Fuel cell automobile energy management method and system

Abstract

The invention discloses an energy management method and system for a fuel cell vehicle. The method comprises the steps of obtaining vehicle driving parameters and power battery parameters; identifying a current driving condition according to the vehicle driving parameters; evaluating the working intensity of the power battery based on the power battery parameters to obtain a working intensity index of the power battery; determining target output power of the fuel cell by adopting a fuzzy control rule according to the driving working condition, the SOC of the power cell, the working intensity index and the required power of the whole vehicle; and managing the energy of the fuel cell vehicle based on the target output power of the fuel cell. The method has the advantages of dynamically adapting to complex working conditions, optimizing the energy distribution efficiency and prolonging the service life of the fuel cell and the power cell.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of hydrogen fuel cell energy management control, and particularly relates to a fuel cell vehicle energy management method and system. BACKGROUND

[0002] There are many technical bottlenecks in the current fuel cell vehicle energy management strategy. The rule-based control strategy mainly relies on preset static threshold parameters, such as controlling the start and stop and power output of the fuel cell only according to the battery SOC threshold. This fixed mode is difficult to adapt to complex and variable actual road conditions. Although the instantaneous optimization strategy such as the equivalent fuel consumption minimum strategy (ECMS) has a certain optimization ability, the equivalent factor is usually fixed, and the optimization effect is significantly reduced under the dynamic driving condition. Although the predictive control strategy has better optimization potential in theory, most schemes need to rely on global path information, resulting in too large calculation amount and being difficult to meet the real-time control demand. More importantly, the existing strategies generally have the problem of single optimization target, excessively focusing on the instantaneous hydrogen consumption index, ignoring the serious influence of frequent start and stop of the fuel cell and load mutation on the system life, and failing to fully consider the influence of the working state of the power battery on the overall performance of the system, lacking the multi-target collaborative optimization ability. These technical defects specifically manifest as: the rule-based control strategy has obvious static characteristics and hysteresis, and cannot be dynamically adjusted according to the real-time traffic condition (such as congestion, high speed, climbing, etc.); at the same time, the existing schemes fail to dynamically integrate the fuel cell life attenuation characteristics and the power battery health state into the real-time control decision, resulting in the decline of the long-term operation economy of the system.

[0003] In view of the above problems, the existing technology needs to be improved. SUMMARY

[0004] The purpose of the present application is to solve the problems in the background art, and to provide a fuel cell vehicle energy management method and system, which has the advantages of dynamically adapting to complex conditions, optimizing energy distribution efficiency, and prolonging the service life of the fuel cell and the power battery.

[0005] The technical scheme adopted by the present application is: a fuel cell vehicle energy management method, obtaining vehicle driving parameters and power battery parameters; identifying the current driving condition according to the vehicle driving parameters; evaluating the working strength of the power battery based on the power battery parameters to obtain a working strength index of the power battery; determining the target output power of the fuel cell according to the driving condition, the power battery SOC, the working strength index, and the demand power of the whole vehicle by using a fuzzy control rule; managing the energy of the fuel cell vehicle based on the target output power of the fuel cell.

[0006] Further, after obtaining the vehicle driving parameters and the power battery parameters, the obtained parameters are filtered and validity checked. The validity check refers to verifying whether the signal value at the corresponding time is within the set range. If not, the average value of the signal values at the previous time and the next time of the corresponding time is calculated, and the average value is used to replace the signal value at the corresponding time. If the signal values at adjacent time points are invalid, the signal value at the previous valid time is used to replace.

[0007] Further, the current driving condition is identified according to the vehicle driving parameters, comprising: A random forest model is trained by historical vehicle driving parameters; Based on the vehicle driving parameters, a plurality of feature parameters in a set time window are extracted, including average speed, speed standard deviation, average absolute acceleration, idle time ratio, ratio of maximum demand power to average demand power; The plurality of feature parameters are input into the random forest model, and the current driving condition is obtained through a voting mechanism. The driving condition is any one of congestion, suburb, highway, and long uphill.

[0008] Further, the working intensity index of the power battery is determined by the following formula: ; Wherein, HI is the working intensity index of the power battery; Q batt is the cumulative throughput of the power battery; Q max is the maximum reference throughput; SOC avg is the average SOC value in a set time period between the current time; T avg is the average battery temperature in a set time period between the current time; T ideal is the optimal working temperature of the battery; T range is the difference between the upper limit value and the lower limit value of the battery working temperature interval; α, β, γ are the first, second and third weight coefficients, respectively; Q batt (k) is the cumulative throughput of the power battery at the kth time; Q batt (k-1) is the cumulative throughput of the power battery at the k-1th time; I batt (k) is the charging current or discharging of the power battery at the kth time; Δt is the sampling period.

[0009] Further, the target output power of the fuel cell is determined according to the driving condition, the power battery SOC, the working intensity index, and the demand power of the whole vehicle by using fuzzy control rules, comprising: The power battery SOC, the working intensity index, and the demand power of the whole vehicle are respectively converted into corresponding fuzzy language variables and the corresponding membership values by corresponding membership functions. Based on driving conditions, power battery SOC, workload index, vehicle power demand, and fuzzy linguistic variables and membership values, multiple target output power fuzzy variables and target output power membership values ​​of fuel cells are determined by searching the fuzzy rule base. For each fuzzy rule, the target output power fuzzy variable and the target output power membership value are obtained. Based on the target output power membership value, the fuzzy interval corresponding to the target output power fuzzy variable is clipped to obtain the clipped fuzzy region. All fuzzy regions are superimposed to obtain a fuzzy set on the output universe of discourse, and the membership value of each power point on the output universe of discourse is determined based on the fuzzy set; The target output power of the fuel cell is determined based on the membership values ​​of all power points in the output universe.

[0010] Further, the step of pruning the fuzzy interval corresponding to the fuzzy variable of the target output power based on the membership value of the target output power includes: All membership values ​​in the fuzzy range of the target output power that are greater than the target output power membership value are changed to the target output power membership value.

[0011] Furthermore, the determination of the membership value of each power point in the output universe based on fuzzy sets includes: For each power point in the output universe of discourse, find all membership values ​​of the fuzzy set at that power point, and take the largest membership value as the membership value of that power point.

[0012] Furthermore, the target output power of the fuel cell is determined using the following formula: ;

[0013] Among them, P fc_cmd P represents the target output power of the fuel cell. i μ(P) represents the i-th discrete point on the output universe of discourse. i ) is P i The corresponding membership value; n is the total number of discrete points in the output universe of discourse.

