Energy-saving running chart based on large model and real-time adjustment method
By constructing and adjusting train timetables based on large models, the problem of insufficient energy-saving effect of traditional methods in complex scenarios is solved, realizing dynamic adjustment and energy optimization of train operation, and improving the transportation efficiency and safety of rail transit.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-16
- Publication Date
- 2026-03-10
AI Technical Summary
Traditional train timetable creation methods are difficult to achieve energy-saving effects and timeliness in complex scenarios, and automatic train operation systems cannot dynamically adjust timetables to cope with real-time interference.
A large-model-based approach is adopted to construct an operation diagram generation network by acquiring macro-operation information. The macro-operation diagram is generated by using line structure encoding, time-series feature extraction and energy consumption prediction modules. Combined with the operation constraint set and micro-level control strategy, the real-time speed and acceleration changes are calculated, and local replanning is triggered when there is a deviation from the planned time. Micro-level and safety control commands are integrated.
It achieves a balance between energy conservation and safety in train operation, and can dynamically adjust the train schedule in the face of real-time interference, thereby improving transportation efficiency and energy utilization.
Smart Images
Figure CN121626224A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of rail transit automation control and intelligent scheduling, in particular to an energy-saving operation diagram and real-time adjustment method based on a large model. BACKGROUND
[0002] With the continuous expansion of the scale of rail transit network and the continuous increase of train operation density, the automation and intelligent level of operation scheduling has a decisive influence on system energy consumption, transportation efficiency and safety. Traditional train operation diagram formulation usually relies on manual experience or deterministic optimization methods based on physical rules, such as the energy-saving curve generation mode of fixed acceleration-constant speed-idling-braking mode. However, such methods are difficult to fully consider the line slope, speed limit changes, vehicle parameters, passenger flow changes and energy consumption differences under different operating conditions, and their energy-saving effect and timeliness are often insufficient in complex scenarios.
[0003] On the other hand, although the automatic train operation system can automatically traction and braking according to the operation diagram, its optimization capability is usually limited to fixed algorithm mechanism, and cannot dynamically adjust the operation diagram when facing real-time interference (such as passenger flow fluctuation, temporary speed limit, influence of previous vehicle, etc.).
[0004] Therefore, it is necessary to improve SUMMARY
[0005] In order to solve the above technical problems, the present application provides an energy-saving operation diagram and real-time adjustment method based on a large model.
[0006] The above invention purpose of the present application is realized by the following technical scheme:
[0007] An energy-saving operation diagram and real-time adjustment method based on a large model, comprising the steps of:
[0008] obtaining macro operation information;
[0009] constructing a large model operation diagram generation network based on the macro operation information, using line structure coding, time sequence feature extraction and energy consumption prediction module to generate a macro operation diagram;
[0010] based on the generated macro operation diagram, constructing an operation constraint set;
[0011] The operation constraint set includes line speed limit, traction / braking power limit, energy consumption target and confidence index;
[0012] According to the operation constraint set and the current actual state of the train, the traction / braking action sequence is generated through the micro-level control strategy, and the real-time speed and acceleration change are calculated, and the speed trajectory is output;
[0013] The speed trajectory is smoothed a second time and reconstructed by combining the line speed limit and the energy consumption target;
[0014] When the time deviation between the planned time and the actual time exceeds the threshold, the local operation map replanning is automatically triggered based on the deviation amount, and the micro-control instructions are updated.
[0015] The micro-control commands and safety control commands are weighted and fused into a final control command, which is then output.
[0016] In a preferred embodiment, the step of constructing a large-scale model operation graph generation network based on the macroscopic operation information, and generating a macroscopic operation graph using line structure encoding, time series feature extraction, and energy consumption prediction modules, includes:
[0017] The segment reference speeds in the macroscopic operation chart are generated using linear interpolation, and the calculation method is as follows:
[0018]
[0019] in, The velocity interpolation increment is calculated from the difference between the starting velocity and the ending velocity of the segment.
[0020] i is the interpolation point number, i = 0, 1, 2, ..., M, where M is the total number of interpolation points in the segment.
