Residual endurance mileage estimation method constructed based on equivalent operation condition of vehicle-mounted hydrogen system

Abstract

The invention provides a remaining endurance mileage estimation method constructed based on an equivalent operation condition of a vehicle-mounted hydrogen system, which comprises the following steps: performing feature extraction on current operation data of a target hydrogen energy vehicle to obtain a current user behavior portrait; inputting the current user behavior portrait and the driving mode to which the current user behavior portrait belongs into a pre-trained time sequence prediction model to obtain a predicted driving mode at the next moment; performing working condition reconstruction on the predicted driving mode according to the historical operation data corresponding to each driving mode to obtain reconstructed operation data in the predicted driving mode; according to the reconstructed operation data and a pre-constructed mapping relation between the operation condition and hydrogen consumption, determining characteristic behavior hydrogen consumption under the reconstructed operation data; and according to the characteristic behavior hydrogen consumption under the reconstruction operation data in the prediction time interval and the current residual hydrogen storage amount of the target hydrogen energy vehicle, calculating to obtain the residual endurance mileage. According to the scheme, multi-dimensional user behaviors are introduced to obtain the reconstructed operation data more conforming to the driving style of the user, so that the remaining endurance mileage is predicted based on the reconstructed operation data, and the prediction accuracy and reliability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management of on-board hydrogen systems, and in particular to a method for estimating remaining cruising range based on equivalent operating conditions of an on-board hydrogen system. Background Art

[0002] Hydrogen systems and hydrogen fuel cells are a pollution-free form of energy. The only byproduct of hydrogen combustion is water. Therefore, they have become an important technical path to achieve low-carbon or even zero-carbon emissions, and bear the heavy responsibility of reducing carbon emissions in the transportation industry. Vehicles equipped with hydrogen systems and hydrogen fuel cells have gradually entered the commercial application stage, but due to the imperfect hydrogen refueling infrastructure and the particularity of hydrogen storage, the existing hydrogen energy supply is difficult. It is necessary to accurately estimate the remaining range in order to reasonably plan and arrange hydrogen energy replenishment, thereby ensuring the normal operation of the vehicle and reducing user inconvenience.

[0003] Existing methods for estimating the remaining range of hydrogen fuel cell vehicles, such as linear regression algorithms based on historical operating conditions and rule-based logic judgment theories based on average energy consumption, only make estimates based on fixed operating conditions and lack consideration of user behavior, resulting in poor estimation accuracy and lack of credibility. Summary of the invention

[0004] The present invention provides a method for estimating the remaining cruising range based on the equivalent operating conditions of an on-board hydrogen system. The operating conditions of the on-board hydrogen system are reconstructed by introducing multi-dimensional user behaviors to obtain predicted conditions that are more in line with the user's driving style, thereby predicting the remaining cruising range based on the predicted conditions and hydrogen consumption, thereby improving the prediction accuracy and reliability.

[0005] In a first aspect, the present invention provides a method for estimating remaining cruising range based on an equivalent operating condition of an on-board hydrogen system, comprising:

[0006] Extract features from the current operating data of the target hydrogen-powered vehicle to obtain the current user behavior profile;

[0007] Inputting the current user behavior profile and the driving mode to which the current user behavior profile belongs into a pre-trained time series prediction model to obtain a predicted driving mode at the next moment;

[0008] Reconstructing the working condition of the predicted driving mode according to the historical operating data corresponding to each driving mode to obtain the reconstructed operating data under the predicted driving mode;

[0009] Determining the characteristic behavior hydrogen consumption under the reconstructed operating data according to the reconstructed operating data and the pre-constructed mapping relationship between the operating condition and the hydrogen consumption;

[0010] The remaining cruising range is obtained based on the characteristic behavior hydrogen consumption under the reconstructed operating data within the prediction time interval and the current remaining hydrogen storage capacity of the target hydrogen-powered vehicle.

[0011] In a second aspect, the present invention provides a remaining cruising range estimation device based on an equivalent operating condition of an on-board hydrogen system, comprising:

[0012] The feature extraction module is used to extract features from the current operating data of the target hydrogen-powered vehicle to obtain the current user behavior portrait;

[0013] A prediction module, used for inputting the current user behavior profile and the driving mode to which the current user behavior profile belongs into a pre-trained time series prediction model to obtain a predicted driving mode at the next moment;

[0014] a reconstruction module, configured to reconstruct the working condition of the predicted driving mode according to the historical operating data corresponding to each of the driving modes, so as to obtain the reconstructed operating data under the predicted driving mode;

[0015] A hydrogen consumption calculation module, used to determine the characteristic behavior hydrogen consumption under the reconstructed operating data according to the reconstructed operating data and a pre-constructed mapping relationship between operating conditions and hydrogen consumption;

[0016] The mileage calculation module is used to calculate the remaining cruising range based on the characteristic behavior hydrogen consumption under the reconstructed operation data within the predicted time interval and the current remaining hydrogen storage capacity of the target hydrogen-powered vehicle.

[0017] In a third aspect, an embodiment of the present invention further provides a computing device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method described in any first aspect of this specification is implemented.

[0018] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, enables the computer to execute the method described in any first aspect of this specification.

[0019] In a fifth aspect, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the steps of the method described in any first aspect of this specification are implemented.

[0020] The embodiment of the present invention provides a method for estimating the remaining cruising range based on the equivalent operating conditions of the on-board hydrogen system. The method introduces a multi-dimensional user behavior profile by extracting features from the operating data, and determines the driving mode to which the user behavior profile belongs. Then, the predicted driving mode at the next moment can be obtained based on the current user behavior profile and the driving mode to which it belongs through a pre-trained time series prediction model. The predicted driving mode is then reconstructed for the predicted driving mode to determine the reconstructed operating data under the preset driving mode. The characteristic behavior hydrogen consumption under the reconstructed operating data is then determined based on the mapping relationship between the pre-constructed operating conditions and hydrogen consumption. The prediction steps are repeated until all the reconstructed operating data and their characteristic behavior hydrogen consumption within the prediction time interval are obtained, so that the remaining cruising range can be predicted based on the current remaining hydrogen storage and the characteristic behavior hydrogen consumption under each reconstructed operating data. In this way, the present invention obtains reconstructed operating data that is more in line with the user's driving style by considering the multi-dimensional operating data prediction, and then predicts the remaining cruising range based on the characteristic behavior hydrogen consumption of the reconstructed operating data, thereby improving the prediction accuracy and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 It is a flow chart of a method for estimating remaining cruising range based on an equivalent operating condition of an on-board hydrogen system provided by an embodiment of the present invention;

[0023] Figure 2 is a flow chart of another remaining cruising range estimation method based on an equivalent operating condition of an on-board hydrogen system provided by an embodiment of the present invention;

[0024] Figure 3 is a hardware architecture diagram of a computing device provided by an embodiment of the present invention;

[0025] Figure 4 It is a structural schematic diagram of a remaining cruising range estimation device constructed based on the equivalent operating conditions of an on-board hydrogen system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0027] The specific implementation of the concept of this application is described below.

