Fuel cell low frequency fluctuation energy consumption optimization system and optimization method

Abstract

The application discloses a kind of fuel cell low-frequency fluctuation energy consumption optimization system and optimization method, and optimization method includes: multi-source operation data millisecond level synchronous acquisition and big data cleaning management, and builds start-stop operating condition special database;Realize bus start-stop operating condition automatic labeling hierarchical identification based on clustering algorithm;Big data fitting start-stop energy consumption loss mapping model, quantifies power fluctuation additional hydrogen consumption;Start-stop operating condition adaptive stack control logic optimization, power battery buffer transient power fluctuation;Vehicle hydrogen consumption data real-time feedback, cloud iterative update energy consumption mapping model and control parameter;And vehicle power distribution instruction output execution.The fuel cell low-frequency fluctuation energy consumption optimization method of the application can solve the industry pain points that traditional fixed control strategy is high in hydrogen consumption in bus start-stop scene, strategy solidification, power and energy consumption cannot be balanced, realizes adaptive hydrogen consumption reduction in whole line, whole life cycle.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for energy management of fuel cell buses, and is particularly applicable to the optimization system and method for low-frequency fluctuation energy consumption of fuel cells in scenarios such as dense urban bus stops, traffic light congestion, low-speed crawling, and frequent start-stop. It belongs to the field of refined optimization control of hydrogen consumption in fuel cell systems and big data-driven energy-saving control technology for fuel cell vehicles. Background Technology

[0002] The core operating conditions of urban fuel cell buses are continuous start-stop and low-speed congestion, with sustained low-frequency and large fluctuations in load power, which are the main sources of additional hydrogen consumption for the entire vehicle. Current mainstream fuel cell energy management solutions in the industry all adopt static fixed-parameter control logic, relying on bench calibration experience to set uniform stack start-stop thresholds and power ramp-up / down slopes. They fail to perform big data adaptive optimization for the specific start-stop scenarios of buses, resulting in four inherent defects:

[0003] Operating condition identification relies on manual calibration and lacks automated data classification capabilities. Existing solutions only rely on simple thresholds for vehicle speed to roughly distinguish between congestion and constant speed. They do not incorporate massive amounts of multi-source time-series data from public transport operations, and cannot automatically classify and label based on start-stop frequency, idling duration, and congestion level. They cannot accurately identify high-energy-consuming operating conditions with heavy continuous start-stop cycles, and cannot switch energy-saving control strategies accordingly.

[0004] The lack of a quantitative start-stop energy loss model, the lack of data support for control parameters, the reliance on bench test experience values ​​for traditional control logic, the failure to statistically analyze the coupling relationship between power fluctuation amplitude and instantaneous hydrogen consumption increment under real line start-stop conditions, the inability to quantify the ineffective hydrogen loss caused by transient load changes in the fuel cell stack, and the lack of real vehicle big data support for power constraint threshold settings

[0005] The power buffering strategy is uniform across the board, making it impossible to balance energy saving and power performance. Existing technologies have a fixed power change rate limit throughout the entire process. If the slope is tightened to reduce hydrogen consumption, the power response during normal driving will deteriorate. If the slope is relaxed to ensure power, the power fluctuates drastically during the start-stop phase and hydrogen consumption increases significantly, making it impossible to achieve differentiated dynamic adjustment based on different scenarios.

[0006] The vehicle-to-cloud closed-loop self-iterative optimization architecture has a fixed control strategy. Traditional controller parameters are only calibrated once at the factory. The massive start-stop and hydrogen consumption data generated by long-term vehicle operation cannot be fed back to the iterative model. It cannot adapt to the changes in operating conditions caused by different city routes, seasonal passenger flow, and fuel cell stack aging. The hydrogen consumption optimization effect continues to decline in the long term.

[0007] Some existing patents only disclose single-point technologies such as "limiting the rate of change of stack power", "power battery buffering load" and "simple working condition identification", but none of them form a complete technical link of big data collection and governance - intelligent classification of start-stop working conditions - energy loss modeling - working condition adaptive control - cloud closed-loop iteration. There is no complete solution for optimizing hydrogen consumption under low-frequency fluctuations in bus start-stop conditions. The measured hydrogen consumption per 100 kilometers under start-stop conditions is 8% to 16% higher, which significantly increases the operating costs of bus companies and restricts the large-scale promotion of fuel cell buses.

