Gas station digital twin system and virtual gas station scene generation method

Abstract

The invention relates to a gas station digital twinning system and a virtual gas station scene generation method, the system comprises a mapping module and a processing module, the mapping module is used for mapping a pre-constructed scene, an equipment three-dimensional model and a vehicle three-dimensional model to a virtual gas station scene; the processing module is used for processing the equipment operation data to obtain equipment state information, processing the video image to obtain vehicle driving information, and mapping the vehicle driving information and the equipment state information into a virtual gas station scene; acquiring historical refueling data of a gas station, analyzing the historical refueling data, predicting the number of refueled vehicles, the refueling amount, the average refueling time and the fuel amount consumption prediction value in a future time period, and predicting the traffic flow; and performing simulation deduction based on the gas station three-dimensional scene information of the virtual gas station scene to obtain an optimal driving route of the vehicle. The prediction and simulation deduction of the operation condition of the gas station in the future period are realized, and the intelligent level and risk early warning capability of the operation of the gas station are greatly improved.

Description

Technical Field

[0001] The present invention relates to a gas station digital twin system and a virtual gas station scene generation method. Background Art

[0002] Gas stations are specialized business places for filling gasoline and diesel tanks for motor vehicles. Compared with other application scenarios, gas stations have very significant particularities, including frequent flow of personnel and vehicles, a large number of equipment types and complex and diverse data, and very high security level requirements. For gas station application scenarios, the existing technology provides a digital twin gas station, which calculates the vehicle refueling time and gas station congestion according to several important time nodes during the refueling process of the refueling vehicle, and can improve the vehicle refueling speed and gas station traffic efficiency to a certain extent. However, this solution lacks video analysis and trajectory tracking of vehicles, cannot predict and simulate future operation conditions, and lacks the ability to warn of risks. Therefore, it is necessary to study a more effective digital twin system for gas station application scenarios to address the shortcomings of existing management methods and improve the management efficiency and intelligence level of gas stations. Summary of the invention

[0003] In order to perform video analysis and trajectory tracking on gas station vehicles and realize the prediction and simulation of future gas station operation conditions, the present invention proposes a gas station digital twin system and a virtual gas station scene generation method. The technical solutions proposed by the present invention are as follows:

[0004] In a first aspect, the present invention provides a gas station digital twin system, comprising: a mapping module and a processing module, wherein the mapping module is connected to the processing module;

[0005] The mapping module is used to map the pre-built scene and equipment three-dimensional model and vehicle three-dimensional model to the virtual gas station scene;

[0006] The processing module is used to obtain the video image in the gas station and the equipment operation data of the gas station facilities and equipment, process the equipment operation data to obtain equipment status information, process the video image to obtain vehicle driving information, and map the vehicle driving information and the equipment status information to the virtual gas station scene;

[0007] Obtain historical refueling data of the gas station, analyze the historical refueling data, predict the number of refueling vehicles, refueling amount, average refueling time and fuel consumption forecast value in the future period, and predict the traffic flow;

[0008] The optimal driving route of the vehicle is obtained by performing simulation based on the three-dimensional scene information of the virtual gas station scene.

[0009] In one or some embodiments, the processing module includes a central control unit and a data processing unit, and the central control unit is connected to the data processing unit;

[0010] The central control unit is used to send the received video image and the equipment operation data to the data processing unit;

[0011] The data processing unit is used to process the device operation data to obtain device status information and process the video image to obtain vehicle driving information;

[0012] Acquire the historical refueling data, analyze the historical refueling data, predict the number of refueling vehicles, refueling amount, average refueling time and fuel consumption forecast value in the future period, and predict the traffic flow;

[0013] Furthermore, a simulation is performed based on the three-dimensional scene information of the virtual gas station scene to obtain the best driving route for the vehicle.

[0014] In one or some embodiments, the processing module further includes: a human-computer interaction unit connected to the central control unit;

[0015] The central control unit is further used to send the device status information and the vehicle driving information to the human-computer interaction unit;

[0016] The human-computer interaction unit is used to map the vehicle driving information and the device status information into the virtual gas station scene, and to display the virtual gas station scene rendered by a 3D engine in real time.

[0017] In one or some embodiments, it further includes: an image acquisition unit connected to the central control unit;

[0018] The image acquisition unit is used to acquire video images in the gas station and send the video images to the central control unit.

[0019] In one or some embodiments, it further includes: a device control unit connected to the central control unit;

[0020] The equipment control unit is used to collect equipment operation data of the gas station facilities and equipment, control and manage the gas station facilities and equipment, and send the equipment operation data to the central control unit.

[0021] In one or some embodiments, it further includes: a data storage unit connected to the data processing unit;

[0022] The data storage unit is used to store the historical refueling data, the equipment status information and the vehicle driving information; the historical refueling data includes date, time period, whether it is a holiday, oil price changes, weather conditions and refueling amount.

[0023] In one or some embodiments, the data processing unit is specifically used for:

[0024] The historical refueling data within a preset time period of the gas station is obtained, and the historical refueling data is input into a pre-established traffic flow prediction model to predict the traffic flow; wherein the historical refueling data includes date, temperature, precipitation, snowfall, whether it is a discount, whether it is a holiday and traffic flow, and the traffic flow prediction model is obtained by the following method:

[0025] Dividing the historical refueling data into a training set and a test set;

[0026] A multivariate linear regression model is used to establish the relationship between traffic volume and date, temperature, precipitation, snowfall, discount, and holiday. The model is expressed as:

[0027] h b (x)=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6+b7x7+e;

[0028] Among them, h b (x) is the traffic volume, x1 is the date of each month, x2 is the temperature, x3 is the precipitation, x4 is the snowfall, x5 is the oil price, x6 is whether it is a discount, x7 is whether it is a holiday, b0, b1, b2, b3, b4, b5, b6, b7 are model parameters, and e is the error term;

[0029] The training set is input into the multivariate linear regression model to obtain the model prediction result. The loss function is determined based on the model prediction result and the actual traffic flow by the following formula:

[0030]

[0031] Among them, J(b) is the loss function, n is the number of data in the training set, and h b (x (i) ) is the model prediction result, y (i) is the actual traffic volume;

[0032] Based on the loss function and the following formula, the model parameters are updated using the gradient descent method to obtain the trained model:

[0033]

[0034] Among them, α is the learning rate, m is the number of data in the training set, is the jth factor affecting traffic flow in the i-th group of data, b j is the jth model parameter, j=1.2....7;

[0035] The trained model is evaluated based on the test set, and the training process and the evaluation process are repeated until preset conditions are met to obtain a traffic flow prediction model.

[0036] In one or some embodiments, the three-dimensional scene information of the gas station based on the virtual gas station scene is simulated and deduced using an A* algorithm to obtain the optimal driving route for the vehicle, specifically including:

[0037] Divide the gas station into nodes according to the three-dimensional scene information of the gas station and generate a node planning map;

[0038] Based on the oil type of each refueling parking space, generate the oil type vector of each refueling space node;

[0039] Generate an oil type vector of the vehicle to be refueled based on the type of oil required by the vehicle to be refueled;

[0040] For the vehicle to be refueled, calculate the inner product of its oil type vector and the oil type vector of each refueling position node, and add the nodes with a multiplication result of 1 to the refueling position list;

[0041] Traverse each refueling station node in the refueling station list, take the station entrance as the starting point, take the refueling station node as the end point, calculate the first optimal trajectory from the starting point to the end point and the corresponding first trajectory cost information according to the node planning diagram; take this end point as the new starting point, determine the second optimal trajectory between the new starting point and the target node and the corresponding second trajectory cost information;

[0042] Determine the overall trajectory cost information based on the first trajectory cost information and the second trajectory cost information, and select the refueling position node with the smallest overall trajectory cost information from the refueling position list to obtain the target refueling position node;

[0043] The first optimal trajectory and the second optimal trajectory corresponding to the target refueling position node are combined to obtain the optimal driving route of the vehicle.

[0044] In one or some embodiments, traversing each refueling position node in the refueling position list, starting from the station entrance and ending at the refueling position node, and calculating the first optimal trajectory from the starting point to the end point and the corresponding first trajectory cost information according to the node planning graph specifically include:

[0045] Perform the following operations for each refueling position node in the refueling position list:

[0046] Taking the entrance as the starting point, adding the starting point and the nodes adjacent to the starting point in the node planning graph to the open list, setting the starting point as the parent node of the nodes adjacent to the starting point, and adding the starting point to the closed list;

[0047] Searching for a designated node that can be reached from the starting point from the open list, and saving the trajectory cost information of the designated node in the trajectory list; wherein the designated node selection condition is that the evaluation function value of the node is the smallest; wherein the evaluation function value is determined by the following formula:

[0048] f1=g+h+t;

[0049] Among them, f1 is the evaluation function value, g is the actual cost of moving from the starting point to the specified node, h is the distance from the specified node to the end point, and t is the waiting time information of the specified node;

[0050] Using the specified node as the new starting point, repeat the search operation until the end point is added to the closed list;

[0051] The trajectory cost information of all nodes in the trajectory list is added together to obtain the first trajectory cost information.

