Fault detection and deterioration estimation device

Abstract

To provide a failure detection and deterioration estimation device capable of highly accurately detecting an abnormality or a failure of a fuel cell even in a case where a cell voltage repeats a temporary fluctuation.SOLUTION: Output data including an FC current i(t), a first cell voltage V1(t), a first feature amount X1(t), and a second feature amount X2(t) of a fuel cell at a time (t) are acquired successively. Next, the time (t) and the first cell voltage V1(t) at a time in which i(t) is a representative current ir are extracted from the output data, and a representative I-V characteristic Vf (t, ir) is calculated from them. Next, the time (t) at which i(t) is the representative current ir, the first feature amount X1(t) and the second feature amount X2(t) at the time at which i(t) is the representative current ir are extracted from the output data and a temporary cell voltage fluctuation Ym (X1, X2) is calculated from them. Next, a mean I-V characteristic Vmean (t, ir) at the time (t) is calculated by subtracting Ym (X1, X2) from Vf (t, ir). Further, on the basis of Vmean (t, ir), it is determined whether or not the fuel cell is failed or deteriorated.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to a fault detection and degradation estimation device, and more particularly to a fault detection and degradation estimation device that can detect abnormalities or faults in fuel cells with high accuracy and at an earlier stage, or that can estimate the degree of degradation of a fuel cell. [Background technology]

[0002] A polymer electrolyte fuel cell (hereinafter simply referred to as a "fuel cell") is equipped with a membrane electrode assembly (MEA) in which catalyst layers containing electrode catalysts are bonded to both sides of an electrolyte membrane. The catalyst layer is the reaction field for electrode reactions and generally consists of a composite of carbon supporting catalyst particles such as platinum and a solid polymer electrolyte (catalyst layer ionomer). In a fuel cell, a gas diffusion layer is usually arranged on the outside of the catalyst layer. A current collector (separator) equipped with a gas flow path is also arranged on the outside of the gas diffusion layer. A fuel cell usually has a structure (fuel cell stack) in which multiple unit cells each consisting of such an MEA, gas diffusion layer, and current collector are stacked.

[0003] When a fuel cell is used as an onboard power source, the voltage of the fuel cell fluctuates greatly depending on the vehicle's driving conditions. When the fuel cell is under low load, the cathode catalyst is exposed to a high potential, which makes it easier for catalytic components to leach out of the cathode catalyst. On the other hand, when the fuel cell is under high load, the cathode catalyst is exposed to a low potential, which makes it easier for the leached catalytic components to re-deposit on the surface of the cathode catalyst. Therefore, when the cathode catalyst is repeatedly exposed to high and low potential states, it gradually deteriorates. When the cathode catalyst deteriorates, the cell voltage of the fuel cell also decreases.

[0004] However, there are cases where the cell voltage of a fuel cell drops due to causes other than the aging of the electrode catalyst. In such cases, if the cause of the drop in cell voltage can be detected early, it may be possible to reduce the costs required for fuel cell maintenance and inspection. For this reason, various proposals have been made regarding fuel cell fault detection.

[0005] For example, Patent Document 1 states: (a) calculating the effective electrode area of ​​the fuel cell so that the effective value of the fuel cell stack voltage in the fuel cell system matches the simulation result by the simulator; (b) When the effective electrode area is outside the normal range, it is determined to be abnormal. A fuel cell monitoring system is disclosed. The document states that such a system can significantly reduce the number of steps required for maintenance and accurately monitor the deterioration status inside the fuel cell power generation system.

[0006] The output of a fuel cell usually changes from moment to moment. Therefore, when the cell voltage measured under certain conditions drops, it is difficult to distinguish, based only on the change in the actual cell voltage, whether the drop in cell voltage is due to the aging of the electrode catalyst or to some other event (i.e., an abnormality or failure of the fuel cell).

[0007] In this regard, Patent Document 1 states that the presence or absence of an abnormality can be detected by determining whether the effective electrode area is outside the normal range. However, Patent Document 1 estimates the effective electrode area from the effective value of the cell voltage, and does not take into account temporary fluctuations in the cell voltage. As a result, a voltage drop due to temporary fluctuations in the cell voltage may be judged as a decrease in the effective electrode area, which could lead to a false determination that an abnormality has occurred, even though the fuel cell is normal. Furthermore, there has been no example of a device proposed to date that can estimate with high accuracy whether or not a fuel cell has malfunctioned and the degree of deterioration. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] Japanese Patent Application Laid-Open No. 2007-305327 Summary of the Invention [Problem to be solved by the invention]

[0009] The problem that the present invention aims to solve is to provide a fault detection and degradation estimation device that can detect abnormalities or failures in a fuel cell with high accuracy and at an earlier stage, even when the cell voltage repeatedly undergoes temporary fluctuations, or that can estimate the degree of degradation of a fuel cell with high accuracy. [Means for solving the problem]

[0010] In order to solve the above problems, a fault detection and degradation estimation device according to the present invention has the following configuration. (A) Fuel cell at time t (a) FC current i(t), (b) First cell voltage V1(t), (c) a first feature quantity X1(t) including one or more parameters selected from the group consisting of the second cell voltage V2(t) and parameters correlated therewith; and (d) A second feature quantity X2(t) including one or more parameters selected from the group consisting of the outside air temperature T(t) and parameters correlated therewith. and a first means for sequentially acquiring output data including the above and storing the same in a memory. Here, the "first cell voltage V1(t)" refers to the cell voltage of the fuel cell at all times when the fuel cell is in the intermittent operation off state. "Second cell voltage V2(t)" refers to the cell voltage of the fuel cell at all times. (B) From the output data, the FC current i(t) is the representative current i r The time t and the first cell voltage V1(t) are extracted, and the representative IV characteristic V f (t,i r ) and storing it in the memory. (C) From the output data, the FC current i(t) is determined to be the representative current i r The time t, the first feature amount X1(t), and the second feature amount X2(t) are extracted, and the temporary cell voltage fluctuation amount Y m a third means for calculating (X1, X2) and storing it in the memory; (D) The above V f (t,i r ) to the Y m By subtracting (X1, X2), the average IV characteristic V at time t is obtained. mean (t,i r and a fourth means for calculating the value of the first parameter and storing it in the memory. (E) The above V mean (t,i r A fifth means for determining whether the fuel cell has failed or deteriorated based on the result of the above. [Effects of the Invention]

[0011] First, the ever-changing FC current i(t), first cell voltage V1(t), first characteristic amount X1(t), and second characteristic amount X2(t) are acquired and stored in memory. Next, the FC current i(t) is the representative current i r When (for example, i r =0.2A / cm 2 When i(t) is i, the time t and V1(t) are read from the memory. r The change in cell voltage over time when f (t,i r ) is calculated. Similarly, the FC current i(t) is the representative current i r When time t, X1(t) and X2(t) are read from memory, and i(t) is i r The temporary cell voltage fluctuation Y m Calculate (X1,X2). Furthermore, the obtained V f (t,i r ) to Y m By subtracting (X1,X2), the average IV characteristic V mean (t,ir ) is calculated.