[0008] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0009] The purpose of this invention is to provide an optimization system for low-frequency fluctuation energy consumption of fuel cells, which can solve the industry pain points of high hydrogen consumption, rigid strategies, and unbalanced power and energy consumption in bus start-stop scenarios using traditional fixed control strategies, and achieve adaptive hydrogen consumption reduction across the entire route and the entire life cycle.

[0010] Another objective of this invention is to provide a method for optimizing the low-frequency fluctuation energy consumption of fuel cells.

[0011] To achieve the above objectives, this invention provides an optimization system for low-frequency fluctuation energy consumption of fuel cells, comprising the following steps: an on-board data acquisition terminal for collecting multi-source time-series data on vehicle speed, braking, fuel cell stack, battery, hydrogen flow, and location; a cloud-based big data platform for storing full operational data, a working condition label library, and a start-stop energy consumption loss mapping model; a VCU-built-in working condition identification algorithm unit for incorporating clustering algorithms and identifying and outputting start-stop congestion working condition levels in real time; a start-stop energy consumption feature database for storing energy consumption samples and constraint thresholds for different routes and congestion levels; a VCU, which is the core control unit and has multiple sets of working condition-specific power allocation logic; a fuel cell stack for stable output of base power and reduction of transient load changes; a power battery system for buffering transient power during start-stop operations; a vehicle drive motor for executing power output; and an on-board hydrogen flow acquisition unit for collecting instantaneous hydrogen consumption data and providing closed-loop iterative feedback signals.

[0012] To achieve the aforementioned objective, this invention also provides an optimization method for low-frequency fluctuation energy consumption of fuel cells, which is used in the optimization system described above. The optimization method includes: millisecond-level synchronous acquisition and big data cleaning and processing of multi-source operating data to build a dedicated database for start-stop conditions; automatic labeling and hierarchical identification of bus start-stop conditions based on clustering algorithms; fitting a start-stop energy consumption loss mapping model with big data to quantify the additional hydrogen consumption due to power fluctuations; adaptive stack control logic optimization for start-stop conditions to buffer transient power fluctuations in the power battery; real-time transmission of vehicle hydrogen consumption data and iterative updating of the energy consumption mapping model and control parameters in the cloud; and execution of vehicle power distribution command output.

[0013] In a preferred embodiment, the steps of synchronously collecting multi-source operating data at millisecond levels and cleaning and processing big data to build a dedicated database for start-stop conditions include: the onboard data acquisition terminal synchronously collects vehicle speed, braking signal, accelerator pedal opening, fuel cell output power, power battery SOC, hydrogen flow, GPS positioning, and time-series data of single start-stop intervals, and achieves millisecond-level synchronous sampling and storage of all signals; big data algorithms are used to identify dirty data such as sensor drift and signal jumps, automatically remove abnormal cycles, and retain valid public transport operation data; and the cleaned and standardized time-series data is uploaded to the cloud big data platform and stored in the start-stop energy consumption characteristic database according to route, congestion level, and vehicle type.

[0014] In a preferred embodiment, the steps for automatic labeling and hierarchical identification of bus start-stop conditions based on clustering algorithms include: using speed range, number of consecutive start-stops, idling duration, and braking frequency as four-dimensional features, and employing the K-means clustering algorithm to automatically label all operational data; setting operating condition judgment rules, classifying ≥3 consecutive start-stops and speeds of 0~20km / h for more than 30 seconds as start-stop congestion conditions; further subdividing operating condition labels into three categories—mild congestion start-stop, moderate congestion start-stop, and severe continuous start-stop—based on two core dimensions: number of consecutive start-stops and low-speed duration; and the VCU's built-in operating condition identification algorithm unit matching the current operating data range in real time, outputting the corresponding operating condition level identifier, and transmitting it to the VCU's internal energy management strategy module.

[0015] In a preferred embodiment, the quantitative boundaries of each level of the operating condition determination rule include: mild congestion start-stop condition: 3 to 5 consecutive start-stops, and a low speed duration of 0 to 20 km / h of 30 to 90 seconds; moderate congestion start-stop condition: 6 to 10 consecutive start-stops, and a low speed duration of 0 to 20 km / h of 91 to 180 seconds; and severe congestion start-stop condition: ≥11 consecutive start-stops, and a low speed duration of 0 to 20 km / h of >180 seconds.

