A method for optimizing loading strategy of vehicle fuel cell system

Abstract

The present invention discloses a method for optimizing the loading strategy of a vehicle fuel cell system, including: dividing different stack output power intervals according to the polarization characteristic curve of the fuel cell stack and the evolution law of the voltage undershoot with the change of the initial current density during dynamic step loading; calculating candidate loading rates according to different stack output power intervals and corresponding reference loading rate values; formulating loading evaluation indicators and constructing a normalized multi-objective loading evaluation function; performing a pairwise power loading test using the candidate loading rates, and determining the optimal loading rate based on the normalized multi-objective loading evaluation function. The present invention takes into account the rapidity of the dynamic response of system loading, the stability of the stack and the reliability of the single cell.

Description

Technical Field

[0001] The present invention relates to the field of automotive fuel cells, and in particular to a method for optimizing a loading strategy of a vehicle fuel cell system. Background Art

[0002] The main differences between automotive fuel cells and stationary fuel cells are reflected in the frequency and amplitude of load changes. When the fuel cell engine controller receives the loading signal of the power request, it takes a certain amount of time for the air flow provided by the air compressor to the stack to meet the gas volume required for the target power. At the same time, the uneven gas concentration and temperature distribution between single cells during the loading process lead to a large single voltage variance coefficient, which will indirectly shorten the remaining service life of the stack. For different loading rates, the dynamic response speed and single voltage consistency are in a contradictory relationship. Therefore, how to quickly and maintain a high single voltage consistency has become a problem to be solved. At present, domestic and foreign researchers rarely study loading decision optimization strategies. Summary of the invention

[0003] This application establishes a normalized multi-objective optimization function to address this problem, mainly optimizing the decision loading rate from the perspectives of dynamic response speed, single cell reliability and system economy to improve the output performance of the fuel cell engine system.

[0004] The present application provides a method for optimizing a loading strategy of a vehicle fuel cell system, which is used to evaluate whether a loading rate is appropriate and to determine an optimal loading method from candidate loading rates. The method comprises the following steps:

[0005] S1: Different stack output power ranges are divided according to the polarization characteristic curve of the fuel cell stack and the evolution law of voltage undershoot with initial current density during dynamic step loading;

[0006] S2: Calculate candidate loading rates according to different stack output power intervals and corresponding reference loading rate values;

[0007] S3: Formulate loading evaluation indicators and construct normalized multi-objective loading evaluation function;

[0008] S4: Perform pairwise power loading tests using candidate loading rates, and determine the optimal loading rate based on a normalized multi-objective loading evaluation function.

[0009] Furthermore, the different battery stack output power intervals include four types, namely: the first area nonlinear low load area, the second area linear low load area, the third area linear medium load area and the fourth area linear high load area; among which low, medium and high loads specifically refer to: according to the load current density from small to large, they are distinguished according to preset thresholds.

[0010] Furthermore, step S2 is specifically as follows:

[0011] S21: determining a candidate loading rate level difference d according to different stack output power intervals and corresponding reference loading rate values;

[0012] S22: Calculate the candidate loading rate according to the number of candidate loading rates, the reference loading rate value, and the candidate loading rate level difference d, as shown in the following formula:

[0013]

[0014] in, represents the reference rate, i represents a non-negative integer, i is 0, indicating that the candidate loading rate is the reference loading rate value, and i is n / 2, indicating the boundary value set for the candidate loading rate; k dj represents the jth candidate loading rate; n is an even number.

[0015] Furthermore, the loading evaluation index includes: the fuel cell system dynamic response speed evaluation index t, the fuel cell dynamic response period load tracking accuracy evaluation index MAE, the loading stability evaluation index V v , Loading reliability evaluation index C v , Economic evaluation indexe s .

