A dynamic pricing method for offline commercial space based on passenger flow prediction

Abstract

This invention discloses a dynamic pricing method for offline commercial spaces based on passenger flow prediction, belonging to the field of commercial pricing technology. The method includes: constructing a price-driven passenger flow evolution model based on multi-source passenger flow observation records for different zones, employing a passenger flow carrying capacity characterization layer, a spatial structure entropy constraint layer, and a price response mapping layer; performing forward price extrapolation based on the price-driven passenger flow evolution model to obtain the passenger flow evolution trajectory for each zone, and performing time integral normalization to generate a predicted passenger flow spatial structure; based on the predicted passenger flow spatial structure, using a dispersion constraint labeling algorithm to mark price changes that cause an increase in the spatial occupancy dispersion index, and then filtering them out to obtain a set of steady-state feasible price changes; combining the set of steady-state feasible price changes with the passenger flow evolution trajectory for each zone to perform a revenue evaluation and obtain a dynamic pricing scheme. This invention improves the stability, sustainability, and decision rationality of pricing strategies through price-driven passenger flow evolution modeling and revenue consistency evaluation.

Description

Technical Field

[0001] This invention relates to the field of commercial pricing technology, and in particular to a dynamic pricing method for offline commercial spaces based on customer flow forecasting. Background Technology

[0002] In recent years, with the development of location awareness and big data analytics, business decision-making based on passenger flow data has become a research hotspot. Methods such as passenger flow statistics, passenger flow forecasting, and spatial analysis are gradually being introduced to characterize the scale, dwell time characteristics, and flow paths of passenger flow in different areas of commercial spaces, thus assisting in pricing. Simultaneously, at the algorithmic level, machine learning methods, especially reinforcement learning frameworks, are being explored. These methods iteratively optimize pricing strategies through historical revenue feedback, enabling price adjustments to have a degree of adaptability and gradually approach revenue targets in complex environments.

[0003] However, existing dynamic pricing methods for offline commercial spaces based on customer flow forecasting or reinforcement learning still have two shortcomings. On the one hand, most existing methods treat customer flow forecasting as an exogenous input, and price adjustments are only passive responses based on the forecast results, making it difficult for pricing strategies to characterize the dynamic coupling relationship between prices and customer flow. On the other hand, the revenue optimization process often focuses on maximizing short-term revenue, paying insufficient attention to the stability of the overall customer flow structure of the commercial space, lacking effective spatial structure constraints, and thus affecting the long-term operational stability and sustainable revenue capacity of the commercial space. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a dynamic pricing method for offline commercial spaces based on passenger flow forecasting to address the problems of insufficient coupling between price decisions and passenger flow evolution, as well as the lack of steady-state constraints on spatial structure.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a dynamic pricing method for offline commercial spaces based on passenger flow prediction. The method includes: collecting observational elements of price changes and pedestrian behavior, and dividing the offline commercial space into zones to obtain multi-source passenger flow observation records for each zone; constructing a price-driven passenger flow evolution model based on these observation records, using a passenger flow carrying capacity characterization layer, a spatial structure entropy constraint layer, and a price response mapping layer; performing forward price extrapolation based on the price-driven passenger flow evolution model to obtain the passenger flow evolution trajectory for each zone, and performing time integral normalization to generate a predicted passenger flow spatial structure; marking price changes that cause an increase in the spatial occupancy dispersion index using a dispersion constraint labeling algorithm based on the predicted passenger flow spatial structure, and then filtering them out to obtain a set of steady-state feasible price changes; and combining the set of steady-state feasible price changes with the passenger flow evolution trajectory for each zone to perform a revenue assessment and obtain a dynamic pricing scheme.

[0007] As a preferred embodiment of the offline commercial space dynamic pricing method based on passenger flow prediction described in this invention, the passenger flow behavior observation elements include the number of people entering and exiting, the duration of stay, and cross-zone movement.

[0008] As a preferred embodiment of the offline commercial space dynamic pricing method based on passenger flow prediction described in this invention, the specific steps for obtaining multi-source passenger flow observation records by zone are as follows: Collect price changes and pedestrian flow behavior data, align them over time, and obtain multi-source pedestrian flow observation data; Identify spatial locations where price changes occur in multi-source passenger flow observation data, and record the spatial locations of the number of people entering and exiting, the duration of stay, and cross-regional movement to obtain co-occurrence combinations of spatial locations; The spatial locations are combined to generate offline commercial space zones, and multi-source passenger flow observation data is collected to obtain multi-source passenger flow observation records for each zone.