[0028] Please refer to Figure 1 , an embodiment of the present invention provides a method for estimating the remaining cruising range based on an equivalent operating condition of an on-board hydrogen system, the method comprising:

[0029] Step 100, extracting features from the current operating data of the target hydrogen-powered vehicle to obtain a current user behavior profile;

[0030] Step 102, inputting the current user behavior profile and the driving mode to which the current user behavior profile belongs into a pre-trained time series prediction model to obtain a predicted driving mode at the next moment;

[0031] Step 104, reconstructing the working condition of the predicted driving mode according to the historical operating data corresponding to each driving mode, and obtaining the reconstructed operating data under the predicted driving mode;

[0032] Step 106, determining the characteristic behavior hydrogen consumption under the reconstructed operating data according to the reconstructed operating data and the pre-constructed mapping relationship between the operating condition and the hydrogen consumption;

[0033] Step 108, calculating the remaining cruising range based on the characteristic behavior hydrogen consumption under the reconstructed operation data within the prediction time interval and the current remaining hydrogen storage of the target hydrogen-powered vehicle.

[0034] In an embodiment of the present invention, an embodiment of the present invention provides a method for estimating the remaining cruising range based on the equivalent operating conditions of the on-board hydrogen system, which introduces a multi-dimensional user behavior profile by extracting features from the operating data, and determines the driving mode to which the user behavior profile belongs, and then obtains the predicted driving mode at the next moment based on the current user behavior profile and the driving mode to which it belongs through a pre-trained time series prediction model, and then reconstructs the predicted driving mode to determine the reconstructed operating data under the preset driving mode, and then determines the characteristic behavior hydrogen consumption under the reconstructed operating data based on the mapping relationship between the pre-constructed operating conditions and hydrogen consumption, and repeats the prediction steps until all the reconstructed operating data and their characteristic behavior hydrogen consumption within the prediction time interval are obtained, so that the remaining cruising range can be predicted based on the current remaining hydrogen storage and the characteristic behavior hydrogen consumption under each reconstructed operating data. In this way, the present invention obtains reconstructed operating data that is more in line with the user's driving style by considering the multi-dimensional operating data prediction, and then predicts the remaining cruising range based on the characteristic behavior hydrogen consumption of the reconstructed operating data, thereby improving the prediction accuracy and reliability.

[0035] Described below Figure 1 How the various steps are performed.

[0036] First, before step 100, the process further includes:

[0037] Divide the historical operation data of the target hydrogen-powered vehicle into equal time segments to obtain a number of historical operation segments including time information and unit time operation data;

[0038] Extract features from several historical operation segments to obtain several segment user portraits;

[0039] The density clustering algorithm is used to classify the fragmented user portraits and determine the driving mode to which each fragmented user portrait belongs.

[0040] It should be noted that historical operation data includes vehicle data, vehicle hydrogen system and hydrogen fuel cell data, power battery data, and motor data; vehicle data includes vehicle speed, accelerator pedal travel, brake pedal travel, etc.; vehicle hydrogen system and hydrogen fuel cell data include operating pressure, hydrogen consumption, hydrogen flow, hydrogen sensor concentration, etc.; power battery data includes voltage, current, temperature, remaining charge SOC, etc.; motor data includes torque, speed, etc. The fragmented user portrait corresponds to the unit time operation data one by one, and the fragmented user portrait is the unit time operation data used to describe the user's driving style.

[0041] Specifically, each segment user portrait is input into the DBSACAN algorithm for data classification to achieve the attribution judgment of each segment user portrait, that is, each attribution category is determined as a driving mode, so as to achieve the determination of the driving mode to which different segment user portraits belong. Among them, different driving modes include different segment user portraits. For example, the historical operation data is divided into equal time segments (the segmentation standard is 1min), and 200 unit time operation data (i.e. 200 historical operation segments) including the segment sequence number are obtained. After feature extraction, 200 segment user portraits are obtained. The DBSACAN algorithm is used to classify all the segment user portraits. Finally, the 200 segment user portraits are divided into A (including 60 segment user portraits), B (including 30 segment user portraits), C (including 40 segment user portraits), D (including 10 segment user portraits), and E (including 60 segment user portraits).

[0042] In a preferred embodiment, feature extraction is performed on several historical operation segments to obtain several segment user portraits, including:

[0043] Obtaining the unit time operation data of the historical operation fragments; wherein the unit time operation data includes the unit time operation data of the whole vehicle data, the hydrogen system operation data, the power battery data, and the motor data;

[0044] Perform feature extraction on each unit time operation data to obtain a number of features; wherein the number of features is the same as the number of quantity types included in the unit time operation data;

[0045] After correlation analysis of the features, effective key features are screened out;

[0046] Generate fragmented user portraits based on effective key features.

[0047] Specifically, the features include average vehicle speed features, acceleration segment ratio, deceleration segment ratio, idling segment ratio, accelerator pedal total duration features, brake pedal total duration features, accelerator pedal average travel features, brake pedal average travel features, operating pressure average features, average hydrogen consumption features, hydrogen flow average features, hydrogen sensor concentration average features, voltage average features, current average features, temperature average features, remaining charge SOC average features, torque average features, and speed average features. The number of valid key features shall not exceed the total number of features.