[0052] In one or some embodiments, the three-dimensional scene information of the gas station includes an entrance, an exit, a refueling island, obstacles, lanes, and refueling parking spaces; the gas station is divided into nodes according to the three-dimensional scene information of the gas station to generate a node planning diagram, including:

[0053] The station entrance, station exit, refueling island, obstacles, lanes, and refueling parking spaces are set as nodes; the station entrance is the starting point, the station exit is the target node, each node includes the location information of the node, and the refueling space node also includes the type of refueling available and the waiting time information;

[0054] The nodes are connected according to the vehicle movement rules to generate the node planning diagram.

[0055] In one or some embodiments, the device control unit comprises:

[0056] Liquid level meter, used to collect oil level, water level and temperature through probe, and calculate oil volume and water volume by comparing with tank capacity table;

[0057] Leak detection sensor, used to collect oil and water leakage information;

[0058] Gas flow meter, used to collect gas velocity and flow data;

[0059] Pressure sensor, used to collect gas pressure data;

[0060] Oil and gas concentration sensor, used to collect gas concentration data.

[0061] In one or some embodiments, the data processing unit is specifically used to:

[0062] Detect and identify the video image taken by the image acquisition unit to obtain the vehicle's license plate information, vehicle model information and color information, track the vehicle's driving trajectory based on the license plate information, vehicle model information and color information, and map the vehicle's driving trajectory from the video to the virtual gas station scene.

[0063] In one or some embodiments, the data processing unit is specifically used for:

[0064] Using the PP-YOLOE method to perform vehicle detection on the video image to obtain vehicle boundary information;

[0065] Based on the vehicle boundary information, a vehicle image of each vehicle is intercepted in the video image, and a PP-OCR method is used to detect and recognize the vehicle image to obtain license plate information;

[0066] The PP-LCNet method is used to perform vehicle model and color recognition on the vehicle image to obtain vehicle model information and color information.

[0067] In one or some embodiments, the data processing unit is further used for:

[0068] Based on the vehicle boundary information, an observation-centered tracker OC-SORT is used to track the vehicle to obtain the image coordinates of the vehicle in the video image;

[0069] The physical coordinates of the vehicle at the gas station are determined based on the image coordinates by the following formula:

[0070]

[0071] Among them, (u,v) represents the image coordinates, (X w ,Y w ,Z w ) represents the physical coordinate, Z c is the scale factor, (u0, v0) represents the center point of the image, dx and dy represent the physical size of the image pixel in the u-axis and v-axis directions, f represents the focal length of the image acquisition unit, R and T represent the rotation matrix and translation vector of the image acquisition unit in the world coordinate system respectively;

[0072] Based on the mapping relationship between the world coordinate system of the gas station and the virtual coordinate system of the virtual gas station scene, the virtual position information corresponding to the physical coordinates is determined, and the vehicle is mapped to the virtual gas station based on the virtual position information.

[0073] In one or some embodiments, the human-computer interaction unit includes:

[0074] A user input device connected to the gas station digital twin software platform, used to collect user control device instructions and send them to the gas station digital twin software platform;

[0075] A gas station digital twin software platform, connected to a display device, for rendering the virtual gas station scene according to the vehicle driving information and the device status information, and for changing the viewing angle of the virtual gas station scene according to instructions of the user control device;

[0076] Display device, used to display the virtual gas station scene rendered by the 3D engine, vehicle driving information and equipment status information of the gas station facilities and equipment.

[0077] In one or some embodiments, the central control unit is also used to calculate the vehicle waiting time in a queue, the vehicle refueling time and the vehicle's stay time in the gas station based on the vehicle driving information, and to issue an alarm message when the equipment status is abnormal or the oil tank balance is insufficient.

[0078] In one or some embodiments, the data processing unit is further used to predict the time to apply for a tanker truck to replenish oil based on the acquired information on the remaining oil volume in the oil tank.

[0079] In a second aspect, the present invention further provides a method for generating a virtual gas station scene, including: a mapping module mapping a pre-built scene and equipment three-dimensional model and a vehicle three-dimensional model to a virtual gas station scene;

[0080] The processing module obtains the video images in the gas station and the equipment operation data of the gas station facilities and equipment, processes the equipment operation data to obtain equipment status information, processes the video images to obtain vehicle driving information, and maps the vehicle driving information and the equipment status information to the virtual gas station scene.

[0081] In a third aspect, the present invention further provides a method for predicting vehicle flow at a gas station, comprising:

[0082] Obtain historical refueling data within a preset time period at a gas station, and input the historical refueling data into a pre-established traffic flow prediction model to predict the traffic flow; wherein the historical refueling data includes date, temperature, precipitation, snowfall, whether it is a discount, whether it is a holiday, and traffic flow, and the traffic flow prediction model is obtained by the following method:

[0083] Dividing the historical refueling data into a training set and a test set;

[0084] A multivariate linear regression model is used to establish the relationship between traffic volume and date, temperature, precipitation, snowfall, discount, and holiday. The model is expressed as:

[0085] h b (x)=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6+b7x7+e;

[0086] Among them, h b (x) is the traffic volume, x1 is the date of each month, x2 is the temperature, x3 is the precipitation, x4 is the snowfall, x5 is the oil price, x6 is whether it is a discount, x7 is whether it is a holiday, b0, b1, b2, b3, b4, b5, b6, b7 are model parameters, and e is the error term;

[0087] The training set is input into the multivariate linear regression model to obtain the model prediction result. The loss function is determined based on the model prediction result and the actual traffic flow by the following formula:

[0088]

[0089] Among them, J(b) is the loss function, n is the number of data in the training set, and h b (x (i) ) is the model prediction result, y (i) is the actual traffic volume;

[0090] Based on the loss function and the following formula, the model parameters are updated using the gradient descent method to obtain a trained model;

[0091]

[0092] Among them, α is the learning rate, m is the number of data in the training set, is the jth factor affecting traffic flow in the i-th group of data, b j is the jth model parameter, j=1.2....7;

[0093] The trained model is evaluated based on the test set, and the training process and the evaluation process are repeated until preset conditions are met to obtain a traffic flow prediction model.

[0094] In a fourth aspect, the present invention further provides a method for planning a driving route of a gas station vehicle, comprising:

[0095] Divide the gas station into nodes according to the three-dimensional scene information of the gas station and generate a node planning map;

[0096] Based on the oil type of each refueling parking space, generate the oil type vector of each refueling space node;

[0097] Generate an oil type vector of the vehicle to be refueled based on the type of oil required by the vehicle to be refueled;

[0098] For the vehicle to be refueled, calculate the inner product of its oil type vector and the oil type vector of each refueling position node, and add the nodes with a multiplication result of 1 to the refueling position list;

[0099] Traverse each refueling station node in the refueling station list, take the station entrance as the starting point, take the refueling station node as the end point, calculate the first optimal trajectory from the starting point to the end point and the corresponding first trajectory cost information according to the node planning diagram; take this end point as the new starting point, determine the second optimal trajectory between the new starting point and the target node and the corresponding second trajectory cost information;

[0100] Determine the overall trajectory cost information based on the first trajectory cost information and the second trajectory cost information, and select the refueling position node with the smallest overall trajectory cost information from the refueling position list to obtain the target refueling position node;

[0101] The first optimal trajectory and the second optimal trajectory corresponding to the target refueling position node are combined to obtain the optimal driving route of the vehicle.

[0102] Based on the above technical solution, the beneficial effects of the present invention compared with the prior art are as follows:

[0103] The gas station digital twin system provided by the embodiment of the present invention processes the equipment operation data through the processing module to obtain the equipment status information, processes the video images in the gas station to obtain the vehicle driving information, integrates the data collected by the image acquisition unit and the equipment control unit, etc., to realize video analysis and trajectory tracking of the gas station vehicles; by analyzing the historical refueling data, the number of refueling vehicles, the refueling amount, the average refueling time and the fuel consumption forecast value in the future time period are predicted; the vehicle flow is obtained by prediction based on the historical refueling data; and the A* algorithm is used to perform path planning based on the three-dimensional scene information of the gas station based on the virtual gas station scene to obtain the optimal driving route of the vehicle. The system has the ability to integrate all equipment and data, and the equipment management method is centralized and efficient. The present invention realizes the prediction and simulation of the operation status of the gas station in the future time period by collecting, analyzing and predicting the key business data, which greatly improves the intelligence level and risk warning capability of the gas station operation.

[0104] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0105] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0106] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0107] Figure 1 is a schematic diagram of the system structure of a gas station digital twin system provided by an embodiment of the present invention;

[0108] Figure 2 is a video analysis flow chart of a data processing unit;

[0109] Figure 3 It is a diagram of the relationship between various coordinate systems;

[0110] Figure 4 This is a node planning diagram provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0111] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0112] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0113] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0114] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0115] Digital twin is a digital concept and technical means. It is based on the integration of data and models. It simulates, verifies, predicts and controls the entire life cycle of physical entities by building accurate digital mapping of physical objects in real time in digital space, based on data integration and analytical prediction, and finally forms an optimized closed loop of intelligent decision-making. The physical objects include physical objects, behaviors and processes. The data involved in building twins include real-time sensor data and operation history data. The integrated models include physical models, mechanism models and process models.