[0009] As a preferred embodiment of the offline commercial space dynamic pricing method based on passenger flow prediction described in this invention, the step of constructing a price-driven passenger flow evolution model based on multi-source passenger flow observation records in different zones, employing a passenger flow carrying capacity characterization layer, a spatial structure entropy constraint layer, and a price response mapping layer, is as follows: The passenger flow carrying capacity characterization layer generates passenger flow carrying capacity characterization entries for each zone by aggregating and organizing the number of people entering and leaving, the duration of stay, and the cross-zone movement from multi-source passenger flow observation records in each zone. The spatial structure entropy constraint layer calculates the spatial structure entropy of the commercial space based on the passenger flow carrying capacity characterization items of the zone, and obtains the spatial structure entropy benchmark. The price response mapping layer aligns the passenger flow carrying capacity profiles of different zones with price changes over time, and generates price response mapping parameters through piecewise monotonic mapping under the constraint of spatial structure entropy benchmark. By hierarchically connecting and constraining the passenger flow carrying capacity characterization layer, the spatial structure entropy constraint layer, and the price response mapping layer, a price-driven passenger flow evolution model is constructed.

[0010] As a preferred embodiment of the offline commercial space dynamic pricing method based on passenger flow prediction described in this invention, the zonal passenger flow evolution trajectory is formed by using the price changes in the multi-source passenger flow observation records of the zonal area as input to perform forward price extrapolation.

[0011] As a preferred embodiment of the offline commercial space dynamic pricing method based on passenger flow prediction described in this invention, the specific steps for generating the predicted passenger flow space structure are as follows: The evolution trajectory of passenger flow in different zones is processed by time integration to generate the time integral of passenger flow in different zones. The time integral of passenger flow in all offline commercial space zones is normalized to generate a predicted passenger flow spatial structure.

[0012] As a preferred embodiment of the offline commercial space dynamic pricing method based on passenger flow prediction described in this invention, the specific steps of marking price changes that cause an increase in the space occupancy dispersion index based on the predicted passenger flow spatial structure using a dispersion constraint labeling algorithm are as follows. The spatial occupancy dispersion index is calculated based on the predicted passenger flow spatial structure, and then arranged in chronological order to generate a spatial occupancy dispersion record sequence. The spatial occupancy dispersion record sequence is cumulatively expanded, and the spatial structure entropy benchmark is used as the constraint boundary to mark the price changes that cause the spatial occupancy dispersion index to rise, thus obtaining a price mark set.

[0013] As a preferred embodiment of the offline commercial space dynamic pricing method based on passenger flow prediction described in this invention, the steady-state feasible price change set is obtained by filtering out price changes marked as causing an increase in the space occupancy dispersion index based on the price label set.

[0014] As a preferred embodiment of the offline commercial space dynamic pricing method based on passenger flow forecasting described in this invention, the specific steps for evaluating revenue by combining the steady-state feasible price change set with the regional passenger flow evolution trajectory are as follows: The set of steady-state feasible price changes and the regional passenger flow evolution trajectory are combined using the time-aligned correlation expansion method to generate the regional revenue contribution trajectory. Based on the revenue contribution trajectory of each region, groups are formed according to price changes, and a revenue contribution consistency score is calculated.

[0015] As a preferred embodiment of the offline commercial space dynamic pricing method based on passenger flow prediction described in this invention, the method of obtaining dynamic pricing schemes refers to sorting the revenue contribution consistency scores in the set of steady-state feasible price changes, and selecting the target price change through the consistency priority screening method.

[0016] The beneficial effects of this invention are as follows: by modeling the evolution of customer flow driven by price, constraining the stability of spatial structure and evaluating the consistency of revenue, the dynamic pricing of offline commercial spaces is optimized, which improves the stability, sustainability and rationality of pricing strategies in complex customer flow environments. It is applicable to offline business operation management and intelligent pricing scenarios based on reinforcement learning. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a dynamic pricing method for offline commercial spaces based on customer flow forecasting.

[0019] Figure 2 A flowchart for obtaining multi-source passenger flow observation records for different zones.

[0020] Figure 3 A flowchart for constructing a price-driven passenger flow evolution model.

[0021] Figure 4 A flowchart for obtaining a dynamic pricing scheme.

[0022] Figure 5 This is a comparison diagram of the evolution of spatial structure entropy.

[0023] Figure 6 This is a comparison chart of the variation in space occupancy dispersion. Detailed Implementation

[0024] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0025] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0026] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0027] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a dynamic pricing method for offline commercial spaces based on customer flow prediction, including the following steps: S1: Collect data on price changes and pedestrian behavior, and divide offline commercial spaces into zones to obtain multi-source pedestrian flow observation records for each zone.

[0028] S1.1: Collect price changes and pedestrian flow behavior observation elements, and perform time alignment to obtain multi-source passenger flow observation data.

[0029] Specifically, price sensors are used to read the price values ​​recorded in electronic price tags and POS terminals. When the price value corresponding to a spatial location is adjusted, the price before the adjustment, the price after the adjustment, and the time information of the price change are recorded and organized into a price change data in chronological order.

[0030] The personnel counting sensor uses infrared beams, video recognition, and millimeter wave sensing to continuously count the number of people entering and leaving a spatial location. It records the number of people entering and leaving the corresponding spatial location within a continuous time scale to obtain the number of people entering and leaving.

