Urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization

Abstract

The invention discloses an urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization, and the method comprises the steps: effectively solving a problem of scheduling decision limitation caused by a single optimization objective in the prior art through comprehensive real-time data processing, multi-objective collaborative optimization and intelligent algorithm iteration solution; the optimal control of the overall operation energy consumption of the rail transit system is realized; the train operation energy efficiency, the regenerated energy utilization rate and the operation punctuality rate are effectively improved, meanwhile, the peak load of a traction power supply system is reduced, the overall energy utilization efficiency and the operation performance of rail transit are remarkably improved, and the dual purposes of energy conservation, consumption reduction and efficient operation of the system are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of urban rail transit operation scheduling, and in particular to an urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization. BACKGROUND

[0002] Urban rail transit has become an important infrastructure for relieving traffic pressure and promoting social and economic development in modern cities due to its efficient, low-pollution and large-capacity transportation characteristics. However, with the rapid expansion of urban rail transit network scale and the continuous increase of traffic volume, train operation energy consumption problem has gradually become prominent, becoming one of the key factors restricting the sustainable development of rail transit. At present, urban rail transit energy consumption optimization scheduling technology mainly focuses on optimizing train departure interval, regenerative braking energy recovery and utilization and dynamic passenger flow response.

[0003] CN113743828A discloses a kind of urban rail transit operation scheduling method, focus on through to direct current traction power supply system modeling and integration regenerative braking energy utilization constraint, to realize the effective control of energy consumption;And CN110901710A discloses a kind of dynamic passenger flow based on train scheduling method, through real-time passenger flow data prediction line interval full load rate and transfer volume, to dynamically adjust train operation plan;However, the above prior art focuses on single or local optimization target, for example, either focus on the utilization of regenerative braking energy, or take passenger flow change as the only optimization reference, lack of overall consideration of multi-objective coordinated optimization, it is difficult to simultaneously consider train operation energy consumption, operation punctuality, regenerative energy utilization efficiency and power supply load management and other multi-dimensional targets.

[0004] Specifically, the existing urban rail transit operation scheduling method has the following shortcomings: first, the optimization target is relatively single or limited, ignoring the interrelation between different targets, and failing to achieve overall optimal scheduling comprehensively;Second, the collection, processing and utilization of real-time data are insufficient, it is difficult to fully and effectively reflect the dynamic characteristics of rail transit system operation state, resulting in limited accuracy of scheduling decision;Third, in terms of solving method, it mainly relies on traditional single-objective optimization or simple mathematical programming solving method, which has insufficient calculation efficiency and real-time performance, and is difficult to meet the actual demand of fast response and efficient processing of large-scale data;In addition, although the existing technical solutions propose a response mechanism for regenerative braking energy utilization and dynamic passenger flow, they lack comprehensive consideration of important factors such as peak load and real-time electricity price fluctuation, resulting in limitations in actual operation effect and failing to fully tap the potential of rail transit energy saving and consumption reduction.

[0005] Therefore, the existing urban rail transit energy consumption scheduling technology generally has the problem of being difficult to simultaneously satisfy multi-objective coordinated optimization and real-time efficient decision-making, and a comprehensive scheduling method considering reduction of traction energy consumption, full recycling of regenerated energy, operation punctuality guarantee and reduction of power grid load is urgently needed, and the present application effectively solves the scheduling decision limitation problem caused by a single optimization target in the prior art through comprehensive real-time data processing, multi-objective coordinated optimization and intelligent algorithm iterative solution, and realizes optimal control of the overall operation energy consumption of the rail transit system. SUMMARY

[0006] This section is intended to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments, and some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, the abstract and the title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0007] In view of the above-mentioned existing problems, the present application is proposed.

[0008] To solve the above technical problems, the present application provides the following technical solutions: collecting train operation conditions, traction power supply parameters, passenger flow space-time distribution and time-of-use electricity price data synchronously at the train side, the line side and the power supply side, and sequentially performing cleaning, integration, conversion and protocol, and outputting a standardized scheduling input data set;

[0009] Taking the data set as input, an energy consumption scheduling model is established, which takes minimization of traction energy consumption, maximization of regenerated energy utilization rate, maximization of operation punctuality rate and minimization of peak load as targets, and takes speed curve, departure interval and energy storage charging and discharging power as decision variables, and is constrained by train safety, line capacity and power supply capacity;

[0010] An artificial intelligence multi-objective optimization algorithm is used to iteratively solve the energy consumption scheduling model to obtain a candidate scheduling scheme set that satisfies real-time constraints;

[0011] According to a rolling horizon evaluation function, the target weight is adaptively determined, the candidate scheduling scheme set is comprehensively evaluated, the optimal scheduling scheme is selected, and the optimal scheduling scheme is real-time issued to the vehicle control system.