[0116] Digital twins originated in the aerospace and military fields. In recent years, they have continued to expand into vertical industries such as smart manufacturing and smart cities, realizing application values ​​such as mechanism description, abnormal diagnosis, risk prediction, and decision-making assistance. At present, digital twin technology has been maturely applied in many fields. For example, the existing technology 202110583517.8 provides a digital twin system for wind power generation, which has good fidelity and real-time performance, and can significantly improve the efficiency and level of digital operation and maintenance of wind turbines; the existing technology 202211315607.X provides a smart canteen management method and system based on digital twins, which realizes decentralized dining and safety control in the canteen by establishing a dangerous warning model for human flow; the existing technology 202211157189.6 provides a digital twin system for industrial production, which can optimize the production process and production plan, so as to better predict and manage production.

[0117] A gas station is a specialized business place for filling gasoline and diesel into motor vehicle tanks. Compared with other application scenarios, gas stations have very significant particularities, including frequent flow of personnel and vehicles, a large number of equipment types and complex and diverse data, very high security level requirements, and so on. For the application scenario of gas stations, the prior art 202110195135.8 provides a digital twin gas station, which calculates the vehicle refueling time and gas station congestion according to several important time nodes of the refueling vehicle during the refueling process, which can improve the vehicle refueling speed and gas station traffic efficiency to a certain extent. However, the solution lacks video analysis and trajectory tracking of vehicles, and cannot correspond each time node in the gas station to multiple vehicles. The inventor found in his work that the currently widely used integrated and intelligent management platform for gas station scenarios has the following main problems: 1) lack of ability to integrate all equipment and data, and loose and inefficient equipment management methods; 2) lack of visualization of business data and processes, and station operation and maintenance methods are not intuitive enough; 3) key business data cannot be effectively collected and analyzed, which is not conducive to the improvement and optimization of business processes; 4) unable to predict and simulate future operation conditions, lack of risk warning capabilities, and cannot meet the inventor's expectations. Therefore, it is necessary to study a more effective digital twin system for gas station application scenarios to cope with the shortcomings of existing management methods and improve the management efficiency and intelligence level of gas stations.

[0118] Embodiment 1

[0119] An embodiment of the present invention provides a gas station digital twin system, comprising: a mapping module and a processing module, wherein the mapping module is connected to the processing module;

[0120] The mapping module is used to map the pre-built scene and equipment three-dimensional model and vehicle three-dimensional model to the virtual gas station scene;

[0121] The processing module is used to obtain the video image in the gas station and the equipment operation data of the gas station facilities and equipment, process the equipment operation data to obtain equipment status information, process the video image to obtain vehicle driving information, and map the vehicle driving information and the equipment status information to the virtual gas station scene;

[0122] Obtain historical refueling data of the gas station, analyze the historical refueling data, predict the number of refueling vehicles, refueling amount, average refueling time and fuel consumption forecast value in the future period, and predict the traffic flow;

[0123] The optimal driving route of the vehicle is obtained by performing simulation based on the three-dimensional scene information of the virtual gas station scene.

[0124] In the above technical solution, the scene and equipment three-dimensional model is used to perform three-dimensional modeling of gas station facilities and equipment, and map the gas station facilities and equipment to the virtual gas station scene; gas station facilities and equipment refer to real physical objects in the physical world, including: 1) buildings (structures) in the gas station, including canopies, gas islands, fields, and roads; 2) oil tanks; 3) underground oil pipelines; 4) gas pumps; 5) gas guns; wherein the oil tanks, underground oil pipelines, gas pumps, and gas guns are connected to the central control unit through the equipment control unit. The scene and equipment three-dimensional model is a three-dimensional digital model constructed for gas station facilities and equipment, which is used to map the physical world to the virtual world and display the virtual gas station scene in the form of a three-dimensional model.

[0125] In the above technical solution, the vehicle three-dimensional model is used to perform three-dimensional modeling on vehicles of different models and map the vehicles to the virtual gas station scene; the vehicle three-dimensional model is a three-dimensional model constructed for different models. In addition to the three-dimensional information of the vehicle, the model also includes information such as the color, texture, and material of the vehicle surface, which is used to generate a realistic three-dimensional scene and display the vehicles in the physical world in the virtual gas station scene.

[0126] In the above technical solution, the three-dimensional models of scenes and equipment and the three-dimensional models of vehicles are constructed in an offline manner. In addition to the three-dimensional information of physical entities, the model also includes information such as the color, texture, and material of the surface of the object. After being processed by the 3D rendering engine, a realistic three-dimensional scene can be obtained. The Unreal Engine 4 3D engine is used to build the three-dimensional models of scenes and equipment. The three-dimensional resources used include the software's built-in models and texture materials, and the animation is completed using AutoDesk 3dsmax. The three-dimensional information of gas station facilities such as canopies, refueling islands, fields, roads, etc. is provided by construction drawings, and the models are created manually on this basis. The three-dimensional models of gas station equipment such as gas pumps, refueling guns, and oil tanks are directly converted from CAD models.

[0127] In the above technical solution, the virtual gas station scene uses an independent virtual coordinate system, and the video information collected by the image acquisition unit uses the image acquisition unit coordinate system. The two are converted to each other with the world coordinate system of the real gas station through a specific mapping method. Furthermore, after the data processing unit obtains the vehicle position information appearing in the video, it can be mapped to the virtual gas station scene to display the vehicle's driving trajectory in real time and accurately.

[0128] In the above technical solution, the three-dimensional model of the vehicle is created manually. A total of 9 models of vehicles are established, including sedans, sports utility vehicles (SUVs), hatchbacks, multi-purpose vehicles (MPVs), vans, pickup trucks, buses, trucks, and station wagons. The vehicle model is stored in OBJ format, and the model contains vehicle color, texture, material, three-dimensional information, etc. In this embodiment, the three-dimensional models of different vehicle models that have been established in the prior art can be directly used, and the specific process of modeling will not be repeated.

[0129] The gas station digital twin system provided by the embodiment of the present invention processes the equipment operation data through the processing module to obtain the equipment status information, processes the video images in the gas station to obtain the vehicle driving information, integrates the data collected by the image acquisition unit and the equipment control unit, etc., to realize video analysis and trajectory tracking of the gas station vehicles; by analyzing the historical refueling data, the number of refueling vehicles, the refueling amount, the average refueling time and the fuel consumption forecast value in the future time period are predicted; the vehicle flow is obtained by prediction based on the historical refueling data; and the A* algorithm is used to perform path planning based on the three-dimensional scene information of the gas station based on the virtual gas station scene to obtain the optimal driving route of the vehicle. The system has the ability to integrate all equipment and data, and the equipment management method is centralized and efficient. The present invention realizes the prediction and simulation of the operation status of the gas station in the future time period by collecting, analyzing and predicting the key business data, which greatly improves the intelligence level and risk warning capability of the gas station operation.

[0130] In an alternative embodiment, referring to Figure 1 As shown, the processing module includes a central control unit and a data processing unit, and the central control unit is connected to the data processing unit;

[0131] The central control unit is used to manage and control the entire gas station digital twin system, respond to user control device instructions and data processing requirements, forward data and transmit control commands. Specifically, it is used to send the received video images and the device operation data to the data processing unit;

[0132] The data processing unit is used to process the device operation data to obtain device status information and process the video image to obtain vehicle driving information;

[0133] Acquire the historical refueling data, analyze the historical refueling data, predict the number of refueling vehicles, refueling amount, average refueling time and fuel consumption forecast value in the future period, and predict the traffic flow;

[0134] Furthermore, a simulation is performed based on the three-dimensional scene information of the virtual gas station scene to obtain the best driving route for the vehicle.

[0135] In an alternative embodiment, referring to Figure 1 As shown, the processing module further includes: a human-computer interaction unit connected to the central control unit;

[0136] The central control unit is further used to send the device status information and the vehicle driving information to the human-computer interaction unit;

[0137] The human-computer interaction unit is used to map the vehicle driving information and the device status information into the virtual gas station scene, and to display the virtual gas station scene rendered by a 3D engine in real time.

[0138] In an alternative embodiment, referring to Figure 1 As shown, the system further includes an image acquisition unit connected to the central control unit;

[0139] The image acquisition unit is used to capture video images in the gas station and send the video images to the central control unit. The image acquisition unit can be an electronic device capable of capturing images, such as a camera. The camera is installed under the canopy of the gas station, and the shooting range covers the driving route of the vehicle during the refueling process. It is usually necessary to install multiple cameras to achieve full coverage of the gas station scene.

[0140] In an alternative embodiment, referring to Figure 1 As shown, the system further includes a device control unit connected to the central control unit;

[0141] The equipment control unit is used to collect equipment operation data of the gas station facilities and equipment, control and manage the gas station facilities and equipment, and send the equipment operation data to the central control unit.

