Mine intersection management and control method based on short-time travel time prediction

Abstract

The present application belongs to the technical field of intersection management and control, and specifically discloses a mine area intersection management and control method based on short-time travel time prediction, comprising the following steps: S1, vehicle-level short-time travel time prediction; S2, passing priority calculation; S3, intersection passing rule establishment; S4, mine area intersection management and control; has the following points: at the vehicle state level, the mine car is combined with the carrying state of the vehicle and the running direction of the vehicle. The more significant running characteristics of the vehicle realize accurate prediction of the travel time; combined with the vehicle dynamic performance, increase the importance weight of the heavy load vehicle, improve the driving continuity of the heavy load vehicle, reduce the cost loss caused by vehicle start and stop, and comprehensively consider the intersection configuration, take the passing time as the index to measure the passing efficiency of the vehicle at the intersection; at the same time, combined with the need of road characteristics and algorithm adaptability, set the phase by the import road, solve the problem of vehicle space conflict, and ensure the safety of passing.

Description

Technical Field

[0001] This invention relates to the field of intersection management and control technology, and more specifically, to a method for managing and controlling intersections in mining areas based on short-term travel time prediction. Background Technology

[0002] Intersections are nodes in road networks. Currently, right-of-way management schemes at intersections are mainly divided into right-of-way management in signalized intersection environments and right-of-way management in unsignalized intersection environments. Right-of-way management in signalized intersection environments mainly optimizes signal parameters such as signal timing, phase sequence, and phase difference to achieve intersection control. The controlled objects are mainly traffic facilities such as traffic lights, and this is also the control method chosen for most urban intersections. Unsignalized intersections mainly rely on setting yield rules to ensure vehicles consciously comply with them to achieve orderly passage. However, with the development of vehicle-to-everything (V2X) technology, some intersection management schemes based on vehicle communication and control technologies have emerged. These schemes mainly target intersections in connected environments, enabling efficient and orderly weaving through intersections by controlling vehicle trajectories and speeds.

[0003] The shortcomings of traditional intersection control methods in special scenarios such as mining areas are mainly reflected in the following aspects:

[0004] (1) The commonly used method of controlling intersections by traffic lights is difficult to implement in mining areas, mainly because most of the roads in mining areas are unstructured, with irregular road configurations that are prone to change, making it difficult to carry out large-scale construction of traffic lights. At the same time, the cost of adjusting for road changes is also high, making it difficult to adapt to the mining environment.

[0005] (2) The traffic environment in the mining area is quite different from that in the city. This is mainly reflected in the traffic composition and traffic priority. The vehicles at the intersections in the mining area are mainly mining trucks. Unlike the situation in the city where all social vehicles enjoy equal right of way, in the mining area, heavy-duty trucks often have a higher traffic priority than empty trucks because of the cost and safety of starting and stopping vehicles. This point is rarely considered in the existing intersection control schemes.

[0006] (3) Most existing signalless intersection control schemes rely on highly reliable communication capabilities and autonomous driving technology. The application scenarios are relatively idealized and often have good results in simulation experiments, but it is difficult to apply them in real-world scenarios, especially in environments with high operational safety requirements, such as mines. Summary of the Invention

[0007] To overcome the shortcomings of existing technologies, this invention aims to propose a routine management method applicable to intersections in mining areas. Considering the unique traffic and geometric characteristics of intersections in mining areas, and taking into account the feasibility of implementation as well as overall traffic efficiency and safety, this invention proposes a virtual signal control method for intersections that combines travel time prediction. On the one hand, by combining travel time prediction, the control accuracy at the vehicle level of the intersection is improved. On the other hand, by using virtual signals, i.e., by uniformly issuing right-of-way to vehicles at the intersection through the control center, the high safety of signal control can be utilized while avoiding the high cost of signal infrastructure construction. This overcomes the shortcomings of traditional traffic control schemes in mining areas and solves the problems of vehicle priority, control accuracy, and portability between different intersections in traditional schemes. It achieves the goal of improving traffic efficiency at intersections in mining areas while ensuring vehicle traffic safety, and also makes the control scheme universally applicable in the environment of variable, irregular, and unstructured roads.

