Fuel cell system

Abstract

Even when the time available for stable acquisition of operating data from the fuel cell module is relatively short, the system can detect abnormalities in the fuel cell module. [Solution] The fuel cell system 1 comprises an acquisition unit 3 that acquires operating data of the fuel cell module FCM at regular intervals Tc1, and a detection unit 4 that uses the operating data acquired by the acquisition unit 3 during a regular interval Tc2 to create a histogram on a two-dimensional coordinate system where the horizontal axis represents the size of the operating data and the vertical axis represents the length of the time the operating data was acquired, and detects that an abnormality has occurred in the fuel cell module FCM based on an index showing the change in the histogram.

Description

Technical Field

[0001] The present invention relates to a fuel cell system including a fuel cell module.

Background Art

[0002] As a fuel cell system, when the variation range of operation data acquired at a time when the operation data of a fuel cell module can be stably acquired becomes equal to or greater than a threshold value, it is detected that an abnormality has occurred in the fuel cell module. As a related technique, there is Patent Document 1.

[0003] However, in the above fuel cell system, when the time during which the operation data can be stably acquired is relatively short, although an abnormality has occurred in the fuel cell module, the variation range of the operation data does not become equal to or greater than the threshold value, and there is a possibility that the abnormality cannot be detected.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] An object according to one aspect of the present invention is to provide a fuel cell system capable of detecting that an abnormality has occurred in a fuel cell module even when the time during which the operation data of the fuel cell module can be stably acquired is relatively short.

Means for Solving the Problems

[0006] One embodiment of the present invention is a fuel cell system comprising a fuel cell module, the system comprising: an acquisition unit that acquires operational data of the fuel cell module at first fixed time intervals; and a detection unit that uses the operational data acquired by the acquisition unit during a second fixed time interval which is longer than the first fixed time interval to create a histogram on a two-dimensional coordinate system where the horizontal axis is the magnitude of the operational data and the vertical axis is the length of the acquisition time of the operational data, and detects that an abnormality has occurred in the fuel cell module based on an index showing the change in the histogram.

[0007] This allows for the acquisition of a second fixed time period sufficient to change the index indicating histogram changes in response to an abnormality in the fuel cell module. As a result, even if the time available for stable acquisition of operational data is relatively short, it is possible to detect an abnormality in the fuel cell module based on the index indicating histogram changes.

[0008] Furthermore, the fuel cell system may be configured such that the fuel cell module comprises a fuel cell stack having a plurality of fuel cell cells, the acquisition unit acquires the voltage of each of the plurality of fuel cell cells at the first fixed time intervals, and the detection unit uses the voltages acquired by the acquisition unit during the second fixed time interval to create a histogram on a two-dimensional coordinate system with the horizontal axis representing the magnitude of the voltage and the vertical axis representing the length of the voltage acquisition time, and detects that an abnormality has occurred in at least one of the plurality of fuel cell cells when the average value of the histogram falls below a threshold.

[0009] Alternatively, the detection unit may be configured to detect that an abnormality has occurred in at least one of the plurality of fuel cell cells when the median value of the histogram falls below a threshold.

[0010] Alternatively, the detection unit may be configured to detect that an abnormality has occurred in at least one of the plurality of fuel cell cells when the standard deviation of the histogram exceeds a threshold.

[0011] Alternatively, the detection unit may be configured to detect that an abnormality has occurred in at least one of the plurality of fuel cell cells when the ratio of the acquisition time corresponding to a voltage below a predetermined voltage in the histogram to the total acquisition time of the histogram exceeds a threshold.

[0012] Alternatively, the detection unit may be configured to create a first histogram on a two-dimensional coordinate system with the horizontal axis representing the voltage magnitude and the vertical axis representing the voltage acquisition time, using the lowest voltage acquired during the second fixed time period from among the voltages acquired by the acquisition unit, and to create a second histogram on the same two-dimensional coordinate system with the horizontal axis representing the voltage magnitude and the vertical axis representing the voltage acquisition time, using the maximum voltage acquired during the second fixed time period from among the voltages acquired by the acquisition unit, and to detect that an abnormality has occurred in at least one of the plurality of fuel cell cells when the dissimilarity between the shapes of the first histogram and the second histogram exceeds a threshold.

[0013] Alternatively, the detection unit may be configured to detect that an abnormality has occurred in at least one of the plurality of fuel cell cells when the degree of dissimilarity between the shape of the histogram created at the first time and the shape of the histogram created at a second time after the first time exceeds a threshold.

[0014] Furthermore, the detection unit may be configured to determine that an abnormality has occurred in the fuel cell cell in which the ratio of the total acquisition time corresponding to the lowest voltage to the total acquisition time of all acquisition times in the histogram is the largest when it detects that an abnormality has occurred in at least one of the plurality of fuel cell cells. [Effects of the Invention]

[0015] According to the present invention, even when the time for stably acquiring the operation data of the fuel cell module is relatively short, it is possible to detect that an abnormality has occurred in the fuel cell module.

Brief Description of the Drawings

[0016] [Figure 1] It is a diagram showing an example of a fuel cell system according to an embodiment. [Figure 2] It is a diagram showing an example of the voltage of a fuel cell during intermittent operation. [Figure 3] It is a flowchart showing an example of the operation of the acquisition unit. [Figure 4] It is a diagram showing an example of a database stored in the storage device. [Figure 5] It is a diagram showing an example of a histogram. [Figure 6] It is a flowchart showing an example of the operation of the detection unit. [Figure 7] It is a diagram showing an example of an index indicating a change in a histogram and an example of a ratio of the total acquisition time corresponding to the minimum voltage to the total acquisition time for all acquisition times in the histogram. [Figure 8] It is a diagram showing an example of a histogram created using the minimum voltage and an example of a histogram created using the maximum voltage.

Embodiments for Carrying Out the Invention

[0017] Hereinafter, embodiments will be described in detail based on the drawings.

[0018] FIG. 1 is a diagram showing an example of a fuel cell system according to an embodiment.

[0019] The fuel cell system 1 shown in FIG. 1 includes a fuel cell module FCM and an information processing device SV such as a physical server or a virtual (cloud) server, and enables the fuel cell module FCM and the information processing device SV to transmit and receive data to and from each other via a network N such as the Internet.

[0020] The fuel cell module (FCM) is installed in vehicles such as forklifts, towing tractors, or automated guided vehicles (AGVs), and supplies power to the load Lo installed in those vehicles. In this configuration, the load Lo shown in Figure 1 is, for example, an inverter circuit that drives a cargo handling device or a drive motor.

[0021] The fuel cell module (FCM) may be installed in stationary generators such as industrial stationary generators, household stationary generators, or emergency stationary generators. In this configuration, the load Lo shown in Figure 1 may be, for example, industrial machinery or home appliances.

[0022] Furthermore, the fuel cell module (FCM) comprises a main unit, the fuel cell stack (FCS), and several types of auxiliary equipment for generating power for the fuel cell stack (FCS).

[0023] In other words, the fuel cell module (FCM) includes a fuel tank (HT), an injector (INJ), a gas-liquid separator (GLS), a hydrogen circulation pump (HP), an exhaust and drain valve (EDV), a diluent (DIL), and a pressure sensor (Sp) as auxiliary hydrogen gas system components.

[0024] Furthermore, the fuel cell module (FCM) is equipped with auxiliary oxidizer gas system components such as an air compressor (ACP) and an air pressure regulating valve (ARV).

[0025] Furthermore, the fuel cell module (FCM) is equipped with cooling system auxiliary components such as a radiator (R), a water pump (WP), and a temperature sensor (St).

[0026] Furthermore, the fuel cell module (FCM) is equipped with electrical auxiliary components such as a DC-DC converter (CNV) and an energy storage device (B).