[0012] When no fault occurs, V f (t,i r ) includes the decrease in cell voltage due to the deterioration of the electrode catalyst over time and the temporary fluctuation in cell voltage Y m (X1, X2). Also, when the fuel cell fails, V f (t,i r ) is further increased by the drop in cell voltage caused by the failure. f (t,i r ) alone, it is difficult to diagnose the fault accurately.

[0013] In contrast, when there is no fault, V mean (t,i r ) includes only the decrease in cell voltage due to the aging of the electrode catalyst. Also, when a fuel cell failure occurs, V mean (t,i r ) is further increased by the drop in cell voltage caused by the failure. mean (t,i r ) changes, V mean (t,i r On the other hand, if no fault is detected, the presence or absence of a fault can be determined from the change in V mean (t,i r ) can be used to estimate the deterioration and IV performance of the fuel cell. [Brief explanation of the drawings]

[0014] [Figure 1] Figure 1(A) is a schematic diagram of the decrease in cell voltage due to the aging of the electrode catalyst, and Figure 1(B) is a schematic diagram of the actual decrease in cell voltage. [Figure 2]Figure 2(A) is a schematic diagram of the change in cell voltage when a failure occurs, in a case where the change in cell voltage includes only a decrease in cell voltage due to the degradation of the electrode catalyst over time. Figure 2(B) is a schematic diagram of the change in cell voltage when a failure occurs, in a case where the change in cell voltage includes both a decrease in cell voltage due to the degradation of the electrode catalyst over time and temporary fluctuations in cell voltage.

[0015] [Figure 3] FIG. 2 is a schematic diagram of a method for calculating the IV characteristics at an arbitrary time t from the running data of a vehicle equipped with a fuel cell. [Figure 4] Figure 4(A) shows an example of the change over time of the representative IV characteristics Vf(t,ir) when the representative current ir is 0.2 A / cm2. Figure 4(B) shows an example of the change over time of the representative IV characteristics Vf(t,ir) when the representative current ir is 0.4 A / cm2. [Figure 5] FIG. 1 is a schematic diagram of a method for calculating Ym(X1, X2) using a neural network.

[0016] [Figure 6] Figure 6(A) is a schematic diagram of a fault detection method using the rate of decrease dVmean(t,ir) / dt of Vmean(t,ir) at time t and the rate of decrease dVmean(tm,ir) / dt of Vmean(tm,ir) at time (tm). Figure 6(B) is a schematic diagram of a fault detection method using only the rate of decrease dVmean(t,ir) / dt of Vmean(t,ir) at time t. [Figure 7] FIG. 1 is a schematic diagram of a fault detection method using a non-fault model Vn(t,ir). DETAILED DESCRIPTION OF THE INVENTION

[0017] An embodiment of the present invention will be described in detail below. [1. Fault detection and degradation estimation methods] The following describes the fault detection and degradation estimation method according to the present invention, focusing on an FC system used in a vehicle. Figure 1(A) shows a schematic diagram of the decrease in cell voltage due to the aging of the electrode catalyst. Figure 1(B) shows a schematic diagram of the actual decrease in cell voltage.

[0018] The platinum catalyst used in an FC stack deteriorates as the vehicle travels, so ideally, the cell voltage V measured under certain conditions will decrease monotonically with the passage of driving time t, as shown in Figure 1(A). However, the cell voltage measured in an actual vehicle contains, as shown in Figure 1(B), not only a monotonically decreasing component due to the deterioration of the catalyst over time, but also a component Y in which the cell voltage temporarily fluctuates around this monotonically decreasing waveform.

[0019] Figure 2(A) shows a schematic diagram of the change in cell voltage when a failure occurs, when the change in cell voltage only includes a decrease in cell voltage due to the degradation of the electrode catalyst over time. Figure 2(B) shows a schematic diagram of the change in cell voltage when a failure occurs, when the change in cell voltage includes both a decrease in cell voltage due to the degradation of the electrode catalyst over time and temporary fluctuations in cell voltage. If the change in the measured cell voltage V contains only a monotonically decreasing component due to the degradation of the electrode catalyst over time, as shown in Figure 1(A), the slope of the monotonically decreasing curve, dV / dt, will change continuously over time. However, if a failure occurs at a certain time t, dV / dt will change discontinuously, as shown in Figure 2(A). Therefore, by continuously monitoring the change in dV / dt, it is possible to determine whether or not a failure has occurred based on the presence or absence of a discontinuous change in dV / dt.

[0020] However, the actually measured cell voltage contains a temporarily fluctuating component Y, as shown in Figure 1(B). Therefore, as shown in Figure 2(B), simply by monitoring the actually measured cell voltage, it is difficult to distinguish whether the fluctuation in the measured cell voltage is due to a cause other than a fault (temporary fluctuation in cell voltage) or a fault.

[0021] Therefore, in the present invention, the temporary fluctuation in cell voltage is estimated using various methods, and the decreasing component of the cell voltage (monotonically decreasing component) due to the deterioration of the electrode catalyst over time is separated by subtracting the temporary fluctuation component from the actually measured cell voltage. Next, the presence or absence of a fault is determined based on the change over time of the separated monotonically decreasing component. This is different from conventional methods. The separation method and determination method will be described in detail later.

[0022] [2. Fault detection and degradation estimation device] A fault detection and degradation estimation device according to the present invention has the following configuration.

