Knowledge-data fusion driven fault diagnosis method and device for vehicle-mounted hydrogen system

Abstract

The application discloses a kind of knowledge-data fusion driven vehicle-mounted hydrogen system fault diagnosis method and device, belong to fault diagnosis technical field.The method includes: obtaining the actual hydrogen output flow of vehicle-mounted hydrogen system, and obtaining the actual hydrogen concentration at the corresponding position using the sensor arranged in the vehicle environment;Determine the current vehicle state;According to the current vehicle state, determine the current dynamic threshold value for fault diagnosis;According to the current dynamic threshold value, the actual hydrogen output flow and the actual hydrogen concentration are detected to determine whether the vehicle-mounted hydrogen system has the fault of hydrogen leakage.The calculation method of threshold value is driven by knowledge and data together, realizes the prediction of hydrogen output flow and hydrogen concentration, and determines the threshold value under the corresponding vehicle state using predicted hydrogen output flow and predicted hydrogen concentration, so that the threshold value can be dynamically adjusted in combination with the actual vehicle state, so that the diagnosis result is more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a knowledge-data fusion driven on-board hydrogen system fault diagnosis method and device. Background Art

[0002] The on-board hydrogen system is the energy storage unit of the fuel cell, and the normal operation of the on-board hydrogen system is crucial.

[0003] The fault diagnosis method for onboard hydrogen systems in related technologies determines whether a fault exists based on whether the data collected by the sensor exceeds a fixed threshold. However, this method suffers from inaccurate diagnostic results. Summary of the Invention

[0004] The present invention provides a knowledge-data fusion-driven onboard hydrogen system fault diagnosis method and device, which can solve the technical problem of inaccurate diagnosis results in related technologies. The technical solution is as follows:

[0005] In one aspect, a knowledge-data fusion-driven onboard hydrogen system fault diagnosis method is provided, the method comprising:

[0006] Obtain the actual hydrogen output flow of the onboard hydrogen system and the actual hydrogen concentration at the corresponding location using sensors deployed in the vehicle environment;

[0007] Determine a current vehicle state, where the vehicle state includes at least one of a power battery state, a vehicle pedal travel, a hydrogen tank temperature, a hydrogen tank pressure, and a hydrogen tank flow rate;

[0008] determining a current dynamic threshold for fault diagnosis based on a current vehicle state;

[0009] detecting the actual hydrogen output flow rate and the actual hydrogen concentration according to the current dynamic threshold value to determine whether a hydrogen leakage fault occurs in the onboard hydrogen system;

[0010] Among them, when the target vehicle state is determined, the calculation method of the corresponding target dynamic threshold includes:

[0011] A knowledge-driven approach is used to build a vehicle power model, which is then used to calculate the vehicle driving data under the current target vehicle state to obtain the ideal output power of the hydrogen system.

[0012] Using a pre-built mapping model, the ideal output power of the hydrogen system is mapped to a predicted hydrogen output flow rate;

[0013] Using a pre-built hydrogen leakage model, the predicted hydrogen output flow rate is substituted into the hydrogen leakage model for calculation to obtain a predicted hydrogen concentration at a specified location in the vehicle hydrogen system; the hydrogen leakage model is used to characterize the correlation between the hydrogen output flow rate and the hydrogen concentration at the specified location when a hydrogen leakage occurs; the specified location is the layout location of the sensor;

[0014] Calculating a first threshold value related to the hydrogen output flow rate using the predicted hydrogen output flow rate;

[0015] calculating a second threshold value related to the hydrogen concentration using the predicted hydrogen concentration;

[0016] The first threshold and the second threshold are integrated to obtain a target dynamic threshold under the target vehicle state.

[0017] On the other hand, a knowledge-data fusion driven on-board hydrogen system fault diagnosis device is provided, the device comprising:

[0018] An acquisition unit, configured to acquire the actual hydrogen output flow of the vehicle-mounted hydrogen system and to acquire the actual hydrogen concentration at a corresponding location using sensors disposed in the vehicle environment;

[0019] a first determining unit, configured to determine a current vehicle state, wherein the vehicle state includes at least one of a power battery state, a vehicle pedal travel, a hydrogen tank temperature, a hydrogen tank pressure, and a hydrogen tank flow rate;

[0020] a second determining unit, configured to determine a current dynamic threshold for fault diagnosis according to a current vehicle state;

[0021] a fault determination unit, configured to detect the actual hydrogen output flow rate and the actual hydrogen concentration according to the current dynamic threshold value to determine whether a hydrogen leakage fault occurs in the on-board hydrogen system;

[0022] Among them, when the target vehicle state is determined, the calculation method of the corresponding target dynamic threshold includes:

[0023] A knowledge-driven approach is used to build a vehicle power model, which is then used to calculate the vehicle driving data under the current target vehicle state to obtain the ideal output power of the hydrogen system.