[0014] Furthermore, the energy management of the fuel cell vehicle based on the target output power of the fuel cell includes: The target output power is controlled by the fuel cell system, and the remaining required power is provided or absorbed by the power battery. The remaining required power = vehicle required power - fuel cell target power.

[0015] A fuel cell vehicle energy management system includes: The parameter acquisition module is used to acquire vehicle driving parameters and power battery parameters; The driving condition recognition module is used to identify the current driving condition based on vehicle driving parameters; The power battery performance evaluation module is used to evaluate the performance of the power battery based on its parameters and obtain the power battery performance index. The output power determination module is used to determine the target output power of the fuel cell based on driving conditions, power battery SOC, workload index, and vehicle power demand using fuzzy control rules. The energy management module is used to manage the energy of fuel cell vehicles based on the target output power of the fuel cell.

[0016] The beneficial effects of this invention are as follows: This invention achieves adaptive optimization control for complex operating conditions by dynamically identifying driving conditions, assessing the working intensity of the power battery in real time, and dynamically adjusting the output power of the fuel cell based on fuzzy control rules. This results in advantages such as improved energy utilization efficiency, extended service life of key components, enhanced system operating economy, and ensuring that the vehicle's power requirements are met in a timely manner. Attached Figure Description

[0017] Figure 1 This is a flowchart of the energy management process of the present invention.

[0018] Figure 2 This is a flowchart illustrating how the present invention uses fuzzy control rules to determine the target output power. Detailed Implementation

[0019] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings. It should be noted that these descriptions are for the purpose of aiding understanding the present invention, but do not constitute a limitation thereof. Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0020] In existing technologies, energy management strategies for fuel cell vehicles generally suffer from static lag, making them unable to adapt to complex, real-time traffic conditions. Rule-based control methods rely on preset fixed thresholds, which are ill-suited for scenarios such as sudden acceleration or prolonged idling. While instantaneous optimization strategies can achieve local optima, their fixed equivalence factors result in poor adaptability to various operating conditions. Predictive strategies are limited by the difficulty of obtaining global path information, thus restricting their practical application. Existing methods primarily focus on instantaneous energy consumption optimization, failing to incorporate fuel cell lifespan degradation and battery health into the dynamic control system, leading to increased long-term operating costs.

[0021] To address the aforementioned issues, researchers noted that traditional strategies, with their single-objective focus, struggle to balance multi-dimensional performance indicators. Analysis revealed that dynamic operating condition identification and multi-parameter collaborative decision-making are key to overcoming existing technological bottlenecks. The initial approach involved establishing a real-time data acquisition system to capture vehicle operating status and battery parameter changes. Next, they explored how to quantify the power battery's workload, avoiding evaluation biases caused by relying solely on SOC parameters. Further research investigated the advantages of fuzzy control rules in complex nonlinear systems, attempting to fuse driving conditions, battery status, and power demand into multi-variable decisions. Ultimately, this resulted in a dynamic power allocation mechanism based on adaptive operating conditions, achieving collaborative optimization between the fuel cell and the power battery.

[0022] Based on the above ideas, this invention proposes an energy management method for fuel cell vehicles, such as... Figure 1 , Figure 2 As shown, it includes the following steps: Obtain vehicle driving parameters and power battery parameters; Identify the current driving condition based on vehicle driving parameters; The working intensity index of the power battery is obtained by evaluating the working intensity of the power battery based on the parameters of the power battery. The target output power of the fuel cell is determined by fuzzy control rules based on driving conditions, power battery SOC, workload index, and vehicle power demand (P_demand). Energy management for fuel cell vehicles is based on the target output power of the fuel cell.

[0023] Among these, vehicle driving parameters refer to physical quantities reflecting the real-time operating status of a vehicle, specifically using data such as vehicle speed, acceleration, and gradient, to construct the foundational dataset for dynamic operating condition identification. Power battery parameters are key indicators characterizing the battery's operating state, specifically using voltage, current, temperature, and SOC parameters, providing multi-dimensional data support for battery load assessment. Driving condition identification refers to classifying vehicle operating modes, specifically using machine learning algorithms to model historical driving data, enabling automatic identification of scenarios such as urban roads, highways, and hill climbs. Work intensity indicators are comprehensive parameters quantifying the power battery's load state, specifically using a weighted calculation of throughput, temperature change rate, and SOC fluctuation, overcoming the limitations of single SOC parameter evaluation. Fuzzy control rules refer to a fuzzy inference system based on expert experience, specifically using multi-input variable membership functions and rule base matching, solving the problem of traditional threshold control's difficulty in handling nonlinear relationships.

[0024] Specifically, a dynamic monitoring network is constructed by continuously collecting driving parameters such as vehicle speed and acceleration through onboard sensors, combined with real-time data on SOC, temperature, voltage, and current obtained from the battery management system. Features such as average vehicle speed and acceleration standard deviation are extracted using a sliding time window and input into a pre-trained random forest model for operating condition classification. Simultaneously, the cumulative throughput and temperature deviation of the power battery within a set period are calculated to generate a comprehensive workload index. The SOC value, workload index, and vehicle power demand are used as input variables for a fuzzy controller, converted into fuzzy linguistic variables through a membership function. Inference operations are performed based on driving conditions, fuzzy linguistic variables, and a pre-set fuzzy rule base. The target power of the fuel cell is obtained by defuzzifying using the area centroid method. Finally, the fuel cell output is adjusted according to the target power, with the remaining power demand compensated by the charging and discharging of the power battery, achieving dynamic power allocation between the two power sources.

[0025] This invention achieves dynamic switching of control strategies through operating condition identification, effectively solving the response lag problem caused by static rules. By integrating the power battery's operating intensity index and SOC parameter, it achieves accurate assessment of battery load status, avoiding misjudgments caused by single parameters in traditional methods. It employs multivariate fuzzy control rules instead of fixed threshold decisions, improving the rationality of power allocation under complex operating conditions while considering both fuel cell lifespan and power battery health. Ultimately, while ensuring the vehicle's power requirements, it reduces lifespan loss caused by frequent fuel cell start-stop cycles, extending the power battery's lifespan. Based on real-time data acquisition and local feature extraction, this invention has stronger operating condition adaptability and real-time response capabilities.

[0026] The present invention further proposes that after obtaining vehicle driving parameters and power battery parameters, the obtained parameters are filtered and validated. Validity validation refers to verifying whether the signal value at the corresponding time is within the set range. If it is not within the set range, the average value of the signal value at the previous time and the next time is calculated and the average value is used to replace the signal value at the corresponding time.

