Dynamic and interactive visual operation management method, device and terminal equipment

Abstract

The embodiments of the present application provide a dynamic and interactive visual operation management method, apparatus, and terminal device. The method includes: obtaining multidimensional operation data of a gas station in various time periods; based on the multidimensional operation data, dynamically modeling the customer traffic lines of the gas station in various time periods using a neural radiation field to obtain a crowd distribution model; classifying the gas station's station type according to the gas station's geographical location, the type of incoming vehicles, and the surrounding traffic conditions; using the crowd distribution model and the gas station's station type to simulate traffic line scheduling for the gas station, and dynamically adjusting the gas station's traffic line scheduling strategy based on the simulated station operation status in various time periods; constructing a dynamic and interactive traffic line optimization model based on multiple traffic line scheduling strategies and corresponding station operation status, and displaying it to the user, so that the user can select the traffic line scheduling plan to be executed based on the traffic line optimization model, thereby realizing intelligent management of the target gas station.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a dynamic and interactive visual operation management method, apparatus, and terminal device. Background Art

[0002] With the continuous increase in vehicle ownership, especially the explosive growth in private cars, daily commuting and refueling have become part of driver's daily lives. However, issues such as whether the gas station has the fuel number for their vehicle, congestion, queues, and estimated refueling wait times all impact the driver's daily experience.

[0003] However, the current management of gas stations is still mainly based on manual services, relying on operators to arrange management processes and manpower based on their own experience. This results in low operational efficiency and poor timeliness, and often leads to problems such as long queues and low traffic efficiency. Congestion, queue jumping, and irregular parking after vehicles enter the station will further exacerbate congestion during peak hours. In addition, safety supervision in the current gas station management process also relies heavily on manual labor, which can easily lead to safety hazards at gas stations due to negligence of operators or inadequate implementation of management regulations. Therefore, in response to the above technical problems, it is urgent to design a new technical solution to solve at least one of the above technical problems. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to provide a dynamic and interactive visual operation management method, device and terminal equipment, aiming to solve the technical problems of low operating efficiency and poor safety supervision effects in traditional gas stations.

[0005] In a first aspect, an embodiment of the present application provides a dynamic and interactive visual operation management method, including:

[0006] Obtain multi-dimensional operational data of gas stations at different time periods;

[0007] Based on the multi-dimensional operational data, a neural radiation field is used to dynamically model the customer flow of the gas station at each time period to obtain a customer flow distribution model at the gas station at each time period;

[0008] Gas stations are classified into station types based on their geographical location, the type of vehicles entering the station, and the surrounding traffic conditions. Station types can be urban commuting or long-distance refueling.

[0009] Using the passenger flow distribution model and the type of gas station, a traffic flow scheduling simulation is performed on the gas station. Based on the simulated station operation status in each time period, the traffic flow scheduling strategy for the gas station is dynamically adjusted to obtain multiple traffic flow scheduling strategies that match the gas station in each time period. The multiple traffic flow scheduling strategies include at least: a gas station personnel scheduling strategy, a gas station vehicle flow planning strategy, a gas station service operation strategy, and a gas station safety management strategy.

[0010] A variety of traffic flow scheduling strategies and corresponding station operation status are constructed into a dynamic and interactive traffic flow optimization model, which is displayed to users, allowing them to select the traffic flow scheduling plan they want to execute based on the traffic flow optimization model, thereby realizing intelligent management of the target gas station.

[0011] In a second aspect, an embodiment of the present application provides a dynamic and interactive visual operation management device, including:

[0012] The acquisition module is used to obtain multi-dimensional operational data of gas stations at different time periods;

[0013] A modeling module, configured to dynamically model the customer flow of the gas station at various time periods using a neural radiation field based on the multi-dimensional operational data, so as to obtain a customer flow distribution model of the gas station at various time periods;

[0014] The classification module is used to classify gas stations into station types based on their geographical location, the type of vehicles entering the station, and the surrounding traffic conditions. The station type can be urban commuting or long-distance supply.

[0015] a simulation module for performing a traffic flow scheduling simulation for the gas station using the passenger flow distribution model and the type of gas station to which the gas station belongs, and dynamically adjusting the traffic flow scheduling strategy for the gas station based on the simulated station operation status in each time period, thereby obtaining a plurality of traffic flow scheduling strategies that match the gas station in each time period; wherein the plurality of traffic flow scheduling strategies include at least: a gas station personnel scheduling strategy, a gas station vehicle flow planning strategy, a gas station service operation strategy, and a gas station safety management strategy;

[0016] The interactive module is used to construct a dynamic and interactive traffic flow optimization model based on multiple traffic flow scheduling strategies and corresponding station operation status, and display the traffic flow optimization model to the user, so that the user can select the traffic flow scheduling plan to be executed based on the traffic flow optimization model, thereby realizing intelligent management of the target gas station.

[0017] In a third aspect, an embodiment of the present application further provides a terminal device, comprising a processor and a memory for storing computer programs; the processor is used to execute the computer program and implement the dynamic and interactive visual operation management method described in the first aspect or any embodiment of the present application when executing the computer program.

[0018] The embodiments of the present application provide a dynamic and interactive visual operation management method, apparatus, and terminal device. In the method, multidimensional operational data of a gas station in various time periods is obtained; based on the multidimensional operational data, a neural radiation field is used to dynamically model the customer traffic flow of the gas station in various time periods to obtain a pedestrian flow distribution model of the gas station in various time periods; the gas station is classified into a station type according to its geographical location, the type of incoming vehicles, and the surrounding traffic conditions; wherein the station type is urban commuting type or long-distance supply type; the pedestrian flow distribution model and the station type are used to simulate the traffic flow scheduling of the gas station, and based on the station operation status in various time periods obtained by the simulation, the traffic flow scheduling strategy for the gas station is dynamically adjusted to obtain multiple traffic flow scheduling strategies that match the gas station in various time periods; wherein the multiple traffic flow scheduling strategies include at least: an in-station personnel scheduling strategy, an in-station vehicle flow planning, an in-station service operation strategy, and an in-station safety management strategy for the gas station; the multiple traffic flow scheduling strategies and the corresponding station operation status are constructed into a dynamic and interactive traffic flow optimization model, which is displayed to a user so that the user can select the traffic flow scheduling plan to be executed based on the traffic flow optimization model, thereby realizing intelligent management of the target gas station. This method can realize intelligent management of gas stations through crowd distribution model and traffic flow scheduling simulation, thereby improving the operational efficiency and safety of gas stations. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flowchart of a dynamic and interactive visual operation management method provided in an embodiment of the present application;

[0020] Figure 2 A schematic diagram of the module structure of a dynamic and interactive visual operation management device provided in an embodiment of the present application;

[0021] Figure 3 A schematic block diagram of the structure of a terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] The embodiments of the present application provide a dynamic and interactive visual operation management method, apparatus, and terminal device. The dynamic and interactive visual operation management method can be applied to a terminal device, which can be a control console in a gas station, or a gas station management system, as well as a mobile terminal that communicates with the management system, such as a mobile phone, virtual reality device, tablet computer, laptop computer, desktop computer, wearable device, or other electronic device. The terminal device can be a server connected to the gas station management system, or a server cluster. The above-mentioned connection method can be implemented through a hardware circuit or a communication module.

[0023] The following is a detailed description of some embodiments of the present application in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Figure 1 , Figure 1 A flowchart of a dynamic and interactive visual operation management method provided in an embodiment of the present application.

[0024] like Figure 1 As shown, the dynamic and interactive visual operation management method includes the following steps:

[0025] Step S101: Acquire multi-dimensional operation data of a gas station in various time periods.

[0026] Step S102: Based on the multi-dimensional operation data, a neural radiation field is used to dynamically model the customer flow of the gas station in each time period to obtain a crowd distribution model of the gas station in each time period.

[0027] In an embodiment of the present application, the multi-dimensional operational data includes at least: the geographical location of the gas station, the type of vehicles entering the station in each time period, the volume of vehicles entering the station, the queuing time, the surrounding traffic conditions, consumption conditions, the working status of the personnel in the station, and the complaint situation.

[0028] The collection of multi-dimensional operational data is the foundation for intelligent gas station management. The following describes some data collection methods in step S101, combining various data types:

[0029] The gas station's geographic location: This can be directly obtained from the latitude and longitude information registered during gas station construction, which can be retrieved from the gas station management system or map annotation system. GPS or Beidou positioning equipment can also be used to accurately locate and calibrate the actual location of the gas station to ensure the accuracy of the geographic location data.

[0030] Vehicle types entering the gas station at different times: High-definition license plate recognition cameras are deployed at gas station entrances. License plate information is captured through license plate recognition technology and then combined with a license plate database to correlate and analyze vehicle types. Furthermore, deep learning image recognition algorithms analyze vehicle appearance features (such as size, shape, and presence of a cargo box) to further accurately determine vehicle type, distinguishing between private cars, trucks, and buses.

[0031] Incoming vehicle flow: Ground induction coils or infrared sensors are installed at the gas station entrance lanes. When vehicles pass by, the sensors detect signal changes and count the number of vehicles passing, thereby obtaining real-time inbound vehicle flow. Furthermore, video analysis technology can be used in conjunction with surveillance cameras at gas station entrances and exits to identify vehicle movement and changes in number in the footage, allowing for real-time monitoring and statistics of vehicle flow, complementing and verifying sensor data.