[0024] Using a pre-built mapping model, the ideal output power of the hydrogen system is mapped to a predicted hydrogen output flow rate;

[0025] Using a pre-built hydrogen leakage model, the predicted hydrogen output flow rate is substituted into the hydrogen leakage model for calculation to obtain a predicted hydrogen concentration at a specified location in the vehicle hydrogen system; the hydrogen leakage model is used to characterize the correlation between the hydrogen output flow rate and the hydrogen concentration at the specified location when a hydrogen leakage occurs; the specified location is the layout location of the sensor;

[0026] Calculating a first threshold value related to the hydrogen output flow rate using the predicted hydrogen output flow rate;

[0027] calculating a second threshold value related to the hydrogen concentration using the predicted hydrogen concentration;

[0028] The first threshold and the second threshold are integrated to obtain a target dynamic threshold under the target vehicle state.

[0029] On the other hand, a computer device is provided, which includes a memory and a processor, the memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory to implement the steps of the above-mentioned knowledge-data fusion driven on-board hydrogen system fault diagnosis method.

[0030] On the other hand, a computer-readable storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned knowledge-data fusion driven vehicle hydrogen system fault diagnosis method are implemented.

[0031] On the other hand, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps of the above-mentioned knowledge-data fusion driven vehicle hydrogen system fault diagnosis method.

[0032] The technical solution provided by the present invention can at least bring the following beneficial effects:

[0033] Taking into account that the vehicle itself has a certain degree of hydrogen leakage during actual driving, and the degree of hydrogen leakage is different under different vehicle states, the thresholds used for hydrogen leakage fault diagnosis under different vehicle states are also different. It is necessary to dynamically determine the thresholds under the corresponding vehicle state in order to use the thresholds under the corresponding vehicle state for fault diagnosis; the threshold calculation method is a fusion of knowledge and data to achieve the prediction of hydrogen output flow and hydrogen concentration, and use the predicted hydrogen output flow and predicted hydrogen concentration to determine the threshold under the corresponding vehicle state, so that the threshold can be dynamically adjusted in combination with the actual vehicle state to make the diagnosis result more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0035] Figure 1 This is a flow chart of a knowledge-data fusion driven vehicle hydrogen system fault diagnosis method provided by one embodiment of the present invention;

[0036] Figure 2 This is a structural diagram of a knowledge-data fusion-driven on-board hydrogen system fault diagnosis device provided by one embodiment of the present invention;

[0037] Figure 3 This is a hardware architecture diagram of a computer device provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0039] Please refer to Figure 1 An embodiment of the present invention provides a knowledge-data fusion driven vehicle hydrogen system fault diagnosis method, the method comprising:

[0040] Step 100: obtaining the actual hydrogen output flow of the vehicle-mounted hydrogen system and obtaining the actual hydrogen concentration at the corresponding position using sensors deployed in the vehicle environment;

[0041] Step 102, determining the current vehicle state, wherein the vehicle state includes at least one of a power battery state, a vehicle pedal travel, a hydrogen tank temperature, a hydrogen tank pressure, and a hydrogen tank flow rate;

[0042] Step 104, determining a current dynamic threshold for fault diagnosis based on the current vehicle state; wherein, when the target vehicle state is determined, the calculation method of the corresponding target dynamic threshold includes: constructing a whole vehicle power model based on knowledge drive, and using the whole vehicle power model to calculate the vehicle driving data under the current target vehicle state to obtain the ideal output power of the hydrogen system; using a pre-constructed mapping model, mapping the ideal output power of the hydrogen system to a predicted hydrogen output flow; using a pre-constructed hydrogen leakage model, substituting the predicted hydrogen output flow into the hydrogen leakage model for calculation to obtain a predicted hydrogen concentration at a specified position in the on-board hydrogen system; the hydrogen leakage model is used to characterize the correlation between the hydrogen output flow and the hydrogen concentration at a specified position when a hydrogen leakage occurs; the specified position is the layout position of the sensor; using the predicted hydrogen output flow to calculate a first threshold related to the hydrogen output flow; using the predicted hydrogen concentration to calculate a second threshold related to the hydrogen concentration; integrating the first threshold and the second threshold to obtain the target dynamic threshold under the target vehicle state;

[0043] Step 106 : Detect the actual hydrogen output flow rate and the actual hydrogen concentration according to the current dynamic threshold value to determine whether a hydrogen leakage fault occurs in the vehicle-mounted hydrogen system.

