AIAgent-based train adjustment methods, systems, media, and equipment

Abstract

The present application relates to the technical field of train operation adjustment, and specifically provides a kind of train automatic operation adjustment method based on AI Agent, train automatic operation adjustment system based on AI Agent, computer readable storage medium, computer equipment, the method therein includes the following steps: train operation state sensing step, it includes: the train real-time state information of train related to the train of this car is sensed;Train operation adjustment scheme generation step, it includes: according to the train operation adjustment model algorithm with reasoning decision function, generates train real-time adjustment scheme;Train operation adjustment scheme verification step, it includes: the train real-time adjustment scheme generated is verified, whether the train real-time adjustment scheme generated is verified is judged;Train operation scheme execution step, it includes: the train operation adjustment scheme verified is issued to train. Through such composition, the autonomous adjustment for train operation can be realized.

Description

Technical Field

[0001] This invention relates to the field of train operation adjustment technology, specifically providing an AI Agent-based automatic train operation adjustment method, an AI Agent-based automatic train operation adjustment system, a computer-readable storage medium, and a computer device. Background Technology

[0002] AIAgent is an intelligent entity that can perceive its environment, make decisions, and execute actions. Compared with traditional training models and large models, it has the ability to perceive, remember, reason, make decisions, and execute. Research on AIAgent has received much attention at home and abroad and has yielded diverse research results. However, it has not yet been applied to the field of train operation adjustment.

[0003] Currently, my country employs a train operation control system based on a train-ground-train architecture and a train operation adjustment system primarily reliant on dispatcher experience. As the road network expands or vehicle density increases, the timeliness of resolving inter-line conflicts through human-machine dialogue becomes unreliable. Therefore, a more effective and flexible solution is needed to meet the demands of ever-increasing traffic volume and real-time decision-making.

[0004] Currently, there are relatively mature vehicle-to-vehicle communication technologies that can realize information exchange between trains; heuristic, reinforcement learning, deep learning, and simulation methods designed for the characteristics of train operation adjustment problems can quickly generate train operation adjustment schemes; mathematical models based on safety constraints, including the minimum duration of train operation activities in sections, the minimum duration of train stopping activities, train arrival and departure time logic, track and train reception and departure capacity, station departure sequence logic, and section overtaking, can be used to verify the rationality of train operation adjustment schemes, but there is no system or method that combines them to realize train autonomous perception, decision-making, and execution.

[0005] Accordingly, a new technical solution is needed in this field to solve the above problems. Summary of the Invention

[0006] In order to at least partially solve the above-mentioned technical problems, the present invention is proposed.

[0007] Specifically, the purpose of this invention is to use an AI Agent system to autonomously perceive the operating status of trains across the entire network and to generate and verify train operation adjustment plans. Based on this, trains can make autonomous adjustments according to the verified decisions.

[0008] In a first aspect, the present invention provides a method for autonomous train operation adjustment based on an AI Agent, the method comprising the following steps:

[0009] The train operation status perception steps include: perceiving the real-time status information of trains related to this train;

[0010] The steps for generating a train operation adjustment plan include: generating a real-time train adjustment plan based on a train operation adjustment model algorithm with reasoning and decision-making capabilities;

[0011] The train operation adjustment plan verification steps include: verifying the generated real-time train adjustment plan and determining whether the generated real-time train adjustment plan passes the verification.

[0012] The steps for implementing the train operation plan include: distributing the verified train operation adjustment plan to the trains.

[0013] This configuration enables the autonomous adjustment of train operation.

[0014] In one possible implementation of the above-mentioned AI Agent-based autonomous train operation adjustment method, the train operation adjustment scheme generation step further includes:

[0015] The planned train timetable is retrieved and compared with the real-time train status information determined by the train operation status perception step to determine the initial train delay information;

[0016] Based on the initial train delay information, a real-time train adjustment plan is generated using a train operation adjustment model algorithm.

[0017] This structure allows for the generation of train operation adjustment plans that are yet to be verified.

[0018] In one possible implementation, in the step of "generating a real-time train adjustment plan based on a train operation adjustment model algorithm with reasoning and decision-making capabilities," the objective function is to minimize the total arrival delay time z.

[0019]

[0020] in, This represents the actual arrival time of train k at station u. Let K be the planned arrival time of train k at station u, where K is the train set and U is the station set.

[0021] This structure provides a possible form for the objective function in optimization computation.

[0022] In one possible implementation of the AI ​​Agent-based train autonomous operation adjustment method, the constraints limiting the feasible solution in the step of "generating a real-time train adjustment scheme based on a train operation adjustment model algorithm with reasoning and decision-making capabilities" include:

[0023] (1) Minimum arrival interval constraint:

[0024]

[0025] in, Let k+1 be the actual arrival time of train k+1 at station u. Let t be the actual arrival time of train k at station u. 到 This is the minimum arrival interval for trains. and The difference is that the actual arrival interval between two adjacent trains at station u should not be less than the minimum arrival interval t. 到 ;

[0026] (2) Minimum departure interval constraint:

[0027]

[0028] in, This represents the actual departure time of train k+1 at station u. Let t be the actual departure time of train k at station u. 发 This is the minimum train departure interval;

[0029] (3) Minimum runtime constraint for the interval:

[0030]

[0031] in, Let this be the actual arrival time of train k at station u+1. Let τ be the actual departure time of train k at station u. 起 Additional time for starting the train, τ 停 Add time for parking, r u,u+1 The minimum runtime between station u and station u+1;

[0032] (4) Minimum stopping time constraint:

[0033]

[0034] in, This represents the actual departure time of train k at station u. This represents the actual arrival time of train k at station u. Let be the minimum stopping time of train k at station u;

[0035] (5) Departure time constraints:

[0036]

[0037] in, This represents the actual departure time of train k at station u. This represents the planned departure time of train k at station u.

