A signal control method and system for intersection emergency vehicles
By constructing a vehicle state matrix using a deep reinforcement learning model and a discretized model, and optimizing traffic signal phase switching, the problem of low efficiency of emergency vehicles at intersections is solved, enabling emergency vehicles to pass quickly without affecting the normal passage of other vehicles.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- WUHAN UNIV OF TECH
- Filing Date
- 2023-07-11
- Publication Date
- 2026-07-21
AI Technical Summary
The existing traffic signal system cannot effectively guarantee the priority passage of emergency vehicles at intersections, especially during peak hours, resulting in low efficiency of emergency vehicle passage and affecting the efficiency of handling emergencies.
By employing a deep reinforcement learning model combined with a discretization model, the state matrix of emergency vehicles and social vehicles is constructed by acquiring the state information of vehicles in the target area. The reward function is used to evaluate the traffic demand and optimize the traffic signal phase switching to ensure the rapid passage of emergency vehicles.
This effectively reduces queuing at intersections, ensures rapid passage for emergency vehicles, minimizes the impact on normal traffic flow, and improves the efficiency of emergency vehicles in complex traffic conditions.
Smart Images
Figure CN116935674B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent transportation technology, specifically to a signal control method and system for emergency vehicles at intersections. Background Technology
[0002] The destructive nature of urban emergencies and their serious threat to public life and property are urging cities to build and improve their urban emergency response and management systems. The traffic efficiency of emergency vehicles such as fire trucks, ambulances, and rescue vehicles directly impacts the efficiency of emergency response; rapid arrival at the scene effectively controls losses caused by emergencies. However, most cities are currently plagued by traffic congestion, especially during peak hours when traffic paralysis is commonplace. The current road traffic environment cannot guarantee priority passage for emergency vehicles, making unimpeded and efficient passage difficult to achieve.
[0003] Traditional adaptive traffic light control typically adjusts signals by sensing the number of vehicles on the road using road sensors. Therefore, it cannot guarantee smooth passage for emergency vehicles at the intersection and cannot meet their urgent passage needs.
[0004] Therefore, optimizing signal control when emergency vehicles pass through intersections to meet their urgent passage needs has become a pressing issue for technical personnel. Summary of the Invention
[0005] In view of this, it is necessary to provide a signal control method and system for emergency vehicles at intersections to meet the emergency passage needs of emergency vehicles.
[0006] To achieve the above objectives, the present invention provides a signal control method for emergency vehicles at intersections, comprising:
[0007] Acquire the status information of vehicles within the target area, including ordinary vehicles and emergency vehicles, and the status information includes speed information and location information;
[0008] The vehicle's state information is processed based on a preset discretization model to obtain the vehicle's state matrix, which includes the state matrix of ordinary social vehicles and the state matrix of the emergency vehicle.
[0009] When a traffic signal phase is input, the state matrix of the ordinary social vehicles and the state matrix of the emergency vehicles are judged based on a preset deep reinforcement learning model to obtain the judgment result.
[0010] The signal phase of the traffic signal at the next moment is determined based on the judgment result.
[0011] In one possible implementation, the state information of the vehicle is processed based on a preset discretization model to obtain a state matrix of the vehicle. This state matrix includes a state matrix for ordinary social vehicles and a state matrix for the emergency vehicle, comprising:
[0012] Discretize the vehicles within the target area to obtain multiple cells containing the vehicles;
[0013] Based on the discretization model, the state information of the vehicle in each cell is transformed into a state matrix, which includes the state matrix of ordinary social vehicles and the state matrix of the emergency vehicle.
[0014] The general social vehicle status matrix includes a general social vehicle speed information matrix and a general social vehicle location information matrix, and the emergency vehicle status matrix includes an emergency vehicle speed information matrix and an emergency vehicle location information matrix.
[0015] In one possible implementation, the DTSE method is used to discretize and model vehicles within the target area.
[0016] In one possible implementation, the expression for processing the state information of the ordinary social vehicles is:
[0017]
[0018] V i，k =V avg ;
[0019] Among them, P i，k V represents the location information of a regular vehicle in the k-th cell of the i-th lane at an intersection. i，k V represents the speed information of ordinary vehicles in the k-th cell of the i-th lane at the intersection. avg This represents the average speed of the vehicles within the cell.