[0031] The video acquisition sensor continuously collects images of people's activities in the space and records the time points when people appear and disappear. The wireless signal sensing sensor obtains the time information of people entering and leaving the space by recording the first appearance time and the last disappearance time of the mobile terminal signal in the space, and obtains the duration of stay based on the time interval between the entry time and the exit time.

[0032] The system uses video acquisition sensors to continuously track changes in the location information of people in different spatial locations. When a person moves from one spatial location to another, it records the starting spatial location, the target spatial location, and the time of the change. The system also organizes multiple location changes to generate cross-zone movement information.

[0033] The elements of human movement observation include the number of people entering and leaving, the length of stay, and cross-regional movement.

[0034] S1.2: Identify the spatial locations where price changes occur in multi-source passenger flow observation data, and record the spatial locations of the number of people entering and leaving, the duration of stay, and cross-regional movement to obtain the co-occurrence combinations of spatial locations.

[0035] Specifically, the observation entries containing price values ​​and corresponding time information are read one by one from the multi-source passenger flow observation data. By comparing the price values ​​of the same spatial location under adjacent time information, the spatial locations where the price values ​​change are obtained.

[0036] For each spatial location where a price value changes, the spatial location corresponding to the number of people entering and exiting, the spatial location corresponding to the duration of stay, and the starting and target spatial locations corresponding to cross-regional movement are synchronously read from multi-source passenger flow observation data.

[0037] The spatial locations where price values ​​change are arranged in parallel with the spatial locations corresponding to the number of people entering and exiting, the duration of stay, and the starting and target spatial locations corresponding to cross-regional movement under the same time information, to obtain the co-occurrence combinations of spatial locations.

[0038] S1.3: Merge the co-occurrence combinations of spatial locations to generate offline commercial space partitions, and combine them with multi-source passenger flow observation data to obtain multi-source passenger flow observation records for each partition.

[0039] Specifically, each spatial location combination record in the spatial location co-occurrence combination is read one by one, and the spatial locations contained in the spatial location combination record are compared; when the same spatial location appears in different spatial location combination records, the same spatial locations are merged to form a spatial location set as an offline commercial space partition.

[0040] The data on price changes, number of people entering and exiting, length of stay, and cross-zone movement in the multi-source passenger flow observation data are read one by one. Based on the spatial location information recorded in the multi-source passenger flow observation data, the corresponding price changes, number of people entering and exiting, length of stay, and cross-zone movement are written into the offline commercial space zones respectively to obtain the multi-source passenger flow observation records of the zones.

[0041] S2: Based on multi-source passenger flow observation records in different regions, a price-driven passenger flow evolution model is constructed by adopting a passenger flow carrying capacity characterization layer, a spatial structure entropy constraint layer, and a price response mapping layer.

[0042] S2.1: The passenger flow carrying capacity characterization layer generates passenger flow carrying capacity characterization entries for each zone by aggregating and organizing the number of people entering and leaving, the length of stay, and the cross-zone movement in the multi-source passenger flow observation records of each zone.

[0043] Specifically, the passenger flow carrying capacity characterization layer reads the multi-source passenger flow observation records of each zone according to the offline commercial space zone; under the same time information, the starting spatial location and target spatial location corresponding to the number of people entering, leaving, stay duration and cross-zone movement recorded in the multi-source passenger flow observation records of each zone are written into the respective offline commercial space zone.

[0044] Within each offline commercial space zone, the number of people entering and leaving is merged and organized, and the total stay time is counted; the number of times people enter and leave across zones is summarized and organized to obtain the number of times people enter and leave across zones.

[0045] The number of people entering, leaving, stay duration, number of times entering across zones and number of times leaving across zones in the same offline commercial space zone at the same time are aggregated and listed side by side, and the corresponding time information and the offline commercial space zone identifier are attached to generate a zone passenger flow carrying capacity characterization item.

[0046] S2.2: The spatial structure entropy constraint layer calculates the spatial structure entropy of commercial space based on the passenger flow carrying capacity characterization items of the partition, and obtains the spatial structure entropy benchmark.

[0047] Specifically, the spatial structure entropy constraint layer reads the passenger flow carrying capacity description items of each zone one by one according to the time information. Under the same time information, it gathers the passenger flow carrying capacity description items corresponding to all offline commercial space zones, and extracts the number of people entering, leaving, stay duration, number of times entering across zones and number of times leaving across zones from the passenger flow carrying capacity description items to form a set of passenger flow carrying capacity scales of each zone.

[0048] The passenger flow carrying capacity of each zone is matched with the time information to obtain the passenger flow carrying capacity value of each zone under the same time information; the passenger flow carrying capacity of all zones under the same time information is counted to form the total carrying capacity; the ratio of the passenger flow carrying capacity of each zone to the total carrying capacity under the corresponding time information is taken as the passenger flow proportion of each zone under the corresponding time information, and arranged according to the spatial location order to generate the passenger flow spatial allocation structure.