[0038] (6) Crossing constraint:

[0039]

[0040] in, This represents the actual departure time of train k at station u. This represents the actual departure time of train g at station u. This represents the actual arrival time of train k at station u. This represents the actual arrival time of train g at station u.

[0041] This structure provides a possible form of constraint conditions in optimization calculations.

[0042] In one possible implementation of the above-mentioned AI Agent-based train autonomous operation adjustment method, the train operation adjustment scheme verification step includes:

[0043] The generated real-time train adjustment scheme is then fed into a mixed-integer linear programming mathematical model based on an event activity graph model to verify the safety constraints, thereby verifying the feasibility of the listed real-time operation adjustment scheme.

[0044] If the verification is successful, proceed to the train operation plan execution steps.

[0045] If the verification fails, the process will return to the train operation adjustment plan generation step and regenerate the train adjustment plan to be verified.

[0046] This configuration allows for the verification of the feasibility of the generated train operation adjustment plan.

[0047] For the aforementioned AI Agent-based train autonomous operation adjustment method, in one possible implementation, the safety constraints in the mixed-integer linear programming mathematical model based on the event activity graph model include:

[0048] (1) Verification of the minimum duration of train operation in the section:

[0049]

[0050] Among them, e j With e i These represent the arrival and departure events of the same train at two adjacent stations, respectively. For event e j The moment of occurrence, For event e i At the moment of occurrence, a i,jIndicates train activity, A run This represents the set of train operation activities within a given section. Indicates the minimum duration of the interval operation;

[0051] (2) Minimum duration test of train stopping activity:

[0052]

[0053] Among them, e j With e i These refer to the departure and arrival events of the same train at the same station. For event e j The moment of occurrence, For event e i At the moment of occurrence, a i,j Indicates train activity, A dwell This indicates a gathering or event at a train stop. Indicates the minimum duration of the stop activity;

[0054] (3) Logical verification of train arrival and departure times:

[0055]

[0056] Among them, e j With e i These refer to the departure and arrival events of the same train at the same station. For event e j The moment of occurrence, For event e i At the moment of occurrence, a i,j Indicates train activity, A station Indicates a collection of activities at the station;

[0057] (4) Inspection of route and train arrival / departure capacity:

[0058]

[0059]

[0060] Among them, e j With e i These refer to the arrival or departure events of two adjacent trains at the same station. For event e j The moment of occurrence, For event e i At the moment of occurrence, a i,j Indicates train activity, A arr A represents the set of activities for the arrival intervals of adjacent trains. dep This represents the set of activities at intervals between the departures of adjacent trains. This represents the minimum duration of train arrival or departure intervals, reflecting the track and train handling capacity. Two adjacent trains cannot simultaneously or within a very short time occupy the same throat area. M is a very large positive number, X(e i ,e j ) for event e i With e j The order of occurrence, train event e i Earlier than e j When it occurs, X(e) i ,e j ) = 1, otherwise X(e i ,e j ) = 0;

[0061] (5) Logical check of station departure order:

[0062] X(e i ,e j )+X(e j ,e i )=1 (55)

[0063] X(e i ,e j ) for event e i With e j The order of occurrence, train event e i Earlier than e j When it occurs, X(e) i ,e j ) = 1, otherwise X(e i ,e j If X(e) = 0, then X(e) = 0. i ,e j If ) = 1, then X(e) j ,e i ) = 0;

[0064] (6) Interval crossing test:

[0065] X(e i ,e j )-X(e g ,e k )=0 (56)

[0066] Event e i With e g These represent the departure and arrival events of the same train at adjacent stations, respectively. Event e j With e k These represent the departure and arrival events of another train at an adjacent station, respectively.

[0067] This structure provides a possible form of constraint conditions for verifying the generated train operation adjustment scheme.

[0068] In one possible implementation of the above-mentioned AI Agent-based train autonomous operation adjustment method, the train operation adjustment model algorithm is a heuristic genetic algorithm.

[0069] Based on this structure, a specific train operation adjustment model algorithm is presented.

[0070] In a second aspect, the present invention provides a computer-readable storage medium including a memory adapted to store a plurality of program codes adapted to be loaded and executed by a processor to perform the AI ​​Agent-based train autonomous operation adjustment method described in any of the preceding claims.

[0071] It is understood that the computer-readable storage medium has all the technical effects of the AI ​​Agent-based train autonomous operation adjustment method described in any of the foregoing, and will not be repeated here.