[0048] Specifically, the following four methods can be used to screen out effective key features after correlation analysis of the features:

[0049] The first method: Use the principal component analysis algorithm to perform correlation analysis and select features with a cumulative contribution rate greater than 90% as effective key features;

[0050] The second method: Use the Relief-F algorithm to perform correlation analysis and select features with a cumulative contribution rate greater than 90% as effective key features;

[0051] The third method: Use the partial principal component analysis algorithm to perform correlation analysis and select features with a cumulative contribution rate greater than 90% as effective key features;

[0052] The fourth method: Use the Pearson correlation analysis algorithm to perform correlation analysis and select features with correlation less than 0.1 as effective key features.

[0053] In the present invention, the current user behavior portrait is a vector composed of effective key features obtained by analyzing the current operation data. The present invention obtains several features by extracting features from multi-dimensional operation data. If all features are considered, the amount of data will increase. Therefore, in order to reduce the dimension while retaining as much information as possible, a correlation analysis method is further used to screen several features to obtain effective key features.

[0054] Step 100 uses the above feature extraction method to obtain the current user behavior portrait under the current operation data. It should be noted that the current user behavior portrait is a segment user portrait, that is, the time length corresponding to the current user behavior portrait is a unit time length.

[0055] In step 102, the pre-trained time series prediction model is trained by the following method:

[0056] Generate a plurality of first sample sets according to the segment user portraits and the driving modes to which the segment user portraits belong;

[0057] The self-attention mechanism Transformer architecture is trained using the first sample set to obtain a pre-trained time series prediction model; wherein the first sample set includes a segmented user portrait at the current moment and the driving mode to which the segmented user portrait belongs as input, and a driving mode to which the segmented user portrait belongs at the next moment as output.

[0058] It should be noted that the time series prediction model in step 102 is the user behavior prediction model, which learns the user's driving style by establishing a Self-attention Transformer algorithm, whose input is the segment user portrait and its driving mode running at the current moment, and the output is the driving mode to which the segment user portrait corresponding to the next moment running at the current moment belongs. The Self-attention Transformer algorithm refers to a series of data-driven models based on the encoder-decoder structure, including a series of encoders and decoders stacked, each encoder module consists of a Self-attention layer and a feedforward neural network, and the decoder has an additional encoder-decoder attention module compared to the encoder, which is used to establish the connection between the encoding and decoding relationship.

[0059] In the present invention, the Transformer architecture of the self-attention mechanism can effectively capture the long-term and short-term dependencies in driving behaviors, such as instantaneous operations such as acceleration and braking, and long-term stable patterns such as driving behavior habits and driving routes. By coupling the Self-attention mechanism, the Transformer can globally perceive the data of all time steps when processing driving behaviors, rather than relying on the time steps of local neighborhoods. In addition, different attention heads in the multi-head attention mechanism can focus on different driving behavior characteristics, so that a parallel processing mode can be used to effectively fuse multi-dimensional fragment user portraits, thereby accurately predicting driving modes and working conditions at future moments.

[0060] With respect to step 104, the predicted driving mode is reconstructed according to the historical operating data corresponding to each driving mode to obtain the reconstructed operating data under the predicted driving mode, including:

[0061] Determine the historical operation data corresponding to each driving mode from the historical operation data of the target hydrogen-powered vehicle, and obtain each historical operation segment included in the predicted driving mode at the next moment;

[0062] Calculating similarity values ​​of the operation segment at the current moment and each historical operation segment included in the predicted driving mode at the next moment, and determining the highest similarity value;

[0063] The unit time operation data included in the historical operation segment corresponding to the highest similarity value is determined as the reconstructed operation data at the next moment corresponding to the operation segment at the current moment.

[0064] In the present invention, the Monte Carlo simulation method is used to achieve the reconstruction of the operating conditions of the on-board hydrogen system at each moment. After the predicted driving mode at the next moment is determined based on step 102, the operating conditions at the next moment are selected from the historical operating segments included in the predicted driving mode, that is, the unit time operating data included in a historical operating segment is determined as the reconstructed operating data at the next moment. In this way, the present invention can not only handle the randomness of multiple dimensions and multiple variables, but also further select the most similar historical operating segment as the reconstructed operating data by calculating the similarity value of the operating segment at the current moment and the next moment, thereby further improving the simulation accuracy.

[0065] In step 106, the pre-established mapping relationship between the operating conditions and the hydrogen consumption is obtained by the following method:

[0066] Establish a nonlinear function relationship with segment user portrait as input and hydrogen consumption as output;

[0067] The optimal solution of the parameters of the nonlinear function relationship is solved to obtain the mapping relationship between the operating condition and the hydrogen consumption.

[0068] It should be noted that the mapping relationship between operating conditions and hydrogen consumption refers to a nonlinear functional relationship established by using characteristic inputs such as vehicle speed, accelerator pedal travel, and decelerator pedal travel in historical operating data, with hydrogen consumption as output. The hydrogen consumption under the predicted user behavior profile in step 106 refers to the hydrogen consumption under the predicted user behavior profile calculated based on the above mapping relationship, with the predicted user behavior profile of the on-board hydrogen system reconstructed as input. Among them, the hydrogen consumption is related to the user behavior profile, that is, related to the characteristic behavior, and there are differences between different users.

[0069] In step 106, the characteristic behavior hydrogen consumption under the reconstructed operation data is determined according to the reconstructed operation data and the pre-constructed mapping relationship between the operation condition and the hydrogen consumption, including:

[0070] Extract features from the reconstructed operation data to obtain a reconstructed user behavior profile under the predicted driving mode;

[0071] The reconstructed user behavior profile is input into the pre-constructed mapping relationship between operating conditions and hydrogen consumption to obtain the characteristic behavior hydrogen consumption under the reconstructed user behavior profile.

