An intelligent management system for online car-hailing

Abstract

The present invention relates to the technical field of intelligent transportation systems, and discloses an intelligent management system for online car-hailing, including: a collection module, a dispatching module, and a pricing module; the collection module obtains real-time order information and passenger historical data and inputs them into the dispatching module and the pricing module; the dispatching module performs vehicle dispatching and dispatch optimization to select the best vehicle based on the acquired real-time order information and passenger historical data; the pricing module dynamically calculates the order price of the best vehicle and adjusts the pricing parameters in real time based on the acquired real-time order information and passenger historical data combined with the pricing parameters of the corresponding time period and route area. The present invention uses big data and machine learning to optimize dispatching and dynamic pricing, improve efficiency and service quality, flexibly respond to changes in supply and demand, reduce empty trips, simplify payment processes, provide a convenient and environmentally friendly travel experience, and improve urban transportation efficiency as a whole.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation systems, and in particular to an intelligent management system for online car-hailing. Background Art

[0002] In the modern urban traffic environment, online ride-hailing services, as a supplement to the traditional taxi industry, have become an indispensable part of urban residents' travel. With the development of mobile Internet technology, online ride-hailing platforms use big data analysis and real-time computing capabilities to optimize vehicle scheduling, route planning, and dynamic pricing strategies to meet the personalized needs of different users, while improving the efficiency of traffic resources and alleviating urban traffic pressure. However, traditional online ride-hailing management methods have some limitations: (1) The early dispatching system was mainly based on the distance priority principle, ignoring other important factors such as real-time traffic conditions, driver service quality, and passenger preferences, resulting in suboptimal dispatching efficiency and service satisfaction. (2) During peak hours or in specific areas, there may be a shortage of vehicles, while during non-peak hours or in other areas, there may be empty vehicles. This imbalance between supply and demand reduces overall operational efficiency. (3) Failure to effectively predict and respond to changes in market demand (such as the impact of weather, holidays, large-scale events, etc. on travel demand) and failure to adjust dispatching strategies in real time to adapt to market changes. (4) Failure to optimize driving routes and reduce empty driving rates will lead to higher energy consumption and emissions, which is not conducive to environmental protection and sustainable development. (5) Once the price parameters of traditional cruising taxi meters are generated, they cannot be remotely and dynamically adjusted according to external factors (such as weather, time, and traffic conditions), which is not conducive to the effective allocation of transportation capacity. (6) The existing management system is relatively cumbersome in the payment process, and the evaluation mechanism is not objective and transparent enough, which affects the continuous improvement of user experience and service quality. Summary of the invention

[0003] In view of the above-mentioned technical deficiencies, the technical problem to be solved by the present invention is to provide an intelligent management system for online car-hailing, which aims to optimize the dispatching process and pricing system of online car-hailing through advanced data analysis, machine learning technology and cloud computing platform, thereby improving the operating efficiency and service quality of the urban transportation system.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides an intelligent management system for online car-hailing, comprising: a collection module, a dispatching module and a pricing module; The collection module obtains real-time order information and passenger historical data and inputs them into the dispatch module and the pricing module; The dispatch module performs vehicle dispatch and dispatch optimization to select the best vehicle based on the acquired real-time order information and passenger history data; The pricing module dynamically calculates the order price of the optimal vehicle and adjusts the pricing parameters in real time based on the acquired real-time order information and passenger history data combined with the pricing parameters of the corresponding time period and route area.

[0005] Furthermore, the dispatching module includes a pre-processing module, a demand forecasting module, an available resource evaluation module and a calculation module; The pre-processing module analyzes real-time order information and passenger history data to build passenger profiles; The real-time order information and passenger history data include the passenger's location, destination, travel time, traffic conditions within a radius set by the passenger's departure point, whether it is a holiday, and the passenger's waiting time; The demand forecasting module predicts the demand for rides in different areas and time periods based on passenger profiles, inputs the available resource assessment module to evaluate the current capacity and compare it with the predicted demand, and calculates the supply and demand coefficient for dispatching orders; The calculation module integrates the data of the preprocessing module and the demand forecasting module, and optimizes the dispatch results through the optimization algorithm.