[0072] Those skilled in the art will understand that all or part of the processes in the AI ​​Agent-based train autonomous operation adjustment method of the present invention can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which includes, but is not limited to, program code for executing the AI ​​Agent-based train autonomous operation adjustment method. For ease of explanation, only the parts relevant to the present invention are shown. The computer program code can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium can include any entity or device capable of carrying the computer program code, a medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content contained in the computer-readable storage medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0073] In a third aspect, the present invention provides a computer device including a memory and a processor, the memory being adapted to store a plurality of program codes adapted to be loaded and executed by the processor to perform the AI ​​Agent-based train autonomous operation adjustment method described in any of the preceding claims.

[0074] It is understood that this device possesses all the technical effects of the AI ​​Agent-based train autonomous operation adjustment method described above, and will not be elaborated further here. This device can be a computer-controlled device comprising various electronic devices.

[0075] In a fourth aspect, the present invention provides a train autonomous operation adjustment system based on an AI Agent, wherein the control system includes a control module configured to execute the train autonomous operation adjustment method based on the AI ​​Agent described in any of the preceding claims.

[0076] It is understood that the AI ​​Agent-based train autonomous operation adjustment system has all the technical effects of the AI ​​Agent-based train autonomous operation adjustment method described above, and will not be repeated here.

[0077] In the description of this invention, a "control module" may include hardware, software, or a combination of both. A module may include hardware circuitry, various suitable sensors, communication ports, memory, and may also include software components, such as program code, or a combination of software and hardware. The processor may be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and / or signal processing capabilities. The processor may be implemented in software, in hardware, or a combination of both. Non-transitory computer-readable storage media include any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc.

[0078] Furthermore, it should be understood that since the control module is only used to illustrate the functional unit in the system corresponding to the AI ​​Agent-based train autonomous operation adjustment method of the present invention, the physical device corresponding to the control module can be the processor itself, or a part of the processor's software, hardware, or a combination of software and hardware. Therefore, the number of control modules is only illustrative. Those skilled in the art will understand that the control module can be adaptively split according to actual conditions. The specific splitting of the control module will not cause the technical solution to deviate from the principle of the present invention; therefore, the technical solutions after splitting will all fall within the protection scope of the present invention. Attached Figure Description

[0079] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings, in which:

[0080] Figure 1 This diagram illustrates a flow chart of an AI Agent-based train autonomous operation adjustment method according to an embodiment of the present invention. Figure 1 ;

[0081] Figure 2 This diagram illustrates a flow chart of an AI Agent-based train autonomous operation adjustment method according to an embodiment of the present invention. Figure 2 ;

[0082] Figure 3 The planned train operation diagram before train adjustment is provided in a scenario where an AI Agent-based train autonomous operation adjustment method according to an embodiment of the present invention is applied.

[0083] Figure 4 The actual train operation diagram after adjustment is shown in a scenario where the train autonomous operation adjustment method based on AI Agent, an embodiment of the present invention, is applied.

[0084] Figure 5 This diagram illustrates the structure of an AI Agent-based autonomous train operation adjustment system according to an embodiment of the present invention. Figure 1 ;as well as

[0085] Figure 6 This diagram illustrates the structure of an AI Agent-based autonomous train operation adjustment system according to an embodiment of the present invention. Figure 2 . Detailed Implementation

[0086] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0087] It should be noted that in the description of this invention, terms such as "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," indicating directional or positional relationships, are based on the directional or positional relationships shown in the accompanying drawings. These are merely for ease of description and do not indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation; therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0088] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "setting," and "connection" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection, an indirect connection through an intermediate medium, or a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0089] Furthermore, to better illustrate the present invention, numerous specific details are provided in the following detailed embodiments. Those skilled in the art should understand that the present invention can be practiced even without certain specific details. In some instances, the principles of AI agents well-known to those skilled in the art are not described in detail, in order to highlight the main points of the present invention.

[0090] Reference Figure 1 and Figure 2 , Figure 1 This diagram illustrates a flowchart of an AI Agent-based autonomous train operation adjustment method according to an embodiment of the present invention. Figure 1 , Figure 2 This diagram illustrates a flowchart of an AI Agent-based autonomous train operation adjustment method according to an embodiment of the present invention. Figure 2 .like Figure 1 and Figure 2 As shown, in one possible implementation, the train autonomous operation adjustment method based on AI Agent mainly includes the following steps:

[0091] S110, Train operation status perception steps, which include: perceiving the real-time status information of trains related to this train (such as the preceding train).

[0092] Real-time train status information includes the train's real-time location, speed, acceleration, and the actual time the train arrives at the station and the actual time the train departs from the station.

[0093] S120, the train operation adjustment plan generation steps, which include: generating a real-time train adjustment plan based on a train operation adjustment model algorithm with reasoning and decision-making functions.

[0094] S130, the train operation adjustment plan verification step, which includes: verifying the generated real-time train adjustment plan and determining whether the generated real-time train adjustment plan passes the verification; if yes, proceed to S140; if no, proceed to S120.

[0095] S140, Train operation plan execution steps, which include: issuing the verified train operation adjustment plan to the train to achieve autonomous train operation adjustment.