[0020] In one possible implementation, the state matrix of the emergency vehicle is:
[0021] E = [d 1，1 v 1，1 d 1，2 v 1，2 ...d i，j v i，j ] T ;
[0022] The state matrix of ordinary social vehicles is as follows:
[0023] N = [P1, 1, V] 1，1 P 1，2 V1，2 ...P i，k V i，k ] T ;
[0024] Where E represents the state matrix of the emergency vehicle, d i，j v represents the distance of the j-th emergency vehicle in the i-th lane from the intersection. i，j Let be the speed of the j-th emergency vehicle in the i-th lane at the intersection.
[0025] In one possible implementation, the state space matrix of the intersection is expressed as:
[0026] s = {E, N};
[0027] Where s represents the state space matrix of the intersection.
[0028] In one possible implementation, the judgment based on a preset judgment model on the state matrix of ordinary social vehicles and the state matrix of emergency vehicles to obtain a judgment result includes:
[0029] When a signal phase is input into the traffic signal system, the state matrices of ordinary vehicles and emergency vehicles are judged based on a preset reward function model to obtain the reward function values of emergency vehicles and ordinary vehicles.
[0030] The reward function values of the emergency vehicles and ordinary vehicles are fed back into the deep reinforcement learning model to obtain the judgment result of the current traffic signal.
[0031] In one possible implementation, the formula for calculating the reward value in the preset reward function model is:
[0032]
[0033] Where 'a' represents the reward coefficient for emergency vehicles, and 'b' represents the reward coefficient for other vehicles. The reward function represents the emergency vehicle. This represents the reward function for social vehicles.
[0034] In one possible implementation, the formula for calculating the reward function value of the emergency vehicle is:
[0035]
[0036] The formula for calculating the reward function value of ordinary social vehicles in the preset reward function model is as follows:
[0037]
[0038] Among them, V E，j V represents the speed of the j-th emergency vehicle. max This represents the maximum value of the emergency vehicle, where m1, k1, m2, and k2 are preset constants, and Q... i C represents the queue length on lane i. max This indicates that the queue of vehicles has reached the maximum capacity of the lane.
[0039] To achieve the above objectives, the present invention also provides a signal control system for emergency vehicles at intersections, comprising:
[0040] The status information module is used to acquire the status information of vehicles in the target area, including ordinary social vehicles and emergency vehicles, and the status information includes speed information and location information.
[0041] The state matrix module is used to process the state information of the vehicle based on a preset discretization model to obtain the state matrix of the vehicle. The state matrix includes the state matrix of ordinary social vehicles and the state matrix of the emergency vehicle.
[0042] The judgment module is used to judge the state matrix of ordinary social vehicles and the state matrix of emergency vehicles based on a preset deep reinforcement learning model when a traffic signal inputs a signal phase, and obtain the judgment result.
[0043] The traffic signal module is used to determine the signal phase of the traffic signal at the next moment based on the judgment result.
[0044] The beneficial effects of the above embodiments are: acquiring the state information of vehicles within the target area; processing the vehicle state information based on a preset discretization model to obtain the vehicle state matrix; when a traffic signal inputs a signal phase, judging the state matrices of ordinary social vehicles and emergency vehicles based on a preset deep reinforcement learning model to obtain a judgment result; and determining the signal phase of the traffic signal at the next moment based on the judgment result. This invention utilizes deep reinforcement learning for traffic light control, which can better adapt to complex traffic conditions, effectively reduce intersection queuing, and coordinate the passage of emergency vehicles and social vehicles when emergency vehicles pass through intersections, ensuring the rapid passage of emergency vehicles without seriously affecting the normal driving of social vehicles. Attached Figure Description
[0045] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0046] Figure 1 A flowchart illustrating an embodiment of a signal control method for emergency vehicles at intersections provided by the present invention;
[0047] Figure 2 A schematic diagram of the signal phase of an embodiment of a signal control method for emergency vehicles at intersections provided by the present invention;
[0048] Figure 3 A schematic diagram of the algorithm framework of an embodiment of the signal control method for emergency vehicles at intersections provided by the present invention;
[0049] Figure 4 This is a schematic diagram of an embodiment of a signal control system for emergency vehicles at intersections provided by the present invention. Detailed Implementation
[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.
[0051] Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor systems and / or microcontroller systems.
[0052] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.
[0053] The present invention provides a signal control method and system for emergency vehicles at intersections, which will be described below.
