerations, and the inefficiency of driver-side guidance strategies.

[0018] Of course, any product implementing this invention does not necessarily need to achieve all of the technical effects described above at the same time.

[0019] Other features and advantages of the invention will become clear from the following detailed description of exemplary embodiments of the invention with reference to the accompanying drawings. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments of the invention and, together with their description, serve to explain the principles of the invention.

[0021] Figure 1 This is a client interface diagram of a method for generating heatmaps based on recent historical order data provided by existing technology; Figure 2 This is a flowchart illustrating a driver-passenger matching guidance method based on dynamic supply and demand forecasting provided by the present invention. Figure 3 This is a data processing flowchart of the driver-passenger matching guidance method based on dynamic supply and demand forecasting provided by the present invention; Figure 4 This is a flowchart illustrating another driver-passenger matching guidance method based on dynamic supply and demand forecasting provided by the present invention. Figure 5 This is a client interface diagram of the driver-passenger matching guidance method based on dynamic supply and demand forecasting provided by the present invention; Figure 6 This is a schematic diagram of the driver-passenger matching guidance system based on dynamic supply and demand forecasting provided by the present invention. Detailed Implementation

[0022] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention.

[0023] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use.

[0024] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0025] In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0026] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be discussed further in subsequent figures.

[0027] Example 1 Figure 2 This is a flowchart illustrating the driver-passenger matching guidance method based on dynamic supply and demand forecasting provided by the present invention. Figure 3 This is a data processing flowchart of the driver-passenger matching guidance method based on dynamic supply and demand forecasting provided by the present invention; Figure 2 This demonstrates the specific execution steps of the driver-passenger matching guidance method. Figure 3From a technical concept perspective, this further explains Figure 2 The data processing for predicting the supply-demand ratio involved in the paper specifically describes the data flow and collaborative relationships among the client, heating service, and algorithm service. Figure 3 In this example, a1 represents the predicted supply-demand ratio data for the next 5 minutes, a2 represents the returned heat map data, a3 represents rendering each geographic grid based on the heat map data, b1 represents querying the predicted order placement data for each geographic grid within the next 5 minutes, b2 represents returning the predicted order placement data, b3 represents calculating the predicted supply-demand ratio data, b2 represents calculating the baseline supply-demand ratio data, b3 represents calculating the heat map data based on the baseline supply-demand ratio data and returning it, and c1 represents calling the prediction model, calculating the predicted order placement data and returning it. This embodiment provides a driver-passenger matching guidance method based on dynamic supply-demand prediction, including: Step S1: After receiving the predicted supply-demand ratio data request sent by client a, the heat service b sends a request to the algorithm service c to obtain the predicted order data based on the predicted supply-demand ratio data. At the same time, it calculates the current idle driver data in each geographic grid in real time based on driver location data and driver dispatch data. Specifically, see [link to relevant documentation] Figure 2 S1 and Figure 3 The data flow between the heating service b and the algorithm service c shows that client a (such as a driver client) sends a request to the heating service b to obtain supply and demand forecasts. This request includes predicted supply and demand ratio data for a specific future time period, which can be within 30 minutes, 25 minutes, 20 minutes, 15 minutes, 10 minutes, or 5 minutes. This embodiment only uses the predicted supply and demand ratio data a1 for the next 5 minutes as an example. After receiving the request from client a, the heating service b, based on the predicted supply and demand ratio data, sends a request to the algorithm service c to query the predicted order placement data for each geographic grid within the next specific time period. For example, it queries the predicted order placement data for each geographic grid within the next 30 minutes, and further queries the predicted order placement data b1 for each geographic grid within the next 5 minutes. This short-term forecast window within the next 5 minutes balances the timeliness and accuracy of the forecast, providing sufficient preparation time for real-time scheduling without causing excessive errors due to excessively long forecast times.

[0028] In ride-hailing platforms, the predicted supply-demand ratio is a core intelligent indicator for the platform to schedule capacity and set dynamic prices. It is a quantitative prediction of the balance between "passenger ride-hailing demand" and "driver capacity supply" at a specific time and in a specific area in the future.

[0029] Predicted ride-hailing order data refers to the number of ride-hailing orders initiated by passengers within a specific geographic area within a predicted future time period (e.g., 5 minutes, 10 minutes, 15 minutes, 20 minutes, 25 minutes, or 30 minutes), as predicted by a predictive model. Taking a 5-minute time period as an example, this predicted ride-hailing order data refers to the total number of ride-hailing orders that passengers are expected to successfully place in each specified geographic grid within the next 5-minute time period. Order data refers to the order data where passengers have completed their ride-hailing and entered the matching pool, excluding cancelled orders.

[0030] In this embodiment, the geographic grid can be a hexagonal geographic grid. The hexagonal geographic grid is a spatial division technology that divides urban space into several regular hexagonal units. It is also often referred to as a cellular grid and is the core spatial carrier for realizing refined management of spatiotemporal supply and demand.

[0031] The hexagonal geographic grid described above has all vertices at the same distance from the center. Some hexagonal geographic grids border six adjacent geographic grids, with uniform boundary lengths, better reflecting the network distribution characteristics of urban roads. The hexagonal geographic grid can seamlessly cover the entire urban space without gaps or overlaps, ensuring that each location (such as driver location or passenger pick-up point) can be uniquely assigned to a geographic grid unit. The color depth of the hexagonal geographic grid is a quantitative visual expression of the core supply and demand indicators within the grid. Different color gradients correspond to the numerical values ​​of the indicators; the darker the color of the hexagonal geographic grid, the higher the demand intensity.

