Train energy consumption prediction method, device and equipment and computer readable storage medium

Abstract

The invention discloses a train energy consumption prediction method, device and equipment and a computer readable storage medium, belongs to the field of railway vehicles, is used for performing energy consumption prediction on a train according to a current working condition, and solves the problem of poor train energy consumption prediction precision. According to the method, the motion characteristic parameter set of the target train under each working condition can be determined according to the kinematics sensing data sequence in the historical data of the target train, and then for any working condition of the target train, the energy consumption prediction model of the target train under the working condition is obtained through training according to the motion characteristic parameter set under the working condition. In the prediction process, the motion characteristic parameters of the target train in the last preset period can be input into the energy consumption prediction model corresponding to the current working condition of the target train, the energy consumption prediction result of the target train is obtained, energy consumption prediction is conducted in a working condition recognition mode, the train energy consumption prediction precision can be greatly improved, and the train energy consumption prediction efficiency is improved. And the guidance value of the train energy consumption prediction result is improved.

Description

Technical Field

[0001] The present invention relates to the field of rail vehicles, and in particular to a train energy consumption prediction method, device, equipment and computer-readable storage medium. Background Art

[0002] The prediction of train energy consumption is of great significance in cost control, train operation optimization and power grid planning. However, the relevant technology lacks a mature train energy consumption prediction method, resulting in poor accuracy in train energy consumption prediction and low guiding value of train energy consumption prediction results.

[0003] Therefore, how to provide a solution to the above technical problems is a problem that those skilled in the art need to solve at present. Summary of the invention

[0004] The purpose of the present invention is to provide a train energy consumption prediction method, device, equipment and computer-readable storage medium. In the present invention, a set of motion characteristic parameters of a target train under various working conditions can be determined based on a kinematic sensor data sequence in historical data of the target train. Then, for any working condition of the target train, an energy consumption prediction model of the target train under the working condition can be trained based on the set of motion characteristic parameters under the working condition. When predicting, the motion characteristic parameters of the target train in the previous preset period can be input into the energy consumption prediction model corresponding to the current working condition of the target train to obtain the energy consumption prediction result of the target train. Since energy consumption prediction is performed by means of working condition identification, the accuracy of train energy consumption prediction can be greatly improved, and the guiding value of the train energy consumption prediction result is improved.

[0005] In order to solve the above technical problems, the present invention provides a train energy consumption prediction method, comprising:

[0006] Determine a set of motion characteristic parameters of the target train under various working conditions according to a kinematic sensor data sequence in historical data of the target train, wherein the kinematic sensor data sequence includes kinematic sensor data of each sampling point;

[0007] For any working condition of the target train, an energy consumption prediction model of the target train under the working condition is trained according to a set of motion characteristic parameters under the working condition;

[0008] Determine the current operating condition of the target train;

[0009] The motion characteristic parameters of the target train in the last preset period are input into the energy consumption prediction model corresponding to the current working condition to obtain the energy consumption prediction result of the target train.

[0010] On the other hand, kinematic sensing data includes velocity and acceleration;

[0011] According to the kinematic sensor data sequence in the historical data of the target train, the motion characteristic parameter set of the target train under various working conditions is determined to include:

[0012] Based on a preset time interval, a kinematic sensor data sequence in the historical data of the target train is divided into a plurality of kinematic sensor data subsequences;

[0013] For any kinematic sensing data subsequence, determining a preset type of motion characteristic parameter of a time interval where the kinematic sensing data subsequence is located according to the kinematic sensing data of each sampling point in the kinematic sensing data subsequence;

[0014] Clustering each of the motion characteristic parameter subsets to obtain a motion characteristic parameter set under each working condition of the target train;

[0015] Among them, a single kinematic sensing data subsequence and its corresponding motion characteristic parameter are collectively regarded as a motion characteristic parameter subset, and the motion characteristic parameter set includes several motion characteristic parameter subsets.

[0016] On the other hand, clustering each of the motion characteristic parameter subsets to obtain a motion characteristic parameter set under each working condition of the target train includes:

[0017] Performing initial clustering on each of the motion characteristic parameter subsets to obtain a motion characteristic parameter set of the target train under each working condition after initial clustering;

[0018] Through correlation analysis and contribution analysis, the various features involved in the motion feature parameter subset are eliminated;

[0019] Secondary clustering is performed on each of the motion feature parameter subsets after feature elimination to obtain a motion feature parameter set under each working condition of the target train.