[0016] Furthermore, the calculation formula of the normalized multi-objective loading evaluation function is as follows:

[0017]

[0018]

[0019] Among them, J i represents the evaluation function; A i and B i It represents the loading test coefficient, and its value is either 0 or 1, which is related to the specific situation of the loading test; w f 、w a 、w s 、w c 、w e are the dynamic weight factors of each evaluation index; t i *,MAE i *,V vi *,C vmaxi *,e si *represents the normalized fuel cell system dynamic response speed evaluation index, fuel cell load tracking accuracy evaluation index during dynamic response, loading stability evaluation index, loading reliability evaluation index, and economic evaluation index respectively; η FCS * represents the normalized fuel cell system efficiency; Pfcsi * represents the normalized fuel cell engine output power; h ef and h p It is also the dynamic weight factor of the economic evaluation index.

[0020] Furthermore, the specific process of determining the optimal loading rate in step S4 is as follows:

[0021] Based on the candidate loading rate i, the candidate loading rate is adjusted, and m groups of different linear loading rate tests are carried out. The group of rates with the highest score J is selected from the m groups of results as the optimal loading rate.

[0022] w f 、w a 、w s 、w c 、w e The calculation formula is as follows:

[0023]

[0024]

[0025]

[0026]

[0027]

[0028] in,

[0029]

[0030]

[0031] h f ,h a ,h s ,h c and h e They represent the preset weight coefficients of the corresponding loading evaluation indicators, and h f +h a +h s +h c +h e =1;

[0032]

[0033]

[0034]

[0035] w fi,i+1 , wSi,i+1 , w ai,i+1 , w Ci,i+1 and w ei,i+1 It means that when the loading rate changes from k i Change to k i+1 , the reference weight of each loading evaluation index to the normalized multi-objective loading evaluation function.

[0036] The beneficial effects provided by the present invention are: taking into account the rapidity of dynamic response of system loading, the stability of the battery stack and the reliability of the single battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic flow chart of the method of the present invention; DETAILED DESCRIPTION

[0038] To make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0039] Before explaining, the following basic concepts of this application are explained.

[0040] Please refer to Table 1, which is the reference loading rate of a 90kW fuel cell engine;

[0041] Table 1 Reference loading rate of 90kW fuel cell engine

[0042]

[0043]

[0044] The purpose of the present invention is to determine whether the loading rate of the fuel cell system is being loaded and whether the optimal rate can be determined from the candidate loading rates. The loading effect evaluation is mainly conducted from five different angles: dynamic response time (rapidity), average voltage change rate (stability) from the first loading to the requested power to the stabilization of the stack voltage, load following accuracy (accuracy), cell voltage consistency (reliability), and system output power and system efficiency (economy).

[0045] The loading rate of the stack at each power point in Table 1 is optimized as a control group. First, the fuel cell engine is operated at rated power. Once the outlet temperature of the fuel cell stack coolant reaches the normal temperature, the variance coefficient of the single cell is counted and the value is used as the single cell voltage consistency reliability threshold.

[0046] In order to protect the proton exchange membrane, the hydrogen and air inlet pressure difference given by the manufacturer is used as the mechanical strength reliability threshold.

[0047] During the loading process, if the variance coefficient or voltage difference of the single cell exceeds the corresponding threshold, it indicates that this power rate point is not suitable for loading.

[0048] On this basis, a normalized multi-objective optimization function is constructed from five perspectives: fuel cell system loading rapidity, stability, dynamic tracking accuracy, reliability and economy, to evaluate the loading effect of the fuel cell system.

[0049] Finally, the loading evaluation method proposed in the present invention is verified based on the degradation effect of the short-term loading cycle test condition and the stack durability test.

[0050] Please refer to Figure 1 , Figure 1 It is a schematic flow chart of the method of the present invention;

[0051] A method for optimizing a loading strategy of a vehicle fuel cell system comprises the following steps:

[0052] S1: Different stack output power ranges are divided according to the polarization characteristic curve of the fuel cell stack and the evolution law of voltage undershoot with initial current density during dynamic step loading;

[0053] It should be noted that the different battery stack output power intervals include four types, namely: the first area nonlinear low load area, the second area linear low load area, the third area linear medium load area and the fourth area linear high load area; among which low, medium and high loads specifically refer to: according to the load current density from small to large, they are distinguished according to preset thresholds.