[0008] In view of this, a first aspect of the present invention is to provide a method for managing mining area intersections based on short-time travel time prediction.

[0009] The first aspect of the present invention provides a method for controlling mining area intersections based on short-term travel time prediction, comprising the following steps: S1, short-term travel time prediction for each vehicle: constructing a long short-term memory neural network model, analyzing the historical travel data uploaded by each vehicle, extracting features consistent with mining area operation as model input and training the model, and predicting the travel time for each vehicle using the trained model; S2, traffic priority calculation: converting all empty and / or heavy vehicles at the intersection into standard vehicle numbers, and calculating the priority of the intersection right-of-way based on the travel time in step S1; S3, intersection traffic rule establishment: setting phases based on the priority in step S2, determining the priority between each phase and adjusting the phases, and delaying the right-of-way for non-queuing vehicles based on the predicted travel time, with each vehicle choosing to proceed or stop according to the right-of-way; S4, mining area intersection control: performing phase control and switching at the intersection according to the intersection traffic rules in step S3.

[0010] This invention provides a method for controlling mining area intersections based on short-time travel time prediction. It considers the unique traffic and geometric characteristics of mining area intersections. Specifically, right-of-way is uniformly issued to vehicles through a control center. On the one hand, by combining travel time prediction, the control accuracy at the vehicle level at the intersection is improved. On the other hand, by using virtual signals—that is, by uniformly issuing right-of-way to vehicles at the intersection through the control center—it can leverage the high security of signal control while avoiding the high cost of building signal infrastructure. This improves the traffic efficiency of mining area intersections while ensuring vehicle traffic safety. Furthermore, it ensures the universal applicability of the control scheme in variable, irregular, and unstructured road environments.

[0011] By combining vehicle power performance to increase the importance weight of heavy-duty vehicles, the driving continuity of heavy-duty vehicles is improved, and the cost loss caused by vehicle start-stop is reduced. At the same time, the intersection configuration is considered, and the passage time is used as an indicator to measure the passage efficiency of vehicles at the intersection. By combining these factors, the real-time calculation and dynamic conversion of vehicle passage priority are realized.

[0012] Phases are set up on an approach lane basis to resolve vehicle space conflicts, ensure traffic safety, and design a right-of-way allocation framework to ensure orderly and efficient vehicle passage at intersections.

[0013] In addition, the technical solutions provided by embodiments of the present invention may also have the following additional technical features:

[0014] In any of the above technical solutions, the model includes: an input layer for converting the historical travel data into a form readable by the model; an LSTM module for processing the model input containing multiple influencing factors and obtaining the features; and an output layer for receiving the obtained features and processing them to derive the travel time.

[0015] In this technical solution, the most important part of the entire model is the multivariate LSTM module. The basic LSTM module is composed of multiple memory neurons connected in chronological order. Each memory neuron has three gate structures: the input gate, the forget gate, and the output gate, which together control the updating and forgetting of information.

[0016] The "forget gate" allows recurrent neural networks to "forget" previously useless information, the "input gate" replenishes new memories, and the "output gate" produces the output at the current moment. A typical gate structure combines a bitwise multiplication operation with a sigmoid function. When the gate is closed (when the sigmoid neural network layer outputs 0), all information cannot pass through; when the gate is open (when the sigmoid neural network layer outputs 1), all information can pass through.

[0017] In any of the above technical solutions, the features include: vehicle status, mining area traffic status, and weather status; wherein, the vehicle status includes: transportation status and vehicle direction of travel.

[0018] In this technical solution, the characteristics of the mining area are analyzed, and features are extracted from three dimensions: vehicle status, mining area traffic status, and weather conditions. At the vehicle status level, the significant operating characteristics of mining cars, including their carrying status and direction of travel, are combined to achieve accurate prediction of travel time.