[0020] On the other hand, through correlation analysis and contribution analysis, the features involved in the motion feature parameter subset are eliminated, including:

[0021] For any feature in the motion feature parameter subset, remove the feature in the motion feature parameter subset whose correlation with the feature is higher than a first preset threshold;

[0022] For any feature in the motion feature parameter subset, the feature in the motion feature parameter subset whose contribution to the working condition clustering is lower than a second preset threshold is eliminated.

[0023] In another aspect, the correlation analysis method comprises Pearson correlation analysis;

[0024] The contribution analysis method includes: a fast Shapley additivity interpretation method.

[0025] On the other hand, for any working condition of the target train, according to the set of motion characteristic parameters under the working condition, the energy consumption prediction model of the target train under the working condition is trained to include:

[0026] For any working condition of the target train, a first energy consumption prediction model of the working condition is trained according to a set of motion characteristic parameters of the working condition;

[0027] Determining whether the prediction accuracy of the first energy consumption prediction model meets the standard;

[0028] If the target is reached, it ends;

[0029] If the standard is not met, the second energy consumption prediction model is trained according to the prediction result of the first energy consumption prediction model and the weight parameters of the first energy consumption prediction model, so that the second energy consumption prediction model is used as the energy consumption prediction model of the target train under the said working conditions.

[0030] On the other hand, the current working condition of the target train is determined to include:

[0031] Based on the kinematic sensor data of the target train in the previous preset period, the current working condition of the target train is determined through a pre-trained language model.

[0032] In order to solve the above technical problems, the present invention also provides a train energy consumption prediction device, comprising:

[0033] A first determination module is used to determine a set of motion characteristic parameters of the target train under various working conditions according to a kinematic sensor data sequence in historical data of the target train, wherein the kinematic sensor data sequence includes kinematic sensor data of each sampling point;

[0034] A training module, for training an energy consumption prediction model of a target train under any working condition of the target train according to a set of motion characteristic parameters under the working condition;

[0035] The second determination module is used to determine the current working condition of the target train;

[0036] The prediction module is used to input the motion characteristic parameters of the target train in the previous preset cycle into the energy consumption prediction model corresponding to the current working condition to obtain the energy consumption prediction result of the target train.

[0037] In order to solve the above technical problems, the present invention also provides a train energy consumption prediction device, comprising:

[0038] Memory for storing computer programs;

[0039] A processor is used to implement the steps of the train energy consumption prediction method as described above when executing the computer program.

[0040] In order to solve the above technical problems, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the train energy consumption prediction method as described above are implemented.

[0041] Beneficial effect: The present invention provides a method for predicting train energy consumption. Considering that train energy consumption is closely related to motion characteristic parameters, and the motion characteristic parameters under different train operating conditions have significant differences, the present invention can determine the motion characteristic parameter set of the target train under various operating conditions based on the kinematic sensor data sequence in the historical data of the target train, and then for any operating condition of the target train, according to the motion characteristic parameter set under the operating condition, an energy consumption prediction model of the target train under the operating condition is trained. When predicting, the motion characteristic parameters of the target train in the previous preset period can be input into the energy consumption prediction model corresponding to the current operating condition of the target train to obtain the energy consumption prediction result of the target train. Since energy consumption prediction is performed by means of operating condition identification, the accuracy of train energy consumption prediction can be greatly improved, and the guiding value of train energy consumption prediction results is improved.

[0042] The present invention also provides a train energy consumption prediction device, equipment and computer-readable storage medium, which have the same beneficial effects as the above train energy consumption prediction method. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the relevant technologies and the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0044] Figure 1 A flow chart of a train energy consumption prediction method provided by the present invention;

[0045] Figure 2 A schematic flow chart of another train energy consumption prediction method provided by the present invention;

[0046] Figure 3 A schematic diagram of a flow chart of a train energy consumption prediction device provided by the present invention;

[0047] Figure 4 A schematic diagram of the flow of a train energy consumption prediction device provided by the present invention. DETAILED DESCRIPTION