[0012] As a preferred scheme of the urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization according to the present application, the output standardized scheduling input data set comprises:

[0013] The train traction current, braking force and speed curve are sampled at a fixed period, and sliding window filtering and missing interpolation are used to remove noise and null values;

[0014] The passenger flow density data is resampled according to a unified mileage time index, and time-space alignment is completed with the train operation data.

[0015] Obtain the feeder voltage, current and time-of-use electricity price through the power supply monitoring interface, and map them to the train data timeline synchronously;

[0016] Interval normalization is performed on all numerical fields, one-hot encoding is performed on category fields, and principal component compression is used on high-dimensional continuous features to retain main information.

[0017] After the data consistency check passes, the train side, line side and power supply side features are merged according to the sampling time, written into a distributed cache, and a standardized scheduling input data set is output.

[0018] As a preferred scheme of the urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization, the energy consumption scheduling model comprises:

[0019] The traction energy consumption, regenerative energy utilization rate, operation punctuality rate and peak load are respectively set as independent optimization targets, and a Pareto relationship is used to retain the target trade-off space;

[0020] The train speed curve is discretized into section target speed pairs, and each pair is constrained by line speed limit and train acceleration capability;

[0021] The departure interval is set as a adjustable timetable parameter, and the lower limit is the sum of the interval running time and the minimum safety tracking interval;

[0022] The energy storage charging and discharging power is limited within the rated power capacity curve of the energy storage device, and is associated with the regenerative braking energy recovery task;

[0023] Meanwhile, train tracking safety constraints, power supply system load margin constraints and bus voltage stability constraints are introduced to form a multi-objective scheduling model description file coupled with trains, lines and power supply, i.e. the energy consumption scheduling model.

[0024] As a preferred scheme of the urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization, the energy consumption scheduling model is iteratively solved by using an artificial intelligence multi-objective optimization algorithm, comprising:

[0025] In the offline stage, historical operation data is used to train the deep neural network predictor to estimate each target function value at a millisecond level;

[0026] In the online stage, the predictor output is used as the fitness function, and an improved fast non-dominated sorting evolutionary algorithm is used to initialize and update the solution set;

[0027] Adaptive crossover and mutation are performed on each generation of population, and an incremental evaluation strategy is used to calculate the target value of the changed part only, thereby shortening the calculation time;

[0028] When the non-dominated solution set is stable, output the Pareto frontier solution set satisfying the real-time constraints as the candidate scheduling scheme set.

[0029] As a preferred scheme of the urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization, the candidate scheduling scheme set specifically includes:

[0030] A target speed curve file corresponding to each train section;

[0031] A train departure plan file specifying the departure time and interval;

[0032] A storage device charge-discharge power time sequence file for energy recovery and peak shaving;

[0033] Each optimization target value and constraint satisfaction flag associated with each scheme.

[0034] As a preferred scheme of the urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization, the target weight is adaptively determined according to the rolling time domain evaluation function, including:

[0035] Real-time monitoring of passenger flow prediction error, energy price fluctuation and power supply load rate in the rolling time domain, and generating corresponding dynamic factors;

[0036] Setting the basic weights of traction energy consumption, regenerative energy utilization rate, running punctuality rate and peak load according to the operation priority table;

[0037] Multiplying the dynamic factors by the basic weights and normalizing the whole to make the sum of the weights equal to 1;

[0038] When the deviation of any index exceeds the threshold value for two consecutive periods, triggering the emergency weight correction strategy, increasing the target index weight and simultaneously reducing the weight of other indexes.

[0039] As a preferred scheme of the urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization, the candidate scheduling scheme set is comprehensively evaluated according to the target weight, including:

[0040] According to the target weight, calculating the weighted total score of each candidate scheme;

[0041] Mapping the weighted score to the interval of 0-1 using the fuzzy satisfaction function, and sorting by satisfaction;

[0042] If the difference between the two schemes with the highest satisfaction is below the set threshold, then compare the regenerative energy utilization rate, and prefer the one with higher utilization rate;

[0043] If it still cannot be distinguished, then perform step-by-step discrimination according to the priority order of peak load, punctuality rate and traction energy consumption, until a unique optimal scheme is generated.