[0021] In a preferred embodiment, the step of constructing a large-scale model operation graph generation network based on the macroscopic operation information, and generating a macroscopic operation graph using line structure encoding, time series feature extraction, and energy consumption prediction modules, further includes:
[0022] The energy consumption evaluation of the macroscopic operation diagram is calculated using a segmented power accumulation method, and the calculation formula is as follows:
[0023]
[0024] in, The traction / braking power of the train at the k-th sampling time is obtained by multiplying the traction force and the speed. Where N is the sampling time interval and N is the total number of samples.
[0025] In a preferred embodiment, the step of generating a traction / braking sequence based on the set of operational constraints and the current actual state of the train using a micro-level control strategy, calculating real-time speed and acceleration changes, and outputting a speed trajectory includes:
[0026] Acceleration is calculated using velocity difference, as shown in the following formula:
[0027]
[0028] in, and These represent the velocities at the next and current moments, respectively.
[0029] Using a fixed time step, the real-time acceleration change trend can be obtained.
[0030] In a preferred embodiment, the step of generating a traction / braking sequence based on the set of operational constraints and the current actual state of the train using a micro-level control strategy, calculating real-time speed and acceleration changes, and outputting a speed trajectory includes:
[0031] The arrival time deviation penalty value of the macroscopic operation chart is calculated in the following way:
[0032]
[0033] in, This refers to the train's actual arrival time at the station.
[0034] The target arrival time is planned for the macro-operational schedule.
[0035] In a preferred embodiment, the secondary smoothing of the speed trajectory and the reconstruction in conjunction with the line speed limit and the energy consumption target include:
[0036] The trajectory smoothing process uses a moving average algorithm for velocity reconstruction, and the calculation method is as follows:
[0037]
[0038] Where L is the length of the smoothing window, representing the number of velocity points participating in the averaging;
[0039] The sampling point velocity within the window;
[0040] The
[0041] Confidence of large model output The formula for dynamically adjusting the length of the smoothing window is as follows:
[0042]
[0043] in, K is the base smooth window length; K is the maximum expansion length.
[0044] In a preferred embodiment, the step of automatically triggering local runtime replanning and updating micro-control instructions based on the deviation when the planned time and actual time exceed a threshold includes:
[0045] The formula for calculating time deviation is as follows:
[0046]
[0047] in, This is the train's current actual running time;
[0048] The reference time for the running chart.
[0049] In a preferred embodiment, the step of automatically triggering local runtime replanning and updating micro-control instructions based on the deviation when the planned time and actual time exceed a threshold further includes:
[0050] When satisfied , When the threshold condition is met, a local runtime replanning is triggered, and the new target velocity value is updated in the following way:
[0051]
[0052] in, The original running chart reference speed;
[0053] The adjustment coefficient is set according to the line conditions.
[0054] In a preferred embodiment, the step of automatically triggering local runtime replanning and updating micro-control instructions based on the deviation when the planned time and actual time exceed a threshold further includes:
[0055] The micro-control command update adopts a progressive incremental update method, and the calculation method is as follows:
[0056]
[0057] in, The incremental change in the control quantity is output by the micro-level control strategy;
[0058] k is the incremental number.
[0059] In a preferred embodiment, the step of fusing the micro-control commands and safety control commands into a final control command through a weighted method and then outputting it includes:
[0060] The weighted fusion into the final control command includes:
[0061]
[0062] in, To incorporate weights, adjustments are made dynamically based on operational status and confidence level.
[0063] The micro-control command is the traction / braking command output by the micro-level control strategy;
[0064] These are safety control commands generated based on train speed limits, acceleration, and safety constraints.