[0048] The core of the present invention is to provide a train energy consumption prediction method, device, equipment and computer-readable storage medium. In the present invention, the motion characteristic parameter set of the target train under various working conditions can be determined according to the kinematic sensor data sequence in the historical data of the target train. Then, for any working condition of the target train, according to the motion characteristic parameter set under the working condition, the energy consumption prediction model of the target train under the working condition is trained. When predicting, the motion characteristic parameters of the target train in the previous preset period can be input into the energy consumption prediction model corresponding to the current working condition of the target train to obtain the energy consumption prediction result of the target train. Since the energy consumption prediction is performed by the working condition identification method, the accuracy of the train energy consumption prediction can be greatly improved, and the guiding value of the train energy consumption prediction result is improved. In order to make the purpose, technical scheme and advantages of the embodiment of the present invention clearer, the technical scheme in the embodiment of the present invention will be clearly and completely described in combination with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0049] Please refer to Figure 1 , Figure 1 A flow chart of a train energy consumption prediction method provided by the present invention, the train energy consumption prediction method comprising:

[0050] S101: determining a set of motion characteristic parameters of the target train under various working conditions according to a kinematic sensor data sequence in historical data of the target train, wherein the kinematic sensor data sequence includes kinematic sensor data of each sampling point;

[0051] Specifically, taking into account the technical problems in the above background technology, and considering that the energy consumption of the train is closely related to the motion characteristic parameters, and the motion characteristic parameters under different train operating conditions have significant differences, the embodiment of the present invention intends to predict the energy consumption of the train under the current operating conditions through the motion characteristic parameters for different operating conditions of the train, so as to achieve efficient and accurate prediction of the energy consumption of the train, and in the embodiment of the present invention, through the motion characteristic parameter set under each operating condition, to train the energy consumption prediction model corresponding to each operating condition, and considering that the motion characteristic parameter set under each operating condition can be obtained through the kinematic sensor data sequence in the historical data of the target train, therefore in this step, the motion characteristic parameter set of the target train under each operating condition can be first determined according to the kinematic sensor data sequence in the historical data of the target train, so as to use it as the data basis for subsequent steps.

[0052] The kinematic sensing data may be of various types, such as velocity and acceleration, etc., which is not limited in the embodiment of the present invention.

[0053] S102: For any working condition of the target train, an energy consumption prediction model of the target train under the working condition is trained according to a set of motion characteristic parameters under the working condition;

[0054] Specifically, after obtaining the set of motion characteristic parameters of each operating condition, it is possible to train an energy consumption prediction model of the target train under any operating condition of the target train based on the set of motion characteristic parameters under the operating condition. In other words, each operating condition of the target train corresponds to an energy consumption prediction model, thereby realizing the energy consumption prediction of the target train under any operating condition.

[0055] S103: Determine the current working condition of the target train;

[0056] Specifically, considering that when the energy consumption of the target train is to be predicted through the above-mentioned energy consumption prediction model, the current operating condition of the target train needs to be clarified, so in this step, the current operating condition of the target train can be first determined, so that the energy consumption prediction model corresponding to the current operating condition can be selected in the subsequent steps to perform energy consumption prediction.

[0057] S104: Inputting the motion characteristic parameters of the target train in the previous preset period into the energy consumption prediction model corresponding to the current working condition to obtain the energy consumption prediction result of the target train.

[0058] Specifically, based on the determined current operating conditions, an energy consumption prediction model corresponding to the current operating conditions can be selected, and the motion characteristic parameters of the target train in the previous preset period can be input into the "energy consumption prediction model corresponding to the current operating conditions" in order to predict the energy consumption of the target train and obtain the energy consumption prediction results.

[0059] The present invention provides a method for predicting train energy consumption. Considering that train energy consumption is closely related to motion characteristic parameters, and that the motion characteristic parameters under different train operating conditions have significant differences, the present invention can determine a set of motion characteristic parameters of a target train under various operating conditions based on a kinematic sensor data sequence in historical data of the target train. Then, for any operating condition of the target train, an energy consumption prediction model of the target train under the operating condition is trained based on the set of motion characteristic parameters under the operating condition. When predicting, the motion characteristic parameters of the target train in the previous preset period can be input into the energy consumption prediction model corresponding to the current operating condition of the target train to obtain an energy consumption prediction result of the target train. Since energy consumption prediction is performed by means of operating condition identification, the accuracy of train energy consumption prediction can be greatly improved, and the guiding value of the train energy consumption prediction result is improved.