[0096] In one possible implementation, step S120 further includes:

[0097] S1201. Call up the planned train timetable and compare it with the real-time train status information determined by the train operation status perception step to determine the initial delay information of the train.

[0098] The planned train timetable includes, but is not limited to, train number, station name, planned arrival time at the station, and planned departure time from the station. The initial train delay information includes, but is not limited to, the delayed train number, the station name where the initial delay occurred, the type of delay, and the delay time.

[0099] S1202. Based on the initial train delay information, a real-time train adjustment plan is generated using a train operation adjustment model algorithm.

[0100] The real-time train adjustment scheme includes, but is not limited to, arrival and departure time decisions, track utilization decisions, and station overtaking decisions. The train operation adjustment model algorithm can be any algorithm with reasoning and decision-making capabilities; for example, the train operation adjustment model algorithm employs a classic heuristic genetic algorithm.

[0101] In one possible implementation, step S1202 further includes:

[0102] In the real-time train adjustment scheme, the ultimate goal of train operation adjustment is to restore delayed trains to their on-time status as much as possible. Since the departure time of a train is calculated by adding the arrival time and stop time, the actual arrival time of the train can be chosen as the decision variable, while the departure time is a variable that changes with the actual arrival time. Therefore, in this embodiment, the objective function for optimizing the real-time train adjustment scheme is to minimize the total arrival delay time z:

[0103]

[0104] in, This represents the actual arrival time of train k at station u. Let K be the planned arrival time of train k at station u, K be the set of trains, and U be the set of stations. The total delay of trains arriving at each station can be measured by minimizing the sum of the differences between the actual arrival time and the planned arrival time of train k at station u.

[0105] In this example, a classic heuristic genetic algorithm can be used to solve the train operation adjustment scheme. The solution steps of the heuristic genetic algorithm include:

[0106] Step 1.1, Chromosome Encoding:

[0107] To ensure a better correspondence between the chromosomes and the adjustments to the train operation plan, the chromosomes are encoded using real numbers. The genes on each chromosome are the numbers of the delayed trains, and the chromosome length is the total number of delayed trains.

[0108] Step 1.2, Population Initialization:

[0109] The departure order of multiple delayed trains at the adjustment station is randomly generated. The randomly generated departure order is likely to be an infeasible solution. Therefore, infeasible solutions are directly eliminated using the constraints.

[0110] In one possible implementation, the constraints limiting the feasible solution include:

[0111] (1) Minimum arrival interval constraint:

[0112]

[0113] in, Let k+1 be the actual arrival time of train k+1 at station u. Let t be the actual arrival time of train k at station u. 到 This is the minimum arrival interval for trains. and The difference is that the actual arrival interval between two adjacent trains at station u should not be less than the minimum arrival interval t. 到 .

[0114] (2) Minimum departure interval constraint:

[0115]

[0116] in, This represents the actual departure time of train k+1 at station u. Let t be the actual departure time of train k at station u. 发 This is the minimum train departure interval. and The difference is that the actual departure interval between two adjacent trains at station u should not be less than the minimum train departure interval t. 发 .

[0117] (3) Minimum runtime constraint for the interval:

[0118]

[0119] in, Let this be the actual arrival time of train k at station u+1. Let τ be the actual departure time of train k at station u. 起 Additional time for starting the train, τ 停 Add time for parking, r u,u+1 The minimum running time between station u and station u+1 is the minimum running time. The train's total running time within the interval should not be less than the minimum running time of the interval.

[0120] (4) Minimum stopping time constraint:

[0121]

[0122] in, This represents the actual departure time of train k at station u. This represents the actual arrival time of train k at station u. Let be the minimum stopping time for train k at station u. The stopping time of the train at the station should not be less than the minimum stopping time of the station.

[0123] (5) Departure time constraints:

[0124]

[0125] in, This represents the actual departure time of train k at station u. This is the planned departure time of train k at station u. For passenger trains, it is generally not allowed to depart earlier than the planned departure time.

[0126] (6) Crossing constraint:

[0127]

[0128] in, This represents the actual departure time of train k at station u. This represents the actual departure time of train g at station u. This represents the actual arrival time of train k at station u. Let g be the actual arrival time of train g at station u. For any two trains departing from the same station, the train that departs first should arrive at the next station first.

[0129] Step 1.3: Construct the fitness function:

[0130]

[0131] Assuming there are n chromosomes, the fitness of the i-th chromosome at station u is calculated as the reciprocal of the total arrival delay time at station u+1.

[0132] Step 3.4, Select Operation:

[0133]

[0134] The selection process uses a selection operation to identify superior parent chromosomes and discards chromosomes with low fitness. Based on the principle that higher-fitting parent chromosomes have a greater probability of being passed on to the next generation, a roulette wheel method is used to select the best parent chromosome. The probability of the i-th chromosome being selected as the parent and passed on to the next generation is P. i The fitness f of the i-th chromosome is calculated as follows: u i The sum of fitness on n chromosomes The percentage.

