Improved thermostat fuel cell automobile energy management strategy based on working condition clustering
Through an improved thermostat energy management strategy based on operating condition clustering, the fuel cell power output under different operating conditions is identified and allocated, and the problems of response hysteresis and fuel cell aging in complex road conditions in the prior art are solved, and the economic and durability optimization of fuel cell vehicles is achieved.
Patent Information
- Application Number
- CN202510188575.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-20
- Publication Date
- 2025-06-27
AI Technical Summary
The existing fuel cell vehicle energy management strategy has lagged response under complex road conditions and cannot meet dynamic power requirements. At the same time, it increases the load fluctuation frequency of fuel cells, resulting in accelerated aging and reduced durability.
The improved thermostat energy management strategy based on working condition clustering is adopted, and the working condition characteristics during the vehicle's driving process are collected, dimensionality reduction processing and cluster analysis are carried out to identify three typical working conditions of urban congestion, suburban roads and highways, and the power output of the fuel cell is dynamically adjusted according to these working condition categories.
The adaptive power distribution of fuel cell vehicles under different operating conditions has been achieved, and the economic benefits of the system and the service life of fuel cells and power batteries have been improved. Simulation verification shows that the total cost has been reduced by 10.56% to 28.73%.
Smart Images
Figure CN120207173A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of energy management of fuel cell vehicles, and particularly to an improved thermostat fuel cell vehicle energy management strategy based on driving condition clustering. Background Art
[0002] With the improvement of environmental protection requirements and the intensification of the energy crisis, fuel cell vehicles (FCEVs), as a type of clean energy vehicle, have gradually become a research hotspot due to their advantages of zero emissions and high efficiency. By the collaborative work of a fuel cell system and energy storage devices such as lithium-ion batteries, fuel cell vehicles can achieve a longer driving range and excellent power performance. However, due to the slow dynamic response speed, high cost, and susceptibility to load fluctuations of fuel cells, a reasonable energy management strategy is crucial for improving fuel economy and extending the system life under different driving conditions.
[0003] The existing energy management strategies for fuel cell vehicles mainly fall into two categories: the thermostat strategy and the power following strategy. The thermostat strategy keeps the output power of the fuel cell stable to reduce the frequent start-stop caused by dynamic loads and the wear of the fuel cell stack, which can extend the service life of the fuel cell system. However, this strategy has a poor response to load fluctuations, easily leading to system response lag and thus being unable to meet the dynamic power demands under complex road conditions. The power following strategy adjusts the output power of the fuel cell by following the vehicle's power demand in real time, reducing the discharge of the lithium-ion battery and improving fuel utilization. However, this strategy increases the load fluctuation frequency of the fuel cell, easily causing accelerated aging of the fuel cell and reducing durability.
[0004] In addition, in a hybrid power system where a fuel cell and a battery work together, the key challenge in achieving energy management lies in the accuracy of driving condition identification. The vehicle has different requirements for power demand fluctuations and response capabilities under different driving conditions (such as urban congestion, highways, suburbs, etc.). Therefore, it is particularly important to adaptively adjust the power distribution between the fuel cell and the battery according to the actual driving conditions. However, traditional strategies have deficiencies in dealing with complex road conditions, are unable to flexibly respond to rapidly changing load demands, and lack effective driving condition identification methods, often resulting in low fuel cell efficiency and shortened system life. Summary of the Invention
[0005] Aiming at the deficiencies existing in the prior art, the purpose of the present invention is to provide an improved thermostat fuel cell vehicle energy management strategy based on driving condition clustering, which can adjust the output power response of the fuel cell in real time, not only improving the economic benefits of the system, but also effectively increasing the service life of the fuel cell and the power battery. To achieve the above object and other advantages according to the present invention, an improved thermostat fuel cell vehicle energy management strategy based on driving condition clustering is provided, including the following steps:
[0006] S1. Extract the characteristics of the driving conditions of the vehicle during actual driving and perform dimensionality reduction processing on the said characteristics;
[0007] S2. Through an improved binary K-means clustering algorithm, cluster the dimensionality-reduced feature data in step S1, and divide the driving conditions of the vehicle into three types of typical driving conditions, including: urban congestion, suburban roads, and highway driving conditions;
[0008] S3. Based on the driving condition categories identified in step S2, dynamically adjust the power output of the fuel cell through an adaptive power distribution strategy to optimize the fuel utilization rate of the vehicle and the service life of the fuel cell.