[0032] Queue Time: By deploying cameras at key locations at gas stations, such as refueling lanes and checkout areas, and using target tracking algorithms in video analytics, we track and record each vehicle from the moment it enters the queue area until it completes service and leaves, thereby calculating the vehicle's queue duration. Furthermore, electronic displays are installed in the queue area to display the estimated wait time for vehicles currently in the queue in real time. This time data can also be collected and analyzed as part of the queue time data.

[0033] Surrounding traffic conditions: Connect to the traffic management department's open data platform to obtain real-time traffic flow, congestion, accident information, and other data on surrounding roads. APIs from third-party mapping services (such as AutoNavi and Baidu Maps) can also be used to obtain road condition information on surrounding roads, including road speeds and traffic control status, to fully understand the traffic situation around the gas station.

[0034] Consumption data: Utilizing the gas station's cash register system, detailed information on each transaction is recorded, including time of purchase, amount, type and quantity of fuel purchased, and details of non-fuel purchases. For mobile payment transactions, data integration with payment platforms (such as WeChat Pay and Alipay) allows for complete transaction data. Furthermore, RFID tags can be attached to convenience store items, and RFID reader / writer devices can be used to quickly and accurately calculate sales figures, providing more detailed data support for consumption data analysis.

[0035] Station personnel work status: Employees are equipped with smart ID badges with built-in positioning chips, accelerometers, and communication modules to collect real-time data such as their location, work hours, and work status (e.g., stationary, moving, busy). Furthermore, surveillance cameras are installed within gas stations, using computer vision technology to analyze employee work behavior, determining whether they are following standard procedures and whether they are engaging in passive work. Furthermore, employee-reported work logs and task completion status can be used to further understand employee work status.

[0036] Complaints: We have established a comprehensive online and offline complaint channel, with complaint boxes set up offline and online complaint portals provided through gas station official websites, mobile apps, WeChat official accounts, and other platforms. After a customer submits a complaint, relevant personnel will promptly enter information such as the complaint content, complaint time, and complaint recipient into the complaint management system for statistical analysis and subsequent processing, thereby comprehensively collecting complaint data.

[0037] Neural Radiance Fields (NeRF) is a technology that represents 3D scenes as neural networks. By learning the color and transparency (radiance field) of each point in space, it can generate high-resolution, high-quality 3D scene renderings. NeRF has been innovatively applied to dynamic scene modeling in gas station customer flow modeling.

[0038] It's important to note that traditional NeRF models are only applicable to static scenarios, whereas customer traffic at gas stations (such as vehicle ingress and egress, personnel movement) is dynamic. Therefore, this application introduces a time coordinate dimension. This allows the model to learn from time-of-day variations (e.g., dense morning rush hour traffic and sporadic nighttime traffic). For example, by labeling time periods with labels like "morning rush hour 7:00-9:00" and "lunchtime off-peak 12:00-14:00," the model automatically identifies customer behavior patterns during different time periods (e.g., morning rush hour customers tend to "fill up and go," while some customers visit convenience stores during lunch). Furthermore, to account for the differences between urban and highway stations, the model also incorporates characteristics such as gas station location (e.g., urban area, highway service area) and vehicle type (private cars, trucks) to generate a unique dynamic model. For example, when trucks arrive at a highway station at night, the model can identify complex traffic patterns, such as trucks taking a break and then shopping after refueling, while urban stations focus on quickly dispersing traffic during the morning rush hour. This enables independent modeling of multiple time periods and accurately matches demand.

[0039] Taking into account the lightweight deployment requirements, in the embodiments of the present application, a MobileNeRF variant model can be further adopted.

[0040] Based on the above principles, in an optional embodiment of step S102, based on the multi-dimensional operational data, a neural radiation field is used to dynamically model the customer flow at the gas station at various time periods to obtain a customer flow distribution model at the gas station at various time periods, including:

[0041] The multidimensional operational data is constructed into corresponding time series data according to the corresponding time periods, and combined with the spatial coordinates corresponding to the geographical location of the gas station to obtain an input vector; a MobileNeRF variant model with an added time coordinate dimension is used to predict the radiation field properties of the gas station in each time period based on the input vector, and obtain the three-dimensional spatial layout of the customer movement lines in the gas station in each time period; based on the three-dimensional spatial layout, the appearance probability, movement trajectory and aggregation density of customers in different functional areas of the gas station in each time period are obtained to form a corresponding crowd distribution model; an online learning mechanism is used, and if it is detected that the gas station layout is adjusted or new equipment is introduced, the newly added data is collected to perform local fine-tuning on the crowd distribution model to update the crowd distribution model.

[0042] In step S102, a refined dynamic analysis of gas station customer traffic patterns is achieved through spatiotemporal fusion modeling and online learning mechanisms. First, the system chronologically arranges the multidimensional data generated during gas station operations, such as traffic flow, queue lengths, and employee work status during each time period. This data is then combined with the spatial coordinates of the gas station, including specific locations such as gas pumps, cash registers, and convenience stores, to form an input vector with both temporal and spatial attributes. This processing method breaks the limitations of traditional data processing, which isolates operational indicators for analysis, and gives data such as traffic flow and staffing a spatiotemporal correlation. For example, during the morning rush hour between 7:00 and 9:00, the traffic flow data at the gas station entrance is bound to the spatial coordinates of the entrance. Combined with the gas station attendant deployment information during this period, it can clearly demonstrate the interactive relationship between space, personnel, and traffic flow during this time period.

[0043] Subsequently, an improved MobileNeRF variant model was adopted, innovatively adding a time dimension to traditional three-dimensional spatial modeling. This model, tailored to actual gas station application scenarios, lightweighted the network structure and, through techniques such as knowledge distillation, compressed the model size, enabling efficient operation on existing edge computing devices at gas stations. Based on the input vector, the model predicts the radiation field properties of the gas station at different time periods and generates a three-dimensional spatial layout of customer traffic during the corresponding time periods. Whether it's the busy morning rush hour with dense traffic and hurried customers, or the sparse nighttime traffic with some customers lingering for extended periods, the model accurately captures these differences. For example, from 10 PM to midnight, the model can clearly show the complete trajectory of truck drivers entering the gas station, from the refueling area to the rest area and then to the convenience store.

[0044] Based on the generated three-dimensional spatial layout, the system further mines the data to extract the probability of customers appearing in different functional areas within the gas station, their movement trajectories, and the density of their gatherings. This analysis revealed that during the morning rush hour, over 70% of customers choose to refuel and then quickly leave, while during the lunch hour, approximately 40% of customers visit the convenience store. By quantifying the concentration of people and vehicles in each area, high-risk areas prone to congestion, such as checkout counters and convenience store entrances and exits, can be accurately identified.

[0045] When a gas station undergoes layout adjustments, such as moving a convenience store entrance from the west to the east, or introducing new equipment like a mobile payment terminal, the system initiates online learning. There's no need to retrain the entire model; instead, the system collects data related to the newly changed area and fine-tunes the local parameters of the crowd distribution model. For example, with the convenience store entrance relocation, the system learned from approximately three days of new data, enabling it to quickly adapt to the new customer traffic flow and promptly identify potential issues arising from the adjustments. For example, the east lane experienced a brief midday traffic jam due to increased traffic flow, and based on this, it provided optimization recommendations for adding guides during that time.

[0046] This approach, based on spatiotemporal fusion modeling and online learning, leverages existing gas station surveillance cameras and other equipment to significantly reduce deployment costs. Furthermore, the ability to analyze data accurately down to specific spatial locations and minute-level time periods significantly improves analytical accuracy. Virtual simulations allow for rapid verification of the effectiveness of different scheduling strategies, shortening the decision-making cycle from several days to less than an hour, significantly improving response speed. Furthermore, as operational data accumulates, the model continuously learns new scenarios and changes, significantly improving the accuracy of predictions for special circumstances such as holiday traffic surges, enabling efficient and precise management of customer traffic at gas stations.

[0047] Furthermore, in step S102, a series of rules can be formulated based on the gas station's layout, traffic regulations, and customer behavior habits. These rules can be used to simulate customer movement paths and behaviors within the gas station. For example, vehicles entering the gas station can be required to drive in specific lanes, and customers can proceed to the cashier to pay after refueling. For example, a Petri net can be used to construct a model, representing the various areas of the gas station (such as the refueling area, cashier area, convenience store, etc.) as nodes, and customer movement and behavior as transitions between nodes. By defining the conditions and rules for these transitions, customer movements within the gas station can be simulated. This allows for the rapid construction of a basic model, enabling preliminary analysis and prediction of customer movement paths.