[0044] In the embodiment of the present invention, considering that the vehicle itself has a certain degree of hydrogen leakage during actual driving, and the degree of hydrogen leakage is different in different vehicle states, the thresholds used for hydrogen leakage fault diagnosis in different vehicle states are also different. It is necessary to dynamically determine the thresholds under the corresponding vehicle state so as to use the thresholds under the corresponding vehicle state for fault diagnosis; the threshold calculation method is driven by the integration of knowledge and data to realize the prediction of hydrogen output flow and hydrogen concentration, and use the predicted hydrogen output flow and predicted hydrogen concentration to determine the threshold under the corresponding vehicle state, so that the threshold can be dynamically adjusted in combination with the actual vehicle state, making the diagnosis result more accurate.

[0045] Once the target vehicle state is determined, the corresponding target dynamic threshold is calculated as follows:

[0046] The calculation process of the dynamic threshold is described below.

[0047] First, a vehicle power model is constructed based on knowledge-driven approach, and the vehicle driving data under the current target vehicle state is calculated using the vehicle power model to obtain the ideal output power of the hydrogen system.

[0048] In an embodiment of the present invention, a vehicle power model is constructed based on knowledge-driven methods, including:

[0049] A1: Establish a vehicle dynamics model based on driving resistance, and use the vehicle dynamics model to establish a power demand model for calculating the vehicle's required power;

[0050] Wherein, the driving resistance F res Including vehicle rolling resistance F roll , vehicle air resistance F air , vehicle slope resistance F grad and vehicle acceleration resistance F acc Then, the vehicle dynamics model based on driving resistance is:

[0051]

[0052] Vehicle rolling resistance F roll The calculation formula is as follows:

[0053]

[0054] Among them, C r represents the rolling resistance coefficient, m represents the mass of the vehicle, and g represents the acceleration due to gravity.

[0055] Vehicle air resistance F air The calculation formula is as follows:

[0056]

[0057] Where ρ represents the air density, C d Represents the air resistance coefficient, A f represents the frontal area of ​​the vehicle, and v represents the vehicle speed.

[0058] Vehicle slope resistance F grad The calculation formula is as follows:

[0059]

[0060] Here, θ represents the slope of the ramp.

[0061] Vehicle acceleration resistance F acc The calculation formula is as follows:

[0062]

[0063] Where δ represents the vehicle rotational mass conversion coefficient, and a represents the vehicle acceleration, which is related to the vehicle's accelerator pedal depth and the accelerator pedal change rate. The calculation equation is:

[0064]

[0065] Among them, G(s pedal) is a function of the accelerator pedal, which calculates the actual driving force given to the vehicle by the motor at different accelerator pedal depths and accelerator pedal change rates, S pedal representing the vehicle accelerator pedal depth, representing the vehicle accelerator pedal change rate.

[0066] After the whole vehicle dynamics model based on the driving resistance is established, the power demand model for calculating the vehicle demand power can be established:

[0067]

[0068] wherein P theo_req is the vehicle demand power.

[0069] A2: based on the motor model and the power demand model, a motor power model for calculating the ideal output power of the motor is established;

[0070] The motor power model is as follows:

[0071]

[0072] wherein P theo_ele_req is the ideal output power of the motor, η ele is the mechanical efficiency of the motor, which is related to the temperature and the motor speed; P FC is the output power of the power battery.

[0073] A3: a whole vehicle power model for calculating the ideal output power of the hydrogen system is established by using the motor power model.

[0074] The whole vehicle power model is as follows:

[0075]

[0076] wherein P FC is the ideal output power of the hydrogen system, α FC is the fuel cell power distribution factor, which is related to the whole vehicle control strategy and the power battery SOC.

[0077] It can be seen that in the embodiment of the present application, the whole vehicle power model refers to a model for power transmission by covering the fuel cell system model, the power battery system model, the motor model and the whole vehicle dynamics model of the hydrogen system, and the whole vehicle dynamics model can calculate the ideal output power of the hydrogen system through the vehicle driving data such as the whole vehicle speed, the accelerator pedal stroke and the accelerator pedal change rate.