[0032] Based on driver location data and driver dispatch data (such as driver dispatch status data), Heat service b calculates the current available driver data in each geographic grid in real time. The current available driver data refers to the number of drivers in the current geographic grid who are ready to accept orders but have not yet received an order. It can be understood as the state of being ready to accept orders but not yet dispatched.

[0033] The aforementioned driver location data includes driver GPS (Global Positioning System) data and driver BeiDou data. Driver GPS data refers to a set of raw positioning, speed measurement, and timing information originating from the Global Positioning System, received, processed, and reported by the GPS chip in the driver's device. Driver BeiDou data refers to a set of raw positioning, navigation, timing, and short message communication information originating from the BeiDou Satellite Navigation System, received, processed, and reported by the BeiDou chip in the driver's device.

[0034] It should be noted that both Heat Map Service b and Algorithm Service c are backend services. Heat Map Service b calculates and analyzes the driver supply and demand situation in each region in real time, generating predicted supply-demand ratio data, baseline supply-demand ratio data, and heat map tier data, and provides real-time monitoring capabilities, offering core data support for driver dispatching, dynamic pricing, and operational decisions. Algorithm Service c is a program package that encapsulates algorithmic logic such as "data prediction," "model calculation," and "intelligent analysis." Both Heat Map Service b and Algorithm Service c run on a server.

[0035] Step S2: Algorithm service c receives the request for predicted order data, calls the prediction model, obtains the predicted order data for each geographic grid, and feeds it back to heat service b. Specifically, see [link to relevant documentation] Figure 2 S2 and Figure 3 The data flow between thermal service b and algorithm service c, c1 and b2 are shown. After receiving the request from thermal service b, algorithm service c calls the prediction model. This prediction model is a pre-trained prediction model. The prediction model is used to calculate the predicted order data in each geographic grid for a specific time period in the future, and feeds the predicted order data back to thermal service b. The above-mentioned calculation of the predicted order data in each geographic grid within a specific time period using the prediction model can be specifically as follows: the predicted order data in each geographic grid within the next 30 minutes can be calculated using the prediction model, and further, the predicted order data in each geographic grid within the next 5 minutes can be calculated using the prediction model. This embodiment uses the calculation of the predicted order data in each geographic grid within the next 5 minutes using the prediction model as an example.

[0036] Step S3: Based on the current available driver data and the predicted order data, obtain the predicted supply-demand ratio data for each geographic grid. Based on the predicted supply-demand ratio data and the supply-demand ratio benchmark data for each geographic grid, match the hot spot data and feed the hot spot data back to client a.

[0037] Specifically, see [link to relevant documentation] Figure 2 S3 and Figure 3 The data flow between client a, heat service b, and algorithm service c is shown. Heat service b receives the predicted order data for each geographic grid in the next 5 minutes. It calculates the predicted supply-demand ratio data for each geographic grid based on the predicted order data and the current idle driver data in each geographic grid in the next 5 minutes. The predicted supply-demand ratio data refers to the ratio between the predicted order data and the current idle driver data in each geographic grid in a specific time period in the future. It reflects the supply and demand matching status of the ride-hailing platform in real time under different time and space conditions, and provides a basis for the ride-hailing platform's scheduling.

[0038] The supply-demand ratio benchmark data can be calculated by taking the median of the supply-demand ratio of all geographic grids in the city as a baseline. It serves as a core reference standard for ride-hailing platforms to determine the matching status of capacity and demand across the city. This avoids interference from extreme values ​​and objectively reflects the neutral supply-demand level of the overall market. Specifically, the city is first divided into several standardized geographic grids, and the real-time supply-demand ratio for each grid is calculated. Then, the supply-demand ratio values ​​of all geographic grids are arranged in ascending order. If the total number of geographic grids is odd, the median value is taken; if it is even, the average of the two median values ​​is taken. This final determined value is the supply-demand ratio benchmark data. For example, if a city is divided into 5 geographic grids, and the supply-demand ratios are sorted as 1, 3, 5, 7, 9, the median of 5 is the supply-demand ratio benchmark data. If there are 6 geographic grids, and the supply-demand ratios are sorted as 2, 3, 4, 5, 7, 9, the median is (4+5) / 2=4.5, and the median of 4.5 is the supply-demand ratio benchmark data. Reference Figure 3 As shown in b5, heat map data is matched based on the percentage between the predicted supply-demand ratio data and the baseline supply-demand ratio data within each geographic grid. Specifically, different percentages between the predicted supply-demand ratio data and the baseline supply-demand ratio data within each geographic grid correspond to different heat map data levels. For example, the larger the percentage between the predicted supply-demand ratio data and the baseline supply-demand ratio data within each geographic grid, the larger the heat map data, indicating a stronger relative demand compared to supply in that region; conversely, the smaller the percentage between the predicted supply-demand ratio data and the baseline supply-demand ratio data within each geographic grid, the smaller the heat map data, indicating insufficient demand in that region. The correlation between the percentage between the predicted supply-demand ratio data and the baseline supply-demand ratio data within each geographic grid and the heat map data can provide multi-dimensional support for ride-hailing platform scheduling, driver order acceptance, and industry management. The matched heat map data is fed back to client A, which can further process the heat map data, such as rendering it for visualization purposes.

[0039] Heat map data is generated by ride-hailing platforms by dividing urban areas into several geographical grids and assigning different heat levels to different grids based on real-time data such as order call volume and currently available drivers within the past few minutes. The heat levels are typically indicated by color intensity, with higher heat levels and darker colors representing stronger demand for rides in that area.