[0006] Furthermore, the demand forecasting module uses time series analysis and machine learning models to predict the ride demand in different areas and time periods, including: Collect historical time series data from different regions and time periods in the past and integrate external factors such as holidays as input features into the machine learning model for demand analysis to obtain predicted demand; Continuously monitor the current call volume and the actual number of available vehicles in each area, calculate the predicted supply and demand coefficient through the historical vehicle data of the current area combined with factors, and calculate the supply and demand adjustment coefficient based on the predicted supply and demand coefficient as follows: Forecast supply and demand coefficient = forecast demand / historical average number of available vehicles Real-time supply and demand coefficient = current call volume / actual number of available vehicles Supply and demand adjustment coefficient = predicted supply and demand coefficient / real-time supply and demand coefficient Among them, if the supply-demand adjustment coefficient is greater than 1, then supply exceeds demand, otherwise supply is less than demand; When supply exceeds demand, the default dispatch area is adjusted according to the supply and demand adjustment coefficient to expand the dispatch area; When supply is less than demand, the default dispatch area is adjusted according to the supply and demand adjustment coefficient to reduce the dispatch area.

[0007] Furthermore, the machine learning model performs demand analysis to obtain predicted demand, including: When a passenger places an order through the platform, the current service area A is calculated based on the passenger's location and service requirements as follows: A = 1 / 2πr 2 Among them, r is the radius that meets the current passenger demand determined by the current capacity, π is the pi, assuming that the current capacity is F, the standard r is 1.5km, and the average capacity when the order completion rate is 80% is Fs, then r= Fs / F*1.5. If r is less than 1.5km after calculation, r minus the standard r is 1.5km; The number of available vehicles and their status in the current service area A are monitored in real time, and dynamic adjustments are made based on the current order volume. The long short-term memory network model is used to predict the supply and demand situation in the future to obtain the predicted demand.

[0008] Furthermore, the calculation module integrates the data of the preprocessing module and the demand forecasting module, and optimizes the dispatching results through the optimization algorithm, including: For each target vehicle T in the current service area A i , calculate its simulated order acceptance probability P, journey time T, driver score R, and passenger waiting coefficient W to construct multi-factor scoring parameters; The multi-factor scoring parameters are calculated to construct the driver scoring matrix Mr and the journey time matrix St, which respectively represent the score and time ratio of each driver relative to other drivers; Based on the constructed driver rating matrix Mr and journey time matrix St, as well as the set rating weights wr and journey time weights st, the final dispatch vector Ve is calculated; Find the vehicle with the largest score in the final dispatch vector Ve, and have this vehicle execute the order.

[0009] Furthermore, for each vehicle C in the current service area A i , calculate its simulated order acceptance probability P, journey time T, driver score R, and passenger waiting coefficient W to construct multi-factor scoring parameters including: Get the number of orders N near the current time point in the historical data, and get the simulated order acceptance probability P=1 / N; Where N is the number of historical orders in the nearby area. The nearby area refers to a circular geo-fence with a radius of d and a distance d from the driver's current coordinates. The current location of the driver and the location where the passenger calls the vehicle are used as the starting point and the end point. The public map connection is used to match the distance, the traffic congestion at that time and the shortest time to obtain the journey time T. Average the driver's past passenger ratings to get the driver rating R; Find the longest waiting time in the current passenger's past orders, and find the average Aw of the top 10% of the maximum waiting time, find the median waiting time Mw of all passengers, and get the waiting coefficient W=Aw / Mw for the passenger.

[0010] Furthermore, the driver rating matrix Mr and the journey time matrix St are constructed to represent the rating and time ratio of each driver relative to other drivers, respectively: For each target vehicle C in the current service area A i , by C1,C2....C N Constitute a target vehicle group C; According to the simulated order acceptance probability P and the target vehicle group C, the vehicle order acceptance probability vector VT=[PC1,PC2,...PC N ]; Calculate the relative value M of a single driver's score relative to other drivers' scores ij =R i / R j ; Among them, R i is the score of the current driver i, R j To compare the ratings of driver j; The relative rating values ​​of all drivers are combined into an N×N rating matrix Mr: Mr = [ [1, M21, ... Mn1] [M12, 1, ... Mn2] ... [M1n, M2n, ... 1] ] Calculate the relative value S of a single driver's journey time T relative to other drivers' journey times ij =1-T i / T j ; Among them, T i is the current journey time of driver i, T j To compare the journey time of driver j; The relative values ​​of the journey times of all drivers are combined into an N×N time matrix St: St = [ [1,S21, ... Sn1] [S12, 1, ... Sn2] ... [S1n, S2n, ... 1] ] Multiply the order acceptance probability vector VT by the scoring matrix Mr to obtain the relative scoring vector Vr, and multiply the order acceptance probability vector VT by the time consumption matrix St to obtain the relative time consumption vector Vt.