[0044] As a preferred scheme of the urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization, the optimal scheduling scheme is selected from the comprehensive evaluation results, and is issued to the vehicle control system in real time, including:

[0045] The candidate optimal scheme is executed on the digital twin simulation platform for rapid safety checking, and the checking content at least includes train collision checking, power supply load margin verification and communication compatibility verification;

[0046] After the checking is passed, the optimal scheme is translated into a speed-time curve file, a departure plan file and a energy storage control instruction file, and is issued to the vehicle control system and the power supply control center through the train communication network and the power supply scheduling network respectively.

[0047] The beneficial effects of the present application are: the present application realizes the comprehensive optimization of the energy consumption scheduling of the urban rail transit system by comprehensively applying multi-source data acquisition and processing technology, multi-objective optimization modeling method, artificial intelligence rapid iteration solution and rolling time domain evaluation strategy and other technical means, effectively improves the energy efficiency, regenerative energy utilization rate and operation punctuality rate of the train operation, and reduces the peak load of the traction power supply system, significantly improves the overall energy utilization efficiency and operation performance of the rail transit, and realizes the dual goals of system energy saving and high efficiency operation. BRIEF DESCRIPTION OF DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0049] Figure 1 The flowchart of the urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present application, not all embodiments.

[0051] Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0052] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. However, it will be apparent to one skilled in the art that the present application can be practiced without the specific details set forth in this description, that the present application can be practiced with other systems, and that the present application can be practiced using different techniques. Therefore, the present application is not limited to the embodiments described herein but rather the scope of the present application is to be given by the appended claims and their equivalents.

[0053] According to an embodiment of the present application, in combination with Figure 1 The flowchart shown in the figure is a city rail energy consumption scheduling method based on artificial intelligence multi-objective optimization, which specifically includes the following steps:

[0054] S1, synchronously collect train operation conditions, traction power supply parameters, passenger flow space-time distribution and time-of-use electricity price data on the train side, line side and power supply side, and sequentially perform cleaning, integration, conversion and protocol, and output standardized scheduling input data set. It needs to be explained that in this step:

[0055] The train traction current, braking force and speed curve are sampled at a fixed period, and sliding window filtering and missing interpolation are used to remove noise and null values;

[0056] Resample the passenger flow density data according to the unified mileage time index, and complete the time-space alignment with the train operation data;

[0057] Obtain the feeder voltage, current and time-of-use electricity price through the power supply monitoring interface, and map it to the train data time axis synchronously;

[0058] Perform interval normalization on all numerical fields, perform one-hot encoding on categorical fields, and use principal component compression on high-dimensional continuous features to retain main information;

[0059] After the data consistency check is passed, the train side, line side and power supply side features are merged according to the sampling time, written into the distributed cache, and the standardized scheduling input data set is output.

[0060] As an example, the train side collects:

[0061] Sensor: traction current, braking force, speed meter;

[0062] Sampling period: 50ms;

[0063] Data format: 32-bit floating point;

[0064] Value range: current 0-4kA, speed 0-120km / h.

[0065] As an example, the line side collects:

[0066] Sensor: video millimeter wave passenger flow detection unit, output passenger flow density and platform identification (string);

[0067] Sampling period: 30s.

[0068] As an example, the power supply side collects:

[0069] Interface: SCADA OPC-UA, fields at least including feeder voltage, feeder current, time-of-use electricity price;

[0070] Sampling period: 60s.

[0071] Exemplarily, the data cleaning of the collected data includes: applying 5-point sliding window median filtering to the traction current and speed curve, if the current value deviates from the window median ± 3 times the root mean square, it is marked as abnormal and set to missing; if the length of continuous missing is ≤3 sampling periods, linear interpolation is performed; if the length of continuous missing is >3 sampling periods, forward filling is performed and the “quality_flag” is marked as “LOW”;

[0072] Further, the integrated processing is performed on the cleaned data, including:

[0073] Unix millisecond timestamp is uniformly used, and all sampling records are rounded up to the nearest 10ms boundary;

[0074] The passenger flow density is mapped to the train mileage-time coordinate system according to the station number-station mileage index, so as to realize the alignment of train-passenger flow space;

[0075] Further, the data conversion is performed on the integrated data, including:

[0076] The continuous fields such as traction current and speed are executed 0-1 interval scaling;

[0077] The discrete fields such as station number and power supply partition are one-hot encoded to generate a Boolean vector;

[0078] Finally, the reduction is performed on the converted data features, including: using principal component compression on the working condition statistical vector ≥50 dimensions, and retaining the first k dimensions (such as k=10) whose cumulative explained variance is ≥95%.