[0019] In any of the above technical solutions, the priority is calculated using the following formula: W i =Q i / T i Among them, W i Q represents the passage priority of the i-th approach lane. i This represents the queue length of the i-th approach lane, used to characterize the traffic state of the intersection, T. i This represents the average travel time of vehicles entering the intersection from the i-th approach lane, and is used to characterize the shape of the intersection.

[0020] In this technical solution, a priority value is determined based on the queue length and average travel time of each approach lane. The higher the priority value, the easier it is to be granted the right of way. The geometry of the intersection is characterized by the average travel time of vehicles through the intersection, and the traffic status of the intersection is characterized by the queue length.

[0021] In any of the above technical solutions, the queue length is calculated using the following formula: Among them, Q i N represents the queue length of the i-th entrance lane. j Let N represent the quantity of vehicle type j, N represent the type of vehicle, and α represent the quantity of vehicle type j. j This represents the weight coefficient of the j-th type of vehicle.

[0022] In this technical solution, the traffic status is characterized by the queue length of the approach lane. The queue length needs to be standardized, that is, a certain type of vehicle is used as the standard, and other vehicles are converted into the standard number of vehicles according to the vehicle's dynamic performance. Since the start-stop time of heavy-duty vehicles is often longer than that of empty vehicles, they will be equivalent to more vehicles in the conversion. The more heavy-duty vehicles there are, the longer the standard queue at the intersection will be.

[0023] In any of the above technical solutions, the weighting coefficient is calculated using the following formula: α j =Nr j / Nr0; where α j Represents the weighting coefficient of the j-th type of vehicle, Nr jThis indicates the specific power of the vehicle, and Nr0 indicates the specific power of the standard vehicle.

[0024] In this technical solution, the weighting coefficient of a vehicle depends entirely on its characteristics, namely its acceleration performance, maximum speed, and braking performance. These parameters represent the biggest differences between vehicles besides their geometric dimensions. The braking distances between existing vehicles are not significantly different, so the main differences lie in the vehicle's acceleration performance and maximum speed, i.e., its power performance. A comprehensive index—specific power—is used to describe the vehicle's power performance. When the specific power of two vehicles is equal, i.e., the ratio is 1, it can be considered that the two vehicles are approximately identical in performance except for their geometric dimensions.

[0025] In any of the above technical solutions, the average travel time is calculated using the following formula: Among them, T i t represents the average travel time t takes for vehicles entering the intersection from the i-th approach lane. m N represents the travel time of the m-th path at the i-th entrance, and N represents the travel time of the m-th path at the i-th entrance. m This represents the number of vehicles traveling from the i-th entrance to the m-th path.

[0026] In this technical solution, the average travel time is measured in units of the approach lanes of the intersection. It is the average time required for vehicles waiting to pass through the intersection in each approach lane. The longer the average travel time, the longer the vehicles in that approach lane will occupy the intersection when they are granted the right of way, which in turn increases the waiting time for vehicles in other approach lanes. Therefore, the approach lane with the longer the average travel time will have a lower traffic priority.

[0027] In any of the above technical solutions, the phase control and conversion steps are specifically as follows: E1, detect and determine whether there is a queue at the current intersection; if there is no queue, proceed to E2; if there is, proceed to E3; E2, allocate right-of-way according to the vehicle distance on a first-come, first-served basis, and return to step E1; E3, calculate the priority of the right-of-way for each phase and issue the right-of-way; E4, determine whether any subsequent vehicles will arrive at the phase with the right-of-way within the time period; if not, return to step E1; if so, proceed to step E5; E5, determine whether the current passage time of the phase with the right-of-way has reached the upper limit; if so, return to step E1; otherwise, proceed to step E6; E6, extend the time period of the phase with the right-of-way by the corresponding number of vehicles, and return to step E4.

[0028] In this technical solution, the specific management of the intersection needs to be adjusted according to the current traffic conditions of the intersection. If there are few vehicles and no queue at the intersection, the right of way is issued according to the principle of "first come, first served". If there are many vehicles and queues occur at the intersection, priority calculation and management are required.