[0054] Figure 1 This is a flowchart illustrating an embodiment of a signal control method for emergency vehicles at intersections provided by the present invention.
[0055] Reference Figure 1 This invention provides a signal control method for emergency vehicles at intersections, comprising:
[0056] S101. Obtain the status information of vehicles in the target area, including ordinary social vehicles and emergency vehicles, and the status information includes speed information and location information;
[0057] S102. The state information of the vehicle is processed based on a preset discretization model to obtain the state matrix of the vehicle. The state matrix includes the state matrix of ordinary social vehicles and the state matrix of the emergency vehicle.
[0058] S103. When a traffic signal inputs a signal phase, the state matrix of the ordinary social vehicles and the state matrix of the emergency vehicles are judged based on a preset deep reinforcement learning model to obtain the judgment result.
[0059] S104. Determine the signal phase of the traffic signal at the next moment based on the judgment result.
[0060] The beneficial effects of the above embodiments are: acquiring the state information of vehicles within the target area; processing the vehicle state information based on a preset discretization model to obtain the vehicle state matrix; when a traffic signal inputs a signal phase, judging the state matrices of ordinary social vehicles and emergency vehicles based on a preset deep reinforcement learning model to obtain a judgment result; and determining the signal phase of the traffic signal at the next moment based on the judgment result. This invention utilizes deep reinforcement learning for traffic light control, which can better adapt to complex traffic conditions, effectively reduce intersection queuing, and coordinate the passage of emergency vehicles and social vehicles when emergency vehicles pass through intersections, ensuring the rapid passage of emergency vehicles without seriously affecting the normal driving of social vehicles.
[0061] Specifically, the target area refers to a designated area within the preset range of the intersection. Taking a crossroads as an example, the target area refers to the area with the crossroads as the center and a preset distance as the radius. The specific preset radius can be flexibly set according to the actual situation.
[0062] In one embodiment, the state information of the vehicle is processed based on a preset discretization model to obtain a state matrix of the vehicle. The state matrix includes a state matrix of ordinary social vehicles and a state matrix of the emergency vehicle, including:
[0063] Discretize the vehicles within the target area to obtain multiple cells containing the vehicles;
[0064] Based on the discretization model, the state information of the vehicle in each cell is transformed into a state matrix, which includes the state matrix of ordinary social vehicles and the state matrix of the emergency vehicle.
[0065] The general social vehicle status matrix includes a general social vehicle speed information matrix and a general social vehicle location information matrix, and the emergency vehicle status matrix includes an emergency vehicle speed information matrix and an emergency vehicle location information matrix.
[0066] Furthermore, let s represent the state space of emergency vehicles and ordinary vehicles, E represent the state space of the emergency lane, and N represent the state space of ordinary vehicles. The speed of each emergency vehicle and the distance between them and their intersection positions are used as the state space matrix of the emergency vehicles, and the speed of each ordinary vehicle and the distance between them and their intersection positions are used as the state space matrix of the emergency vehicles.
[0067] In one embodiment, the DTSE method is used to discretize and model vehicles within the target area.
[0068] Specifically, the expression for processing the state information of the ordinary social vehicles is as follows:
[0069]
[0070] V i，k =V Avg ;
[0071] Among them, P i，k V represents the location information of a regular vehicle in the k-th cell of the i-th lane at an intersection. i，k V represents the speed information of ordinary vehicles in the k-th cell of the i-th lane at the intersection. avg This represents the average speed of the vehicles within the cell.
[0072] Furthermore, the state matrix of the emergency vehicle is as follows:
[0073] E = [d 1，1 v 1，1 d 1，2 v 1，2 ...d i，j v i，j ] T ;
[0074] The state matrix of ordinary social vehicles is as follows:
[0075] N = [P] 1，1 V 1，1 P 1，2 V 1，2 ...P i，k V i，k ] T ;
[0076] Where E represents the state matrix of the emergency vehicle, di，j v represents the distance of the j-th emergency vehicle in the i-th lane from the intersection. i，j Let be the speed of the j-th emergency vehicle in the i-th lane at the intersection.
[0077] The state space matrix s of the intersection consists of the state matrix E of emergency vehicles and the state matrix N of ordinary vehicles. The expression of the state space matrix of the intersection is:
[0078] s = {E, N};
[0079] Where s represents the state space matrix of the intersection.