[0027] The vehicle driving parameters and power battery parameters are signals cyclically acquired through the vehicle's CAN bus, including vehicle speed, accelerator pedal opening, total power battery current, total power battery voltage, power battery SOC, power battery temperature, and fuel cell output current and voltage. Filtering refers to eliminating high-frequency noise interference in the signal using digital filters, specifically first-order low-pass filters such as moving average filters or Butterworth filters, to smooth vehicle speed fluctuations and power battery current transients. Validity verification involves judging the signal's reasonableness based on physical constraints. This can be achieved by setting sensor ranges or equipment operating limits, for example, limiting the vehicle speed signal to the 0-200 km / h range. When an abnormal value exceeding this range is detected, a data replacement mechanism for adjacent time moments is triggered.

[0028] Specifically, filtering suppresses random noise introduced during signal acquisition using time-domain or frequency-domain algorithms, ensuring smooth input data for the subsequent operating condition identification module. Validity verification further identifies outliers caused by sensor malfunctions or transient interference through a dynamic threshold detection mechanism, using the average of valid values ​​from adjacent time points to replace outliers while preserving data temporal continuity. The combined effect of these two processes ensures that the vehicle driving parameters and power battery parameters input to the control strategy conform to physical laws and equipment operating characteristics, preventing misjudgments of operating conditions or inaccurate power distribution due to abnormal data.

[0029] This invention employs a dual data preprocessing approach combining filtering and validity verification. By eliminating high-frequency noise, it designs a time-series-based intelligent replacement algorithm for occasional outliers. This effectively solves the information loss problem caused by directly removing data in traditional methods, thereby improving the robustness of the control strategy.

[0030] Through the above technical solution, the present invention can significantly reduce the risk of parameter distortion caused by sensor noise or transient interference, and ensure that the input data received by the subsequent operating condition identification, power battery status assessment and power distribution modules have physical rationality and temporal consistency, thereby providing a reliable data foundation for the stable operation of the fuel cell vehicle energy management system.

[0031] This invention further proposes a method for identifying the current driving condition based on vehicle driving parameters, including training a random forest model using historical vehicle driving parameters; extracting multiple feature parameters within a set time window based on the vehicle driving parameters; inputting the multiple feature parameters into the random forest model; and obtaining the current driving condition through a voting mechanism.

[0032] The Random Forest model is a machine learning model ensembled from multiple decision trees, specifically trained using historical driving data, used to handle nonlinear relationships of multidimensional features. Average vehicle speed is the arithmetic mean of vehicle speeds within a set time window, calculated using the sliding window method, reflecting driving intensity. Speed ​​standard deviation refers to the dispersion of vehicle speed within a set time window, calculated using statistical methods, characterizing speed fluctuations. Mean absolute acceleration is the average absolute value of acceleration within a set time window, obtained through integration, used to quantify driving intensity. Idle time ratio refers to the percentage of time within a set time window where vehicle speed is zero, statistically analyzed using a timer, reflecting traffic congestion characteristics. The ratio of maximum demand power to average demand power refers to the relative intensity of peak power within a set time window, extracted using extreme value detection algorithms, characterizing instantaneous load changes.

[0033] Specifically, the random forest model trained on historical data can establish a mapping relationship between multi-dimensional features and driving conditions. During real-time operation, the system collects vehicle driving data at fixed time windows, simultaneously calculating five feature parameters: average vehicle speed, standard deviation of vehicle speed, average absolute acceleration, idling time ratio, and power ratio. These parameters construct a condition feature vector from five dimensions: driving intensity, fluctuation characteristics, operational intensity, traffic conditions, and load changes. After the feature vector is input into the random forest model, each decision tree independently makes a classification decision. Finally, the results from each tree are integrated through a majority voting mechanism to output the judgment result for the current driving condition. This process achieves real-time matching of complex driving conditions through dynamic feature extraction and ensemble learning mechanisms.

[0034] In some embodiments, the specific steps for identifying the current driving condition described above are as follows: a. Feature extraction: Set a sliding time window of length T (e.g., 60 seconds) and calculate the feature parameters within the window in real time: average vehicle speed, standard deviation of vehicle speed (characterizing vehicle speed fluctuation, this value is higher during congestion), average absolute acceleration (reflecting the intensity of driving), idling time ratio (an important feature for distinguishing congestion), and the ratio of maximum demand power to average demand power (used to identify high-load conditions such as rapid acceleration and hill climbing).

[0035] b. Operating condition identification and classification: Offline training: On a PC, a random forest model is trained using a large amount of historical driving data. During training, the depth and number of trees are controlled to ensure that the model is lightweight.

[0036] Model solidification: The trained model parameters (including splitting features, splitting thresholds, and leaf node outputs for each tree) are integrated into the VCU software in the form of lookup tables or hard-coded conditional statements.

[0037] Online inference: For each new set of features extracted, the VCU runs the solidified model: the feature values ​​are compared with the node conditions of each tree in turn, and finally the working condition identification result is obtained through the "voting" mechanism (e.g., 1-congestion, 2-suburbs, 3-highway, 4-long uphill).

[0038] This invention utilizes multi-dimensional feature fusion and machine learning models to automatically capture the dynamic characteristics differences under various operating conditions. For example, in congested conditions, the system identifies the problem through a combination of high idling time ratio and low average vehicle speed; in high-speed conditions, it determines the problem through a combination of low vehicle speed standard deviation and high average power demand. This dynamic feature combination mechanism overcomes the limitations of traditional static thresholds.

[0039] Through the above technical solution, this invention can accurately identify the driving condition type of a vehicle in real time, providing precise condition input for subsequent energy allocation strategies. The dynamic condition identification mechanism effectively solves the misjudgment problem caused by the single feature dimension of traditional methods, and at the same time improves the classification robustness in complex traffic environments through ensemble learning. The accurate condition judgment results enable the energy management system to dynamically adjust the power allocation ratio between the fuel cell and the power battery, thereby optimizing the overall vehicle energy utilization efficiency.

[0040] This invention further proposes a method for evaluating the working intensity of power batteries based on big data, including the following steps: a. Real-time throughput calculation: The cumulative throughput Q of the power battery since the start of the previous charging cycle (i.e., the vehicle startup time) is calculated using the ampere-hour integral method. batt Its discretization calculation formula is: ; Among them, Q batt (k) represents the cumulative throughput of the power battery at time k; Q batt (k-1) represents the cumulative throughput of the power battery at time k-1; I batt (k) represents the charging or discharging current of the power battery at time k; Δt represents the sampling period; the absolute value of the current indicates that both charging and discharging losses are included.