[0023] [Configuration 1] A fault detection and degradation estimation device having the following configuration. (A) Fuel cell at time t (a) FC current i(t), (b) First cell voltage V1(t), (c) a first feature quantity X1(t) including one or more parameters selected from the group consisting of the second cell voltage V2(t) and parameters correlated therewith; and (d) A second feature quantity X2(t) including one or more parameters selected from the group consisting of the outside air temperature T(t) and parameters correlated therewith. and a first means for sequentially acquiring output data including the above and storing the same in a memory. Here, the "first cell voltage V1(t)" refers to the cell voltage of the fuel cell at all times when the fuel cell is in the intermittent operation off state. "Second cell voltage V2(t)" refers to the cell voltage of the fuel cell at all times. (B) From the output data, the FC current i(t) is the representative current i r The time t and the first cell voltage V1(t) are extracted, and the representative IV characteristic V f (t,i r ) and storing it in the memory. (C) From the output data, the FC current i(t) is determined to be the representative current i rThe time t, the first feature amount X1(t), and the second feature amount X2(t) are extracted, and the temporary cell voltage fluctuation amount Y m a third means for calculating (X1, X2) and storing it in the memory; (D) The above V f (t,i r ) to the Y m By subtracting (X1, X2), the average IV characteristic V at time t is obtained. mean (t,i r and a fourth means for calculating the value of the first parameter and storing it in the memory. (E) The above V mean (t,i r A fifth means for determining whether the fuel cell has failed or deteriorated based on the result of the above.

[0024] [Configuration 2] The failure detection and degradation estimation device according to configuration 1, wherein the parameter correlated with V2(t) includes one or more selected from the group consisting of an FC intermittent operation flag, FC power, air flow rate, air compressor motor rotation speed, hydrogen pump power consumption, hydrogen pump rotation speed, FC inlet pressure (anode), EV cooling water pump rotation speed, FC cooling water pump rotation speed, FC converter current, required power for the vehicle system, required current value for the vehicle system, FC inlet pressure (cathode), anode gas flow rate, air compressor power consumption, FC power generation state, and A / C power consumption.

[0025] [Configuration 3] 3. The fault detection and degradation estimation device according to configuration 1 or 2, wherein the parameter correlated with T(t) includes one or more selected from the group consisting of a radiator outlet coolant temperature, an air compressor motor temperature, an air compressor inverter temperature, a hydrogen pump motor temperature, an FC outlet coolant temperature, and temperatures of each auxiliary component.

[0026] [Configuration 4] The third means calculates the Y m 4. The fault detection and degradation estimation device according to any one of configurations 1 to 3, further comprising means for calculating (X1, X2). Y m (X1,X2)=a0+Σa 1p X1(t) p +Σa 2q X2(t) q …(2) however, p and q are integers equal to or greater than 1, a0, a 1p , and a 2q are the representative currents i r Coefficients identified for each

[0027] [Configuration 5] The third means calculates the Y m 5. The fault detection and degradation estimation device according to any one of configurations 1 to 4, further comprising means for calculating (X1, X2).

[0028] [Configuration 6] The fifth means is The above V mean (t,i r ) Decrease rate dV mean (t,i r ) / dt is calculated sequentially, A means for determining that the fuel cell has failed when any one or more of the following formulas (3.1) to (3.6) is satisfied: 6. The fault detection and degradation estimation device according to any one of configurations 1 to 5, comprising: |dV mean (t,i r ) / dt-dV mean (tm,i r ) / dt|>ε 31 …(3.1) |dV mean (t,i r ) / dt-dV mean (tm,i r ) / dt|≧ε 31 …(3.2) |[dV mean (t,i r ) / dt] / [dV mean (tm,i r ) / dt]|>ε 32 …(3.3) |[dV mean (t,i r ) / dt] / [dV mean (tm,i r ) / dt]|≧ε 32 …(3.4) |dV mean (t,i r ) / dt|>ε 33 …(3.5) |dV mean (t,i r ) / dt|≧ε 33 …(3.6) however, dV mean (tm,i r ) / dt is the average IV characteristic V at time (tm) (m>0) mean (tm,i r ) decrease rate, ε 31 , ε 32 , ε 33 are the thresholds for determining whether or not a fault has occurred.

[0029] [Configuration 7] The memory stores the FC current i(t) and the representative current i r Non-faulty model V when n (t,i r ) is stored, The failure detection and degradation estimation device according to any one of configurations 1 to 6, wherein the fifth means includes a means for determining that the fuel cell has failed when any one or more of the following formulas (4.1) to (4.4) are satisfied: |V mean (t,i r )-V n (t,i r )|>ε 41 …(4.1) |V mean (t,i r )-V n (t,i r )|≧ε 41 …(4.2) |V mean (t,i r ) / V n (t,i r )|<ε42 …(4.3) |V mean (t,i r ) / V n (t,i r )|≦ε 42 …(4.4) however, "Non-faulty Model V n (t,i r ) is a model showing the relationship between the usage history of a fuel cell having the same specifications as the fuel cell that is the subject of fault detection and degradation estimation and the decrease in cell voltage caused by the degradation of the electrode catalyst over time; ε 41 , ε 42 are the thresholds for determining whether or not a fault has occurred.

[0030] [Configuration 8] The above V n (t,i r )teeth, (a) a physical model, or (b) V from time zero to time t f (t,i r ) approximated by an rth degree polynomial (r is an integer between 1 and 10) 8. The fault detection and degradation estimation device according to configuration 7,

[0031] [2.1. First means] The first method is to calculate the fuel cell current at time t. (a) FC current i(t), (b) First cell voltage V1(t), (c) a first feature quantity X1(t) including one or more parameters selected from the group consisting of the second cell voltage V2(t) and parameters correlated therewith; and (d) A second feature quantity X2(t) including one or more parameters selected from the group consisting of the outside air temperature T(t) and parameters correlated therewith. and stores the output data in a memory.

[0032] [2.1.1. Fuel cell] In the present invention, the type of fuel cell is not particularly limited, and examples of fuel cells to which the present invention can be applied include polymer electrolyte fuel cells and solid oxide fuel cells.

[0033] [2.1.2. FC current i(t)] The FC current i(t) is used to select data to be used for fault detection and degradation estimation. Normally, even if the degradation state of the fuel cell is the same, if the FC current i(t) changes, the cell voltage V(t) will also change accordingly. Therefore, to accurately detect faults and estimate degradation, it is necessary to use the cell voltage V(t) when the FC current i(t) is at a specific value. In the present invention, the FC current i(t) is selected from the stored data as a specific value (representative current i r ) is extracted, and the extracted cell voltage V(t) is used to perform fault detection and degradation estimation. This will be described later.