[0072] In a preferred embodiment, finding the optimal solution for the parameters of the nonlinear functional relationship includes:

[0073] S1: Initialize the population, determine the initial temperature, temperature cooling rate, cutoff temperature, maximum number of iterations, initial energy of prey, and feasible domain of feasible solutions of parameters, and randomly generate a set of potential feasible solutions in the feasible domain;

[0074] S2: In the current iteration, the potential feasible solutions are optimized in the feasible domain based on the Harris Hawk algorithm to obtain the first updated feasible solution output in the exploration phase;

[0075] S3: Determine whether to switch from the exploration phase to the development phase according to the prey energy of the current iteration and the number of iterations; if yes, execute step S4; otherwise, execute step S1;

[0076] S4: determining an optimization strategy according to the prey energy, and optimizing the first updated feasible solution using the optimization strategy to obtain a second updated feasible solution;

[0077] S5: respectively using the optimal solution in the current iteration and the second updated feasible solution to calculate the degree of fit of the nonlinear function relationship. If the calculated degree of fit of the second updated feasible solution is lower than the degree of fit of the optimal solution in the current iteration, then executing step S6; otherwise, executing step S7;

[0078] S6: Calculate the acceptance probability of the second updated feasible solution according to the initial temperature, the temperature cooling rate, the second updated feasible solution and the prey position of the current iteration, so as to accept the second updated feasible solution according to the acceptance probability; if the second updated feasible solution is accepted, execute step S7; if the second updated feasible solution is not accepted, retain the optimal solution in the current iteration, update the current temperature, and execute step S8;

[0079] S7: Determine the second updated feasible solution as the optimal solution of the current iteration;

[0080] S8: Determine whether the current number of iterations reaches the maximum number of iterations or the current temperature reaches the cutoff temperature. If so, output the optimal solution; otherwise, use the optimal solution as the initial solution for the next iteration and return to step S2.

[0081] It should be noted that, in step S6, the prey position of the current iteration is the position of the individual with the best fitness.

[0082] In a preferred embodiment, the acceptance probability of the second updated feasible solution is determined by the following formula:

[0083] T t+1 =α·T t

[0084]

[0085] Specifically, P(T t+1 ) is the acceptance probability of the second updated feasible solution at the t+1th iteration; T0 is the initial temperature; T t+1 , T t are the temperatures at the t+1th iteration and the tth iteration respectively; α is the temperature cooling rate, α∈[0,1]; k is the Boltzmann constant; EX(t+1) is the fitness value of the second updated feasible solution at the t+1th iteration; E Xrabbit is the fitness value of the current optimal fitness position.

[0086] In the present invention, the Harris Eagle algorithm uses fewer parameters, has high stability, and converges quickly. The algorithm switches between the search phase and the development phase to adapt to the changing environment and the escape mode of the prey, thereby improving the efficiency and accuracy of obtaining the optimal solution. At the same time, in the process of optimizing using the Harris Eagle algorithm, when calculating the feasible solution at the next moment from the feasible solution at the current moment, the acceptance probability of the feasible solution is changed from 100% acceptance of the original algorithm to acceptance based on probability, and the acceptance probability is calculated by the above formula, thereby improving the local optimization ability, quickly determining a better solution, and improving the solution efficiency.

[0087] In step 108, if Figure 2 As shown in FIG. 1 , the remaining cruising range is calculated based on the characteristic behavior hydrogen consumption under the reconstructed operation data within the prediction time interval and the current remaining hydrogen storage of the target hydrogen-powered vehicle, including:

[0088] A1: Extract features from the current running data to obtain the current user behavior profile. Then input the current user behavior profile and the driving mode to which the current user behavior profile belongs into the pre-trained time series prediction model to obtain the predicted driving mode at the next moment.

[0089] A2: Reconstruct the working condition of the predicted driving mode according to the historical operating data corresponding to each driving mode to obtain the reconstructed operating data under the predicted driving mode;

[0090] A3: Determine whether the current cumulative prediction time length is greater than a preset time threshold. If so, obtain the reconstructed operation data within the prediction time interval and execute step A4; otherwise, use the reconstructed operation data at the next moment as the current operation data and return to step A1;

[0091] A4: Input the reconstructed user behavior portrait obtained by feature extraction of the reconstructed operation data within the prediction time interval into the mapping relationship to obtain the characteristic behavior hydrogen consumption under each reconstructed operation data;

[0092] A5: Accumulate the characteristic behavior hydrogen consumption under the reconstructed operation data in the order of prediction time to obtain the total hydrogen consumption; when the total hydrogen consumption is the same as the current remaining hydrogen storage, calculate the sum of the mileage of the reconstructed operation data included in the total hydrogen consumption to obtain the remaining cruising range.

[0093] Specifically, the Monte Carlo simulation method can be used to continuously predict the reconstructed operation data within a period of time in the future by repeatedly executing steps A1 to A3, and then the reconstructed operation data is spliced ​​in chronological order to obtain the equivalent working condition within the predicted time interval, and the equivalent working condition is converted into a power working condition through an energy flow model to realize the reconstruction of the power working condition of the on-board hydrogen system; and then the reconstructed user behavior portrait within the time is obtained by feature extraction of the reconstructed operation data within the time, so as to input the reconstructed user behavior portrait into the mapping relationship to obtain the characteristic behavior hydrogen consumption under each reconstructed operation data. For example, the current user behavior portrait is used as the initial portrait, and the current moment is recorded as t0. After the reconstructed operation data at the moment t1 is predicted according to the initial portrait, the reconstructed user behavior portrait at the moment t1 is obtained through feature extraction, and the reconstructed operation data at the moment t2 is predicted according to the reconstructed user behavior portrait at the moment t1, and the reconstructed user behavior portrait at the moment t2 is obtained after feature extraction, and the reconstructed operation data at the moment t3 is predicted according to the reconstructed user behavior portrait at the moment t2, and the reconstructed user behavior portrait at the moment t3 is obtained after feature extraction, and the cycle is repeated until the preset time threshold tn is reached and the cycle stops. It should be noted that the preset time threshold is determined according to actual application conditions.

[0094] In the present invention, the remaining range of the vehicle is obtained by reading the current remaining hydrogen storage of the target hydrogen-powered vehicle and the hydrogen consumption of each reconstructed operation data calculated in step 106, and summing up the predicted user behavior profiles when the remaining hydrogen storage is 0. The specific calculation formula is as follows:

[0095]

[0096] Among them, CapacityHremain is the current remaining hydrogen storage capacity, CapacityH per_seg (i) is the hydrogen consumption of the i-th reconstructed operating data, mileageremain is the remaining cruising range, mileage per_seg (i) is the mileage of the i-th reconstructed operation data. It should be noted that the reconstructed operation data includes the vehicle speed per unit time, that is, the average vehicle speed, so the mileage of each predicted user behavior profile can be determined based on the vehicle speed and unit time.