[0078] The motor model calculates the actual power demand of the power system based on the power demand output by the vehicle dynamics model, taking into account factors such as motor speed, temperature, and operating efficiency. Furthermore, the hydrogen system's fuel cell system model and power battery system model, taking into account the vehicle's energy management strategy, allocate the actual power demand to the hydrogen system's fuel cell system model and power battery system model, thereby calculating the ideal output power of the hydrogen system and the ideal output power of the motor.

[0079] A4: Based on the actual vehicle operating power, the unknown parameters in the vehicle power model are identified to obtain a vehicle power model that does not contain unknown parameters.

[0080] In the embodiment of the present invention, the calculation formulas involved in A1-A4 above constitute the vehicle power model in the embodiment of the present invention. Since the vehicle power model involves unknown parameters, it is necessary to use the actual vehicle operating power to identify the unknown parameters.

[0081] These unknowns are taken as parameters to be identified, and the parameters to be identified include: rolling resistance coefficient C r , air resistance coefficient C d , the vehicle's rotating mass conversion coefficient δ and the motor's mechanical efficiency η ele .

[0082] In an embodiment of the present invention, these parameters to be identified can be identified through an ensemble learning algorithm. The ensemble learning algorithm can include basic ensemble learning algorithm models such as random decision forest models, Adaboost algorithm models, and XGBoost algorithm models, as well as deep ensemble learning algorithm models based on shallow learners such as ELM, LSTM, and GRU. It can also include fusion algorithm models that optimize and adjust ensemble learning algorithm parameters using particle swarm optimization algorithms, Osprey algorithms, and the like.

[0083] Preferably, the ensemble learning algorithm of the embodiment of the present invention includes: Fireworks Algorithm FWA, Harris Hawk Optimization Algorithm HHO, and Black Hole Optimization Algorithm BHO. Specifically, the identification process of the parameters to be identified in the vehicle power model using the FWA-HHO-BHO ensemble learning algorithm includes:

[0084] A41: Based on the upper and lower limits of the solution range of each parameter to be identified, multiple groups of potential feasible solutions are randomly generated; each group of potential feasible solutions includes solutions for four parameters to be identified;

[0085] A42: For each set of potential feasible solutions, perform the following steps: substitute the potential feasible solution into the vehicle power model to evaluate the fit of the set of potential feasible solutions; if the fit exceeds a set fit threshold, determine that the set of feasible solutions performs well, and perform explosion spark processing based on the fit of the set of potential feasible solutions to generate a first number of sparks within a first explosion range; if the fit does not exceed the set fit threshold, determine that the set of feasible solutions performs poorly, and perform explosion spark processing based on the fit of the set of potential feasible solutions to generate a second number of sparks within a second explosion range; wherein the first explosion range is smaller than the second explosion range, and the first number of sparks is greater than the second number of sparks; and each spark corresponds to a newly generated feasible solution;

[0086] A43: Substituting each newly generated feasible solution into the vehicle power model to evaluate the fitness of each newly generated feasible solution; classifying the newly generated feasible solutions based on the average fitness and variance of the newly generated feasible solutions to obtain feasible solutions with fitness above the average and feasible solutions with fitness below the average; executing step A44 for feasible solutions with fitness above the average, and executing step A45 for feasible solutions with fitness below the average;

[0087] A44: The feasible solutions with a degree of fit higher than the average are input into the Harris Hawk optimization algorithm as the initial population of the Harris Hawk optimization algorithm to explore potential feasible solutions, output the optimal solution using the maximum number of iterations, and execute step A46.

[0088] Specifically, in the exploration phase, the Harris Hawk Optimization Algorithm develops two different exploration strategies based on the positions of other hawks and the positions they stay on random branches to update the initial population position, and simultaneously balances the global and local exploration in the optimal solution solving process.

[0089] In the transition phase, the prey escape energy is calculated, and whether the Harris Hawk optimization algorithm should transition from the exploration phase to the exploitation phase is determined by determining whether the absolute value of the prey escape energy is less than 1. If the prey escape energy is less than 1, the Harris Hawk transitions from the exploration phase to the exploitation phase, and adopts soft and hard attack methods based on whether the prey successfully escapes.

[0090] Among them, the criterion for successful prey escape is that the absolute value of the prey escape energy is greater than 0.5;

[0091] If the prey successfully escapes, a soft attack is adopted to update the position of the Harris hawk. The update equation is:

[0092]

[0093] Among them, X best represents the best fitting eagle among all current eagles, and J represents the jump strength from 0 to 2;

[0094] If the prey is not successfully escaped, the hard attack mode is taken to update the position of the Harris hawk, and the update equation is:

[0095]

[0096] Determine whether the maximum number of iterations is reached. If not, input the current positions of all hawks as the initial population into the Harris hawk optimization algorithm. If the maximum number of iterations is reached, output the current optimal hawk position as the optimal solution.