[0009] Preferably, 19 driving condition characteristic parameters during the vehicle driving process are collected in step S1, such as speed, acceleration, and the proportion of parking time. Use principal component analysis (PCA) to perform dimensionality reduction on these characteristics, reduce the calculation amount and remove redundancy, and finally retain 5 principal components, which can cover 87.4% of the information volume.
[0010] Preferably, use the binary K-means clustering algorithm to cluster the dimensionality-reduced data, and divide the driving conditions into three categories: urban congestion, suburban roads, and highways. Through the iterative process, continuously bisect the clusters until the preset number of categories is reached to minimize the sum of squared errors.
[0011] Preferably, according to the clustering results, set different fuel cell power outputs for each driving condition category to adapt to different driving requirements. Through the adaptive power distribution strategy, optimize the fuel utilization rate and the service life of the fuel cell.
[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: By extracting the driving condition characteristics from historical data and combining principal component analysis with an improved K-means clustering method, different driving conditions are divided into several categories, and based on this, the adaptive power distribution of the fuel cell is realized, achieving the comprehensive optimization of economy and durability. This method can adjust the output power response of the fuel cell in real time, not only improving the economic benefits of the system, but also effectively increasing the service life of the fuel cell and the power battery.
[0013] Through an improved thermostat energy management strategy, adaptive power distribution of fuel cell vehicles under different working conditions is achieved, enabling the vehicle to balance between dynamic demand response and system life. Simulation verification shows that compared with traditional thermostat strategies and power following strategies, the present invention has significant advantages in achieving durability and fuel economy, effectively improving the overall economic efficiency of the system, and is applicable to fuel cell vehicle applications under various complex working conditions. Description of the Drawings
[0014] Figure 1 It is a schematic flow chart of an improved thermostat fuel cell vehicle energy management strategy based on working condition clustering according to the present invention;
[0015] Figure 2 It is a process diagram of working condition feature extraction and dimensionality reduction of an improved thermostat fuel cell vehicle energy management strategy based on working condition clustering according to the present invention;
[0016] Figure 3 It is an operation flow chart of a working condition identifier based on improved K-means clustering of an improved thermostat fuel cell vehicle energy management strategy according to the present invention. Detailed Embodiment
[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0018] Refer to Figure 1 , an improved thermostat fuel cell vehicle energy management strategy based on working condition clustering, includes the following steps:
[0019] S1. Working condition feature extraction and feature dimensionality reduction
[0020] In this embodiment, first, the working condition data of the vehicle during actual driving is collected, including 19 characteristic parameters such as vehicle speed, acceleration, proportion of parking time, and driving mileage. The specific parameters are shown in the following table:
[0021] Table 1 Working condition characteristic parameters and their meanings
[0022]
[0023]
[0024] These parameters are used to characterize different operating conditions of the vehicle, covering typical driving scenarios such as urban congestion, suburban roads, and highways. Since directly using 19 characteristic parameters will lead to large computational amounts and redundant features, principal component analysis is adopted to reduce the dimensionality of the 19 characteristic parameters. The specific dimensionality reduction process is as follows:
[0025] Select representative operating conditions for combination to form an operating condition cycle, and divide the above combined operating conditions at a fixed step length with a period of 30 seconds. The original combined operating conditions are divided into 264 operating condition segments, and the 19 characteristic data of the 264 samples are standardized to remove the influence of dimension and numerical differences, forming an initial standardized matrix as shown in the following formula:
[0026]
[0027] where X ij is the j operating condition data of the i operating condition segment.