[0097] After the remaining cruising range is obtained in step 108, the following steps are further included:

[0098] According to the characteristic behavior hydrogen consumption and the current remaining hydrogen storage under the reconstructed operation data, the instantaneous remaining cruising range at the time of the reconstructed operation data is calculated;

[0099] Each reconstructed operation data and the instantaneous remaining cruising range at the time of the reconstructed operation data are used as a feature combination, and the feature combinations are arranged in the order of time information to obtain a feature combination sequence;

[0100] The self-attention mechanism Transformer architecture is used to extract features from feature combination sequences to obtain high-dimensional features;

[0101] Concatenate the feature combination sequence and the high-dimensional feature to obtain a high-dimensional feature vector;

[0102] The high-dimensional feature vector is input into the pre-trained XGboost model and the corrected remaining range is output.

[0103] In a preferred embodiment, the pre-trained XGboost model is trained by the following method:

[0104] Based on historical operation data, the actual remaining range of the target hydrogen-powered vehicle in each segment user profile is obtained;

[0105] The user profile of each segment and the actual remaining range at the time of the user profile of the segment are used as training feature combinations, and the feature combinations are arranged in the order of time information to obtain a training feature combination sequence;

[0106] The self-attention mechanism Transformer architecture is used to extract features from the training feature combination sequence to obtain training high-dimensional features;

[0107] Concatenate the training feature combination sequence and the training high-dimensional feature to obtain a training high-dimensional feature vector;

[0108] Generate a plurality of second sample sets based on the training high-dimensional feature vector and the actual remaining cruising range at the most recent running time corresponding to the training high-dimensional feature vector;

[0109] The XGboost model is trained using the second sample set to obtain a pre-trained XGboost model; wherein the second sample set includes the training high-dimensional feature vector as input and the actual remaining cruising range at the most recent running time corresponding to the training high-dimensional feature vector as output.

[0110] Specifically, the input of Transformer needs to be constructed in the form of a time series, that is, a data sequence containing multiple time steps, representing the changes in the vehicle status over the past period of time. The feature dimension of the fragment user portrait is Q, the training feature combination is Q+1, the time window of the past period of time is selected as n minutes, and the data batch size is m, then the shape of the input data is: Inputshape = (m,n,Q+1). Then, through the self-attention mechanism of Transformer, it is possible to learn the correlation between the features of each moment and the features of other moments. For example, it is possible to capture the acceleration behavior, speed fluctuations, SOC consumption patterns of the vehicle over the past period of time, and how these features affect the current vehicle status and endurance prediction. After encoding through multiple layers of Transformer, Transformer will output a shape: Transfomeroutputshape = (m,n,d model ), where d model It is a high-dimensional feature dimension, representing the correlation between different input features. Although Transformer extracts high-dimensional features of time series, the original input features of each segment user portrait also contain important instantaneous information. Therefore, it is necessary to concatenate the high-dimensional features generated by Transformer with the original features at the current moment to obtain a more comprehensive high-dimensional feature vector. Concatenated shape = (m, n, d model +Q+1).

[0111] Specifically, the training process of the XGBoost model continuously iterates to build new regression trees, each time building a new tree to reduce the previous prediction error:

[0112] 1) Initial prediction: predict the remaining range through simple average or other initial strategies;

[0113] 2) Calculate the residual: Calculate the residual of each sample based on the difference between the initial prediction value and the actual experimental test; the residual is used to represent the error that needs to be further optimized in the model;

[0114] 3) Fitting residuals: In each iteration, the model builds a new regression tree specifically to fit these residuals in an attempt to reduce the error;

[0115] 4) Update prediction: After the new regression tree is generated, the model's prediction value will be updated to make it closer to the actual result;

[0116] 5) Repeat process: This process will continue to iterate and build multiple regression trees until the predetermined number of trees is reached or the error converges.

[0117] In the present invention, the calculated remaining cruising range is corrected by the Transformer-XGboost algorithm model, thereby combining the time series feature extraction capability of Transformer and the regression capability of XGBoost with the actual experimental test results to complete the correction and improve the accuracy and precision of the remaining cruising range. XGBoost also uses regularization technology to prevent the model from overfitting, and by constraining the complexity of the model, it ensures that the model performs well on the training set and maintains good generalization performance on the test set.

[0118] like Figure 3 , Figure 4 As shown, an embodiment of the present invention provides a remaining cruising range estimation device based on the equivalent operating conditions of the vehicle-mounted hydrogen system. The device embodiment can be implemented by software, or by hardware or a combination of software and hardware. From the hardware level, Figure 3 As shown in FIG. 1 , a hardware architecture diagram of a computing device where a remaining cruising range estimation device based on an equivalent operating condition of an onboard hydrogen system provided by an embodiment of the present invention is located is shown in FIG. Figure 3 In addition to the processor, memory, network interface, and non-volatile memory shown in the figure, the computing device in which the device is located in the embodiment may also generally include other hardware, such as a forwarding chip responsible for processing messages, etc. Taking software implementation as an example, Figure 4 As shown, as a device in a logical sense, the CPU of the computing device in which it is located reads the corresponding computer program in the non-volatile memory into the memory and runs it. This embodiment provides a remaining cruising range estimation device based on the equivalent operating condition of the vehicle-mounted hydrogen system, and the device includes:

[0119] The feature extraction module 400 is used to extract features from the current operating data of the target hydrogen-powered vehicle to obtain a current user behavior profile;

[0120] Prediction module 402, used to input the current user behavior profile and the driving mode to which the current user behavior profile belongs into a pre-trained time series prediction model to obtain a predicted driving mode at the next moment;

[0121] A reconstruction module 404, configured to reconstruct the working condition of the predicted driving mode according to the historical operating data corresponding to each driving mode, and obtain the reconstructed operating data under the predicted driving mode;

[0122] A hydrogen consumption calculation module 406 is used to determine the characteristic behavior hydrogen consumption under the reconstructed operating data according to the reconstructed operating data and the pre-constructed mapping relationship between the operating condition and the hydrogen consumption;

[0123] The mileage calculation module 408 is used to calculate the remaining cruising range according to the characteristic behavior hydrogen consumption under the reconstructed operation data within the prediction time interval and the current remaining hydrogen storage of the target hydrogen-powered vehicle.