[0097] A45: Input the feasible solution with a fitting degree lower than the average value into the black hole algorithm as the initial population to output the optimal solution using the maximum number of iterations, and perform step A46;

[0098] Specifically, the initial population solution fitting degree evaluation: calculate the difference between the model output real part and imaginary part impedance at the collection frequency point and the actual measured value to evaluate the fitting degree of the potential solution, and select the best candidate solution with the best fitting degree as the black hole X BH , and other solutions are set as stars X, which move around the gravitational field of the black hole;

[0099] Solution update: stars (i.e. other candidate solutions) will be attracted by the gravity of the black hole and gradually approach the black hole. The update formula for each candidate solution is:

[0100]

[0101] Where X i (t) is the position of candidate solution i at t iterations, r is a random number between 0 and 1, used to simulate the different effects of different stars on gravity.

[0102] Event horizon determination: black holes have an important feature, "event horizon", which refers to the gravitational boundary of the black hole, and any object entering this boundary will be swallowed by the black hole. In the black hole algorithm, when a candidate solution (star) enters the event horizon (i.e. too close to the black hole), the solution will be swallowed and a new solution will be randomly generated. The radius of the event horizon is defined as:

[0103]

[0104] Where X BH is the fitness value of the black hole, is the sum of the fitness values of all candidate solutions.

[0105] When the distance of a star satisfies the following conditions:

[0106] ​

[0107] The candidate solution will be considered to have entered the event horizon, the solution will be swallowed up, and new candidate solutions will be randomly generated to explore the entire search space.

[0108] Randomly generate new solutions: The swallowed candidate solutions will randomly generate new solutions in the search space to maintain the ability of global search. The new solutions are randomly initialized in the entire search space, and the formula is as follows:

[0109]

[0110] where X new is the new solution, r is a random number between 0 and 1, LB is the lower bound of the feasible solution, and UB is the upper bound.

[0111] Determine whether the maximum number of iterations has been reached. If not, all current stars and black hole positions are re-input into the black hole algorithm as the initial population. If the maximum number of iterations has been reached, the current black hole position is output as the optimal solution.

[0112] A46: The optimal solution calculated by the black hole algorithm and Harris Eagle algorithm is re-input into the vehicle power model as a potential feasible solution.

[0113] After obtaining a vehicle power model without unknown parameters, the ideal output power of the hydrogen system can be accurately calculated by collecting vehicle driving data and inputting it into the vehicle power model. This data supports the reverse prediction of the hydrogen system output flow rate and hydrogen concentration. The vehicle driving data includes vehicle speed, accelerator pedal travel, accelerator pedal change rate, battery SOC, ambient temperature, and road slope.

[0114] Then, the ideal output power of the hydrogen system is mapped to the predicted hydrogen output flow rate using the pre-built mapping model.

[0115] In the embodiment of the present invention, after the ideal output power of the hydrogen system is calculated, in order to be able to reversely predict the hydrogen output flow rate of the hydrogen system, it is necessary to pre-build a mapping model that can map the ideal output power of the hydrogen system to the hydrogen output flow rate.

[0116] In one embodiment of the present invention, the mapping model may be constructed in the following manner:

[0117] B1: Identify the various driving conditions that the vehicle may encounter;

[0118] B2: For each driving condition, the following steps are performed: obtaining a dataset of the vehicle state that changes over time under the driving condition based on experimental data; the dataset includes the correspondence between the hydrogen system output power and the hydrogen output flow rate; constructing a time-domain convolutional network, and using the dataset to train the time-domain convolutional network so that the time-domain convolutional network maps the input output power to the hydrogen output flow rate.

[0119] After determining the driving conditions, the output power of the hydrogen system is continuously adjusted under the driving conditions, and the hydrogen output flow rate at different output powers is measured to obtain a data set. The hydrogen output flow rate can be obtained through a hydrogen system bench leak test.

[0120] Since the state of the vehicle at the current moment is correlated with the state at the previous moment during driving, and in order to be able to use the data set to train the time domain convolutional network, in an embodiment of the present invention, the data set can be segmented using a time window to obtain multiple data segments of the same time length, and there is a time sequence between the multiple data segments. Each data segment includes the hydrogen system output power and the hydrogen output flow rate, and then the hydrogen system output power of the multiple data segments is used as the input of the time domain convolutional network, and the hydrogen output flow rate of the multiple data segments is used as the output of the time domain convolutional network, so that a trained mapping model can be obtained.