[0142] The equipment control unit includes a variety of sensors and controllers connected to various facilities and equipment in the gas station. Under the premise of ensuring communication security, it is connected to the central control unit through a wired or wireless network. Specifically, it includes: 1) Liquid level meter, which collects original data such as oil level height, water level height, temperature, etc. through a probe, and can calculate oil volume, water volume and other data by comparing with the tank capacity table and using the volume calculation formula; 2) Leak detection sensor, which is used to collect oil and water leakage information; 3) Gas flow meter, which is used to collect gas flow rate and flow data; 4) Pressure sensor, which is used to collect gas pressure data; 5) Oil and gas concentration sensor, which is used to collect gas concentration data.

[0143] In an alternative embodiment, referring to Figure 1 As shown, the system further includes: a data storage unit connected to the data processing unit;

[0144] The data storage unit is used to store the historical refueling data, the equipment status information and the vehicle driving information; the historical refueling data stored in the data storage unit includes the vehicle license plate number, driving track, refueling amount, refueling amount, refueling time, weather conditions and oil price adjustment, etc., which are used for historical data analysis and future data prediction of gas stations. The historical refueling data can be used through the big data analysis function of the data processing unit to analyze and predict the congestion situation and oil consumption of gas stations.

[0145] In the above technical solution, the data storage unit uses the relational database MySQL to store historical refueling data, including vehicle driving trajectory, refueling amount, refueling amount, refueling time and weather conditions, etc. The data can be stored on a local storage medium or on a remote server and accessed through the network.

[0146] In the above technical solution, the data processing unit is used to process various data distributed by the central control unit, and send the processing results to the central control unit and the data storage unit according to the requirements. Specifically, it is used to process the equipment operation data to obtain equipment status information, process the video image to obtain vehicle driving information, and send the equipment status information and the vehicle driving information to the central control unit and the data storage unit.

[0147] The gas station digital twin system provided by the present invention can operate in the following four modes: historical playback, real-time control, future prediction and simulation deduction. Historical playback refers to obtaining the gas station operation data of a specified historical period and performing statistical analysis on key business data. Real-time control refers to displaying vehicle information in real time in a virtual gas station scene, while displaying the operation data of the gas station equipment in a graphical manner. Future prediction refers to analyzing historical refueling data to obtain predicted values ​​for the number of refueling vehicles, refueling volume, average refueling time, etc. in future time periods. Simulation deduction refers to the simulation of virtual refueling of vehicles according to the actual refueling process. Among them:

[0148] 1. Historical playback

[0149] By reading the historical refueling data in the data storage unit, the gas station operation data of the specified historical period can be obtained, including the driving trajectory of the refueling vehicles, the amount of refueling and the remaining oil in the tank, etc., which can be played in the form of 3D animation through the human-computer interaction unit. The system can also count the total number of refueling vehicles, total refueling volume, average refueling time and other key business data in the specified historical period, and give the change curve and change law of the data over a period of time.

[0150] 2. Real-time control

[0151] In real-time control mode, the human-computer interaction unit displays the vehicle information in the virtual gas station scene in real time, and graphically displays the status information of gas station equipment such as gas pumps and liquid level meters, and issues an alarm signal when the equipment status is abnormal. Managers can use the human-computer interaction unit to achieve comprehensive, intuitive and efficient gas station data browsing and equipment management.

[0152] 3. Future predictions

[0153] Based on big data technology, historical refueling data is analyzed, including date, time period, weather conditions, whether it is a holiday, oil price adjustment, etc., to obtain predicted values ​​of the total number of refueling vehicles, total refueling volume, average refueling time, etc. in a certain period of time in the future. This makes it easier for managers to prepare oil products in advance, activate an appropriate number of refueling machines and refueling guns, and equip sufficient refueling staff. If the gas station load is exceeded, a risk warning will be issued in advance.

[0154] 4. Simulation

[0155] Given the number of refueling vehicles and the type of oil required within a certain period of time, the system simulates the actual refueling process of the gas station. The system automatically assigns fuel dispensers and fuel guns to vehicles, and simulates the driving routes and stop times of vehicles according to the refueling sequence. The remaining oil in the tank is simulated according to the oil consumption, and the application of tank trucks to refill oil is made to achieve a closed-loop simulation of the gas station operation process.

[0156] The gas station digital twin system provided by the embodiment of the present invention belongs to the field of smart gas station management technology. The gas station facilities and equipment in the physical world are mapped to the virtual gas station scene through the scene and equipment three-dimensional models, and the gas station facilities and equipment in the physical world are displayed in the virtual gas station scene; the vehicle three-dimensional model is used to display the vehicle in the physical world in the virtual world; the video images in the gas station are collected by the image acquisition device, and the equipment operation data of the gas station facilities and equipment and the control and management of the gas station facilities and equipment are collected by the equipment control unit, which can effectively collect and analyze key business data, which is conducive to the improvement and optimization of business processes; the equipment operation data is processed by the data processing unit to obtain equipment status information, and the video images in the gas station collected by the image acquisition unit are processed to obtain vehicle driving information, and the data collected by the image acquisition unit and the equipment control unit are used to collect the data of the gas station facilities and equipment. According to the integration of data, the video analysis and trajectory tracking of gas station vehicles can be realized; by analyzing the historical refueling data in the data storage unit, the number of refueling vehicles, refueling volume, average refueling time and fuel consumption forecast values ​​in the future period can be predicted; based on the historical refueling data, the diffusion convolution recursive neural network method is used to predict the vehicle flow; and based on the three-dimensional scene information of the virtual gas station scene, the A* algorithm is used for path planning to obtain the best driving route for the vehicle, which has the ability to integrate all equipment and data, and the equipment management method is centralized and efficient; the present invention realizes the prediction and simulation of the future operation of the gas station through the collection, analysis and prediction of key business data, which greatly improves the intelligent level of gas station operation and risk warning ability. The human-computer interaction unit can display the virtual gas station scene rendered by the 3D engine in real time, providing a way to visualize business data and processes, and the site operation and maintenance methods are more intuitive.

[0157] The present invention efficiently integrates various types of business data in gas stations and provides visualization solutions. By establishing a centralized gas station operation and maintenance platform, the efficiency of gas station operation and maintenance is improved. By collecting and analyzing key business data such as oil consumption and vehicle flow, data support is provided for optimizing gas station business processes, and the intelligence level and risk warning capabilities of gas station operation and maintenance are improved. In addition, the present invention comprehensively utilizes technologies such as the Internet of Things (data collection and transmission), artificial intelligence (video analysis), big data (data analysis and prediction), and virtual reality (3D modeling and visualization) to achieve a highly simulated and real-time gas station digital twin system, which can greatly improve the management and control level and risk warning capabilities of gas stations.

[0158] In an optional embodiment, the human-computer interaction unit includes:

[0159] A user input device connected to the gas station digital twin software platform, used to collect user control device instructions and send them to the gas station digital twin software platform;

[0160] A gas station digital twin software platform, connected to a display device, for rendering the virtual gas station scene according to the vehicle driving information and the device status information, and for changing the viewing angle of the virtual gas station scene according to instructions of the user control device;

[0161] Display device, used to display the virtual gas station scene rendered by the 3D engine, vehicle driving information and equipment status information of the gas station facilities and equipment.

[0162] The human-computer interaction unit displays the virtual gas station scene, vehicle driving information, and equipment status information of the gas station equipment rendered by the 3D engine in real time through the display device. Managers can use the mouse to observe the gas station facilities and vehicles in the virtual gas station scene from any perspective, and can also send control instructions to designated gas station equipment based on the equipment status information of the gas station equipment.

[0163] In an optional embodiment, the central control unit is also used to calculate the waiting time in the queue, the refueling time and the residence time of the vehicle in the gas station according to the vehicle driving information, and to issue an alarm message when the equipment status is abnormal or the oil tank is insufficient. The time from the vehicle entering the gas station to the time it stops in front of the gas pump to start refueling is the waiting time in the queue. The time from stopping in front of the gas pump to driving away from the gas pump is the refueling time of the vehicle. The time from entering the gas station to leaving the gas station is the residence time of the vehicle in the gas station. Furthermore, the congestion of the gas station can be known based on the waiting time in the queue of the vehicle, and the waiting time required for subsequent vehicles to refuel can be predicted. The waiting time in the queue of the vehicle can be calculated by obtaining the number of vehicles in front of the current vehicle, the location information of each vehicle, the type of oil required and the refueling time of the vehicle, and the status information of each gas pump (i.e., whether it is idle). The refueling time can be calculated based on the remaining oil information of the vehicle.

[0164] In an optional embodiment, when the system operates in real-time control mode, the central control unit sends the data and status information collected by the device control unit to the human-computer interaction unit for display in a graphical manner. The video images collected by the image acquisition unit are sent to the data processing unit, which analyzes and processes the video and then sends the returned results to the human-computer interaction unit. The data processing unit has a video data analysis function. The video data analysis refers to automatically detecting and identifying license plates from the video taken by the image acquisition unit, identifying vehicle colors and models, tracking vehicle driving trajectories, and mapping vehicle driving trajectories from the video to the virtual gas station scene. Reference Figure 2 As shown, the data processing unit is specifically used for:

[0165] S101, using the PP-YOLOE method to perform vehicle detection on the video image to obtain vehicle boundary information;

[0166] The present invention uses the PP-YOLOE method based on deep learning to realize vehicle detection. PP-YOLOE is a network model with high precision and fast reasoning speed, including modules such as backbone network CSPRepResNet, feature fusion CSPPAN, lightweight ET-Head and dynamic matching algorithm TAL. In the training stage, the Objects365 large-scale data set containing 365 categories and millions of data volumes is used as the training set. In the reasoning stage, target detection is performed on each frame image in the video, and the corresponding bounding box is recorded for the target belonging to the vehicle category, and the detection results of other categories are directly ignored. The present invention can select the existing vehicle detection model based on the PP-YOLOE network for vehicle detection, and the training process of the model will not be repeated.