[0124] In some specific embodiments, the feature extraction module 400 can be used to execute the above step 100, the prediction module 402 can be used to execute the above step 102, the reconstruction module 404 can be used to execute the above step 104, the hydrogen consumption calculation module 406 can be used to execute the above step 106, and the mileage calculation module 408 can be used to execute the above step 108.

[0125] In one embodiment of the present invention, the device further includes a preprocessing module, which is used to perform the following operations:

[0126] Divide the historical operation data of the target hydrogen-powered vehicle into equal time segments to obtain a number of historical operation segments including time information and unit time operation data;

[0127] Extract features from several historical operation segments to obtain several segment user portraits;

[0128] The density clustering algorithm is used to classify the fragmented user portraits and determine the driving mode to which each fragmented user portrait belongs.

[0129] In one embodiment of the present invention, the preprocessing module is further used to perform the following operations:

[0130] Obtaining the unit time operation data of the historical operation fragments; wherein the unit time operation data includes the unit time operation data of the whole vehicle data, the hydrogen system operation data, the power battery data, and the motor data;

[0131] Perform feature extraction on each unit time operation data to obtain a number of features; wherein the number of features is the same as the number of quantity types included in the unit time operation data;

[0132] After correlation analysis of the features, effective key features are screened out;

[0133] Generate fragmented user portraits based on effective key features.

[0134] In one embodiment of the present invention, the device further includes a first training module, and the first training module is further configured to perform the following operations:

[0135] Generate a plurality of first sample sets according to the segment user portraits and the driving modes to which the segment user portraits belong;

[0136] The self-attention mechanism Transformer architecture is trained using the first sample set to obtain a pre-trained time series prediction model; wherein the first sample set includes a segmented user portrait at the current moment and the driving mode to which the segmented user portrait belongs as input, and a driving mode to which the segmented user portrait belongs at the next moment as output.

[0137] In one embodiment of the present invention, the reconstruction module 404 is further configured to perform the following operations:

[0138] Determine the historical operation data corresponding to each driving mode from the historical operation data of the target hydrogen-powered vehicle, and obtain each historical operation segment included in the predicted driving mode at the next moment;

[0139] Calculating similarity values ​​of the operation segment at the current moment and each historical operation segment included in the predicted driving mode at the next moment, and determining the highest similarity value;

[0140] The unit time operation data included in the historical operation segment corresponding to the highest similarity value is determined as the reconstructed operation data at the next moment corresponding to the operation segment at the current moment.

[0141] In one embodiment of the present invention, the apparatus further includes a second training module, and the second training module is further configured to perform the following operations:

[0142] The pre-built mapping relationship between operating conditions and hydrogen consumption is obtained by the following method:

[0143] Establish a nonlinear function relationship with segment user portrait as input and hydrogen consumption as output;

[0144] The optimal solution of the parameters of the nonlinear function relationship is solved to obtain the mapping relationship between operating conditions and hydrogen consumption.

[0145] In one embodiment of the present invention, the second training module is further configured to perform the following operations:

[0146] S1: Initialize the population, determine the initial temperature, temperature cooling rate, cutoff temperature, maximum number of iterations, initial energy of prey, and feasible domain of feasible solutions of parameters, and randomly generate a set of potential feasible solutions in the feasible domain;

[0147] S2: In the current iteration, the potential feasible solutions are optimized in the feasible domain based on the Harris Hawk algorithm to obtain the first updated feasible solution output in the exploration phase and the second updated feasible solution output in the development phase;

[0148] S3: Determine whether to switch from the exploration phase to the development phase according to the prey energy of the current iteration and the number of iterations; if yes, execute step S4; otherwise, execute step S1;

[0149] S4: determining an optimization strategy according to the prey energy, and optimizing the first updated feasible solution using the optimization strategy to obtain a second updated feasible solution;

[0150] S5: respectively using the optimal solution in the current iteration and the second updated feasible solution to calculate the degree of fit of the nonlinear function relationship. If the calculated degree of fit of the second updated feasible solution is lower than the degree of fit of the optimal solution in the current iteration, then executing step S6; otherwise, executing step S7;

[0151] S6: Calculate the acceptance probability of the second updated feasible solution according to the initial temperature, the temperature cooling rate, the second updated feasible solution and the prey position of the current iteration, so as to accept the second updated feasible solution according to the acceptance probability; if the second updated feasible solution is accepted, execute step S7; if the second updated feasible solution is not accepted, retain the optimal solution in the current iteration, update the current temperature, and execute step S8;

[0152] S7: Determine the second updated feasible solution as the optimal solution of the current iteration;

[0153] S8: Determine whether the current number of iterations reaches the maximum number of iterations or the current temperature reaches the cutoff temperature. If so, output the optimal solution; otherwise, use the optimal solution as the initial solution for the next iteration and return to step S2.

[0154] In one embodiment of the present invention, the reconstruction module 404 is further configured to perform the following operations:

[0155] Extract features from the reconstructed operation data to obtain a reconstructed user behavior profile under the predicted driving mode;

[0156] The reconstructed user behavior profile is input into the pre-constructed mapping relationship between operating conditions and hydrogen consumption to obtain the characteristic behavior hydrogen consumption under the reconstructed user behavior profile.

[0157] In one embodiment of the present invention, the remaining cruising range is calculated based on the predicted hydrogen consumption under the user behavior profile within the predicted time interval and the current remaining hydrogen storage of the target hydrogen-powered vehicle, including:

[0158] A1: Extract features from the current running data to obtain the current user behavior profile. Then input the current user behavior profile and the driving mode to which the current user behavior profile belongs into the pre-trained time series prediction model to obtain the predicted driving mode at the next moment.

[0159] A2: Reconstruct the working condition of the predicted driving mode according to the historical operating data corresponding to each driving mode to obtain the reconstructed operating data under the predicted driving mode;

[0160] A3: Determine whether the current cumulative prediction time length is greater than a preset time threshold. If so, obtain the reconstructed operation data within the prediction time interval and execute step A4; otherwise, use the reconstructed operation data at the next moment as the current operation data and return to step A1;

[0161] A4: Input the reconstructed user behavior portrait obtained by feature extraction of the reconstructed operation data within the prediction time interval into the mapping relationship to obtain the characteristic behavior hydrogen consumption under each reconstructed operation data;

[0162] A5: Accumulate the characteristic behavior hydrogen consumption under the reconstructed operation data in the order of prediction time to obtain the total hydrogen consumption; when the total hydrogen consumption is the same as the current remaining hydrogen storage, calculate the sum of the mileage of the reconstructed operation data included in the total hydrogen consumption to obtain the remaining cruising range.