[0135] Step 3.5, Crossover and Mutation Operation:

[0136] Since the encoded chromosomes represent the departure order of delayed trains, crossover between two parent chromosomes will inevitably result in an infeasible solution, regardless of the method used. Therefore, population diversity is achieved through mutation. Each chromosome corresponds to a departure order; mutating only one gene position will result in two identical trains in the departure order, inevitably leading to an infeasible solution. To ensure the generated individual represents a feasible solution, the genes at two positions are randomly swapped, thereby increasing population diversity and preventing the population from getting trapped in local optima.

[0137] Step 3.6, Stop sign.

[0138] Stop adjusting when the maximum number of iterations is reached.

[0139] In one possible implementation, step S130 further includes:

[0140] The generated real-time train adjustment plan is input into a mixed-integer linear programming mathematical model based on an event activity graph model to verify the safety constraints, thereby verifying the feasibility of the listed real-time operation adjustment plan. If the verification passes, proceed to S140; if the verification fails, jump back to S120 and regenerate the train adjustment plan to be verified.

[0141] In one possible implementation, the safety constraints in the mixed-integer linear programming mathematical model based on the event activity graph model include:

[0142] (1) Verification of the minimum duration of train operation in the section:

[0143]

[0144] Among them, e j With e i These represent the arrival and departure events of the same train at two adjacent stations, respectively. For event e j The moment of occurrence, For event e i At the moment of occurrence, a i,j Indicates train activity, A run This represents the set of train operation activities within a given section. This indicates the minimum duration of the interval operation activity, and checks whether the train's interval operation meets the minimum duration of the interval operation activity.

[0145] (2) Minimum duration test of train stopping activity:

[0146]

[0147] Among them, e j With e i These refer to the departure and arrival events of the same train at the same station. For event e j The moment of occurrence, For event e i At the moment of occurrence, a i,j Indicates train activity, A dwell This indicates a gathering or event at a train stop. This indicates the minimum duration of the station stop activity, and checks whether the train's station stop activity meets the minimum duration requirement.

[0148] (3) Logical verification of train arrival and departure times:

[0149]

[0150] Among them, e j With e i These refer to the departure and arrival events of the same train at the same station. For event e j The moment of occurrence, For event e i At the moment of occurrence, a i,j Indicates train activity, A station This represents the set of station activities, including stopping activities and passing activities, and checks whether the train's station activities satisfy the logic that the departure event occurs later than the arrival event.

[0151] (4) Inspection of route and train arrival / departure capacity:

[0152]

[0153]

[0154] Among them, e j With e i These represent the arrival (departure) events and arrival (departure) events of two adjacent trains at the same station, respectively. For event e j The moment of occurrence, For event e i At the moment of occurrence, a i,j Indicates train activity, A arr A represents the set of activities for the arrival intervals of adjacent trains. dep This represents the set of activities at intervals between the departures of adjacent trains. This represents the minimum duration of train arrival (departure) intervals, reflecting the track and train handling capacity. Two adjacent trains cannot simultaneously or within a very short time occupy the same throat area. M is a very large positive number, X(ei ,e j ) for event e i With e j The order of occurrence, train event e i Earlier than e j When it occurs, X(e) i ,e j ) = 1, otherwise X(e i ,e j If ) = 0, check whether the minimum duration of the arrival (departure) interval activity is met between two adjacent trains.

[0155] (5) Logical check of station departure order:

[0156] X(e i ,e j )+X(e j ,e i )=1 (55)

[0157] X(e i ,e j ) for event e i With e j The order of occurrence, train event e i Earlier than e j When it occurs, X(e) i ,e j ) = 1, otherwise X(e i ,e j If X(e) = 0, then X(e) = 0. i ,e j If ) = 1, then X(e) j ,e i If ) = 0, check whether the train departure sequence at the station is logical.

[0158] (6) Interval crossing test:

[0159] X(e i ,e j )-X(e g ,e k )=0 (56)

[0160] Event e i With e g These represent the departure and arrival events of the same train at adjacent stations, respectively. Event e j With e k These represent the departure and arrival events of another train at an adjacent station, respectively. The two adjacent trains are prohibited from crossing each other within the section. The departure order of the two trains at the current station is consistent with the arrival order at the next station. This checks whether the running lines of the two adjacent trains intersect within the section.

[0161] The train autonomous operation adjustment method of the AI ​​Agent of the present invention will be further described in detail below with reference to a specific example. Obviously, the following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the invention.

[0162] Specifically, taking the operation scenario of a high-speed train from Guangzhou South to Chenzhou West dispatching section as an example, this invention provides a detailed description of an AI Agent-based train autonomous operation adjustment method according to one embodiment.

[0163] In this example, the planned train timetable includes:

[0164] (1) Train number, including: G66, G6030, G1110, G822, G1112, G1008.

[0165] (2) Station names, including: Guangzhou South, Guangzhou North, Qingyuan, Yingde West, Shaoguan, Lechang East, and Chenzhou West.

[0166] (3) The time when the train is scheduled to arrive at the station and the time when it is scheduled to depart from the station.

[0167] The planned train timetable is shown in Table 1 and Figure 3 As shown in the example, the information in the planned train timetable is ('G66', 'Shaoguan', 'dep', 10:48). Comparing this with the real-time train status information ('G66', 'Shaoguan', 'dep', 11:28), the initial delay information is obtained as ('G66', 'Shaoguan', 'dep', 1400). It can be seen that the delay information includes the delayed train number 'G66', the station name where the initial delay occurred 'Shaoguan', the delay type 'dep' (departure delay), and the delay time of 1400 seconds.