[0041] b. Generation of Intensity Index (HI): A comprehensive model considering throughput, average SOC, and average battery temperature is established to calculate the performance index HI of the power battery. The value range is [0, 1], with a higher value indicating higher system fatigue. The calculation formula is as follows: ; Among them, Q batt For the cumulative throughput of the power battery; Q maxMaximum reference throughput capacity; SOC avg The average SOC value over a set time period from the current time; T avg The average battery temperature over a set time period between the current moment and the current time; T ideal The optimal operating temperature for the battery; T range α represents the difference between the upper and lower limits of the battery's operating temperature range; α, β, and γ are the first, second, and third weighting coefficients, respectively.

[0042] It should be understood that cumulative throughput refers to the total amount of electricity passed through the power battery during charging and discharging. This can be updated in real-time using a recursive formula, reflecting the battery's cyclical usage. Maximum reference throughput refers to the maximum theoretical charge and discharge capacity the power battery can withstand within its lifespan. This can be achieved using the battery's factory calibration value or experimental test values, used to quantify the degree of battery lifespan degradation. Average SOC value refers to the average state of charge of the battery over a set time period. This can be calculated using the sliding window method, used to characterize the stability of the battery's energy state. Average battery temperature refers to the average battery temperature over a set time period. This can be achieved by taking the arithmetic mean of data collected by a temperature sensor, used to reflect the thermal state of the battery's operating environment. Optimal battery operating temperature refers to the midpoint of the temperature range where the power battery performs best. This can be achieved using the midpoint of the recommended operating temperature range provided by the battery manufacturer. The battery operating temperature range span refers to the difference between the highest and lowest permissible operating temperatures of the power battery. This can be calculated using the safe temperature range specified in the battery's technical manual. Weighting coefficients α, β, and γ are used to adjust the contribution of different factors to the performance indicators, and can be determined through experimental calibration or optimization algorithms.

[0043] Specifically, during vehicle operation, the charging and discharging current of the power battery is collected in real time, and the cumulative throughput is updated recursively. This, combined with a preset maximum reference throughput, is used to calculate a relative ratio, reflecting the battery's cycle aging degree. Simultaneously, the average State of Charge (SOC) is calculated over a set time period to assess fluctuations in the battery's energy state. Battery temperature parameters are quantified by calculating the deviation between the actual average temperature and the optimal operating temperature, considering the range of the operating temperature. The three weighting coefficients are dynamically adjusted based on the vehicle's actual operating conditions; for example, the weighting coefficient γ for the temperature term can be increased in high-temperature environments. By weighting and fusing the parameters from the three dimensions—cumulative throughput, SOC stability, and temperature impact—a comprehensive evaluation model is constructed, outputting a real-time power battery performance index. This index serves as an input parameter for the fuzzy controller, working in conjunction with driving conditions and the vehicle's power demand to dynamically adjust the fuel cell's output power allocation strategy.

[0044] This invention establishes a multi-factor coupled model that not only monitors battery energy state but also incorporates cumulative throughput to reflect cycle life degradation and combines temperature parameters to assess thermal management status, forming a multi-dimensional battery health assessment system. This invention optimizes parameter contributions under different environments through adjustable weighting coefficients, improving the adaptability of the assessment model.

[0045] Through the above technical solution, this invention can dynamically quantify the comprehensive workload of the power battery under complex operating conditions and accurately assess the impact of battery health status on lifespan. By calculating the workload index in real time, it provides precise battery status input for fuel cell power distribution, effectively balancing the fuel cell output power and the power battery workload while meeting the vehicle's power requirements, avoiding overcharging and discharging and high-temperature operation, and extending the power battery's service life.

[0046] This invention further proposes a method for determining the target output power of a fuel cell using fuzzy control rules based on driving conditions, battery SOC, workload indicators, and vehicle power requirements, comprising the following steps: a. Blurring: The three input quantities—power battery SOC, workload index, and vehicle power demand—are converted into corresponding fuzzy linguistic variables (e.g., low, medium, high) using their respective membership functions, and their corresponding membership values ​​μ are determined. The driving scenario type is a clear value (e.g., congestion, suburbs, highway, long uphill), so it can directly participate in rule judgment without fuzzification. The membership value for each driving scenario is set to 1.

[0047] b. Fuzzy reasoning: Based on the type and membership value of driving conditions, as well as the fuzzy linguistic variables and membership values ​​of power battery SOC, working intensity index, and vehicle power demand, multiple target output power fuzzy variables and target output power membership values ​​of the fuel cell are determined by reasoning through a pre-set fuzzy rule base.

[0048] The fuzzy rule base embodies the core control strategy. Each rule has a basically the same form: IF (condition 1 AND / OR condition 2...), THEN (output conclusion).

[0049] For example: IF Scenario IS Congested (μ=1) AND HI IS High (μ=0.8), THEN P fc_cmd In IS (μ=min(1,0.8)=0.8).

[0050] When calculating the target output power membership value for each rule, for rules connected by AND, the membership value is the minimum value among the membership values ​​of each condition; for rules connected by OR, the membership value is the maximum value among the membership values ​​of each condition.

[0051] For each fuzzy rule, the target output power fuzzy variable and the fuel cell membership value are obtained. Based on the target output power membership value, the fuzzy interval corresponding to the fuel cell fuzzy variable is clipped to obtain the clipped fuzzy region.

[0052] c. Defuzzification: All fuzzy regions are superimposed to obtain a fuzzy set on the output universe of discourse. The membership value of each power point of the fuel cell is determined based on the fuzzy set. The target output power of the fuel cell is determined based on the membership values ​​of all power points of the fuel cell.

[0053] The membership function is a mathematical function that maps input parameters to fuzzy linguistic variables. It can be implemented using triangular or trapezoidal functions; for example, when dividing SOC into low, medium, and high levels, a corresponding trapezoidal membership function can be set. The fuzzy rule base is a set of rules containing multiple conditional logical combinations, which can be defined using an if-then form. For example, when the driving condition is congested and SOC is low, a low-power fuzzy variable is output. The pruning operation refers to truncating the output fuzzy interval based on the membership value. This can be done by taking the minimum value; for example, the portion of the output power fuzzy interval above the membership value is truncated to the membership value. The superimposed fuzzy set refers to merging multiple pruned fuzzy regions. This can be done by combining the maximum membership values ​​at each point; for example, at the power point of 50kW, the maximum membership value among all rules is taken as the composite result.

[0054] Specifically, the power battery's State of Charge (SOC), workload index, and vehicle power demand are converted into linguistic variables through fuzzification. For example, SOC can be divided into three fuzzy levels: low, medium, and high. Driving conditions serve as conditional variables for fuzzy rules, such as urban, highway, and hill-climbing conditions. Each rule in the fuzzy rule base contains multiple condition combinations. For instance, when the driving condition is highway and the SOC is medium, the corresponding target output power fuzzy variable is matched. The output fuzzy variables generated by each rule are clipped by membership values ​​and then superimposed to form a comprehensive fuzzy set. Finally, the target power value is determined through a defuzzification process. This process optimizes power allocation under different operating conditions by dynamically adjusting the weights of the fuzzy rules, such as prioritizing the reduction of fuel cell power fluctuations when the power battery is under high load.