[0029] Traffic rights are assigned based on priority. To reduce the number of times vehicles stop and improve traffic efficiency, the right-of-way needs to be extended appropriately based on the time interval between vehicle arrivals (travel time prediction). At the same time, there is an upper limit to the right-of-way for each phase to prevent a certain phase from continuously obtaining the right-of-way.

[0030] In any of the above technical solutions, the time period is 4s-8s, and the upper limit is 40s-50s.

[0031] In this technical solution, the time period is the maximum time for the right-of-way of the current approach lane to be empty, which is the time for 1-2 vehicles to pass through the intersection. Based on experience, it can be set to 4s-8s. The upper limit is set to avoid dangerous situations such as vehicles in other phases waiting for too long and running into the intersection. That is, the driver's tolerance for stopping. This value is determined based on experience and is set to a range of 40s-50s.

[0032] The beneficial effects of this invention compared to the prior art are as follows:

[0033] This paper proposes an LSTM-based method for travel time prediction. It analyzes the characteristics of the mining area and extracts features from three dimensions: vehicle status, mining area traffic status, and weather conditions. At the vehicle status level, it combines the vehicle's carrying status and the more significant operating characteristics of mining cars in the direction of vehicle movement to achieve accurate travel time prediction.

[0034] It can effectively solve the problem of vehicle traffic priority in mining areas. By reconstructing queue length through vehicle standardization, it increases the importance weight of heavy-duty vehicles by combining vehicle power performance, thereby improving the driving continuity of heavy-duty vehicles and reducing the cost losses caused by vehicle start-stop. At the same time, it comprehensively considers the intersection configuration and uses the passage time as an indicator to measure the passage efficiency of vehicles at the intersection, realizing the real-time calculation and dynamic conversion of vehicle traffic priority.

[0035] By analyzing phase conflicts, the conflicting directions of vehicles are separated. In addition, considering road characteristics and the need for algorithm adaptability, phases are set on the approach lanes to solve the problem of vehicle spatial conflicts, ensure traffic safety, and integrate inductive signal control logic to design a right-of-way allocation framework to ensure the orderly and efficient passage of vehicles at intersections.

[0036] Additional aspects and advantages of embodiments of the invention will become apparent in the following description or may be learned by practice of embodiments of the invention. Attached Figure Description

[0037] The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of the invention.

[0038] Figure 1 This is a flowchart of the present invention;

[0039] Figure 2 This is a schematic diagram of an irregular intersection in a mining area according to the present invention. Detailed Implementation

[0040] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0041] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0042] Please see Figure 1-2 The first aspect of the present invention provides a method for controlling mining area intersections based on short-time travel time prediction, comprising the following steps:

[0043] The first step is vehicle-level short-term travel time prediction: First, vehicle driving data is collected through onboard equipment to form a historical database. The data is then analyzed to mine the characteristics of the historical data and extract features that are consistent with the operating characteristics of the mining area. A long short-term memory neural network (LSTM) is built and trained using historical data until the model's prediction results meet the requirements, that is, accurate prediction of vehicle-level travel time is achieved.

[0044] Specifically, the method for vehicle-level short-time travel time prediction is as follows:

[0045] The entire process is divided into two parts: extracting the factors affecting travel time and constructing the LSTM neural network.

[0046] Based on the operational characteristics of the mining area, the analysis of factors affecting travel time is mainly conducted from three perspectives: vehicle-related factors, road traffic factors, and weather factors. The specific analysis is shown in the table below:

[0047]

[0048]

[0049] An LSTM network model is constructed, which consists of three parts: the input layer is responsible for transforming the preprocessed data into a form that the model can read; the LSTM module is mainly responsible for processing the input data containing multiple influencing factors and obtaining feature information; the output layer receives the obtained feature information, processes it, and obtains the final travel time prediction result.

