A hydrogen energy unmanned aerial vehicle fuel cell residual power detection method

Abstract

The application relates to the technical field of electrical performance testing, and discloses a residual power detection method based on a hydrogen energy unmanned aerial vehicle fuel cell, which comprises the following steps: monitoring the pressure data, internal temperature data and environmental temperature data of a hydrogen storage tank of the hydrogen energy unmanned aerial vehicle in real time during a current task, recording the output voltage and output current of the fuel cell at each moment, and obtaining a large amount of historical tasks; obtaining the residual hydrogen amount at the current moment; obtaining a plurality of power change modes of the current task, obtaining the local stack descending state at each moment, obtaining the stack health state at each moment of the current task, obtaining the internal temperature state at each moment, obtaining a plurality of reference working states of the current task at the current moment, obtaining the impedance and the electric energy conversion parameter of the current task at the current moment, and obtaining the residual power of the current task at the current moment. The application aims to solve the problem that the detection result is inaccurate when the residual power is determined by the hydrogen storage amount due to the battery performance aging.

Description

Technical Field

[0001] This invention relates to the field of electrical performance testing technology, and more specifically to a method for detecting the remaining power of a hydrogen-powered drone fuel cell. Background Technology

[0002] Hydrogen-powered drones, which use hydrogen fuel cells as their power source, have advantages such as long endurance, zero emissions, and strong adaptability to low temperatures. They have shown great potential in fields such as power line inspection, environmental monitoring, and emergency rescue. Due to their long endurance and strong adaptability to low temperatures, the prediction of the lifespan of the hydrogen-powered drone fuel cell and the detection of the remaining power are key factors affecting the reliability and practicality of the drone. Timely detection of the remaining power can effectively ensure the long-endurance operation of hydrogen-powered drones.

[0003] Normally, the remaining power of a hydrogen-powered drone fuel cell is detected based on the pressure inside the hydrogen storage tank. However, due to their strong adaptability to extreme environments and typically long range, hydrogen-powered drones inevitably cause wear and tear on the fuel cell. Therefore, the health status of the fuel cell stack and the impact of extreme environments on the membrane electrode assembly and catalyst activity will affect the fuel cell stack efficiency, which in turn affects the determination of the remaining power based on the remaining hydrogen. It is necessary to consider changes in stack efficiency in real time when detecting the remaining power. Summary of the Invention

[0004] This invention provides a method for detecting the remaining power of a hydrogen-powered drone fuel cell, to solve the problem that existing methods for determining remaining power based on hydrogen storage are inaccurate due to battery aging. The specific technical solution adopted is as follows:

[0005] This invention proposes a method for detecting the remaining power of a hydrogen-powered drone fuel cell, the method comprising the following steps:

[0006] The system monitors the hydrogen storage tank pressure, internal temperature, and ambient temperature data in real time for the current mission of the hydrogen-powered drone, records the output voltage and output current of the fuel cell at various times, and acquires a large amount of historical mission data.

[0007] Based on the pressure and internal temperature data of the hydrogen storage tank, the remaining hydrogen quantity at the current moment is obtained; based on the differences in output voltage and output current changes at different times of the current task, several power change modes and their rated power of the current task are identified, and then the local stack decline state at each moment is obtained; the rated power decline trend of each power change mode is analyzed, and combined with the temporal distribution of the power change mode in the current task, the stack health status at each moment of the current task is obtained.

[0008] The time lag correlation between the current mission's internal temperature data and the ambient temperature data is analyzed, and the internal temperature data is corrected to obtain the internal temperature state at each moment. Combining the stack health status and historical missions, several reference operating states of the current mission at the current moment are obtained. Based on the changes in the output voltage and output current of the current mission up to the current moment, the impedance and energy conversion parameters of the current mission at the current moment are obtained.

[0009] Based on the remaining hydrogen quantity of the current mission at the current moment, combined with impedance and electrical energy conversion parameters, the remaining electrical energy of the current mission at the current moment can be obtained.

[0010] Optionally, the remaining amount of hydrogen at the current moment can be obtained using the following method:

[0011] The volume of the hydrogen storage tank, the current pressure data of the hydrogen storage tank, and the internal temperature data are input into the gas state equation to obtain the amount of hydrogen in the hydrogen storage tank at the current moment, which is taken as the amount of remaining hydrogen at the current moment.

[0012] Optionally, the specific method for obtaining several power change modes and their rated power for the current task includes:

[0013] The product of the output voltage and output current at any given moment in the current task is taken as the output power at that moment. A coordinate system is constructed with time on the horizontal axis and power on the vertical axis. Based on the output power at each moment in the current task, the output power-time curve of the current task is obtained. The slope of the output power at adjacent moments is calculated and taken as the degree of power change at the next moment.

[0014] Density clustering is performed on all moments of the current task. The distance metric is the absolute value of the difference between the power change levels at each moment, resulting in several clusters. The time interval formed by consecutively distributed moments in the same cluster is taken as a power change pattern.

[0015] The maximum output power in any power change mode is taken as the rated power of that power change mode.

[0016] Optionally, the specific method for obtaining the local stack degradation state at each time step includes:

[0017] For any power change mode, if the power change at all times in the power change mode is less than or equal to 0, the power change mode is a declining power change mode; for any time, the ratio of the absolute value of the mean of the power change at all times from the first time to the current time in the power change mode to the standard deviation is obtained, and the product of the ratio and the power deviation at the current time is taken as the local stack declining state at the current time.

[0018] If the power change rate at all times in the power change mode is greater than or equal to 0, the power change mode is an upward power change mode. For any given time, the ratio of the mean to the standard deviation of the power change rate at all times from that time to the last time in the power change mode is obtained, and the product of this ratio and the power deviation at that time is taken as the local stack decline state at that time.