[0121] Different driving conditions correspond to different mapping models. After the current driving condition, the ideal output power of the hydrogen system can be input into the mapping model corresponding to the current driving condition to obtain the output hydrogen output flow.

[0122] Next, using a pre-built hydrogen leakage model, the predicted hydrogen output flow rate is substituted into the hydrogen leakage model for calculation to obtain the predicted hydrogen concentration at a specified location in the on-board hydrogen system; the hydrogen leakage model is used to characterize the correlation between the hydrogen output flow rate and the hydrogen concentration at a specified location when a hydrogen leakage occurs; the specified location is the layout location of the sensor.

[0123] In an embodiment of the present invention, whether there is a hydrogen leak in the vehicle-mounted hydrogen system can be determined based on the hydrogen concentration. Under normal circumstances, there is a nonlinear functional relationship between the hydrogen output flow rate and the hydrogen concentration at a specified location. If the hydrogen concentration changes relative to the hydrogen concentration under normal circumstances, it indicates that there is a possibility of hydrogen leakage.

[0124] Based on this, it is necessary to build a hydrogen leakage model that can characterize the correlation between the hydrogen output flow rate and the hydrogen concentration at a specified location when a hydrogen leak occurs. In one implementation, this can be achieved through the following steps:

[0125] C1: Based on the hydrogen leakage test, the hydrogen concentration at different detection locations under different hydrogen output flow rates is obtained. The hydrogen concentration is collected by sensors placed in the vehicle environment or density space environment. During the hydrogen leakage test, the output of the hydrogen system flow valve is controlled to adjust the hydrogen output flow rate, and the hydrogen concentration at the corresponding detection location is detected by the sensor.

[0126] In the embodiment of the present invention, the detection locations may include various valves and pumps in the hydrogen system and locations near the hydrogen bottle and the hydrogen fuel cell.

[0127] C 2: Construct a hydrogen leakage model based on the gas diffusion equation; this hydrogen leakage model can characterize the relationship between the hydrogen output flow rate and the hydrogen concentration at each location when hydrogen leakage occurs;

[0128] In the embodiment of the present invention, the hydrogen leakage model based on the gas diffusion equation is:

[0129]

[0130] in, represents the hydrogen concentration at the ith spatial grid point at the t+1th time step, represents the hydrogen velocity field at the i-th spatial grid point, D represents the hydrogen diffusion coefficient, S i Represents the amount of hydrogen leaked from the leak source.

[0131] The conversion relationship between hydrogen concentration and hydrogen pressure is:

[0132]

[0133] in, represents the hydrogen pressure at the i-th spatial grid point at the t-th time step, R represents the ideal gas constant, and T represents the absolute temperature.

[0134] The conversion relationship between hydrogen output flow and hydrogen velocity field is:

[0135]

[0136] Where A represents the cross-sectional area through which hydrogen flows, Represents the hydrogen output flow rate at time step t.

[0137] The conversion relationship between hydrogen concentration and hydrogen pressure and the conversion relationship between hydrogen output flow and hydrogen velocity field are substituted into the hydrogen leakage model to obtain a hydrogen leakage model that correlates the hydrogen output flow and the hydrogen concentration at each spatial grid point. At this time, the hydrogen leakage amount of the leakage source in the hydrogen leakage model is an unknown quantity.

[0138] C3: Using the hydrogen concentrations at different detection positions under different hydrogen output flow rates, parameter fitting is performed on the hydrogen leakage amount of the leakage source in the hydrogen leakage model to obtain a constructed hydrogen leakage model.

[0139] In the embodiment of the present invention, the hydrogen leakage amount S of the leakage source is i The fitting method can be achieved by combining the simulated annealing algorithm SA and the Harris Hawk optimization algorithm HHO.

[0140] Specifically, first set the initial temperature T of the simulated annealing algorithm initial , temperature cooling rate α, cut-off temperature T end , the maximum number of iterations T, and the upper and lower bounds UB and LB of the feasible solution, and on this basis, a set of potential feasible solutions X is randomly generated based on the upper and lower bounds of the feasible solution domain; then the Harris Eagle optimization algorithm is used to explore the feasible solutions to obtain the optimal solution.