[0065] The second objective of this invention is achieved through the following technical solution:
[0066] An energy-saving operation diagram and real-time adjustment system based on a large model includes:
[0067] Module 1: Obtaining Macro-level Operational Information;
[0068] The second module: Based on the macroscopic operation information, a large model operation graph generation network is constructed, and the macroscopic operation graph is generated using the line structure encoding, time series feature extraction and energy consumption prediction modules;
[0069] The third module: Based on the generated macroscopic operation graph, construct the set of operational constraints;
[0070] The fourth module: The set of operational constraints includes line speed limits, traction / braking power limits, energy consumption targets, and confidence indices;
[0071] The fifth module: Based on the set of operational constraints and the current actual state of the train, it generates a traction / braking action sequence through a micro-level control strategy, calculates real-time speed and acceleration changes, and outputs the speed trajectory;
[0072] Module 6: Perform secondary smoothing on the speed trajectory and reconstruct it by combining the line speed limit and the energy consumption target;
[0073] Module 7: When the time deviation between the planned time and the actual time exceeds the threshold, the local running map replanning is automatically triggered based on the deviation amount, and the micro-control instructions are updated.
[0074] Module 8: The micro-control commands and safety control commands are weighted and fused into a final control command, which is then output.
[0075] The above-mentioned objective three of this application is achieved through the following technical solution:
[0076] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described energy-saving operation diagram and real-time adjustment method based on a large model.
[0077] The fourth objective of this application is achieved through the following technical solution:
[0078] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described energy-saving operation diagram and real-time adjustment method based on a large model.
[0079] In summary, this application includes at least one of the following beneficial technical effects:
[0080] This invention provides a method for energy-saving operation diagrams and real-time adjustments based on a large model. The method includes acquiring macro-level operational information required for train operation, such as track gradient curves, speed limit curves, vehicle dynamics parameters, historical operation diagrams, energy-saving operation samples, passenger flow status, and operating timetable requirements. Based on the acquired data, a large model operation diagram generation network is constructed. Through track structure encoding, temporal feature extraction, and energy consumption estimation modules, a macro-level energy-saving operation diagram is generated, including reference speed, inter-station travel time, stop time, and section passage time. Based on the macro-level operation diagram, an operational constraint set is further generated, including speed limits, acceleration limits, traction / braking power limits, energy consumption targets, and a confidence index output by the large model. Based on the operational constraint set and the train's current real-time speed, acceleration, and position, a traction / braking action sequence is generated through a micro-control strategy, and the acceleration is calculated using a difference formula to achieve continuous updates to the train's dynamic state. The speed sequence generated by the micro-control is smoothed, and based on the track speed limit, energy consumption target, and confidence index Γ... The speed curve is reconstructed to ensure that the final operating speed meets the combined requirements of safety, stability and energy saving. When the actual train operation time deviates from the reference timetable by more than a preset threshold, the local timetable replanning module is triggered to correct the reference speed according to the deviation and automatically generate a new target speed trajectory. Finally, the micro-control commands and safety control commands are fused in a weighted manner to generate the final traction / braking output of the train, thereby achieving the unity of energy saving and safety goals. Attached Figure Description
[0081] Figure 1 This is the overall flowchart of this application;
[0082] Figure 2 This is a flowchart of an embodiment of S2 in this application;
[0083] Figure 3 This is a flowchart of embodiments S4-S5 of this application;
[0084] Figure 4 This is a flowchart of an embodiment of S6 of this application. Detailed Implementation
[0085] The following is in conjunction with the appendix Figures 1-4 This application will be described in further detail.
[0086] In the example, such as Figure 1 As shown, this application discloses an energy-saving operation diagram and real-time adjustment method based on a large model, which specifically includes the following steps:
[0087] S1: Obtain macroscopic operational information;
[0088] S2: Based on the macroscopic operation information, a large model operation graph generation network is constructed, and the macroscopic operation graph is generated using the line structure encoding, time series feature extraction and energy consumption prediction modules;
[0089] S3: Based on the generated macroscopic operation map, construct an operation constraint set, which includes line speed limit, traction / braking power limit, energy consumption target and confidence index;
[0090] S4: Based on the set of operating constraints and the current actual state of the train, generate a traction / braking action sequence through a micro-level control strategy, calculate real-time speed and acceleration changes, and output the speed trajectory;
[0091] S5: Perform secondary smoothing on the speed trajectory and reconstruct it by combining the line speed limit and the energy consumption target;
[0092] S6: When the time deviation between the planned time and the actual time exceeds the threshold, the local running chart replanning is automatically triggered based on the deviation amount, and the micro-control instructions are updated.