[0011] Furthermore, based on the constructed driver rating matrix Mr and journey time matrix St, as well as the set rating weight wr and journey time weight wt, the final dispatch vector Ve=wr*Vr+wt / W*Vt is calculated.

[0012] Furthermore, the pricing module dynamically calculates the order price of the optimal vehicle based on the acquired real-time order information and passenger history data combined with pricing parameters of the corresponding time period and route area, including: Calculate the estimated price of this trip based on the booking time, the navigation route of the starting point and the end point, and the pricing parameters of the corresponding time period and the area passed through; The pricing parameters and order information are sent to the on-board meter in an encrypted manner to achieve remote dynamic price adjustment; The vehicle terminal recalculates the price according to the pricing parameters issued in real time, and displays the recalculated price on the display screen of the vehicle terminal in real time; After the order is completed, the current pricing parameters will automatically become invalid, and the vehicle terminal will revert to the original pricing parameters; The pricing parameters include basic fee, mileage fee, duration fee, surcharge and discount.

[0013] Furthermore, the pricing module adjusts the pricing parameters in real time according to the supply and demand adjustment coefficient of the dispatching module provided by the dispatching module; Among them, if the supply-demand adjustment coefficient is > 1, the basic fee, mileage fee, duration fee will be reduced, and the surcharge will be reduced or cancelled; If the supply-demand adjustment factor is < 1, the basic fee, mileage fee, duration fee and surcharge will increase. The beneficial effects of the present invention are: 1. This invention uses advanced data analysis and machine learning technology to optimize the dispatching process by not only considering the distance factor, but also comprehensively evaluating multiple aspects of information such as real-time traffic conditions, driver service quality, and passenger preferences. This intelligent dispatching method significantly improves dispatching efficiency and service satisfaction, ensuring that every passenger can get the most suitable vehicle service.

[0014] 2. The pricing module of the present invention can dynamically calculate the order price of the optimal vehicle based on real-time order information and passenger historical data, combined with pricing parameters for corresponding time periods and route areas, and adjust pricing parameters in real time. During peak hours or when demand surges in a specific area, prices can be flexibly adjusted to reflect the actual supply and demand relationship, thereby ensuring platform revenue and providing users with a transparent and reasonable pricing mechanism.

[0015] 3. The present invention uses the supply and demand adjustment coefficient to dynamically adjust the dispatch area and pricing parameters, effectively dealing with the phenomenon of vehicle supply and demand imbalance during peak hours or in specific areas, and reducing the situation of empty vehicles during non-peak hours. It not only improves the overall operational efficiency, but also guides users' travel choices through reasonable pricing strategies, further optimizing resource allocation.

[0016] 4. The intelligent route planning of the present invention reduces unnecessary driving distance, reduces energy consumption and emissions, and contributes to the realization of environmental protection and sustainable development goals. At the same time, the flexibility and transparency of the pricing system enhances the user's sense of trust, simplifies the payment process, and makes the entire travel experience more convenient and efficient. In addition, the objective and transparent evaluation mechanism promotes the continuous improvement of service quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A logical schematic diagram of an online car-hailing intelligent management system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] Example 1

[0021] like Figure 1 As shown, this embodiment provides an intelligent management system for online car-hailing, including: a collection module, an order dispatching module and a pricing module; the collection module obtains real-time order information and passenger historical data and inputs them into the order dispatching module and the pricing module.