[0049] For each time information, the spatial structure entropy of the commercial space is calculated based on the spatial distribution structure of passenger flow, and the spatial structure entropy obtained under different time information is arranged and organized in chronological order; the time average of the spatial structure entropy obtained within the observation time range is taken as the benchmark of spatial structure entropy.

[0050] The expression for calculating the spatial structure entropy is: ; in, For spatial structure entropy, The number of zones for offline commercial space. Index numbers for the partitioning of offline commercial spaces. For the first The capacity of each offline commercial space zone to handle customer traffic at the corresponding time. It is the natural logarithm function.

[0051] It should be noted that in the expression for calculating the spatial structure entropy, the passenger flow carrying capacity of each offline commercial space zone is calculated by ratio using the dimension of people, forming a dimensionless passenger flow proportion. The independent variable of the natural logarithm function is a dimensionless quantity, and the dimensions are kept consistent and reasonable.

[0052] S2.3: The price response mapping layer aligns the passenger flow carrying capacity characterization items of the zones with price changes over time, and generates price response mapping parameters through piecewise monotonic mapping under the constraint of the spatial structure entropy benchmark.

[0053] Specifically, the price response mapping layer reads the passenger flow capacity profile items and price changes of each zone according to the offline commercial space zone identifier and time information. By comparing the time information of the passenger flow capacity profile items and price changes of each zone, a one-to-one correspondence is formed between each price change and the passenger flow capacity profile item recorded under the same time information.

[0054] Price changes within the same offline commercial space are sorted by price value, and the sorted price changes are divided into continuous price ranges. Within each price range, the time-aligned passenger flow carrying capacity profiles are aggregated and organized to obtain passenger flow carrying capacity aggregated profiles corresponding to the price range.

[0055] The passenger flow carrying capacity aggregation entries corresponding to adjacent price ranges are checked sequentially. When the direction of change of the passenger flow carrying capacity aggregation entries is inconsistent with the direction of price change (for example, when the direction of price change is an increase, but the passenger flow carrying capacity aggregation entries show an increase in carrying capacity, it indicates that the direction of change of the passenger flow carrying capacity aggregation entries is inconsistent with the direction of price change), the adjacent price ranges are merged and the corresponding passenger flow carrying capacity aggregation entries are regenerated.

[0056] The final price range, the corresponding passenger flow carrying capacity aggregation items, and the spatial structure entropy benchmark constraint information are arranged side by side to generate price response mapping parameters.

[0057] S2.4: Connect and couple the passenger flow carrying capacity characterization layer, spatial structure entropy constraint layer, and price response mapping layer in a hierarchical manner to construct a price-driven passenger flow evolution model.

[0058] Specifically, the passenger flow carrying capacity characterization items formed in the passenger flow carrying capacity characterization layer, the spatial structure entropy benchmark formed in the spatial structure entropy constraint layer, and the price response mapping parameters formed in the price response mapping layer are aligned and organized one by one according to the offline commercial space partition identifier and time information, so that each offline commercial space partition is simultaneously associated with the corresponding passenger flow carrying capacity characterization items, spatial structure entropy benchmark, and price response mapping parameters under each time information.

[0059] The price response mapping parameter is used as the mapping basis for the effect of price changes on the passenger flow carrying capacity of the zones; the spatial structure entropy benchmark is used as the constraint condition to limit the magnitude of changes in passenger flow carrying capacity caused by price changes; and the passenger flow carrying capacity of the zones is used as the starting characterization basis for passenger flow evolution.

[0060] By jointly applying the price response mapping parameters and the spatial structure entropy benchmark, the passenger flow carrying capacity characterization items of each zone are updated under the influence of price changes, forming a continuous and traceable process of passenger flow carrying capacity change in the time dimension. The evolutionary relationship composed of the update process of passenger flow carrying capacity characterization items, the price response mapping parameters, and the spatial structure entropy benchmark is expressed in a unified manner, generating a price-driven passenger flow evolution model.

[0061] The multi-source passenger flow observation records of different zones and the corresponding price changes are input into the price-driven passenger flow evolution model. The price changes and the passenger flow carrying capacity of different zones are aligned and organized according to the time information. Based on the spatial structure entropy benchmark, the passenger flow carrying capacity changes in different price change intervals are gradually solidified to solidify the correspondence between price changes and the evolution of passenger flow carrying capacity of different zones, and the trained price-driven passenger flow evolution model is obtained.

[0062] It should be noted that the constraint coupling uses the spatial structure entropy benchmark as a constraint condition to limit the range of price changes described in the price response mapping layer, so that when price changes act on the passenger flow carrying capacity characterization items of the partition, they do not exceed the stable range of passenger flow spatial distribution defined by the spatial structure entropy benchmark.