[0079] It should be noted that the data consistency check in the embodiment includes checking the monotonicity of the timestamp, the legality of the value range and the consistency of the station, if the record failure rate >0.1%, an alarm is triggered and written into the audit log;

[0080] Finally, the standardized scheduling input data set is generated, which is represented by a structured format (Parquet), and an example is as follows:

[0081] {

[0082] “ts”: int64, “train_id”: string, “km_post”: float,

[0083] "speed_norm": float, "traction_curr_norm": float,

[0084] "passenger_density": float, "price_norm": float,

[0085] "pca_feat_1"... "pca_feat_k": float, "quality_flag": string

[0086] }

[0087] The dataset is stored in the HDFS partition directory / rail / standardized / yyyyMMdd / HH / ;

[0088] And synchronously write to Redis cache (key: ts|train_id, expire 10min) for subsequent model real-time calling.

[0089] It should be noted that this step effectively integrates multi-dimensional data such as train running state, passenger flow trend and power supply condition, which is used to comprehensively reflect the real-time running state of urban rail transit system, which is conducive to subsequent optimization model accurate modeling, ensures that data-driven decision-making has higher credibility and accuracy, and improves the reliability of data basis for energy scheduling decision-making.

[0090] S2, with the dataset as input, an energy consumption scheduling model is established, which minimizes the traction energy consumption, maximizes the regenerative energy utilization rate, maximizes the operation punctuality rate, and minimizes the peak load, with the speed curve, departure interval and energy storage charging and discharging power as decision variables, and is subject to the constraints of train safety, line capacity and power supply capacity. Among them, it should be noted that this step:

[0091] The traction energy consumption, regenerative energy utilization rate, operation punctuality rate and peak load are set as independent optimization objectives respectively, and the Pareto relationship is used to reserve the objective trade-off space;

[0092] The train speed curve is discretized into section target speed pairs, each pair is subject to line speed limit and train acceleration capability constraints;

[0093] The departure interval is taken as the adjustable timetable parameter, and its lower limit is the sum of the interval running time and the minimum safety tracking interval;

[0094] The energy storage charging and discharging power is limited within the rated power capacity curve of the energy storage device, and is associated with the regenerative braking energy recovery task;

[0095] At the same time, train tracking safety constraints, power supply system load margin constraints and bus voltage stability constraints are introduced to form a multi-objective scheduling model description file for train-line-power supply coupling.

[0096] It should be noted that the above-mentioned multi-objective scheduling model description file is written into a model description file that can be directly parsed by the artificial intelligence multi-objective optimization algorithm as the final expression of the energy consumption scheduling model.

[0097] As an example, the mathematical expression of the energy consumption scheduling model is:

[0098]

[0099] Where u is the target speed decision vector of each sub-section, v is the departure interval decision vector of each train section, w is the charge and discharge power vector of the energy storage device at discrete moments, t is a continuous time variable, T is the length of the evaluation cycle, α(t) is the time weight function constructed based on the train load rate, P tr (u, t) is the traction power curve obtained by speed vector mapping, β(t) is the time weight function constructed based on the regenerative energy absorption rate, P re (w, t) is the regenerative energy absorption power curve obtained by energy storage power mapping, γ is the punctuality sensitivity coefficient, M is the number of evaluation sections, η k (v k ) is the departure interval v k The corresponding on-time arrival satisfaction function, P gr (u, w, t) is the grid-side instantaneous power curve after traction-energy storage coupling, P max is the upper limit of the power peak allowed by the power supply system, F(·) is a multi-objective function vector containing four components;

[0100] It should be further explained that in the comprehensive target vector F(u,v,w)=(F1,F2,F3,F4), the four components correspond to the following optimization objectives in sequence:

[0101] Traction energy consumption target:

[0102]

[0103] Renewable energy utilization rate target:

[0104]

[0105] On-time performance targets:

[0106]

[0107] Peak load target:

[0108]

[0109] The components of the comprehensive target vector F(u, v, w) after fractional normalization belong to the interval [0, 1], the smaller the first component value (F1) represents the lower the unit weighted time energy consumption, the greater the second component value (F2) represents the higher the regenerative energy utilization rate, the closer the third component value (F3) to 1 represents the higher the punctuality rate, and the smaller the fourth component value (F4) represents the lower the peak power load. When the four components simultaneously reach the set threshold interval, it is considered as an optimal solution.