[0019] Optionally, the specific method for obtaining the stack health status at each moment of the current task includes:

[0020] In the output power-time curve, the rated power and the first moment of the corresponding power change mode in each power decrease change mode are extracted as the rated power coordinate points of each power decrease change mode. The least squares method is used to fit a straight line to each rated power coordinate point and the fitting slope is obtained.

[0021] Based on the fitting slope and rated power of the power decrease mode, combined with the rated power and time sequence distribution of the power increase mode, the overall stack decrease state of each power decrease mode is obtained.

[0022] The overall stack decline state of each rising power change mode is set to 1. The product of the local stack decline state at any time and the overall stack decline state of the power change mode at that time is taken as the stack health state at that time.

[0023] Optionally, the specific method for obtaining the overall stack decline state of each power decline mode includes:

[0024]

[0025]

[0026] in, Indicates the first The fitted slope of each power change pattern, This indicates the rated power of the first power change mode of the current task. Indicates the first Rated power for each power reduction variation mode; Indicates the first The power increase of a decreasing power change pattern Indicates the current task up to the [number]th [number]. The number of rising power change modes preceding each falling power change mode. Indicates the first The rated power corresponding to each power decrease change mode at different times. Indicates the current task up to the [number]th [number]. Before the first power change pattern The rated power corresponding to each rising power change mode at a given time. Indicates the current task up to the [number]th [number]. Before the first power change pattern Rated power for each rising power change mode, Indicates up to the number Before the first power change pattern The rated power of the decreasing power change mode preceding the increasing power change mode.

[0027] Optionally, the specific method for obtaining the internal temperature state at each time point includes:

[0028] Analyze the changes in the temporal lag correlation between the current task's internal temperature data and the ambient temperature data, and obtain several lag time quantities and their correlation coefficients;

[0029] Arrange the correlation coefficients in ascending order of lag time to obtain the temperature-related change sequence. Obtain several maxima in the temperature-related change sequence and use the lag time corresponding to the maximum value among all maxima as the lag interference factor.

[0030] For any given moment, the ambient temperature data at that moment is obtained. The ambient temperature data at that moment is obtained by subtracting the hysteresis interference factor from the ambient temperature data at that moment, and is used as the reference ambient temperature data at that moment. The difference between the ambient temperature data at that moment and the reference ambient temperature data is obtained. The temperature data obtained by subtracting the difference from the internal temperature data at that moment is used as the internal temperature state at that moment.

[0031] Optionally, the specific method for obtaining the aforementioned lag time quantities and their correlation coefficients is as follows:

[0032] Based on the ambient temperature data and the internal temperature data of the hydrogen storage tank at each moment of the current mission, the initial ambient temperature sequence and the initial internal temperature sequence of the current mission are obtained. The Pearson correlation coefficient between the two temperature sequences is calculated as the initial correlation coefficient. The first internal temperature data of the initial internal temperature data sequence and the first ambient temperature data of the initial ambient temperature sequence are removed respectively, and the first correlation coefficient is recalculated. This process is repeated to remove the preceding and following elements in the temperature sequence and calculate the second correlation coefficient until two temperature data remain in both temperature sequences, at which point the correlation coefficient update calculation is stopped.

[0033] For the first correlation coefficient, the product of the total number of elements removed from the corresponding single temperature sequence and the acquisition time interval is used as the lag time corresponding to the first correlation coefficient, thus obtaining the lag time corresponding to each correlation coefficient.

[0034] Optionally, the specific method for obtaining several reference working states of the current task at the current moment includes:

[0035] A two-dimensional sample space is constructed based on the fuel cell health status and internal temperature status. The fuel cell health status and internal temperature status of the current task at the current moment are mapped to sample points in the two-dimensional sample space. A large number of fuel cell health status and internal temperature status at various moments of historical tasks are obtained and mapped to the two-dimensional sample space to obtain a number of sample points. Density clustering is performed on all sample points, and the distance metric is the Euclidean distance between sample points. The time corresponding to all sample points in the cluster to which the sample point of the current task at the current moment belongs is used as the reference working state of the current task at the current moment.

[0036] Optionally, the impedance and energy conversion parameters of the current task at the current moment can be obtained using the following method:

[0037] Obtain the ratio of output voltage to output current at any moment in the current task, and use it as the load resistance at that moment. Arrange the load resistances at each moment up to the current moment in chronological order to obtain the resistance change sequence at the current moment. Obtain the power deviation at each moment up to the current moment in the current task, and arrange them in chronological order to obtain the power deviation sequence at the current moment. Obtain the resistance change sequence and power deviation sequence for each reference operating state at the current moment.

[0038] For any reference operating state, obtain the Euclidean distance between the sample point corresponding to the reference operating state and the sample point corresponding to the current time, and obtain the DTW distance between the resistance change sequence of the reference operating state and the resistance change sequence of the current time. The product of the Euclidean distance and the DTW distance is inversely normalized and used as the impedance reference coefficient of the reference operating state. The impedance reference coefficients of all reference operating states are weighted and normalized, and the result is used as the impedance reference weight of each reference operating state. The load resistances at the corresponding times of all reference operating states are weighted and summed according to the impedance reference weights, and the result is used as the impedance of the current task at the current time.

[0039] For any reference operating state, based on the power deviation sequence of that reference operating state and the power deviation sequence at the current moment, combined with the Euclidean distance between the corresponding sample points, the power reference coefficient of that reference operating state is obtained, and then the power reference weight of each reference operating state is obtained. The power deviation of all reference operating states at the corresponding moment is weighted and summed according to the power reference weight, and the result is used as the corrected power deviation of the current task at the current moment. The ratio of the corrected power deviation to the rated power of the power change mode at the current moment is obtained, and the difference obtained by subtracting the ratio from 1 is used as the power conversion parameter of the current task at the current moment.