[0016] In a preferred embodiment, the steps of fitting a start-stop energy consumption loss mapping model using big data to quantify the additional hydrogen consumption due to power fluctuations include: retrieving massive historical data of the same route and vehicle model from the start-stop energy consumption feature database, and statistically analyzing the instantaneous additional hydrogen consumption corresponding to different power fluctuation amplitudes; fitting a three-dimensional mapping relationship between start-stop frequency, power fluctuation range, and additional hydrogen consumption to construct a start-stop energy consumption loss mapping model, and calibrating the optimal power fluctuation constraint thresholds corresponding to light / medium / heavy start-stop operations; and sending the model and the grading thresholds to the local storage of the vehicle controller.

[0017] In a preferred embodiment, the steps for optimizing the adaptive stack control logic under start-stop conditions and buffering transient power fluctuations in the power battery include: after receiving the start-stop condition identifier, the VCU automatically switches to a dedicated energy-saving mode for start-stop, lowers the upper limit of the stack power rise / fall slope, and strictly limits the instantaneous power fluctuation range; all transient power demands during vehicle start-up and braking are buffered by the charging and discharging of the power battery, and the fuel cell maintains a stable and constant output, avoiding frequent start-stop and large load changes of the stack; a dedicated SOC working range for start-stop conditions is configured to prevent secondary energy consumption caused by overcharging and over-discharging of the battery, and continuously ensure the battery's transient buffering capacity; and when identifying non-start-stop conditions such as constant speed and climbing, the normal power response parameters are automatically restored to ensure that the vehicle's acceleration and climbing performance are not limited.

[0018] In a preferred embodiment, the steps of real-time transmission of vehicle hydrogen consumption data and iterative updating of the energy consumption mapping model and control parameters in the cloud include: the on-board hydrogen flow acquisition unit continuously collects instantaneous hydrogen consumption, comprehensive hydrogen consumption per 100 kilometers, and fuel cell stack efficiency data; real-time operation data is transmitted back to the cloud start-stop energy consumption characteristic database to update the start-stop energy consumption loss mapping model and correct the power constraint thresholds corresponding to each congestion level; and the VCU downloads the iteratively optimized control parameters according to a preset cycle, preferably with a 24-hour complete update cycle, and performs full power constraint parameter synchronization during the daily high-pressure idle period of the vehicle.

[0019] In a preferred embodiment, the steps of real-time transmission of vehicle hydrogen consumption data and iterative updating of energy consumption mapping model and control parameters in the cloud also include: simultaneously configuring an event-triggered update mechanism, pushing an update command to the VCU after the cloud energy loss mapping model completes the iterative update, and automatically pulling the latest graded control threshold when the VCU is powered on and started again; verifying the parameter version number to the cloud every 30 minutes during vehicle operation, and if there are unsynchronized update packages, completing the incremental update after the vehicle is turned off and shut down.

[0020] In a preferred embodiment, the steps of executing the vehicle power distribution command output include: the VCU issues commands for fuel cell output power and power battery charging and discharging power based on the current operating condition level and the parameters after iteration, performs low-frequency fluctuation energy consumption optimization under start-stop conditions in real time, and synchronously stores the operating data for subsequent model iteration.

[0021] Compared with existing technologies, the optimization system and method for low-frequency fluctuation energy consumption of fuel cells in this invention have the following beneficial effects: Targeted hydrogen consumption reduction across specific scenarios, with quantifiable optimization effects: Specifically addressing the pain point of frequent start-stop low-frequency fluctuations in urban public transportation, the system dynamically constrains stack power fluctuations based on big data modeling. Actual measurements show a 14.5% reduction in hydrogen consumption per 100 kilometers under start-stop conditions on the same route, significantly reducing the daily hydrogen procurement costs for public transportation companies. Compared to existing single power limiting schemes, energy saving is improved by more than 6%, demonstrating unexpected technological benefits. Data-driven adaptive optimization, overcoming the limitations of traditional fixed control: Abandoning static calibration based on bench experience, the system autonomously identifies operating conditions and updates control parameters based on massive amounts of real-vehicle operation data. It can adapt to the congestion characteristics of different cities and routes, with versatility and practicality far superior to existing general-purpose energy management strategies. Balancing energy saving and vehicle power performance: The system tightens stack power response only in start-stop congestion scenarios, restoring conventional control parameters under constant speed, climbing, and high-speed conditions, without sacrificing vehicle acceleration and climbing power, thus resolving the technical contradiction of existing technologies where "energy saving results in weak power, and good power results in high hydrogen consumption." Extending stack lifespan and reducing total lifecycle costs: Significantly reducing frequent stack start-ups and shutdowns, and large-scale power load changes, slowing down membrane electrode assembly (MEA) and catalyst degradation, extending stack lifespan, and simultaneously reducing both hydrogen consumption and maintenance / depreciation operating costs. Fully automated closed-loop control, reducing manual calibration workload: The vehicle autonomously completes data acquisition, operating condition identification, and model iteration during operation, eliminating the need for engineers to repeatedly calibrate control parameters for each circuit, significantly shortening the fuel cell vehicle matching and development cycle, demonstrating significant industrialization value. Complete patent barrier, differentiating it from existing single-point, fragmented technologies: Existing patents only disclose single technologies such as power buffering and simple operating condition identification. This invention integrates big data governance, operating condition classification, energy consumption modeling, operating condition-specific control, and cloud-based iteration into a complete closed-loop system. The four innovative points are mutually coupled and synergistic, with no prior complete technical solutions, demonstrating sufficient novelty and inventiveness. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the overall process of the optimization method according to an embodiment of the present invention;