[0093] S7: The micro-control instructions and safety control instructions are weighted and merged into a final control instruction, which is then output.
[0094] In this embodiment, the macroscopic operation information includes line conditions, train dynamic parameters, gradient curves, speed limit curves, historical timetables, energy-saving operation samples, passenger flow status, and operating timetable requirements; the macroscopic operation diagram includes the macroscopic operation diagram of running time, reference speed, station dwell time, and section passage time.
[0095] Figure 2 S2 includes:
[0096] S21: The segment reference speeds in the macroscopic operation chart are generated using linear interpolation, and the calculation method is as follows:
[0097] ;
[0098] S22: Wherein, The velocity interpolation increment is calculated from the difference between the starting velocity and the ending velocity of the segment.
[0099] S23: i is the interpolation point number, i = 0, 1, 2, ..., M, where M is the total number of interpolation points in the segment.
[0100] S24: The energy consumption evaluation of the macroscopic operation diagram is calculated using a segmented power accumulation method, and the calculation formula is as follows:
[0101] ;
[0102] S25: Among them, The traction / braking power of the train at the k-th sampling time is obtained by multiplying the traction force and the speed.
[0103] S26: Where N is the sampling time interval and N is the total number of samples.
[0104] In this embodiment, the use of a large model to predict and optimize the global operation graph at this stage is the core link of the present invention to achieve "multi-source data-driven energy-saving operation".
[0105] 1. Overall Model Structure
[0106] The large model consists of four parts:
[0107] Multimodal coding layer: Encodes line geometry, dynamic parameters, historical patterns, and passenger flow information into vector sequences respectively;
[0108] Temporal inference layer: Employs a Transformer architecture to enable the model to capture long-range dependencies, such as the impact of long-term slope on running time and the impact of passenger flow on braking distance;
[0109] Energy consumption prediction branch: Based on the encoded time-series features, power and energy consumption prediction is performed on each speed sequence segment to calculate the energy consumption target E;
[0110] Confidence estimation branch: Outputs confidence Γ by methods such as predicting variance or feature entropy, which is used for adaptive smoothing in S5.
[0111] 2. Generation of section reference speed
[0112] The model output provides the starting and ending velocities of the segment. The segment velocity curve is generated by interpolation using the following formula:
[0113]
[0114] Where Δv is the difference between the start and end speeds of the section divided by the number of interpolation points M. This method ensures that the speed curve is continuous, smooth, and meets the physical constraints of operation.
[0115] Engineering tips: In sections with steep slopes and frequent speed limit changes, a higher interpolation density can be set; a physical check should be performed on the speed changes, and if the speed exceeds the limit, Δv should be adjusted back.
[0116] 3. Energy Consumption Estimation
[0117] Energy consumption is calculated using a piecewise power integral formula:
[0118]
[0119] in F_k represents the traction / braking force.
[0120] 4. The role of confidence level output Γ
[0121] Confidence level is used to: determine whether the running graph is reliable; determine the length L of the S5 smoothing window; and serve as a micro-level heuristic signal to measure the aggressiveness of action updates.
[0122] Figure 3 S4 includes:
[0123] S41: The acceleration is calculated using velocity difference, as shown in the following formula: ;
[0124] S42: Wherein, and These represent the velocities at the next and current moments, respectively.
[0125] S43: Using a fixed time difference as the time step, the real-time acceleration change trend can be obtained.
[0126] S44: The arrival time deviation penalty value of the macroscopic operation chart is calculated in the following way: ;
[0127] S45: Among them, This refers to the train's actual arrival time at the station.
[0128] S46: The target arrival time is planned for the macro-operational schedule.
[0129] In this embodiment, at the micro-control layer, the system generates traction or braking force at each time step and calculates speed and acceleration in real time.
[0130] 1. State update and action solving
[0131] At each time step, the system calculates the traction / braking command u based on the current speed, position, constraint set, and macroscopic reference speed.
[0132] This action is solved using an MDP (Markov Decision Process) or an optimizer, while satisfying physical constraints.