[0055] This invention utilizes a dynamic matching mechanism based on a fuzzy rule base to collaboratively optimize the power battery's state and operating characteristics. For example, it automatically reduces the fuel cell's start-stop frequency under congested conditions and dynamically increases power output during hill-climbing conditions. By using the power battery's operating intensity index as a fuzzy input, this invention effectively controls the depth of charge and discharge of the power battery.

[0056] Through the above technical solutions, this invention achieves dynamic optimization and adjustment of fuel cell output power, solving the problem of poor adaptability of traditional methods to complex operating conditions. By integrating multi-dimensional parameter fuzzy processing, it can balance instantaneous power demand and long-term equipment losses, such as automatically limiting the charging and discharging rate of the power battery under high load conditions to extend its lifespan. The logical combination mechanism of fuzzy rules enhances the flexibility of the control strategy; for example, when the SOC is low, it prioritizes ensuring stable fuel cell output and avoids over-discharging of the power battery. The superimposed fuzzy set synthesis method ensures the global optimality of multi-rule decision-making; for example, under multiple constraints, it selects the power point with the highest comprehensive membership degree as the optimal solution.

[0057] This invention further proposes a pruning operation for the fuzzy interval of the target output power during the fuzzy control process, specifically including changing all the parts of the fuzzy interval with a membership value greater than the membership value of the target output power to that membership value.

[0058] The target output power membership value refers to the confidence level corresponding to the power variable output by each fuzzy rule. Specifically, it can be determined through the logical operation results of the membership values ​​of each input variable during fuzzy inference, and is used to quantify the influence weight of the current rule on the final decision. The pruning operation refers to the mathematical processing of adjusting the shape of the membership function graph. Specifically, it can use a truncation method to flatten the portion of the original membership curve exceeding a specific threshold. This operation can eliminate abrupt changes in the membership distribution caused by differences in confidence levels between different rules.

[0059] Specifically, after each fuzzy rule is activated, the original shape of the membership curve corresponding to its output power fuzzy variable may exhibit peaks or steepness. By truncating the region of the curve that is higher than the current rule's membership value and replacing it with a fixed value, the peak region of the superimposed total membership distribution curve is limited to a reasonable range. For example, when the power fuzzy variable output by a rule corresponds to the "medium power" interval and has a membership value of 0.6, the original membership function in the medium power interval might reach 0.8. In this case, the region between 0.8 and 0.6 is uniformly set to 0.6. This processing method can balance the weight relationship between multiple rules, avoid a single high-confidence rule excessively dominating the final decision, and thus ensure the continuity of power output commands.

[0060] This invention, through a membership degree pruning mechanism, effectively constrains the influence range of each rule on the final decision, resulting in a smoother transition characteristic in the superimposed membership degree distribution. Through this technical solution, the invention can significantly reduce the frequency of abrupt changes in fuel cell output power commands, especially under conditions of frequent vehicle acceleration and deceleration. By suppressing drastic changes in the membership degree distribution, it ensures a smooth transition of power output commands. This control characteristic reduces the dynamic stress on the fuel cell system, effectively avoiding mechanical damage to the proton exchange membrane caused by power step changes, thereby extending the service life of the fuel cell stack.

[0061] This invention further proposes that, when determining the membership value of each power point in the output universe of discourse during fuzzy control, all membership values ​​in the fuzzy set are traversed for each power point, and the maximum value is selected as the final membership value of that point.

[0062] Here, the fuzzy set refers to the comprehensive membership distribution formed by superimposing all clipped fuzzy regions. This can be achieved using a multi-rule superposition algorithm, and its function is to integrate the influence of different fuzzy rules on the output power. The membership value quantifies the degree to which a power point belongs to a certain fuzzy linguistic variable, and can be calculated using a membership function. Its function is to provide a quantitative basis for power point decision-making. The output universe of discourse refers to the possible range of values ​​for the target output power of the fuel cell. Specifically, it can be a discretized set of power points (e.g., from 0 to 100kW, one point for every 1kW), and its function is to define the physical boundary of the fuzzy control output. The maximum value operation selects the highest value from multiple membership values, which can be achieved using a traversal comparison algorithm. Its function is to retain the rule output that has the most significant impact on the current power point.

[0063] Specifically, during the execution of fuzzy control rules, the fuzzy regions generated by each rule are clipped and superimposed to form a comprehensive fuzzy set. For each discrete power point in the output universe of discourse, the system iterates through all superimposed membership values ​​and records all the corresponding values ​​for that point. By comparing the magnitudes of these values, the system determines the maximum value as the final membership value for that power point. This approach ensures that when multiple rules have different effects on the same power point, only the control signal with the strongest intensity is retained, thereby eliminating the mutual cancellation effect between different rules. For example, under vehicle acceleration conditions, if a power point simultaneously has a high membership degree of a high-power demand rule and a low membership degree of a low-power protection rule, taking the maximum value will prioritize responding to the acceleration demand.

[0064] This invention employs a maximum value filtering mechanism to accurately capture the control rules most decisive for the current operating condition, avoiding the decrease in decision-making accuracy caused by membership averaging. Through this technical solution, the invention effectively solves the problem of insufficient output accuracy in fuzzy control, ensuring that the decision-making process for the target power of the fuel cell accurately reflects the most critical control requirements in actual operating conditions. This solution improves the clarity of power allocation decisions, enabling the control system to respond quickly to changes in vehicle status while maintaining the collaborative working efficiency of the power battery and fuel cell.

[0065] This invention further proposes a method for determining the target output power of a fuel cell using the following formula: ; Among them, P fc_cmd P represents the target output power of the fuel cell. i μ(P) represents the i-th discrete point on the output universe of discourse. i ) is P i The corresponding membership value; n is the total number of discrete points in the output universe of discourse.

[0066] In this context, discrete power points in the output universe refer to power values ​​that divide the output power range of the fuel cell into several equally or unequally spaced intervals, used to construct the output quantization benchmark for fuzzy control. The membership value refers to the degree to which each discrete power point is activated during fuzzy inference, specifically calculated using a membership function, and is used to characterize the reliability of that power point as the final output value. The total number of discrete points, n, refers to the number of power points divided in the output universe, which can be set to 50-100 points based on the controller's computing power to balance computational accuracy and real-time requirements.