[0168] Table 1. Planned train timetable for Guangzhou South - Chenzhou West

[0169]

[0170]

[0171] The aforementioned AI Agent-based autonomous train operation adjustment method was used for optimization calculations to obtain the adjusted and validated train timetable. The adjusted train timetable is shown in Table 2. Figure 4 As shown.

[0172] Table 2 Adjusted train timetable for Guangzhou South - Chenzhou West

[0173]

[0174] As can be seen, in the preferred embodiment of the present invention, by applying the AI ​​Agent to the field of train operation adjustment, rapid and effective adjustments can be made to train operation in the target scheduling section. Based on this, those skilled in the art can determine the scale of the target scheduling section according to actual needs.

[0175] On the other hand, the present invention also provides a reference Figure 5 and Figure 6 , Figure 5 This diagram illustrates the structure of an AI Agent-based autonomous train operation adjustment system according to an embodiment of the present invention. Figure 1 , Figure 6 This diagram illustrates the structure of an AIAgent-based autonomous train operation adjustment system according to an embodiment of the present invention. Figure 2 .like Figure 5 and Figure 6 As shown, in one possible implementation, the AI ​​Agent-based autonomous train operation adjustment system mainly includes:

[0176] The train operation status sensing module is configured to sense the real-time status information of trains related to this train (such as the preceding train).

[0177] Real-time train status information includes the train's real-time location, speed, acceleration, and the actual time the train arrives at the station and the actual time the train departs from the station.

[0178] The train operation adjustment plan generation module is configured to generate real-time train adjustment plans based on a train operation adjustment model algorithm with reasoning and decision-making capabilities.

[0179] The train operation adjustment scheme verification module is configured to: verify the real-time train adjustment scheme and determine whether the generated real-time train adjustment scheme passes the verification; if yes, it will proceed to the train operation autonomous adjustment implemented by the train operation scheme execution module; if no, it will proceed to the train operation adjustment scheme regenerated by the train operation adjustment scheme generation module.

[0180] The train operation plan execution module is configured to: distribute the verified train operation adjustment plan to the train to enable the train to adjust its operation autonomously.

[0181] In one specific implementation, the train autonomous operation adjustment system based on AI Agent mainly includes a train autonomous operation adjustment Agent module, a perception module, a memory module, a reasoning and decision-making module, a verification and interpretation module, and an execution module.

[0182] The Train Autonomous Operation Adjustment Agent module, as the control module of the AI ​​Agent-based train autonomous operation adjustment system, possesses various control rules for other functional modules. Furthermore, as the central hub of the AI ​​Agent-based train autonomous operation adjustment system, one function of the Train Autonomous Operation Adjustment Agent module is to receive information output from other functional modules, such as real-time train status information from the perception module, real-time train adjustment plans from the reasoning and decision-making module, and real-time train verification results from the verification and interpretation module, and then transmit this information to the memory module for short-term memory storage. Another function of the Train Autonomous Operation Adjustment Agent module is to transmit the necessary information to other modules by accessing long-term or short-term memory within the memory module when calling them. For example, when calling the train delay information analysis submodule of the reasoning and decision-making module, the planned train timetable from the long-term memory submodule and the real-time train status from the short-term memory submodule are transmitted to the train delay information analysis submodule; when calling the verification and interpretation module, the real-time train adjustment plan from the short-term memory submodule is transmitted to the verification and interpretation module; and when calling the execution module, the real-time train verification results from the short-term memory submodule are transmitted to the execution module.

[0183] The sensing module primarily utilizes vehicle-to-vehicle communication technology to perceive the real-time status of the preceding vehicle (e.g., through parameters such as position, speed, and acceleration). Based on the real-time status of both the current train and the preceding train, the current train adjusts its own real-time status (also presented through parameters such as position, speed, and acceleration). This aims to ensure a consistently safe distance between trains. By leveraging the sensing module to help each train acquire real-time parameters such as position, speed, and acceleration from its associated trains, each train can effectively perceive the real-time operating status of the entire network. From the acquired position, speed, and acceleration parameters, it can sensitively detect whether the preceding and following trains are accelerating / decelerating or maintaining a safe distance. Based on this, the train predicts the trajectory of the preceding vehicle by acquiring its status information, and then calculates in real time the safe speed to ensure that it will not collide with the preceding vehicle. For example, if the preceding vehicle takes deceleration measures, the train can obtain the magnitude of the preceding vehicle's deceleration at this time, and thus adjust its own deceleration in a timely manner, so that the two trains can maintain a safe distance continuously. In addition, each train can output the real-time positions of the preceding vehicle and itself, as well as the actual arrival and departure times of each station along the route, which are sensed by the sensing module, to the train autonomous operation adjustment agent module.