[0034] [2.1.3. First cell voltage V1(t)] "First cell voltage V1(t)" refers to the cell voltage of the fuel cell at all times when the fuel cell is in the intermittent operation off state. V1(t) is the representative IV characteristic V f (t,i r ) In other words, V1(t) is used to estimate the degradation of the electrocatalyst over time.

[0035] For example, a fuel cell vehicle is usually equipped with a fuel cell and a secondary battery as a power source. In this case, the fuel cell does not generate power all the time, but alternates between a state in which power is being generated (intermittent operation off state) and a state in which power generation is stopped (intermittent operation on state) depending on the power required for the vehicle. Therefore, V includes not only the cell voltage when the fuel cell is generating power, but also the cell voltage when it is stopped. f (t,i r) does not allow accurate estimation of the decrease in cell voltage due to the aging of the electrode catalyst. f (t,i r ) is calculated using V1(t).

[0036] [2.1.4. First feature X1(t)] The "first characteristic amount X1(t)" refers to one or more parameters selected from the group consisting of the second cell voltage V2(t) and parameters correlated therewith. A "correlated parameter" refers to a parameter that can be regarded as the same as the second cell voltage V2(t), or a parameter that can be converted into the second cell voltage V2(t). The "second cell voltage V2(t)" refers to the cell voltage of the fuel cell at all times. That is, the second cell voltage V2(t) includes not only the cell voltage when the fuel cell is generating electricity (when intermittent operation is off), but also the cell voltage when the fuel cell is stopped (when intermittent operation is on). X1(t) is the temporary cell voltage fluctuation Y m It is one of the two parameters used to calculate (X1, X2).

[0037] One of the main causes of temporary cell voltage fluctuations is fluctuations in the power demanded from the fuel cell. Furthermore, fluctuations in the power demand are primarily manifested as fluctuations in V2(t). Therefore, V2(t) has a strong correlation with temporary cell voltage fluctuations. When V2(t) is used as X1(t), Y m (X1,X2) can be estimated with high accuracy.

[0038] In addition, Y m When estimating (X1, X2), another parameter correlated with V2(t) may be used as the first feature amount X1(t) instead of or in addition to V2(t). Specific examples of such parameters include the following: X1(t) may use any one of the following parameters, or may use a combination of two or more parameters.

[0039] (a) Flag indicating the intermittent operation state of the fuel cell (FC intermittent operation flag). (b) Electric power generated by the fuel cell (FC power). (c) The flow rate of air supplied to the cathode of the fuel cell (air flow rate). (d) Motor speed of the air compressor that supplies air to the cathode of the fuel cell. (e) Power consumption of the hydrogen pump to supply hydrogen to the fuel cell anode. (f) The rotation speed of the hydrogen pump that supplies hydrogen to the anode of the fuel cell. (g) Pressure of hydrogen supplied to the fuel cell anode (FC inlet pressure (anode)).

[0040] (h) The rotation speed of the pump that supplies cooling water to the drive system, such as the motor and inverter (EV cooling water pump rotation speed). (i) The rotational speed of the cooling water pump for supplying cooling water to the fuel cell (the rotational speed of the FC cooling water pump). (j) FC converter current. (k) Power requirements for vehicle systems. (l) Current requirements for vehicle systems.

[0041] (m) Pressure of air supplied to the cathode of the fuel cell (FC inlet pressure (cathode)). (n) The flow rate of hydrogen supplied to the anode of the fuel cell (anode gas flow rate). (o) Power consumption of the air compressor to supply air to the cathode of the fuel cell. (p) FC power generation status (power generation, non-power generation, startup, completion sequence, etc.). (q) Air conditioning power consumption (A / C power consumption).

[0042] For example, if X1(t) is the FC intermittent operation flag, the flag is set to "1" when the fuel cell is in operation, and to "0" when the fuel cell is stopped. In this case, if fluctuations in the FC intermittent operation flag over time are averaged, the averaged fluctuations in the FC intermittent operation flag have a strong correlation with fluctuations in V2(t). Therefore, even if the averaged FC intermittent operation flag is used instead of V2(t), Y m (X1,X2) can be estimated with high accuracy.

[0043] The same applies to FC power, air flow rate, etc., and these parameters have a positive correlation with V2(t). In other words, basically, the larger V2(t) becomes, the larger these parameters become. Therefore, even if these parameters are used as X1(t) instead of V2(t), Y m (X1,X2) can be estimated with high accuracy.

[0044] [2.1.5. Second feature X2(t)] The "second feature amount X2(t)" refers to one or more parameters selected from the group consisting of the outside air temperature T(t) and parameters correlated therewith. A "correlated parameter" refers to a parameter that can be regarded as equivalent to the outside air temperature T(t) or a parameter that can be converted to the outside air temperature T(t). X2(t) is the temporary cell voltage fluctuation Y m This is the other of the two parameters used to calculate (X1, X2).

[0045] Temporary fluctuations in cell voltage are caused not only by fluctuations in the power demanded from the fuel cell, but also by fluctuations in the outside air temperature T(t). Fluctuations in T(t) affect not only the power generation efficiency of the fuel cell, but also the power consumption of the fuel cell's auxiliary equipment. Therefore, T(t) has a strong correlation with temporary fluctuations in cell voltage. If T(t) is used as X2(t), then Y m (X1,X2) can be estimated with high accuracy.

[0046] In addition, Y m When estimating (X1, X2), another parameter correlated with T(t) may be used as the second feature amount X2(t) instead of or in addition to T(t). Specific examples of such parameters include the following: For X2(t), any one of the following parameters may be used, or two or more parameters may be used in combination.

[0047] (a) Coolant temperature at the radiator outlet for cooling the fuel cell. (b) Motor temperature of the air compressor that supplies air to the cathode of the fuel cell. (c) inverter temperature of the air compressor that supplies air to the cathode of the fuel cell; (d) Motor temperature of the hydrogen pump that supplies hydrogen to the anode of the fuel cell. (e) Temperature of the cooling water discharged from the fuel cell (FC outlet cooling water temperature). (f) Temperature of each auxiliary component other than those mentioned above.

[0048] For example, if X2(t) is the coolant temperature at the outlet of the radiator for cooling the fuel cell, fluctuations in the coolant temperature are strongly correlated with fluctuations in T(t). Generally, the higher T(t) is, the higher the coolant temperature is. Therefore, even if the coolant temperature is used instead of T(t), Y m (X1,X2) can be estimated with high accuracy.