[0027] Furthermore, the fuel cell module FCM also includes a voltage detection unit VD, a memory device Str1, a communication device Com1, and a control device Cnt.

[0028] A fuel cell stack (FCS) consists of three fuel cell cells C1 to C3 connected in series, and generates electricity through an electrochemical reaction between hydrogen gas (anode gas) and oxygen (cathode gas) in the air. When fuel cell cells C1 to C3 are not distinguished, they are simply referred to as fuel cell cell C. The number of fuel cell cells C in a fuel cell stack (FCS) is not limited to three; it may be two, four or more. A fuel cell C is, for example, a polymer electrolyte fuel cell (PEFC).

[0029] The fuel tank (HT) is a storage container for hydrogen gas. The hydrogen gas stored in the fuel tank (HT) is supplied to the fuel cell stack (FCS) via the injector (INJ).

[0030] The injector (INJ) regulates the flow rate of hydrogen gas supplied to the fuel cell stack (FCS).

[0031] The gas-liquid separator (GLS) separates unreacted hydrogen gas from liquid water discharged from the fuel cell stack (FCS).

[0032] The hydrogen circulation pump (HP) resupplies the hydrogen gas separated by the gas-liquid separator (GLS) back to the fuel cell stack (FCS).

[0033] The pressure sensor Sp detects the pressure P of the hydrogen gas supplied to the fuel cell stack FCS and sends the detected pressure P to the control device Cnt.

[0034] The exhaust drain valve (EDV) sends the liquid water separated by the gas-liquid separator (GLS) to the diluent (DIL). The liquid water sent to the diluent (DIL) is stored in a tank within the DIL. In addition, hydrogen gas and air discharged from the fuel cell stack (FCS) merge in the diluent (DIL) and are discharged to the outside or inside of the fuel cell module (FCM).

[0035] The air compressor (ACP) compresses air supplied from outside the fuel cell module (FCM) and delivers it to the fuel cell stack (FCS).

[0036] The air pressure regulating valve (ARV) adjusts the pressure and flow rate of air supplied to the fuel cell stack (FCS).

[0037] The radiator R exchanges heat between the coolant discharged from the fuel cell module (FCM) and the outside air. For example, the coolant is a liquid such as water containing additives such as antifreeze, rust inhibitors, and antioxidants.

[0038] The water pump (WP) supplies the cooling water, which has been heated by the radiator (R), to the fuel cell stack (FCS).

[0039] The temperature sensor St detects the temperature of the cooling water and sends the detected temperature to the control device Cnt.

[0040] The DC-DC converter (CNV) converts the voltage output from the fuel cell stack (FCS) to a predetermined voltage. The power output from the DC-DC converter (CNV) is supplied to various auxiliary equipment and loads (Lo).

[0041] Energy storage device B is composed of a lithium-ion battery or lithium-ion capacitor and is connected between the DC-DC converter CNV and the load Lo. If the supplied power, which is the difference between the power output from the DC-DC converter CNV and the total power supplied to each auxiliary device, is greater than the required power requested from outside the fuel cell module FCM (for example, a control unit not shown that controls the operation of load Lo), then the required power is supplied to load Lo, and the surplus power is supplied to energy storage device B. When power is supplied from the DC-DC converter CNV to energy storage device B, energy storage device B is charged and its charge rate (the ratio of remaining capacity to the full charge capacity of energy storage device B [%]) increases. Also, when regenerative power supplied from load Lo to the fuel cell module FCM is supplied to energy storage device B, energy storage device B is charged and its charge rate increases. Furthermore, if the supplied power is less than the required power, that supplied power is supplied to load Lo, and the deficit is supplied from energy storage device B to load Lo. When power is supplied from energy storage device B to load Lo, energy storage device B is discharged and its charge level decreases.

[0042] The voltage detection unit VD is composed of an IC (Integrated Circuit) and detects the voltage V1 of fuel cell C1, the voltage V2 of fuel cell C2, and the voltage V3 of fuel cell C3, and sends the detected voltages V1 to V3 to the control device Cnt.

[0043] The memory device Str1 is composed of components such as RAM (Random Access Memory) and ROM (Read Only Memory), and stores operational data, which will be described later.

[0044] Communication device Com1 sends and receives data with information processing device SV via network N.

[0045] The control device Cnt is composed of, for example, a microcomputer and comprises a power generation control unit 2 and an acquisition unit 3. For example, the power generation control unit 2 and the acquisition unit 3 are realized when the microcomputer executes a program stored in the memory device Str1.

[0046] The power generation control unit 2 controls the operation of each auxiliary device in order to control the power generation of the fuel cell stack (FCS). Specifically, the power generation control unit 2 changes the target power generation according to the charge level of the energy storage device B, and controls the operation of each auxiliary device so that the power generation of the fuel cell stack (FCS) follows the target power generation, using PI (Proportional-Integral) control, etc. For example, when the load Lo is in a high load state and the required power is relatively large, the energy storage device B is unlikely to reach a fully charged state, and when the required power fluctuates, the target power generation fluctuates to a value greater than zero. In such cases, the power generation control unit 2 performs normal operation in which the fuel cell stack (FCS) generates power. Also, when the load Lo is in a low load state and the required power is relatively small, the energy storage device B is likely to reach a fully charged state, and the target power generation is likely to become zero. In such cases, when the target power generation becomes zero, the power generation control unit 2 performs intermittent operation in which the fuel cell stack (FCS) does not generate power. During normal operation, the voltage of fuel cell cell C fluctuates in accordance with fluctuations in the target power generation. During intermittent operation, however, since the target power generation is zero, the voltage of fuel cell cell C does not fluctuate but gradually decreases. In other words, the time during which intermittent operation is performed is the time when operational data can be stably acquired.

[0047] Figure 2 schematically shows an example of voltage change in fuel cell C during intermittent operation. In the two-dimensional coordinate system shown in Figure 2, the horizontal axis represents time and the vertical axis represents voltage. The solid line in Figure 2 shows an example of voltage change in fuel cell C when no abnormality occurs, and the dashed line in Figure 2 shows an example of voltage change in fuel cell C when an abnormality occurs.

[0048] As shown in Figure 2, the rate of voltage drop per unit time of fuel cell cell C when an abnormality occurs is greater than the rate of voltage drop per unit time of fuel cell cell C when no abnormality occurs. In other words, during intermittent operation, the voltage of fuel cell cell C decreases gradually over time when no abnormality occurs, but the voltage of fuel cell cell C decreases sharply over time when an abnormality occurs.

[0049] Thus, the voltage of fuel cell C when an abnormality occurs and the voltage of fuel cell C when an abnormality does not occur have different change trends (the rate of decrease in the voltage of fuel cell C per unit time). If this difference in change trends can be recognized using parameters obtainable from fuel cell C, it is possible to detect that an abnormality has occurred in at least one of the multiple fuel cell Cs. The detection unit 4, described later, creates a histogram using the voltage of each fuel cell C that changes over time, recognizes the difference in the fluctuation trend of the voltage of fuel cell C using an index that shows the change in the created histogram, and detects that at least one fuel cell C is deteriorating. However, if an abnormality in fuel cell C is detected when the voltage of fuel cell C falls below the threshold Vth, an abnormality can be detected if intermittent operation continues from the start of intermittent operation (time t10) until the voltage of fuel cell C falls below the threshold Vth (time t11). However, there is a concern that an abnormality cannot be detected if intermittent operation ends before time t11.

[0050] Furthermore, the acquisition unit 3 shown in Figure 1 acquires operational data at regular intervals Tc1 (first regular time) when the fuel cell module FCM is in operation, and sends the acquired operational data to the communication device Com1, associating it with the acquisition time. The regular time Tc1 can be, for example, the clock period of the control device Cnt or any arbitrary time. The operational data can be, for example, voltages V1 to V3 detected by the voltage detection unit VD or pressure P detected by the pressure sensor Sp.