[0023] Figure 2 This is a system architecture diagram of an adaptive adjustment system for big data operating condition identification and control strategy according to an embodiment of the present invention.

[0024] Figure 3 This is a logic block diagram of big data label classification for bus congestion start-stop conditions according to an embodiment of the present invention. Detailed Implementation

[0025] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings, but it should be understood that the scope of protection of the present invention is not limited to the specific embodiments.

[0026] Unless otherwise expressly stated, throughout the specification and claims, the term "comprising" or its variations such as "including" or "comprises" shall be understood to include the stated elements or components without excluding other elements or other components.

[0027] like Figures 1 to 3 As shown, an optimization system for low-frequency fluctuation energy consumption of fuel cells according to a preferred embodiment of the present invention includes the following steps: an on-board data acquisition terminal, used to collect multi-source time-series data on vehicle speed, braking, fuel cell stack, battery, hydrogen flow, and location; a cloud-based big data platform, used to store full operational data, a working condition label library, and a start-stop energy consumption loss mapping model; a VCU with a built-in working condition identification algorithm unit, used to carry a clustering algorithm and identify and output the start-stop congestion working condition level in real time; a start-stop energy consumption feature database, used to store energy consumption samples and constraint thresholds by route and congestion level; a VCU, which is the core control unit, with multiple sets of working condition power allocation logic built-in; a fuel cell stack, used to smoothly output basic power and reduce transient load changes; a power battery system, used to buffer transient power during start-stop; a vehicle drive motor, used to execute power output; and an on-board hydrogen flow acquisition unit, used to collect instantaneous hydrogen consumption data and provide closed-loop iterative feedback signals.

[0028] According to a preferred embodiment of the present invention, a method for optimizing low-frequency fluctuation energy consumption of fuel cells is used in the aforementioned optimization system. The optimization method includes: millisecond-level synchronous acquisition and big data cleaning and processing of multi-source operating data to build a dedicated database for start-stop operating conditions; automatic labeling and hierarchical identification of bus start-stop operating conditions based on clustering algorithms; fitting a start-stop energy consumption loss mapping model with big data to quantify the additional hydrogen consumption due to power fluctuations; adaptive stack control logic optimization for start-stop operating conditions to buffer transient power fluctuations in the power battery; real-time transmission of vehicle hydrogen consumption data and iterative updating of the energy consumption mapping model and control parameters in the cloud; and execution of vehicle power allocation commands. The technical benefits include: solving the problems of asynchronous multi-source data and interference from invalid data; providing a high-precision standardized data foundation for operating condition identification and energy consumption modeling; and eliminating control strategy deviations caused by data errors.

[0029] In some implementations, the steps of multi-source operational data synchronous acquisition at millisecond levels and big data cleaning and processing to build a dedicated database for start-stop conditions include: the onboard data acquisition terminal synchronously collects vehicle speed, braking signals, accelerator pedal opening, fuel cell output power, power battery SOC, hydrogen flow, GPS positioning, and time-series data of single start-stop intervals, achieving millisecond-level synchronous sampling and storage of all signals; big data algorithms are used to identify dirty data such as sensor drift and signal jumps, automatically eliminating abnormal cycles and retaining valid public transport operation data; and the cleaned and standardized time-series data is uploaded to a cloud-based big data platform and stored in a start-stop energy consumption characteristic database categorized by route, congestion level, and vehicle type. The technical role of this approach is to automatically distinguish start-stop scenarios of different energy consumption levels, replacing manual calibration of operating condition thresholds and providing accurate judgment criteria for differentiated energy-saving control.