[0028] According to the following formula, calculate the correlation coefficient matrix of the standardized data to analyze the correlation between each feature:
[0029]
[0030] where r ij is the correlation coefficient between the characteristic parameters x i and x j . r ij is equal in magnitude to r ji .
[0031] Calculate the eigenvalues and eigenvectors of the correlation coefficient matrix R, and select the first 5 principal components with larger eigenvalues to ensure that the cumulative contribution rate exceeds 85%, representing the original 19 features. After dimensionality reduction, 5 principal components are retained, and these principal components can effectively represent the driving characteristics of the vehicle and cover 87.4% of the information volume. The information of each operating condition can be expressed by the product of the five principal component eigenvectors and the corresponding coefficients, as shown in the following formula:
[0032]
[0033] F ij = a j1 y i1 + a j2 y i2 +…+ a jp y ip (i = 1, 2, …, n, j = 1, 2, …, k)
[0034] where F ij is the operating condition information, a is the coefficient, and y is the principal component eigenvector
[0035] For the specific process, seeFigure 2
[0036] S2. Improved K-means clustering
[0037] Using the improved binary K-means clustering algorithm, cluster the five-dimensional feature data after dimensionality reduction in S1, and divide the driving conditions of the vehicle into three typical conditions: urban congestion, suburban roads, and highway conditions. The specific steps are as follows:
[0038] S21 Initialization: Treat the sample data points of all 264 condition segments as a single cluster, and calculate the mean of the five principal component features of all sample data points as the initial centroid.
[0039] S22 Evaluation matrix: Create a matrix with the same number of rows and columns of two as the number of sample points to store the centroid assigned to each point and the distance from the point to the centroid.
[0040] S23 Iteration process: When the number of clusters is less than the preset number of categories (3 in this example), perform the following steps:
[0041] Set a very large parameter "minimum SSE".
[0042] Traverse each cluster and perform 2-means clustering on the current cluster, that is, divide the cluster into two sub-clusters.
[0043] Calculate the SSE (sum of squared errors) of the newly formed sub-clusters and add it to the SSE of the clusters that have not been traversed to obtain the current total SSE.
[0044] If the current total SSE is less than the recorded minimum SSE, update the minimum SSE and record the number of the cluster that was bisected currently.
[0045] Update the total number of clusters, update the first column of the evaluation matrix, and assign the newly formed sub-clusters to new category numbers.
[0046] Update the centroid coordinates, and use the clustering center of the newly formed sub-clusters as the new centroid.
[0047] Update the centroid assignment and error of each point in the divided cluster.
[0048] S24 Termination condition: When the number of clusters reaches the preset number of categories, end the iteration.
[0049] S25 Output result: Return the final K clusters and the centroid coordinates of each cluster.
[0050] Through the above steps, 264 working condition segments can be effectively classified into 3 categories according to 5 principal component features, enabling the segments in each category to represent the main features of the category as much as possible, while there are obvious differences in working condition features between different categories. This improved K-means clustering algorithm, namely binary K-means clustering, can better handle scenarios with a large sample size and reduce the risk of falling into local optima. The specific process is as Figure 3 .
[0051] S3. Working condition classifier
[0052] Based on the working condition categories identified in step S2, the power output of the fuel cell is dynamically adjusted through an adaptive power allocation strategy to optimize the fuel utilization rate of the vehicle and the lifespan of the fuel cell. The specific power allocation strategy is as follows:
[0053] Urban congestion working condition, i.e., working condition category 1: When the vehicle is in an urban congestion state with low speed and frequent stops, the fuel cell output is set to 50% of the rated power. This low power output is suitable for the driving requirements of low speed and low acceleration, and helps to reduce the impact of power fluctuations on the fuel cell.