[0163] In one embodiment of the present invention, the mileage calculation module 408 is further configured to perform the following operations:

[0164] According to the characteristic behavior hydrogen consumption and the current remaining hydrogen storage under the reconstructed operation data, the instantaneous remaining cruising range at the time of the reconstructed operation data is calculated;

[0165] Each reconstructed operation data and the instantaneous remaining cruising range at the time of the reconstructed operation data are used as a feature combination, and the feature combinations are arranged in the order of time information to obtain a feature combination sequence;

[0166] The self-attention mechanism Transformer architecture is used to extract features from feature combination sequences to obtain high-dimensional features;

[0167] Concatenate the feature combination sequence and the high-dimensional feature to obtain a high-dimensional feature vector;

[0168] The high-dimensional feature vector is input into the pre-trained XGboost model and the corrected remaining range is output.

[0169] It is understandable that the structure illustrated in the embodiment of the present invention does not constitute a specific limitation on a remaining cruising range estimation device based on the equivalent operating condition of an on-board hydrogen system. In other embodiments of the present invention, a remaining cruising range estimation device based on the equivalent operating condition of an on-board hydrogen system may include more or fewer components than shown in the figure, or combine certain components, or split certain components, or arrange the components differently. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0170] The information interaction, execution process and other contents between the modules in the above-mentioned device are based on the same concept as the embodiment of the method of the present invention. For the specific contents, please refer to the description in the embodiment of the method of the present invention, and no further description is given here.

[0171] An embodiment of the present invention also provides a computing device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, a remaining cruising range estimation method based on the equivalent operating conditions of the on-board hydrogen system in any embodiment of the present invention is implemented.

[0172] An embodiment of 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 processor executes a remaining cruising range estimation method based on the equivalent operating conditions of an on-board hydrogen system in any embodiment of the present invention.

[0173] An embodiment of the present application also provides a computer program product, which includes a computer program. A processor of a computer device reads the computer program from a computer-readable storage medium, and the processor executes the computer program, so that the computer device executes any of the above-mentioned embodiments. A method for estimating remaining cruising range based on the equivalent operating conditions of a vehicle-mounted hydrogen system.

[0174] Specifically, a system or device equipped with a storage medium can be provided, on which software program code that implements the functions of any of the above-mentioned embodiments is stored, and a computer (or CPU or MPU) of the system or device can be enabled to read and execute the program code stored in the storage medium.

[0175] In this case, the program code itself read from the storage medium can realize the function of any one of the above-mentioned embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.

[0176] The storage medium embodiments for providing the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), a magnetic tape, a non-volatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer by a communication network.

[0177] In addition, it should be clear that the functions of any of the above embodiments can be implemented not only by executing the program code read by the computer, but also by enabling the operating system on the computer to complete part or all of the actual operations based on the instructions of the program code.

[0178] In addition, it can be understood that the program code read from the storage medium is written to a memory provided in an expansion board inserted into the computer or to a memory provided in an expansion module connected to the computer, and then based on the instructions of the program code, a CPU installed on the expansion board or expansion module is enabled to perform part or all of the actual operations, thereby realizing the functions of any of the above-mentioned embodiments.

[0179] It should be noted that, in this article, 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 any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical factors in the process, method, article or device including the elements.

[0180] A person of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiments; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, etc., various media that can store program codes.

[0181] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating remaining cruising range based on equivalent operating conditions of a vehicle-mounted hydrogen system, characterized in that: include: Extract features from the current operating data of the target hydrogen-powered vehicle to obtain the current user behavior profile; Inputting the current user behavior profile and the driving mode to which the current user behavior profile belongs into a pre-trained time series prediction model to obtain a predicted driving mode at the next moment; Reconstructing the working condition of the predicted driving mode according to the historical operating data corresponding to each driving mode to obtain the reconstructed operating data under the predicted driving mode; Determining the characteristic behavior hydrogen consumption under the reconstructed operating data according to the reconstructed operating data and the pre-constructed mapping relationship between the operating condition and the hydrogen consumption; The remaining cruising range is obtained based on the characteristic behavior hydrogen consumption under the reconstructed operating data within the prediction time interval and the current remaining hydrogen storage capacity of the target hydrogen-powered vehicle.

2. The method according to claim 1, characterized in that Before extracting features from the current operating data of the target hydrogen-powered vehicle, the method further includes: Dividing the historical operation data of the target hydrogen-powered vehicle into equal time segments to obtain a plurality of historical operation segments including time information and unit time operation data; Extract features from the historical operation segments to obtain user portraits of the segments; A density clustering algorithm is used to classify the segmented user portraits to determine the driving mode to which each segmented user portrait belongs.

3. The method according to claim 2, characterized in that The pre-trained time series prediction model is trained by the following method: generating a plurality of first sample sets according to the segment user portraits and the driving modes to which the segment user portraits belong; The self-attention mechanism Transformer architecture is trained using the first sample set to obtain the pre-trained time series prediction model; wherein the first sample set includes the segment user portrait at the current moment and the driving mode to which the segment user portrait belongs as input, and the driving mode to which the segment user portrait belongs at the next moment as output; and / or, Reconstructing the working condition of the predicted driving mode according to the historical operating data corresponding to each driving mode to obtain the reconstructed operating data under the predicted driving mode includes: Determining the historical operation data corresponding to each of the driving modes from the historical operation data of the target hydrogen-powered vehicle, and obtaining each historical operation segment included in the predicted driving mode at the next moment; respectively calculating similarity values ​​between the operation segment at the current moment and each historical operation segment included in the predicted driving mode at the next moment, and determining the highest similarity value; The unit time operation data included in the historical operation segment corresponding to the highest similarity value is determined as the reconstructed operation data at the next moment corresponding to the operation segment at the current moment.