[0048] Further, optionally, in step S102, a large amount of customer behavior data can be collected, such as the customer's dwell time in various areas, movement direction, and probability of selecting a certain service, and then a probabilistic statistical method can be used to build a model. By analyzing historical data, the probability of customer behavior in different situations can be estimated, thereby predicting the customer's movement path. For example, a Markov chain model can be used to describe the probability of a customer transitioning between different states at a gas station. For example, after refueling in the gas station, there is a certain probability that the customer will go to the cashier to pay, and there is also a certain probability that they will go to the convenience store to shop before paying. By estimating these transition probabilities, the customer's entire movement process can be simulated. Therefore, the probabilistic statistical method can use historical data to reflect the general patterns of customer behavior and provide relatively accurate predictions for some common situations.

[0049] Step S103: Classify the gas station into a station type based on the gas station's location, the type of vehicles entering the station, and the surrounding traffic conditions. The station type can be either a city commuting type or a long-distance refueling type.

[0050] Urban commuter gas stations primarily serve daily commuters within the city and are typically located near major urban arterials, ring roads, or around office buildings and residential areas. These stations experience distinct morning and evening peak traffic patterns. Between 7:00 AM and 9:00 AM, and 5:00 PM and 7:00 PM, large numbers of private cars and ride-hailing services, primarily small family sedans, are concentrated there. Drivers often seek quick refueling before continuing their journeys, placing high demands on fuel efficiency. Station layouts often feature multiple parallel lanes to reduce waiting times, while also introducing quick payment methods like mobile payment and license plate recognition to further speed up refueling. Non-fuel services primarily include breakfast, coffee, and convenience foods to meet the immediate needs of commuters.

[0051] Long-distance refueling gas stations are mostly located in highway service areas and along national or provincial highways. They primarily serve long-distance trucks, buses, and self-driving tour vehicles. Due to the nature of long-distance driving, drivers require more than just refueling at gas stations; they also require comprehensive services such as rest, dining, and vehicle maintenance. These stations are large, with dedicated areas for large parking spaces and dedicated lanes for trucks to accommodate the entry, exit, and parking of large vehicles. In terms of fuel supply, demand for diesel far exceeds gasoline, meeting the fuel needs of trucks. Furthermore, stations are equipped with restaurants, convenience stores, driver lounges, and even offer services such as tire changes and minor vehicle repairs. These stations form a refueling hub for long-distance travel, ensuring drivers and passengers receive adequate rest and supplies after long drives, ensuring a safe and smooth onward journey.

[0052] Based on the above characteristics, for example, in step S103, the geographical location of the gas station, the types of incoming vehicles, and the surrounding traffic conditions are obtained. If the gas station is located on a main urban road or commercial district, it is determined whether the proportion of private cars among the incoming vehicles exceeds a first preset ratio; the first preset ratio is a fixed value or is set based on the city's size. If the proportion of private cars among the incoming vehicles exceeds the first preset ratio, it is determined whether the surrounding traffic trends conform to the characteristics of morning and evening rush hour commuting. If so, the gas station is determined to be an urban commuting station. If the gas station is located in a highway service area or along a national highway, it is determined whether the proportion of trucks and long-distance buses among the incoming vehicles exceeds a second preset ratio; the second preset ratio is a fixed value or is set based on the city's size. If the proportion of trucks and long-distance buses among the incoming vehicles exceeds the second preset ratio, it is determined whether the traffic around the gas station is primarily long-distance transit traffic. If so, the gas station is determined to be a long-distance supply station.

[0053] In addition to urban commuting and long-distance refueling types, gas station types can be further subdivided based on geographic location, vehicle type, traffic characteristics, and service functions. The following are several expanded types and their characteristics:

[0054] Highway Service Areas: Located within highway service areas, with no nearby residential or commercial areas, these areas serve only long-distance transit vehicles. These primarily consist of long-distance buses, freight trucks, and private cars, with trucks typically accounting for over 30%. Traffic is highly unidirectional (along the highway), with peak hours concentrated on holidays and during morning and evening long-distance travel. These stations typically feature large refueling areas, dedicated lanes for trucks, and parking areas for hazardous chemicals. They also offer a variety of long-distance services, including fast-food restaurants, rest areas, and vehicle maintenance. Fuel prices are generally higher than in urban areas but lower than at highway stations.

[0055] Suburban hubs: Located at the intersection of main arterial roads in the suburbs, at the entrance to a ring road, or in the urban-rural fringe. These hubs serve a mix of urban commuter vehicles (such as commuters' cars) and transit vehicles (such as logistics trucks and tourist buses). They accommodate urban commuter traffic during peak hours, see a large number of transit vehicles during off-peak hours, and may see a surge in tourist traffic on weekends. These hubs primarily cater to both the needs of small cars for quick refueling and trucks for temporary parking. Lane design should be tailored to vehicle type. They also offer convenience stores, simple dining options (such as hot-plates), and in-vehicle retail. Charging stations or LNG refueling stations can be added to meet the needs of new energy logistics vehicles.

[0056] Community-friendly: Located within residential areas, commercial districts, or office buildings, with a service radius typically less than 3 kilometers. Small private cars, ride-hailing services, and taxis are the primary users, with trucks rarely entering. Traffic is dispersed but frequent, with peak times during lunch and after get off work (6:00 PM to 9:00 PM). Stops are short (refueling and leaving immediately). The site layout is compact, prioritizing traffic efficiency (e.g., multi-lane parallel design). Non-fuel services primarily cater to high-frequency, essential needs (e.g., breakfast, cold drinks, and emergency medications). Quick payment options are supported (e.g., license plate recognition and app-based refueling reservations), reducing waiting times.

[0057] Scenic Area Dedicated: Located directly adjacent to tourist attractions, resorts, or on highway exits leading to scenic areas. These areas are primarily used by tour buses and private cars, with holiday traffic accounting for over 70% of annual traffic. They are highly seasonal, with traffic surges on weekends and short holidays, often leading to queues and congestion. This type of station primarily requires dedicated lanes and temporary parking for buses to prevent mixing with cars. Non-fuel services focus on tourist amenities (such as maps, sunscreen, and scenic area ticket sales). Fuel prices may be slightly higher in scenic areas, but value-added services such as car washes and luggage storage are offered to attract customers.

[0058] Logistics Park Type: Located adjacent to logistics parks, industrial parks, or freight hubs (such as ports and railway freight stations). Heavy trucks, cold chain trucks, and container trucks primarily account for over 90% of the fleet. Traffic flows continuously day and night, with peak refueling times from early morning to mid-morning (e.g., 3:00 AM to 10:00 AM), when diesel demand dominates. This type of station primarily requires extra-large refueling spaces (supporting trucks over 16 meters long) and dual-gun, high-flow diesel fuel dispensers. It also provides rest areas for truck drivers (such as showers, laundry facilities, and dining options), tire replacement, and simple repair services. It also supports business-to-business services such as fleet batch settlement and fuel card pre-deposits. Some stations can also integrate with logistics management systems.

[0059] Rural Economy Stations are located along main town roads, county highways, or near agricultural production areas. They primarily utilize agricultural tricycles, small trucks, motorcycles, and local private cars. Traffic is dispersed, with short-term peaks occurring on market days (e.g., one or two fixed days per week) or during busy farming seasons (e.g., sowing and harvesting). These stations are smaller, and fuel prices are generally lower than those in urban areas, prioritizing value for money. Their non-fuel business focuses on the retail of agricultural supplies (e.g., fertilizers and pesticides), agricultural machinery parts, and daily necessities. Cash and credit services are supported to accommodate rural customer needs.

[0060] New Energy Dedicated (Integrated Trend): Deployed in areas with high NEV penetration (such as first-tier cities and electric vehicle industrial parks). These primarily feature electric and hybrid vehicles, with conventional fuel vehicles accounting for less than 20%. Charging demand is fragmented (e.g., recharging during commutes), with peak hours overlapping with grid off-peak periods (e.g., at night). Charging stations (primarily DC fast charging) replace traditional gas pumps, with supporting energy storage equipment to balance grid loads. Charging waiting services (such as coffee bars, office areas, and pet storage) are provided. New energy facilities such as photovoltaic and battery swap stations can be integrated to form comprehensive energy service stations.

[0061] This refined classification helps gas stations accurately locate sites based on traffic flow scheduling needs. For example, if a site is adjacent to both a highway exit and an industrial park, it can be classified as a "highway-logistics complex" to simultaneously meet the refueling and rest needs of both transit long-distance vehicles and local trucks, thus avoiding resource mismatch.

[0062] Step S104: Using the crowd distribution model and the type of station to which the gas station belongs, a traffic flow scheduling simulation is performed on the gas station, and based on the station operation status in each time period obtained by the simulation, the traffic flow scheduling strategy of the gas station is dynamically adjusted to obtain multiple traffic flow scheduling strategies that match the gas station in each time period.

[0063] Among them, the various traffic flow scheduling strategies include at least: in-station personnel scheduling strategy, in-station vehicle flow line planning, in-station service operation strategy and in-station safety management strategy.

[0064] The traffic flow scheduling strategy is a systematic solution based on the actual operational status of gas stations, comprehensively planning and dynamically adjusting the management of people, vehicles, services, and safety within the station. Based on data from a passenger flow distribution model and the characteristics of gas station types (such as urban commuter and long-distance refueling), it simulates gas station operations at different times in a virtual environment, generating optimized strategies tailored to each scenario.