[0050] The most crucial part of the entire model is the multivariate LSTM module. A basic LSTM module consists of multiple memory neurons connected sequentially over time. Each memory neuron contains three gates: an input gate, a forget gate, and an output gate, which collectively control the updating and forgetting of information. The "forget gate" allows the recurrent neural network to "forget" previously useless information, the "input gate" replenishes new memories, and the "output gate" generates the output for the current time step. The typical gate structure combines a bitwise multiplication operation with a sigmoid function. When the gate is closed (when the sigmoid neural network layer output is 0), all information cannot pass through; when the gate is open (when the sigmoid neural network layer output is 1), all information can pass through.

[0051] The network model is trained using historical driving data from the mining area, and the resulting model that meets the usage requirements is saved and can be used later.

[0052] The second step is to calculate the traffic priority: by using the location and status information uploaded by the vehicles, the number of vehicles of each type at the intersection is counted. Combined with the relevant calculation of the vehicle conversion factor, the empty and heavy vehicles at the intersection are converted into the standard number of vehicles to achieve the standardization of the queue length at the intersection. Then, combined with historical driving data, the average passage time of vehicles in the intersection is calculated, and the traffic priority of the intersection is calculated in a comprehensive manner.

[0053] Specifically, traffic priority is measured from two dimensions: intersection traffic status and intersection shape. Traffic status is characterized by the queue length at the approach lanes. This queue length needs to be standardized, meaning a specific type of vehicle is used as the standard, and other vehicles are converted into a standard number based on their dynamic performance. Since heavy-duty vehicles often have longer start-stop times than empty vehicles, they will be equivalent to a larger number of vehicles in the calculation. The more heavy-duty vehicles there are, the longer the standard queue at the intersection will be. The specific calculation formula is as follows:

[0054] α j =Nr j / Nr0(1)

[0055]

[0056] Where, α j Nr represents the weighting coefficient of the j-th type of vehicle. j This represents the specific power of the vehicle, Nr0 represents the specific power of the standard vehicle, and Q... i N represents the standard queue length of the i-th entrance lane. j Let N represent the actual number of vehicles of type j, and N represent the specific vehicle type.

[0057] On the other hand, the geometry of an intersection is characterized by the average travel time of vehicles through the intersection, specifically calculated as follows:

[0058]

[0059] Among them, T i Let t represent the average travel time of vehicles entering the intersection from the i-th approach lane. m N represents the travel time of the m-th path at the i-th entrance, where N is the travel time. m This represents the number of vehicles traveling from the i-th entrance to the m-th path.

[0060] Finally, a priority value is determined based on the queue length and average transit time of each entrance lane. The higher the priority value, the easier it is to be granted the right of way. The specific calculation formula is as follows:

[0061] W i =Q i / T i (4)

[0062] Among them, W i This indicates the passage priority of the i-th entrance lane.

[0063] The third step is to establish intersection traffic rules: Based on the priority calculated in the previous step, the proposed phase priority control algorithm is used to determine the priority between each phase and flexibly and dynamically adjust the phases. This allows for rapid switching of right-of-way according to traffic demand. At the same time, the right-of-way is appropriately extended for non-queuing vehicles based on the travel time prediction results. The right-of-way is uniformly issued to vehicles through the control center, and vehicles choose to drive or stop according to the right-of-way.

[0064] Specifically, the management of intersections needs to be adjusted according to the current traffic conditions. If there are few vehicles and no queues, the right of way is issued according to the "first-come, first-served" principle. If there are many vehicles and queues have formed, priority calculation and management are required. The right of way is then issued according to the priority. It should also be noted that in order to reduce the number of times vehicles stop and improve traffic efficiency, the right of way should be extended appropriately based on the time interval between vehicle arrivals (travel time prediction). In addition, there is an upper limit to the right of way for each phase to prevent a certain phase from having the right of way all the time.

[0065] The fourth step is to perform phase control and switching at the intersection according to the intersection traffic rules;

[0066] Specifically, the phase control and conversion steps in the fourth step are as follows:

[0067] Step 1: Initialize and check the queue length at the current intersection. If there is no queue, proceed to Step 2; otherwise, proceed to Step 3.