[0049] The same is true for the air compressor motor temperature, the air compressor inverter temperature, etc., and these parameters have a positive correlation with T(t). In other words, the higher T(t) is, the larger these parameters become. Therefore, even if these parameters are used as X2(t) instead of T(t), Y m (X1,X2) can be estimated with high accuracy.

[0050] [2.2. Second means] The second method is to determine whether the FC current i(t) is a representative current i r The time t and the first cell voltage V1(t) are extracted, and the representative IV characteristic V f (t,i r ) and store it in memory.

[0051] "Representative IV characteristics V f (t,i r )" means that the FC current i(t) is the representative current i r This refers to a group of data that shows the relationship between time t and cell voltage V(t) when "Representative current i r " refers to the FC current i(t) that serves as the reference when comparing the cell voltage V(t) when there is no fault with the cell voltage V(t) when there is a fault when fault detection and degradation estimation are performed.

[0052] Representative current i r may be one, or may be two or more. In other words, fault detection is (a) Any one of the representative currents i r (e.g., i r =0.2A / cm 2 When f (t,i r ) may be judged to be a failure when the following conditions are met, or (b) Two or more representative currents i r (e.g., i r =0.2A / cm 2 , 0.4A / cm 2 , and 0.6A / cm 2 When f (t,i r ) may be judged to be faulty when they simultaneously satisfy the requirements described below.

[0053] [2.2.1. Calculation of IV characteristics at any time t] Figure 3 shows a schematic diagram of a method for calculating the IV characteristics at any time t from driving data of a vehicle equipped with a fuel cell. After a fuel cell vehicle is completed, when the vehicle is actually driven, the cell voltage V(t), FC current i(t), etc. change from moment to moment depending on the driving conditions. These are stored sequentially in memory as vehicle measurement data.

[0054] When a vehicle is actually driven, the fuel cell's power generation conditions vary widely, from light load (low current, high voltage) to heavy load (high current, low voltage). Meanwhile, the electrode catalyst gradually deteriorates over time. Therefore, data on the electrode catalyst deterioration over time within a certain time interval (time t ± α) can be considered "data at a certain time t." Therefore, by extracting various data consisting of combinations of FC current i(t) and first cell voltage V1(t) from the data within time (t±α), the IV characteristics at a certain time t can be obtained. The value of α is not particularly limited, and an optimal value can be selected depending on the purpose. The value of α is usually about 1 to 10 hours.

[0055] The upper diagram in Figure 3 shows an example of the IV characteristics when the running time is 0 h (i.e., when the running time can be considered to be 0 h). Each "○" represents extracted data. In the example shown in Figure 3, when the FC current i(t) is 0.2, 0.4, 0.6, 0.8, and 1.0 A / cm 2 The figure shows the first cell voltage V1(t) when the running time is 0 h and the current is any current i(t). By approximating this data with an appropriate function, it is possible to calculate the first cell voltage V1(t) when the running time is 0 h and the current is any current i(t). For example, if these data are approximated with a cubic function, the IV characteristics when the running time is 0 h can be expressed by the following equation (1.1).

[0056] V f1 (t,i)=a1i 3 +b1i 2 +c1i+d1…(1.1) however, V f1(t, i) is the first cell voltage when the running time is 0 h and the FC current is i(t), a1, b1, c1, and d1 are coefficients identified for each travel time.

[0057] The middle and bottom panels of Figure 3 show examples of IV characteristics when the driving time is 3000 hours and 5000 hours, respectively. The same applies when the driving time is 3000 hours or 5000 hours; simply extract various data from the accumulated data consisting of combinations of FC current i(t) and first cell voltage V1(t) that can be considered to be driving times of 3000 hours or 5000 hours, and approximate these with an appropriate function. For example, if this data is approximated with a cubic function, the IV characteristics when the driving time is 3000 hours or 5000 hours can be expressed by the following equation (1.2) or equation (1.3), respectively.

[0058] V f2 (t,i)=a2i 3 +b2i 2 +c2i+d2…(1.2) however, V f2 (t, i) is the first cell voltage when the running time is 3000 h and the FC current is i(t), a2, b2, c2, and d2 are coefficients identified for each travel time.

[0059] V f3 (t,i)=a3i 3 +b3i 2 +c3i+d3…(1.3) however, V f3 (t, i) is the first cell voltage when the running time is 5000 h and the FC current is i(t), a3, b3, c3, and d3 are coefficients identified for each travel time.

[0060] Generally, the longer the driving time, the more the electrode catalyst deteriorates over time. f2 (t,i) is usually V f1 (t,i) is lower than Vf3 (t,i) is usually V f2 The value will be lower than (t,i).

[0061] [2.2.2. Calculation of representative IV characteristics] If the IV characteristics at any time t are known, the FC current i(t) can be calculated as the representative current i r The time t and the first cell voltage V1(t) when f (t,i r ) can be calculated.

[0062] Figure 4(A) shows the representative current i r =0.2A / cm 2 Representative IV characteristics when V f (t,i r ) over time. Figure 4(B) shows an example of the change in the representative current i r =0.4A / cm 2 Representative IV characteristics when V f (t,i r ) over time is shown below. Figures 4(A) and 4(B) show the results of the data shown in Figure 3 when the FC current i(t) was 0.2 A / cm 2 or 0.4A / cm 2 The first cell voltage V1(t) at time t is extracted and V1(t) is plotted against time t.

[0063] Typical IV characteristics V f (t,i r ) is the temporary fluctuation of the cell voltage Y m (X1,X2) and other components V mean (t,i r ) and V mean (t,i r ) usually consists of only a drop in cell voltage due to the degradation of the electrode catalyst over time. However, when a fuel cell fails, V mean (t,i r ) is further increased by the drop in cell voltage due to a fault. Therefore, in order to accurately determine the fault, V f(t,i r ) to Y m Subtract (X1,X2) and V mean (t,i r ) must be separated. To do this, first m You need to know (X1,X2).

[0064] [2.3. Third means] The third method is to determine whether the FC current i(t) is a representative current i r The time t, the first feature value X1(t), and the second feature value X2(t) are extracted, and the temporary cell voltage fluctuation Y m This is a means of calculating (X1, X2) and storing it in memory.