[0054] As an implementation, the present application defines the fuel cell system power range (P1 is the idle power, P 21 is the peak power) is discretized into 20 small intervals (see Table 1).

[0055] First, the polarization curve is divided into a nonlinear region and an ohmic linear region. Since the peak power point of the system is not in the mass transfer nonlinear region, this region is not considered. The division results are: P1-P3 is the activation nonlinear region, P4-P 21 It is the Ohm linear region.

[0056] Secondly, the load current density is divided into intervals. Below 0.4A / cm2 is the low load area (P1-P4), 0.4-1.2A / cm2 (P5-P13) is the medium load area, and above 1.2A / cm2 is the high load area (P14-P21). The division of load current density is mainly based on the evolution law of voltage undershoot with the change of initial current density during dynamic step loading: when the step amplitude is 0.4A / cm2 and the load current density is lower than 0.4A / cm2, the loading transient voltage undershoot increases with the increase of initial load current density. When the load current density is in the range of 0.4A / cm2-1.2A / cm2, the transient voltage undershoot decreases slowly with the increase of initial load current density. When the load current density is higher than 1.2A / cm2, the transient voltage undershoot decreases rapidly with the increase of initial load current density.

[0057] Finally, the previous two power division methods are combined (see Table 2): P1-P3 is a nonlinear low load area (Zone 1), P4 is a linear low load area (Zone 2), P5-P13 is a linear medium load area (Zone 3), and P14-P21 is a linear high load area (Zone 4).

[0058] Table 2 Power range division

[0059] Power range Low middle high Nonlinear <![CDATA[P1 - P3 (Zone 1)]]> Ohm Linear <![CDATA[P4 (Zone 2)]]> <![CDATA[P5-P 13 (Zone 3)]]> <![CDATA[P 14 -P 21 (Zone 4)]]>

[0060] S2: Calculate candidate loading rates according to different stack output power intervals and corresponding reference loading rate values;

[0061] It should be noted that step S2 is specifically as follows:

[0062] S21: determining a candidate loading rate level difference d according to different stack output power intervals and corresponding reference loading rate values;

[0063] It should be noted that the candidate rate level difference refers to the tolerance of the arithmetic progression of the rate values ​​to be optimized for each group of loading conditions, which is the rate level difference d. The higher the power range (1-4 zones), the smaller the rate level difference. Because the higher the power, the greater the variance coefficient of the single battery, and the worse the reliability of the battery stack. Therefore, in order to protect the battery stack, the decision rate difference for loading in the high power range should be small. The specific size of the level difference d must be determined manually based on the reference rate value corresponding to a certain power point in Table 1;

[0064] S22: Calculate the candidate loading rate according to the number of candidate loading rates, the reference loading rate value, and the candidate loading rate level difference d, as shown in the following formula (1):

[0065]

[0066] in, represents the reference rate, i represents a non-negative integer, i is 0, indicating that the candidate loading rate is the reference loading rate value, and i is n / 2, indicating the boundary value set for the candidate loading rate; k dj represents the jth candidate loading rate; n is an even number.

[0067] S3: Formulate loading evaluation indicators and construct normalized multi-objective loading evaluation function;

[0068] It should be noted that the loading evaluation index includes: the fuel cell system dynamic response speed evaluation index t, the fuel cell dynamic response load tracking accuracy evaluation index MAE, the loading stability evaluation index V v , Loading reliability evaluation index C v , Economic evaluation indexe s .

[0069] The concepts of each indicator are as follows:

[0070] Fuel cell system dynamic response speed evaluation index: The loading dynamic response time is used to characterize the loading speed. When the test platform sends a request loading signal instruction to the fuel cell engine, the timing t0 begins. When the actual output power of the stack reaches the target requested power value for the first time, the timing t1 is started, and the loading dynamic response time is t0-t1.