[0060] Based on the above embodiments:

[0061] As an optional embodiment, the kinematic sensing data includes velocity and acceleration;

[0062] According to the kinematic sensor data sequence in the historical data of the target train, the motion characteristic parameter set of the target train under various working conditions is determined to include:

[0063] Based on a preset time interval, a kinematic sensor data sequence in the historical data of the target train is divided into a plurality of kinematic sensor data subsequences;

[0064] For any kinematic sensor data subsequence, determining a preset type of motion characteristic parameter of the time interval where the kinematic sensor data subsequence is located according to the kinematic sensor data of each sampling point in the kinematic sensor data subsequence;

[0065] Clustering each motion characteristic parameter subset to obtain a motion characteristic parameter set under each working condition of the target train;

[0066] Among them, a single kinematic sensing data subsequence and its corresponding motion characteristic parameter are collectively regarded as a motion characteristic parameter subset, and the motion characteristic parameter set includes several motion characteristic parameter subsets.

[0067] Specifically, considering that the types of motion characteristic parameters contained in the kinematic sensor data are limited, for example, only two types of parameters, speed and acceleration, and more types of motion characteristic parameters are conducive to efficient and accurate energy consumption prediction, the embodiment of the present invention intends to obtain more types of motion characteristic parameters based on the kinematic sensor data sequence. Therefore, in the embodiment of the present invention, the kinematic sensor data sequence in the historical data of the target train can be first divided into several kinematic sensor data subsequences based on a preset time interval, and then for any kinematic sensor data subsequence, the kinematic sensor data of each sampling point in the kinematic sensor data subsequence is used to determine the preset type of motion characteristic parameters of the time interval in which the kinematic sensor data subsequence is located. At this point, multiple motion characteristic parameter subsets can be obtained, and a single motion characteristic parameter subset includes a kinematic sensor data subsequence and its corresponding motion characteristic parameters. Then, by clustering each motion characteristic parameter subset, the motion characteristic parameter set under each working condition of the target train can be obtained.

[0068] Specifically, the preset type can also be set independently, for example, it can include acceleration time percentage, deceleration time percentage, average acceleration, average deceleration, speed standard deviation and acceleration standard deviation, etc., which is not limited in the embodiment of the present invention.

[0069] Specifically, the clustering method used for clustering may also be independently set, and may include, for example, SVM (Support Vector Machine Clustering, support vector machine), etc., which is not limited in the embodiment of the present invention.

[0070] The preset time interval may be set independently, for example, may be set to 10 minutes, etc., and the embodiment of the present invention does not limit this.

[0071] As an optional embodiment, clustering each motion feature parameter subset to obtain a motion feature parameter set under each working condition of the target train includes:

[0072] Performing initial clustering on each subset of motion characteristic parameters to obtain a set of motion characteristic parameters of the target train under each working condition after initial clustering;

[0073] Through correlation analysis and contribution analysis, the various features involved in the motion feature parameter subset are eliminated;

[0074] The motion feature parameter subsets after feature elimination are clustered again to obtain the motion feature parameter sets under various working conditions of the target train.

[0075] Specifically, considering that after "determining the preset type of motion characteristic parameters of the time interval in which the kinematic sensor data subsequence is located based on the kinematic sensor data of each sampling point in the kinematic sensor data subsequence", among the preset types of motion characteristic parameters determined, if the correlation between multiple types of motion characteristic parameters is strong, then the contributions of the multiple types of motion characteristic parameters with strong correlation to energy consumption prediction are relatively repeated, which will lead to redundant data processing, and some types of motion characteristic parameters have a low contribution to working condition clustering, then these motion characteristic parameters with a low contribution to working condition clustering will affect the accuracy of energy consumption prediction; therefore, in the process of clustering each motion characteristic parameter subset, the embodiment of the present invention can first cluster each motion characteristic parameter subset. The initial clustering is performed to obtain a set of motion characteristic parameters of the target train under various working conditions after the initial clustering. Then, the various features involved in the motion characteristic parameter subset are eliminated through correlation analysis and contribution analysis. The secondary clustering is performed on each motion characteristic parameter subset after feature elimination to obtain a set of motion characteristic parameters of the target train under various working conditions. After feature elimination, "some types of motion characteristic parameters among multiple types of motion characteristic parameters with strong correlation" and "some types of motion characteristic parameters with low contribution to working condition clustering" can be eliminated. On the one hand, the data processing volume is reduced and the model training efficiency is improved. On the other hand, the interference of low-contribution data on the energy consumption prediction results can be reduced, thereby improving the accuracy of the energy consumption prediction results.