[0040] Compared with existing technologies, the driver-passenger matching guidance method based on dynamic supply and demand forecasting provided in this embodiment achieves at least the following beneficial effects: The driver-passenger matching guidance method based on dynamic supply and demand forecasting provided in this embodiment involves the following steps: After receiving a request for predicted supply-demand ratio data from the client, the heat service initiates a request to the algorithm service to obtain predicted order placement data based on the predicted supply-demand ratio data. Simultaneously, it calculates the current available driver data in each geographic grid in real time based on driver location data and driver dispatch data. Upon receiving the request for predicted order placement data, the algorithm service calls the prediction model to obtain the predicted order placement data for each geographic grid and feeds it back to the heat service. Based on the current available driver data and the predicted order placement data, it obtains the predicted supply-demand ratio data for each geographic grid. Based on the predicted supply-demand ratio data for each geographic grid and the supply-demand ratio benchmark data, it matches the heat service level data and feeds the heat service level data back to the client. The above solution achieves the following technical effects: First, through the synergy between heat map service b and algorithm service c, and intelligent scheduling guidance based on predictive models, the traditional "historical heat map" is upgraded to a "dynamic probability field (calculating the order acceptance probability of each area in the city within a specific future time period and guiding drivers to locations with higher probabilities)." By predicting order density at the minute level, the accuracy of driver order acceptance prediction is significantly improved, effectively alleviating the information bias of "hot areas not being hot," and greatly enhancing driver trust and satisfaction. Second, through the synergy of predicted supply-demand ratio data, supply-demand ratio benchmark data, and heat map level data, drivers are guided from areas with "high driver concentration" to areas with "demand exceeding supply," thereby achieving a dynamic and balanced allocation of the ride-hailing platform's overall capacity, reducing the system's empty-running rate, improving overall order acceptance efficiency, and thus optimizing the overall allocation of transportation resources. Third, the ride-hailing platform significantly shortens the average order acceptance time for drivers. This optimization directly improves the driver's operational efficiency per unit time, manifested in an increase in average response volume and average order completion volume. Higher efficiency leads to more stable income, thereby enhancing driver stickiness and satisfaction with the platform. Ultimately, a stable and sufficient supply of transportation capacity, coupled with a better passenger travel experience, jointly drove the continuous growth of core indicators such as total order volume, market share, and overall financial performance of the platform. Therefore, this embodiment solves the technical problems in the prior art, such as the inability to predict short-term demand changes due to data lag and distortion, the lack of consideration for dynamic supply and demand balance, and the inefficiency of driver-side guidance strategies.

[0041] In one alternative embodiment, the expression for the prediction model is: , In the formula, Represents the predictive model. This represents the total data for all geographic networks. This represents the volume-weighted value. Indicates the current number Predicted order placement data for each geographic grid. Indicates the current number TTOP represents the actual order data for each geographic grid.

[0042] Specifically, continue to refer to Figure 2 S2 and Figure 3 As shown in algorithm service c, the prediction model uses the total data of all geographic networks, the transaction volume-weighted value, and the current [number] [item]. Predicted order data for the current geographic grid, the current order number. The prediction model is constructed using real order placement data from each geographic grid and all real order placement data from all geographic grids. This model provides the core data basis and decision support for the algorithm service c.

[0043] The total data for all the above geographic networks refers to the total number of geographic grid units that are fully or partially covered within the target city, based on a pre-defined geographic grid size.

[0044] The volume-weighted value is a composite indicator, not a single numerical value. It refers to the value obtained by weighting or summing a certain "attribute value" to be evaluated, using the transaction volume of each geographic grid as the weight. It reflects the degree to which the active trading area contributes to the overall performance of the attribute, and the transaction volume is the number of valid orders completed within each geographic grid.

[0045] Current number The predicted order data for each geographic grid is not the actual value, but rather the result of inference based on historical data and related characteristics (such as time characteristics, environmental characteristics, etc.).

[0046] Current number The actual order data for a geographic grid refers to the order number within the current time period. The total number of valid ride-hailing orders actually generated and recorded by the ride-hailing platform in each geographic grid is an objective and true result of order statistics.

[0047] The actual order data for all geographic grids refers to the total number of order requests successfully submitted by passengers and entered into the order matching system across all geographic grids during the study period.

[0048] Algorithm service c obtains an accurate estimate of the predicted order data within a geographic grid for a specific future time period by calling this prediction model. This is the current... Combined with the predicted order data from the first geographic grid, the current order data is... The real order data from each geographic grid will serve as the core input, directly driving the intelligent scheduling system to make forward-looking deployments of transportation capacity and providing key basis for dynamic pricing strategies, ultimately achieving the core goals of supply and demand balance and experience optimization.

[0049] In one optional embodiment, the volume-weighted value is determined by the average revenue of historical orders across the grid and the average revenue of historical orders across the entire grid, and the volume-weighted value is expressed as:

[0050] In the formula, This represents the average revenue from historical orders within the grid. This represents the average revenue from all historical orders across the entire grid.

[0051] Specifically, the average revenue from historical orders in the aforementioned grid refers to a specific target grid (such as the first grid). The average revenue of all valid ride-hailing orders within a set historical statistical period (in geographical grids). The average historical order revenue across all grids refers to the overall average revenue of all valid ride-hailing orders within the set historical statistical period across all geographical grids within the entire city or target service area defined by the ride-hailing platform.

[0052] The transaction volume-weighted value is obtained by comparing the average revenue of historical orders in the grid with the average revenue of historical orders across the entire grid.

[0053] By adopting the above approach, the transaction volume-weighted values ​​not only provide drivers with more valuable information on expected revenue, helping them make operational decisions, but also allow them to precisely focus their operational efforts on the areas that have the greatest impact on the platform's overall revenue. Simultaneously, this shifts the pricing strategy from "perception-driven" to "data-driven." By adjusting the prices of specific high-weighted grids, the platform's overall revenue level and supply-demand balance can be most effectively leveraged.