[0067] Specifically, after obtaining the membership distribution over the output universe through fuzzy inference, a weighted summation operation is performed on each discrete power point and its corresponding membership value. This weighting process involves multiplying each power point value by its corresponding membership value, summing the results, and then dividing by the sum of all membership values ​​to obtain a precise continuous power output value. This mathematical processing method transforms the originally discrete membership distribution into a continuous power adjustment command, effectively eliminating the abrupt changes caused by directly selecting the power point with the highest membership degree in traditional methods. The setting of the total number of discrete points, n, ensures the smoothness of the output power while avoiding the impact of excessive computation on the real-time performance of the vehicle controller.

[0068] This invention achieves a smooth transition in power output while maintaining computational efficiency through a normalized weighted average algorithm, overcoming the problem of abrupt control command changes caused by discretization processing in traditional methods under dynamic operating conditions. Through this technical solution, the invention effectively solves the power system impact problem caused by step changes in fuel cell power output, enabling the energy management system to generate continuous and smooth power commands based on fuzzy inference results. This significantly improves the smoothness of dynamic response during vehicle acceleration and deceleration, while ensuring the real-time computing capability of the onboard controller.

[0069] This invention further proposes energy management for fuel cell vehicles based on the target output power of the fuel cell. This includes sending the calculated optimal target output power to the fuel cell controller, controlling the fuel cell system to output power at this power, and providing or absorbing the remaining required power by the power battery. The remaining required power is equal to the total vehicle required power minus the target power of the fuel cell. If the remaining required power exceeds the upper limit of the power battery's absorption capacity, the target output power of the fuel cell is limited to the total vehicle required power minus the upper limit of the power battery's absorption capacity.

[0070] The target output power of the fuel cell refers to the power value that the fuel cell system should output, dynamically calculated through fuzzy control rules. Specifically, this can be achieved through online optimization calculations using a fuzzy inference engine combined with real-time operating parameters. This power value balances fuel cell efficiency and lifespan degradation. The remaining power demand refers to the difference between the total power required for vehicle operation and the actual output power of the fuel cell. This can be achieved by real-time monitoring of the vehicle's power demand using power sensors and subtracting it from the fuel cell's output power. This difference is dynamically compensated by the power battery through charging and discharging operations.

[0071] Specifically, when the vehicle's power demand changes, the fuel cell system adjusts to the target output power according to fuzzy control rules, ensuring it operates within its optimal efficiency range. At this time, the battery discharges or charges based on the remaining power demand: if the remaining power demand is positive, the battery releases electrical energy to fill the power gap; if it is negative, the battery absorbs excess energy for charging. This dynamic allocation mechanism, through power decoupling control, allows the battery to handle high-frequency power fluctuations, while the fuel cell maintains a relatively stable output. For example, during vehicle acceleration, the fuel cell maintains a constant power output, while the battery provides a burst of current to meet acceleration demands; during regenerative braking, the fuel cell power is reduced, and the battery quickly absorbs regenerative braking energy. Thus, the fuel cell avoids the accumulation of mechanical stress on the proton exchange membrane caused by frequent load changes, while the depth of charge and discharge of the battery is effectively controlled through real-time power compensation.

[0072] This invention employs a dynamic power decoupling mechanism to separate steady-state power from transient power, ensuring that the fuel cell always operates within a preset high-efficiency range. Compared to static rule-based control where the fuel cell is forced to respond to all power fluctuations, this invention reduces the fuel cell power change rate by approximately 60%, while simultaneously reducing the number of charge-discharge cycles of the power battery by 30%. Furthermore, in existing technologies, the power battery serves only as a passive energy buffer, while this invention, through a bidirectional power processing mechanism, enables it to actively participate in power regulation, thereby enhancing the system's dynamic response capability.

[0073] Through the above technical solution, this invention effectively reduces the lifespan degradation of fuel cells caused by power fluctuations, extending the stack lifespan by more than 20%. Simultaneously, through real-time compensation of remaining power by the power battery, the vehicle's power response speed under complex operating conditions is improved by approximately 15%, and the energy recovery efficiency is increased to over 92%. This solution also achieves coordinated and optimized control of the fuel cell and the power battery, maintaining the overall system efficiency within a high-efficiency range of 85%-90%.

[0074] The following is a detailed explanation of the process of determining the target output power of a fuel cell using fuzzy control rules.

[0075] 1. Definition and fuzzification of input and output variables 1.1 Input variables include: HI: Work intensity index, range [0, 1]; SOC: Battery state of charge, range [0, 100] (%); P_demand: Vehicle power demand, range [0, P_max] (kW). Scenario: Operating condition identification result, a clear value (1, 2, 3...), corresponding to the aforementioned 1-congestion, 2-suburbs, 3-highway, 4-long uphill.

[0076] 1.2 Output variable: P fc_cmd Target output power of fuel cell, range [0, P] fc_cmd (Rated power of fuel cell stack).

[0077] 1.3 Fuzzification: Convert the input value into the membership degree (μ) of the corresponding fuzzy set. Define 2-3 fuzzy sets for each input variable.

[0078] The fuzzy set of HI (Work Intensity Index) is: Low, Medium, High. Membership functions: triangular and trapezoidal functions are used. For example: μ_Low(HI): When HI is close to 0, the membership degree is close to 1; it drops to 0 when HI>0.3. μ_Medium(HI): The membership degree is 1 when HI=0.5, and drops to 0 when HI=0.2 and HI=0.6. μ_High(HI): When HI is close to 1, the membership degree is close to 1; it drops to 0 when HI<0.7.

[0079] Example: If HI=0.25, it may belong to both the Low set and the Medium set. In this case, the fuzzy set with the larger membership degree shall prevail.

[0080] Fuzzy sets of SOC (State of Charge): Low, Medium, High; Universe of discourse: [0, 100] %; Defined as Low: 0% - 30% (low charge, needs charging), Medium: 20% - 80% (moderate charge, can be charged and discharged), High: 70% - 100% (high charge, can accept high-power discharge). Fuzzy sets of P_demand (vehicle power demand): Small, Medium, Large; Universe of discourse: [0, P_max]kW; Defined as Small: 0 - 20% P_max (low load, such as idling, slow driving), Medium: 10% - 60% P_max (medium load, such as cruising), Large: 50% - 100% P_max (high load, such as rapid acceleration, long uphill).

[0081] Scenario: This is a clear input that directly participates in rule judgment and does not need to be fuzzed. The membership degree of each scenario is 1. For example: 1: Congestion, 2: Suburban, 3: Highway, 4: Climbing.