[0076] The clustering methods used in the initial clustering and the secondary clustering may be the same or different, which is not limited in the embodiment of the present invention.

[0077] As an optional embodiment, the features involved in the motion feature parameter subset are eliminated by using a correlation analysis method and a contribution analysis method, including:

[0078] For any feature in the motion feature parameter subset, remove the feature in the motion feature parameter subset whose correlation with the feature is higher than a first preset threshold;

[0079] For any feature in the motion feature parameter subset, the feature in the motion feature parameter subset whose contribution to the working condition clustering is lower than a second preset threshold is eliminated.

[0080] Specifically, as described above, in order to better eliminate some types of motion feature parameters in the "multiple types of motion feature parameters with strong correlation" in the motion feature parameter subset, an embodiment of the present invention can pre-set a first preset threshold, and for any feature in the motion feature parameter subset, the feature in the motion feature parameter subset whose correlation with the feature is higher than the first preset threshold is eliminated. It is worth mentioning that the motion feature parameters eliminated in this link can be consistent in each motion feature parameter subset, that is, for any motion feature parameter subset, the type of motion feature parameters eliminated based on the correlation analysis method can be the same, which means that one of the motion feature parameter subsets can be selected, and the step of "for any feature in the motion feature parameter subset, eliminating the feature in the motion feature parameter subset whose correlation with the feature is higher than the first preset threshold" is performed on the selected motion feature parameter subset, so as to determine the type of motion feature parameters to be eliminated obtained by the correlation analysis method, and then eliminate the motion feature parameters of the type to be eliminated in each motion feature parameter subset.

[0081] Among them, in order to better eliminate the "motion feature parameters with low contribution to working condition clustering" in the motion feature parameter subset, a second preset threshold can be set for the contribution in an embodiment of the present invention, and then for any feature in the motion feature parameter subset, the feature in the motion feature parameter subset whose contribution to working condition clustering is lower than the second preset threshold is eliminated. It is worth mentioning that the motion feature parameters eliminated in this link can be consistent in each motion feature parameter subset, that is, one of the motion feature parameter subsets can be selected, and the type of motion feature parameters to be eliminated can be determined through the step of "for any feature in the motion feature parameter subset, the feature in the motion feature parameter subset whose contribution to working condition clustering is lower than the second preset threshold is eliminated", and then the motion feature parameters of the type to be eliminated in each motion feature parameter subset can be eliminated.

[0082] Specifically, the first preset threshold and the second preset threshold can be set independently, and the embodiment of the present invention does not limit this.

[0083] As an optional embodiment, the correlation analysis method includes Pearson correlation analysis;

[0084] Contribution analysis methods include: Fast Shapley additivity interpretation method.

[0085] Specifically, both the similarity analysis method and the contribution analysis method can be of various types. For example, the similarity analysis method is Pearson correlation analysis, and the contribution analysis method can include FastShap, etc., which is not limited in the embodiment of the present invention.

[0086] Of course, in addition to the above methods, the similarity analysis method and the contribution analysis method may also be of other types, which are not limited in the embodiment of the present invention.

[0087] As an optional embodiment, for any working condition of the target train, according to the set of motion characteristic parameters under the working condition, the energy consumption prediction model of the target train under the working condition is trained to include:

[0088] For any working condition of the target train, a first energy consumption prediction model of the working condition is trained according to a set of motion characteristic parameters of the working condition;

[0089] Determining whether the prediction accuracy of the first energy consumption prediction model meets the standard;

[0090] If the target is reached, it ends;

[0091] If the standard is not met, the second energy consumption prediction model is trained according to the prediction result of the first energy consumption prediction model and the weight parameters of the first energy consumption prediction model, so that the second energy consumption prediction model can be used as the energy consumption prediction model of the target train under the working condition.