[0184] The memory module is divided into a long-term memory submodule and a short-term memory submodule. The long-term memory submodule is mainly used to store basic data, such as the line topology, train operation performance, planned train timetables, and arrival and departure line operation plans. The short-term memory submodule is mainly used to represent the train autonomous operation adjustment agent module's ability to remember the phase information output by the perception module, reasoning and decision-making module, and verification and interpretation module. The phase information mainly includes the train's real-time status, real-time train delay information, real-time train adjustment plan, and real-time train verification results.

[0185] The reasoning and decision-making module mainly includes a train delay information analysis submodule and a train operation adjustment plan generation submodule. The train delay information analysis submodule is mainly used to obtain real-time train delay information, and the train operation adjustment plan generation submodule is mainly used to provide real-time train operation adjustment plans.

[0186] In one possible implementation, the train delay information analysis submodule is mainly used to: obtain real-time train delay information by comparing the planned train timetable with the actual arrival and departure times of each station along the route of each train, based on the basic data (such as the planned train timetable) stored in the long-term memory submodule and the stage information (such as the real-time status of the train) stored in the short-term memory submodule.

[0187] In one possible implementation, the train operation adjustment scheme generation submodule is mainly used to: based on real-time train delay information in the stage information and planned train timetables in the basic data, and by calling mature train operation adjustment model algorithms, provide a real-time train operation adjustment scheme, such as arrival and departure time decision, track utilization decision and station overtaking decision.

[0188] For example, the train operation adjustment model algorithm adopts the classic heuristic genetic algorithm. Here, a chromosome real number encoding is designed for the train operation adjustment problem. For example, the gene is the number of the delayed train, the chromosome length is the number of all delayed trains, the initial population is the departure order of all trains that need to be adjusted at the adjustment station, the fitness function is the reciprocal of the total weighted arrival delay time of the next station after the adjustment station, the selection operation is based on the principle that the higher the fitness of the parent chromosome, the greater the probability of inheritance to the next generation, and the crossover and mutation operation randomly swaps the positions of genes at two locations to increase the diversity of the population and prevent the optimization from getting trapped in a local optimum.

[0189] The verification and interpretation module is mainly used to input the real-time train operation adjustment scheme determined by the reasoning and decision-making module into the pre-constructed mixed-integer linear programming mathematical model based on event activity graphs. The constraints include the minimum duration of train interval operation activities, the minimum duration of train stop activities, train arrival and departure time logic, line and train reception and departure capacity, station departure sequence logic, interval overtaking, etc. The arrival and departure time decisions of the real-time train operation adjustment scheme are tested by inputting them into various time interval constraints defined by the constraints. For example, it tests whether the difference between the arrival time of each train at the next station and the departure time of the previous station is greater than the minimum interval running time, whether the difference between the departure time and the arrival time of each train at each station is greater than the minimum stop time, whether the difference between the arrival times of two adjacent trains at each station is greater than the minimum arrival interval, whether the difference between the departure times is greater than the minimum departure interval, etc. By checking whether the constraints are met, the feasibility of the real-time train operation adjustment scheme determined by the reasoning and decision-making module is verified, as well as whether there are any logical and safety errors in the determined real-time train operation adjustment scheme.

[0190] If a train real-time operation adjustment plan passes verification, it will return a "verification passed" result to the train autonomous operation adjustment agent module. This means the agent module can only proceed to the execution function module upon receiving a "verification passed" result. If the plan fails verification, it will return a "verification failed" result to the agent module. Upon receiving this result, the agent module will invoke the reasoning and decision-making module to generate another real-time adjustment plan. Based on this, the agent module will again invoke the verification and interpretation module to test the newly generated plan until it receives a "verification passed" result, at which point it will invoke the execution function module.

[0191] The execution function module is mainly used to send the real-time train operation adjustment plan to the train corresponding to the "passed verification" train real-time verification result information, so as to realize the train's autonomous operation adjustment.

[0192] It should be noted that although the steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effects of the present invention, different steps do not necessarily have to be executed in such an order. They can be executed simultaneously or in other orders, and some steps can be added, replaced or omitted.

[0193] It should be noted that although the AI ​​Agent-based train autonomous operation adjustment method described above has been used as an example, those skilled in the art will understand that the present invention is not limited thereto. In fact, users can flexibly adjust the relevant steps and parameters according to actual application scenarios and other factors.