[0051] Communication device Com1 transmits the operational data received from acquisition unit 3 to information processing device SV via network N. Note that communication device Com2 may be provided in control device Cnt.

[0052] Here, Figure 3 is a flowchart showing an example of the operation of the acquisition unit 3.

[0053] First, the acquisition unit 3 determines that intermittent operation has started (step S11: Yes), and also determines that the temperature detected by the temperature sensor St is above a predetermined temperature (step S12: Yes), and acquires operational data and transmits the acquired operational data to the information processing device SV via the communication device Com1 (step S13). The predetermined temperature is the minimum value of the cooling water temperature when the voltage fluctuation of the fuel cell C is relatively small, for example, 50°C. In this way, when intermittent operation is in progress and the cooling water temperature is above the predetermined temperature, the voltage fluctuation of the fuel cell C can be suppressed and gradually decreased. Therefore, calculating a histogram using the operational data acquired when these conditions are met is effective in improving the accuracy of anomaly detection. However, if the fuel cell module FCM is in a relatively high-temperature environment and the temperature of the fuel cell C tends to exceed the predetermined temperature, the process in step S12 may be omitted. In this case, the acquisition unit 3 determines that intermittent operation has started (step S11: Yes), and then executes the process in step S13.

[0054] Next, if the acquisition unit 3 determines that the intermittent operation has not finished (step S14: No), after a certain period of time Tc1 has elapsed (step S15: Yes), it acquires the operation data again and transmits it to the information processing device SV (step S13). If it determines that the intermittent operation has finished (step S14: Yes), it terminates the operation data acquisition process. In other words, the acquisition unit 3 repeatedly acquires operation data at regular intervals of Tc1 from the start to the end of the intermittent operation.

[0055] Furthermore, the acquisition unit 3 may be configured to switch a valid flag, which indicates whether or not the data is valid as operational data for creating a histogram described later, from off to on when it determines that intermittent operation has started and that the temperature detected by the temperature sensor St is above a predetermined temperature, and to transmit the operational data along with the valid flag to the information processing device SV. In this configuration, when intermittent operation starts, the acquisition unit 3 transmits the operational data to the information processing device SV at regular intervals Tc1.

[0056] Furthermore, the acquisition unit 3 may be configured to acquire operational data at regular intervals Tc1 and store it in the storage device Str1, and to collect all the operational data stored in the storage device Str1 and transmit it from the fuel cell module FCM to the information processing device SV at regular intervals Tc2 (second regular interval) that are longer than Tc1. For example, if the regular interval Tc2 is set to a relatively short time (e.g., 8 hours), and the time during which the above conditions are met is relatively short, the histogram described later may fluctuate due to noise contained in the operational data. Also, if the regular interval Tc2 is set to a relatively long time (e.g., 1 month), the changes in the operational data may be smoothed out, and even if an abnormality occurs in the fuel cell cell C, it may not be possible to sufficiently change the indicator showing the change in the histogram described later. Therefore, it is desirable to set the constant time Tc2 to a length of time (for example, 40 hours = 8 hours × 5 days) that can suppress fluctuations in the index showing the change in the histogram, as described later, due to noise included in the operating data, and that can sufficiently change the index showing the change in the histogram when an abnormality occurs in at least one fuel cell cell C.

[0057] Furthermore, the information processing device SV shown in Figure 1 comprises a communication device Com2, a storage device Str2, and a processor Pro.

[0058] The communication device Com2 is composed of, for example, a router, and receives operational data transmitted from the fuel cell module FCM via network N.

[0059] The memory device Str2 is composed of components such as RAM and ROM, and stores operational data received by the communication device Com2.

[0060] Figure 4 shows an example of a database stored in the memory device Str2. The database shown in Figure 4 contains multiple records RC (records RC10~RC1300, records RC20~RC2300, records RC30~RC3300, and records RC40~RC4300). The "Acquisition Time" column of each record RC stores the time when the operational data was acquired. The "Identifier" column of each record RC stores information to identify the fuel cell module FCM that is the source of the operational data. The "Operational Data Do1" column of each record RC stores voltage V1, the "Operational Data Do2" column stores voltage V2, and the "Operational Data Do3" column stores voltage V3. The "Enabled Flag" column of each record RC stores whether it is on or off.

[0061] Each record RC shown in Figure 4 stores operational data acquired every second over a 40-hour operating period of the fuel cell module FCM. Specifically, records RC10 to RC1300 store voltages V1 to V3 acquired every second during the 5-minute intermittent operation from "2024 / 10 / 7 9:30:00" to "2024 / 10 / 7 9:35:00", along with the identifier FCM1 and the enabled flag "On". Records RC20 to RC2300 store voltages V1 to V3 acquired every second during the 5-minute normal operation from "2024 / 10 / 7 9:35:30" to "2024 / 10 / 7 9:40:30", along with the identifier FCM1 and the enabled flag "Off". Records RC30 to RC3300 store the voltages V1 to V3 acquired every second during the 5-minute intermittent operation from "2024 / 10 / 9 16:15:00" to "2024 / 10 / 9 16:20:00", along with the identifier FCM1 and the enabled flag "On". Records RC40 to RC4300 store the voltages V1 to V3 acquired every second during the 5-minute intermittent operation from "2024 / 10 / 11 13:45:00" to "2024 / 10 / 11 13:50:00", along with the identifier FCM1 and the enabled flag "On".

[0062] Furthermore, the processor Pro shown in Figure 1 includes a detection unit 4. For example, the detection unit 4 is realized when the processor Pro executes a program stored in the memory device Str2.

[0063] The detection unit 4 uses the operational data acquired during a certain period of time Tc2 from among the operational data stored in the memory device Str2 (for example, operational data with the valid flag shown in Figure 4 turned on) to create a histogram on a two-dimensional coordinate system where the horizontal axis represents the size of the operational data and the vertical axis represents the length of the operational data acquisition time.

[0064] For example, consider creating a histogram using each record RC shown in Figure 4.

[0065] First, the detection unit 4 extracts records RC10-RC1300, RC30-RC3300, and RC40-RC4300 from records RC10-RC1300, RC20-RC2300, RC30-RC3300, and RC40-RC4300 as records RC with the valid flag set to "on".

[0066] Next, the detection unit 4 extracts the minimum operating data (minimum voltage Vmin) from each extracted record RC. Specifically, the detection unit 4 extracts "0.90" from record RC10, "0.90" from record RC11, "0.88" from record RC12, ... and "0.85" from record RC1300. Furthermore, the detection unit 4 extracts "0.90" from record RC30, "0.88" from record RC31, "0.87" from record RC32, ... and "0.85" from record RC3300. Furthermore, the detection unit 4 extracts "0.90" from record RC40, "0.88" from record RC41, "0.87" from record RC42, ... and "0.84" from record RC4300. For example, if four "0.90" values ​​are extracted, three "0.88" values ​​are extracted, two "0.87" values ​​are extracted, and one "0.84" value is extracted, the acquisition time for "0.90" will be 4 seconds, the acquisition time for "0.88" will be 3 seconds, the acquisition time for "0.87" will be 2 seconds, and the acquisition time for "0.84" will be 1 second.

[0067] Next, the detection unit 4 creates a histogram using the extracted operational data. Specifically, the detection unit 4 creates a histogram by setting the acquisition time on the horizontal axis (frequency) for "0.90" on the horizontal axis (class) to "4", the acquisition time on the horizontal axis (frequency) for "0.88" on the horizontal axis (class) to "3", the acquisition time on the horizontal axis (frequency) for "0.87" on the horizontal axis (class) to "2", and the acquisition time on the horizontal axis (frequency) for "0.84" on the horizontal axis (class) to "1".