[0167] S102, intercepting a vehicle image of each vehicle in the video image based on the vehicle boundary information, and detecting and recognizing the vehicle image using a PP-OCR method to obtain license plate information;

[0168] According to the output result of S101, each vehicle is intercepted in the input image, and then the PP-OCR method is used to detect and recognize the license plate. Specifically, the PP-OCR model is used to realize license plate recognition, which includes two stages. The first stage uses the DB method to realize text detection, and the second stage uses the CRNN method to realize text recognition. A text direction classifier is added between the detection and recognition stages to cope with text recognition in different directions. After the license plate is successfully recognized, it is used as the unique identifier of the vehicle for subsequent vehicle tracking. The present invention can select the existing PP-OCR model for license plate recognition, and the training process of the model will not be repeated.

[0169] S103, using the PP-LCNet method to perform vehicle model and color recognition on the vehicle image to obtain vehicle model information and color information;

[0170] First, each vehicle is intercepted in the input image according to the output result of S101, and then the PP-LCNet method is used to realize the recognition of vehicle color and model. PP-LCNet is an efficient lightweight network model. Compared with other lightweight models, the model can improve the accuracy of the model without increasing the reasoning time. The vehicle color is divided into 10 categories including yellow, orange, green, gray, red, blue, white, gold, brown and black, and the vehicle model is divided into 9 categories including sedan, sports utility vehicle (SUV), hatchback, multi-purpose vehicle (MPV), van, pickup, bus, truck and travel car, corresponding to the established vehicle three-dimensional model. Correct vehicle color and vehicle model information are conducive to generating a realistic virtual gas station scene. The present invention can select the existing PP-LCNet model for vehicle color and vehicle model recognition, and the training process of the model is not repeated.

[0171] S104, sending the license plate information, vehicle model information and color information to the central control unit and the data storage unit respectively.

[0172] In an optional embodiment, the data processing unit is further used for:

[0173] S105, based on the vehicle boundary information, using an observation-centered tracker OC-SORT to track the vehicle to obtain image coordinates of the vehicle in the video image;

[0174] Based on the vehicle boundary detected in stage S101, the observation-centered tracker OC-SORT is used to track the vehicle. This method is a simple and efficient online tracking method that can better adapt to the nonlinear motion of the target and improve the robustness to occlusion. By binding the vehicle's driving trajectory with the license plate information obtained in stage S102, the vehicle's waiting time in the queue, refueling time, departure time, and refueling location can be obtained.

[0175] S106. Determine the physical coordinates of the vehicle at the gas station based on the image coordinates by the following formula:

[0176]

[0177] The physical position of the vehicle at the gas station is calculated based on the pixel position of the vehicle in the video image obtained in S105. The video information collected by the image acquisition unit adopts the image coordinate system, and the gas station adopts the world coordinate system, and the two satisfy the relationship shown in the above formula. According to the above formula, the physical coordinates of the vehicle at the gas station are determined to obtain the above vehicle driving information.

[0178] Among them, (u,v) represents the image coordinates, (X w ,Y w ,Z w) represents the physical coordinate, Z c is the scale factor, (u0, v0) represents the center point of the image, dx and dy represent the physical size of the image pixel in the u-axis and v-axis directions, f represents the focal length of the image acquisition unit, R and T represent the rotation matrix and translation vector of the image acquisition unit in the world coordinate system respectively; according to Zhang Zhengyou's calibration method, these unknown parameters can be calculated by taking multiple black and white grid images at different angles by the image acquisition unit. Figure 3 As shown, the camera coordinate system is (X C ,Y C ,Z C ), the image pixel coordinate system is (X W ,Y W ,Z W )

[0179] S107. Determine virtual position information corresponding to the physical coordinates based on a mapping relationship between a world coordinate system of the gas station and a virtual coordinate system of the virtual gas station scene, and map the vehicle to the virtual gas station based on the virtual position information.

[0180] After obtaining the image coordinates of the vehicle in the video image, i.e., the pixel position, the vehicle in the world coordinate system (Y W ,Y W ,Z W ) position. Furthermore, the virtual gas station scene uses an independent virtual coordinate system (X V ,Y V ,Z V ), the origin of the virtual coordinate system is O V is the center point of the gas station scene, refer to Figure 3 As shown, the direction of the coordinate axis is the same as that of the world coordinate system. By setting the origin and scale correspondence, the position of the vehicle in the world coordinate system can be corresponded to the virtual coordinate system through simple translation and scaling, which is used to display the vehicle 3D model in the gas station scene and display the vehicle driving trajectory in an animated manner. The processing results obtained in stages S101 to S107 are sent to the central control unit and the data storage unit respectively, including license plate information, color information, vehicle model information, and vehicle driving trajectory in the video image and the physical world.

[0181] In an optional embodiment, the data processing unit also includes a big data analysis function, which can obtain a predicted value of fuel consumption in the future based on historical refueling data, including date, time period, whether it is a holiday, oil price changes, weather conditions, refueling amount, etc. When the system works in the future prediction mode, the data processing unit uses a multivariate linear regression model to predict oil consumption and vehicle flow. The historical refueling data stored in the data storage unit, including date, time period, whether it is a holiday, weather conditions, refueling amount, etc., is used as an algorithm input to obtain the number of refueling vehicles and fuel consumption predictions in the future. The data processing unit is specifically used to:

[0182] The historical refueling data within a preset time period of the gas station is obtained, and the historical refueling data is input into a pre-established traffic flow prediction model to predict the traffic flow; wherein the historical refueling data includes date, temperature, precipitation, snowfall, whether there is a discount, whether it is a holiday and traffic flow.

[0183] The above-mentioned preset time period can be set according to actual needs. For example, the date, temperature, precipitation, snowfall, oil price, whether there are any promotional activities (discounts, etc.), whether it is a holiday, and traffic flow information (statisticed by day) of a gas station in the past year or half a year can be obtained. Based on the data in these dimensions, the traffic flow situation of the gas station in the future can be predicted.

[0184] The vehicle flow prediction model is obtained in the following way:

[0185] S201, dividing the historical refueling data into a training set and a test set;

[0186] Before dividing the historical refueling data, the data is preprocessed, such as cleaning the data, filling or deleting missing values, and processing outliers. The specific methods can refer to the description in the prior art and will not be repeated here. Most of the data is divided into a training set, and a small part is divided into a test set. The ratio can be set according to actual needs, such as 4:1.

[0187] S202. Use a multiple linear regression model to establish the relationship between traffic volume and date, temperature, precipitation, snowfall, discount, and holiday. The model is expressed as:

[0188] h b (x)=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6+b7x7+e;

[0189] Among them, h b(x) is the traffic volume, x1 is the date of each month, x2 is the temperature (℃), x3 is the precipitation (mm), x4 is the snowfall (mm), x5 is the oil price (yuan), x6 is whether it is a discount (1 or 0), x7 is whether it is a holiday (1 or 0), b0, b1, b2, b3, b4, b5, b6, b7 are model parameters, and e is the error term;

[0190] S203, input the training set into the multivariate linear regression model to obtain the model prediction result, and determine the loss function based on the model prediction result and the actual vehicle flow by the following formula:

[0191]

[0192] Among them, J(b) is the loss function, n is the number of data in the training set, and h b (x (i) ) is the model prediction result, y (i) is the actual traffic volume;

[0193] The model is trained and the parameters of the model are randomly initialized. The present invention selects the mean square error as the loss function. The gradient descent method is used to update the model parameters. The smaller the loss function, the better the fitting effect. The partial derivative of the loss function with respect to the parameters can be obtained:

[0194]

[0195] therefore:

[0196]

[0197] S204, based on the loss function and the following formula, using the gradient descent method to update the model parameters to obtain a trained model;

[0198]

[0199] Among them, α is the learning rate, m is the number of data in the training set, is a multiple linear regression model, i.e. date, temperature, precipitation, snowfall, discount, holiday, b j is the jth model parameter, j=1.2....7;

[0200] S205, evaluating the trained model based on the test set, repeating the training process and the evaluation process until a preset condition is met, thereby obtaining a traffic flow prediction model;

[0201] Evaluate the model after each training based on the test set to determine whether the trained model meets the preset conditions. Use the trained model to predict the traffic flow results of the test set, evaluate the predicted results with the actual results, and use the mean square error as an evaluation method to measure the degree of fit of the trained model. If the mean square error is greater than the preset error threshold, repeat the above steps S203 to S204, and the model parameters will gradually converge as the learning rate decreases. Then iterate the training model for 30 times, generally not more than 50 times, until the model converges, stop training, and obtain a trained traffic flow prediction model.