[0030] In some implementations, the steps for automatic labeling and hierarchical identification of bus start-stop conditions based on clustering algorithms include: using speed range, number of consecutive start-stops, idling duration, and braking frequency as four-dimensional features, and employing the K-means clustering algorithm to automatically label all operational data; setting operating condition judgment rules, classifying ≥3 consecutive start-stops and speeds of 0~20km / h for more than 30 seconds as start-stop congestion conditions; further subdividing operating condition labels into three categories—mild congestion start-stop, moderate congestion start-stop, and severe continuous start-stop—based on two core dimensions: number of consecutive start-stops and low-speed duration; and the VCU's built-in operating condition identification algorithm unit matching the current operating data range in real time, outputting the corresponding operating condition level identifier, and transmitting it to the VCU's internal energy management strategy module.

[0031] In some implementations, the quantitative boundaries for each level of the operating condition judgment rules include: Mild congestion start-stop condition: 3-5 consecutive start-stop cycles, 30-90 seconds of low speed at 0-20 km / h; Moderate congestion start-stop condition: 6-10 consecutive start-stop cycles, 91-180 seconds of low speed at 0-20 km / h; and Severe congestion start-stop condition: ≥11 consecutive start-stop cycles, >180 seconds of low speed at 0-20 km / h. The technical function is to automatically distinguish start-stop scenarios of different energy consumption levels, replacing manual calibration of operating condition thresholds and providing accurate judgment criteria for differentiated energy-saving control.

[0032] In some implementations, the steps of fitting a start-stop energy consumption loss mapping model using big data to quantify the additional hydrogen consumption due to power fluctuations include: retrieving massive historical data from the same route and vehicle model within the start-stop energy consumption characteristic database, and statistically analyzing the instantaneous additional hydrogen consumption corresponding to different power fluctuation amplitudes; fitting a three-dimensional mapping relationship between start-stop frequency, power fluctuation range, and additional hydrogen consumption to construct a start-stop energy consumption loss mapping model, and calibrating the optimal power fluctuation constraint thresholds for light / medium / heavy start-stop operations; and distributing the model and tiered thresholds to the local storage of the vehicle controller. The technical role of this approach is to abandon traditional bench calibration, quantify hydrogen loss caused by low-frequency fluctuations using real bus operation data, and achieve data-driven control parameters.

[0033] In some implementations, the adaptive stack control logic optimization for start-stop conditions and the steps for the power battery to buffer transient power fluctuations include: after receiving the start-stop condition identifier, the VCU automatically switches to a dedicated start-stop energy-saving mode, lowers the upper limit of the stack power rise / fall slope, and strictly limits the instantaneous power fluctuation amplitude; all transient power demands during vehicle start-up and braking are buffered by the power battery charging and discharging, and the fuel cell maintains a stable and constant output, avoiding frequent start-stop operations and large load changes of the stack; a dedicated SOC working range for start-stop conditions is configured to prevent secondary energy consumption caused by overcharging and over-discharging of the battery, continuously ensuring the battery's transient buffering capacity; and when identifying non-start-stop conditions such as constant speed and climbing, the normal power response parameters are automatically restored to ensure that the vehicle's acceleration and climbing performance are not restricted. The technical role of this is to dynamically balance hydrogen consumption and power performance according to different scenarios, suppressing ineffective hydrogen consumption caused by low-frequency power fluctuations during start-stop operations from the source.

[0034] In some implementations, the steps of real-time transmission of vehicle hydrogen consumption data and iterative updates of the energy consumption mapping model and control parameters in the cloud include: the on-board hydrogen flow acquisition unit continuously collecting instantaneous hydrogen consumption, comprehensive hydrogen consumption per 100 kilometers, and fuel cell stack efficiency data; real-time operation data being transmitted back to the cloud-based start-stop energy consumption characteristic database to update the start-stop energy consumption loss mapping model and correct the power constraint thresholds corresponding to each congestion level; and the VCU downloading the iteratively optimized control parameters according to a preset cycle, preferably with a 24-hour complete update cycle, and performing full power constraint parameter synchronization during the daily high-pressure idle period of the vehicle.