[0068] Then, the detection unit 4 creates a new histogram using the operational data acquired during the next fixed time period Tc2. In other words, the detection unit 4 repeatedly performs the histogram creation process.

[0069] Figure 5 shows an example of a histogram. In Figures 5(a) to 5(c), the horizontal axis of the two-dimensional coordinate system represents the magnitude of the voltage, and the vertical axis represents the length of the acquisition time. In the bar graphs shown in Figures 5(a) to 5(c), the white portion represents the acquisition time corresponding to voltage V1, the portion shaded with an upward-sloping diagonal line represents the acquisition time corresponding to voltage V2, and the portion shaded with a downward-sloping diagonal line represents the acquisition time corresponding to voltage V3.

[0070] Figure 5(a) shows the state before the fuel cell module (FCM) is put into use, and a histogram created using the voltages V1 to V3 over a certain period of time Tc2, assuming that no abnormalities have occurred in any of the fuel cell cells C1 to C3.

[0071] Figure 5(b) shows a histogram created using voltages V1 to V3 over a certain period of time Tc2, in a situation where a predetermined amount of time has elapsed since the start of use of the fuel cell module (FCM), and signs of abnormality have begun to appear only in fuel cell cell C1.

[0072] Figure 5(c) shows a histogram created using voltages V1 to V3 over a certain period of time Tc2, in a situation where a predetermined amount of time has elapsed since the start of use of the fuel cell module FCM, and an abnormality has occurred only in fuel cell cell C1.

[0073] In the histogram shown in Figure 5(a), the total acquisition time corresponding to voltage V3 (total area of ​​the shaded portion with a downward-sloping diagonal line) is greater than the total acquisition time corresponding to voltage V1 (total area of ​​the white portion) and the total acquisition time corresponding to voltage V2 (total area of ​​the shaded portion with an upward-sloping diagonal line), indicating that voltage V3 was the lowest voltage most frequently during a certain period of time Tc2. Furthermore, in the histogram shown in Figure 5(a), since no abnormalities occurred in any of the fuel cell cells C1 to C3, the acquisition times corresponding to voltages V1 to V3 are concentrated at relatively high voltages (classes).

[0074] In the histogram shown in Figure 5(b), the total acquisition time corresponding to voltage V2 (total area of ​​the shaded portion with upward-sloping diagonal lines) is larger than the total acquisition time corresponding to voltage V1 (total area of ​​the white portion) and the total acquisition time corresponding to voltage V3 (total area of ​​the shaded portion with downward-sloping diagonal lines), indicating that voltage V2 was at its lowest voltage most frequently during a certain period of time Tc2. Also, in the histogram shown in Figure 5(b), since signs of abnormality are beginning to appear only in fuel cell cell C1, the acquisition times corresponding to voltages V2 and V3 are concentrated at relatively high voltages (classes), but the acquisition time corresponding to voltage V1 is somewhat spread out between relatively high and relatively low voltages (classes).

[0075] In the histogram shown in Figure 5(c), the total acquisition time corresponding to voltage V1 (total area of ​​the white portion) is greater than the total acquisition time corresponding to voltage V2 (total area of ​​the portion shaded with upward-sloping diagonal lines) and the total acquisition time corresponding to voltage V3 (total area of ​​the portion shaded with downward-sloping diagonal lines), indicating that voltage V1 was at its lowest voltage most frequently during a certain period of time Tc2. Furthermore, in the histogram shown in Figure 5(c), since the anomaly occurred only in fuel cell cell C1, the acquisition time corresponding to voltage V1 is spread over a wide range from relatively high voltages (classes) to relatively low voltages (classes).

[0076] Thus, if each acquisition time (bar graph) in the histogram is concentrated at relatively high voltages (classes) and does not spread much between relatively high and relatively low voltages (classes), it can be estimated that no abnormalities have occurred in any of the fuel cell cells C.

[0077] Furthermore, if the histogram's acquisition time (bar graph) is spread over a wide range from relatively high voltages (classes) to relatively low voltages (classes), it can be estimated that an abnormality has occurred in at least one fuel cell C (for example, a voltage drop in fuel cell C due to deterioration of fuel cell C).

[0078] In other words, by recognizing the trend in the histogram's changes, it is possible to grasp the trend of abnormal conditions in fuel cell cell C.

[0079] Therefore, in this embodiment, an indicator showing changes in the histogram is used to detect when an abnormality occurs in the fuel cell module (FCM). The operational data used to create the histogram is not particularly limited, as long as it is a value that changes the indicator showing changes in the histogram as the abnormal state of the fuel cell module (FCM) progresses.

[0080] Here, Figure 6 is a flowchart showing an example of the abnormality detection process performed by the detection unit 4. Note that the abnormality detection process shown in Figure 6 may be repeatedly executed at regular intervals Tc2, or it may be executed at any timing after a regular interval Tc2 has elapsed.

[0081] First, the detection unit 4 creates a histogram using operating data for a certain period of time Tc2 (step S21), and then determines an index that indicates the change in the histogram from the created histogram (step S22).

[0082] Next, the detection unit 4 detects whether or not an abnormality has occurred in the fuel cell module (FCM) based on an index showing changes in the histogram (step S23).

[0083] If the detection unit 4 detects that no abnormality has occurred in the fuel cell module (FCM) (step S23: No), it terminates the abnormality detection process and waits until the next abnormality detection process.

[0084] On the other hand, if the detection unit 4 detects that an abnormality has occurred in the fuel cell module (FCM) (step S23: Yes), it turns on the abnormality flag in the operating data corresponding to that fuel cell module (FCM) (step S24), terminates the abnormality determination process, and waits until the next abnormality detection process. The operating data corresponding to the abnormality flag can be used to review how to use the fuel cell module (FCM) and to understand its status. In addition, the operating data corresponding to the abnormality flag may be used to improve the quality during maintenance or manufacturing of the fuel cell module (FCM).

[0085] Furthermore, the detection unit 4 may be provided in the control device Cnt. In this configuration, the detection unit 4 uses the operation data acquired during a certain period of time Tc2 from among the operation data stored in the memory device Str1 to create a histogram on a two-dimensional coordinate system where the horizontal axis represents the size of the operation data and the vertical axis represents the length of the operation data acquisition time. Based on an index showing the changes in the created histogram, the detection unit 4 detects that an abnormality has occurred in the fuel cell system 1.

[0086] In other words, the detection unit 4 uses the operational data acquired by the acquisition unit 3 during a certain period of time Tc2 to create a histogram on a two-dimensional coordinate system where the horizontal axis represents the size of the operational data and the vertical axis represents the length of the operational data acquisition time. Based on an index showing the changes in the created histogram, the detection unit 4 detects that an abnormality has occurred in the fuel cell module (FCM).

[0087] This allows for the acquisition of a certain amount of operational data (Tc2) that can change the index indicating histogram changes in response to an abnormality in the fuel cell module (FCM). By setting Tc2 for a certain period of time, it becomes possible to detect an abnormality in the fuel cell module (FCM) based on the index indicating histogram changes, even when the time for which operational data can be stably acquired, i.e., the time during which intermittent operation is performed, is relatively short.

[0088] <Example 1> In Example 1, the mean value Vmean of the histogram is used as an indicator to show the change in the histogram.

[0089] Here, Figure 7(a) shows an example of the average value Vmean of the histograms created at regular intervals Tc2. In the two-dimensional coordinate system shown in Figure 7(a), the horizontal axis represents time [h] and the vertical axis represents voltage [V]. Each point on the two-dimensional coordinate system shown in Figure 7(a) represents the average value Vmean obtained at regular intervals Tc2. Also, time t20 < time t21 < time t22 < time t23. Time t20 is an arbitrary time at the time of factory shipment of the fuel cell module FCM, and it is assumed that no degradation abnormality has occurred in any fuel cell cell C at time t20, and the same applies to the two-dimensional coordinate systems shown in Figures 7(b) to 7(e) described later. Furthermore, it is assumed that at least one fuel cell cell C has an abnormality (for example, a voltage drop in fuel cell C due to degradation of fuel cell C) from time t21 onward, and the same applies to the two-dimensional coordinate systems shown in Figures 7(b) to 7(e) described later.