[0092] Specifically, considering that the accuracy of the first energy consumption prediction model trained by the data set is not good enough, the second energy consumption prediction model is trained by the prediction result of the first energy consumption prediction model and the weight parameters of the second energy consumption model, so that the second energy consumption prediction model with higher energy consumption prediction accuracy can be obtained. Therefore, in the embodiment of the present invention, for any working condition of the target train, the first energy consumption prediction model of the working condition can be trained according to the motion characteristic parameter set of the working condition, and then it is determined whether the prediction accuracy of the first energy consumption prediction model meets the standard. If the standard is met, the subsequent steps can be omitted. If the standard is not met, the second energy consumption prediction model can be trained according to the prediction result of the first energy consumption prediction model and the weight parameters of the first energy consumption prediction model, so as to use the second energy consumption prediction model as the energy consumption prediction model of the target train under the working condition, thereby improving the energy consumption prediction accuracy under the working condition.

[0093] Among them, the first energy consumption prediction model and the second energy consumption prediction model can be constructed based on various types of machine learning algorithms, such as XGBoost (eXtreme Gradient Boosting), etc., which is not limited in the embodiment of the present invention.

[0094] As an optional embodiment, determining the current operating condition of the target train includes:

[0095] Based on the kinematic sensor data of the target train in the previous preset period, the current working condition of the target train is determined through a pre-trained language model.

[0096] Specifically, considering that the "kinematic sensor data of the target train in the previous preset period" is closely related to the current operating condition of the target train, the current operating condition of the target train can be automatically identified based on the "kinematic sensor data of the target train in the previous preset period" through a pre-trained language model. Therefore, a pre-trained language model can be set in an embodiment of the present invention, and based on the kinematic sensor data of the target train in the previous preset period, the current operating condition of the target train can be determined through the pre-trained language model, which can reduce labor costs and improve the speed of operating condition confirmation.

[0097] Of course, in addition to this specific form, the current operating condition of the target train may also be determined in other ways, which are not limited in the embodiment of the present invention.

[0098] Specifically, the construction process of the pre-trained language model may include: designing the prompt words of the pre-trained prediction model through "determining the analysis logic of the working condition of the target train based on kinematic sensor data", and then setting the ambient temperature (for example, 25°C) and passenger load factor (for example, 34%) of the pre-trained prediction model to complete the construction of the pre-trained language model; and recording the API (Application Programming Interface) of the pre-trained language model so that the API can be called later to use the pre-trained language model.

[0099] Specifically, the prompt words for the pre-trained prediction model may include: You are a traffic engineering expert specializing in vehicle dynamics analysis. Based on the following data, please determine whether the vehicle's current driving condition is starting, accelerating, constant speed, or braking. Provide detailed analysis and reasons. The data are as follows: current speed (m / s); current acceleration (m / s²); Please explain your analysis process step by step and give your final judgment result.

[0100] Among them, the "analysis logic for determining the working condition of the target train based on kinematic sensor data" can be set independently. For example, when the kinematic sensor data includes speed and acceleration, the "analysis logic for determining the working condition of the target train based on kinematic sensor data" can include: Analyze the current speed and acceleration: The magnitude and direction of the current speed and acceleration are important bases for preliminary judgment of the working condition. The starting condition is usually accompanied by a lower speed and positive acceleration. The acceleration condition is usually characterized by a higher speed accompanied by a positive acceleration. The uniform speed condition has a higher speed and the acceleration is close to zero. The braking condition is usually characterized by a higher speed accompanied by a negative acceleration.

[0101] In addition, it is worth mentioning that based on the kinematic sensor data of the target train in the previous preset period, the current working condition of the target train can be determined by the pre-trained language model, including:

[0102] Based on the kinematic sensor data and future speed of the target train in the previous preset cycle, the current working condition of the target train is determined through the pre-trained language model.