[0202] By using the trained traffic flow prediction model to predict new samples, we can obtain the traffic flow situation of the gas station in the future based on the temperature, precipitation, snowfall, oil price, whether there is a discount, and whether it is a holiday.

[0203] Similarly, a multivariate linear regression model is used to establish the relationship between oil consumption and date, temperature, precipitation, snowfall, discount, and holiday, and an oil consumption prediction model can be established to predict oil consumption. Similarly, by analyzing the historical refueling data, the number of refueling vehicles, refueling volume, average refueling time, etc. in the future period can be predicted, which will not be repeated here.

[0204] In an optional embodiment, the device control unit is also used to collect the remaining oil volume information of the oil tank, and the data processing unit is also used to predict the time to apply for a tanker to replenish oil products based on the remaining oil volume information of the oil tank collected by the device control unit. According to the refueling volume in the historical refueling data, the daily oil consumption can be obtained, and the remaining oil volume information of the oil tank is divided by the daily oil consumption to obtain the oil available time, thereby obtaining the time to apply for a tanker to replenish oil products.

[0205] In an optional embodiment, when the system works in the simulation mode, the data processing unit can simulate the actual oil gun oil products and refueling process of the gas station according to the number and model of refueling vehicles and the number and type of required oil products within a given time period. The classic A* algorithm is used to simulate the vehicle's driving trajectory, and the three-dimensional scene information of the gas station, including the refueling island, obstacles, lanes, refueling parking spaces and other information, is used as the algorithm input to give the optimal vehicle driving route. The refueling time is calculated according to the amount of required oil products, and the refueling process of all vehicles is simulated in sequence. The remaining oil in the tank is simulated according to the oil consumption and the application for oil tankers to replenish oil products is made to realize the closed-loop simulation of the gas station operation process. The three-dimensional scene information of the gas station based on the virtual gas station scene is simulated using the A* algorithm to obtain the optimal driving route for the vehicle, which specifically includes:

[0206] S301, dividing the gas station by nodes according to the three-dimensional scene information of the gas station, and generating a node planning map;

[0207] S302, generating an oil type vector of each refueling position node based on the oil type of each refueling position;

[0208] Each refueling station node contains the type of fuel that can be refueled in the parking space, that is, the above-mentioned fuel type, which is mapped into a vector of length 3, corresponding to 92, 95, and 98 oils. For example: for the refueling station node representing 95 oil, its refueling vector is [0,1,0].

[0209] S303, generating an oil type vector of the vehicle to be refueled based on the type of oil required by the vehicle to be refueled;

[0210] The vehicle to be refueled has a refueling vector, which is mapped to a vector of length 3, corresponding to 92, 95, and 98 gasoline. For example, [0,1,0] means that the vehicle needs to be refueled with 95 gasoline.

[0211] S304, for the vehicle to be refueled, calculate the inner product of its oil type vector and the oil type vector of each refueling position node, and add the node whose multiplication result is 1 to the refueling position list;

[0212] Before planning the driving route, four lists are defined: 1) a refueling position list, which is used to store the location information of refueling; 2) an open list, which is used to store the node information that currently needs to be considered; 3) a closed list, which stores all nodes that no longer need to be paid attention to; 4) a trajectory list, which is used to store multiple predicted trajectories and the cost information of the trajectory.

[0213] If the inner product of the oil type vector of the vehicle to be refueled and the oil type vector of a refueling position node is 1, it means that the oil type of the refueling position node is the same as the oil type of the vehicle to be refueled, and the refueling position node is added to the refueling position list.

[0214] S305, traverse each refueling position node in the refueling position list, take the station entrance as the starting point, take the refueling position node as the end point, calculate the first optimal trajectory from the starting point to the end point and the corresponding first trajectory cost information according to the node planning diagram; take this end point as the new starting point, determine the second optimal trajectory between the new starting point and the target node and the corresponding second trajectory cost information;

[0215] S306: determine the overall trajectory cost information based on the first trajectory cost information and the second trajectory cost information, and select the refueling position node with the smallest overall trajectory cost information from the refueling position list to obtain the target refueling position node.

[0216] For each refueling position node, the first trajectory cost information and the second trajectory cost information are added together to obtain the overall trajectory cost information, and the refueling position node with the smallest overall trajectory cost information is selected from the refueling position list to obtain the target refueling position node.

[0217] S307: Combine the first optimal trajectory and the second optimal trajectory corresponding to the target refueling position node to obtain an optimal driving route for the vehicle.

[0218] In an optional embodiment, traversing each refueling position node in the refueling position list, starting from the station entrance and ending at the refueling position node, and calculating the first optimal trajectory from the starting point to the end point and the corresponding first trajectory cost information according to the node planning graph specifically includes:

[0219] Perform the following operations for each refueling position node in the refueling position list:

[0220] S3051, taking the entrance as the starting point, adding the starting point and the nodes adjacent to the starting point in the node planning graph to the open list, setting the starting point as the parent node of the nodes adjacent to the starting point, and adding the starting point to the closed list;

[0221] Searching for a designated node that can be reached from the starting point from the open list, and saving the trajectory cost information of the designated node in the trajectory list; wherein the designated node selection condition is that the evaluation function value of the node is the smallest; wherein the evaluation function value is determined by the following formula:

[0222] f1=g+h+t;

[0223] Among them, f1 is the evaluation function value, g is the actual cost of moving from the starting point to the specified node, which refers to the distance from the starting point S to the specified node, not the straight-line distance, but the total distance after the distances of adjacent nodes are added together, that is, first obtain the g value of its parent node, and then add the distance moved by the node in the straight or diagonal direction of the parent node; h is the distance from the specified node to the end point, and according to the connection between the nodes, it can only move horizontally or vertically; t is the time information required to wait for the specified node; f1 can be used to judge the cost required to walk from the starting point through the node to the end point.

[0224] Traverse the open list and calculate the evaluation function value f1 of each node in the list. Find the node with the smallest f1 and get the specified node.

[0225] S3052, taking the designated node as a new starting point, repeating the search operation until the end point is added to the closed list;

[0226] With the designated node as the new starting point, the search operation of step S3052 is repeated until the end point is added to the closed list. To determine the path of the trajectory, one can start from the end point E and search for its parent node step by step until the starting point S. This trajectory is the first optimal trajectory of the refueling position node.

[0227] S3053: Add the trajectory cost information of all nodes in the trajectory list to obtain first trajectory cost information.

[0228] Starting from the end point, move toward the parent node until returning to the starting point, and connect all the specified nodes to get the first optimal trajectory from the refueling node to the end point. Add the g value of the end node of the first optimal trajectory and the t value of all nodes on the path as the cost of the first half of the vehicle's total trajectory to get the first trajectory cost information.

[0229] Similarly, the second optimal trajectory of the new starting point and target node can be determined, and the g value of the end node of the second optimal trajectory and the t value of all nodes on the path are added as the cost of the second half of the vehicle's total trajectory to obtain the second trajectory cost information. After both parts of the trajectory are predicted, the two parts of the trajectory are spliced ​​and the costs are added to obtain the complete trajectory and its overall trajectory cost information.

[0230] Until all the refueling station nodes in the refueling station list have calculated the optimal trajectory and their overall trajectory cost information, the one with the lowest overall trajectory cost information is selected as the optimal driving route of the vehicle in the complete trajectory.

[0231] In an optional embodiment, the three-dimensional scene information of the gas station in the above step S301 includes an entrance, an exit, a refueling island, obstacles, lanes, and refueling parking spaces; the gas station is divided into nodes according to the three-dimensional scene information of the gas station to generate a node planning diagram, including:

[0232] S3011, setting the station entrance, station exit, refueling island, obstacle, lane, and refueling parking space as nodes; wherein the station entrance is the starting point, the station exit is the target node, each node includes the location information of the node, and the refueling space node also includes the type of refueling available and the waiting time information;

[0233] The gas station is divided into nodes. In addition to the entrance as the starting point and the exit as the end point, each refueling parking space and the waiting area next to the refueling parking space are set as nodes. Figure 4 As shown, the dark dots represent waiting parking spaces. Each refueling node contains the refueling types that can be refueled in the parking space, which are mapped to vectors of length 3, corresponding to 92, 95, and 98 oils. For example: For the refueling node representing 95 oil, its refueling vector is [0,1,0]. The refueling type of the waiting parking space is [0,0,0]. At the same time, each node contains location information, which is used to calculate the distance between nodes. The distance here is the Euclidean distance. Among them, x and y represent the horizontal coordinate and vertical coordinate of the node respectively. In addition, the refueling node representing the refueling parking space also includes the waiting time information (there is a vehicle in front refueling, and the following vehicle needs to wait).

[0234] S3012, connecting the nodes according to the vehicle movement rules to generate the node planning diagram;

[0235] After the nodes are defined, the connections between them need to be defined. The nodes are connected according to the vehicle movement rules (the vehicle cannot move sideways or backward). The vehicle can only move between nodes according to the defined connections.