[0065] In terms of staffing strategies, we flexibly deploy staff based on the business needs of gas stations at different times. For example, during the morning rush hour at urban commuter stations, if we predict a surge in refueling vehicles, we'll temporarily relocate convenience store employees to the refueling lanes to help guide vehicles and expedite the refueling process. At night, at long-distance refueling stations, we consider that truck drivers may need more rest and shopping services, so we appropriately increase the number of convenience store guides and service staff. By rationally arranging staff positions and working hours, we can ensure service quality while avoiding idle or overworked personnel, thereby improving employee efficiency and customer satisfaction.

[0066] In-station traffic flow planning aims to optimize vehicle routes within gas stations, reducing congestion and wait times. For urban commuter gas stations, where vehicles spend less time and seek faster transit, one-way streets and clear signage are designed to prevent vehicles from crossing each other. At long-distance refueling stations, given the large size of trucks and the difficulty they pose in entering and exiting, dedicated refueling and parking areas are designated, with separate routes planned to ensure smooth access for large vehicles. Reasonable traffic flow planning can maintain orderly traffic flow within the station, reduce the likelihood of accidents, and improve overall operational efficiency.

[0067] The station's in-station service operations strategy focuses on improving the service quality and commercial value of gas stations. In terms of service content, urban commuter stations will cater to the time-sensitive nature of customers during the morning rush hour by offering convenient items such as quick breakfasts and ready-to-drink coffee. Long-distance refueling stations will also offer services necessary for long-distance driving, such as vehicle maintenance and tire changes. Operationally, customer wait times will be shortened through optimized checkout processes and the addition of mobile payment terminals. Furthermore, by predicting consumer demand based on traffic flow models, merchandise displays and promotional activities will be adjusted. For example, discounts will be offered during the lunch break to attract customers and increase non-fuel business revenue.

[0068] The station's safety management strategy prioritizes ensuring the safe operation of gas stations. By simulating pedestrian and vehicle traffic in various scenarios, potential safety hazards are identified and countermeasures are formulated in advance. For example, when simulations predict excessive traffic flow during a certain period, potentially causing congestion, station safety inspections are strengthened and dedicated personnel are assigned to guide vehicles to prevent collisions. Given the frequent traffic and hazardous chemical transport vehicles entering and exiting long-distance refueling stations, a strict hazardous chemical vehicle inspection process and emergency response plan have been established to ensure that gas stations can respond quickly and effectively to sudden safety incidents, protecting the lives and property of personnel.

[0069] These interrelated and synergistic traffic flow scheduling strategies contribute to an efficient, safe, and orderly gas station operation system. By precisely matching demand across time periods and station types, they significantly improve gas station operational efficiency, reduce customer wait times, and enhance gas station management efficiency. At the same time, they effectively mitigate safety risks, providing strong support for the sustainable development and competitiveness of gas stations.

[0070] In an optional embodiment of step S104, the crowd distribution model is imported into a three-dimensional virtual simulation environment, and an initial traffic flow scheduling strategy is set in combination with the type of gas station. In the three-dimensional virtual simulation environment, changes in traffic volume and fluctuations in consumer demand in different time periods are simulated, and the simulation monitoring system collects real-time data on tidal lane switching, idle personnel, personnel scheduling, queue status, complaint rate, and average customer travel time. Based on the simulation results collected in real time, a reinforcement learning algorithm is used to iteratively optimize the initial traffic flow scheduling strategy, adjust the number of personnel configurations within the station, the traffic flow guidance plan, the number of open service windows, the tidal lane division method, the lane switching method, and the safety inspection route, and generate a variety of traffic flow scheduling strategies suitable for different time periods.

[0071] The core principle of the above steps is to combine the crowd distribution model with the actual operation of the gas station, use a virtual simulation environment to simulate the changes in traffic volume and consumer demand in different time periods, and collect relevant data in real time. Then, use the reinforcement learning algorithm to iteratively optimize the initial traffic flow scheduling strategy to generate an efficient and reasonable traffic flow scheduling strategy that is more suitable for different time periods.

[0072] Specifically, a crowd flow distribution model is first imported into a 3D virtual simulation environment. This model, built based on historical operational data and real-time monitoring data from the gas station, reflects the flow of people and vehicles within the station at different times. An initial traffic flow scheduling strategy is then developed based on the type of gas station. For example, a city commuter gas station might have more attendants and more refueling lanes open during the morning rush hour to accommodate peak traffic. Long-distance refueling stations might have dedicated truck refueling and rest areas, with corresponding staffing.

[0073] In a 3D virtual simulation environment, traffic flow changes and consumer demand fluctuations at different times are simulated. A real-time monitoring system collects key indicators such as tidal lane switching, idle personnel, personnel scheduling, queue status, complaint rates, and average customer travel time. This data serves as input for a reinforcement learning algorithm to evaluate the effectiveness of current strategies.

[0074] Based on real-time simulation results, a reinforcement learning algorithm is used to iteratively optimize the initial traffic flow scheduling strategy. Based on the effectiveness of the current strategy, the reinforcement learning algorithm automatically adjusts parameters such as station staffing, traffic flow guidance, number of service windows, tidal lane divisions, lane switching methods, and safety inspection routes to find the optimal scheduling strategy. For example, if queues in refueling lanes are too long during a certain period, the algorithm may recommend adding more staff or adjusting the traffic flow guidance plan to reduce queue times.

[0075] Through continuous iterative optimization, we ultimately generate a variety of traffic flow scheduling strategies tailored to different time periods. These strategies can be dynamically adjusted based on the actual operating conditions of gas stations, improving operational efficiency and service quality, reducing customer wait times and complaint rates, while ensuring safe operation of gas stations.

[0076] After optimizing through the above steps, the previously long average waiting times were significantly improved. By adjusting staffing, increasing the number of gas station attendants, and optimizing traffic flow guidance, vehicles were able to enter and exit the gas station in a more orderly manner, significantly shortening average customer travel time and significantly reducing the complaint rate. During off-peak hours, the number of service windows open was appropriately reduced based on changes in consumer demand, and safety inspection routes were adjusted, improving resource utilization efficiency while ensuring safety. This optimization strategy allows for flexible adjustments to gas station operations based on the characteristics of different time periods, improving overall operational effectiveness.

[0077] Alternatively, after collecting real-time data on tidal lane switching, idle personnel, personnel scheduling, queuing status, complaint rates, and average customer travel time through a simulation monitoring system in step S104, a non-dominated sorting genetic algorithm (NSGA-II) can be used, with customer waiting time, employee workload, and non-refueling consumption conversion rate as optimization targets. Decision variables are encoded as chromosomes, and a Pareto strategy solution set is generated based on the real-time collected simulation results. Multiple non-dominated solution sets are screened according to the Pareto rank, with each non-dominated solution corresponding to a set of initial traffic flow scheduling strategies. A Bayesian optimization dynamic algorithm is employed, using a Gaussian process to model the mapping relationship between strategy parameters and the objective function in the initial traffic flow scheduling strategy. Using the expected improvement (EI) function or the probability improvement (PI) function, the Pareto strategy solution set is searched for the optimal strategy solution for the current simulation conditions. The Pareto strategy solutions are sorted according to the collected function values, generating a recommended strategy list containing the confidence level of each Pareto strategy solution. Based on this recommended strategy list, the optimal traffic flow scheduling strategy for different time periods is selected.

[0078] In this application, the NSGA-II algorithm, along with Bayesian optimization, was integrated to achieve multi-objective collaborative optimization in a gas station traffic flow scheduling simulation. As a multi-objective evolutionary algorithm, NSGA-II uses customer wait time, employee workload, and non-gas consumption conversion rate as optimization objectives. Decision variables such as staffing levels and traffic flow guidance schemes are encoded as chromosomes. By simulating real-time data collected by a monitoring system (such as tidal lane switching and queue status), NSGA-II performs population iterations and selects mutually non-dominant strategy solutions based on the Pareto rank. This ensures that each solution strikes a balance between different objectives, avoiding the compromises that can result from single-objective optimization.

[0079] It can be explained that a Pareto strategy solution refers to a combination of strategies in a multi-objective optimization problem that cannot improve the performance of other objectives without degrading the performance of at least one objective. In simple terms, these strategy solutions each have their own advantages and disadvantages and are mutually non-dominant. For example, strategy A can reduce customer wait times but increase employee workload, while strategy B can increase customer conversion rates but slightly increase wait times. There's no absolute superiority or inferiority between these strategies; rather, they form a balance between different objectives.

[0080] In practical applications, Pareto strategy solutions are generated by multi-objective optimization algorithms such as NSGA-II, forming a set of multiple possible strategy solutions for further screening and optimization by subsequent algorithms (such as Bayesian optimization). This helps decision makers select the strategy that best suits the current scenario based on real-time needs (such as prioritizing efficiency during peak hours and taking into account consumption conversion during off-peak hours).