[0071] Evaluation index of load tracking accuracy during dynamic response of fuel cell: Dynamic tracking capability refers to the error between the actual output power of the fuel cell stack and the target planned power. The calculation formula is as follows. MAE represents the average error during dynamic response of loading, l represents the number of samples, and f i Indicates the actual power, y i Indicates the target planned power (transient power that increases linearly at the decision rate). If the decision rate is set reasonably, the actual output power of the stack will lag the theoretical power for a short time, indicating that the system has a strong ability to dynamically track loads.

[0072]

[0073] Loading stability evaluation index: During the rising stage of the actual output power of the fuel cell, the amount of water produced at the cathode continues to increase. When loaded to the target power, the water balance begins to be re-established due to the concentration gradient of water inside the stack. The process of re-establishing water balance is also the process of the stack voltage returning to the steady-state voltage.

[0074] Different loading rates result in different water production speeds and voltage change rates, which in turn affect the stability of the stack. The voltage change rate during this period is calculated and used to characterize the stability of the stack loading (as shown in the following formula (3)). This period refers to the time it takes for the voltage to rise from the minimum value to 98.2% of the steady-state voltage value.

[0075]

[0076] Loading reliability evaluation index: using single battery variance coefficient C v Or minimum distance V r To evaluate the uniformity between battery cells, these two statistical indicators are selected as representatives of reliable operation of the battery stack. Their expressions are formulas 4 and 5. Vi represents the voltage of a single battery; is the average single cell voltage; n is the number of cells.

[0077]

[0078]

[0079] In the embodiment of the present application, the single cell variance coefficient C is used v express;

[0080] System economic evaluation index: System economic performance mainly depends on the system net output power (engine power) and hydrogen consumption. It is characterized by system efficiency and fuel cell engine output power. System efficiency is expressed by formula 6, η FCS Represents the efficiency of the fuel cell system; P F is the engine output power (kW); m H2 is the hydrogen flow rate (g / s); LHV H2 It is the low calorific value of hydrogen; its size is 1.2*10^5kJ / kg.

[0081]

[0082] As an example, the loading effect is evaluated from five perspectives: rapidity of dynamic response of loading, load tracking accuracy, stability of battery stack voltage, reliability of single cell operation, and system economy. A weight coefficient is set for each indicator to characterize the importance of the evaluation indicator on the loading effect. The specific expressions are shown in formulas 7-8.

[0083] S4: Perform pairwise power loading tests using candidate loading rates, and determine the optimal loading rate based on a normalized multi-objective loading evaluation function.

[0084] In each loading condition, m groups of different linear loading rate tests are carried out, and the group of rates with the highest total score J is selected from the m groups of results as the optimal loading rate.

[0085] If the variance coefficient of a single cell or the pressure difference between hydrogen and air inlets exceeds the threshold, the loading rate is considered to be unsuitable for loading under this condition. Under normal temperature in steady-state conditions, the variance coefficient of a single cell when the stack outputs peak power is used as the threshold. The stack inlet pressure difference threshold is the manufacturer's recommended value of 0.5 bar.

[0086] In order to quantify the contribution of each evaluation index to the objective function, they are normalized respectively.

[0087] As shown in Formula 9: t*,MAE*,V v *,C vmax *, and e s *Represent the normalized values ​​of loading evaluation indicators.

[0088]

[0089]

[0090]

[0091]

[0092] As the loading conditions change, the weight of each evaluation index on the objective function will also change, so the weight coefficient needs to be adjusted dynamically.

[0093] In each loading condition, for a rate interval (k from k i Change to k i+1 ), the reference value calculation formula of the loading evaluation weight coefficient is shown in formula 10-12. fi,i+1 , w Si,i+1 , w ai,i+1 , w Ci,i+1 and w ei,i+1 It means that when the loading rate changes from k i Change to k i+1 When the load is adjusted, the dynamic response time, the average load tracking error, the average voltage change rate, the maximum variance coefficient of the single battery and the system economy are the reference weights of the normalized multi-objective loading evaluation function.