[0141] In practical applications, if the location of the leakage source is determined, then when generating potential feasible solutions, the hydrogen leakage amount at other locations can be set to 0, and the hydrogen leakage amount at the leakage source location can be randomly selected within the boundary range, thereby obtaining a set of potential feasible solutions at different locations.

[0142] It should be noted that the Harris Hawk optimization algorithm in this step can refer to the Harris Hawk optimization algorithm described in step 100, and more siege strategies can be formulated when formulating the siege strategy, such as soft siege strategy, hard siege strategy, soft siege strategy with progressive rapid dive, and soft siege strategy with progressive rapid dive, so as to obtain a more accurate optimal solution.

[0143] In this way, a hydrogen leakage model can be obtained for characterizing the correlation between the hydrogen output flow rate and the hydrogen concentration at each location when hydrogen leakage occurs.

[0144] Finally, a first threshold value related to the hydrogen output flow rate is calculated using the predicted hydrogen output flow rate; a second threshold value related to the hydrogen concentration is calculated using the predicted hydrogen concentration; and the first and second threshold values ​​are integrated to obtain a target dynamic threshold value under the target vehicle state.

[0145] In the embodiment of the present invention, when the target vehicle state is determined, the power battery state in the target vehicle state is determined, and the hydrogen output flow corresponding to different power requirements under the power battery state can be predicted. and the hydrogen tank temperature in the target vehicle state is determined, whereby the hydrogen concentration corresponding to the hydrogen tank temperature can be predicted;

[0146] Based on this, the first threshold and the second threshold Calculated by the following formula:

[0147]

[0148] in, 、 are the predicted hydrogen output flow rate and the predicted hydrogen concentration, respectively. α1 and α2 are both adjustment factors. δ1 is the standard deviation of the error between the predicted hydrogen output flow rate and the measured hydrogen output flow rate. δ2 is the standard deviation of the error between the predicted hydrogen concentration and the measured hydrogen concentration. The adjustment factor can be an array of length 2, based on which the upper and lower boundaries of the first threshold and the upper and lower boundaries of the second threshold can be generated.

[0149] The upper boundary of the target dynamic threshold is the sum of the upper boundary of the first threshold and the upper boundary of the second threshold; the lower boundary of the target dynamic threshold is the sum of the lower boundary of the first threshold and the lower boundary of the second threshold.

[0150] The above calculation method can be used to obtain dynamic thresholds under different vehicle states, so a threshold table can be generated, which includes dynamic thresholds corresponding to different vehicle states.

[0151] Then, in step 104, determining the current dynamic threshold for fault diagnosis based on the current vehicle state may be implemented as follows:

[0152] Obtaining a threshold table; the threshold table includes dynamic thresholds corresponding to different vehicle states;

[0153] The threshold value table is queried by a table lookup method to obtain a current dynamic threshold value corresponding to the current vehicle state.

[0154] When the vehicle state includes the power battery, vehicle pedal travel, hydrogen bottle temperature, hydrogen bottle pressure, and hydrogen bottle flow, the vehicle state is a combination of the values ​​of the above parameters. It can be seen that the number of combinations is very large, so it is impossible to generate a threshold table covering the full range of combinations. Based on this, the required test data can be pre-stored on the server side. When the vehicle state is determined, the server side uses the vehicle state to calculate the dynamic threshold according to the above dynamic threshold calculation method, and then diagnoses the hydrogen system fault. That is:

[0155] In step 104, the current dynamic threshold for fault diagnosis is determined based on the current vehicle state. Another implementation method may include: taking the current vehicle state as the target vehicle state, and calculating the target dynamic threshold using the target dynamic threshold calculation method to obtain the current dynamic threshold for fault diagnosis.

[0156] Furthermore, after obtaining a current dynamic threshold for fault diagnosis based on the current vehicle state, in step 106, detecting the actual hydrogen output flow rate and the actual hydrogen concentration based on the current dynamic threshold to determine whether a hydrogen leakage fault occurs in the onboard hydrogen system may include:

[0157] Calculating the sum of the actual hydrogen output flow rate and the actual hydrogen concentration;

[0158] It is determined whether the sum value is between the lower boundary of the target dynamic threshold and the upper boundary of the target dynamic threshold; if not, it is determined that a hydrogen leakage fault has occurred.

[0159] The embodiments of the present invention can effectively identify hydrogen system leakage problems under large flow leakage conditions and improve the safety protection capability of the hydrogen system.