[0133] 2. Acceleration Calculation
[0134] Acceleration is calculated using velocity difference:
[0135]
[0136] This formula has the following advantages: it is simple to implement in engineering; it closely matches real-time speed measurements; and it can be used for immediate speed limit and safety constraint judgments.
[0137] 3. Legality check of actions
[0138] For each control output The system needs to check whether the following conditions are met: maximum acceleration limit; maximum power limit; maximum deceleration safety limit (such as emergency braking curve).
[0139] If an illegality is detected, a forced rollback to security control will be initiated. .
[0140] Figure 3 S5 includes:
[0141] S51: The trajectory smoothing process uses a moving average algorithm for velocity reconstruction, and the calculation method is as follows: ;
[0142] S52: Where L is the length of the smoothing window, representing the number of velocity points participating in the averaging;
[0143] S53: The sampling point velocity within the window;
[0144] S54: Confidence level of the large model output The formula for dynamically adjusting the length of the smoothing window is as follows: ;
[0145] S55: Among them, Based on the smooth window length;
[0146] S56:K represents the maximum extended length.
[0147] In this embodiment, the present invention avoids high-frequency speed oscillations caused by micro-control through a smoothing algorithm, and ensures that the final speed curve can be used for real train operation.
[0148] 1. Moving average smoothing
[0149] This invention employs a sliding window for smoothing:
[0150]
[0151] L will be adaptively adjusted in the next step.
[0152] 2. Confidence-driven window length adjustment
[0153] The window length L is calculated as follows:
[0154]
[0155] Therefore, it can be seen that when Γ is high (the model is highly confident), L≈L0, and the response is sensitive; when Γ is low (the model is uncertain), L increases, and the trajectory is smoother and more robust.
[0156] 3. Secondary security check after smoothing
[0157] Smoothing may lead to: local overspeeding; deviation from the planned arrival time. Therefore, a secondary verification is required after smoothing, and the speed trajectory may need to be locally compressed or expanded if necessary.
[0158] Figure 4 S6 includes:
[0159] S61: The formula for calculating time deviation is as follows: ;
[0160] S62: Wherein, This is the train's current actual running time;
[0161] S63: For reference time in the running chart;
[0162] S64: When satisfied , When the threshold condition is met, a local runtime replanning is triggered, and the new target velocity value is updated in the following way: ;
[0163] S65: Among them, The original running chart reference speed;
[0164] S66: The adjustment coefficient is set according to the line conditions;
[0165] S67: The micro-control command update adopts a progressive incremental update method, and the calculation method is as follows: ;
[0166] S68: Among them, The incremental change in the control quantity is output by the micro-level control strategy;
[0167] S69: Incremental number.
[0168] In this embodiment, this stage ensures that the train can be dynamically adjusted according to disturbances and deviations during operation, achieving a balance between punctuality and energy efficiency.
[0169] 1. Deviation detection
[0170] Real-time system calculations:
[0171]
[0172] when Time triggers replanning.
[0173] 2. Speed correction of local running chart
[0174] Speed correction method:
[0175]
[0176] The setting of λ affects: recovery speed; comfort; power consumption. Segmented λ is generally used (with variations depending on long slopes or short sections).
[0177] 3. Final control fusion
[0178] Control fusion formula:
[0179]
[0180] Dynamic adjustment strategy: When the security risk is high: α is small, tending to... When the trajectory is stable: a larger α value enhances the energy-saving control effect.
[0181] This combination of safety and energy saving meets both the ATO and train control system interface specifications.
[0182] S7 includes:
[0183] S71: The weighted fusion into the final control command includes: ;
[0184] S72: Among them, To incorporate weights, adjustments are made dynamically based on operational status and confidence level.
[0185] S73: The micro-control command is the traction / braking command output by the micro-level control strategy;
[0186] S74: These are safety control commands generated based on train speed limits, acceleration, and safety constraints.
[0187] In this embodiment, the final control command is generated by fusing the energy-saving control output and the safety control module:
[0188]
[0189] α can be dynamically adjusted based on confidence level Γ, operating status, and risk level. The safety control module is responsible for: speed limit control; maximum acceleration / deceleration protection; and emergency braking management. It ensures that the system meets operational safety requirements while achieving energy savings.