[0022] The dispatch module performs vehicle dispatch and dispatch optimization to select the best vehicle based on the real-time order information and passenger history data. The dispatch module includes a preprocessing module, a demand forecasting module, an available resource assessment module and a calculation module; the preprocessing module analyzes real-time order information and passenger historical data to build a passenger portrait; among them, the real-time order information and passenger historical data include passenger location, destination, travel time, traffic conditions within a radius set by the passenger's departure point, whether it is a holiday and the passenger's waiting time.

[0023] The demand forecasting module predicts the demand for rides in different areas and time periods based on passenger portraits, inputs the available resource assessment module to evaluate the current capacity and compare it with the predicted demand, and calculates the supply and demand coefficient for dispatching orders.

[0024] The calculation module integrates the data of the preprocessing module and the demand forecasting module, and optimizes the dispatch results through the optimization algorithm.

[0025] Preferably, the demand forecasting module uses time series analysis and machine learning models to predict the ride demand in different regions and time periods, including: collecting historical time series data of different regions and time periods in the past period of time and integrating external factors including holidays as input features into the machine learning model for demand analysis to obtain the predicted demand as follows: When a passenger places an order through the platform, the current service area A is calculated based on the passenger's location and service requirements as follows: A = 1 / 2πr 2 Among them, r is the radius that meets the current passenger demand determined by the current capacity, π is the pi, assuming that the current capacity is F, the standard r is 1.5km, and the average capacity when the order completion rate is 80% is Fs, then r= Fs / F*1.5. If r is less than 1.5km after calculation, r minus the standard r is 1.5km; Furthermore, the number of available vehicles and their status in the current service area A are monitored in real time, and dynamically adjusted based on the current order volume. The long short-term memory network model is used to predict the supply and demand situation in the future to obtain the predicted demand as follows: set up the long short-term memory network model and initialize the weights; clean the data, select features, convert timestamps, and standardize / normalize; divide the data into training set, validation set, and test set; perform iterative training through forward propagation, loss calculation, back propagation, and weight update; use the validation set to monitor indicators such as loss and accuracy, and stop early when necessary; use an independent test set to perform a final evaluation of the trained model; adjust the model parameters according to the performance of the validation set to find the optimal configuration.

[0026] Specifically, the current call volume and the actual number of available vehicles in each area are continuously monitored, and the predicted supply and demand coefficient is calculated through the historical vehicle data of the current area combined with factors, and the supply and demand adjustment coefficient is calculated based on the predicted supply and demand coefficient as follows: Forecast supply and demand coefficient = forecast demand / historical average number of available vehicles Real-time supply and demand coefficient = current call volume / actual number of available vehicles Supply and demand adjustment coefficient = predicted supply and demand coefficient / real-time supply and demand coefficient Among them, if the supply and demand adjustment coefficient is greater than 1, the supply exceeds the demand, otherwise the supply is less than the demand; when the supply exceeds the demand, the default dispatch area is adjusted according to the supply and demand adjustment coefficient to expand the dispatch area; when the supply is less than the demand, the default dispatch area is adjusted according to the supply and demand adjustment coefficient to reduce the dispatch area.

[0027] Specifically, the calculation module integrates the data of the preprocessing module and the demand prediction module, and optimizes the dispatch results through the optimization algorithm, including: each vehicle C in the current service area A i , calculate its simulated order acceptance probability P, journey time T, driver score R, and passenger waiting coefficient W to construct multi-factor scoring parameters including: Get the number of orders N near the current time point in the historical data, and get the simulated order acceptance probability P=1 / N; where N is the number of historical orders in the nearby area, and the nearby area refers to a circular geographic fence with a radius of d, centered on the driver's current coordinates, and a distance d from the driver's order reception distance. The driver's current location and the passenger's boarding location are used as the starting and ending points. By calling the public map connection, the algorithm matching the distance, the current traffic congestion, and the shortest time is used to obtain the journey time T. Average the driver's past passenger ratings to obtain the driver rating R. Find the longest waiting time in the current passenger's past orders, and find the average Aw of the top 10% of the maximum waiting time, find the median waiting time Mw of all passengers, and get the waiting coefficient W=Aw / Mw for the passenger.