[0094] The larger the change in the normalized index (the term with an absolute value sign in Formula 10), the more sensitive the evaluation index is to the change in loading rate. Changing the index can more easily increase the objective function value, and the corresponding weight factor should be larger. Therefore, the weight factor of the loading evaluation index is proportional to the absolute value of the change in the normalized index (Formula 10).

[0095] The arithmetic mean of the reference weight coefficients of all rate intervals is calculated as the reference weight of the loading condition (Equation 13).

[0096]

[0097]

[0098]

[0099]

[0100] In addition, due to different loading conditions, the mechanisms that affect the output performance of the fuel cell system are different, and the performance indicators that are focused on are different.

[0101] In order to highlight the key loading evaluation index, it is necessary to increase the reference weight factor value of the index and correct the weight of the index.

[0102] The calculation formula of the weight factor after loading index correction is shown in formula 15. Among them, h f ,h a ,h s ,h c and h e Denote the subjective weight coefficients of the corresponding evaluation indicators respectively. The subjective weight coefficients are determined according to the evaluation indicators of key concern, and the sum of the subjective weight coefficients is ensured to be 1 (Formula 14).

[0103] h f +h a +h s +h c +h e =1 (14)

[0104]

[0105]

[0106]

[0107]

[0108]

[0109] As an embodiment, in order to verify the effectiveness of the fuel cell system loading evaluation method, a short-term loading dynamic response test is performed. The loading dynamic response test is repeated three times with the original loading strategy (Table 2), and the evaluation results are recorded in the second column of Table 3. The rate value in Table 2 is replaced with the optimized rate result on the host computer. The loading dynamic response test is also repeated three times with the optimized loading rate, and the results are recorded in the third column of Table 3 (it should be noted that Table 3 only gives a schematic diagram of the recording table), and the evaluation indicators before and after optimization are compared.

[0110] The specific loading dynamic response test steps are carried out as follows:

[0111] a) After the warm-up process is completed, return to idle state and run for 10 seconds;

[0112] b) Load to the 10% PE (system rated power) power point of dynamic response in accordance with the prescribed loading method, and operate stably at this power point for at least 1 minute;

[0113] c) The test platform sends a dynamic step working instruction to the fuel cell engine. At the same time, the test platform is loaded according to the prescribed loading method until the cutoff point of the dynamic step (10% PE) is reached. The fuel cell engine continues to operate stably at this power point for at least 10 minutes.

[0114] d) Repeat steps b and c twice.

[0115] Table 3 Short-term test verification of engine loading strategy

[0116]

[0117]

[0118] Verification by long-term durability test. The evaluation method can also be verified by a long-term durability test of the fuel cell. Before the test, a steady-state characteristic test of the fuel cell system is carried out (select 10% PE, 20% PE, …, PE, a total of 10 power points, and each power point runs stably for at least 3 minutes). The stack current and voltage of each power point are recorded, and the polarization characteristic curve of the fuel cell stack is plotted. A 50-hour durability test is carried out on the fuel cell engine, and then a steady-state test is carried out on the same 10 power points. The voltage and decay rate of each power point are counted (see Table 4. It should be noted that Table 4 only gives a schematic diagram of the recording table).

[0119] Table 4 Engine loading strategy durability test verification indicators

[0120] Stack voltage decay rate Minimum single voltage decay rate Original loading strategy Optimizing loading strategies

[0121] The beneficial effects of the present invention are:

[0122] 1. Based on the two reliability thresholds in the fuel cell system loading evaluation method, it can be judged whether the loading rate is too large.

[0123] 2. The optimization method of the invention is essentially to control the rate of change of load power, which is easy to implement on the vehicle.

[0124] 3. This evaluation method mainly takes the dynamic response time, the rate of change of the battery stack voltage and the consistency of the single cell as the optimization targets, taking into account the rapidity of the dynamic response of the system loading, the stability of the battery stack and the reliability of the single cell.

[0125] 4. The evaluation method adopted is based on the idealized idea of ​​calculus small intervals. The entire power interval is first discretized, and the linear loading rate of each small interval is used as the decision variable to be optimized.