[0065] For fuel cells with the same specifications as the fuel cell to be controlled, m A function representing the relationship between X1 and X2 or an equivalent function is obtained and stored in memory. In actual fault detection, first, from the accumulated data, the FC current i(t) is selected as the representative current i r When (a) Time t, (b) a first feature quantity X1(t) including the second cell voltage V2(t) and / or a parameter equivalent thereto; and (c) A second feature quantity X2(t) including the outside air temperature T(t) and / or a parameter equivalent thereto Extract. Next, X1(t) and X2(t) are substituted into a previously obtained function or its equivalent, and Y m Calculate (X1,X2).

[0066] Y m The method for calculating (X1, X2) is not particularly limited, and the most suitable method can be selected depending on the purpose. m Specific methods for calculating (X1, X2) include the following.

[0067] [2.3.1. Method using multiple regression model] The third method is to calculate Y using the following formula (2): m It may also include means for calculating (X1, X2). Y m (X1,X2)=a0+Σa 1p X1(t) p +Σa 2q X2(t) q …(2) however, p and q are integers equal to or greater than 1, a0, a 1p , and a 2q are the representative currents i r Coefficients identified for each

[0068] Analysis of vehicle measurement data by the inventors of the present application revealed that the temporary cell voltage fluctuation Y is highly correlated with the first feature amount X1(t) and the second feature amount X2(t). Based on this knowledge, formula (2) is m This is a multiple regression model that approximates (the estimated value of Y) by the sum of a pth-order polynomial of X1(t) and a qth-order polynomial of X2(t).

[0069] Using equation (2), Y m When calculating (X1, X2), a test is conducted in advance on a fuel cell with the same specifications as the fuel cell being the target of fault detection, and the actual cell voltage fluctuation Y when X1(t) and X2(t) fluctuate is calculated as the fluctuation Y in the model formula expressed by equation (2). m The orders p and q of X1(t) and X2(t) and the coefficients a0, a 1p , and a 2q Each coefficient is determined by the representative current i r Furthermore, these identified coefficients are stored in memory. In actual fault detection, the acquired X1(t) and X2(t) are substituted into equation (2). This allows the FC current i(t) to be calculated as the representative current i r Y when m (X1,X2) can be calculated.

[0070] [2.3.2. Method using neural networks] The third means is to use a neural network to calculate the Y m It may also include means for calculating (X1, X2). For a fuel cell with the same specifications as the fuel cell being the target of fault detection, we used a neural network to calculate Y m The relationship between X1 and X2 is learned and a prediction model is constructed. In actual fault detection, the obtained X1(t) and X2(t) are input to the constructed prediction model. This allows the FC current i(t) to be calculated as the representative current i r Y when m (X1,X2) can be output.

[0071] Figure 5 shows the Y m A schematic diagram of the calculation method for (X1, X2) is shown in Figure 5. This is an example of a neural network, and Y m This is a model that predicts the following.

[0072] [2.4. Fourth means] The fourth method is V f (t,i r ) to Y m By subtracting (X1,X2), the average IV characteristic V at time t is obtained. mean (t,i r ) and store it in memory.

[0073] "Average IV characteristic V mean (t,i r )" means that the FC current i(t) is the representative current i r This refers to a group of data that shows the relationship between time t and cell voltage V, excluding temporary fluctuations in cell voltage, when When no fault occurs, V mean (t,i r ) ideally includes only the decrease in cell voltage due to the aging of the electrode catalyst. However, in the absence of a fault, V mean (t,i r) is the sum of the decrease in cell voltage due to the deterioration of the electrode catalyst over time and the decrease in cell voltage due to a failure. mean (t,i r ) can be used to accurately detect faults.

[0074] [2.5. Fifth means] The fifth method is V mean (t,i r ) is a means for determining whether the fuel cell has failed or deteriorated. In the present invention, V mean (t,i r The fault detection method using the ) is not particularly limited, and the most suitable method can be selected depending on the purpose. Specific fault detection methods include the following. Any one of these may be used, or two or more may be used in combination as far as physically possible.

[0075] 2.5.1. dV mean (t,i r ) / dt based fault detection method] The fifth measure is V mean (t,i r ) Decrease rate dV mean (t,i r ) / dt is calculated sequentially, A means for determining that the fuel cell has failed when any one or more of the following formulas (3.1) to (3.6) is satisfied: It may also include:

[0076] |dV mean (t,i r ) / dt-dV mean (tm,i r ) / dt|>ε 31 …(3.1) |dV mean (t,i r ) / dt-dV mean (tm,i r ) / dt|≧ε 31 …(3.2) |[dV mean (t,ir ) / dt] / [dV mean (tm,i r ) / dt]|>ε 32 …(3.3) |[dV mean (t,i r ) / dt] / [dV mean (tm,i r ) / dt]|≧ε 32 …(3.4) |dV mean (t,i r ) / dt|>ε 33 …(3.5) |dV mean (t,i r ) / dt|≧ε 33 …(3.6) however, dV mean (tm,i r ) / dt is the average IV characteristic V at time (tm) (m>0) mean (tm,i r ) decrease rate, ε 31 , ε 32 , ε 33 are the thresholds for determining whether or not a fault has occurred.

[0077] "tm" represents the time when past data was acquired to compare with the current data acquired at time t. The value of m is not particularly limited, and an optimal value can be selected depending on the purpose.

[0078] Figure 6(A) shows the V mean (t,i r ) Decrease rate dV mean (t,i r ) / dt and V at time (tm) mean (tm,i r ) Decrease rate dV mean (tm,i r ) / dt at time t. mean (t,i r ) Decrease rate dV mean (t,i r) / dt. As mentioned above, if a fault occurs, V mean (t,i r ) is added with the decrease in cell voltage due to the fault, so V mean (t,i r ) is smaller than the value expected based solely on the aging of the electrode catalyst. As a result, as shown in Figure 6, dV mean (t,i r ) / dt becomes more negative than would be expected based solely on the degradation of the electrocatalyst over time.