[0081] Based on the above concepts, the Pareto rank method is used to select mutually non-dominated strategy solutions. The key is to hierarchically classify all strategy solutions using a multi-objective optimization algorithm, eliminating "dominated" solutions and retaining "non-dominated" solutions. The specific process is as follows: First, all strategies are placed into the same set. The performance of any two strategy solutions on multiple optimization objectives (such as customer wait time, employee load, and purchase conversion rate) is then compared one by one. If a strategy solution is inferior to another strategy solution on all objectives, it is called a "dominated solution" (for example, strategy A has a longer wait time, higher employee load, and lower purchase conversion rate, and is completely dominated by strategy B) and is eliminated from the solution set. If two strategy solutions each have their own advantages and disadvantages (for example, strategy C has a shorter wait time but higher employee load, while strategy D has a slightly longer wait time but lower load), they are mutually "non-dominated" and are retained in the solution set. Through this pairwise comparison, all strategy solutions are classified into different Pareto ranks: Rank 1 represents the set of completely non-dominated optimal solutions (no other solution dominates them), Rank 2 represents the set of solutions dominated only by Rank 1 solutions, and so on. The final set of mutually non-dominant strategy solutions is the level 1 Pareto optimal solution set, where each solution represents an effective balance between multiple objectives. There is no absolutely better solution, providing a variety of candidate strategies for subsequent optimization.

[0082] Building on NSGA-II, the Bayesian Optimization algorithm utilizes Gaussian processes to construct a probabilistic mapping model between policy parameters and objective functions, describing the impact of policy parameter changes on objectives such as customer wait time. Using the expected improvement (EI) function or the probability improvement (PI) function, the algorithm efficiently searches for the Pareto policy solution set generated by NSGA-II that is most likely to optimize the objective under the current simulation conditions, while also accounting for policy uncertainty to avoid falling into local optima. The algorithm ranks Pareto policy solutions based on the collected function values ​​and generates a list of recommended policies with associated confidence levels, providing decision makers with a quantitative guide to the pros and cons of these policies.

[0083] Taking the morning rush hour at a commuter gas station in a certain city as an example, NSGA-II first generated multiple policy solutions: some prioritized reducing customer wait times but increased staff load, while others balanced customer conversion at the expense of slightly longer wait times. Bayesian optimization, combined with real-time simulated data such as traffic surges and excessive queue lengths, quickly identified a Pareto solution combination: adding two mobile payment guides, opening tidal lanes, and offering a full-amount gas promotion at the convenience store. This strategy reduced wait times, increased customer conversion rates, and kept staff load within an acceptable range. The algorithm placed this strategy at the top of the recommended list and assigned a high confidence rating, prompting managers to quickly adopt it.

[0084] For example, at 7:00 a.m., a commuter gas station in a city entered its morning rush hour. As traffic density on the main road surged, the station's cameras captured an influx of 15 new vehicles per minute, far exceeding the average for the morning rush hour. At this point, the system initiated a strategy optimization process. First, the NSGA-II algorithm, based on historical data and real-time monitoring of traffic flow and consumer demand, rapidly generated eight differentiated strategy solutions. Strategy A minimized customer wait times by ensuring full staffing, but this required employees to work intensely for two hours straight. Strategy B, employing a "three gas station attendants + one shopping guide" configuration, attempted to boost customer conversions through convenience store promotions, but this extended the average wait time to seven minutes.

[0085] Faced with an emergency situation where the queue length exceeded 10 vehicles, the Bayesian optimization algorithm immediately stepped in. Based on Gaussian processes, it constructed a probabilistic model from historical policy data, simulating the impact of different policy parameter combinations on the objective function. When it detected that mobile payment usage was only 30% and 80% of the vehicles in the queue were private cars, the algorithm quickly identified key variables within the Pareto solution set: adding mobile payment guides to reduce manual checkout pressure, opening tidal lanes to increase vehicle throughput, and combining them with convenience store promotions like "spend 50 and get 10 off" to attract customers without incurring additional labor costs.

[0086] After 200 rapid iterations, the system output the optimal strategy: immediately dispatch two additional mobile payment guides to the entrance, temporarily convert one exit lane to a tidal lane, and activate the convenience store's electronic screens to display promotional information. This strategy, validated in simulations, reduced average wait times from 7 minutes to 4 minutes, maintained employee workload at a reasonable 75%, and increased non-fuel consumption conversion rates by an estimated 18%. The system assigned this solution a "92% confidence rating" top recommendation and simultaneously pushed execution instructions via in-station announcements and employee smart devices.

[0087] At 7:15 AM, with the implementation of the new strategy, the previously chaotic traffic flow was diverted in an orderly manner under the guidance of guides. The tidal lanes, after being activated, served an additional 12 vehicles per hour. At the convenience store entrance, drivers, attracted by promotional information, completed their purchases while waiting to refuel. Real-time data showed that non-fuel sales increased by 22% year-on-year compared to the morning peak. By 8:30 AM, when the peak subsided, the station's customer complaint rate had dropped by 60% compared to the previous day. Staff maintained efficient service through a rotating shift system, validating the effectiveness of the algorithmic recommendation strategy in balancing multiple objectives.

[0088] Through the above steps, gas stations can achieve a dynamic balance of multiple objectives. Compared with traditional single optimization methods, it can not only reduce the average waiting time of customers by more than 30%, but also increase the non-gas consumption conversion rate by 15%. At the same time, it can control the standard deviation of employee load within a reasonable range, avoid excessive fatigue or idle resources, and significantly improve the overall operational efficiency and service quality of gas stations.

[0089] Step S105: Construct a dynamic and interactive traffic flow optimization model based on multiple traffic flow scheduling strategies and corresponding station operation status, and display the traffic flow optimization model to the user, so that the user can select the traffic flow scheduling plan to be executed based on the traffic flow optimization model, thereby realizing intelligent management of the target gas station.

[0090] In the embodiment of the present application, a variety of traffic flow scheduling strategies and site operation status are constructed into a dynamic and interactive traffic flow optimization model, creating an immersive management experience for users.

[0091] It's understandable that users can freely manipulate the three-dimensional virtual scene and intuitively experience the impact of different strategies on gas station operations. For example, by dragging employee models to different locations, one can see changes in customer density and fluctuations in related indicators in real time, making it feel like they're adjusting strategies in a real gas station operating environment. Furthermore, the traffic flow optimization model supports parallel comparison of multiple strategies, visually presenting the differences between each strategy in key indicators such as customer wait time, employee workload, and non-gas consumption conversion rate, helping users quickly weigh the pros and cons and make the best decision.

[0092] The traffic flow optimization model also provides powerful data analysis tools. Users can use simple operations, such as clicking on a specific area on the spatiotemporal heat map, to delve deeper into detailed data for that area over different time periods and uncover underlying patterns. By entering a specific indicator, the model automatically analyzes the strategy parameters that influence that indicator and provides targeted adjustment recommendations.

[0093] Furthermore, the traffic flow optimization model supports natural language interaction, allowing users to query and execute strategies through voice commands, significantly lowering the barrier to entry and improving the convenience and fluidity of interaction. This high level of interactivity allows users to move beyond passive information receivers and instead deeply participate in gas station operational management decisions, truly achieving human-machine collaboration for intelligent management.

[0094] In an optional embodiment of step S105, multiple traffic flow scheduling strategies and corresponding station operation statuses are constructed into a dynamic and interactive traffic flow optimization model, including:

[0095] Various traffic flow scheduling strategies and their corresponding station operational statuses are mapped to the corresponding data mapping space of the gas station's internal functional areas, according to the matching traffic flow scheduling strategies. This generates a data point cloud corresponding to each functional area. Using density peak clustering, the data point cloud is divided into multiple data clusters. High-density policy state clusters for different time periods are selected, and within these high-density policy state clusters, sub-clusters are generated based on the similarity of policy parameters. The high-density policy state clusters include at least a cluster with optimal morning peak efficiency and a cluster with enhanced nighttime safety inspections. Sub-clusters of the optimal morning peak efficiency cluster include clusters corresponding to different tidal lane scheduling methods and different combinations of gas station attendants and cashiers. Outlier policy clusters are identified using isolation forests, automatically triggering the corresponding risk warning process. Outlier policy clusters include at least policy configurations that lead to queue overflows, policy configurations that lead to excessive complaint rates, and policy configurations that lead to staff overload. Each data cluster is assigned a policy maturity label based on historical execution counts and effectiveness scores. For the optimized data point cloud, dynamic operational effect diagrams and corresponding explanatory information are generated for each data cluster at different time periods to construct a corresponding traffic flow optimization model. The explanatory information at least includes the causal relationship between the dynamic operational effect diagrams and the traffic flow scheduling strategy. The dynamic operational effect diagrams in the traffic flow optimization model are used to reflect at least the following: the gas station's staff scheduling status at various time periods, predicted traffic flow, predicted consumption, predicted queue status, and predicted work status of station personnel.

[0096] Specifically, through data mapping, intelligent clustering, and visualization techniques, the abstract traffic flow scheduling strategies and operational status are transformed into intuitive and interactive decision-making tools. First, traffic flow scheduling strategies (such as staffing and lane planning) and corresponding operational data (traffic volume, queue lengths, etc.) for different time periods are mapped into a three-dimensional data space according to the functional areas within the gas station (fueling area, convenience store, cashier), forming a data point cloud containing spatiotemporal information. Each data point represents the operational status of a specific strategy in a specific area and time period.