[0063] S3: Perform forward price extrapolation based on the price-driven passenger flow evolution model to obtain the passenger flow evolution trajectory of each zone, and perform time integral normalization to generate the predicted passenger flow spatial structure.

[0064] S3.1: Based on the price-driven passenger flow evolution model, price changes in multi-source passenger flow observation records of different zones are used as input to perform forward price extrapolation and obtain the passenger flow evolution trajectory of different zones.

[0065] Specifically, based on the price-driven passenger flow evolution model, the multi-source passenger flow observation records of each zone are sorted and organized according to time information; the price changes of the corresponding offline commercial space zones in the multi-source passenger flow observation records of each zone are read one by one, and the price changes are aligned and combined with the passenger flow carrying capacity characterization items of the zone under the same time information to form the price forward extrapolation input items.

[0066] The price forward extrapolation input items are sequentially input into the price-driven passenger flow evolution model according to the time information sequence. The price response mapping parameters in the price-driven passenger flow evolution model are used to update the passenger flow carrying capacity characterization items of each zone, thereby obtaining the passenger flow carrying capacity change records of each time information under the influence of price changes.

[0067] By serially organizing the records of changes in passenger flow in the same offline commercial space under continuous time information, the evolution trajectory of passenger flow in each zone can be obtained.

[0068] S3.2: Perform time integration on the evolution trajectory of passenger flow in each zone to generate the time integral of passenger flow in each zone.

[0069] Specifically, the evolution trajectory of passenger flow in each zone is organized sequentially according to time information to obtain a sequence of passenger flow carrying capacity changes for each offline commercial space zone under continuous time information; the passenger flow carrying capacity change records corresponding to adjacent time information are read in sequence, and the time interval between time information is used as the cumulative scale to statistically analyze the passenger flow carrying capacity change records of each zone, forming a cumulative passenger flow volume of each zone that accumulates continuously with time information.

[0070] The cumulative passenger flow of the same offline commercial space within the time range is summarized and organized to generate the passenger flow time integral of the zone.

[0071] S3.3: Normalize the time integral of passenger flow in all offline commercial space zones to generate the predicted passenger flow spatial structure.

[0072] Specifically, time information is matched and organized with the time points of passenger flow in each zone to form a set of time points of passenger flow covering all offline commercial space zones under the same time information; the total amount of the set of time points of passenger flow in each zone is calculated, and the ratio of the time points of passenger flow in each offline commercial space zone to the total amount of the set of time points of passenger flow in each zone is used as the passenger flow percentage of each offline commercial space zone under the corresponding time information.

[0073] The proportion of customer flow corresponding to each offline commercial space zone is arranged and organized according to spatial location to generate a predicted customer flow spatial structure.

[0074] It should be noted that the predicted passenger flow spatial structure describes the distribution of passenger flow proportions in offline commercial space zones within the future time range under the influence of price changes. The predicted passenger flow spatial structure is obtained by time integration and normalization of the passenger flow evolution trajectory of the zones, reflecting the degree of occupancy of different offline commercial space zones in the overall passenger flow, and providing a basis for subsequent space occupancy dispersion assessment and dynamic pricing scheme generation.

[0075] like Figure 5The simulation illustrates the differences in the stability of predicted customer flow structure in offline commercial spaces under different pricing schemes. A control group, an ablation group, and a complete group were set up and run under the same customer flow input conditions, differing only in the modeling and constraints used in the pricing schemes. The control group uses a basic pricing scheme that does not introduce price-driven customer flow evolution coupling; it only adjusts prices based on the current customer flow state, without constructing a mapping relationship between price changes and subsequent customer flow evolution, nor performing price forward inference, and therefore does not generate a predicted customer flow spatial structure. Ablation group one introduces price-driven customer flow evolution modeling and price forward inference on top of the control group, generating a predicted customer flow spatial structure, but does not calculate the spatial structure entropy index for the predicted structure. Ablation group two further uses spatial structure entropy calculation to limit the drastic fluctuations in spatial structure caused by price inference, but does not perform subsequent dispersion filtering. The complete group simultaneously introduces price-driven customer flow evolution modeling, price forward inference, spatial structure entropy constraints, and subsequent constraints, constituting the dynamic pricing scheme proposed in this invention. Figure 5 It is evident that the complete group exhibits a more gradual trend in spatial structure entropy change over most time steps, indicating that introducing spatial structure entropy constraints during the prediction of passenger flow spatial structure generation helps to improve the stability of spatial structure evolution.

[0076] S4: Based on the predicted passenger flow spatial structure, use the dispersion constraint labeling algorithm to mark the price changes that cause the space occupancy dispersion index to rise, and then filter them out to obtain the set of steady-state feasible price changes.

[0077] S4.1: Calculate the space occupancy dispersion index based on the predicted passenger flow spatial structure, and generate a space occupancy dispersion record sequence by arranging them in chronological order.