[0068] Step 2: Allocate passage rights according to the "first-come, first-served" principle based on vehicle distance, and then return to Step 1;

[0069] Step 3: Calculate the priority right-of-way for each phase and issue the right-of-way;

[0070] Step 4: Determine whether any subsequent vehicles arrive within a certain time interval, 4s-8s. If not, proceed to Step 1; if so, continue to Step 5. The preferred time interval is 5s.

[0071] Step 5: Determine if the current passage time has reached the upper limit, 40s-50s. If yes, return to Step 1; otherwise, continue to Step 6. The preferred upper limit is 40s.

[0072] Step 6: Extend the current phase's passage time to allow subsequent vehicles to pass, and then return to Step 4.

[0073] In the description of this invention, it should be understood that the terms "longitudinal", "lateral", "up", "down", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this invention, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0074] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for controlling intersections in mining areas based on short-time travel time prediction, characterized in that, Includes the following steps: S1, Vehicle-level short-term travel time prediction: Construct a long short-term memory neural network model, analyze the historical travel data uploaded by the vehicle, extract features that conform to the operation of the mining area as model input and train the model, and use the trained model to predict the travel time of the vehicle. S2, Priority calculation: Convert all empty and / or heavy vehicles at the intersection into standard vehicle numbers, and combine them with the travel time mentioned in step S1 to calculate the priority of the vehicle's right of way at the current intersection. S3, intersection traffic rules establishment: Based on the priority mentioned in step S2, phases are set up on the approach lanes, the priority between each phase is determined and the phases are adjusted. Combined with the predicted travel time, the right of way for non-queuing vehicles is delayed, and each vehicle chooses to drive or stop according to the right of way. S4, Mining Area Intersection Control: Based on the intersection traffic rules in step S3, phase control and conversion are performed on the intersection; The priority is calculated using the following formula: ； in, Indicates the passage priority of the i-th entrance lane. This represents the queue length of the i-th approach lane, used to characterize the traffic state of the intersection. This represents the average travel time of vehicles entering the intersection from the i-th approach lane, and is used to characterize the shape of the intersection; The queue length is calculated using the following formula: ； in, Indicates the queue length of the i-th entrance lane. Indicates the quantity of the j-th type of vehicle, Indicates the type of vehicle, Indicates the first j Weighting coefficients for different types of vehicles; The weighting coefficients are calculated using the following formula: ； in, Indicates the first j Weighting coefficients for different types of vehicles Indicates the first j Power-to-weight ratio of the vehicle This indicates the specific power of a standard vehicle; The average travel time is calculated using the following formula: ； in, This represents the average travel time of vehicles entering the intersection from the i-th approach lane. This represents the travel time of the m-th path at the i-th entrance. This represents the number of vehicles traveling from the i-th entrance to the m-th path; The phase control and conversion steps are as follows: E1: Detect and determine if there is a queue at the current intersection. If there is no queue, proceed to E2; otherwise, proceed to E3. E2: Allocate passage rights according to the first-come, first-served principle based on vehicle distance, and return to step E1; E3, calculate the priority of the right of way for each phase and issue the right of way; E4. Determine whether any subsequent vehicles arrive at each phase within the time period. If not, return to step E1; otherwise, proceed to step E5. E5, determine whether the current passage time of the phase with the right of way has reached the upper limit. If yes, return to step E1; otherwise, proceed to step E6. E6, extend the time period for the phase with right-of-way by the corresponding number of vehicles, and return to step E4.

2. The method for controlling mining area intersections based on short-time travel time prediction according to claim 1, characterized in that, The model includes: The input layer is used to transform the historical travel data into a form that the model can read. The LSTM module is used to process the model input, which contains multiple influencing factors, and to obtain the features; The output layer is used to receive the acquired features and process them to derive the travel time.

3. The method for controlling mining area intersections based on short-time travel time prediction according to claim 1, characterized in that, The features include: vehicle status, mine area traffic status, and weather status; The vehicle status includes: the carrying status and the vehicle's direction of travel.

4. The method for controlling mining area intersections based on short-time travel time prediction according to claim 3, characterized in that, The time period is 4s-8s, and the upper limit is 40s-50s.

Images