[0097] Next, the data point cloud is processed using a density peak clustering algorithm. This algorithm can automatically identify densely distributed areas of data and divide the data point cloud into multiple data clusters, each of which represents a combination of strategies and operating status with similar characteristics. For example, by analyzing data during the morning rush hour, the algorithm can filter out the "morning rush hour efficiency cluster." The strategies in this cluster are generally effective in reducing customer waiting time; while in nighttime data, a "nighttime safety inspection enhancement cluster" will be formed, which includes strategies focused on safety management. To further refine the strategy types, within the high-density clusters, the system performs a secondary division based on the similarity of strategy parameters (such as tidal lane switching time and personnel ratio) to generate sub-strategy clusters, allowing managers to more accurately compare the details of different strategies.

[0098] To ensure strategy feasibility, the system uses an isolation forest algorithm to detect outlier strategy clusters. These strategies often lead to queue overflows, soaring complaint rates, or employee overload due to improper parameter settings. Once identified, the system immediately triggers a risk warning, prompting managers to avoid adopting them. Furthermore, based on historical execution counts and effectiveness scores, each data cluster is assigned a strategy maturity label, such as "Golden Strategy Cluster" (over 100 executions and a satisfaction rating exceeding 90%), to help managers quickly identify reliable strategies.

[0099] Finally, the optimized data point cloud is transformed into a dynamic operational rendering, visualizing the gas station's staffing, traffic flow, and consumption trends at each time period through 3D animations and heat maps. Explanatory information is automatically generated, clearly demonstrating the causal relationship between strategy and operational effectiveness. For example, the model allows managers to visually see how a strategy like "adding one attendant and switching to a tidal lane 30 minutes earlier" can shorten morning rush hour queues and balance employee workloads.

[0100] Taking a gas station in a certain city as an example, a traffic flow optimization model constructed using this method helped managers quickly identify an efficient strategy for the morning rush hour: a 5:2 ratio of gas station attendants to cashiers, alternating tidal lane switching, and limited-time convenience store promotions. This approach ensured both fuel efficiency and increased non-fuel consumption. The model also promptly warned of a strategy that could lead to a surge in complaints due to excessive staffing constraints, thus mitigating operational risks. Ultimately, this model reduced wait times during the morning rush hour at the gas station by 35%, increased non-fuel consumption by 20%, and reduced the standard deviation of employee workload by 40%, significantly improving the accuracy and efficiency of operational management.

[0101] Further optionally, in the above steps, for the optimized data point cloud, a dynamic operation effect diagram and corresponding explanatory information are generated for each data cluster in different time periods. After the corresponding traffic flow optimization model is constructed, the adaptability between the site type to which the gas station belongs and the various traffic flow scheduling strategies in each time period can be evaluated based on the site operation status to obtain a adaptability report; the adaptability report includes at least: the adaptability between the site type and the dynamic scheduling strategy, traffic flow scheduling bottlenecks, and adjustment suggestions. In this case, the traffic flow scheduling strategies with a adaptability lower than the set threshold in the traffic flow optimization model and the corresponding dynamic operation effect diagram can be hidden. In this case, the display effect of the traffic flow optimization model on the traffic flow scheduling strategy in each time period can also be dynamically adjusted according to the adaptability, so that users pay priority attention to the traffic flow scheduling strategies with higher adaptability.

[0102] The above steps evaluate the adaptability of strategies, dynamically screen and optimize the traffic flow optimization model, and ensure that managers can quickly access high-value strategies. This is based on real-time operational status data from gas stations, matching and analyzing station type characteristics at different time periods with traffic flow scheduling strategies. An algorithm is used to calculate the adaptability of each strategy for the corresponding station type and time period. For example, factors such as traffic flow, consumer demand, and personnel load are comprehensively considered to determine whether the strategy aligns with the station's operational characteristics. If a strategy significantly increases operating costs while reducing customer wait times and does not meet the station's cost control objectives, its adaptability will be reduced.

[0103] The calculated fitness data is used to generate a fitness report, which not only includes the fitness values ​​of the strategy and site type, but also locates traffic scheduling bottlenecks by comparing actual operating results with expected goals. For example, it may be found that unreasonable tidal lane switching during a certain period of time leads to vehicle congestion, or that staffing cannot meet service needs. Specific adjustment suggestions are then made to address these bottlenecks, such as optimizing lane switching rules and reallocating personnel positions.

[0104] Based on the compatibility report, the system processes the traffic flow optimization model. Strategies with compatibility below the set threshold, along with their dynamic operational renderings, are directly hidden to prevent managers from being distracted by low-value information. Simultaneously, the model's display is dynamically adjusted based on compatibility. Highly compatible strategies are highlighted, their display area is enlarged, and animation duration is increased to enhance their prominence within the interface. Strategies with lower compatibility but still valuable for reference are presented with faded display and a smaller display area.

[0105] Taking the morning rush hour at a commuter gas station in a certain city as an example, the system-generated adaptability report showed that while a strategy of "reducing the number of gas station attendants to reduce costs" could reduce costs, it significantly increased customer wait times, seriously inconsistent with the station's focus on fast service during the morning rush hour. The adaptability was only 30%, far below the set threshold of 60%, so this strategy and its corresponding dynamic operational rendering were hidden. However, another strategy, "increasing mobile payment guides and opening tidal lanes," had an adaptability of 85%. In the model display, the dynamic operational rendering of this strategy was enlarged to full screen, with vehicle routes and personnel scheduling highlighted in bold colors. The interface also prominently displayed adjustment suggestions and expected results for the strategy, allowing managers to quickly identify and adopt it.

[0106] By following these steps, managers can quickly focus on strategies that are highly adaptable and better meet actual operational needs when reviewing traffic flow optimization models, reducing information screening time and improving decision-making efficiency. Furthermore, the clear bottleneck analysis and adjustment suggestions in the adaptability report provide managers with a clear direction for optimizing gas station operations, helping to improve overall operational efficiency and service quality, and avoiding problems such as low operational efficiency and customer churn caused by adopting inappropriate strategies.

[0107] In another optional embodiment of step S105, a plurality of traffic flow scheduling strategies and corresponding station operation statuses are constructed into a dynamic and interactive traffic flow optimization model, which is displayed to the user, allowing the user to select a traffic flow scheduling solution to be executed based on the traffic flow optimization model, including:

[0108] Based on the characteristics of historical time periods, the matching traffic flow scheduling strategy is loaded into the gas station management system in advance, and the pre-alert mechanism is triggered within the set time before the start of the corresponding time period; the strategy overview and risk warnings for different time periods are pushed to the station management personnel, and the management personnel confirm whether to execute the traffic flow scheduling plan; if the traffic flow scheduling plan is determined to be executed, the station terminal prompts, mobile terminal push, and voice broadcast are used to send prompts for pending operations that match the traffic flow scheduling plan to be executed to the station execution personnel in different time periods; if the station execution personnel fail to execute as planned, an abnormal alarm is triggered and backup strategy suggestions are automatically pushed.

[0109] The above steps achieve a closed-loop process from decision-making to implementation of traffic flow scheduling strategies through automated strategy loading, pre-alert mechanisms, and multi-channel execution management. Its core principle is to pre-store proven efficient strategies in the gas station management system based on historical time period characteristics (such as peak traffic flow times during the morning rush hour and lunchtime consumption habits), and to proactively push and monitor the execution of strategies in combination with time-triggered rules. The system analyzes the typical operating status of each time period in historical data (such as traffic density during the morning rush hour from 7:00 to 9:00 and employee load thresholds), matches the corresponding traffic flow scheduling strategies in advance (such as adding gas station attendants and opening tidal lanes), and triggers a pre-alert 30 minutes before the start of the time period to ensure that managers have sufficient time to prepare.

[0110] At the execution level, the system pushes strategy overviews (such as projected traffic volume and staffing adjustments) and risk warnings (such as the potential risk of queue overflow) to station managers, who then confirm and initiate the execution process. Specific operational prompts are issued to executive personnel through multiple channels, including on-site terminals (such as gas station displays and cash register computers), mobile apps (such as employee-specific scheduling software), and voice broadcasts. For example, a voice broadcast reminds gas station attendants to arrive at their posts five minutes before the morning rush hour, while mobile devices push tidal lane switching time points and guidance gesture specifications. If the executive fails to complete the operation within the specified time (such as failing to switch to the tidal lane on time), the system automatically detects the anomaly and triggers an alarm, while also pushing a backup strategy (such as temporarily enabling manual guidance and diversion) to avoid operational disruptions caused by human delays.

[0111] Taking a commuter gas station in a certain city as an example, the system predicted based on historical data that a traffic surge would occur during Tuesday's morning rush hour at 7:30 AM. Therefore, the "All-Staff Gas Station + Mobile Payment Specialist Pre-positioning" strategy was pre-loaded into the management system at 7:00 AM, and a reminder containing the strategy details and congestion risk warnings was pushed to the station manager. After the station manager confirmed the implementation, the system notified all employees via voice broadcast at 7:25 AM to enter peak mode. The lane numbers and guidance procedures for each employee were also displayed on the gas station attendants' smart ID cards. At 7:35 AM, the system detected that a gas station attendant had failed to guide vehicles to the newly added tidal lane as planned. The system immediately sent an alert to the attendant's mobile device and simultaneously recommended that nearby employees provide temporary support. Thanks to the timely response, the station's average wait time during the morning rush hour was kept below 4 minutes, a 30% reduction compared to the same period last week, and there were no customer complaints due to inadequate policy implementation.