[0080] Figure 2 This is a schematic diagram of the signal phase of an embodiment of a signal control method for emergency vehicles at intersections provided by the present invention.
[0081] Generally, there are two methods for controlling traffic signals: fixed phase sequence control and variable phase sequence control. In actual traffic control systems, most traffic lights switch according to a fixed phase sequence to prevent traffic chaos; therefore, this system also uses a fixed phase sequence for control. This system adopts the following... Figure 2 The four-phase signal control scheme shown switches phases in the following order: East-West (EW), East-West Left Turn (EWL), North-South Straight (NS), and North-South Left Turn (NSL).
[0082] For each time step t, based on the preset deep reinforcement learning model, it is determined whether the signal phase needs to be switched according to the current state matrix, and the output action at∈A is: if at=1, the traffic light switches to the next phase; otherwise, if at=0, the current signal state remains unchanged.
[0083] The specific conditional expression is:
[0084] A = {1, 0};
[0085] Table A outputs the judgment action.
[0086] In one embodiment, the judgment based on a preset judgment model on the state matrix of ordinary social vehicles and the state matrix of emergency vehicles to obtain a judgment result includes:
[0087] When a signal phase is input into the traffic signal system, the state matrices of ordinary vehicles and emergency vehicles are judged based on a preset reward function model to obtain the reward function values of emergency vehicles and ordinary vehicles.
[0088] The reward function values of the emergency vehicles and ordinary vehicles are fed back into the deep reinforcement learning model to obtain the judgment result of the current traffic signal.
[0089] Specifically, the formula for calculating the reward value in the preset reward function model is as follows:
[0090]
[0091] Where 'a' represents the reward coefficient for emergency vehicles, and 'b' represents the reward coefficient for other vehicles. The reward function represents the emergency vehicle. This represents the reward function for social vehicles.
[0092] The deep reinforcement learning model judges the merits of the signal phase strategy adopted by comparing the reward function values of emergency vehicles and ordinary social vehicles, thereby continuously optimizing the deep reinforcement learning model. In this invention, the reward functions of emergency vehicles and social vehicles are calculated separately and then linearly combined, as shown in the above calculation formula.
[0093] It should be explained that the present invention adopts a normalized pressure calculation method for the reward function of both emergency vehicles and social vehicles. Emergency vehicles use their driving speed as the evaluation index, while social vehicles use queuing as their evaluation index.
[0094] The formula for calculating the normalized pressure value of emergency vehicles is:
[0095]
[0096] The formula for calculating the normalized pressure value of ordinary social vehicles is:
[0097]
[0098] ρ E，j ∈[0,1];
[0099] ρ N，i ∈[0,1];
[0100] Where, ρ E，j For the normalized pressure of the j-th emergency vehicle, v E，j The speed of the j-th emergency vehicle, ρ N，i For lane normalization pressure, Q i Let m1, k1, m2, and k2 be the queue length on lane i, where m1, k1, m2, and k2 are all preset constants.
[0101] When the emergency vehicle speed is 0, the normalized pressure value is at most 1; when the emergency vehicle speed reaches V... max At that time, the normalized pressure value is at its minimum of 0, resulting in... When the lane queue length is 0, the normalized pressure value is 0; when the vehicle queue reaches the lane's maximum capacity C... max When the normalized pressure value is 1, we get
[0102] Substituting ω1 into the formula for calculating the normalized pressure value of emergency vehicles, we get:
[0103]
[0104] Substituting ω2 into the formula for calculating the normalized pressure value of emergency vehicles, we get:
[0105]
[0106] The faster the emergency vehicle travels, the smaller the normalized pressure value, and the greater the reward value of the intelligent algorithm; conversely, the slower the vehicle travels, the smaller the reward value of the intelligent algorithm.
[0107] The formula for calculating the reward function value of the emergency vehicle is as follows:
[0108]
[0109] The shorter the queue length, the lower the normalized pressure value, and the greater the reward value of the intelligent algorithm; conversely, the longer the queue length, the smaller the reward value of the intelligent algorithm.
[0110] The formula for calculating the reward function value of ordinary social vehicles in the preset reward function model is as follows:
[0111]
[0112] Among them, V E，j V represents the speed of the j-th emergency vehicle. max This represents the maximum value of emergency vehicles, where m1 and k1 are preset constants, and Q... i C represents the queue length on lane i. max This indicates that the queue of vehicles has reached the maximum capacity of the lane.