[0236] Embodiment 2

[0237] The embodiment of the present invention also provides a method for generating a virtual gas station scene, comprising:

[0238] The mapping module maps the pre-built scene and equipment 3D models and vehicle 3D models to the virtual gas station scene;

[0239] The processing module obtains the video images in the gas station and the equipment operation data of the gas station facilities and equipment, processes the equipment operation data to obtain equipment status information, processes the video images to obtain vehicle driving information, and maps the vehicle driving information and the equipment status information to the virtual gas station scene.

[0240] In an embodiment of the present invention, the virtual gas station scene generation method corresponds to the gas station digital twin system described in the above embodiment one. Its specific implementation process can refer to the process of applying the gas station digital twin system to generate a virtual gas station scene in the above embodiment one. The repeated parts will not be repeated here.

[0241] Embodiment 3

[0242] The embodiment of the present invention also provides a method for predicting vehicle flow at a gas station, comprising:

[0243] Obtain historical refueling data within a preset time period at a gas station, and input the historical refueling data into a pre-established traffic flow prediction model to predict the traffic flow; wherein the historical refueling data includes date, temperature, precipitation, snowfall, whether it is a discount, whether it is a holiday, and traffic flow, and the traffic flow prediction model is obtained by the following method:

[0244] Dividing the historical refueling data into a training set and a test set;

[0245] A multivariate linear regression model is used to establish the relationship between traffic volume and date, temperature, precipitation, snowfall, discount, and holiday. The model is expressed as:

[0246] h b (x)=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6+b7x7+e;

[0247] Among them, h b (x) is the traffic volume, x1 is the date of each month, x2 is the temperature, x3 is the precipitation, x4 is the snowfall, x5 is the oil price, x6 is whether it is a discount, x7 is whether it is a holiday, b0, b1, b2, b3, b4, b5, b6, b7 are model parameters, and e is the error term;

[0248] The training set is input into the multivariate linear regression model to obtain the model prediction result. The loss function is determined based on the model prediction result and the actual traffic flow by the following formula:

[0249]

[0250] Among them, J(b) is the loss function, n is the number of data in the training set, and h b (x (i) ) is the model prediction result, y (i) is the actual traffic volume;

[0251] Based on the loss function and the following formula, the model parameters are updated using the gradient descent method to obtain a trained model;

[0252]

[0253] Among them, α is the learning rate, m is the number of data in the training set, is the jth factor affecting traffic flow in the i-th group of data, b j is the jth model parameter, j=1.2....7;

[0254] The trained model is evaluated based on the test set, and the training process and the evaluation process are repeated until preset conditions are met to obtain a traffic flow prediction model.

[0255] In an embodiment of the present invention, the gas station traffic flow prediction method corresponds to the gas station digital twin system described in the above embodiment one. Its specific implementation process can refer to the process of using the gas station digital twin system to predict traffic flow in the above embodiment one. The repeated parts will not be repeated here.

[0256] Embodiment 4

[0257] The embodiment of the present invention also provides a method for planning a driving route of a gas station vehicle, comprising:

[0258] Divide the gas station into nodes according to the three-dimensional scene information of the gas station and generate a node planning map;

[0259] Based on the oil type of each refueling parking space, generate the oil type vector of each refueling space node;

[0260] Generate an oil type vector of the vehicle to be refueled based on the type of oil required by the vehicle to be refueled;

[0261] For the vehicle to be refueled, calculate the inner product of its oil type vector and the oil type vector of each refueling position node, and add the nodes with a multiplication result of 1 to the refueling position list;

[0262] Traverse each refueling station node in the refueling station list, take the station entrance as the starting point, take the refueling station node as the end point, calculate the first optimal trajectory from the starting point to the end point and the corresponding first trajectory cost information according to the node planning diagram; take this end point as the new starting point, determine the second optimal trajectory between the new starting point and the target node and the corresponding second trajectory cost information;

[0263] Determine the overall trajectory cost information based on the first trajectory cost information and the second trajectory cost information, and select the refueling position node with the smallest overall trajectory cost information from the refueling position list to obtain the target refueling position node;

[0264] The first optimal trajectory and the second optimal trajectory corresponding to the target refueling position node are combined to obtain the optimal driving route of the vehicle.

[0265] In an embodiment of the present invention, the gas station vehicle route planning method corresponds to the gas station digital twin system described in the above embodiment one. Its specific implementation process can refer to the process of planning the vehicle route using the gas station digital twin system in the above embodiment one. The repeated parts will not be repeated here.

[0266] In this document, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. The orientation or position relationship indicated by the terms "upper", "lower", etc. is based on the orientation or position relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0267] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict. The present invention is not limited to any single aspect, any single embodiment, or any combination and / or permutation of these aspects and / or embodiments. Each aspect and / or embodiment of the present invention can be used alone, or in combination with one or more other aspects and / or other embodiments.

[0268] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A gas station digital twin system, characterized in that: include: A mapping module and a processing module, wherein the mapping module is connected to the processing module; The mapping module is used to map the pre-built scene and equipment three-dimensional model and vehicle three-dimensional model to the virtual gas station scene; The processing module is used to obtain the video image in the gas station and the equipment operation data of the gas station facilities and equipment, process the equipment operation data to obtain equipment status information, process the video image to obtain vehicle driving information, and map the vehicle driving information and the equipment status information to the virtual gas station scene; Obtain historical refueling data of the gas station, analyze the historical refueling data, predict the number of refueling vehicles, refueling amount, average refueling time and fuel consumption forecast value in the future period, and predict the traffic flow; The optimal driving route of the vehicle is obtained by performing simulation based on the three-dimensional scene information of the virtual gas station scene.

2. The gas station digital twin system according to claim 1, characterized in that: The processing module includes a central control unit and a data processing unit, and the central control unit is connected to the data processing unit; The central control unit is used to send the received video image and the equipment operation data to the data processing unit; The data processing unit is used to process the device operation data to obtain device status information and process the video image to obtain vehicle driving information; Acquire the historical refueling data, analyze the historical refueling data, predict the number of refueling vehicles, refueling amount, average refueling time and fuel consumption forecast value in the future period, and predict the traffic flow; Furthermore, a simulation is performed based on the three-dimensional scene information of the virtual gas station scene to obtain the best driving route for the vehicle.

3. The gas station digital twin system according to claim 2, characterized in that: Also includes: The processing module also includes: a human-computer interaction unit connected to the central control unit; The central control unit is further used to send the device status information and the vehicle driving information to the human-computer interaction unit; The human-computer interaction unit is used to map the vehicle driving information and the device status information into the virtual gas station scene, and to display the virtual gas station scene rendered by a 3D engine in real time.

4. The gas station digital twin system according to claim 2, characterized in that: Also includes: An image acquisition unit connected to the central control unit; The image acquisition unit is used to acquire video images in the gas station and send the video images to the central control unit.

5. The gas station digital twin system according to claim 2, characterized in that: Also includes: A device control unit connected to the central control unit; The equipment control unit is used to collect equipment operation data of the gas station facilities and equipment, control and manage the gas station facilities and equipment, and send the equipment operation data to the central control unit.

6. The gas station digital twin system according to claim 2, characterized in that: Also includes: a data storage unit connected to the data processing unit; The data storage unit is used to store the historical refueling data, the equipment status information and the vehicle driving information; the historical refueling data includes date, time period, whether it is a holiday, oil price changes, weather conditions and refueling amount.

7. The gas station digital twin system according to claim 2, characterized in that: The data processing unit is specifically used for: The historical refueling data within a preset time period of the gas station is obtained, and the historical refueling data is input into a pre-established traffic flow prediction model to predict the traffic flow; wherein the historical refueling data includes date, temperature, precipitation, snowfall, whether it is a discount, whether it is a holiday and traffic flow, and the traffic flow prediction model is obtained by the following method: Dividing the historical refueling data into a training set and a test set; A multivariate linear regression model is used to establish the relationship between traffic volume and date, temperature, precipitation, snowfall, discount, and holiday. The model is expressed as: h b (x)=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6+b7x7+e; Among them, h b (x) is the traffic volume, x1 is the date of each month, x2 is the temperature, x3 is the precipitation, x4 is the snowfall, x5 is the oil price, x6 is whether it is a discount, x7 is whether it is a holiday, b0, b1, b2, b3, b4, b5, b6, b7 are model parameters, and e is the error term; The training set is input into the multivariate linear regression model to obtain the model prediction result. The loss function is determined based on the model prediction result and the actual traffic flow by the following formula: Among them, J(b) is the loss function, n is the number of data in the training set, and h b (x (i) ) is the model prediction result, y (i) is the actual traffic volume; Based on the loss function and the following formula, the model parameters are updated using the gradient descent method to obtain a trained model; Among them, α is the learning rate, m is the number of data in the training set, is the jth factor affecting traffic flow in the i-th group of data, b j is the jth model parameter, j=1.2....7; The trained model is evaluated based on the test set, and the training process and the evaluation process are repeated until preset conditions are met to obtain a traffic flow prediction model.