[0112] Through full-chain management, the efficiency of policy implementation is significantly improved. The pre-loading mechanism for historical data matching strategies reduces decision-making time from 20 minutes for manual decision-making to seconds for automatic triggering. Multi-channel prompts ensure over 99% accuracy in the transmission of execution instructions. The combination of abnormal alerts and backup strategies reduces the impact of human operational errors by 85%. In actual application, after a chain of gas stations deployed this mechanism, the compliance rate of policy execution at each site increased from 68% to 92%, the fluctuation range of time-based operational indicators (such as morning peak service efficiency and lunchtime consumption conversion rate) decreased by 40%, and management costs decreased by 15%, truly realizing the empowerment of operational efficiency through intelligent management.

[0113] In the embodiments of this application, the crowd distribution model constructed based on the neural radiation field can accurately capture subtle changes in customer traffic patterns at different times of the day at gas stations. For example, it can identify the fast-paced demand of customers at urban commuter stations during the morning rush hour, such as the quick-paced demand of customers who fill up and leave, and the complex traffic patterns of truck drivers at long-distance supply stations who refuel, rest, and shop at night. This precise modeling provides a solid data foundation for subsequent traffic flow scheduling simulations, allowing station personnel scheduling strategies and traffic flow planning to be highly tailored to actual operational scenarios. For example, by predicting morning rush hour traffic volume based on the crowd distribution model, pre-positioning gas station attendants to lane entrances, and rationally arranging mobile payment specialists, customers' average waiting time can be significantly shortened, queue lengths can be significantly reduced, and operational efficiency can be effectively improved. At the same time, in terms of safety management, combining sensor data with the neural radiation field model can monitor abnormal conditions within the gas station in real time, such as excessive oil and gas concentrations and illegal personnel operations. Once a risk is detected, an alarm can be quickly triggered and the corresponding emergency plan can be activated, nipping safety hazards in the bud, achieving a shift from passive response to active prevention, and comprehensively improving the safety of gas stations. In terms of interactivity, the embodiment of the present application constructs a variety of traffic flow scheduling strategies and site operation status into a dynamic and interactive traffic flow optimization model, creating an immersive management experience for users and effectively improving the management efficiency of gas stations.

[0114] See also Figure 2 , Figure 2A dynamic interactive visual operation management device is provided in an embodiment of the present application. The dynamic interactive visual operation management device includes: an acquisition module for acquiring multi-dimensional operation data of a gas station in each time period; a modeling module for dynamically modeling the customer movement lines of the gas station in each time period using a neural radiation field based on the multi-dimensional operation data to obtain a crowd distribution model of the gas station in each time period; a classification module for classifying the type of station to which the gas station belongs according to the geographical location of the gas station, the type of vehicle entering the station, and the surrounding traffic conditions; wherein the station type is urban commuting type or long-distance supply type; a simulation module for using the crowd distribution model and the gas station to which the station belongs Type, conducts a traffic flow scheduling simulation for the gas station, and based on the station operation status in each time period obtained by the simulation, dynamically adjusts the traffic flow scheduling strategy for the gas station, and obtains a variety of traffic flow scheduling strategies that match the gas station in each time period; wherein, the multiple traffic flow scheduling strategies at least include: on-site personnel scheduling strategy for the gas station, on-site vehicle flow line planning, on-site service operation strategy, and on-site safety management strategy; an interactive module, used to construct a variety of traffic flow scheduling strategies and corresponding station operation status into a dynamic and interactive traffic flow optimization model, display the traffic flow optimization model to the user, and enable the user to select the traffic flow scheduling plan to be executed based on the traffic flow optimization model, thereby realizing intelligent management of the target gas station. In some embodiments, the dynamic and interactive visual operation management device can be applied to a terminal device. It should be noted that, for the convenience and simplicity of description, the specific working process of the dynamic and interactive visual operation management device described above can refer to the corresponding process in the aforementioned dynamic and interactive visual operation management method embodiment, and will not be repeated here.

[0115] See also Figure 3 , Figure 3 A schematic block diagram of the structure of a terminal device provided in an embodiment of the present application.

[0116] like Figure 3 As shown, the terminal device 300 includes a processor 301 and a memory 302, and the processor 301 and the memory 302 are connected via a bus 303, such as I 2 C bus. Specifically, processor 301 is used to provide computing and control capabilities to support the operation of the entire terminal device. Processor 301 can be a central processing unit, or it can be other general-purpose processors, digital signal processors, application-specific integrated circuits, field programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or any conventional processor. Specifically, memory 302 can be a Flash chip, a read-only memory disk, an optical disk, a USB flash drive, or a mobile hard drive, etc.

[0117] Those skilled in the art will understand that Figure 3 The structure shown in is only a block diagram of a part of the structure related to the embodiment of the present application, and does not constitute a limitation on the terminal device to which the embodiment of the present application is applied. The specific server may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements. Among them, the processor is used to run a computer program stored in the memory, and implement any one of the dynamic and interactive visual operation management methods provided by the embodiment of the present application when executing the computer program. It should be noted that technical personnel in the relevant field can clearly understand that for the convenience and conciseness of description, the specific working process of the terminal device described above can refer to the aforementioned dynamic and interactive visual operation management method embodiment, and will not be repeated here.

[0118] An embodiment of the present application also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any dynamic and interactive visual operation management method provided in the description of the embodiment of the present application.

Claims

1. A dynamic and interactive visual operation management method, characterized in that: The method comprises: Obtain multi-dimensional operational data of gas stations at different time periods; Based on the multi-dimensional operation data, a neural radiation field is used to dynamically model the customer movement lines of the gas station in each time period to obtain a crowd distribution model of the gas station in each time period, including: constructing the multi-dimensional operation data into corresponding time series data according to the corresponding time period, and combining it with the spatial coordinates corresponding to the geographical location of the gas station to obtain an input vector; using a MobileNeRF variant model with an added time coordinate dimension, based on the input vector, predicting the radiation field properties of the gas station in each time period, and obtaining a three-dimensional spatial layout of the customer movement lines in the gas station in each time period; based on the three-dimensional spatial layout, obtaining the appearance probability, movement trajectory and aggregation density of customers in different functional areas of the gas station in each time period to form a corresponding crowd distribution model; using an online learning mechanism, if it is detected that the gas station layout is adjusted or new equipment is introduced, the newly added data is collected to perform local fine-tuning on the crowd distribution model to update the crowd distribution model of the gas station in each time period; Gas stations are classified into station types based on their geographical location, the type of vehicles entering the station, and the surrounding traffic conditions. Station types can be urban commuting or long-distance refueling. Using the passenger flow distribution model and the type of gas station, a traffic flow scheduling simulation is performed on the gas station. Based on the simulated station operation status in each time period, the traffic flow scheduling strategy for the gas station is dynamically adjusted to obtain multiple traffic flow scheduling strategies that match the gas station in each time period. The multiple traffic flow scheduling strategies include at least: a gas station personnel scheduling strategy, a gas station vehicle flow planning strategy, a gas station service operation strategy, and a gas station safety management strategy. A variety of traffic flow scheduling strategies and corresponding station operation status are constructed into a dynamic and interactive traffic flow optimization model, including: mapping the various traffic flow scheduling strategies and corresponding station operation status to the data mapping space corresponding to the gas station according to the internal functional areas of the gas station matched with the traffic flow scheduling strategies, and obtaining the data point cloud corresponding to each functional area; through density peak clustering, multiple data clusters are obtained in the data point cloud, and high-density strategy state clusters under different time periods are screened. Within the high-density strategy state cluster, the clusters are subdivided according to the similarity of strategy parameters to generate sub-strategy clusters; wherein, the high-density strategy state cluster includes at least: the optimal efficiency cluster during the morning peak period and the nighttime safety inspection reinforcement cluster; the sub-clusters of the optimal efficiency cluster during the morning peak period include: different tidal Lane scheduling method clusters, data clusters corresponding to gas station attendant and cashier combinations with different personnel ratios; use isolation forest to mark outlier strategy clusters and automatically trigger corresponding risk warning processes; where outlier strategy clusters include at least: strategy configurations that lead to queue overflow, strategy configurations that lead to complaint rate overflow, and strategy configurations that lead to personnel overload; based on the historical number of executions and effect scores, label each data cluster with a strategy maturity label; for the optimized data point cloud, generate dynamic operation effect diagrams obtained by the effects of each data cluster at different time periods and corresponding explanatory information, and construct the corresponding traffic flow optimization model; the explanatory information at least includes: the causal relationship between the dynamic operation effect diagram and the traffic flow scheduling strategy; The traffic flow optimization model is displayed to the user, so that the user can select the traffic flow scheduling plan to be executed based on the traffic flow optimization model, thereby realizing intelligent management of the target gas station.