[0078] Specifically, the predicted passenger flow spatial structure is first organized sequentially according to time information to form a sequence of predicted passenger flow spatial structures covering continuous time information; under each time information, the passenger flow percentage corresponding to each offline commercial space zone in the predicted passenger flow spatial structure is read one by one; based on the passenger flow percentage of all offline commercial space zones under the same time information, the spatial occupancy dispersion index under the corresponding time information is calculated, and the expression is: ; in, This is an index of spatial occupancy dispersion. For the first The proportion of customer traffic in the predicted customer traffic structure of each offline commercial space zone. This serves as a reference value for the even distribution of customer traffic proportions within the zoning of offline commercial spaces.

[0079] It should be noted that the expression for calculating the space occupancy dispersion index involves a dimensionless passenger flow percentage, and the reference value is dimensionless. Therefore, the expression for calculating the space occupancy dispersion index has a unified and reasonable dimension.

[0080] The spatial occupancy dispersion index corresponding to each time information is paired and recorded with the corresponding time information; the spatial occupancy dispersion index is arranged and organized according to the chronological order of the time information to generate a spatial occupancy dispersion record sequence that reflects the relationship between spatial occupancy dispersion and time.

[0081] S4.2: Accumulate and expand the spatial occupancy dispersion record sequence, and use the spatial structure entropy benchmark as the constraint boundary to mark the price changes that cause the spatial occupancy dispersion index to rise, and obtain the price mark set.

[0082] Specifically, the spatial occupancy dispersion index corresponding to adjacent time information is compared item by item to obtain the change in spatial occupancy dispersion between adjacent time information; the change in spatial occupancy dispersion is arranged in the order of time information, and the sum of the change in spatial occupancy dispersion corresponding to the current time information and the cumulative change in spatial occupancy dispersion of the previous time information is taken as the cumulative change in spatial occupancy dispersion; the cumulative changes in spatial occupancy dispersion between all adjacent time information are summarized to generate a cumulative expansion sequence of spatial occupancy dispersion changes.

[0083] The spatial structure entropy benchmark is used as a constraint boundary (to limit the stable range of overall customer flow distribution in commercial space), and the cumulative unfolded sequence of spatial occupancy dispersion change is matched with the spatial structure entropy benchmark under the corresponding time information. When the spatial occupancy dispersion change exceeds the constraint boundary, the price change associated with the corresponding time information is marked as the price change that causes the spatial occupancy dispersion index to rise. All marked price changes are summarized and organized according to time information to obtain a price mark set.

[0084] S4.3: Based on the price tag set, filter out price changes that are marked as causing an increase in the spatial occupancy dispersion index to obtain the set of steady-state feasible price changes.

[0085] Specifically, the price changes obtained from the multi-source passenger flow observation records of the zones are sorted according to the time information order to form a price change item sequence; the price changes in the price change item sequence are read one by one and compared with the price changes recorded in the price mark set.

[0086] Under the same time information and the same spatial location conditions, the price changes in the price change item sequence are compared item by item with the price changes in the price tag set; when a price change appears in the price tag set, the corresponding price change is deleted from the price change item sequence; when a price change does not appear in the price tag set, the corresponding price change is retained; the retained price changes are summarized and organized in chronological order to form a steady-state feasible price change set containing only untagged price changes.

[0087] It should be noted that the dispersion constraint labeling algorithm sorts and accumulates the spatial occupancy dispersion index obtained from the predicted passenger flow spatial structure in a time sequence, and combines it with the spatial structure entropy benchmark as the constraint boundary to label the situations that cause the spatial occupancy dispersion index to rise under different price change conditions. This enables the identification and screening of price changes that may disrupt the stability of passenger flow occupancy in commercial spaces, providing a basis for obtaining the set of steady-state feasible price changes in the future.

[0088] like Figure 6 The results shown are based on simulations of the control group, ablation group, and complete group. The control group did not introduce price-driven passenger flow evolution coupling, nor did it impose any spatial constraints on price changes; the price adjustment process did not consider the predicted space occupancy status. Ablation group one introduced price-driven passenger flow evolution modeling and forward price extrapolation, but did not impose constraints on spatial structure or space occupancy status, therefore price changes could still lead to fluctuations in space occupancy distribution. Ablation group two introduced spatial structure entropy constraints in the pricing process, limiting the overall stability of the predicted passenger flow structure, but did not mark or filter out price changes that might increase space occupancy dispersion. The complete group introduced price change marking and filtering based on space occupancy dispersion, eliminating price changes that caused imbalances in spatial resource distribution. Figure 6 It is evident that the complete group maintained a continuous level of spatial occupancy dispersion throughout the entire simulation period, indicating that the dispersion constraint screening can effectively suppress the spatial occupancy imbalance caused by price adjustments, thereby improving the balance and sustainability of offline commercial space operation.

[0089] S5: Combine the set of steady-state feasible price changes with the evolution trajectory of passenger flow in different areas to conduct revenue assessment and obtain a dynamic pricing scheme.