[0054] In one alternative embodiment, the expression for predicting the supply-demand ratio data is: , In the formula, This indicates the predicted supply-demand ratio data. This represents the predicted order data for a specific future time period. This indicates the current number of idle drivers.

[0055] Specifically, continue to refer to Figure 2 S3 and Figure 3 As shown in b and b3 of the heating service, the predicted supply-demand ratio is calculated based on the ratio between the predicted order demand in the short term (e.g., 5-30 minutes) and the current idle driver data.

[0056] If the predicted supply-demand ratio is greater than 1, it indicates that supply (supply side) is insufficient to meet demand (demand side), and passengers will have difficulty hailing a ride and experience longer waiting times. If the predicted supply-demand ratio is approximately equal to 1, then supply and demand are basically in balance. If the predicted supply-demand ratio is less than 1, it indicates that supply (supply side) exceeds demand (demand side), meaning that drivers may face idle waiting times and reduced income.

[0057] Heating service b monitors and predicts supply-demand ratio data. When the predicted supply-demand ratio is greater than 1, it injects "dispatch guidance" in advance to reduce pressure. When the predicted supply-demand ratio is less than 1, it provides guidance information to drivers, thereby maintaining the stable and efficient operation of the entire two-sided market and ultimately achieving a win-win-win situation where passengers can quickly hail a ride, drivers' income is maximized, and the platform efficiency is optimized.

[0058] In one alternative embodiment, continue to refer to Figure 3 As shown in the heating service b, the supply-demand ratio benchmark data is the median of the supply-demand ratio of all non-zero heating grids at the benchmark time in a specific future time period, based on the driver's capacity level.

[0059] Specifically, driver capacity level refers to the driver's order-accepting level, which includes Economy, Premium, Executive, Business, and Luxury. These are common service levels or vehicle classifications on ride-hailing platforms. They are mainly in a step-by-step upgrade relationship "from low to high," with the core differences being vehicle standards, service experience, price, driver requirements, and target customer groups.

[0060] In a non-zero thermal grid, the thermal grid divides the target city into virtual, fixed-size hexagonal geographic grids. Each hexagonal geographic grid is a "thermal grid," which is the smallest geographic unit for ride-hailing platforms to calculate and schedule supply and demand; non-zero means that at the "baseline time," the demand of this hexagonal geographic grid is not zero.

[0061] First, determine the baseline time and driver capacity level. For example, if the baseline time is 16:00 today, and the driver capacity level is "Luxury," then the service is filtered to identify all non-zero hot grids at the baseline time. For instance, service b retrieves data from the 15:30 to 16:00 time window, iterates through all geographic grids in the city, and filters out grids with predicted order volume greater than zero. Assume there are 10,000 grids within a certain ring road in a city, of which 3,000 are active grids with demand at this time. These 3,000 are the "all non-zero hot grids at the baseline time." Next, predict the supply-demand ratio for a specific future time period, such as 17:00 to 17:30. For each of the 3,000 grids, service b uses historical big data, real-time trends, and predictive models to predict order volume and available driver data for the 17:00-17:30 time period, calculating the predicted supply-demand ratio for each grid. Finally, the median of the supply-demand ratio is calculated to obtain the baseline supply-demand ratio data, such as the predicted supply-demand ratio data for each of the 3000 grid cells. 1, 2, 3,..., Sort these 3000 values ​​from smallest to largest, and take the median of the supply-demand ratio of this set of data (i.e., the average of the 1500th and 1501st values). This median supply-demand ratio is "the baseline data of the supply-demand ratio seen by luxury car drivers at the baseline time (16:00) for the future evening peak (17:00-17:30)".

[0062] The median supply-demand ratio allows us to better reflect the supply and demand situation in the "middle level" area where a driver is located, and how a driver would react if randomly staying in a certain area with demand.

[0063] By adopting the above solution, drivers are directed to areas with better future supply and demand, reducing their empty driving rate and waiting time, improving order matching speed and passenger ride success rate, and maximizing the platform's overall operational efficiency (total transaction volume, order completion rate, etc.). Furthermore, drivers can clearly perceive that upgrading their driver capacity level not only earns them honor badges but also provides them with tangible high-quality order opportunities and regional priority. This is the core motivation for incentivizing drivers to provide better service and remain online for longer periods.

[0064] In one optional embodiment, a baseline supply-demand percentage is obtained based on the predicted supply-demand ratio data and the baseline supply-demand ratio data. The baseline supply-demand percentage is expressed as follows: , In the formula, Indicates the baseline supply and demand percentage. This represents the baseline data for the supply-demand ratio; Specifically, continue to refer to Figure 3 As shown in Figures b and b4 of the heating service, the benchmark supply-demand percentage is used to match the heating level data. The benchmark supply-demand percentage is obtained based on the ratio between the predicted supply-demand ratio data and the benchmark supply-demand ratio data. If the predicted supply-demand ratio data is larger, the benchmark supply-demand percentage is larger, there are more predicted order data in the area, and the probability that drivers can accept orders is greater. If the predicted supply-demand ratio data is smaller, the benchmark supply-demand percentage is smaller, there are fewer predicted order data in the area, and the probability that drivers can accept orders is smaller.

[0065] By adopting the above scheme, based on the baseline supply and demand percentage, drivers are not only guided to areas with "high predicted order data", but also to areas with "relatively less competition and a higher probability of accepting orders", thus achieving an upgrade from one-way information display to two-way supply and demand scheduling.