[0040] The beneficial effects of this invention are as follows: This invention analyzes the power change of the output voltage and current changes during the current task, quantifies the stack health status at each moment based on the local and overall power decline trends, and considers the lag effect of ambient temperature on internal temperature to obtain the internal temperature status. This allows for the acquisition of a reference operating state for the current moment from historical tasks. Furthermore, based on the resistance and power changes exhibited by the output voltage and current, the impedance and catalyst activity of the reference operating state are comprehensively analyzed to obtain impedance and energy conversion parameters for the current moment. Specifically, the remaining hydrogen quantity is obtained in real time using the gas state equation. The output power change of the current task is analyzed based on the output voltage and current, and the power change mode is divided according to the output power change trend, and its rated power is obtained. The deviation between the output power and the rated power within the power change mode is analyzed, and the local stack decline state is quantified by combining the power change trend, reflecting the power... The study examines the performance of the fuel cell stack in different power states. Simultaneously, the rated power change in the overall declining power change mode reflects the overall decline in stack status and eliminates the influence of the rising power change mode, thus obtaining the stack health status based on the overall declining stack status. Considering the lag effect of ambient temperature on the internal temperature of the hydrogen storage tank, the study adjusts and obtains the internal temperature status. Based on cluster analysis of the stack health status and internal temperature status in historical missions, a reference operating state is selected. Based on the resistance and power changes of the reference operating state, the current impedance and catalyst activity are weighted to obtain the resistance and power, thereby reflecting the current impedance and energy conversion parameters. Combined with the remaining hydrogen quantity, the study estimates the remaining energy, eliminating the influence of stack performance and health status on remaining energy detection under long-duration or extreme environmental conditions for hydrogen-powered drones, thus improving the accuracy of remaining energy detection for hydrogen-powered drone fuel cells. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a schematic flowchart of a method for detecting the remaining power of a hydrogen-powered drone fuel cell, provided as an embodiment of the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Please see Figure 1 The diagram illustrates a flowchart of a method for detecting the remaining power of a hydrogen-powered drone fuel cell, provided by an embodiment of the present invention. The method includes the following steps:

[0045] Step S001: Monitor the hydrogen storage tank pressure data, internal temperature data and ambient temperature data in real time for the current mission of the hydrogen-powered drone, record the output voltage and output current of the fuel cell at each moment, and obtain a large amount of historical mission data.

[0046] The purpose of this embodiment is to reduce the interference of complex environmental conditions on the remaining power detection of hydrogen-powered drone fuel cells, based on the detection of remaining power through the remaining hydrogen quantity, and considering the impact of battery stack health status and performance aging. This requires real-time monitoring of hydrogen storage tank pressure data and internal temperature data to estimate the remaining hydrogen quantity, while monitoring the ambient temperature of the hydrogen-powered drone to account for changes in extreme external environments. Battery stack analysis requires real-time monitoring of the voltage and current of the hydrogen-powered drone's fuel cells.

[0047] Specifically, during the execution of its current mission, the hydrogen-powered drone periodically monitors the output voltage and current of its fuel cell at 1-minute intervals using voltage and current sensors built into the fuel cell. The fuel cell is connected to the drone's hydrogen storage tank, which is equipped with pressure and temperature sensors. The pressure sensor monitors the internal pressure of the tank in real time, providing pressure data, while the temperature sensor monitors the internal temperature, providing internal temperature data, also at 1-minute intervals. An external temperature sensor on the drone collects the external temperature data in real time, providing ambient temperature data, also at 1-minute intervals. This provides the data for the hydrogen storage tank pressure, internal temperature, ambient temperature, output voltage, and output current at various times.

[0048] Furthermore, a large number of historical missions completed by the same type of hydrogen-powered drones were acquired, and the internal temperature data, ambient temperature data, output voltage and output current at each moment of the historical missions were recorded. The sampling time interval of the historical missions was the same as that of the current mission.

[0049] It should be noted that while the remaining hydrogen quantity can be estimated using the hydrogen storage tank pressure, changes in the working environment and the duration of continuous operation will affect the fuel cell performance. Therefore, it is necessary to analyze the current health status of the fuel cell stack, quantifying its performance by analyzing the trends in output voltage and current. Furthermore, the analysis of the current mission's operating status also needs to consider the ambient temperature and humidity to reflect the extreme environmental conditions encountered by the hydrogen-powered drone. The stack impedance status needs to be obtained through high-frequency impedance testing, which is not suitable for impedance status analysis under the operating conditions of the hydrogen-powered drone. However, similar operating conditions will result in similar stack performance and thus similar impedance status, allowing the acquisition of the current mission's impedance status. Based on the remaining hydrogen quantity and the impedance status, the remaining power can be estimated.

[0050] Step S002: Based on the pressure data and internal temperature data of the hydrogen storage tank, obtain the remaining hydrogen quantity at the current moment; based on the differences in output voltage and output current changes at various moments of the current task, divide the current task into several power change modes and their rated power, and then obtain the local stack decline state at each moment; analyze the rated power decline trend of each power change mode, and combine the time sequence distribution of the power change mode in the current task to obtain the stack health status at each moment of the current task.

[0051] Preferably, in one embodiment of the present invention, the method for obtaining the remaining hydrogen quantity at the current moment based on the hydrogen storage tank pressure data and internal temperature data includes:

[0052] It should be noted that, since the volume of the hydrogen storage tank is fixed and the pressure and internal temperature of the hydrogen storage tank are monitored in real time, the amount of remaining hydrogen at the current moment can be obtained through the gas state equation.