[0110] Preferably, through this step, the synergistic optimization of multiple objectives is achieved, the internal correlation between different operation objectives is systematically considered, the systematization and integrity of rail transit dispatching decision are improved, the dispatching strategy is more accurately and efficiently guided to meet diversified operation demands, thereby obtaining a globally optimized operation strategy, ensuring operation safety while achieving the goal of energy saving and consumption reduction.

[0111] S3, using an artificial intelligence multi-objective optimization algorithm to iteratively solve the energy consumption scheduling model to obtain a candidate scheduling scheme set that satisfies real-time constraints. It should be noted that in this step:

[0112] (1) Offline predictor training

[0113] Collect the standardized scheduling input data set and the corresponding historical scheduling results of the last 12 months, automatically align the passenger flow peak, holidays and abnormal operation segments;

[0114] Select a deep neural network containing a time series convolution layer, a bidirectional gated recurrent layer and an attention aggregation layer for multi-task learning of the four objectives of traction energy consumption, regenerative energy utilization rate, operation punctuality rate and peak load;

[0115] Use a hierarchical hold-out strategy to divide the training set, validation set and test set, and jointly evaluate the model performance with the target mean square error and the consistency rate of the Pareto ranking;

[0116] When the average error of the four objectives is lower than the threshold set by the operator and the consistency rate is not less than 0.9, freeze the network parameters, export the "predictor-vN" model weight file, and load it in the online stage;

[0117] (2) Online population initialization

[0118] Read the standardized input data segment within the last hour as the environment state vector at the beginning of each scheduling period;

[0119] According to the environment state vector, call "predictor-vN" to quickly sample and evaluate the speed curve, departure interval and energy storage power, forming 200 initial individuals;

[0120] Filter the initial individuals by train safety, line capacity and power supply capacity constraints, remove solutions that do not meet the constraints, and retain about 150 legal individuals as the population G0;

[0121] (3) Adaptive crossover-mutation and incremental evaluation

[0122] For the population G t , retrieve the historical good solution fragments of the approximate environment in the running account, compare with the variance of the target components in the current population, and dynamically set the crossover probability P c and the mutation probability P m ( high variance → low P c , high P m ; low variance → high P c , low P m ) ;

[0123] For the new individuals generated by crossover and mutation, only recalculate the time slice target affected by the change variable, and the unchanged part inherits the parent's evaluation;

[0124] Merge the new and old individuals, maintain diversity using fast non-dominated sorting and crowding distance, and intercept the first 150 individuals to generate the next generation G {t+1} ;

[0125] (4) Convergence monitoring and real-time constraint judgment

[0126] When the similarity of the Pareto frontiers of the last two generations is greater than or equal to 0.95 and the total calculation time has exceeded the scheduling period × 0.6, the algorithm is considered to have converged;

[0127] Perform hard constraint checking and safety timing simulation on all individuals in the converged frontier, and remove schemes that cause following conflicts and voltage drops;

[0128] If the number of candidate solutions after removal is less than 5, reduce the similarity threshold and repeat steps (3) to (4) until the number requirement is met;

[0129] (5) Output candidate scheduling scheme set

[0130] Generate three types of instruction files for each candidate solution:

[0131] Target speed curve file: CSV format, fields include section number, distance from station, target speed;

[0132] Train departure plan file: JSON format, records train ID, departure time, and interval between trains in the section;

[0133] Energy storage power sequence file: Parquet format, fields include time stamp, power value and working mode identifier (charge / discharge / standby) ;

[0134] The calculation and accompanying four target standardization values and "constraint_pass" Boolean label are uniformly packaged as a "solution package-UID";

[0135] All solution packages are stored in the dispatch buffer and a "candidate solution ready" event is published through the message queue for the rolling evaluation module in step S4 to call.