[0090] In the example shown in Figure 7(a), no abnormalities occurred in any of the fuel cell cells C between time t20 and time t21, so the average value Vmean is kept within a predetermined range of change relative to a constant value Vc1. For example, the constant value Vc1 is the approximate straight line (regression line) obtained from the average value Vmean between time t20 and time t21. In other words, if no abnormalities occur in any of the fuel cell cells C, the average value Vmean of the histogram remains almost unchanged.

[0091] Furthermore, in the example shown in Figure 7(a), the average Vmean gradually decreases from time t23 onward. In other words, if an abnormality occurs in at least one fuel cell cell C, the average Vmean of the histogram changes.

[0092] Thus, there is a correlation between the occurrence of an abnormality in the fuel cell module (FCM) and the change in the histogram's mean value, Vmean. Therefore, the mean value of the histogram, Vmean, can be used as an indicator of the histogram's changes. Furthermore, the smaller the mean value Vmean, the larger the change in the histogram, and the higher the probability that an abnormality has occurred in the fuel cell module (FCM).

[0093] The mean value, Vmean, may be calculated using a general method for determining the mean of a histogram.

[0094] For example, consider a case in a histogram where the frequency "acquisition time Tα" belongs to the class "voltage Vα", the frequency "acquisition time Tβ" belongs to the class "voltage Vβ", and the frequency "acquisition time Tγ" belongs to the class "voltage Vγ".

[0095] In this case, the detection unit 4 calculates the average value Vmean by calculating (voltage Vα × acquisition time Tα + voltage Vβ × acquisition time Tβ + voltage Vγ × acquisition time Tγ) / (acquisition time Tα + acquisition time Tβ + acquisition time Tγ). When the detection unit 4 determines that the average value Vmean is less than or equal to the threshold Th1, it detects that an abnormality has occurred in at least one fuel cell cell C. It is desirable that the threshold Th1 be set to a value smaller than the constant value Vc1, taking into account the variability of the average value Vmean.

[0096] This makes it possible to detect an anomaly in at least one fuel cell cell C based on the mean value Vmean, which is an indicator of changes in the histogram.

[0097] Furthermore, by setting Tc2 for a certain period of time so that an amount of voltage V1 to V3 can be acquired that is sufficient to change the average value Vmean in the event of an abnormality in at least one fuel cell cell C, it is possible to detect that an abnormality has occurred in at least one fuel cell cell C based on the average value Vmean, even if the time for which voltage V1 to V3 can be stably acquired is relatively short.

[0098] Furthermore, the detection unit 4 may be configured to detect that an abnormality has occurred in at least one fuel cell cell C if it determines that the average value Vmean is less than or equal to the threshold Th1 multiple times in a row.

[0099] Thus, when an abnormality is detected in at least one fuel cell cell C when the average value Vmean is determined to be less than or equal to the threshold Th1 multiple times in a row, the accuracy of abnormality detection can be improved by reducing false detections due to noise contained in the voltage, compared to when an abnormality is detected in at least one fuel cell cell C when the average value Vmean is determined to be less than or equal to the threshold Th1 only once.

[0100] <Example 2> In Example 2, the median of the histogram, Vmedi, is used as an indicator to show the change in the histogram.

[0101] Figure 7(b) shows an example of the median Vmedi of the histograms created at regular intervals of time Tc2. In the two-dimensional coordinate system shown in Figure 7(b), the horizontal axis represents time [h] and the vertical axis represents voltage [V]. Each point on the two-dimensional coordinate system shown in Figure 7(b) represents the median Vmedi obtained at regular intervals of time Tc2.

[0102] In the example shown in Figure 7(b), no degradation abnormalities occurred in any of the fuel cell cells C between time t20 and time t21, so each median Vmedi is kept within a predetermined range of change relative to a constant value Vc2. For example, the constant value Vc2 is the approximate straight line (regression line) obtained from each median Vmedi between time t20 and time t21. That is, if no degradation abnormalities occur in any of the fuel cell cells C, the median Vmedi of the histogram remains almost unchanged.

[0103] Furthermore, in the example shown in Figure 7(b), the median Vmedi gradually decreases from time t23 onward. In other words, if an abnormality occurs in at least one fuel cell cell C, the median Vmedi of the histogram changes.

[0104] Thus, there is a correlation between the occurrence of an abnormality in the fuel cell module (FCM) and the change in the median value Vmedi of the histogram. Therefore, the median value Vmedi of the histogram can be used as an indicator of the change in the histogram. Furthermore, the smaller the median Vmedi, the larger the change in the histogram, and the higher the probability that an abnormality has occurred in the fuel cell module (FCM).

[0105] The median Vmedi may be obtained using a general method for calculating the median of a histogram.

[0106] For example, consider a histogram where the class "Voltage Vα" corresponds to the frequency "Acquisition Time Tα", the class "Voltage Vβ" corresponds to the frequency "Acquisition Time Tβ", and the class "Voltage Vγ" corresponds to the frequency "Acquisition Time Tγ", and where acquisition times Tα=5, Tβ=10, and Tγ=15, and voltage Vα < voltage Vβ < voltage Vγ.

[0107] In this case, the detection unit 4 calculates "15" as the median of acquisition time Tα + acquisition time Tβ + acquisition time Tγ, and arranges the voltages Vα to Vγ in ascending order. It then sets the "voltage Vβ" which is the class obtained by counting the acquisition times sequentially from voltage Vα by the calculated median value, as the median value of the histogram, Vmedi. When the detection unit 4 determines that the median value Vmedi of the histogram is less than or equal to the threshold Th2, it detects that an abnormality has occurred in at least one fuel cell cell C. It is desirable that the threshold Th2 be set to a value smaller than the constant value Vc2 by a value that takes into account the variability of the median value Vmedi.

[0108] This makes it possible to detect an anomaly in at least one fuel cell cell C based on the median Vmedi, which is an indicator of the change in the histogram.

[0109] Furthermore, by setting Tc2 for a certain period of time so that voltages V1 to V3 can be acquired in amounts sufficient to change the median Vmedi in response to an abnormality occurring in at least one fuel cell cell C, it is possible to detect an abnormality in at least one fuel cell cell C based on the median Vmedi, even if the time for which voltages V1 to V3 can be stably acquired is relatively short.

[0110] Furthermore, since the variation in each median Vmedi shown in Figure 7(b) is smaller than the variation in each mean Vmean shown in Figure 7(a), the threshold Th2 can be set closer to a constant value Vc2, thereby shortening the time it takes from the occurrence of an abnormality in the fuel cell C until the detection unit 4 detects the abnormality.

[0111] Furthermore, the detection unit 4 may be configured to detect that an abnormality has occurred in at least one fuel cell cell C when it determines that the median value Vmedi of the histogram is less than or equal to the threshold Th2 multiple times in a row.

[0112] Thus, when an abnormality is detected in at least one fuel cell cell C when the median Vmedi is determined to be below the threshold Th2 multiple times in a row, the accuracy of abnormality detection can be improved by reducing false detections due to noise in the voltage, compared to when an abnormality is detected in at least one fuel cell cell C when the median Vmedi is determined to be below the threshold Th2 only once.

[0113] <Example 3> In Example 3, the standard deviation Vsd of the histogram (a value indicating the variability of the histogram) is used as an indicator to show the change in the histogram.