[0053] Specifically, the volume of the hydrogen storage tank, the current pressure data of the hydrogen storage tank, and the internal temperature data are input into the gas state equation to obtain the amount of hydrogen in the hydrogen storage tank at the current moment, which is taken as the amount of remaining hydrogen at the current moment. The gas state equation is a well-known technology and will not be described in detail in this embodiment.

[0054] It should be further noted that the changing trends of output voltage and output current can reflect the changes in the output power of the fuel cell stack. Long-term use will lead to changes in the health status of the fuel cell stack and a decrease in output power, which is reflected in a certain degree of decrease in both voltage and current. By analyzing the decrease in voltage and current, we can analyze the overall changing trend of voltage and current under the current mission, as well as the impact of the real-time working status of the hydrogen-powered UAV on the health status of the fuel cell stack. That is, the decrease in different working states under the current mission is different, so as to obtain the real-time health status of the fuel cell stack.

[0055] Preferably, in one embodiment of the present invention, based on the differences in output voltage and output current changes at various times during the current task, several power change modes and their rated power for the current task are determined, thereby obtaining the local stack degradation state at each time. The specific method includes:

[0056] The product of the output voltage and output current at any given moment in the current task is taken as the output power at that moment. A coordinate system is constructed with time on the horizontal axis and power on the vertical axis. Based on the output power at each moment in the current task, the output power-time curve of the current task is obtained. The slope of the output power at adjacent moments is calculated and used as the degree of power change at the next moment. The slope calculation is based on the output power at adjacent moments and the corresponding moments, which is a well-known technique and will not be described in detail in this embodiment. It is worth noting that the degree of power change at the first moment in the current task is set as the degree of power change at the second moment. Density clustering is performed on all moments of the current task using DBSCAN clustering. The distance metric is the absolute value of the difference between the degree of power change at each moment, resulting in several clusters. The time period consisting of consecutive moments in the same cluster is taken as a power change pattern. That is, one cluster may correspond to multiple power change patterns. Thus, the time period consisting of several consecutive moments yields several power change patterns.

[0057] Furthermore, for any power change mode, if the degree of power change at all times in the power change mode is greater than or equal to 0, the power change mode is an upward power change mode. The maximum output power in the power change mode (the output power at the last moment) is taken as the rated power of the power change mode segment, and the difference between the output power and the rated power at each moment in the power change mode segment (the difference is obtained by subtracting the output power from the rated power) is taken as the power deviation at each moment.

[0058] Similarly, if the power change is less than or equal to 0 at all times in the power change mode, the power change mode is a decreasing power change mode. The maximum output power in the power change mode (the output power at the first time) is taken as the rated power of the power change mode segment, and the difference between the output power and the rated power at each time in the power change mode segment (the difference is obtained by subtracting the output power from the rated power) is taken as the power deviation at each time.

[0059] Furthermore, for any moment in any power decline mode, the ratio of the absolute value of the mean of the power change degree from the first moment to that moment (including the first moment and that moment) to the standard deviation is obtained, and the product of the ratio and the power deviation at that moment is taken as the local stack decline state at that moment; it should be noted that the local stack decline state at the first moment in the power change mode is directly set to 0.

[0060] Similarly, for any moment in any rising power change mode, the ratio of the mean to the standard deviation of the power change degree of all moments (including the last moment and the current moment) from that moment to the last moment in that power change mode is obtained, and the product of the ratio and the power deviation at that moment is taken as the local stack decline state at that moment; it should be noted that the local stack decline state at the last moment in the power change mode is directly set to 0.

[0061] It should be noted that in local power change patterns, the power deviation can initially reflect the decrease in output power. A larger average absolute value of the power change and a similar decreasing trend indicate that the stack state is steadily decreasing under this power change pattern, exhibiting a significant decrease in stack performance, and thus the local stack decline is greater. Similarly, for an increase in output power, a larger average power change and a smaller fluctuation in the increasing trend indicate a stable and significant increase in power under this power change pattern, and the greater the influence of the previous decreasing trend in the power change pattern, the greater the corresponding decrease in stack performance.

[0062] It should be noted that, in order to prevent the denominator from being 0 in this embodiment, a hyperparameter is added to both the numerator and denominator when calculating the ratio. In this embodiment, the hyperparameter is described as 0.1.

[0063] It should be further explained that after obtaining the local stack decline state, it is necessary to further analyze the overall stack health state based on the rated power of each power change mode. The continuous decline of rated power reflects the decline of stack health state. At the same time, it is necessary to quantify the impact of the rising rated power, eliminate the influence of the rising stack health state, and thus quantify the overall decline performance of stack health state.

[0064] Preferably, in one embodiment of the present invention, the rated power decrease trend of each power change mode is analyzed, and the stack health status at each moment of the current task is obtained by combining the time-series distribution of the power change modes. The specific method includes:

[0065] In the output power-time curve, the rated power and the first moment of each power change mode in each power decrease mode are extracted as the rated power coordinate points of each power decrease mode. The least squares method is used to fit a straight line to each rated power coordinate point, and the fitting slope is obtained. The least squares method is a well-known technique and will not be elaborated upon in this embodiment. Starting from the rated power coordinate point of the second power decrease mode, a straight line is fitted from the rated power coordinate point of the first power decrease mode to the rated power coordinate point of the second power decrease mode to obtain the fitting slope of the second power decrease mode, and so on. The fitting slope of the first power decrease pattern is the distance from the rated power coordinate point of the first power decrease pattern to the first power decrease pattern. The fitting slope is obtained by fitting the rated power coordinate points of each power decrease pattern. It should be noted that the fitting slope for the first power decrease pattern is set to 0; then the... Overall stack decline state of each power decline mode The calculation method is as follows:

[0066]

[0067]

[0068] in, Indicates the first The fitted slope of each power change pattern, This indicates the rated power of the first power change mode of the current task. Indicates the first Rated power for each power reduction variation mode; Indicates the first The power increase of a decreasing power change pattern Indicates the current task up to the [number]th [number]. The number of rising power change modes preceding each falling power change mode. Indicates the first The rated power corresponding to each power decrease change mode at different times. Indicates the current task up to the [number]th [number]. Before the first power change pattern The rated power corresponding to each rising power change mode at a given time. Indicates the current task up to the [number]th [number]. Before the first power change pattern Rated power for each rising power change mode, Indicates up to the number Before the first power change pattern The rated power of the decreasing power change mode preceding the increasing power change mode.