[0090] S5.1: Combine the set of steady-state feasible price changes with the regional passenger flow evolution trajectory using the time-aligned correlation expansion method to generate the regional revenue contribution trajectory.

[0091] Specifically, the set of steady-state feasible price changes is organized sequentially according to time information to form a sequence of steady-state feasible price change entries arranged by time information; the offline commercial space zoning identifiers and time information are used to organize the zoning passenger flow evolution trajectory to form a sequence of zoning passenger flow evolution trajectory entries covering each offline commercial space zoning and arranged by time information.

[0092] The time-aligned correlation expansion method is adopted to expand the steady-state feasible price change item with the partition customer flow evolution trajectory items corresponding to all offline commercial space partitions under the same time information in a one-to-many correlation, so that each steady-state feasible price change item corresponds to a set of partition customer flow evolution trajectories including all offline commercial space partitions.

[0093] Based on the offline commercial space zoning identifiers, the evolution trajectory of passenger flow in each zone is organized, and the corresponding price change information is recorded in conjunction with the steady-state feasible price change entries to generate the zone revenue contribution trajectory.

[0094] It should be noted that the alignment correlation expansion method is used to correlate different price changes with passenger flow evolution information. It synchronizes the set of steady-state feasible price changes and the passenger flow evolution trajectory of each region according to unified time information, so that the price change records under the same time information and the corresponding passenger flow evolution records of each region are established to provide a structured basis for subsequent revenue assessment and price screening.

[0095] S5.2: Group the revenue contribution trajectories based on price changes and calculate the revenue contribution consistency score.

[0096] Specifically, price changes in the steady-state feasible price change set are used as group identifiers, and the revenue contribution trajectories corresponding to the same price change in the regional revenue contribution trajectory are grouped into the same group; based on each price change group, the revenue change records corresponding to each offline commercial space zone in the regional revenue contribution trajectory are read in chronological order, and the revenue contribution consistency score is calculated, expressed as: ; in, Consistency score for contribution to revenue. The amount of time information covered by grouping price changes. Grouping price changes by time information The next Record the revenue changes of each offline commercial space zone. For time information The average revenue change record for all offline commercial space zones in the region.

[0097] It should be noted that, since the revenue change records are only compared and proportionalized between the same physical quantities during the calculation process, the resulting revenue contribution consistency score is a dimensionless index, and the overall dimensions remain consistent.

[0098] S5.3: Sort the revenue contribution consistency scores in the set of steady-state feasible price changes, and select the target price change through the consistency priority screening method to form a dynamic pricing scheme.

[0099] Specifically, each price change in the set of steady-state feasible price changes is used as an index to centrally organize the corresponding revenue contribution consistency scores, and the price changes in the set of steady-state feasible price changes are uniformly sorted according to the magnitude of the revenue contribution consistency scores.

[0100] The consistency-first screening method is used to read the price changes sequentially from the ranking results, and the price change with the highest revenue contribution consistency score that has not been screened out is selected as the target price change in combination with the set of steady-state feasible price changes. The target price change is determined as the price adjustment method to be implemented by the offline commercial space in the corresponding decision-making cycle, thus forming a dynamic pricing scheme.

[0101] It should be noted that dynamic pricing schemes refer to price adjustment methods determined by sorting and filtering different price changes within the constraints of the set of steady-state feasible price changes, combined with the consistency score of regional customer flow evolution trajectory and revenue contribution; it can achieve synergistic consistency of revenue changes in offline commercial spaces and provide an executable and sustainable basis for price adjustments in offline commercial spaces within the corresponding decision-making cycle.

[0102] The consistency-first screening method sorts the revenue contribution consistency scores of price changes in the set of steady-state feasible price changes, prioritizes price changes that can generate synchronous revenue changes in different offline commercial space zones, and gradually screens them according to the revenue contribution consistency scores from high to low when there are multiple optional price changes, to ensure that the dynamic pricing scheme has sustainability and synergy at the overall operation level.

[0103] In summary, this invention optimizes the dynamic pricing of offline commercial spaces through price-driven customer flow evolution modeling, spatial structure stability constraints, and revenue consistency assessment. This improves the stability, sustainability, and decision rationality of pricing strategies in complex customer flow environments and is applicable to offline commercial operation management and intelligent pricing scenarios based on reinforcement learning.

[0104] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for dynamic pricing of offline commercial space based on passenger flow prediction, characterized in that: include, Collect data on price changes and pedestrian behavior, and divide offline commercial spaces into zones to obtain multi-source pedestrian flow observation records for each zone; Based on multi-source passenger flow observation records in different regions, a price-driven passenger flow evolution model is constructed by employing a passenger flow carrying capacity characterization layer, a spatial structure entropy constraint layer, and a price response mapping layer. Based on the price-driven passenger flow evolution model, forward price extrapolation is performed to obtain the passenger flow evolution trajectory of each zone, and time integral normalization is performed to generate the predicted passenger flow spatial structure. Based on the predicted passenger flow spatial structure, price changes that cause an increase in the spatial occupancy dispersion index are marked by a dispersion constraint labeling algorithm and then filtered out to obtain a set of steady-state feasible price changes. By combining the set of steady-state feasible price changes with the evolution trajectory of passenger flow in different areas, a revenue assessment is conducted to obtain a dynamic pricing scheme.