[0028] Preferably, the multi-factor scoring parameters are calculated to construct a driver scoring matrix Mr and a journey time matrix St, which respectively represent the score and time ratio of each driver relative to other drivers, including: For each target vehicle C in the current service area A i , by C1,C2....C N Constitute the target vehicle group C; according to the simulated order acceptance probability P and the target vehicle group C, obtain the vehicle order acceptance probability vector VT=[PC1,PC2,...PC N ]; Calculate the relative value M of a single driver's score relative to other drivers' scores ij =R i / R j ; Among them, R i is the score of the current driver i, R j To compare the ratings of driver j, the relative rating values ​​of all drivers are combined into an N×N rating matrix Mr: Mr = [ [1, M21, ... Mn1] [M12, 1, ... Mn2] ... [M1n, M2n, ... 1] ] Furthermore, the relative value S of a single driver's journey time T relative to other drivers' journey times is calculated. ij =1-T i / T j ; Among them, T i is the current journey time of driver i, T j To compare the journey time of driver j, the relative values ​​of the journey time of all drivers are combined into an N×N time matrix St: St = [ [1,S21, ... Sn1] [S12, 1, ... Sn2] ... [S1n, S2n, ... 1] ] Preferably, the order acceptance probability vector VT is multiplied by the scoring matrix Mr to obtain the relative scoring vector Vr, and the order acceptance probability vector VT is multiplied by the time matrix St to obtain the relative time vector Vt; based on the constructed driver scoring matrix Mr and the journey time matrix St, as well as the set scoring weight wr and time weight wt, the final dispatch vector Ve=wr*Vr+wt / W*Vt is calculated; find the vehicle with the largest score in the final dispatch vector Ve, and have the vehicle execute the order.

[0029] Furthermore, the pricing module dynamically calculates the order price of the optimal vehicle and adjusts the pricing parameters in real time based on the acquired real-time order information and passenger history data combined with the pricing parameters of the corresponding time period and route area. The estimated price of this trip is calculated based on the booking time, the navigation routes of the starting point and the end point, combined with the pricing parameters of the corresponding time period and the area passed through; the pricing parameters and order information are uniformly sent to the on-board meter in an encrypted manner to achieve remote dynamic price adjustment; the on-board terminal recalculates the price based on the pricing parameters sent in real time, and displays the recalculated price on the display screen of the on-board terminal in real time; after the order is completed, the current pricing parameters are automatically invalid, and the on-board terminal reverts to the original pricing parameters; the pricing parameters include basic fees, mileage fees, duration fees, surcharges and discounts; at the same time, the pricing module adjusts the pricing parameters in real time based on the supply and demand adjustment coefficient of the dispatch module provided by the dispatch module; if the supply and demand adjustment coefficient is > 1, the basic fee, mileage fee, duration fee is reduced, and the surcharge is reduced or cancelled; if the supply and demand adjustment coefficient is < 1, the basic fee, mileage fee, duration fee and surcharge are increased.

[0030] Example 2

[0031] Based on Example 1, in order to facilitate making more reasonable dispatching decisions in different environments, especially in areas with poor air quality, densely populated areas or major events, electric vehicles or hybrid vehicles are dispatched first, and an environmental index E can be introduced.

[0032] Specifically, the environmental index E is related to PM2.5 concentration, geographical location and major events; if the PM2.5 concentration in the area is > 75 micrograms / cubic meter, setting the environmental index E indicates a preference for electric vehicles; if the order is located in a densely populated area, setting the environmental index E indicates a preference for electric vehicles to reduce pollution and noise; if there are major events on that day such as sports meetings, marathons, and Environmental Protection Day on June 5, setting the environmental index E indicates a preference for electric vehicles to support environmental protection initiatives.

[0033] According to the above conditions, the environmental index E can be simply binary processed, that is, 0 or 1, E=1 means that the order tends to be electric vehicles; E=0 means there is no special tendency.

[0034] The weights of PM2.5 concentration, geographical location and major events are further set, and the formula for constructing the environmental index E is: E = ω1×S1+ω2×S2+ω3×S3 Among them, ω1 is the weight of PM2.5 concentration, S1 is, ω2 is the weight of geographical location, S2 is, and ω3 is the weight of major events. S1, S2 and S3 are the binary scores of 0 or 1 for PM2.5 concentration, geographical location and major events respectively.