[0103] Specifically, the prompt words of the pre-trained prediction model can be designed through "the analysis logic of determining the working condition of the target train based on kinematic sensor data and future speed". This requires predicting the future speed of the target train based on kinematic sensor data. In this case, the prompt words of the pre-trained prediction model can include: You are a traffic engineering expert who focuses on vehicle dynamics analysis. Based on the following data, please determine whether the vehicle's current driving condition is starting, accelerating, constant speed or braking. Provide detailed analysis and reasons. The data are as follows: current speed (m / s); current acceleration (m / s²); speed in the future preset time interval (for example, 10 seconds) (recorded once per second) (m / s); Please explain your analysis process step by step and finally give the judgment result. In this case, "the analysis logic of determining the working condition of the target train based on kinematic sensor data and future speed" can include: (1) Analysis of current speed and acceleration: The magnitude and direction of current speed and acceleration are important bases for preliminary judgment of working conditions. Starting conditions are usually accompanied by low speed and positive acceleration. Acceleration conditions are usually characterized by high speed accompanied by positive acceleration. The uniform speed condition is characterized by a high speed and an acceleration close to zero. The braking condition is usually characterized by a high speed accompanied by a negative acceleration. (2) Predicting future speed trends: Based on the speed data of a preset time interval in the future, observe the speed change trend. If the speed gradually increases (acceleration is positive), it indicates that the vehicle is in an acceleration condition. If the speed remains unchanged (acceleration is close to zero), it indicates that the vehicle is in a uniform speed condition. If the speed gradually decreases (acceleration is negative), it indicates that the vehicle is in a braking condition. Comprehensive judgment: Based on the current speed, acceleration, and future speed trends, the final judgment is made.

[0104] There are many methods for predicting future speed, such as various machine learning algorithms, which may include: (1) normalizing the "kinematic sensor data of the target train in the previous preset period", (2) constructing a speed prediction model based on LSTM (Long Short-Term Memory), setting the input gate, forget gate and output gate of the speed prediction model. (3) Calculating the model output, predicting the speed change of the target train in the future preset time interval, and denormalizing the speed prediction data to obtain the final predicted speed.

[0105] Specifically, in order to better illustrate the embodiments of the present invention, please refer to Figure 2 , Figure 2 A flow chart of another train energy consumption prediction method provided by the present invention is shown in FIG. Figure 2 The right column shows the process of building a "condition-based energy consumption prediction model" based on a subset of motion feature parameters, while the left column shows that based on the kinematic sensor data of the target train in the previous preset period, the future speed is predicted through the speed prediction model, and the condition judgment is made based on the pre-trained language model, and finally the energy consumption is predicted using the corresponding energy consumption prediction model.

[0106] Please refer to Figure 3 , Figure 3 A schematic diagram of the structure of a train energy consumption prediction device provided by the present invention, the train energy consumption prediction device comprises:

[0107] A first determination module 31 is used to determine a set of motion characteristic parameters of the target train under various working conditions according to a kinematic sensor data sequence in historical data of the target train, wherein the kinematic sensor data sequence includes kinematic sensor data of each sampling point;

[0108] A training module 32 is used for training an energy consumption prediction model of the target train under any working condition of the target train according to a set of motion characteristic parameters under the working condition;

[0109] A second determination module 33 is used to determine the current operating condition of the target train;

[0110] The prediction module 34 is used to input the motion characteristic parameters of the target train in the previous preset period into the energy consumption prediction model corresponding to the current working condition to obtain the energy consumption prediction result of the target train.

[0111] For an introduction to the train energy consumption prediction device provided in an embodiment of the present invention, please refer to the aforementioned embodiment of the train energy consumption prediction method, and the embodiment of the present invention will not be described in detail here.

[0112] Please refer to Figure 4 , Figure 4A schematic diagram of the structure of a train energy consumption prediction device provided by the present invention, the train energy consumption prediction device comprises:

[0113] A memory 41, used for storing computer programs;

[0114] The processor 42 is used to implement the steps of the train energy consumption prediction method in the above-mentioned embodiment when executing the computer program.

[0115] For an introduction to the train energy consumption prediction device provided in an embodiment of the present invention, please refer to the aforementioned embodiment of the train energy consumption prediction method, and the embodiment of the present invention will not be described in detail here.

[0116] The present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the train energy consumption prediction method in the aforementioned embodiment are implemented.

[0117] For an introduction to the computer-readable storage medium provided in an embodiment of the present invention, please refer to the aforementioned embodiment of the train energy consumption prediction method, and the embodiment of the present invention will not be described in detail here.

[0118] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should also be noted that in this specification, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the term "include", "comprise" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such process, method, article or equipment. In the absence of more restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method, article or equipment including the element.