[0114] Figure 7(c) shows an example of the standard deviation Vsd of the histogram created at regular intervals of time Tc2. In the two-dimensional coordinate system shown in Figure 7(c), the horizontal axis represents time [h] and the vertical axis represents voltage [V]. Each point on the two-dimensional coordinate system shown in Figure 7(c) represents the standard deviation Vsd obtained at regular intervals of time Tc2.

[0115] In the example shown in Figure 7(c), no abnormalities occurred in any of the fuel cell cells C between time t20 and time t21, so each standard deviation Vsd is kept within a predetermined range of change relative to a constant value Vc3. For example, the constant value Vc3 is the approximate straight line (regression line) obtained from each standard deviation Vsd between time t20 and time t21. In other words, if no abnormalities occur in any of the fuel cell cells C, the standard deviation Vsd of the histogram remains almost unchanged.

[0116] Furthermore, in the example shown in Figure 7(c), each standard deviation Vsd gradually increases from time t22 onward. In other words, if an abnormality occurs in at least one fuel cell cell C, the standard deviation Vsd of the histogram changes.

[0117] Thus, there is a correlation between the occurrence of an abnormality in the fuel cell module (FCM) and the change in the histogram's standard deviation (Vsd). Therefore, the histogram's standard deviation (Vsd) can be used as an indicator of the histogram's changes. Furthermore, the larger the standard deviation (Vsd), the greater the change in the histogram, and the higher the likelihood that an abnormality has occurred in the fuel cell module (FCM).

[0118] The standard deviation Vsd may be calculated using the general method for determining the standard deviation of a histogram.

[0119] For example, consider a case in a histogram where the class "Voltage Vα" corresponds to the frequency "Acquisition Time Tα", the class "Voltage Vβ" corresponds to the frequency "Acquisition Time Tβ", and the class "Voltage Vγ" corresponds to the frequency "Acquisition Time Tγ".

[0120] In this case, the detection unit 4 calculates the mean value Vmean of the histogram, and then calculates (acquisition time Tα + acquisition time Tβ + acquisition time Tγ) / ((voltage Vα - mean value Vmean) 2 +(Voltage Vβ - Mean value Vmean) 2 +(Voltage Vγ - Average Value Vmean) 2 The square root of the result calculated is taken as the standard deviation Vsd. When the detection unit 4 determines that the standard deviation Vsd is greater than or equal to the threshold Th3, it detects that a degradation abnormality has occurred in at least one fuel cell cell C. It is desirable that the threshold Th3 be set to a value greater than the constant value Vc3, taking into account the variability of the standard deviation Vsd in the histogram.

[0121] This makes it possible to detect an anomaly in at least one fuel cell cell C based on the standard deviation Vsd, which is an indicator of changes in the histogram.

[0122] Furthermore, by setting Tc2 to a certain period of time so that voltages V1 to V3 can be acquired in amounts that change the standard deviation Vsd in the event of an abnormality in at least one fuel cell cell C, it is possible to detect that an abnormality has occurred in at least one fuel cell cell C based on the standard deviation Vsd, even if the time for which voltages V1 to V3 can be stably acquired is relatively short.

[0123] Furthermore, the timing at which the increase in the standard deviation Vsd associated with the occurrence of an anomaly (time t22), as shown in Figure 7(c), begins (time t23), is earlier than the timing at which the decrease in the mean Vmean associated with the occurrence of an anomaly (time t23), as shown in Figure 7(a), and the timing at which the decrease in the median Vmedi associated with the occurrence of an anomaly (time t23), as shown in Figure 7(b). Therefore, the time required from the actual occurrence of an anomaly in the fuel cell C until the detection unit 4 detects the anomaly can be shortened.

[0124] Furthermore, the detection unit 4 may be configured to detect that a degradation abnormality has occurred in at least one fuel cell cell C when it determines that the standard deviation Vsd of the histogram is equal to or greater than the threshold Th3 multiple times in a row.

[0125] Thus, when the standard deviation Vsd is determined to be greater than or equal to the threshold Th3 multiple times in a row, and a degradation anomaly is detected in at least one fuel cell cell C, the accuracy of anomaly detection can be improved by reducing false detections due to noise contained in the voltage, compared to when the standard deviation Vsd is determined to be greater than or equal to the threshold Th3 only once.

[0126] <Example 4> In Example 4, the ratio Ra of the acquisition time corresponding to a voltage below a predetermined voltage Vra in the histogram, relative to the total acquisition time of the histogram, is used as an indicator of the change in the histogram. The predetermined voltage Vra is the maximum voltage of the fuel cell cell C where the abnormality is occurring, for example, 0.5[V] or 0.4[V].

[0127] Here, Figure 7(d) shows an example of a percentage Ra created at regular intervals of time Tc2. In the two-dimensional coordinate system shown in Figure 7(d), the horizontal axis represents time [h] and the vertical axis represents the percentage [%]. Furthermore, each point on the two-dimensional coordinate system shown in Figure 7(d) represents the percentage Ra obtained at regular intervals of time Tc2.

[0128] In the example shown in Figure 7(d), no abnormalities occur in any of the fuel cell cells C between time t20 and time t21, so the ratio Ra is kept within a predetermined range of change relative to a constant value Rc. For example, the constant value Rc is an approximate straight line (regression line) obtained from each ratio Ra between time t20 and time t21. That is, if no abnormalities occur in any of the fuel cell cells C, the ratio Ra remains almost unchanged.

[0129] Furthermore, in the example shown in Figure 7(d), the ratio Ra gradually increases from time t22 onward. In other words, if an abnormality occurs in at least one fuel cell cell C, the ratio Ra changes.

[0130] Thus, because there is a correlation between the occurrence of an abnormality in the fuel cell module (FCM) and the change in the proportion Ra, the proportion Ra can be used as an indicator of the change in the histogram. Furthermore, the larger the proportion Ra, the greater the change in the histogram, and the higher the probability that an abnormality has occurred in the fuel cell module (FCM).

[0131] For example, consider a histogram where "acquisition time Tα" as a frequency belongs to the class "voltage Vα", "acquisition time Tβ" as a frequency belongs to the class "voltage Vβ", and "acquisition time Tγ" as a frequency belongs to the class "voltage Vγ", and voltage Vα is less than or equal to a predetermined voltage Vra.

[0132] In this case, the detection unit 4 uses the result of the calculation (voltage Vα × acquisition time Tα) / ((voltage Vα × acquisition time Tα) + (voltage Vβ × acquisition time Tβ) + (voltage Vγ × acquisition time Tγ)) as the ratio Ra. When the detection unit 4 determines that the ratio Ra is greater than or equal to the threshold Th4, it detects that an abnormality has occurred in at least one fuel cell cell C. It is desirable that the threshold Th4 be set to a value greater than the constant value Rc, taking into account the variation in the ratio Ra.

[0133] This makes it possible to detect an anomaly in at least one fuel cell cell C based on the percentage Ra, which is an indicator of the change in the histogram.

[0134] Furthermore, by setting Tc2 to a certain period of time so that voltages V1 to V3 can be acquired in amounts that change the ratio Ra in response to an abnormality occurring in at least one fuel cell cell C, it is possible to detect an abnormality in at least one fuel cell cell C based on the ratio Ra, even if the time for which voltages V1 to V3 can be stably acquired is relatively short.

[0135] Furthermore, the timing at which the percentage Ra increases due to the occurrence of an abnormality (time t22), as shown in Figure 7(d), is earlier than the timing at which the average value Vmean begins to decrease due to the occurrence of an abnormality (time t23), as shown in Figure 7(a), and the timing at which the median value Vmedi begins to decrease due to the occurrence of an abnormality (time t23), as shown in Figure 7(b). Therefore, the time required from the occurrence of an abnormality in the fuel cell C until the detection unit 4 detects the abnormality can be shortened.