[0194] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for autonomous train operation adjustment based on AI Agent, characterized in that, The method includes the following steps: The train operation status perception steps include: perceiving the real-time status information of trains related to this train; The steps for generating a train operation adjustment plan include: generating a real-time train adjustment plan based on a train operation adjustment model algorithm with reasoning and decision-making capabilities; The train operation adjustment plan verification steps include: verifying the generated real-time train adjustment plan and determining whether the generated real-time train adjustment plan passes the verification. The steps for implementing the train operation plan include: distributing the verified train operation adjustment plan to the trains; The verification steps for the train operation adjustment plan specifically include: The generated real-time train adjustment scheme is then fed into a mixed-integer linear programming mathematical model based on an event activity graph model to verify the safety constraints, thereby verifying the feasibility of the listed real-time operation adjustment scheme. If the verification is successful, proceed to the train operation plan execution steps. Specifically, the safety constraints in the mixed-integer linear programming mathematical model based on the event-activity graph model include: (1) Verification of the minimum duration of train operation in the section: （51） in, and These represent the arrival and departure events of the same train at two adjacent stations, respectively. For the event The moment of occurrence, For the event The moment of occurrence, Indicates train activity. This represents the set of train operation activities within a given section. Indicates the minimum duration of the interval operation; (2) Verification of the minimum duration of train stopping activities: （52） in, and These refer to the departure and arrival events of the same train at the same station. For the event The moment of occurrence, For the event The moment of occurrence, Indicates train activity. This indicates a gathering or event at a train stop. Indicates the minimum duration of the stop activity; (3) Logical verification of train arrival and departure times: （53） in, and These refer to the departure and arrival events of the same train at the same station. For the event The moment of occurrence, For the event The moment of occurrence, Indicates train activity. Indicates a collection of activities at the station; (4) Inspection of the route and train arrival / departure capacity: （541） （542） in, and These refer to the arrival or event of two adjacent trains at the same station. For the event The moment of occurrence, For the event The moment of occurrence, Indicates train activity. This represents the set of activities at intervals between the arrival of adjacent trains. This represents the set of activities at intervals between the departures of adjacent trains. This represents the minimum duration of train arrival (departure) intervals, reflecting the track and train handling capacity. It indicates that two adjacent trains cannot simultaneously or within a very short time occupy the same throat area. M is a very large positive number. For the event and The order of occurrence, the train incident Earlier When it happens ,otherwise ; (5) Logical check of station departure order: （55） For the event and The order of occurrence, the train incident Earlier When it happens ,otherwise ,like ,but ; (6) Interval crossover test: （56） event and These represent the departure and arrival events of the same train at adjacent stations, respectively. and These represent the departure and arrival events of another train at an adjacent station, respectively.

2. The train autonomous operation adjustment method based on AI Agent according to claim 1, characterized in that, The train operation adjustment plan generation step further includes: The planned train timetable is retrieved and compared with the real-time train status information determined by the train operation status perception step to determine the initial train delay information; Based on the initial train delay information, a real-time train adjustment plan is generated using a train operation adjustment model algorithm.

3. The train autonomous operation adjustment method based on AI Agent according to claim 2, characterized in that, In the step of "generating a real-time train adjustment plan based on a train operation adjustment model algorithm with reasoning and decision-making capabilities", the objective function is to minimize the total arrival delay time z. （1） in, Let K be the actual arrival time of train K at station U. Let K be the planned arrival time of train k at station u, where K is the train set and U is the station set.

4. The train autonomous operation adjustment method based on AI Agent according to claim 2, characterized in that, In the step of "generating a real-time train adjustment plan based on a train operation adjustment model algorithm with reasoning and decision-making capabilities", the constraints limiting the feasible solution include: (1) Minimum arrival interval constraint: （21） in, Let k+1 be the actual arrival time of train k+1 at station u. Let K be the actual arrival time of train K at station U. This is the minimum arrival interval for trains. and The difference is that the actual arrival interval between two adjacent trains at station U should not be less than the minimum arrival interval of the trains. ; (2) Minimum departure interval constraint: （22） in, This represents the actual departure time of train k+1 at station u. This represents the actual departure time of train k at station u. This is the minimum train departure interval; (3) Minimum runtime constraint for the interval: （23） in, Let this be the actual arrival time of train k at station u+1. This represents the actual departure time of train k at station u. Additional time is added for starting the car. Additional time for parking, The minimum runtime between station u and station u+1; (4) Minimum stopping time constraint: （24） in, This represents the actual departure time of train k at station u. Let K be the actual arrival time of train K at station U. Let be the minimum stopping time of train k at station u; (5) Departure time constraints: （25） in, This represents the actual departure time of train k at station u. This refers to the planned departure time of train k at station u. (6) Crossover constraint: （26） in, This represents the actual departure time of train k at station u. This represents the actual departure time of train g at station u. Let K be the actual arrival time of train K at station U. This represents the actual arrival time of train g at station u.

5. The train autonomous operation adjustment method based on AI Agent according to claim 1, characterized in that, The verification steps for the train operation adjustment plan include: The generated real-time train adjustment scheme is then fed into a mixed-integer linear programming mathematical model based on an event activity graph model to verify the safety constraints, thereby verifying the feasibility of the listed real-time operation adjustment scheme. If the verification fails, the process will return to the train operation adjustment plan generation step and regenerate the train adjustment plan to be verified.

6. The train autonomous operation adjustment method based on AI Agent according to claim 1, characterized in that, The train operation adjustment model algorithm is a heuristic genetic algorithm.

7. A computer-readable storage medium comprising a memory adapted to store a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by a processor to perform the AI ​​Agent-based train autonomous operation adjustment method as described in any one of claims 1 to 6.

8. A computer device, the device comprising a memory and a processor, the memory being adapted to store a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by the processor to perform the AI ​​Agent-based train autonomous operation adjustment method as described in any one of claims 1 to 6.

9. A train autonomous operation adjustment system based on AI Agent, characterized in that, The adjustment system includes a control module configured to execute the AI ​​Agent-based train autonomous operation adjustment method as described in any one of claims 1 to 6.

Images