[0035] When E=1, orders tend to be dispatched to electric vehicles, and the final dispatch vector Ve=(wr * Vr+wt / W*Vt)*E.

[0036] When E=0, there is no special tendency, and the final single vector Ve=wr * Vr+wt / W*Vt is assigned.

[0037] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent management system for online car-hailing, characterized in that: include: Collection module, dispatch module and pricing module; The collection module obtains real-time order information and passenger historical data and inputs them into the dispatch module and the pricing module; The dispatch module performs vehicle dispatch and dispatch optimization to select the best vehicle based on the acquired real-time order information and passenger history data; The pricing module dynamically calculates the order price of the optimal vehicle and adjusts the pricing parameters in real time based on the acquired real-time order information and passenger history data combined with the pricing parameters of the corresponding time period and route area.

2. The intelligent management system for online car-hailing according to claim 1, characterized in that: The dispatch module includes a preprocessing module, a demand forecasting module, an available resource evaluation module and a calculation module; The pre-processing module analyzes real-time order information and passenger history data to build passenger profiles; The real-time order information and passenger history data include the passenger's location, destination, travel time, traffic conditions within a radius set by the passenger's departure point, whether it is a holiday, and the passenger's waiting time; The demand forecasting module predicts the demand for rides in different areas and time periods based on passenger profiles, inputs the available resource assessment module to evaluate the current capacity and compare it with the predicted demand, and calculates the supply and demand coefficient for dispatching orders; The calculation module integrates the data of the preprocessing module and the demand forecasting module, and optimizes the dispatch results through the optimization algorithm.

3. The intelligent management system for online car-hailing according to claim 2, characterized in that: The demand forecasting module uses time series analysis and machine learning models to predict the demand for rides in different areas and time periods, including: Collect historical time series data from different regions and time periods in the past and integrate external factors such as holidays as input features into the machine learning model for demand analysis to obtain predicted demand; Continuously monitor the current call volume and the actual number of available vehicles in each area, calculate the predicted supply and demand coefficient through the historical vehicle data of the current area combined with factors, and calculate the supply and demand adjustment coefficient based on the predicted supply and demand coefficient as follows: Forecast supply and demand coefficient = forecast demand / historical average number of available vehicles Real-time supply and demand coefficient = current call volume / actual number of available vehicles Supply and demand adjustment coefficient = predicted supply and demand coefficient / real-time supply and demand coefficient Among them, if the supply-demand adjustment coefficient is greater than 1, then supply exceeds demand, otherwise supply is less than demand; When supply exceeds demand, the default dispatch area is adjusted according to the supply and demand adjustment coefficient to expand the dispatch area; When supply is less than demand, the default dispatch area is adjusted according to the supply and demand adjustment coefficient to reduce the dispatch area.

4. The intelligent management system for online car-hailing according to claim 3, characterized in that: The machine learning model performs demand analysis to obtain predicted demand, including: When a passenger places an order through the platform, the current service area A is calculated based on the passenger's location and service requirements as follows: A = 1 / 2πr 2 Among them, r is the radius that meets the current passenger demand determined by the current capacity, π is the pi, assuming that the current capacity is F, the standard r is 1.5km, and the average capacity when the order completion rate is 80% is Fs, then r= Fs / F*1.

5. If r is less than 1.5km after calculation, r minus the standard r is 1.5km; The number of available vehicles and their status in the current service area A are monitored in real time, and dynamic adjustments are made based on the current order volume. The long short-term memory network model is used to predict the supply and demand situation in the future to obtain the predicted demand.

5. The intelligent management system for online car-hailing according to claim 2, characterized in that: The calculation module integrates the data of the pre-processing module and the demand forecasting module, and optimizes the dispatch results through the optimization algorithm, including: For each target vehicle T in the current service area A i , calculate its simulated order acceptance probability P, journey time T, driver score R, and passenger waiting coefficient W to construct multi-factor scoring parameters; The multi-factor scoring parameters are calculated to construct the driver scoring matrix Mr and the journey time matrix St, which respectively represent the score and time ratio of each driver relative to other drivers; Based on the constructed driver rating matrix Mr and journey time matrix St, as well as the set rating weights wr and journey time weights st, the final dispatch vector Ve is calculated; Find the vehicle with the largest score in the final dispatch vector Ve, and have that vehicle execute the order.