[0190] Furthermore, the macroscopic reference velocity filling adopts a piecewise interpolation formula:
[0191]
[0192] In the above formula is the speed interpolation increment; i is the interpolation point number; N is the number of segmented interpolations. This formula can generate a continuous speed curve based on the start and end speeds of a segment, improving the usability of the timetable.
[0193] Furthermore, energy consumption estimation employs a piecewise power integral model:
[0194]
[0195] In the above formula Traction / braking power; The sampling interval is defined. This energy consumption model serves as the basis for evaluating the performance of the runtime graph.
[0196] Furthermore, the acceleration generated by the micro-control action is calculated using a differential form:
[0197]
[0198] This formula simplifies the acceleration calculation for real-time control, making the controller more suitable for high-frequency operation.
[0199] Furthermore, the speed smoothing process employs a sliding window:
[0200]
[0201] In the above formula, L is the length of the smoothing window, which is dynamically adjusted by the confidence level Γ.
[0202] Furthermore, the relationship between the smoothing window and the confidence level is as follows:
[0203]
[0204] In the above formula, L0 is the base window and K is the maximum expansion ratio.
[0205] Furthermore, the correction speed after replanning the running graph is:
[0206]
[0207] In the above formula, λ is the speed correction coefficient.
[0208] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.
[0209] In one embodiment, a large-model-based energy-saving operation chart and real-time adjustment system is provided, which corresponds one-to-one with the large-model-based energy-saving operation chart and real-time adjustment method described in the above embodiments. The large-model-based energy-saving operation chart and real-time adjustment system includes:
[0210] Module 1: Obtaining Macro-level Operational Information;
[0211] The second module: Based on the macroscopic operation information, a large model operation graph generation network is constructed, and the macroscopic operation graph is generated using the line structure encoding, time series feature extraction and energy consumption prediction modules;
[0212] The third module: Based on the generated macroscopic operation graph, construct the set of operational constraints;
[0213] The fourth module: The set of operational constraints includes line speed limits, traction / braking power limits, energy consumption targets, and confidence indices;
[0214] The fifth module: Based on the set of operational constraints and the current actual state of the train, it generates a traction / braking action sequence through a micro-level control strategy, calculates real-time speed and acceleration changes, and outputs the speed trajectory;
[0215] Module 6: Perform secondary smoothing on the speed trajectory and reconstruct it by combining the line speed limit and the energy consumption target;
[0216] Module 7: When the time deviation between the planned time and the actual time exceeds the threshold, the local running map replanning is automatically triggered based on the deviation amount, and the micro-control instructions are updated.
[0217] Module 8: The micro-control commands and safety control commands are weighted and fused into a final control command, which is then output.
[0218] Specific limitations regarding the energy-saving operation diagram and real-time adjustment system based on a large model can be found in the limitations of the energy-saving operation diagram and real-time adjustment method based on a large model mentioned above, and will not be repeated here. Each module in the aforementioned energy-saving operation diagram and real-time adjustment system based on a large model can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.
[0219] In one embodiment, a computer device is provided, which may be a server. The computer device includes a processor, memory, a network interface, and a database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage medium. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements an energy-saving operation diagram and real-time adjustment method based on a large model.
[0220] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps:
[0221] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:
[0222] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0223] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.
[0224] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.
Claims
1. A large model-based energy-saving operation graph and real-time adjustment method, characterized in that: macroscopic operation information is obtained; a large model operation graph generation network is constructed based on the macroscopic operation information, line structure coding, time series feature extraction, and energy consumption prediction modules are used to generate a macroscopic operation graph; based on the generated macroscopic operation graph, an operation constraint set is constructed; the operation constraint set includes line speed limit, traction / braking power limit, energy consumption target, and confidence index; according to the operation constraint set and the current actual state of the train, a traction / braking action sequence is generated through a micro-level control strategy, and real-time speed and acceleration changes are calculated to output a speed trajectory; the speed trajectory is subjected to secondary smoothing processing, and is reconstructed in combination with the line speed limit and the energy consumption target; when the time deviation between the planned time and the actual time exceeds a threshold value, local operation graph re-planning is automatically triggered according to the deviation amount, and the micro-control instruction is updated; the micro-control instruction and the safety control instruction are fused into a final control instruction through a weighted manner and output.