[0069] It should be noted that the rated power for the decreasing power change mode is compared with the rated power for the first power change mode in the current mission. This represents the initial rated power for the current task, which is unaffected by the degradation of the fuel cell stack performance. The larger the deviation, the greater the overall degradation of the fuel cell stack performance under the power degradation mode. Simultaneously, the smaller the negative and smaller the fitting slope, the greater the degradation of the rated power and the greater the overall degradation state of the fuel cell stack. The earlier the power degradation mode precedes the power degradation mode, and the closer its time distribution is to the power degradation mode, the greater the rated power is compared to the power degradation mode, and the more it needs to be corrected to eliminate the influence of the power degradation. Then, the power degradation amount is weighted and corrected to finally obtain the overall degradation state of the fuel cell stack.

[0070] Furthermore, the overall stack decline state of each rising power change mode is set to 1, that is, the overall stack performance change of rising power change mode is not considered, and the local stack decline state is directly analyzed. The product of the local stack decline state at any time and the overall stack decline state of the power change mode at that time is taken as the stack health state at that time, thus obtaining the stack health state at each time of the current task.

[0071] Thus, the remaining hydrogen quantity is obtained in real time through the gas state equation, the output power change of the current task is analyzed based on the output voltage and output current, and the power change mode is divided according to the output power change trend and its rated power is obtained. The deviation between the output power and the rated power within the power change mode is analyzed, and the local stack decline state is quantified in combination with the power change trend, reflecting the stack state decline performance within the power change mode. At the same time, the rated power change of the overall declining power change mode can reflect the overall stack state decline performance and eliminate the influence of the rising power change mode. Therefore, the stack health state is obtained based on the overall stack decline state.

[0072] Step S003: Analyze the time lag correlation between the current task's internal temperature data and the ambient temperature data, correct the internal temperature data to obtain the internal temperature state at each moment; combine the stack health status and historical tasks to obtain several reference operating states of the current task at the current moment; for the current task's output voltage and output current up to the current moment, based on the changes in output voltage and output current of the reference operating states, obtain the impedance and power conversion parameters of the current task at the current moment.

[0073] It should be noted that, due to the complex and variable working environment of hydrogen-powered drones, including extreme environmental conditions and drastic temperature changes at high altitudes (such as rapidly ascending from 30°C on the ground to -20°C at high altitude), the change in the working environment of the hydrogen-powered drone will affect the internal temperature of the hydrogen storage tank. Therefore, in order to obtain historical mission data similar to the current working conditions at the current moment, it is necessary to further consider the influence of ambient temperature on the fuel cell stack health status. However, the influence of ambient temperature on the internal temperature of the hydrogen storage tank has a lag. By eliminating the lag, the influence of the current ambient temperature on the internal temperature can be quantified, and combined with the fuel cell stack health status, the working status of the current mission at the current moment can be obtained.

[0074] Preferably, in one embodiment of the present invention, the method for analyzing the temporal lag correlation between the current task's internal temperature data and the ambient temperature data, and correcting the internal temperature data to obtain the internal temperature state at each moment, includes:

[0075] Based on the ambient temperature data and the internal temperature data of the hydrogen storage tank at each moment of the current mission, the initial ambient temperature sequence and the initial internal temperature sequence of the current mission are obtained. The Pearson correlation coefficient between the two temperature sequences is calculated and used as the initial correlation coefficient. The first internal temperature data is removed from the initial internal temperature sequence to obtain the first internal temperature sequence. The last internal temperature data is removed from the initial ambient temperature sequence to obtain the first ambient temperature sequence. The Pearson correlation coefficient between the first internal ambient temperature sequence and the first ambient temperature sequence is recalculated and used as the first correlation coefficient. This process is repeated for each internal temperature sequence, removing the first internal ambient temperature data (the first data in the initial internal temperature sequence) from the first internal temperature sequence. (Two), for the first ambient temperature sequence, remove the last ambient temperature data (the second to last in the initial ambient temperature sequence), recalculate the Pearson correlation coefficient and use it as the second correlation coefficient, and so on, until only two temperature data remain in both temperature sequences, at which point the correlation coefficient update calculation stops; then for the first correlation coefficient, multiply the total number of elements removed from the corresponding single temperature sequence by the acquisition time interval, and use this as the lag time corresponding to the first correlation coefficient, where the total number of elements removed for the first correlation coefficient is 1, the total number of elements removed for the second correlation coefficient is 2, and so on, to obtain the lag time corresponding to each correlation coefficient.

[0076] Furthermore, the correlation coefficients are arranged in ascending order of lag time to obtain a temperature-related change sequence. Several maxima are obtained from the temperature-related change sequence, and the lag time corresponding to the maximum value among all maxima is used as the lag interference factor. For the internal temperature data at any given time, the ambient temperature data at that time is obtained. The ambient temperature data corresponding to the time obtained by subtracting the lag interference factor from the current time is used as the reference ambient temperature data at that time. The difference between the ambient temperature data at that time and the reference ambient temperature data is obtained. The temperature data obtained by subtracting the difference from the internal temperature data at that time is used as the internal temperature state at that time. It should be noted that if the time obtained by subtracting the lag interference factor from the current time is outside the time range of the current task, that is, the corresponding time does not exist in the time range of the current task, then the internal temperature data at that time is directly used as the internal temperature state at that time.