8. The gas station digital twin system according to claim 1, characterized in that: The three-dimensional scene information of the gas station based on the virtual gas station scene is simulated and deduced using the A* algorithm to obtain the best driving route for the vehicle, specifically including: Divide the gas station into nodes according to the three-dimensional scene information of the gas station and generate a node planning map; Based on the oil type of each refueling parking space, generate the oil type vector of each refueling space node; Generate an oil type vector of the vehicle to be refueled based on the type of oil required by the vehicle to be refueled; For the vehicle to be refueled, calculate the inner product of its oil type vector and the oil type vector of each refueling position node, and add the nodes with a multiplication result of 1 to the refueling position list; Traverse each refueling station node in the refueling station list, take the station entrance as the starting point, take the refueling station node as the end point, calculate the first optimal trajectory from the starting point to the end point and the corresponding first trajectory cost information according to the node planning diagram; take this end point as the new starting point, determine the second optimal trajectory between the new starting point and the target node and the corresponding second trajectory cost information; Determine the overall trajectory cost information based on the first trajectory cost information and the second trajectory cost information, and select the refueling position node with the smallest overall trajectory cost information from the refueling position list to obtain the target refueling position node; The first optimal trajectory and the second optimal trajectory corresponding to the target refueling position node are combined to obtain the optimal driving route of the vehicle.

9. The gas station digital twin system according to claim 8, characterized in that: The traversing each refueling position node in the refueling position list, taking the station entrance as the starting point and the refueling position node as the end point, and calculating the first optimal trajectory from the starting point to the end point and the corresponding first trajectory cost information according to the node planning graph specifically includes: Perform the following operations for each refueling position node in the refueling position list: Taking the entrance as the starting point, adding the starting point and the nodes adjacent to the starting point in the node planning graph to the open list, setting the starting point as the parent node of the nodes adjacent to the starting point, and adding the starting point to the closed list; Searching for a designated node that can be reached from the starting point from the open list, and saving the trajectory cost information of the designated node in the trajectory list; wherein the designated node selection condition is that the evaluation function value of the node is the smallest; wherein the evaluation function value is determined by the following formula: f1=g+h+t; Among them, f1 is the evaluation function value, g is the actual cost of moving from the starting point to the specified node, h is the distance from the specified node to the end point, and t is the waiting time information of the specified node; Using the specified node as the new starting point, repeat the search operation until the end point is added to the closed list; The trajectory cost information of all nodes in the trajectory list is added together to obtain the first trajectory cost information.

10. The gas station digital twin system according to claim 8, characterized in that: The three-dimensional scene information of the gas station includes an entrance, an exit, a refueling island, obstacles, lanes, and refueling parking spaces; the gas station is divided into nodes according to the three-dimensional scene information of the gas station to generate a node planning diagram, including: The station entrance, station exit, refueling island, obstacles, lanes, and refueling parking spaces are set as nodes; the station entrance is the starting point, the station exit is the target node, each node includes the location information of the node, and the refueling space node also includes the type of refueling available and the waiting time information; The nodes are connected according to the vehicle movement rules to generate the node planning diagram.

11. The gas station digital twin system according to claim 4, characterized in that: The device control unit comprises: Liquid level meter, used to collect oil level, water level and temperature through probe, and calculate oil volume and water volume by comparing with tank capacity table; Leak detection sensor, used to collect oil and water leakage information; Gas flow meter, used to collect gas velocity and flow data; Pressure sensor, used to collect gas pressure data; Oil and gas concentration sensor, used to collect gas concentration data.

12. The gas station digital twin system according to claim 2, characterized in that: The data processing unit is specifically used for: Detect and identify the video image taken by the image acquisition unit to obtain the vehicle's license plate information, vehicle model information and color information, track the vehicle's driving trajectory based on the license plate information, vehicle model information and color information, and map the vehicle's driving trajectory from the video to the virtual gas station scene.

13. The gas station digital twin system according to claim 2, characterized in that: The data processing unit is specifically used for: Using the PP-YOLOE method to perform vehicle detection on the video image to obtain vehicle boundary information; Based on the vehicle boundary information, a vehicle image of each vehicle is intercepted in the video image, and a PP-OCR method is used to detect and recognize the vehicle image to obtain license plate information; The PP-LCNet method is used to perform vehicle model and color recognition on the vehicle image to obtain vehicle model information and color information.

14. The gas station digital twin system according to claim 13, characterized in that: The data processing unit is further used for: Based on the vehicle boundary information, an observation-centered tracker OC-SORT is used to track the vehicle to obtain the image coordinates of the vehicle in the video image; The physical coordinates of the vehicle at the gas station are determined based on the image coordinates by the following formula: Among them, (u,v) represents the image coordinates, (X w ,Y w ,Z w ) represents the physical coordinate, Z c is the scale factor, (u0, v0) represents the center point of the image, dx and dy represent the physical size of the image pixel in the u-axis and v-axis directions, f represents the focal length of the image acquisition unit, R and T represent the rotation matrix and translation vector of the image acquisition unit in the world coordinate system respectively; Based on the mapping relationship between the world coordinate system of the gas station and the virtual coordinate system of the virtual gas station scene, the virtual position information corresponding to the physical coordinates is determined, and the vehicle is mapped to the virtual gas station based on the virtual position information.

15. The gas station digital twin system according to claim 3, characterized in that: The human-computer interaction unit comprises: A user input device connected to the gas station digital twin software platform, used to collect user control device instructions and send them to the gas station digital twin software platform; A gas station digital twin software platform, connected to a display device, for rendering the virtual gas station scene according to the vehicle driving information and the device status information, and for changing the viewing angle of the virtual gas station scene according to instructions of the user control device; Display device, used to display the virtual gas station scene rendered by the 3D engine, vehicle driving information and equipment status information of the gas station facilities and equipment.

16. The gas station digital twin system according to claim 2, characterized in that: The central control unit is also used to calculate the waiting time of the vehicle in the queue, the refueling time of the vehicle and the residence time of the vehicle in the gas station according to the vehicle driving information, and to issue an alarm message when the equipment status is abnormal or the oil tank is insufficient.

17. The gas station digital twin system according to claim 2, characterized in that: The data processing unit is also used to predict the time to apply for a tanker truck to replenish oil products based on the acquired information on the remaining oil volume in the oil tank.

18. A method for generating a virtual gas station scene, characterized in that: include: The mapping module maps the pre-built scene and equipment 3D models and vehicle 3D models to the virtual gas station scene; The processing module obtains the video images in the gas station and the equipment operation data of the gas station facilities and equipment, processes the equipment operation data to obtain equipment status information, processes the video images to obtain vehicle driving information, and maps the vehicle driving information and the equipment status information to the virtual gas station scene.

19. A method for predicting vehicle flow at a gas station, characterized in that: include: Obtain historical refueling data within a preset time period at a gas station, and input the historical refueling data into a pre-established traffic flow prediction model to predict the traffic flow; wherein the historical refueling data includes date, temperature, precipitation, snowfall, whether it is a discount, whether it is a holiday, and traffic flow, and the traffic flow prediction model is obtained by the following method: Dividing the historical refueling data into a training set and a test set; A multivariate linear regression model is used to establish the relationship between traffic volume and date, temperature, precipitation, snowfall, discount, and holiday. The model is expressed as: h b (x)=b0+b1x1+b2x2+b3x3+b4x4+b5x5+b6x6+b7x7+e; Among them, h b (x) is the traffic volume, x1 is the date of each month, x2 is the temperature, x3 is the precipitation, x4 is the snowfall, x5 is the oil price, x6 is whether it is a discount, x7 is whether it is a holiday, b0, b1, b2, b3, b4, b5, b6, b7 are model parameters, and e is the error term; The training set is input into the multivariate linear regression model to obtain the model prediction result. The loss function is determined based on the model prediction result and the actual traffic flow by the following formula: Among them, J(b) is the loss function, n is the number of data in the training set, and h b (x (i) ) is the model prediction result, y (i) is the actual traffic volume; Based on the loss function and the following formula, the model parameters are updated using the gradient descent method to obtain a trained model; Among them, α is the learning rate, m is the number of data in the training set, is the jth factor affecting traffic flow in the i-th group of data, b j is the jth model parameter, j=1.2....7; The trained model is evaluated based on the test set, and the training process and the evaluation process are repeated until preset conditions are met to obtain a traffic flow prediction model.

20. A method for planning a vehicle driving route at a gas station, characterized in that: include: Divide the gas station into nodes according to the three-dimensional scene information of the gas station and generate a node planning map; Based on the oil type of each refueling parking space, generate the oil type vector of each refueling space node; Generate an oil type vector of the vehicle to be refueled based on the type of oil required by the vehicle to be refueled; For the vehicle to be refueled, calculate the inner product of its oil type vector and the oil type vector of each refueling position node, and add the nodes with a multiplication result of 1 to the refueling position list; Traversing each refueling position node in the refueling position list, taking the station entrance as the starting point and the refueling position node as the end point, and calculating the first optimal trajectory from the starting point to the end point and the corresponding first trajectory cost information according to the node planning graph; Taking the end point as the new starting point, determining the second optimal trajectory between the new starting point and the target node and the corresponding second trajectory cost information; Determine the overall trajectory cost information based on the first trajectory cost information and the second trajectory cost information, and select the refueling position node with the smallest overall trajectory cost information from the refueling position list to obtain the target refueling position node; The first optimal trajectory and the second optimal trajectory corresponding to the target refueling position node are combined to obtain the optimal driving route of the vehicle.

Images