[0066] In one alternative embodiment, continue to refer to Figure 3 As shown in the heating service b, the heating level data includes the first heating level data, the second heating level data, the third heating level data, and the fourth heating level data. The first heating level data is greater than the second heating level data, the second heating level data is greater than the third heating level data, and the third heating level data is greater than the fourth heating level data. When the baseline supply-demand percentage is (250%, ∞), the heat level data is the first heat level data; when the baseline supply-demand percentage is (100%, 250%), the heat level data is the second heat level data; when the baseline supply-demand percentage is (80%, 100%), the heat level data is the third heat level data; when the baseline supply-demand percentage is (0%, 80%), the heat level data is the fourth heat level data.

[0067] Specifically, the data for the first, second, third, and fourth heat levels decrease sequentially.

[0068] When the baseline supply-demand percentage is (250%, ∞), the heat level data is the first heat level data, which is the hottest heat level data. This hottest heat level data indicates that supply is far less than demand, and there is a serious imbalance between supply and demand. When the baseline supply and demand percentage is (100%, 250%), the heat level data is the second heat level data, which is the secondary heat level data. This secondary heat level data indicates that the supply is much less than the demand, resulting in a supply and demand imbalance. When the baseline supply and demand percentage is (80%, 100%), the hot spot data is the third hot spot data, which is the next hot spot data. This next hot spot data indicates that supply and demand are basically balanced, and the predicted order data is roughly the same as the current idle driver data. When the baseline supply-demand percentage is (0, 80%), the hot spot data is the fourth hot spot data. The fourth hot spot data is the non-hot spot data. The non-hot spot data indicates that the supply exceeds the demand and there is excess capacity. This can be understood as the current number of idle drivers being large, while the predicted number of orders is small.

[0069] When the baseline supply and demand percentage is zero, the geographic grid is not displayed.

[0070] For example: Continuing with the division of a city into 5 geographic grids, the supply-demand ratios, after being sorted, are 1, 3, 5, 7, and 9. The median of 5 is the baseline supply-demand ratio. For instance: the baseline supply-demand ratio is 5, the predicted supply-demand ratio for the driver's grid is 6, and the heat level falls within the secondary heat level. The baseline supply-demand percentage is calculated using the formula for the baseline supply-demand percentage. When the baseline supply and demand percentage is (100%, 250%), 100% < 120% ≤ 250%, which means it falls into the baseline supply and demand percentage of (100%, 250%), corresponding to the second heat level data, i.e. the secondary heat level data.

[0071] By adopting the above scheme, based on different values ​​of the baseline supply and demand percentage, it is possible to quickly match the data of the first, second, third, and fourth heat levels, which greatly reduces the computational complexity of heat service b. At the same time, it is possible to quickly understand the supply and demand relationship between the current idle driver data and the predicted number of orders, thereby guiding scheduling and market strategies.

[0072] In one alternative embodiment, see Figure 4 As shown, Figure 4 This is a flowchart illustrating another driver-passenger matching guidance method based on dynamic supply and demand forecasting provided by the present invention; Heat service b receives the predicted supply-demand ratio data, and sends a request for predicted order placement data to algorithm service c based on the predicted supply-demand ratio data. Simultaneously, before real-time statistical analysis of current available driver data within each geographic grid based on driver location data and driver dispatch data, the following steps are also included: S0 acquires historical order data, driver location data, and driver dispatch data. It then uses a deep learning algorithm to train a prediction model on the historical order data, driver location data, and driver dispatch data to obtain a prediction model. The prediction model is used to predict the order dispatch data in each geographic grid within a specific future time period.

[0073] Specifically, regarding the training phase of the prediction model, this embodiment uses a deep learning algorithm, inputting historical order data, driver location data, and driver dispatch data to complete model training. At the deep learning algorithm level, a spatiotemporal convolutional network can be used: first, the geographical grid of the target city is treated as graph nodes, and this network is used to mine the spatial correlation features between grids. Simultaneously, CNN (Convolutional Neural Network) / LSTM (Long Short-Term Memory) components are used to capture the temporal variation patterns of the data. Since spatiotemporal convolutional networks are a mature and readily available deep learning algorithm, this embodiment does not limit its specific implementation.

[0074] The aforementioned historical order data refers to the complete record of all completed or cancelled orders on ride-hailing platforms within a certain period (e.g., the past 3 months or 1 year). It primarily reflects the spatiotemporal distribution patterns of regional travel demand and serves as core training data for supply and demand forecasting models. This historical order data includes basic order attribute data, spatiotemporal characteristic data, business attribute data, and order status-derived data. Basic order attribute data includes unique information such as order number, order creation / completion time, and driver / passenger identification. Spatiotemporal characteristic data includes spatiotemporal information such as origin / destination latitude and longitude, functional area, and trip trajectory sequence. Business attribute data includes transaction details such as order type, estimated / actual mileage duration, and cost composition. Order status-derived data includes status information such as order completion / cancellation status, cancellation entity, and cancellation time.

[0075] The aforementioned driver location data refers to the geographical location and trajectory data of drivers on the road, collected in real time or periodically by ride-hailing platforms. It primarily reflects the spatial distribution of current transportation capacity and is a key basis for real-time matching and capacity scheduling. This driver location data specifically includes the following dimensions: 1. Basic location attributes, such as driver ID, vehicle ID, location collection timestamp, real-time latitude and longitude coordinates, positioning accuracy, and positioning device information; 2. Vehicle status attributes, including the driver's current service status: empty / carrying passengers / resting; 3. Dynamic trajectory attributes, such as driving speed, driving direction, cumulative mileage, current road name, and congestion status.