[0113] Figure 3 This is a schematic diagram of the algorithm framework of an embodiment of the signal control method for emergency vehicles at intersections provided by the present invention.
[0114] In traditional deep Q-learning algorithms, action selection and evaluation are coupled, using the same network to select actions and evaluate state-action. That is, the target network selects each possible action based on the environment state and calculates the Q value, using a greedy strategy to select the maximum Q value. This may lead to overestimation of the Q value, and the final learning result is not optimal.
[0115] Reference Figure 3To improve the efficiency and effectiveness of deep reinforcement learning algorithms in addressing the problem that traditional inductive intersection signal control cannot adequately meet the emergency passage needs of emergency vehicles, the DDNQ and Prioritized ReplayDQN algorithms are combined.
[0116] Among them, s t Let a represent the state matrix. t Indicates the selection action, r t Let θ represent the reward function, θ represent the network parameters, and e represent the network parameters. i Indicates the current training batch, Y t DDNQ The target value is represented by Sum Tree, Q-Network is represented by Q-Network, and TargetNetwork is represented by the target network.
[0117] Figure 4 This is a schematic diagram of an embodiment of a signal control system for emergency vehicles at intersections provided by the present invention.
[0118] Reference Figure 4 The present invention also provides a signal control system for emergency vehicles at intersections, comprising:
[0119] The status information module 401 is used to acquire the status information of vehicles in the target area road, including ordinary social vehicles and emergency vehicles, and the status information includes speed information and location information.
[0120] The state matrix module 402 is used to process the state information of the vehicle based on a preset discretization model to obtain the state matrix of the vehicle. The state matrix includes the state matrix of ordinary social vehicles and the state matrix of the emergency vehicle.
[0121] The judgment module 403 is used to judge the state matrix of the ordinary social vehicles and the state matrix of the emergency vehicles based on a preset deep reinforcement learning model when a traffic signal inputs a signal phase, and obtain the judgment result.
[0122] Traffic signal module 404 is used to determine the signal phase of the traffic signal at the next moment based on the judgment result.
[0123] The beneficial effects of the above embodiments are as follows: the state information module 401 acquires the state information of vehicles within the target area road; the state matrix module 402 processes the vehicle state information based on a preset discretization model to obtain the vehicle state matrix; the judgment module 403 judges the state matrices of ordinary social vehicles and emergency vehicles based on a preset deep reinforcement learning model when a traffic signal inputs a signal phase, and obtains the judgment result; the traffic signal module 404 determines the signal phase of the traffic signal at the next moment based on the judgment result. This invention utilizes deep reinforcement learning for traffic light control, which can better adapt to complex traffic conditions, effectively reduce intersection queuing, and coordinate the passage of emergency vehicles and social vehicles when emergency vehicles pass through intersections, ensuring the rapid passage of emergency vehicles without seriously affecting the normal driving of social vehicles.
[0124] The above embodiments provide a signal control system for an emergency vehicle at an intersection, which can realize the technical solutions described in the above embodiments of the signal control method for an emergency vehicle at an intersection. The specific implementation principles of each module or unit can be found in the corresponding content of the embodiment of the signal control method for an emergency vehicle at an intersection, and will not be repeated here.
[0125] The above provides a detailed description of the signal control method and system for emergency vehicles at intersections provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A signal control method for emergency vehicles at intersections, characterized in that, include: Acquire the status information of vehicles within the target area, including ordinary vehicles and emergency vehicles, and the status information includes speed information and location information; The vehicle's state information is processed based on a preset discretization model to obtain the vehicle's state matrix, which includes the state matrix of ordinary social vehicles and the state matrix of the emergency vehicle. When a traffic signal phase is input, the state matrix of the ordinary social vehicles and the state matrix of the emergency vehicles are judged based on a preset deep reinforcement learning model to obtain the judgment result. The signal phase of the traffic signal at the next moment is determined based on the judgment result; The state matrix of ordinary social vehicles and the state matrix of emergency vehicles are judged based on a preset deep reinforcement learning model to obtain the judgment result, including: When a signal phase is input into the traffic signal system, the state matrices of ordinary vehicles and emergency vehicles are judged based on a preset reward function model to obtain the reward function values of emergency vehicles and ordinary vehicles. The reward function values of the emergency vehicles and ordinary vehicles are fed back into the deep reinforcement learning model to obtain the judgment result of the current traffic signal. The formula for calculating the reward value in the preset reward function model is as follows: ； Where 'a' represents the reward coefficient for emergency vehicles, and 'b' represents the reward coefficient for other vehicles. The reward function represents the emergency vehicle. The reward function represents the reward function for social vehicles, and the formula for calculating the reward function value for emergency vehicles is as follows: ； The formula for calculating the reward function value of ordinary social vehicles in the preset reward function model is as follows: ； in, Indicates the speed of the j-th emergency vehicle. This indicates the maximum value for emergency vehicles. , , and As a preset constant, Indicates the queue length on lane i. This indicates that the queue of vehicles has reached the maximum capacity of the lane.