[0082] P fc_cmd The fuzzy sets of the output are: Very_Small, Small, Medium, Large, Very_Large; the universe of discourse is [0,P]. fc_cmd kW; Definition: Assigns a specific range of power values ​​to its output fuzzy set.

[0083] 2. Establishment of a fuzzy rule base This is the core step of the control strategy, a series of IF-THEN statements based on expert knowledge. The rule base needs to cover typical cases of all input combinations. Below are some rule examples; only fuzzy sets are shown in the examples, and the corresponding membership values ​​are not displayed: 2.1, IF Scenario IS Congestion AND HI IS High, THEN P fc_cmd IS Medium. Explanation: Core protection rule. In congested conditions and with high system fatigue, the fuel cell is forced to operate at a stable, medium power level, ignoring frequent fluctuations in demand, which are then handled by the battery.

[0084] 2.2, IF SOC IS Low AND P_demand IS Large, THEN P fc_cmd IS Large. Explanation: Minimum guarantee rule. If the battery capacity is insufficient but the power demand is high, the fuel cell power must be increased to meet the power requirements and charge the battery. This priority is higher than lifespan protection.

[0085] 2.3、IF Scenario IS Highway AND HI IS Low AND SOC IS Medium, THENP fc_cmd IS Large. Explanation: Optimization rule. Stable operating conditions, low system fatigue, and moderate power allow the fuel cell to operate in a high-efficiency, high-power region, reducing battery consumption.

[0086] 2.4, IF P_demand IS Small AND SOC IS High, THEN P fc_cmd IS Very_Small. Explanation: Energy-saving rule. With low power demand and sufficient electricity, the fuel cell power can be significantly reduced to idle or near-shutdown, saving hydrogen.

[0087] 2.5, IF Scenario IS Climbing, THEN P fc_cmd IS Large. Explanation: Special operating condition rule. Upon detecting a ramp-up and anticipating continued high power demand, the fuel cell is instructed to increase power in advance to prevent battery depletion.

[0088] 3. Fuzzy reasoning and defuzzification: This involves transforming fuzzy rules into precise output P. fc_cmd The calculation process.

[0089] 3.1 Fuzzy Reasoning: For each rule, calculate its membership value (μ). For rules connected by AND, the membership value is the minimum of the membership values ​​of each condition; for rules connected by OR, the membership value is the maximum of the membership values ​​of each condition.

[0090] Example: For rule 2.1: IF Scenario IS Congestion (1.0) AND HI IS High (0.8), THEN ...

[0091] The membership value of this rule is μ_rule1 = min(1.0, 0.8) = 0.8.

[0092] Use this μ to trim the output fuzzy set corresponding to the rule conclusion, and finally obtain a trimmed fuzzy output region.

[0093] 3.2 Defuzzification: The cropped fuzzy sets corresponding to all activated rules are superimposed to form a total, complex-shaped output membership function. The centroid of this total output membership function is calculated; the x-coordinate of this centroid is the final precise output value P. fc_cmd .

[0094] Formula for calculating the center of gravity: .

[0095] Where: P i It is the i-th discrete power point on the output universe of discourse (e.g., one point for every 1kW from 0 to 100kW), μ(P i ) is the discrete point P i The corresponding membership value, where n is the total number of discrete points in the output universe of discourse.

[0096] 4. Issuance and execution of instructions The optimal target output power value of the fuel cell calculated in step 3 is sent to the fuel cell controller to control the fuel cell system to output power at this value. The remaining power demand (total power demand - target power of the fuel cell) is provided or absorbed by the power battery pack.

[0097] This invention further proposes a fuel cell vehicle energy management system for implementing the above-mentioned fuel cell vehicle energy management method, comprising: The parameter acquisition module is used to acquire vehicle driving parameters and power battery parameters; The driving condition recognition module is used to identify the current driving condition based on vehicle driving parameters; The power battery performance evaluation module is used to evaluate the performance of the power battery based on its parameters and obtain the power battery performance index. The output power determination module is used to determine the target output power of the fuel cell based on driving conditions, power battery SOC, workload index, and vehicle power demand using fuzzy control rules; the energy management module is used to manage the energy of the fuel cell vehicle based on the target output power of the fuel cell.

[0098] The system comprises several modules: The parameter acquisition module, which collects real-time vehicle operating status and battery status data (using onboard sensors and a battery management system), and the output power determination module. The parameter acquisition module collects parameters such as vehicle speed, acceleration, battery temperature, and charging / discharging current to provide a data foundation for dynamic control. The driving condition recognition module classifies driving conditions based on vehicle driving parameters. It uses a random forest algorithm to perform pattern recognition on feature parameters within a set time window, such as calculating average vehicle speed and speed standard deviation to determine complex driving conditions in real time. The power battery performance evaluation module quantifies battery health. It employs a weighted calculation model based on cumulative throughput, SOC change rate, and temperature deviation to dynamically assess battery wear and avoid the limitations of a single SOC indicator. The output power determination module is a controller based on multi-factor collaborative decision-making. It uses a fuzzy logic algorithm to integrate driving condition type, battery status, and power demand, and dynamically optimizes power output through membership functions and a rule base. The energy management module is a device that coordinates the power distribution between the fuel cell and the power battery. Specifically, it can use a residual power compensation algorithm to control the battery charging and discharging behavior based on the difference between the target power and the power required by the vehicle.

[0099] Specifically, the parameter acquisition module continuously collects vehicle speed and acceleration signals through onboard sensors, and simultaneously acquires the SOC, temperature, and current data of the power battery through the battery management system. The driving condition recognition module extracts feature parameters such as the average vehicle speed and speed standard deviation within the current time window, inputs them into a pre-trained random forest model for condition classification, such as outputting a congested condition or a high-speed condition. The power battery workload assessment module, based on the cumulative throughput formula, combined with the current SOC fluctuation and the degree of temperature deviation from the optimal operating range, generates quantitative indicators reflecting the battery's health status. The output power determination module uses the driving condition type, workload indicators, SOC, and vehicle power demand as fuzzy control inputs, converts them into fuzzy linguistic variables through a membership function, performs inference calculations based on a preset rule base, and finally obtains the target power value of the fuel cell through defuzzification processing. The energy management module controls the power battery to perform power compensation or energy recovery based on the difference between the target power and the vehicle power demand, for example, supplementing the difference in power when the fuel cell output power is insufficient.

[0100] This invention achieves dynamic operating condition adaptation through an operating condition identification module, solving the control lag problem of static rules in complex traffic environments; it quantifies battery loss through a workload assessment module, overcoming the limitation of traditional methods that rely solely on SOC; and it adopts fuzzy control rules to fuse multi-dimensional parameters, reducing fuel cell load fluctuations while meeting real-time power requirements, thus achieving coordinated control of lifespan management and economic optimization.