[0076] The aforementioned driver dispatch data refers to full data reflecting the daily / weekly online service behavior patterns of drivers. Its core characteristic is the time-based supply characteristics of transportation capacity, used to analyze peak and trough patterns in capacity supply and driver operational preferences. This driver dispatch data specifically includes the following dimensions: 1. Basic dispatch attributes, such as driver identification, vehicle identification, dispatch date; online time, offline time; cumulative online time and effective service time for the day. 2. Dispatch behavior attributes, such as online location, offline location, number of orders accepted, number of orders completed, number of orders rejected, number of orders canceled, empty mileage, and empty mileage rate for the day. 3. Dispatch status attributes, such as driver dispatch status; driver dispatch status refers to the driver's dispatch status and vehicle return status. When a driver is in dispatch status, they are marked as an idle driver.

[0077] By adopting the above solution, it is possible to predict the demand for order placement in specific time periods and target geographical areas in advance, guide drivers to high-demand areas in advance, reduce drivers' empty waiting time in low-demand areas, and improve drivers' order-taking efficiency and income.

[0078] In one alternative embodiment, combined with Figure 3 and Figure 5 As shown, Figure 5 This is a client interface diagram of the driver-passenger matching guidance method based on dynamic supply and demand forecasting provided by the present invention. Figure 5 From the perspective of the client-side implementation, it demonstrates Figure 3 The heat map data (first heat map data, second heat map data, third heat map data and fourth heat map data) involved in the process is rendered as geographic grids of different colors on the client map (i.e. dynamic predicted heat map). Figure 5 In the C region, the predicted supply-demand ratio is reflected in the fact that the darker the color of the geographic grid, the more orders are predicted, and the higher the probability of drivers accepting orders; Figure 5 Below, data cards are also displayed. For example, the predicted supply-demand ratio data for area C is broken down and displayed, showing the predicted order placement data for a specific future time period (e.g., an estimated 4 to 6 calls in the next 5 minutes), historical order placement data for the same period (e.g., 4 calls in the last 5 minutes), and current available driver data (currently 0 available vehicles). This driver-passenger matching guidance method also includes: client a initiating a request to heat service b to obtain the predicted supply-demand ratio data; Client A receives heat level data from heat service B, renders each geographic grid based on the heat level data, and generates a dynamic predicted heat map. The heat level data is determined by the predicted supply-demand ratio data and the supply-demand ratio baseline data within each geographic grid.

[0079] Specifically, continue to refer to Figure 3 As shown in a1, when the driver opens client a (such as a driver app), client a actively sends a request to the heat service b. The request content is: the predicted supply-demand ratio data for a specific time period in the future, such as the predicted supply-demand ratio data for the next 30 minutes, and further limited to the predicted supply-demand ratio data for the next 5 minutes.

[0080] Continue to refer to Figure 5 and Figure 3 As shown in Figure a3, client a, based on the received heat map data (such as first, second, third, and fourth heat map data), renders the corresponding geographic grid on the map using different color depths, ultimately generating and displaying an intuitive, dynamically predicted heat map to guide drivers to areas with higher future demand. Specifically, client a divides the target city map into small geographic grids. Each grid is assigned a "heat map level" based on the predicted passenger demand intensity for a specific future time period (e.g., 5 to 30 minutes), and is visually encoded using different color depths. Color mapping: The first heatmap is mapped to a dark orange geographic grid d (the darkest color with the highest saturation), the second heatmap is mapped to an orange geographic grid e, the third heatmap is mapped to a light orange geographic grid f, and the fourth heatmap is mapped to a lighter orange-beige geographic grid h. The colors transition between the dark orange geographic grid d, the orange geographic grid e, the light orange geographic grid f, and the lighter orange-beige geographic grid h.

[0081] Dynamic rendering technology: Based on the heat level data fed back by heat service b, the prediction is recalculated every 5-10 minutes and the dynamic prediction heat map is refreshed; color gradient animation is used between the first, second, third and fourth heat level data to avoid visual jumps; the dynamic prediction heat map is semi-transparently overlaid on the map without affecting the viewing of road network information.

[0082] This embodiment upgrades the calculation scope of the heatmap from 'completed order density', which reflects historical conditions, to 'supply-demand ratio of predicted order data to current available driver data', which predicts future supply and demand. Based on the calculated degree of supply and demand imbalance, a dynamic predictive heatmap is ultimately formed to guide drivers through color depth rendering.

[0083] Dynamically predictive heat maps can not only guide drivers to areas with high future demand, thereby reducing the time and fuel costs of blind cruising, but also shorten the average passenger waiting time, increase the order completion rate and user satisfaction. At the same time, the capacity can be deployed in advance in demand areas, shortening the response time.

[0084] Optionally, based on the driver's capacity level, a dynamic forecast heatmap can be viewed. For example, if a driver's capacity level is Economy, that driver can only view the dynamic forecast heatmap for Economy vehicles within the same city; if a driver's capacity level is Premium, that driver can view not only the dynamic forecast heatmap for Premium vehicles within the same city, but also the dynamic forecast heatmaps for Economy and Premium vehicles within the same city. The ride-hailing platform implements a 'high-quality capacity pre-schedule strategy,' prioritizing high-service-level / high-compliance-level drivers to predicted high-demand hotspot areas based on the dynamic forecast heatmap. This strategy aims to improve the order matching efficiency and income expectations of high-level drivers through forward-looking capacity deployment, while optimizing the overall passenger experience.

[0085] Example 2 Reference Figure 6 As shown, Figure 6 This is a schematic diagram of the driver-passenger matching guidance system based on dynamic supply and demand forecasting provided by the present invention. This embodiment provides a driver-passenger matching guidance system based on dynamic supply and demand forecasting, including: The first data request module 10 is used to receive the predicted supply-demand ratio data request sent by the client, and then, based on the predicted supply-demand ratio data, send a request to the algorithm service to obtain the predicted order data. At the same time, it uses driver location data and driver dispatch data to statistically analyze the current idle driver data in each geographic grid in real time. The model calling module 11 is coupled to the first data request module 10 and the data matching module 12 respectively. It is used to receive the request to obtain the predicted order data, call the prediction model, obtain the predicted order data of the geographic grid, and feed it back to the heat service. The data matching module 12, coupled to the model calling module 11, is used to receive the predicted order data, obtain the predicted supply-demand ratio data in each geographic grid based on the current idle driver data and the predicted order data, match the hot spot data with the predicted supply-demand ratio data in each geographic grid and the supply-demand ratio benchmark data, and feed the hot spot data back to the client.