2. The signal control method for emergency vehicles at intersections according to claim 1, characterized in that, The process of processing the vehicle's state information based on a preset discretization model to obtain the vehicle's state matrix includes: Discretize the vehicles within the target area to obtain multiple cells containing the vehicles; Based on the discretization model, the state information of the vehicle in each cell is transformed into a state matrix, which includes the state matrix of ordinary social vehicles and the state matrix of the emergency vehicle. The general social vehicle status matrix includes a general social vehicle speed information matrix and a general social vehicle location information matrix, and the emergency vehicle status matrix includes an emergency vehicle speed information matrix and an emergency vehicle location information matrix.
3. The signal control method for emergency vehicles at intersections according to claim 2, characterized in that, The DTSE method is used to discretize and model vehicles within the target area.
4. The signal control method for emergency vehicles at intersections according to claim 2, characterized in that, The expression for processing the state information of the ordinary social vehicles is: ； ； in, Represented as the intersection number The first lane Location information of ordinary vehicles within each cell. Indicates the intersection number The first lane Speed information for ordinary vehicles within each cell. This represents the average speed of the vehicles within the cell.
5. The signal control method for emergency vehicles at intersections according to claim 4, characterized in that, The state matrix of the emergency vehicle is: ； The state matrix of ordinary social vehicles is as follows: ； Where E represents the state matrix of the emergency vehicle. Indicates the intersection number The first lane The distance of each emergency vehicle from the intersection For the intersection The first lane The speed of the emergency vehicles.
6. The signal control method for emergency vehicles at intersections according to claim 5, characterized in that, The expression for the state space matrix of the intersection is: ； Where s represents the state space matrix of the intersection.
7. A signal control system for emergency vehicles at intersections, characterized in that, include: The status information module is used to acquire the status information of vehicles in the target area, including ordinary social vehicles and emergency vehicles, and the status information includes speed information and location information. The state matrix module is used to process the state information of the vehicle based on a preset discretization model to obtain the state matrix of the vehicle. The state matrix includes the state matrix of ordinary social vehicles and the state matrix of the emergency vehicle. The judgment module is used to judge the state matrix of ordinary social vehicles and the state matrix of emergency vehicles based on a preset deep reinforcement learning model when a traffic signal inputs a signal phase, and obtain the judgment result. The traffic signal module is used to determine the signal phase of the traffic signal at the next moment based on the judgment result; The state matrix of ordinary social vehicles and the state matrix of emergency vehicles are judged based on a preset deep reinforcement learning model to obtain the judgment result, including: When a signal phase is input into the traffic signal system, the state matrices of ordinary vehicles and emergency vehicles are judged based on a preset reward function model to obtain the reward function values of emergency vehicles and ordinary vehicles. The reward function values of the emergency vehicles and ordinary vehicles are fed back into the deep reinforcement learning model to obtain the judgment result of the current traffic signal. The formula for calculating the reward value in the preset reward function model is as follows: ； Where 'a' represents the reward coefficient for emergency vehicles, and 'b' represents the reward coefficient for other vehicles. The reward function represents the emergency vehicle. The reward function represents the reward function for social vehicles, and the formula for calculating the reward function value for emergency vehicles is as follows: ； The formula for calculating the reward function value of ordinary social vehicles in the preset reward function model is as follows: ； in, Indicates the speed of the j-th emergency vehicle. This indicates the maximum value for emergency vehicles. , , and As a preset constant, Indicates the queue length on lane i. This indicates that the queue of vehicles has reached the maximum capacity of the lane.