2. The method for dynamic pricing of offline commercial spaces based on customer flow forecasting as described in claim 1, characterized in that: The observed elements of pedestrian flow behavior include the number of people entering and leaving, the duration of stay, and cross-regional movement.

3. The method for dynamic pricing of offline commercial spaces based on customer flow forecasting as described in claim 2, characterized in that: The specific steps for obtaining multi-source passenger flow observation records by region are as follows. Collect price changes and pedestrian flow behavior data, align them over time, and obtain multi-source pedestrian flow observation data; Identify spatial locations where price changes occur in multi-source passenger flow observation data, and record the spatial locations of the number of people entering and exiting, the duration of stay, and cross-regional movement to obtain co-occurrence combinations of spatial locations; The spatial locations are combined to generate offline commercial space zones, and multi-source passenger flow observation data is collected to obtain multi-source passenger flow observation records for each zone.

4. The method for dynamic pricing of offline commercial spaces based on customer flow forecasting as described in claim 3, characterized in that: Based on multi-source passenger flow observation records in different zones, a price-driven passenger flow evolution model is constructed using a passenger flow carrying capacity characterization layer, a spatial structure entropy constraint layer, and a price response mapping layer. The specific steps are as follows. The passenger flow carrying capacity characterization layer generates passenger flow carrying capacity characterization entries for each zone by aggregating and organizing the number of people entering and leaving, the duration of stay, and the cross-zone movement from multi-source passenger flow observation records in each zone. The spatial structure entropy constraint layer calculates the spatial structure entropy of the commercial space based on the passenger flow carrying capacity characterization items of the zone, and obtains the spatial structure entropy benchmark. The price response mapping layer aligns the passenger flow carrying capacity profiles of different zones with price changes over time, and generates price response mapping parameters through piecewise monotonic mapping under the constraint of spatial structure entropy benchmark. By hierarchically connecting and constraining the passenger flow carrying capacity characterization layer, the spatial structure entropy constraint layer, and the price response mapping layer, a price-driven passenger flow evolution model is constructed.

5. The method for dynamic pricing of offline commercial spaces based on customer flow forecasting as described in claim 4, characterized in that: The passenger flow evolution trajectory of the zone is based on a price-driven passenger flow evolution model, which uses price changes in multi-source passenger flow observation records of the zone as input to perform forward price extrapolation.

6. The method for dynamic pricing of offline commercial spaces based on customer flow forecasting as described in claim 5, characterized in that: The specific steps for generating the predicted passenger flow spatial structure are as follows: The evolution trajectory of passenger flow in different zones is processed by time integration to generate the time integral of passenger flow in different zones. The time integral of passenger flow in all offline commercial space zones is normalized to generate a predicted passenger flow spatial structure.

7. The method for dynamic pricing of offline commercial spaces based on passenger flow forecasting as described in claim 6, characterized in that: The step of using a dispersion constraint labeling algorithm to mark price changes that cause an increase in the space occupancy dispersion index, based on the predicted passenger flow spatial structure, is as follows: The spatial occupancy dispersion index is calculated based on the predicted passenger flow spatial structure, and then arranged in chronological order to generate a spatial occupancy dispersion record sequence. The spatial occupancy dispersion record sequence is cumulatively expanded, and the spatial structure entropy benchmark is used as the constraint boundary to mark the price changes that cause the spatial occupancy dispersion index to rise, thus obtaining a price mark set.

8. The method for dynamic pricing of offline commercial spaces based on passenger flow forecasting as described in claim 7, characterized in that: The set of steady-state feasible price changes is obtained by filtering out price changes marked as causing an increase in the spatial occupancy dispersion index, based on the set of price labels.

9. The method for dynamic pricing of offline commercial spaces based on passenger flow forecasting as described in claim 8, characterized in that: The revenue assessment, which combines the set of steady-state feasible price changes with the evolution trajectory of passenger flow in different zones, involves the following specific steps. The set of steady-state feasible price changes and the regional passenger flow evolution trajectory are combined using the time-aligned correlation expansion method to generate the regional revenue contribution trajectory. Based on the revenue contribution trajectory of each region, groups are formed according to price changes, and a revenue contribution consistency score is calculated.

10. The method for dynamic pricing of offline commercial spaces based on passenger flow forecasting as described in claim 9, characterized in that: The acquisition of dynamic pricing schemes refers to sorting the revenue contribution consistency scores in the set of steady-state feasible price changes, and selecting the target price change through a consistency-first screening method.

Images