[0077] It should be noted that by removing the first few internal temperature data points from the internal temperature sequence and the last few environmental temperature data points from the environmental temperature sequence, and comparing the correlation coefficients before and after removal, the highest correlation coefficient indicates that the overall temperature change after removal shows the strongest correlation. This means that the removed part is likely to be lagging, thus obtaining the lag time and correcting the internal temperature data.

[0078] It should be further noted that a large number of historical missions performed by the same type of hydrogen-powered drones, as well as the testing process under extreme environments, can be used as reference missions for analysis of the current mission. Since the water management status and catalyst activity of the fuel cell membrane electrode cannot be monitored in real-time flight missions, it is necessary to use the impedance status determined by reference to similar working states to infer the impedance status of the current mission at the current moment.

[0079] Preferably, in one embodiment of the present invention, several reference operating states of the current task at the current moment are obtained by combining the stack health status and historical tasks, including the following specific method:

[0080] A two-dimensional sample space is constructed based on the fuel cell health status and internal temperature status. The fuel cell health status and internal temperature status of the current task at the current moment are mapped to sample points in the two-dimensional sample space. At the same time, the fuel cell health status and internal temperature status of a large number of historical tasks at various moments are obtained and mapped to the two-dimensional sample space to obtain a number of sample points. DBSCAN clustering is performed on all sample points, and the Euclidean distance between sample points is used as the distance metric. The time corresponding to all sample points in the cluster to which the sample point of the current task at the current moment belongs (a number of time points in a number of historical tasks) is used as the reference working state of the current task at the current moment.

[0081] It should be noted that cluster analysis based on working state is used, with the working state of the reference task at the corresponding moment in the same cluster serving as the reference for the current task at the current moment. Based on the distance between data points in the cluster, since it is an impedance state analysis, the activity of the membrane electrode and catalyst can be expressed by load changes and power changes quantified by output voltage and current, respectively. Therefore, the similarity of load changes and the similarity of the deviation of output power from normal power are used as limiting weights to obtain the reference weights of each sample point, and the corresponding impedance state is weighted and inferred.

[0082] Preferably, in one embodiment of the present invention, the method for obtaining the impedance and power conversion parameters of the current task at the current moment based on the changes in the output voltage and output current of the current task up to the current moment, according to the changes in the output voltage and output current of the reference operating state, includes the following specific method:

[0083] Obtain the ratio of output voltage to output current at any moment in the current task, and use it as the load resistance at that moment. Arrange the load resistances at all moments up to the current moment in chronological order to obtain the resistance change sequence at the current moment. Obtain the power deviation at each moment up to the current moment in the current task, and arrange them in chronological order to obtain the power deviation sequence at the current moment. Obtain the resistance change sequence and power deviation sequence for each reference operating state at the current moment. The method is to obtain the sequence composed of the load resistance time sequence up to the corresponding moment and the sequence composed of the power deviation time sequence in the historical task where each reference operating state is located.

[0084] Furthermore, for any reference operating state, the Euclidean distance between the sample point corresponding to the reference operating state and the sample point corresponding to the current time is obtained, and the DTW distance between the resistance change sequence of the reference operating state and the resistance change sequence of the current time is obtained. The product of the Euclidean distance and the DTW distance is inversely normalized and used as the impedance reference coefficient of the reference operating state. The impedance reference coefficients of all reference operating states are weighted and normalized, and the result is used as the impedance reference weight of each reference operating state. The load resistances at the corresponding times of all reference operating states are weighted and summed according to the impedance reference weights, and the result is used as the impedance of the current task at the current time.

[0085] Furthermore, for any reference operating state, the DTW distance between the power deviation sequence of that reference operating state and the power deviation sequence at the current time is obtained. The product of this distance and the Euclidean distance is inversely normalized to obtain the power reference coefficient for that reference operating state. The power reference coefficients of all reference operating states are weighted and normalized to obtain the power reference weight for each reference operating state. The power deviations at corresponding times of all reference operating states are weighted and summed according to the power reference weights to obtain the corrected power deviation for the current task at the current time. The ratio of the corrected power deviation to the rated power of the power change mode at the current time is obtained. The difference obtained by subtracting the ratio from 1 is used as the power conversion parameter for the current task at the current time.

[0086] It should be noted that the resistance and power deviations at the current moment are obtained by weighting the values ​​based on the resistance and power deviations at the reference operating state. This means that since the voltage and current are still changing at the current moment, the resistance and power deviations may change subsequently and cannot directly reflect the impedance and catalyst activity at the current moment. Therefore, it is necessary to refer to the operating state to obtain the values. At the same time, the weighted quantification is performed based on the Euclidean distance between the sample points and the DTW distance corresponding to the resistance change sequence and the power deviation sequence. The smaller the Euclidean distance, the closer the health state and internal temperature state of the fuel cell stack are, and the greater the reference value. The smaller the DTW distance, the closer the voltage and current change patterns are, further enhancing the confidence of the fuel cell stack state changes.

[0087] It should be noted that this embodiment adopts... The model is used to represent the inverse proportional relationship and for normalization processing. This represents an exponential function with the natural constant as its base. As input to the model, implementers can set inverse proportional functions and normalization functions according to the actual situation.