[0079] Therefore, as shown in equation (3.1) or equation (3.2), dV at the current time t mean (t,i r ) / dt and dV at a certain time in the past (tm) mean (tm,i r ) / dt and the absolute value of the difference between them is the threshold ε 31 When it exceeds ε 31 If this is the case, it can be determined that a failure has occurred (FIG. 6(A)). Alternatively, as shown in equation (3.3) or equation (3.4), dV mean (tm,i r ) / dt vs. dV mean (t,i r ) / dt ratio has a certain absolute value, ε 32 When it exceeds ε 31 If this is the case, it can be determined that a failure has occurred (FIG. 6(A)). Alternatively, as shown in equation (3.5) or equation (3.6), dV mean (t,i r ) / dt has a certain absolute value, ε 33 When it exceeds ε 33 If this is the case, it can be determined that a failure has occurred (FIG. 6(B)).

[0080] On the other hand, if no fault is detected, V mean (t,i r) represents an estimated value of the IV characteristics that takes into account the decrease in cell voltage due to the degradation of the electrode catalyst over time. mean (t,i r ) makes it possible to estimate deterioration, predict fuel consumption, predict cruising range, etc.

[0081] 2.5.2. Non-faulty model V n (t,i r ) Fault detection method] The memory stores the FC current i(t) as the representative current i r Non-faulty model V when n (t,i r ) may be stored. In this case, the fifth means may include means for determining that the fuel cell has failed when any one or more of the following formulas (4.1) to (4.4) is satisfied.

[0082] |V mean (t,i r )-V n (t,i r )|>ε 41 …(4.1) |V mean (t,i r )-V n (t,i r )|≧ε 41 …(4.2) |V mean (t,i r ) / V n (t,i r )|<ε 42 …(4.3) |V mean (t,i r ) / V n (t,i r )|≦ε 42 …(4.4) However, ε 41 , ε 42 are the thresholds for determining whether or not a fault has occurred.

[0083] [A. Non-faulty Model V n (t,i r )] "Non-faulty Model V n(t,i r )" refers to a model that shows the relationship between the usage history of a fuel cell with the same specifications as the fuel cell being the target of fault detection and the decrease in cell voltage caused by the deterioration of the electrode catalyst over time. In other words, the non-fault model V n (t,i r ) refers to a model that shows the relationship between usage history and cell voltage when it is assumed that no failures have occurred.

[0084] V n (t,i r ) is not particularly limited as long as it can derive the relationship between the usage history and the cell voltage. n (t,i r ) for example, (a) Physical model, (b) V from time zero to time t f (t,i r ) approximated by an rth degree polynomial (r is an integer between 1 and 10) etc.

[0085] A.1. Physical Model V n (t,i r ) may be a physical model. A "physical model" is a model that can estimate the time-dependent deterioration of the electrode catalyst using a theoretical formula and estimate the cell voltage V(t) at time t based on the estimated time-dependent deterioration of the electrode catalyst. For example, Reference 1 discloses a method for predicting fuel cell catalyst degradation. Using the method described in this document, it is possible to estimate the platinum catalyst surface area ECSA(t) at time t when the cell voltage V(t), FC current i(t), and cell temperature T(t) at time t are given. Furthermore, by substituting the obtained ECSA(t) into the following equation (5), it is possible to determine whether the FC current i(t) is equal to the representative current i at any time t. r When the cell voltage V is not faulty, n (t,i r ) can be calculated. [Reference 1] JP 2010-236989 A

[0086] Vn (t,i r )=V(0,i r )+K·ln{ECSA(t) / ECSA(0)} …(5) however, V n (t,i r ) is the cell voltage [V] at time t when there is no fault, i r is the representative current [A / cm 2 ] t is the time [s], K is a constant.

[0087] [A.2. r-degree polynomial] V n (t,i r ) is V from time zero to time t f (t,i r ) may be a model obtained by approximating it with an r-th degree polynomial (r is an integer between 1 and 10). V, including temporary cell voltage fluctuations f (t,i r ) with an rth order polynomial (a relatively low order polynomial), an average IV characteristic can be obtained in which temporary fluctuations in cell voltage are canceled out. When no faults occur, the IV characteristic approximated with an rth order polynomial does not include the drop in cell voltage caused by faults. mean (t,i r ) will have a similar shape.

[0088] However, when a failure occurs, the IV characteristics approximated by the rth order polynomial are calculated using data that includes both the IV characteristics before and after the failure. Therefore, the rth order polynomial does not easily reflect the drop in cell voltage caused by the failure immediately after the failure occurs. On the other hand, V mean (t,i r ) immediately reflects the drop in cell voltage due to a fault. Therefore, the IV characteristics approximated by an rth-order polynomial and V mean (t,i r ) to determine whether or not there is a malfunction.

[0089] A.3. Detection Methods Figure 7 shows the non-faulty model Vn (t,i r ) is shown in Figure 7. When a fault occurs, V mean (t,i r ) is V n (t,i r ) is lower than that. Therefore, as shown in equation (4.1) or equation (4.2), V at the current time t mean (t,i r ) and V n (t,i r ) and the absolute value of the difference between 41 When it exceeds ε 41 If this is the case, it can be determined that a failure has occurred. Alternatively, as shown in equation (4.3) or equation (4.4), V n (t,i r ) / dt vs. V mean (t,i r ) the absolute value of the ratio is a threshold ε 42 When it is less than ε 42 If the following is true, it can be determined that a fault has occurred.

[0090] On the other hand, if no fault is detected, V mean (t,i r ) as well as V n (t,i r ) also represents an estimated value of the IV characteristics that takes into account the decrease in cell voltage due to the aging of the electrode catalyst. mean (t,i r ) or V n (t,i r ) makes it possible to estimate deterioration, predict fuel consumption, predict cruising range, etc.

[0091] [4. Effect] First, the ever-changing FC current i(t), first cell voltage V1(t), first characteristic amount X1(t), and second characteristic amount X2(t) are acquired and stored in memory. Next, the FC current i(t) is the representative current i r When (for example, i r =0.2A / cm 2When i(t) is i, the time t and V1(t) are read from the memory. r The change in cell voltage over time when f (t,i r ) is calculated. Similarly, the FC current i(t) is the representative current i r When time t, X1(t) and X2(t) are read from memory, and i(t) is i r The temporary cell voltage fluctuation Y m Calculate (X1,X2). Furthermore, the obtained V f (t,i r ) to Y m By subtracting (X1,X2), the average IV characteristic V mean (t,i r ) is calculated.

[0092] When no fault occurs, V f (t,i r ) includes the decrease in cell voltage due to the deterioration of the electrode catalyst over time and the temporary fluctuation in cell voltage Y m (X1, X2). Also, when the fuel cell fails, V f (t,i r ) is further increased by the drop in cell voltage caused by the failure. f (t,i r ) alone, it is difficult to diagnose the fault accurately.