[0126] 5. The method for determining the weight coefficient of the normalized multi-objective evaluation function is relatively objective. First, the reference weight coefficient is determined based on the dynamic weight factor method, and then the reference weight coefficient is corrected according to the voltage loss characteristics of the fuel cell stack polarization curve at different powers.

[0127] 6. The method of the present invention can be verified by a short-term loading dynamic response test or a long-term durability test.

[0128] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for optimizing a loading strategy of a vehicle fuel cell system, characterized in that: The following steps are involved: S1: Different stack output power ranges are divided according to the polarization characteristic curve of the fuel cell stack and the evolution law of voltage undershoot with initial current density during dynamic step loading; S2: Calculate candidate loading rates according to different stack output power intervals and corresponding reference loading rate values; S3: Formulate loading evaluation indicators and construct normalized multi-objective loading evaluation function; The loading evaluation indexes include: the fuel cell system dynamic response speed evaluation index t, the fuel cell dynamic response period load tracking accuracy evaluation index MAE, the loading stability evaluation index V v , Loading reliability evaluation index C v , Economic evaluation indexe s ; The calculation formula of the normalized multi-objective loading evaluation function is as follows: Among them, J i represents the evaluation function; A i and B i It represents the loading test coefficient, and its value is 0 or 1, which is related to the specific situation of the loading test; f 、w a 、w s 、w c 、w e are the dynamic weight factors of each evaluation index; t i *,MAE i *,V vi *,C vmaxi *,e si *represents the normalized fuel cell system dynamic response speed evaluation index, fuel cell load tracking accuracy evaluation index during dynamic response, loading stability evaluation index, loading reliability evaluation index, and economic evaluation index respectively; η FCS * represents the normalized fuel cell system efficiency; P fcsi * represents the normalized fuel cell engine output power; h ef and h p It is also the dynamic weight factor of the economic evaluation index; S4: Perform pairwise power loading tests using candidate loading rates, and determine the optimal loading rate based on a normalized multi-objective loading evaluation function.

2. The method for optimizing a vehicle fuel cell system loading strategy according to claim 1, characterized in that: The different battery stack output power intervals include four types, namely: a first area nonlinear low load area, a second area linear low load area, a third area linear medium load area and a fourth area linear high load area; among which low, medium and high loads specifically refer to: according to the load current density from small to large, they are distinguished according to preset thresholds.

3. The method for optimizing a vehicle fuel cell system loading strategy according to claim 2, wherein: Step S2 is specifically as follows: S21: determining a candidate loading rate level difference d according to different stack output power intervals and corresponding reference loading rate values; S22: Calculate the candidate loading rate according to the number of candidate loading rates, the reference loading rate value, and the candidate loading rate level difference d, as shown in the following formula: in, represents the reference rate, i represents a non-negative integer, i is 0, indicating that the candidate loading rate is the reference loading rate value, and i is n / 2, indicating the boundary value set for the candidate loading rate; k dj represents the jth candidate loading rate; n is an even number.

4. The method for optimizing a loading strategy of a vehicle fuel cell system according to claim 1, characterized in that: The specific process of determining the optimal loading rate in step S4 is as follows: Based on the candidate loading rate i, the candidate loading rate is adjusted, and m groups of different linear loading rate tests are carried out. The group of rates with the highest score J is selected from the m groups of results as the optimal loading rate.

5. The method for optimizing a loading strategy of a vehicle fuel cell system according to claim 1, characterized in that: w f 、w a 、w s 、w c 、w e The calculation formula is as follows: in, h f ,h a ,h s ,h c and h e They represent the preset weight coefficients of the corresponding loading evaluation indicators, and h f +h a +h s +h c +h e =1; w fi,i+1 , w Si,i+1 , w ai,i+1 , w Ci,i+1 and w ei,i+1 It means that when the loading rate changes from k i Change to k i+1 , the reference weight of each loading evaluation index to the normalized multi-objective loading evaluation function.

Images