2. The large model-based energy-saving operation graph and real-time adjustment method according to claim 1, characterized in that, the large model-based energy-saving operation graph and real-time adjustment method, characterized in that: the segment reference speed of the macroscopic operation graph is generated by linear interpolation, and the calculation method is as follows: wherein, is the speed interpolation increment, calculated from the difference between the start and end speed of the segment; i is the interpolation point number, i=0, 1, 2, …, M, and M is the total number of interpolation points of the segment.
3. The method of claim 2, wherein the method further comprises: the large model-based energy-saving operation graph and real-time adjustment method, characterized in that: the energy consumption evaluation of the macroscopic operation graph is calculated by using the segmented power accumulation method, and the calculation formula is as follows: wherein, Pki is the traction / braking power of the train at the kth sampling time, obtained from the product of the traction force and the speed; is the sampling time interval, and N is the total number of samples.
4. The large model-based energy-saving operation graph and real-time adjustment method according to claim 1, characterized in that, the large model-based energy-saving operation graph and real-time adjustment method, characterized in that: the acceleration is calculated by differentiating the speed, and the formula is as follows: wherein, with respectively denote the speed at the next time instant and the current time instant. For time step, the real-time acceleration change trend can be obtained by fixed time difference.
5. The large model-based energy-saving operation graph and real-time adjustment method of claim 1, characterized in that: the arrival time deviation penalty value of the operation graph is calculated by the following method: wherein, is the actual arrival time of the train; is the target arrival time of the macroscopic train plan.
6. The large model-based energy-saving operation graph and real-time adjustment method according to claim 1, characterized in that, the large model-based energy-saving operation graph and real-time adjustment method, characterized in that: the trajectory smoothing processing adopts a sliding average algorithm for speed reconstruction, and the calculation method is as follows: where L is the smoothing window length, representing the number of speed points participating in the average; Vw is the velocity of the sampling point within the window; the Confidence of large model output For dynamically adjusting the smoothing window length, the formula is as follows: wherein is the base smoothing window length; K is the maximum spread length.
7. The large model-based energy-saving operation graph and real-time adjustment method according to claim 1, characterized in that, when the time deviation between the planned time and the actual time exceeds a threshold value, local operation graph re-planning is automatically triggered according to the deviation amount, and the micro-control instruction is updated, including: the time deviation calculation formula is as follows: wherein, is the current actual running time of the train; Reference time for the working diagram.
8. The large model-based energy-saving operation graph and real-time adjustment method according to claim 7, characterized in that, when the time deviation between the planned time and the actual time exceeds a threshold value, local operation graph re-planning is automatically triggered according to the deviation amount, and the micro-control instruction is updated, including: When the condition for the threshold is met , The local run graph is re-planned and the new target speed value is updated by: wherein, is the original schedule reference speed; are adjustment factors set according to line conditions.
9. The large model-based energy-saving operation graph and real-time adjustment method according to claim 8, characterized in that, when the time deviation between the planned time and the actual time exceeds a threshold value, local operation graph re-planning is automatically triggered according to the deviation amount, and the micro-control instruction is updated, including: The micro-control instruction update adopts a progressive incremental update mode, and the calculation mode is: wherein, is a control quantity variation increment, output by the micro-layer control strategy; k is an incremental number.
10. The large model-based energy-saving operation graph and real-time adjustment method according to claim 1, characterized in that, The micro-control instruction and the safety control instruction are fused into a final control instruction by a weighting mode and output, including: The micro-control instruction and the safety control instruction are fused into a final control instruction by a weighting mode and output, including: The micro-control instruction and the safety control instruction are fused into a final control instruction by a weighting mode and output, including: wherein, is a fusion weight, dynamically adjusted according to the operating state and the confidence level; The micro-control instruction is a traction / braking instruction output by the micro-layer control strategy. Safety control instructions generated based on train speed limits, acceleration, and safety constraints.