[0093] In contrast, when there is no fault, V mean (t,i r ) includes only the decrease in cell voltage due to the aging of the electrode catalyst. Also, when a fuel cell failure occurs, V mean (t,i r ) is further increased by the drop in cell voltage caused by the failure. mean (t,i r ) changes, V mean (t,i r On the other hand, if no fault is detected, the presence or absence of a fault can be determined from the change in V mean(t,i r ) can be used to estimate the deterioration and IV performance of the fuel cell.

[0094] Although the embodiments of the present invention have been described in detail above, the present invention is not limited to the above embodiments, and various modifications are possible within the scope of the gist of the present invention. [Industrial Applicability]

[0095] The failure detection and degradation estimation device according to the present invention can be used for determining failures, estimating degradation, predicting fuel consumption, predicting cruising distance, etc. of a fuel cell in a vehicle equipped with a fuel cell.

Claims

1. A fault detection and degradation estimation device having the following configuration. (A) The fuel cell at time t (a) FC current i(t), (b) First cell voltage V 1 (t), (c) Second cell voltage V 2 (t), and a first feature quantity X including one or more parameters selected from a group consisting of parameters correlated therewith. 1 (t), and (d) a second feature quantity X including one or more parameters selected from the group consisting of the outside air temperature T(t) and parameters correlated therewith; 2 (t) and a first means for sequentially acquiring output data including the above and storing the same in a memory. However, "first cell voltage V 1 "(t)" refers to the cell voltage of the fuel cell at all times when the fuel cell is in the intermittent operation off state. "Second cell voltage V 2 "(t)" refers to the cell voltage of the fuel cell at all times. (B) From the output data, the FC current i(t) is the representative current i r the time t and the first cell voltage V 1 (t) are extracted, and the representative IV characteristics V f (t,i r ) and storing it in the memory. (C) From the output data, the FC current i(t) is determined to be the representative current i r the time t when 1 (t), and the second feature amount X 2 (t) are extracted and used to calculate the temporary cell voltage fluctuation Y m (X 1 ,X 2 and a third means for calculating and storing the calculated value in the memory. (D) the V f (t,i r ) to the above Y m (X 1 ,X 2 ) to obtain the average IV characteristic V at the time t. mean (t,i r and a fourth means for calculating and storing the calculated value in the memory. (E) The V mean (t,i r A fifth means for determining whether the fuel cell has failed or deteriorated based on the result of the above.

2. The V 2 2. The failure detection and degradation estimation device according to claim 1, wherein the parameter correlated with (t) includes one or more selected from the group consisting of an FC intermittent operation flag, FC power, air flow rate, air compressor motor rotation speed, hydrogen pump power consumption, hydrogen pump rotation speed, FC inlet pressure (anode), EV cooling water pump rotation speed, FC cooling water pump rotation speed, FC converter current, required power for the vehicle system, required current value for the vehicle system, FC inlet pressure (cathode), anode gas flow rate, air compressor power consumption, FC power generation state, and A / C power consumption.

3. 2. The fault detection and degradation estimation device according to claim 1, wherein the parameters correlated with T(t) include one or more selected from the group consisting of a radiator outlet coolant temperature, an air compressor motor temperature, an air compressor inverter temperature, a hydrogen pump motor temperature, an FC outlet coolant temperature, and temperatures of each auxiliary component.

4. The third means calculates the Y m (X 1 ,X 2 2. The fault detection and degradation estimation device according to claim 1, further comprising means for calculating Y m (X 1 ,X 2 )=a 0 +Σa 1p X 1 (t) p +Σa 2q X 2 (t) q …(2) however, p and q are each an integer of 1 or more, a 0 , a 1p , and a 2q are the representative currents i r Coefficients identified for each

5. The third means uses a neural network to calculate the Y m (X 1 ,X 2 2. The fault detection and degradation estimation device according to claim 1, further comprising means for calculating

6. The fifth means is Foreword V mean (t,i r )のdecrease speed dV mean (t,i r ) / dtをCalculate it one by one, A means for determining that the fuel cell has failed when any one or more of the following formulas (3.1) to (3.6) is satisfied: The fault detection and degradation estimation device according to claim 1 , comprising: |dV mean (t,i r ) / dt-dV mean (t-m,i r ) / dt|>ε 31 …(3.1) |dV mean (t,i r ) / dt-dV mean (t-m,i r ) / dt|≧ε 31 …(3.2) |[dV mean (t,i r ) / dt] / [dV mean (t-m,i r ) / dt]|>ε 32 …(3.3) |[dV mean (t,i r ) / dt] / [dV mean (t-m,i r ) / dt]|≧ε 32 …(3.4) |dV mean (t,i r ) / dt|>ε 33 …(3.5) |dV mean (t,i r ) / dt|≧ε 33 …(3.6) however, dV mean (t-m,i r ) / dt is the average IV characteristic V at time (t-m) (m>0) mean (t-m,i r ) decrease rate, ε 31 , ε 32 , ε 33 are the thresholds for determining whether or not a fault has occurred.

7. The memory stores the FC current i(t) and the representative current i r A non-faulty model V when n (t,i r ) is stored, 2. The failure detection and degradation estimation device according to claim 1, wherein the fifth means includes means for determining that the fuel cell has failed when any one or more of the following formulas (4.1) to (4.4) is satisfied: |V mean (t,i r )-V n (t,i r )|>ε 41 …(4.1) |V mean (t,i r )-V n (t,i r )|≧ε 41 …(4.2) |V mean (t,i r ) / V n (t,i r )|<ε 42 …(4.3) |V mean (t,i r ) / V n (t,i r )|≦ε 42 …(4.4) however, "Non-faulty model V n (t,i r ) is a model showing the relationship between the usage history of a fuel cell having the same specifications as the fuel cell that is the subject of fault detection and degradation estimation and the decrease in cell voltage caused by the degradation of the electrode catalyst over time; ε 41 , ε 42 are the thresholds for determining whether or not a fault has occurred.

8. The V n (t,i r )teeth, (a) a physical model, or (b) the V from time zero to the time t f (t,i r ) is approximated by an rth degree polynomial (r is an integer between 1 and 10) 8. The fault detection and degradation estimation device according to claim 7, wherein:

Images