[0086] In one alternative embodiment, the expression for the prediction model is: , In the formula, Represents the predictive model. This represents the total data for all geographic networks. This represents the volume-weighted value. Indicates the current number Predicted order placement data for each geographic grid. Indicates the current number TTOP represents the actual order data for each geographic grid.

[0087] In one optional embodiment, the volume-weighted value is determined by the average revenue of historical orders across the grid and the average revenue of historical orders across the entire grid, and the volume-weighted value is expressed as:

[0088] In the formula, This represents the average revenue from historical orders within the grid. This represents the average revenue from all historical orders across the entire grid.

[0089] In one alternative embodiment, the expression for predicting the supply-demand ratio data is: , In the formula, This indicates the predicted supply-demand ratio data. This represents the predicted order data for a specific future time period. This indicates the current number of idle drivers.

[0090] In one alternative embodiment, the supply-demand ratio baseline data is the median of the supply-demand ratio over a specific future time period, based on the driver's capacity level and filtering all non-zero thermal grids at the baseline time.

[0091] In one optional embodiment, a baseline supply-demand percentage is obtained based on the predicted supply-demand ratio data and the baseline supply-demand ratio data. The baseline supply-demand percentage is expressed as follows: , In the formula, Indicates the baseline supply and demand percentage. This represents the baseline data for the supply-demand ratio.

[0092] In one optional embodiment, the heat level data includes first heat level data, second heat level data, third heat level data, and fourth heat level data, wherein the first heat level data is greater than the second heat level data, the second heat level data is greater than the third heat level data, and the third heat level data is greater than the fourth heat level data. When the baseline supply-demand percentage is (250%, ∞), the heat level data is the first heat level data; when the baseline supply-demand percentage is (100%, 250%), the heat level data is the second heat level data; when the baseline supply-demand percentage is (80%, 100%), the heat level data is the third heat level data; when the baseline supply-demand percentage is (0%, 80%), the heat level data is the fourth heat level data.

[0093] In one optional embodiment, the driver-passenger matching guidance system further includes a training module for acquiring historical order data, driver location data, and driver dispatch data, and using a deep learning algorithm to train the historical order data, driver location data, and driver dispatch data to obtain a prediction model. The prediction model is used to predict order dispatch data in each geographic grid within a specific future time period.

[0094] In one optional embodiment, it further includes a second data request module and a grid rendering module. The second data request module is used by the client to initiate a request to the heat service to obtain the predicted supply-demand ratio data. The grid rendering module is used by the client to receive the heat level data fed back by the heat service, and to render each geographic grid according to the heat level data to generate a dynamic predicted heat map. The heat level data is determined by the predicted supply-demand ratio data and the supply-demand ratio benchmark data in each geographic grid.

[0095] It should be noted that the specific limitations of a driver-passenger matching guidance system can be found in the limitations of a driver-passenger matching guidance method described above, and will not be repeated here. The modules in the aforementioned driver-passenger matching guidance system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0096] As can be seen from the above embodiments, the driver-passenger matching guidance system based on dynamic supply and demand forecasting provided in this embodiment achieves at least the following beneficial effects: The driver-passenger matching guidance system based on dynamic supply and demand forecasting provided by this invention comprises: a first data request module, used by the heat service to receive a request for predicted supply-demand ratio data from a client, and then, based on the predicted supply-demand ratio data, to initiate a request to the algorithm service to obtain predicted order placement data; simultaneously, it uses driver location data and driver dispatch data to statistically analyze the current available driver data in each geographic grid in real time; a model invocation module, coupled to both the first data request module and the data matching module, is used to receive the request for predicted order placement data, invoke the prediction model, obtain the predicted order placement data for the geographic grid, and feed it back to the heat service; and a data matching module, coupled to the model invocation module, is used to receive the predicted order placement data, obtain the predicted supply-demand ratio data for each geographic grid based on the current available driver data and the predicted order placement data, and match the heat service level data based on the predicted supply-demand ratio data for each geographic grid and the supply-demand ratio benchmark data. By feeding back heat map data to the client, the collaborative efforts of the first data request module, model invocation module, and data matching module achieve the following technical effects: First, intelligent scheduling guidance based on predictive models upgrades the traditional "historical heat map" to a "dynamic probability field (calculating the probability of order acceptance in each area of ​​the city within a specific future time period and guiding drivers to locations with higher probabilities)." Through minute-level order density prediction, the accuracy of driver order acceptance prediction is significantly improved, effectively alleviating the information bias of "hot areas not being hot," and greatly enhancing driver trust and satisfaction. Second, drivers are guided from areas with "high driver concentration" to areas with "demand exceeding supply," thereby achieving a dynamic and balanced allocation of the ride-hailing platform's overall capacity, reducing the system's empty mileage rate, improving overall order acceptance efficiency, and thus optimizing the overall allocation of capacity resources. Third, the ride-hailing platform significantly shortens the average order acceptance time for drivers. This optimization directly improves the driver's operational efficiency per unit time, manifested in an increase in average response volume and average order completion volume. Higher efficiency leads to more stable income, thereby enhancing driver stickiness and satisfaction with the platform. Ultimately, a stable and sufficient supply of transportation capacity, coupled with a better passenger travel experience, jointly drove the continuous growth of core indicators such as total order volume, market share, and overall financial performance of the platform. Therefore, this embodiment solves the technical problems in the prior art, such as the inability to predict short-term demand changes due to data lag and distortion, the lack of consideration for dynamic supply and demand balance, and the inefficiency of driver-side guidance strategies.