[0054] Suburban road working condition, i.e., working condition category 2: When the vehicle is in a suburban road working condition with medium speed, the fuel cell output is set to 80% of the rated power to ensure sufficient power support for the vehicle when driving at medium speed.
[0055] Highway working condition, i.e., working condition category 3: When the vehicle is in a highway driving state, the fuel cell output is set to 110% of the rated power to meet the requirements of high speed and large acceleration.
[0056] Under this strategy, the vehicle can automatically adapt to changes in the driving environment, select the optimal power output according to the identified working condition category, and achieve adaptive adjustment of energy management. The simulation results show that the strategy proposed in the present invention optimizes the fuel utilization efficiency under the three working conditions. Compared with the traditional thermostat and power following strategies, the total cost is reduced by 10.56% (compared with the thermostat strategy) and 28.73% (compared with the power following strategy). In addition, the state of the power battery remains stable during the simulation, and the power fluctuations of the fuel cell system are significantly reduced, extending the service life of the system.
[0057] The number of devices and the processing scale described herein are used to simplify the description of the present invention, and applications, modifications, and variations of the present invention will be apparent to those skilled in the art. Although the embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. It can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily achieved. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to specific details and the illustrated examples described herein.
Claims
1. An improved thermostat fuel cell vehicle energy management strategy based on operating condition clustering, characterized in that: The following steps are involved: S1, extracting features of the working conditions of the vehicle during actual driving and performing dimensionality reduction processing on the features; S2, clustering the feature data after dimension reduction in step S1 by using an improved binary K-means clustering algorithm, and dividing the vehicle driving conditions into three typical conditions, including: urban congestion, suburban roads and highway conditions; S3. Based on the operating condition category identified in step S2, dynamically adjust the power output of the fuel cell through an adaptive power allocation strategy to optimize the fuel utilization of the vehicle and the life of the fuel cell.
2. The improved thermostat fuel cell vehicle energy management strategy based on operating condition clustering as claimed in claim 1, characterized in that: The feature dimension reduction in step S1 specifically includes the following steps: S11, selecting representative working conditions for combination, combining them into working condition cycles, and dividing the combined working conditions into fixed step lengths with a cycle of 30 seconds; S12, dividing the original combined working condition into a number of working condition segments, standardizing the characteristic data of the several working condition segments, removing the influence of dimension and value differences, and forming a matrix of initial matrix standardization; S13, calculating the correlation coefficient matrix of the standardized data to analyze the correlation between the features; S14. Calculate the eigenvalues and eigenvectors of the correlation coefficient matrix R, and select the first five principal components with the largest eigenvalues.
3. The improved thermostat fuel cell vehicle energy management strategy based on operating condition clustering as claimed in claim 1, characterized in that: In step S2, clustering the feature data after dimensionality reduction in step S1 by using an improved binary K-means clustering algorithm specifically includes the following steps: S21, initialization: the sample data points of several working condition segments are regarded as a cluster, and the mean of the five principal component features of all sample data points is calculated as the initial centroid; S22, evaluation matrix: create a matrix with two rows and columns equal to the number of sample points, used to store the centroid assigned to each point and the distance from the point to the centroid; S23, iterating, and ending the iteration when the number of clusters reaches the preset number of categories; S24. Output result: Return the final K clusters and the centroid coordinates of each cluster.
4. The improved thermostat fuel cell vehicle energy management strategy based on operating condition clustering as claimed in claim 1, characterized in that: The power allocation strategy in step S3 is as follows: Urban congestion conditions: When the vehicle is in a low-speed, frequently stopped urban congestion state, the fuel cell output is set to 50% of the rated power; Suburban road conditions: When the vehicle is in a medium-speed suburban road condition, the fuel cell output is set to 80% of the rated power; Highway condition: When the vehicle is in highway driving state, the fuel cell output is set to 110% of the rated power.