[0136] As an example, the candidate solution set includes:

[0137] Solution UID = A17F-3B2: traction energy consumption 0.42, regenerative utilization 0.83, punctuality rate 0.91, peak load 0.35, constraint_pass = true;

[0138] Solution UID = D04C-8E1: traction energy consumption 0.40, regenerative utilization 0.78, punctuality rate 0.89, peak load 0.38, constraint_pass = true;

[0139] Solution UID = F92B-5D0: traction energy consumption 0.45, regenerative utilization 0.85, punctuality rate 0.93, peak load 0.33, constraint_pass = true.

[0140] Preferably, the present embodiment more efficiently searches the optimal solution space by using artificial intelligence algorithms than traditional mathematical programming methods, and fully utilizes historical data to train artificial intelligence predictors, greatly shortens the calculation time, improves the real-time response ability of the algorithm and the feasibility of the scheduling solution, so that the solution solving process can adapt to the changes in the operation state in real time, greatly enhancing the dynamic adaptability of the scheduling decision.

[0141] S4, according to the rolling time domain evaluation function, the target weight is determined adaptively, the candidate scheduling solution set is comprehensively evaluated, the optimal scheduling solution is selected and is issued to the vehicle control system in real time. Wherein, the present step needs to be explained:

[0142] Real-time monitoring of passenger flow prediction error, energy price fluctuation and power supply load rate in the rolling time domain, and generating corresponding dynamic factors;

[0143] According to the operation priority table, the basic weights of traction energy consumption, regenerative energy utilization rate, running punctuality rate and peak load are set;

[0144] After multiplying the dynamic factors and the basic weights, the whole is normalized, so that the sum of the weights is 1;

[0145] When the deviation of any index exceeds the threshold value for two consecutive periods, the emergency weight correction strategy is triggered, the target index weight is increased and the weights of other indexes are simultaneously reduced;

[0146] According to the target weight, a weighted total score is calculated for each candidate scheme;

[0147] The weighted score is mapped to the interval of 0-1 using a fuzzy satisfaction function, and ranked by satisfaction;

[0148] If the difference between the two schemes with the highest satisfaction is below a set threshold, then the regenerative energy utilization rate is compared, and the one with higher utilization rate is preferred;

[0149] If still cannot be distinguished, then the peak load, punctuality rate, and traction energy consumption are sequentially judged according to their priority order, until a unique optimal scheme is generated;

[0150] The candidate optimal scheme is executed on a digital twin simulation platform for rapid safety checking, which at least includes train-to-train conflict checking, power supply load margin verification, and communication compatibility verification;

[0151] After the checking is passed, the optimal scheme is translated into a speed-time curve file, a departure plan file, and an energy storage control instruction file, and is respectively issued to the vehicle control system and the power supply control center through the train communication network and the power supply dispatching network.

[0152] Preferably, the present application can respond to changes in real-time operating conditions such as passenger flow fluctuations, electricity price changes, and load demand by dynamically adjusting the evaluation weight, ensuring the flexibility and adaptability of scheduling decisions, ensuring that scheduling decisions are accurately and efficiently executed, and significantly improving the efficiency of rail transit system operation.

[0153] The aforementioned method of executing rapid safety checking on the candidate optimal scheme on a digital twin simulation platform can be performed using existing techniques and means, which will not be described in detail in this example.

[0154] It should be noted that the above examples are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and they should be included in the scope of the claims of the present application.

Claims

1. An urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization, characterized in that: include: Data on train operating conditions, traction power supply parameters, passenger flow temporal and spatial distribution, and time-of-use electricity prices are collected synchronously on the train, line, and power supply sides. Data is then cleaned, integrated, converted, and reduced in sequence to output standardized scheduling input data sets. Using the dataset as input, an energy consumption scheduling model is established with the goals of minimizing traction energy consumption, maximizing regenerative energy utilization, maximizing on-time operation, and minimizing peak load. The model uses speed curves, departure intervals, and energy storage charging and discharging power as decision variables, and is subject to constraints such as train safety, line capacity, and power supply capacity. Iteratively solving the energy consumption scheduling model using an artificial intelligence multi-objective optimization algorithm to obtain a set of candidate scheduling solutions that meet real-time constraints; The target weight is adaptively determined according to the rolling horizon evaluation function, the candidate scheduling scheme set is comprehensively evaluated, the optimal scheduling scheme is selected and sent to the vehicle control system in real time.