2. The method according to claim 1, characterized in that The multi-dimensional operation data includes at least: the geographical location of the gas station, the type of vehicles entering the station at different time periods, the volume of vehicles entering the station, the queuing time, the surrounding traffic conditions, consumption conditions, the working status of the station staff, and the number of complaints.

3. The method according to claim 1, characterized in that The gas station is classified into the following types based on its geographical location, the type of vehicles entering the station, and the surrounding traffic conditions: Obtain the geographical location of the gas station, the type of vehicles entering the station, and the surrounding traffic conditions; If the gas station is located on a main urban road or in a commercial area, determine whether the proportion of private cars among the incoming vehicles exceeds a first preset ratio; the first preset ratio may be a fixed value or set based on the size of the city; if the proportion of private cars among the incoming vehicles exceeds the first preset ratio, determine whether the surrounding traffic change trend conforms to the characteristics of morning and evening rush hour commuting; if it conforms to the characteristics of morning and evening rush hour commuting, determine that the gas station belongs to the urban commuting type; If the gas station is located in a highway service area or along a national highway, determine whether the proportion of trucks and long-distance buses among the types of vehicles entering the station exceeds a second preset ratio; the second preset ratio is a fixed value or is set based on the size of the city; if the proportion of trucks and long-distance buses among the types of vehicles entering the station exceeds the second preset ratio, determine whether the traffic around the gas station is mainly long-distance transit traffic; if it is mainly long-distance transit traffic, determine that the gas station belongs to a long-distance supply type.

4. The method according to claim 1, wherein The traffic flow scheduling simulation of the gas station is performed using the crowd flow distribution model and the type of gas station. Based on the simulated station operation status in each time period, the traffic flow scheduling strategy of the gas station is dynamically adjusted to obtain multiple traffic flow scheduling strategies that match the gas station in each time period, including: Importing the crowd flow distribution model into a three-dimensional virtual simulation environment, and setting an initial traffic flow scheduling strategy based on the type of gas station; In a 3D virtual simulation environment, the changes in traffic flow and fluctuations in consumer demand at different times are simulated, and the simulation monitoring system collects real-time data on tidal lane switching, idle personnel, personnel scheduling, queue status, complaint rate, and average customer travel time; Based on the simulation results collected in real time, a reinforcement learning algorithm is used to iteratively optimize the initial traffic flow scheduling strategy. The system adjusts the number of station personnel, traffic flow guidance plan, number of open service windows, tidal lane division method, lane switching method, and safety inspection routes to generate a variety of traffic flow scheduling strategies suitable for different time periods.

5. The method according to claim 4, characterized in that After collecting the real-time information of tidal lane switching, idle personnel, personnel scheduling, queue status, complaint rate, and average customer travel time through the simulation monitoring system, it also includes: Using the non-dominated sorting genetic algorithm (NSGA-II), the optimization targets are customer waiting time, employee workload, and non-refueling consumption conversion rate. Decision variables are encoded as chromosomes, and Pareto strategy solution sets are generated based on real-time simulation results. Multiple non-dominated solution sets are selected based on the Pareto rank, and each non-dominated solution corresponds to a set of initial route scheduling strategies. Adopting the Bayesian optimization dynamic algorithm, the Gaussian process is used to model the mapping relationship between the strategy parameters and the objective function in the initial route scheduling strategy. The expected improvement EI function or the probability improvement PI function is used to search for the optimal strategy solution for the current simulation conditions in the Pareto strategy solution set. Sort the Pareto strategy solutions according to the acquisition function values ​​and generate a recommended strategy list containing the confidence level of each Pareto strategy solution; Based on the recommended strategy list, the optimal route scheduling strategy suitable for different time periods is selected.

6. The method according to claim 1, characterized in that After generating the dynamic operation effect diagrams and corresponding explanatory information for the optimized data point cloud at different time periods and the corresponding movement line optimization model, the process further includes: Based on the station operation status, the compatibility between the station type and various traffic flow scheduling strategies in each time period is evaluated to generate a compatibility report; the compatibility report includes at least: the compatibility between the station type and the dynamic scheduling strategy, traffic flow scheduling bottlenecks, and adjustment suggestions; Hide the traffic flow scheduling strategies with fitness levels lower than a set threshold in the traffic flow optimization model, and the corresponding dynamic operation effect diagrams; and / or, The display effect of the movement line optimization model on the movement line scheduling strategy in each time period is dynamically adjusted according to the adaptability, so that users pay priority attention to the movement line scheduling strategy with higher adaptability.

7. The method according to claim 1, characterized in that The multiple traffic flow scheduling strategies and corresponding station operation statuses are constructed into a dynamic and interactive traffic flow optimization model, which is displayed to the user, allowing the user to select the traffic flow scheduling solution to be executed based on the traffic flow optimization model, including: Based on the characteristics of historical time periods, the matching traffic flow scheduling strategy is loaded into the gas station management system in advance, and the pre-alert mechanism is triggered within a set time before the start of the corresponding time period; Push strategy overviews and risk warnings for different time periods to station managers, who then confirm whether to implement the route scheduling plan; Once the route scheduling plan is confirmed, prompts for pending operations matching the route scheduling plan will be sent to the station execution personnel at different times using in-station terminal prompts, mobile push notifications, and voice broadcasts. If the site's executors fail to execute according to plan, an abnormal alarm will be triggered and backup strategy recommendations will be automatically pushed.

8. A dynamic and interactive visual operation management device, characterized in that: The device comprises the following modules: The acquisition module is used to obtain multi-dimensional operational data of gas stations at different time periods; A modeling module, configured to dynamically model the customer flow of the gas station at various time periods using a neural radiation field based on the multi-dimensional operational data, so as to obtain a customer flow distribution model of the gas station at various time periods; Among them, the modeling module is specifically used to: construct the multi-dimensional operation data into corresponding time series data according to the corresponding time period, and combine it with the spatial coordinates corresponding to the geographical location of the gas station to obtain an input vector; use the MobileNeRF variant model with the time coordinate dimension added to predict the radiation field properties of the gas station in each time period based on the input vector, and obtain the three-dimensional spatial layout of the customer movement lines in the gas station in each time period; based on the three-dimensional spatial layout, obtain the appearance probability, movement trajectory and aggregation density of customers in different functional areas of the gas station in each time period to form a corresponding crowd distribution model; use an online learning mechanism, if it is detected that the gas station layout is adjusted or new equipment is introduced, collect new data to perform local fine-tuning on the crowd distribution model to update the crowd distribution model of the gas station in each time period; The classification module is used to classify gas stations into station types based on their geographical location, the type of vehicles entering the station, and the surrounding traffic conditions. The station type can be urban commuting or long-distance supply. a simulation module for performing a traffic flow scheduling simulation for the gas station using the passenger flow distribution model and the type of gas station to which the gas station belongs, and dynamically adjusting the traffic flow scheduling strategy for the gas station based on the simulated station operation status in each time period, thereby obtaining a plurality of traffic flow scheduling strategies that match the gas station in each time period; wherein the plurality of traffic flow scheduling strategies include at least: a gas station personnel scheduling strategy, a gas station vehicle flow planning strategy, a gas station service operation strategy, and a gas station safety management strategy; An interactive module is used to construct a dynamic and interactive traffic flow optimization model based on multiple traffic flow scheduling strategies and corresponding station operation status. The model is displayed to the user, allowing the user to select the traffic flow scheduling plan to be executed based on the model, thereby realizing intelligent management of the target gas station; Among them, when the interactive module constructs a variety of traffic flow scheduling strategies and corresponding site operation status into a dynamic and interactive traffic flow optimization model, it is specifically used to: map the various traffic flow scheduling strategies and corresponding site operation status, according to the internal functional areas of the gas station matched with the traffic flow scheduling strategies, to the data mapping space corresponding to the gas station, and obtain the data point cloud corresponding to each functional area; through density peak clustering, multiple data clusters are obtained in the data point cloud, and high-density strategy state clusters under different time periods are screened, and sub-strategy clusters are generated according to the similarity of strategy parameters within the high-density strategy state cluster; wherein the high-density strategy state cluster includes at least: the optimal efficiency cluster during the morning peak period, the nighttime safety inspection reinforcement cluster; sub-clusters of the optimal efficiency cluster during the morning peak period It includes: data clusters corresponding to different tidal lane scheduling method clusters, and gas station attendants and cashier combinations with different personnel ratios; uses isolation forest to mark outlier strategy clusters and automatically triggers the corresponding risk warning process; among them, outlier strategy clusters include at least: strategy configurations that lead to queue overflow, strategy configurations that lead to complaint rate overflow, and strategy configurations that lead to personnel overload; based on the historical number of executions and effect scores, the strategy maturity label of each data cluster is marked; for the optimized data point cloud, the dynamic operation effect diagram obtained by the action of each data cluster in different time periods and the corresponding explanatory information are generated to construct the corresponding traffic line optimization model; the explanatory information at least includes: the causal relationship between the dynamic operation effect diagram and the traffic line scheduling strategy.

9. A terminal device, characterized in that: The terminal device includes a processor and a memory; The memory is used to store computer programs; The processor is configured to execute the computer program and implement the dynamic interactive visual operation management method according to any one of claims 1 to 7 when executing the computer program.

Images