[0136] Furthermore, the detection unit 4 may be configured to detect that a degradation abnormality has occurred in at least one fuel cell cell C when it determines that the ratio Ra is equal to or greater than the threshold Th4 multiple times in a row.

[0137] Thus, when an abnormality is detected in at least one fuel cell cell C when the ratio Ra is determined to be above the threshold Th4 multiple times in a row, the accuracy of abnormality detection can be improved by reducing false detections due to noise contained in the voltage, compared to when an abnormality is detected in at least one fuel cell cell C when the ratio Ra is determined to be above the threshold Th4 only once.

[0138] <Example 5> In Example 5, the dissimilarity D of the histogram shape is used as an indicator of the change in the histogram.

[0139] Figure 7(e) shows an example of the dissimilarity D for each histogram shape created at regular intervals of time Tc2. In the two-dimensional coordinate system shown in Figure 7(e), the horizontal axis represents time [h], and the vertical axis represents the magnitude of the dissimilarity D. Furthermore, each point on the two-dimensional coordinate system shown in Figure 7(e) represents the dissimilarity D obtained at regular intervals of time Tc2.

[0140] In the example shown in Figure 7(e), no abnormalities occurred in any of the fuel cell cells C between time t20 and time t21, so the dissimilarity D is kept within a predetermined range of change based on a constant value Dc. For example, the constant value Dc is the approximate straight line (regression line) obtained from each dissimilarity D between time t20 and time t21. In other words, if no abnormalities occur in any of the fuel cell cells C, the dissimilarity D remains almost unchanged.

[0141] Furthermore, in the example shown in Figure 7(e), the dissimilarity D gradually increases from time t22 onward. In other words, the dissimilarity D changes after a predetermined time has elapsed since an abnormality occurred in at least one fuel cell cell C.

[0142] Thus, since there is a correlation between the occurrence of an abnormality in the fuel cell module (FCM) and a change in the dissimilarity index D, the dissimilarity index D can be used as an indicator of the change in the histogram. Furthermore, the larger the dissimilarity index D, the greater the change in the histogram, and the higher the probability that an abnormality has occurred in the fuel cell module (FCM).

[0143] Here, Figure 8(a) shows the state before the fuel cell module FCM is put into use, and no abnormalities have occurred in any of the fuel cell cells C1 to C3. The histogram created using only the minimum voltage Vmin (first histogram) is shown. Figure 8(b) shows the state before the fuel cell module FCM is put into use, and no abnormalities have occurred in any of the fuel cell cells C1 to C3. The histogram created using only the maximum voltage Vmax (second histogram) is shown. Figure 8(c) shows the state after a predetermined time has elapsed since the start of use of the fuel cell module FCM, and no abnormalities have occurred in any of the fuel cell cells C1 to C3. The histogram created using only the minimum voltage Vmin (first histogram) is shown. Figure 8(d) shows the state after a predetermined time has elapsed since the start of use of the fuel cell module FCM, and no abnormalities have occurred in any of the fuel cell cells C1 to C3. The histogram created using only the maximum voltage Vmax (second histogram) is shown. Furthermore, Figure 8(e) shows a histogram (first histogram) created using only the minimum voltage Vmin when a predetermined time has elapsed since the start of use of the fuel cell module FCM and an abnormality has occurred in at least one fuel cell cell C. Furthermore, Figure 8(f) shows a histogram (second histogram) created using only the maximum voltage Vmax when a predetermined time has elapsed since the start of use of the fuel cell module FCM and an abnormality has occurred in at least one fuel cell cell C. In other words, the shape of the histogram shown in Figure 8(a) and the shape of the histogram shown in Figure 8(b) are similar, but the shape of the histogram shown in Figure 8(c) and the shape of the histogram shown in Figure 8(d) are not similar. Also, the shape of the histogram shown in Figure 8(a) and the shape of the histogram shown in Figure 8(c) are similar, but the shape of the histogram shown in Figure 8(a) and the shape of the histogram shown in Figure 8(e) are not similar. Thus, determining whether the shape of the histogram created using only the minimum voltage is similar to the shape of the histogram created using only the maximum voltage is an effective way to detect an abnormality in fuel cell cell C.Furthermore, determining whether the shape of past histograms created using only the lowest voltage is similar to the shape of current histograms created using only the lowest voltage is an effective way to detect an abnormality in fuel cell cell C.

[0144] For example, the detection unit 4 calculates the Wasserstein distance or KL distance between the histogram shown in Figure 8(e) and the histogram shown in Figure 8(f), and sets the result of this calculation as the dissimilarity D. Alternatively, the detection unit 4 calculates the Wasserstein distance or KL distance between the histogram shown in Figure 8(a), created at the first time step, and the histogram shown in Figure 8(e), created at a second time step after the first time step, and sets the result of this calculation as the dissimilarity D. When the detection unit 4 determines that the dissimilarity D is greater than or equal to the threshold Th5, it detects that an abnormality has occurred in at least one fuel cell cell C. It is desirable that the threshold Th5 be set to a value greater than a constant value Dc, taking into account the variability of the dissimilarity D. The Wasserstein distance or KL distance should be calculated using a general calculation method. Furthermore, the dissimilarity D is not limited to the Wasserstein distance or KL distance, as long as it is a value that can indicate the degree to which the shapes of the histograms being compared are not similar.

[0145] This makes it possible to detect that an anomaly has occurred in at least one fuel cell cell C based on the dissimilarity D, which is an indicator of changes in the histogram.

[0146] Furthermore, by setting Tc2 to a certain period of time so that voltages V1 to V3 can be acquired in amounts sufficient to change the dissimilarity D in the event of an abnormality in at least one fuel cell cell C, it is possible to detect an abnormality in at least one fuel cell cell C based on the dissimilarity D, even if the time for which voltages V1 to V3 can be stably acquired is relatively short.

[0147] Furthermore, the timing at which the increase in dissimilarity D due to the occurrence of an anomaly (time t22), as shown in Figure 7(d), begins is earlier than the timing at which the decrease in the mean value Vmean due to the occurrence of an anomaly (time t23), as shown in Figure 7(a), and the timing at which the decrease in the median value Vmedi due to the occurrence of an anomaly (time t23), as shown in Figure 7(b). Therefore, the time required from the occurrence of an anomaly in the fuel cell cell C until the detection unit 4 detects the anomaly can be shortened.

[0148] Furthermore, the detection unit 4 may be configured to detect that an abnormality has occurred in at least one fuel cell cell C if it determines that the dissimilarity D is equal to or greater than the threshold Th5 multiple times in a row.

[0149] Thus, when an abnormality is detected in at least one fuel cell cell C when the dissimilarity D is determined to be above the threshold Th5 multiple times in a row, the accuracy of abnormality detection can be improved by reducing false detections due to noise in the voltage, compared to when an abnormality is detected in at least one fuel cell cell C when the dissimilarity D is determined to be above the threshold Th5 only once.

[0150] <Method for identifying fuel cell cell C that has experienced degradation abnormalities> Figure 7(f) shows the percentage Ri of the acquisition time corresponding to each fuel cell C out of the total acquisition time of the histogram. In the two-dimensional coordinate system shown in Figure 7(f), the horizontal axis represents time [h] and the vertical axis represents percentage [%]. The solid line in Figure 7(f) shows Ri1, which is the percentage of the acquisition time corresponding to fuel cell C1 out of the total acquisition time of the histogram. The dashed line in Figure 7(f) shows Ri2, which is the percentage of the acquisition time corresponding to fuel cell C2 out of the total acquisition time of the histogram. The dashed line in Figure 7(f) shows Ri3, which is the percentage of the acquisition time corresponding to fuel cell C3 out of the total acquisition time of the histogram.