2. The urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization according to claim 1 is characterized in that: The output standardized scheduling input data set includes: The train traction current, braking force and speed curves are sampled at a fixed period, and sliding window filtering and missing interpolation are used to remove noise and null values; Resample the passenger flow density data according to a unified mileage-time index and complete time-space alignment with the train operation data; Obtain feeder voltage, current, and time-of-use electricity price through the power supply monitoring interface and synchronously map them to the train data timeline; Perform interval normalization on all numerical fields, perform one-hot encoding on categorical fields, and then use principal component compression on high-dimensional continuous features to retain the main information; After the data consistency check is passed, the train side, line side and power supply side features are merged according to the sampling time, written into the distributed cache, and the standardized scheduling input data set is output.

3. The urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization according to claim 1 or 2 is characterized in that: The establishing of the energy consumption scheduling model includes: Traction energy consumption, regenerative energy utilization, on-time operation rate, and peak load are set as independent optimization targets, and the Pareto relationship is used to preserve the trade-off space of each target. The train speed profile is discretized into segment target speed pairs, each of which is constrained by the line speed limit and the train acceleration capability; The departure interval is used as an adjustable timetable parameter, and its lower limit is the sum of the interval running time and the minimum safe tracking interval; Limit the energy storage charging and discharging power to the rated power capacity curve of the energy storage device and associate it with the regenerative braking energy recovery task; At the same time, train tracking safety constraints, power supply system load margin constraints and bus voltage stability constraints are introduced to form a multi-objective scheduling model description file for train-line-power supply coupling, which is the energy consumption scheduling model.

4. The urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization according to claim 3 is characterized in that: The energy consumption scheduling model is iteratively solved using an artificial intelligence multi-objective optimization algorithm, including: In the offline phase, historical operational data is used to train a deep neural network predictor to estimate the values ​​of each objective function in milliseconds. In the online stage, the predictor output is used as the fitness function, and the improved fast non-dominated sorting evolutionary algorithm is used to initialize and update the solution set; Perform adaptive crossover and mutation on each generation of the population, and use an incremental evaluation strategy to calculate only the target value of the changed part, shortening the calculation time; When the non-dominated solution set is stable, the Pareto front solution set that meets the real-time constraints is output as the candidate scheduling solution set.

5. The urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization according to claim 4 is characterized in that: The candidate scheduling solution set specifically includes: Target speed curve files corresponding to each train section; Train departure schedule document that clearly specifies departure times and intervals between trains; Time series files of charge and discharge power of energy storage devices used for energy recovery and peak shaving; The optimization objective values ​​and constraint satisfaction marks associated with each solution.

6. The urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization according to claim 1 is characterized in that: The target weight is adaptively determined based on the rolling horizon evaluation function, including: Monitor passenger flow forecast errors, energy price fluctuations, and power supply load rates in real time within a rolling time domain and generate corresponding dynamic factors; Set basic weights for traction energy consumption, regenerative energy utilization, on-time operation rate, and peak load according to the operational priority table; Multiplying the dynamic factor by the basic weight and normalizing the whole so that the sum of the weights is 1; When the deviation of any indicator exceeds the threshold for two consecutive periods, the emergency weight correction strategy is triggered to increase the weight of the target indicator and simultaneously reduce the weights of other indicators.

7. The urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization according to claim 1 or 6 is characterized in that: Based on the target weights, a comprehensive evaluation is performed on the candidate scheduling solution set, including: Calculating a weighted total score for each candidate solution according to the target weight; Use fuzzy satisfaction function to map weighted scores to the range of 0 to 1 and sort them by satisfaction; If the difference between the two solutions with the highest satisfaction is lower than the set threshold, the utilization rate of renewable energy is compared and the one with higher utilization rate is preferred; If it is still impossible to distinguish, the system will be judged step by step in the order of priority of peak load, punctuality and traction energy consumption until the only optimal solution is generated.

8. The urban rail energy consumption scheduling method based on artificial intelligence multi-objective optimization according to claim 7 is characterized in that: The optimal dispatching plan is selected from the comprehensive evaluation results and sent to the vehicle control system in real time, including: Perform rapid safety verification on the candidate optimal solution on the digital twin simulation platform. The verification content includes at least inter-train conflict check, power supply load margin verification, and communication compatibility verification. After verification, the optimal solution will be translated into speed-time curve files, departure plan files and energy storage control instruction files, and sent to the vehicle control system and power supply control center through the train communication network and power supply dispatching network respectively.

Images