6. The intelligent management system for online car-hailing according to claim 5, characterized in that: For each vehicle C in the current service area A i , calculate its simulated order acceptance probability P, journey time T, driver score R, and passenger waiting coefficient W to construct multi-factor scoring parameters including: Get the number of orders N near the current time point in the historical data, and get the simulated order acceptance probability P=1 / N; Where N is the number of historical orders in the nearby area. The nearby area refers to a circular geo-fence with a radius of d and a distance d from the driver's current coordinates. The current location of the driver and the location where the passenger calls the vehicle are used as the starting point and the end point. The public map connection is used to match the distance, the traffic congestion at that time and the shortest time to obtain the journey time T. Average the driver's past passenger ratings to get the driver rating R; Find the longest waiting time in the current passenger's past orders, and find the average Aw of the top 10% of the maximum waiting time, find the median waiting time Mw of all passengers, and get the waiting coefficient W=Aw / Mw for the passenger.

7. The intelligent management system for online car-hailing according to claim 4, characterized in that: Construct the driver rating matrix Mr and the journey time matrix St, which respectively represent the rating and time ratio of each driver relative to other drivers: For each target vehicle C in the current service area A i , by C1,C2....C N Constitute a target vehicle group C; According to the simulated order acceptance probability P and the target vehicle group C, the vehicle order acceptance probability vector VT=[PC1,PC2,...PC N ]; Calculate the relative value M of a single driver's score relative to other drivers' scores ij =R i / R j ; Among them, R i is the score of the current driver i, R j To compare the ratings of driver j; The relative rating values ​​of all drivers are combined into an N×N rating matrix Mr: Mr = [ [1, M21, ... Mn1] [M12, 1, ... Mn2] ... [M1n, M2n, ... 1] ] Calculate the relative value S of a single driver's journey time T relative to other drivers' journey times ij =1-T i / T j ; Among them, T i is the current journey time of driver i, T j To compare the journey time of driver j; The relative values ​​of the journey times of all drivers are combined into an N×N time matrix St: St = [ [1,S21, ... Sn1] [S12, 1, ... Sn2] ... [S1n, S2n, ... 1] ] Multiply the order acceptance probability vector VT by the scoring matrix Mr to obtain the relative scoring vector Vr, and multiply the order acceptance probability vector VT by the time consumption matrix St to obtain the relative time consumption vector Vt.

8. The intelligent management system for online car-hailing according to claim 4, characterized in that: Based on the constructed driver rating matrix Mr and journey time matrix St, as well as the set rating weight wr and journey time weight wt, the final dispatch vector Ve=wr*Vr+wt / W*Vt is calculated.

9. The intelligent management system for online car-hailing according to claim 1, characterized in that: The pricing module dynamically calculates the order price of the optimal vehicle based on the acquired real-time order information and passenger history data combined with the pricing parameters of the corresponding time period and route area, including: Calculate the estimated price of this trip based on the booking time, the navigation route of the starting point and the end point, and the pricing parameters of the corresponding time period and the area passed through; The pricing parameters and order information are sent to the on-board meter in an encrypted manner to achieve remote dynamic price adjustment; The vehicle terminal recalculates the price according to the pricing parameters issued in real time, and displays the recalculated price on the display screen of the vehicle terminal in real time; After the order is completed, the current pricing parameters will automatically become invalid, and the vehicle terminal will revert to the original pricing parameters; The pricing parameters include basic fee, mileage fee, duration fee, surcharge and discount.

10. An intelligent management system for online car-hailing according to any one of claims 1 or 3, characterized in that: The pricing module adjusts the pricing parameters in real time according to the supply and demand adjustment coefficient of the dispatching module provided by the dispatching module; Among them, if the supply-demand adjustment coefficient is > 1, the basic fee, mileage fee, duration fee will be reduced, and the surcharge will be reduced or cancelled; If the supply-demand adjustment factor is < 1, the basic fee, mileage fee, duration fee and surcharge will increase.

Images