4. The method according to claim 2, characterized in that: The pre-constructed mapping relationship between the operating conditions and the hydrogen consumption is obtained by the following method: Establishing a nonlinear function relationship with the segment user portrait as input and hydrogen consumption as output; Solving the optimal solution of the parameters of the nonlinear function relationship to obtain a mapping relationship between the operating condition and hydrogen consumption; and / or, Determining the characteristic behavior hydrogen consumption under the reconstructed operation data according to the reconstructed operation data and the pre-constructed mapping relationship between the operation condition and the hydrogen consumption includes: Performing feature extraction on the reconstructed operation data to obtain a reconstructed user behavior profile under the predicted driving mode; The reconstructed user behavior profile is input into the pre-constructed mapping relationship between the operating condition and the hydrogen consumption to obtain the characteristic behavior hydrogen consumption under the reconstructed user behavior profile.

5. The method according to claim 4, characterized in that The step of solving the optimal solution of the parameters of the nonlinear functional relationship includes: S1: Initialize the population, determine the initial temperature, temperature cooling rate, cutoff temperature, maximum number of iterations, initial energy of prey and feasible domain of feasible solutions of the parameters, and randomly generate a set of potential feasible solutions in the feasible domain; S2: In the current iteration, the potential feasible solution is optimized in the feasible domain based on the Harris Hawk algorithm to obtain the first updated feasible solution output in the exploration phase; S3: Determine whether to switch from the exploration phase to the development phase according to the prey energy of the current iteration and the number of iterations; if yes, execute step S4; otherwise, execute step S1; S4: determining an optimization strategy according to the prey energy, and optimizing the first updated feasible solution by using the optimization strategy to obtain a second updated feasible solution; S5: respectively using the optimal solution in the current iteration and the second updated feasible solution to calculate the degree of fit of the nonlinear function relationship, if the calculated degree of fit of the second updated feasible solution is lower than the degree of fit of the optimal solution in the current iteration, executing step S6; otherwise, executing step S7; S6: Calculate the acceptance probability of the second updated feasible solution according to the initial temperature, the temperature cooling rate, the second updated feasible solution and the prey position of the current iteration, so as to accept the second updated feasible solution according to the acceptance probability; if the second updated feasible solution is accepted, execute step S7; if the second updated feasible solution is not accepted, retain the optimal solution in the current iteration, update the current temperature, and execute step S8; S7: Determine the second updated feasible solution as the optimal solution of the current iteration; S8: Determine whether the current number of iterations reaches the maximum number of iterations or whether the current temperature reaches the cutoff temperature. If so, output the optimal solution; otherwise, use the optimal solution as the initial solution for the next iteration and return to step S2.

6. The method according to claim 1, characterized in that The remaining cruising range is calculated based on the characteristic behavior hydrogen consumption under the reconstructed operation data within the predicted time interval and the current remaining hydrogen storage of the target hydrogen-powered vehicle, including: A1: extracting features from the current running data to obtain the current user behavior profile; then inputting the current user behavior profile and the driving mode to which the current user behavior profile belongs into the pre-trained time series prediction model to obtain a predicted driving mode at the next moment; A2: reconstructing the working condition of the predicted driving mode according to the historical operating data corresponding to each driving mode to obtain the reconstructed operating data under the predicted driving mode; A3: Determine whether the current cumulative prediction time length is greater than a preset time threshold. If so, obtain the reconstructed operation data within the prediction time interval and execute step A4; otherwise, use the reconstructed operation data at the next moment as the current operation data and return to step A1; A4: Inputting the reconstructed user behavior portrait obtained by feature extraction of the reconstructed operation data within the predicted time interval into the mapping relationship to obtain the characteristic behavior hydrogen consumption under each of the reconstructed operation data; A5: Accumulate the characteristic behavior hydrogen consumption under the reconstructed operation data in the order of predicted time to obtain the total hydrogen consumption; when the total hydrogen consumption is the same as the current remaining hydrogen storage capacity, calculate the sum of the mileage of the reconstructed operation data included in the total hydrogen consumption to obtain the remaining cruising range.

7. The method according to any one of claims 2 to 6, characterized in that: The feature extraction of the plurality of historical operation segments is performed to obtain a plurality of segment user portraits, including: Acquire the unit time operation data of the historical operation segment; wherein the unit time operation data includes vehicle data, hydrogen system operation data, power battery data, and motor data; Performing feature extraction on each of the unit time operation data to obtain a number of features; After performing correlation analysis on the features, effective key features are screened out; Generate the segment user portrait based on the effective key features; and / or, After obtaining the remaining cruising range, the method further includes: Calculate the instantaneous remaining cruising range at the time of the reconstructed operation data according to the characteristic behavior hydrogen consumption under the reconstructed operation data and the current remaining hydrogen storage amount; Taking each of the reconstructed operating data and the instantaneous remaining cruising range at the time of the reconstructed operating data as a feature combination, and arranging the feature combinations in the order of the time information to obtain a feature combination sequence; A self-attention mechanism Transformer architecture is used to extract features from the feature combination sequence to obtain high-dimensional features; Concatenate the feature combination sequence and the high-dimensional feature to obtain a high-dimensional feature vector; The high-dimensional feature vector is input into the pre-trained XGboost model, and the corrected remaining cruising range is output.

8. A device for estimating remaining cruising range based on equivalent operating conditions of a vehicle-mounted hydrogen system, characterized in that: include: The feature extraction module is used to extract features from the current operating data of the target hydrogen-powered vehicle to obtain the current user behavior portrait; A prediction module, used for inputting the current user behavior profile and the driving mode to which the current user behavior profile belongs into a pre-trained time series prediction model to obtain a predicted driving mode at the next moment; a reconstruction module, configured to reconstruct the working condition of the predicted driving mode according to the historical operating data corresponding to each of the driving modes, so as to obtain the reconstructed operating data under the predicted driving mode; A hydrogen consumption calculation module, used to determine the characteristic behavior hydrogen consumption under the reconstructed operating data according to the reconstructed operating data and a pre-constructed mapping relationship between operating conditions and hydrogen consumption; The mileage calculation module is used to calculate the remaining cruising range based on the characteristic behavior hydrogen consumption under the reconstructed operation data within the predicted time interval and the current remaining hydrogen storage capacity of the target hydrogen-powered vehicle.

9. A computing device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute the method according to any one of claims 1 to 7.

Images