[0101] Through the above technical solutions, this invention can adapt to various real-time changing driving conditions such as urban congestion and highways, and reduce lifespan loss caused by frequent start-stop cycles by dynamically adjusting the fuel cell output power. Quantitative assessment of the power battery's workload effectively avoids battery performance degradation caused by overcharging and over-discharging. The multi-factor fusion decision-making mechanism of fuzzy control rules ensures that the power distribution scheme meets the vehicle's power requirements while also considering the efficiency of the fuel cell system, thereby achieving optimized control of vehicle energy management under complex operating conditions.

[0102] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process may be rearranged without departing from the scope of this disclosure. The appended method claims provide elements of various steps in an exemplary order and are not intended to limit the scope to the specific order or hierarchy described.

[0103] Those skilled in the art will also understand that the various illustrative logic blocks, units, and steps listed in the embodiments of the present invention can be implemented by electronic hardware, computer software, or a combination of both. To clearly demonstrate the substitutability of hardware and software, the functions of the various illustrative components, units, and steps described above have been generally described. Whether such functionality is implemented through hardware or software depends on the specific application and the overall system design requirements. Those skilled in the art can implement the described functions using various methods for each specific application, but such implementation should not be construed as exceeding the scope of protection of the embodiments of the present invention.

[0104] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Contents not described in detail in this specification belong to prior art known to those skilled in the art.

Claims

1. A method for energy management of a fuel cell vehicle, characterized in that: Obtain vehicle driving parameters and power battery parameters; Identify the current driving condition based on vehicle driving parameters; The working intensity index of the power battery is obtained by evaluating the working intensity of the power battery based on the parameters of the power battery. The target output power of the fuel cell is determined by fuzzy control rules based on driving conditions, power battery SOC, workload index, and vehicle power demand. Energy management for fuel cell vehicles is based on the target output power of the fuel cell.

2. The energy management method for fuel cell vehicles according to claim 1, characterized in that: After obtaining the vehicle driving parameters and power battery parameters, the obtained parameters are filtered and validated. The validation means verifying whether the signal value at the corresponding time is within the set range. If it is not within the set range, the average value of the signal value at the previous time and the next time is calculated and the average value is used to replace the signal value at the corresponding time. If the signal value at an adjacent time point is invalid, the signal value at the previous valid time point is used to replace it.

3. The energy management method for fuel cell vehicles according to claim 1, characterized in that, The step of identifying the current driving condition based on vehicle driving parameters includes: A random forest model was trained using historical vehicle driving parameters. Based on vehicle driving parameters, multiple feature parameters within a set time window are extracted. These multiple feature parameters include average vehicle speed, vehicle speed standard deviation, average absolute acceleration, idling time ratio, and the ratio of maximum demand power to average demand power. Multiple feature parameters are input into a random forest model, and the current driving conditions are obtained through a voting mechanism.

4. The energy management method for fuel cell vehicles according to claim 1, characterized in that: The operating intensity index of the power battery is determined by the following formula: ； Wherein, HI represents the operating intensity index of the power battery; Q batt For the cumulative throughput of the power battery; Q max Maximum reference throughput capacity; SOC avg The average SOC value over a set time period from the current time; T avg The average battery temperature over a set time period between the current moment and the current time; T ideal The optimal operating temperature for the battery; T range Q represents the difference between the upper and lower limits of the battery's operating temperature range; α, β, and γ are the first, second, and third weighting coefficients, respectively; batt (k) represents the cumulative throughput of the power battery at time k; Q batt (k-1) represents the cumulative throughput of the power battery at time k-1; I batt (k) represents the charging or discharging current of the power battery at time k; Δt represents the sampling period.

5. The energy management method for fuel cell vehicles according to claim 1, characterized in that, The determination of the target output power of the fuel cell using fuzzy control rules based on driving conditions, battery SOC, workload indicators, and vehicle power requirements includes: The three input quantities, namely, the SOC of the power battery, the working intensity index, and the power demand of the whole vehicle, are converted into corresponding fuzzy linguistic variables through the corresponding membership functions and the corresponding membership values ​​are determined. Based on driving conditions, power battery SOC, workload index, vehicle power demand, and fuzzy linguistic variables and membership values, multiple target output power fuzzy variables and target output power membership values ​​of fuel cells are determined by searching the fuzzy rule base. For each fuzzy rule, the target output power fuzzy variable and the target output power membership value are obtained. Based on the target output power membership value, the fuzzy interval corresponding to the target output power fuzzy variable is clipped to obtain the clipped fuzzy region. All fuzzy regions are superimposed to obtain a fuzzy set on the output universe of discourse, and the membership value of each power point on the output universe of discourse is determined based on the fuzzy set; The target output power of the fuel cell is determined based on the membership values ​​of all power points in the output universe.

6. The energy management method for fuel cell vehicles according to claim 5, characterized in that, The step of cropping the fuzzy interval corresponding to the fuzzy variable of the target output power based on the membership value of the target output power includes: All membership values ​​in the fuzzy range of the target output power that are greater than the target output power membership value are changed to the target output power membership value.

7. The energy management method for fuel cell vehicles according to claim 5, characterized in that, The determination of the membership value of each power point in the output universe based on fuzzy sets includes: For each power point in the output universe of discourse, find all membership values ​​of the fuzzy set at that power point, and take the largest membership value as the membership value of that power point.

8. The energy management method for fuel cell vehicles according to claim 1, characterized in that, The target output power of the fuel cell is determined using the following formula: ； Among them, P fc_cmd P represents the target output power of the fuel cell. i μ(P) represents the i-th discrete point on the output universe of discourse. i ) is P i The corresponding membership value; n is the total number of discrete points in the output universe of discourse.

9. The energy management method for fuel cell vehicles according to claim 1, characterized in that, The energy management of fuel cell vehicles based on the target output power of the fuel cell includes: The target output power is controlled by the fuel cell system, and the remaining required power is provided or absorbed by the power battery. The remaining required power = vehicle required power - fuel cell target power.

10. An energy management system for a fuel cell vehicle, characterized in that: include The parameter acquisition module is used to acquire vehicle driving parameters and power battery parameters; The driving condition recognition module is used to identify the current driving condition based on vehicle driving parameters; The power battery performance evaluation module is used to evaluate the performance of the power battery based on its parameters and obtain the power battery performance index. The output power determination module is used to determine the target output power of the fuel cell based on driving conditions, power battery SOC, workload index, and vehicle power demand using fuzzy control rules. The energy management module is used to manage the energy of fuel cell vehicles based on the target output power of the fuel cell.

Images