[0088] Therefore, considering the lag effect of ambient temperature on the internal temperature of the hydrogen storage tank, the internal temperature status is adjusted and obtained. Based on the cluster analysis of the stack health status and internal temperature status in historical missions, a reference operating state is selected. Based on the resistance and power changes of the reference operating state, the current impedance and catalyst activity are weighted and obtained to reflect the current impedance and power conversion parameters.

[0089] Step S004: Based on the remaining hydrogen amount of the current task at the current moment, combined with impedance and power conversion parameters, obtain the remaining power of the current task at the current moment.

[0090] It should be noted that the impact of the stack health status on the estimation of remaining electricity from the remaining hydrogen quantity is mainly manifested in the changes in impedance and catalyst activity. When the impedance and energy conversion parameters (catalyst activity) are known, the remaining electricity can be estimated by combining the remaining hydrogen quantity.

[0091] Specifically, based on the existing hydrogen fuel cell power estimation model, the remaining hydrogen quantity at the current moment of the current task is input, and the impedance and power conversion parameters are used as the load impedance and catalyst activity at the current moment as inputs into the model. The estimated remaining power quantity at the current moment is then output. The hydrogen fuel cell power estimation model based on the remaining hydrogen quantity, load impedance and catalyst activity is existing technology and will not be described in detail in this embodiment.

[0092] Thus, by analyzing the power change of the output voltage and current changes during the current mission, the health status of the fuel cell stack at each moment is quantified based on the local power decrease and the overall power decrease trend. The internal temperature status is obtained by considering the lag effect of ambient temperature on internal temperature. This allows the reference operating state at the current moment to be obtained from historical missions. Furthermore, based on the resistance and power changes shown by the output voltage and current, the impedance and catalyst activity of the reference operating state are comprehensively analyzed. The impedance and energy conversion parameters at the current moment are obtained, and the remaining power is estimated by combining the remaining hydrogen quantity. This eliminates the influence of fuel cell stack performance and health status on the remaining power detection under long-endurance or extreme environment operation of hydrogen-powered drones, and improves the accuracy of fuel cell remaining power detection for hydrogen-powered drones.

[0093] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for detecting the remaining power of a hydrogen-powered drone fuel cell, characterized in that, The method includes the following steps: The system monitors the hydrogen storage tank pressure, internal temperature, and ambient temperature data in real time for the current mission of the hydrogen-powered drone, records the output voltage and output current of the fuel cell at various times, and acquires a large amount of historical mission data. Based on the pressure and internal temperature data of the hydrogen storage tank, the remaining hydrogen quantity at the current moment is obtained; based on the differences in output voltage and output current changes at different times of the current task, several power change modes and their rated power of the current task are identified, and then the local stack decline state at each moment is obtained; the rated power decline trend of each power change mode is analyzed, and combined with the temporal distribution of the power change mode in the current task, the stack health status at each moment of the current task is obtained. The time lag correlation between the current mission's internal temperature data and the ambient temperature data is analyzed, and the internal temperature data is corrected to obtain the internal temperature state at each moment. Combining the stack health status and historical missions, several reference operating states of the current mission at the current moment are obtained. Based on the changes in the output voltage and output current of the current mission up to the current moment, the impedance and energy conversion parameters of the current mission at the current moment are obtained. Based on the remaining hydrogen quantity of the current mission at the current moment, combined with impedance and electrical energy conversion parameters, the remaining electrical energy of the current mission at the current moment can be obtained.

2. The method for detecting remaining power based on a hydrogen-powered drone fuel cell according to claim 1, characterized in that, The remaining hydrogen quantity at the current moment is obtained using the following method: The volume of the hydrogen storage tank, the current pressure data of the hydrogen storage tank, and the internal temperature data are input into the gas state equation to obtain the amount of hydrogen in the hydrogen storage tank at the current moment, which is taken as the amount of remaining hydrogen at the current moment.

3. The method for detecting remaining power based on a hydrogen-powered drone fuel cell according to claim 1, characterized in that, The specific methods for obtaining several power change modes and their rated power for the current task include: The product of the output voltage and output current at any given moment in the current task is taken as the output power at that moment. A coordinate system is constructed with time on the horizontal axis and power on the vertical axis. Based on the output power at each moment in the current task, the output power-time curve of the current task is obtained. The slope of the output power at adjacent moments is calculated and taken as the degree of power change at the next moment. Density clustering is performed on all moments of the current task. The distance metric is the absolute value of the difference between the power change levels at each moment, resulting in several clusters. The time interval formed by consecutively distributed moments in the same cluster is taken as a power change pattern. The maximum output power in any power change mode is taken as the rated power of that power change mode.

4. The method for detecting remaining power based on a hydrogen-powered drone fuel cell according to claim 1, characterized in that, The specific methods for obtaining the local stack degradation state at each time point are as follows: For any power change mode, if the power change at all times in the power change mode is less than or equal to 0, the power change mode is a declining power change mode; for any time, the ratio of the absolute value of the mean of the power change at all times from the first time to the current time in the power change mode to the standard deviation is obtained, and the product of the ratio and the power deviation at the current time is taken as the local stack declining state at the current time. If the power change rate at all times in the power change mode is greater than or equal to 0, the power change mode is an upward power change mode. For any given time, the ratio of the mean to the standard deviation of the power change rate at all times from that time to the last time in the power change mode is obtained, and the product of this ratio and the power deviation at that time is taken as the local stack decline state at that time.

5. The method for detecting remaining power based on a hydrogen-powered drone fuel cell according to claim 4, characterized in that, The specific methods for obtaining the stack health status at each moment of the current task are as follows: In the output power-time curve, the rated power and the first moment of the corresponding power change mode in each power decrease change mode are extracted as the rated power coordinate points of each power decrease change mode. The least squares method is used to fit a straight line to each rated power coordinate point and the fitting slope is obtained. Based on the fitting slope and rated power of the power decrease mode, combined with the rated power and time sequence distribution of the power increase mode, the overall stack decrease state of each power decrease mode is obtained. The overall stack decline state of each rising power change mode is set to 1. The product of the local stack decline state at any time and the overall stack decline state of the power change mode at that time is taken as the stack health state at that time.