[0097] While specific embodiments of the invention have been described in detail by way of examples, those skilled in the art should understand that the examples are for illustrative purposes only and not intended to limit the scope of the invention. Those skilled in the art should understand that modifications can be made to the above embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.

Claims

1. A driver-passenger matching guidance method based on dynamic supply and demand forecasting, characterized in that, include: After receiving the predicted supply-demand ratio data request sent by the client, the heating service initiates a request to the algorithm service to obtain the predicted order data based on the predicted supply-demand ratio data. At the same time, it calculates the current idle driver data in each geographic grid in real time based on driver location data and driver dispatch data. The algorithm service receives a request for predicted order data, calls the prediction model, obtains the predicted order data for each geographic grid, and feeds it back to the heat service. Based on the current available driver data and the predicted order data, the predicted supply-demand ratio data for each geographic grid is obtained. Based on the predicted supply-demand ratio data and the supply-demand ratio benchmark data for each geographic grid, the hot spot data is matched and the hot spot data is fed back to the client.

2. The driver-passenger matching guidance method based on dynamic supply and demand forecasting according to claim 1, characterized in that, The expression for the prediction model is: ， In the formula, Represents the predictive model. This represents the total data for all geographic networks. This represents the volume-weighted value. Indicates the current number Predicted order placement data for each geographic grid. Indicates the current number TTOP represents the actual order data for each geographic grid.

3. The driver-passenger matching guidance method based on dynamic supply and demand forecasting according to claim 2, characterized in that, The transaction volume-weighted value is determined by the average revenue of historical orders in the grid and the average revenue of historical orders across the entire grid. The transaction volume-weighted value is expressed as follows: ， In the formula, This represents the average revenue from historical orders within the grid. This represents the average revenue from all historical orders across the entire grid.

4. The driver-passenger matching guidance method based on dynamic supply and demand forecasting according to claim 1, characterized in that, The expression for the predicted supply-demand ratio data is: ， In the formula, This indicates the predicted supply-demand ratio data. This represents the predicted order data for a specific future time period. This indicates the current number of idle drivers.

5. The driver-passenger matching guidance method based on dynamic supply and demand forecasting according to claim 1, characterized in that, The supply-demand ratio benchmark data is the median of the supply-demand ratio of all non-zero thermal grids at the benchmark time over a specific future time period, based on driver capacity levels.

6. The driver-passenger matching guidance method based on dynamic supply and demand forecasting according to claim 1, characterized in that, Based on the predicted supply-demand ratio data and the baseline supply-demand ratio data, the baseline supply-demand percentage is obtained, which is expressed as: ， In the formula, Indicates the baseline supply and demand percentage. This represents the baseline data for the supply-demand ratio.

7. The driver-passenger matching guidance method based on dynamic supply and demand forecasting according to claim 6, characterized in that, The heat level data includes a first heat level data, a second heat level data, a third heat level data, and a fourth heat level data. The first heat level data is greater than the second heat level data, the second heat level data is greater than the third heat level data, and the third heat level data is greater than the fourth heat level data. When the baseline supply-demand percentage is (250%, ∞), the heat level data is the first heat level data; When the baseline supply and demand percentage is (100%, 250%), the heat level data is the second heat level data; When the baseline supply and demand percentage is (80%, 100%), the heat level data is the third heat level data; When the baseline supply-demand percentage is (0, 80%), the heat level data is the fourth heat level data.

8. The driver-passenger matching guidance method based on dynamic supply and demand forecasting according to claim 1, characterized in that, The heat service receives predicted supply-demand ratio data, sends a request for predicted order placement data to the algorithm service based on the predicted supply-demand ratio data, and before simultaneously calculating the current available driver data in each geographic grid based on driver location data and driver dispatch data, it also includes: Historical order data, driver location data, and driver dispatch data are acquired. A deep learning algorithm is used to train the historical order data, driver location data, and driver dispatch data to obtain the prediction model. The prediction model is used to predict the order dispatch data in each geographic grid within a specific future time period.

9. The driver-passenger matching guidance method based on dynamic supply and demand forecasting according to claim 1, characterized in that, Also includes: The client initiates a request to the heating service to obtain the predicted supply-demand ratio data; The client receives heat level data fed back by the heat service, renders each geographic grid based on the heat level data, and generates a dynamic predicted heat map. The heat level data is determined by the predicted supply-demand ratio data and the supply-demand ratio benchmark data within each geographic grid.

10. A driver-passenger matching guidance system based on dynamic supply and demand forecasting, characterized in that, include: The first data request module is used to receive the predicted supply-demand ratio data request sent by the client, and then, based on the predicted supply-demand ratio data, initiate a request to the algorithm service to obtain the predicted order data. At the same time, it uses driver location data and driver dispatch data to statistically analyze the current idle driver data in each geographic grid in real time. The model invocation module is coupled to the first data request module and the data matching module respectively. It is used to receive the request to obtain the predicted order data, invoke the prediction model, obtain the predicted order data of the geographic grid, and feed it back to the heat service. The data matching module, coupled to the model calling module, is used to receive the predicted order placement data, obtain the predicted supply-demand ratio data in each geographic grid based on the current idle driver data and the predicted order placement data, match the hot spot data based on the predicted supply-demand ratio data and the supply-demand ratio benchmark data in each geographic grid, and feed the hot spot data back to the client.

Images