[0119] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A train energy consumption prediction method, characterized in that: include: Determine a set of motion characteristic parameters of the target train under various working conditions according to a kinematic sensor data sequence in historical data of the target train, wherein the kinematic sensor data sequence includes kinematic sensor data of each sampling point; For any working condition of the target train, an energy consumption prediction model of the target train under the working condition is trained according to a set of motion characteristic parameters under the working condition; Determine the current operating condition of the target train; The motion characteristic parameters of the target train in the last preset period are input into the energy consumption prediction model corresponding to the current working condition to obtain the energy consumption prediction result of the target train.

2. The train energy consumption prediction method according to claim 1, characterized in that: Kinematic sensing data includes velocity and acceleration; According to the kinematic sensor data sequence in the historical data of the target train, the motion characteristic parameter set of the target train under various working conditions is determined to include: Based on a preset time interval, a kinematic sensor data sequence in the historical data of the target train is divided into a plurality of kinematic sensor data subsequences; For any kinematic sensing data subsequence, determining a preset type of motion characteristic parameter of a time interval where the kinematic sensing data subsequence is located according to the kinematic sensing data of each sampling point in the kinematic sensing data subsequence; Clustering each motion characteristic parameter subset to obtain a motion characteristic parameter set under each working condition of the target train; Among them, a single kinematic sensing data subsequence and its corresponding motion characteristic parameter are collectively regarded as a motion characteristic parameter subset, and the motion characteristic parameter set includes several motion characteristic parameter subsets.

3. The train energy consumption prediction method according to claim 2, characterized in that: Clustering each of the motion characteristic parameter subsets to obtain a motion characteristic parameter set under each working condition of the target train includes: Performing initial clustering on each of the motion characteristic parameter subsets to obtain a motion characteristic parameter set of the target train under each working condition after initial clustering; Through correlation analysis and contribution analysis, the various features involved in the motion feature parameter subset are eliminated; Secondary clustering is performed on each of the motion feature parameter subsets after feature elimination to obtain a motion feature parameter set under each working condition of the target train.

4. The train energy consumption prediction method according to claim 3, characterized in that: Through correlation analysis and contribution analysis, the various features involved in the motion feature parameter subset are eliminated, including: For any feature in the motion feature parameter subset, remove the feature in the motion feature parameter subset whose correlation with the feature is higher than a first preset threshold; For any feature in the motion feature parameter subset, the feature in the motion feature parameter subset whose contribution to the working condition clustering is lower than a second preset threshold is eliminated.

5. The train energy consumption prediction method according to claim 4, characterized in that: The correlation analysis method includes Pearson correlation analysis; The contribution analysis method includes: a fast Shapley additivity interpretation method.

6. The train energy consumption prediction method according to claim 1, characterized in that: For any working condition of the target train, according to the set of motion characteristic parameters under the working condition, the energy consumption prediction model of the target train under the working condition is trained to include: For any working condition of the target train, a first energy consumption prediction model of the working condition is trained according to a set of motion characteristic parameters of the working condition; Determining whether the prediction accuracy of the first energy consumption prediction model meets the standard; If the target is reached, it ends; If the standard is not met, the second energy consumption prediction model is trained according to the prediction result of the first energy consumption prediction model and the weight parameters of the first energy consumption prediction model, so that the second energy consumption prediction model is used as the energy consumption prediction model of the target train under the said working conditions.

7. The train energy consumption prediction method according to any one of claims 1 to 6, characterized in that: Determine the current working condition of the target train including: Based on the kinematic sensor data of the target train in the previous preset period, the current working condition of the target train is determined through a pre-trained language model.

8. A train energy consumption prediction device, characterized in that: include: A first determination module is used to determine a set of motion characteristic parameters of the target train under various working conditions according to a kinematic sensor data sequence in historical data of the target train, wherein the kinematic sensor data sequence includes kinematic sensor data of each sampling point; A training module, for training an energy consumption prediction model of a target train under any working condition of the target train according to a set of motion characteristic parameters under the working condition; The second determination module is used to determine the current working condition of the target train; The prediction module is used to input the motion characteristic parameters of the target train in the previous preset cycle into the energy consumption prediction model corresponding to the current working condition to obtain the energy consumption prediction result of the target train.

9. A train energy consumption prediction device, characterized in that: include: Memory for storing computer programs; A processor is used to implement the steps of the train energy consumption prediction method as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the train energy consumption prediction method according to any one of claims 1 to 7 are implemented.

Images