6. The method for detecting remaining power based on a hydrogen-powered drone fuel cell according to claim 5, characterized in that, The specific method for obtaining the overall stack decline state of each power decline mode is as follows: in, Indicates the first The fitted slope of each power change pattern, This indicates the rated power of the first power change mode of the current task. Indicates the first Rated power for each power reduction variation mode; Indicates the first The power increase of a decreasing power change pattern Indicates the current task up to the [number]th [number]. The number of rising power change modes preceding each falling power change mode. Indicates the first The rated power corresponding to each power decrease change mode at different times. Indicates the current task up to the [number]th [number]. Before the first power change pattern The rated power corresponding to each rising power change mode at a given time. Indicates the current task up to the [number]th [number]. Before the first power change pattern Rated power for each rising power change mode, Indicates up to the number Before the first power change pattern The rated power of the decreasing power change mode preceding the increasing power change mode.

7. The method for detecting remaining power based on a hydrogen-powered drone fuel cell according to claim 1, characterized in that, The specific method for obtaining the internal temperature state at each time point is as follows: Analyze the changes in the temporal lag correlation between the current task's internal temperature data and the ambient temperature data, and obtain several lag time quantities and their correlation coefficients; Arrange the correlation coefficients in ascending order of lag time to obtain the temperature-related change sequence. Obtain several maxima in the temperature-related change sequence and use the lag time corresponding to the maximum value among all maxima as the lag interference factor. For any given moment, the ambient temperature data at that moment is obtained. The ambient temperature data at that moment is obtained by subtracting the hysteresis interference factor from the ambient temperature data at that moment, and is used as the reference ambient temperature data at that moment. The difference between the ambient temperature data at that moment and the reference ambient temperature data is obtained. The temperature data obtained by subtracting the difference from the internal temperature data at that moment is used as the internal temperature state at that moment.

8. The method for detecting remaining power based on a hydrogen-powered drone fuel cell according to claim 7, characterized in that, The specific method for obtaining the aforementioned lag time quantities and their correlation coefficients is as follows: Based on the ambient temperature data and the internal temperature data of the hydrogen storage tank at each moment of the current mission, the initial ambient temperature sequence and the initial internal temperature sequence of the current mission are obtained. The Pearson correlation coefficient between the two temperature sequences is calculated as the initial correlation coefficient. The first internal temperature data of the initial internal temperature data sequence and the first ambient temperature data of the initial ambient temperature sequence are removed respectively, and the first correlation coefficient is recalculated. This process is repeated to remove the preceding and following elements in the temperature sequence and calculate the second correlation coefficient until two temperature data remain in both temperature sequences, at which point the correlation coefficient update calculation is stopped. For the first correlation coefficient, the product of the total number of elements removed from the corresponding single temperature sequence and the acquisition time interval is used as the lag time corresponding to the first correlation coefficient, thus obtaining the lag time corresponding to each correlation coefficient.

9. The method for detecting the remaining power of a hydrogen-powered drone fuel cell according to claim 1, characterized in that, The specific methods for obtaining several reference working states of the current task at the current moment include: A two-dimensional sample space is constructed based on the fuel cell health status and internal temperature status. The fuel cell health status and internal temperature status of the current task at the current moment are mapped to sample points in the two-dimensional sample space. A large number of fuel cell health status and internal temperature status at various moments of historical tasks are obtained and mapped to the two-dimensional sample space to obtain a number of sample points. Density clustering is performed on all sample points, and the distance metric is the Euclidean distance between sample points. The time corresponding to all sample points in the cluster to which the sample point of the current task at the current moment belongs is used as the reference working state of the current task at the current moment.

10. A method for detecting the remaining power of a hydrogen-powered drone fuel cell according to claim 9, characterized in that, The impedance and energy conversion parameters of the current task at the current moment are obtained using the following method: Obtain the ratio of output voltage to output current at any moment in the current task, and use it as the load resistance at that moment. Arrange the load resistances at each moment up to the current moment in chronological order to obtain the resistance change sequence at the current moment. Obtain the power deviation at each moment up to the current moment in the current task, and arrange them in chronological order to obtain the power deviation sequence at the current moment. Obtain the resistance change sequence and power deviation sequence for each reference operating state at the current moment. For any reference operating state, obtain the Euclidean distance between the sample point corresponding to the reference operating state and the sample point corresponding to the current time, and obtain the DTW distance between the resistance change sequence of the reference operating state and the resistance change sequence of the current time. The product of the Euclidean distance and the DTW distance is inversely normalized and used as the impedance reference coefficient of the reference operating state. The impedance reference coefficients of all reference operating states are weighted and normalized, and the result is used as the impedance reference weight of each reference operating state. The load resistances of all reference operating states at the corresponding time are weighted and summed according to the impedance reference weights, and the result is used as the impedance of the current task at the current time. For any reference operating state, based on the power deviation sequence of that reference operating state and the power deviation sequence at the current moment, combined with the Euclidean distance between the corresponding sample points, the power reference coefficient of that reference operating state is obtained, and then the power reference weight of each reference operating state is obtained. The power deviation of all reference operating states at the corresponding moment is weighted and summed according to the power reference weight, and the result is used as the corrected power deviation of the current task at the current moment. The ratio of the corrected power deviation to the rated power of the power change mode at the current moment is obtained, and the difference obtained by subtracting the ratio from 1 is used as the power conversion parameter of the current task at the current moment.

Images