[0035] In some implementations, the real-time transmission of vehicle hydrogen consumption data and the iterative updating of the energy consumption mapping model and control parameters in the cloud also include: simultaneously configuring an event-triggered update mechanism; after the cloud-based energy loss mapping model completes its iterative update, it pushes an update command to the VCU; the VCU automatically retrieves the latest graded control thresholds upon the next power-on start-up; and verifying the parameter version number to the cloud every 30 minutes during vehicle operation. If there are any unsynchronized update packages, incremental updates are completed after the vehicle is shut down. The technical advantages include: employing a three-layer cycle download mechanism of "daily scheduled full update + power-on event-triggered update + driving version verification," which avoids the large amount of data transmission during driving occupying CAN bus bandwidth and interfering with real-time vehicle control, while ensuring that the start-stop constraint parameters after cloud iteration are quickly distributed to adapt to changes in route congestion; the 24-hour update cycle matches the daily bus maintenance operation habits, balancing communication load and model iteration timeliness; and technical personnel in the relevant field can flexibly adjust the cycle length according to fleet size and route change frequency.

[0036] In some implementations, the steps for executing the vehicle power distribution command output include: the VCU issuing commands for fuel cell output power and power battery charging and discharging power based on the current operating condition level and the parameters after iteration, completing the low-frequency fluctuation energy consumption optimization under start-stop conditions in real time, and synchronously storing the operating data for subsequent model iterations.

[0037] In some embodiments, the fuel cell low-frequency fluctuation energy consumption optimization system and method of the present invention rely on massive multi-source big data from public transportation operations to construct a complete hydrogen consumption optimization closed-loop system for low-frequency fluctuation scenarios during start-stop operations. The core technical features include: ① a dedicated big data governance and automatic hierarchical labeling system for bus start-stop operations; ② a start-stop power fluctuation-additional hydrogen consumption mapping model based on real vehicle data; ③ a collaborative control logic that dynamically switches stack power constraints and power battery transient buffering under different operating conditions; and ④ an on-board-cloud data interconnection and continuous model iteration self-learning architecture. The overall solution addresses the industry pain points of traditional fixed control strategies in bus start-stop scenarios, such as high hydrogen consumption, rigid strategies, and an inability to balance power and energy consumption, achieving adaptive hydrogen consumption reduction across the entire route and the entire lifecycle. The following are simulation comparison test data:

[0038] Test conditions: The same 12-meter fuel cell bus, the same congested bus route in the same city, and the same hydrogen supply and thermal management system were used. Traditional fixed control strategies and the optimized scheme of this invention were run separately, and 1000 sets of start-stop cycle data were continuously collected.

[0039] Power fluctuation index: Traditional strategy stack power fluctuation is ±28kW; this solution controls the fluctuation within ±8kW, reducing the fluctuation amplitude by 71.4%;

[0040] Instantaneous hydrogen consumption indicators: The peak hydrogen consumption during start-stop of the traditional strategy is 0.92 g / s; the peak hydrogen consumption of this scheme is 0.76 g / s, representing a 17.4% reduction in instantaneous peak hydrogen consumption.

[0041] Overall hydrogen consumption per 100 kilometers: Traditional strategy 10.8 kg / 100km; This solution 9.23 kg / 100km, reducing overall hydrogen consumption under start-stop conditions by 14.5%;

[0042] Stack degradation index: Under the same operating mileage, the frequency of stack power change is reduced by 62%, which effectively slows down the aging of membrane electrode and catalyst, and reduces the maintenance cost throughout the entire life cycle.

[0043] In summary, the optimization system and method for low-frequency fluctuation energy consumption of fuel cells of the present invention have the following advantages: Targeted hydrogen consumption reduction across specific scenarios, with quantifiable optimization effects: Specifically addressing the pain point of frequent start-stop low-frequency fluctuations in urban public transportation, it dynamically constrains stack power fluctuations based on big data modeling. Actual measurements show a 14.5% reduction in hydrogen consumption per 100 kilometers under start-stop conditions on the same route, significantly reducing the daily hydrogen procurement costs for public transportation companies. Compared to existing single power limiting schemes, energy saving is improved by more than 6%, demonstrating unexpected technical effects. Data-driven adaptive optimization, overcoming the limitations of traditional fixed control: Abandoning static calibration based on bench experience, it autonomously identifies operating conditions and updates control parameters based on massive amounts of real-vehicle operation data. It can adapt to the congestion characteristics of different cities and routes, with versatility and practicality far superior to existing general-purpose energy management strategies. Balancing energy saving and vehicle power performance: Only tightening the stack power response in start-stop congestion scenarios, restoring conventional control parameters under constant speed, climbing, and high-speed conditions, without sacrificing vehicle acceleration and climbing power, resolving the technical contradiction of existing technologies where "energy saving results in weak power, and good power results in high hydrogen consumption." Extending stack lifespan and reducing total lifecycle costs: Significantly reducing frequent stack start-ups and shutdowns, and large-scale power load changes, slowing down membrane electrode assembly (MEA) and catalyst degradation, extending stack lifespan, and simultaneously reducing both hydrogen consumption and maintenance / depreciation operating costs. Fully automated closed-loop control, reducing manual calibration workload: The vehicle autonomously completes data acquisition, operating condition identification, and model iteration during operation, eliminating the need for engineers to repeatedly calibrate control parameters for each circuit, significantly shortening the fuel cell vehicle matching and development cycle, demonstrating significant industrialization value. Complete patent barrier, differentiating it from existing single-point, fragmented technologies: Existing patents only disclose single technologies such as power buffering and simple operating condition identification. This invention integrates big data governance, operating condition classification, energy consumption modeling, operating condition-specific control, and cloud-based iteration into a complete closed-loop system. The four innovative points are mutually coupled and synergistic, with no prior complete technical solutions, demonstrating sufficient novelty and inventiveness.