[0151] In the example shown in Figure 7(f), due to manufacturing variations in fuel cell cells C1 to C3, the ratio Ri3 is at its highest from time t20 to time t22. Also, in the example shown in Figure 7(f), due to an abnormality occurring in fuel cell cell C1 at time t21, the ratio Ri1 is at its highest from time t23 onward.

[0152] Furthermore, in the examples shown in Figures 7(a) to 7(e), the detection unit 4 does not detect an abnormality at time t20 to time t21, but it does detect an abnormality at least from time t23 onwards.

[0153] In other words, when an abnormality is detected in at least one fuel cell cell C, the fuel cell C corresponding to the largest proportion Ri can be identified as the fuel cell cell C experiencing the abnormality.

[0154] Therefore, when the detection unit 4 detects that an abnormality has occurred in at least one of the fuel cell cells C1 to C3, it determines that the abnormality is in the fuel cell C where the proportion Ri of the acquisition time within the total acquisition time of the histogram is the largest. For example, when the detection unit 4 detects that an abnormality has occurred in at least one fuel cell C from time t23 onwards in Figure 7(f), and determines that the proportion Ri1 is the largest, it identifies fuel cell C1 corresponding to the proportion Ri1 as the fuel cell C where the abnormality has occurred.

[0155] Furthermore, the identification of the faulty fuel cell C can be used to review the usage of the fuel cell module (FCM) and to understand its condition. Additionally, the identification of the faulty fuel cell C may be used to improve the quality during maintenance and manufacturing of the fuel cell module (FCM).

[0156] Furthermore, the present invention is not limited to the embodiments described above, and various improvements and modifications are possible without departing from the spirit of the invention.

[0157] <Variation> In the above embodiment, the voltage of the fuel cell cell C is used to determine an index that shows the change in the histogram, but the operating data used to determine the index that shows the change in the histogram is not particularly limited.

[0158] For example, the detection unit 4 may be configured to create a histogram using the pressure P over a certain period of time Tc2 minutes, and to detect an abnormality (e.g., an increase in pressure loss) in the hydrogen circulation pump HP based on an index showing the change in the histogram.

[0159] Even with this configuration, by setting Tc2 for a certain period of time so that a pressure P sufficient to change the index indicating the histogram change when an abnormality occurs in the hydrogen circulation pump HP can be acquired, it is possible to detect an abnormality in the hydrogen circulation pump HP based on the index indicating the histogram change, even if the time for which pressure P can be stably acquired is relatively short. [Explanation of Symbols]

[0160] 1. Fuel cell system 2. Power generation control unit 3 Acquisition part 4. Detection Unit FCM Fuel Cell Module Lo load FCS Fuel Cell Stack VD Voltage Detection Unit HT fuel tank INJ Injector sp pressure sensor GLS gas-liquid separator HP Hydrogen Circulation Pump EDV Exhaust Drain Valve DIL Diluent ACP Air Compressor ARV Air Pressure Regulating Valve R Radiator WP Water Pump St temperature sensor CNV DC-DC converter B Energy storage device Str1, Str2 storage device Com1, Com2 communication device Cnt control unit SV Information Processing Device Pro Processor

Claims

1. A fuel cell system comprising a fuel cell module, An acquisition unit that acquires the operating data of the fuel cell module at first fixed time intervals, A detection unit that, using the operational data acquired by the acquisition unit during a second fixed time period longer than the first fixed time period, creates a histogram on a two-dimensional coordinate system where the horizontal axis represents the size of the operational data and the vertical axis represents the length of the acquisition time of the operational data, and detects that an abnormality has occurred in the fuel cell module based on an index showing the change in the histogram. A fuel cell system equipped with the following features.

2. A fuel cell system according to claim 1, The fuel cell module comprises a fuel cell stack having a plurality of fuel cell cells, The acquisition unit acquires the voltage of each of the plurality of fuel cell cells at the first fixed time intervals, The detection unit uses the voltages acquired by the acquisition unit during the second fixed time to create a histogram on a two-dimensional coordinate system with the horizontal axis representing the voltage magnitude and the vertical axis representing the voltage acquisition time. When the average value of the histogram falls below a threshold, the detection unit detects that an abnormality has occurred in at least one of the multiple fuel cell cells. Fuel cell system.

3. A fuel cell system according to claim 1, The fuel cell module comprises a fuel cell stack having a plurality of fuel cell cells, The acquisition unit acquires the voltage of each of the plurality of fuel cell cells at the first fixed time intervals, The detection unit uses the voltages acquired by the acquisition unit during the second fixed time to create a histogram on a two-dimensional coordinate system with the horizontal axis representing the voltage magnitude and the vertical axis representing the voltage acquisition time. When the median of the histogram falls below a threshold, the detection unit detects that an abnormality has occurred in at least one of the multiple fuel cell cells. Fuel cell system.

4. A fuel cell system according to claim 1, The fuel cell module comprises a fuel cell stack having a plurality of fuel cell cells, The acquisition unit acquires the voltage of each of the plurality of fuel cell cells at the first fixed time intervals, The detection unit uses the voltages acquired by the acquisition unit during the second fixed time to create a histogram on a two-dimensional coordinate system with the horizontal axis representing the voltage magnitude and the vertical axis representing the voltage acquisition time. When the standard deviation of the histogram exceeds a threshold, the detection unit detects that an abnormality has occurred in at least one of the multiple fuel cell cells. Fuel cell system.

5. A fuel cell system according to claim 1, The fuel cell module comprises a fuel cell stack having a plurality of fuel cell cells, The acquisition unit acquires the voltage of each of the plurality of fuel cell cells at the first fixed time intervals, The detection unit uses the voltages acquired by the acquisition unit during the second fixed time to create a histogram on a two-dimensional coordinate system with the horizontal axis representing the magnitude of the voltage and the vertical axis representing the length of the voltage acquisition time. When the ratio of the acquisition time corresponding to a voltage below a predetermined voltage in the histogram to the total acquisition time of the histogram exceeds a threshold, the detection unit detects that an abnormality has occurred in at least one of the multiple fuel cell cells. Fuel cell system.

6. A fuel cell system according to claim 1, The fuel cell module comprises a fuel cell stack having a plurality of fuel cell cells, The acquisition unit acquires the voltage of each of the plurality of fuel cell cells at the first fixed time intervals, The detection unit creates a first histogram on a two-dimensional coordinate system with the horizontal axis representing the voltage magnitude and the vertical axis representing the voltage acquisition time, using the lowest voltage acquired during the second fixed time period from among the voltages acquired by the acquisition unit, and creates a second histogram on the same two-dimensional coordinate system with the horizontal axis representing the voltage magnitude and the vertical axis representing the voltage acquisition time, using the maximum voltage acquired during the second fixed time period from among the voltages acquired by the acquisition unit, and detects that an abnormality has occurred in at least one of the plurality of fuel cell cells when the dissimilarity between the shapes of the first histogram and the second histogram exceeds a threshold. Fuel cell system.

7. A fuel cell system according to claim 1, The fuel cell module comprises a fuel cell stack having a plurality of fuel cell cells, The acquisition unit acquires the voltage of each of the plurality of fuel cell cells at the first fixed time intervals, The detection unit uses the voltages acquired by the acquisition unit during the second fixed time period to create a histogram on a two-dimensional coordinate system with the horizontal axis representing the voltage magnitude and the vertical axis representing the voltage acquisition time. When the dissimilarity between the shape of the histogram created at the first time point and the shape of the histogram created at the second time point (after the first time point) exceeds a threshold, the detection unit detects that an abnormality has occurred in at least one of the plurality of fuel cell cells. Fuel cell system.

8. A fuel cell system according to claims 2 to 7, When the detection unit detects that an abnormality has occurred in at least one of the plurality of fuel cell cells, it determines that the abnormality has occurred in the fuel cell cell whose acquisition time accounts for the largest proportion of the total acquisition time of the histogram. Fuel cell system.

Images