[0044] The foregoing description of specific exemplary embodiments of the invention is for illustrative and explanatory purposes. These descriptions are not intended to limit the invention to the precise forms disclosed, and it will be apparent that many changes and variations can be made in accordance with the foregoing teachings. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application, thereby enabling those skilled in the art to implement and utilize various different exemplary embodiments of the invention, as well as various different choices and variations. The scope of the invention is intended to be defined by the claims and their equivalents.

Claims

1. An optimization system for low-frequency fluctuation energy consumption of a fuel cell, characterized in that, Includes the following steps: The vehicle-mounted data acquisition terminal is used to collect multi-source time-series data on vehicle speed, braking, fuel cell stack, battery, hydrogen flow, and location. The cloud-based big data platform is used to store all operational data, operating condition tag library, and start-stop energy loss mapping model; The VCU has a built-in operating condition recognition algorithm unit, which is used to carry out clustering algorithms and identify and output the start-stop congestion level in real time. Start-stop energy consumption characteristic database, which is used to store energy consumption samples and constraint thresholds by line and by congestion level; VCU is the core control unit, which has multiple sets of power distribution logic for different operating conditions. Fuel cell stacks are used to smoothly output base power and reduce transient load changes; The power battery system is used to buffer transient power during start-stop operations; The vehicle's drive motor, which is used to perform power output; and The on-board hydrogen flow acquisition unit is used to collect instantaneous hydrogen consumption data and provide closed-loop iterative feedback signals.

2. A method for optimizing low-frequency fluctuation energy consumption of a fuel cell, used in the optimization system as described in claim 1, characterized in that, The optimization method includes: Millisecond-level synchronous collection and big data cleaning and management of multi-source operational data; establishing a dedicated database for start-up and shutdown conditions. Automatic labeling and hierarchical identification of bus start-stop conditions based on clustering algorithm; Big data fitting model of start-stop energy consumption loss mapping to quantify additional hydrogen consumption due to power fluctuations; Optimize the adaptive stack control logic under start-stop conditions to buffer transient power fluctuations in the power battery; Real-time transmission of vehicle hydrogen consumption data, with cloud-based iterative updates to the energy consumption mapping model and control parameters; and The vehicle power distribution command is executed.

3. The optimization method as described in claim 2, characterized in that, The steps for millisecond-level synchronous acquisition and big data cleaning and processing of multi-source operational data to build a dedicated database for start-up and shutdown conditions include: The vehicle-mounted data acquisition terminal synchronously collects vehicle speed, braking signal, accelerator pedal opening, fuel cell output power, power battery SOC, hydrogen flow, GPS positioning, and time sequence data of single start-stop intervals, and achieves millisecond-level synchronous sampling and storage of all signals. Big data algorithms are used to identify dirty data such as sensor drift and signal jumps, automatically removing abnormal loops and retaining valid bus operation data; and After cleaning, standardized time-series data is uploaded to the cloud big data platform and stored in the start-stop energy consumption characteristic database according to route, congestion level and vehicle type.

4. The optimization method as described in claim 2, characterized in that, The steps for automatic labeling and hierarchical identification of bus start-stop conditions based on clustering algorithms include: Using speed range, number of consecutive starts and stops, idling duration, and braking frequency as four-dimensional features, the K-means clustering algorithm is used to automatically label all operational data. The system establishes rules for determining traffic conditions, classifying a situation as start-stop congestion involving at least three consecutive starts and stops, or a vehicle speed of 0-20 km / h for more than 30 seconds. Based on two core dimensions—the number of consecutive starts and stops and the duration of low-speed driving—it further categorizes this into three types of conditions: mild congestion start-stop, moderate congestion start-stop, and severe continuous start-stop. The VCU's built-in operating condition identification algorithm unit matches the current operating data range in real time and outputs the corresponding operating condition level identifier, which is then transmitted to the VCU's internal energy management strategy module.

5. The optimization method as described in claim 4, characterized in that, The quantization boundaries of each level of the defined operating condition determination rule include: Light congestion start-stop condition: 3 to 5 consecutive start-stop cycles, 0 to 20 km / h low speed duration 30 to 90 seconds; Moderate congestion start-stop conditions: 6-10 consecutive start-stop cycles, 91-180 seconds of low speed (0-20 km / h); and Severe traffic jam start-stop conditions: ≥11 consecutive start-stop cycles, and continuous low speed duration of 0-20km / h >180s.

6. The optimization method as described in claim 2, characterized in that, The steps for quantifying the additional hydrogen consumption due to power fluctuations in the big data fitting start-stop energy loss mapping model include: Retrieve massive amounts of historical data from the same route and vehicle model within the start-stop energy consumption characteristic database, and statistically analyze the instantaneous additional hydrogen consumption corresponding to different power fluctuation amplitudes; A three-dimensional mapping relationship between start-stop frequency, power fluctuation range, and additional hydrogen consumption was fitted to construct a start-stop energy loss mapping model, and the optimal power fluctuation constraint thresholds for light, medium, and heavy start-stop operations were calibrated; and... The model and grading thresholds are sent to the local storage of the vehicle controller.

7. The optimization method as described in claim 2, characterized in that, The steps for optimizing the adaptive stack control logic under start-stop conditions and buffering transient power fluctuations in the power battery include: After receiving the start-stop condition flag, the VCU automatically switches to the start-stop dedicated energy-saving mode, lowers the upper limit of the stack power rise and fall slope, and strictly limits the instantaneous power change amplitude. The power demand during vehicle start-up and braking is buffered by the charging and discharging of the power battery, while the fuel cell maintains a stable and constant output, avoiding frequent start-stop and large load changes of the fuel cell stack. Configure a dedicated SOC operating range for start-stop conditions to prevent secondary energy consumption caused by battery overcharging and over-discharging, and continuously ensure the battery's transient buffering capability; and When recognizing non-start-stop operating conditions such as constant speed and hill climbing, it automatically restores the normal power response parameters to ensure that the vehicle's power acceleration and hill climbing performance are not limited.

8. The optimization method as described in claim 2, characterized in that, The steps for real-time transmission of vehicle hydrogen consumption data and iterative updating of the energy consumption mapping model and control parameters in the cloud include: The onboard hydrogen flow acquisition unit continuously collects data on instantaneous hydrogen consumption, comprehensive hydrogen consumption per 100 kilometers, and fuel cell stack operating efficiency. Real-time operational data is transmitted back to the cloud-based start-stop energy consumption characteristic database to update the start-stop energy loss mapping model and correct the power constraint thresholds corresponding to each congestion level; and The VCU downloads the iteratively optimized control parameters according to a preset cycle, preferably with a 24-hour complete update cycle, and performs full power constraint parameter synchronization during the daily high-pressure stationary period of the vehicle.

9. The optimization method as described in claim 8, characterized in that, The steps of real-time transmission of vehicle hydrogen consumption data and iterative updating of the energy consumption mapping model and control parameters in the cloud also include: Simultaneously, an event-triggered update mechanism is configured. After the cloud-based energy loss mapping model completes its iterative update, an update command is pushed to the VCU. The VCU automatically retrieves the latest graded control thresholds upon the next power-on startup. During vehicle operation, the parameter version number is checked to the cloud every 30 minutes. If there are any unsynchronized update packages, incremental updates are completed after the vehicle is turned off and shut down.

10. The optimization method as described in claim 2, characterized in that, The steps for executing the vehicle power distribution command output include: Based on the current operating condition level and the parameters after iteration, the VCU issues commands for fuel cell output power and power battery charging and discharging power, and performs real-time optimization of low-frequency fluctuation energy consumption during start-stop conditions, while simultaneously storing operating data for subsequent model iterations.

Images