[0160] Please refer to Figure 2 The embodiment of the present invention provides a knowledge-data fusion driven vehicle hydrogen system fault diagnosis device, the device comprising:

[0161] An acquisition unit 200 is used to acquire the actual hydrogen output flow rate of the vehicle-mounted hydrogen system and to acquire the actual hydrogen concentration at the corresponding position using sensors arranged in the vehicle environment;

[0162] A first determining unit 202 is configured to determine a current vehicle state, where the vehicle state includes at least one of a power battery state, a vehicle pedal stroke, a hydrogen tank temperature, a hydrogen tank pressure, and a hydrogen tank flow rate;

[0163] A second determining unit 204 is configured to determine a current dynamic threshold for fault diagnosis according to a current vehicle state;

[0164] a fault determination unit 206, configured to detect the actual hydrogen output flow rate and the actual hydrogen concentration according to the current dynamic threshold value to determine whether a hydrogen leakage fault occurs in the onboard hydrogen system;

[0165] Among them, when the target vehicle state is determined, the calculation method of the corresponding target dynamic threshold includes:

[0166] A knowledge-driven approach is used to build a vehicle power model, which is then used to calculate the vehicle driving data under the current target vehicle state to obtain the ideal output power of the hydrogen system.

[0167] Using a pre-built mapping model, the ideal output power of the hydrogen system is mapped to a predicted hydrogen output flow rate;

[0168] The predicted hydrogen output flow is substituted into a hydrogen leakage model to obtain a predicted hydrogen concentration at a specified position in the vehicle-mounted hydrogen system, the hydrogen leakage model being used to represent a correlation between the hydrogen output flow and the hydrogen concentration at the specified position when hydrogen leakage occurs, and the specified position being a layout position of the sensor;

[0169] A first threshold value related to the hydrogen output flow is calculated using the predicted hydrogen output flow;

[0170] A second threshold value related to the hydrogen concentration is calculated using the predicted hydrogen concentration;

[0171] The first threshold value and the second threshold value are integrated to obtain a target dynamic threshold value under a target vehicle state.

[0172] In an embodiment of the present application, the current dynamic threshold value for fault diagnosis is determined according to the current vehicle state, including:

[0173] A threshold table is obtained, and the threshold table includes dynamic threshold values corresponding to different vehicle states;

[0174] The threshold table is queried by a table lookup method to obtain a current dynamic threshold value corresponding to the current vehicle state;

[0175] Or,

[0176] The current dynamic threshold value for fault diagnosis is determined according to the current vehicle state, including:

[0177] The current vehicle state is taken as a target vehicle state, and a target dynamic threshold value is calculated using the calculation method of the target dynamic threshold value to obtain the current dynamic threshold value for fault diagnosis.

[0178] In an embodiment of the present application, the whole vehicle power model is constructed based on knowledge driving, including:

[0179] A whole vehicle dynamics model based on driving resistance is established, and a power demand model for calculating vehicle demand power is established using the whole vehicle dynamics model, and the driving resistance includes vehicle rolling resistance, vehicle air resistance, vehicle slope resistance and vehicle acceleration resistance;

[0180] Based on the motor model and the power demand model, a motor power model for calculating ideal output power of the motor is established;

[0181] The whole vehicle power model for calculating ideal output power of the hydrogen system is established using the motor power model;

[0182] Unknown parameters in the whole vehicle power model are identified based on actual whole vehicle operating power to obtain a whole vehicle power model not containing unknown parameters.

[0183] In one embodiment of the present invention, the method of constructing the hydrogen leakage model includes:

[0184] Based on the hydrogen leakage test, the hydrogen concentration at different detection locations under different hydrogen output flow rates is obtained. The hydrogen concentration is collected by sensors placed in the vehicle environment or density space environment. When conducting the hydrogen leakage test, the output of the hydrogen system flow valve is controlled to adjust the hydrogen output flow rate, and the sensor detects the hydrogen concentration at the corresponding detection location;

[0185] Construct a hydrogen leakage model based on the gas diffusion equation; this hydrogen leakage model can characterize the relationship between the hydrogen output flow rate and the hydrogen concentration at each location when hydrogen leakage occurs;

[0186] The hydrogen concentrations at different detection positions under different hydrogen output flow rates are used to perform parameter fitting on the hydrogen leakage amount of the leakage source in the hydrogen leakage model to obtain a constructed hydrogen leakage model.

[0187] In one embodiment of the present invention, the first threshold and the second threshold Calculated by the following formula:

[0188]

[0189] in, 、 are the predicted hydrogen output flow rate and the predicted hydrogen concentration, respectively. α1 and α2 are adjustment factors. δ1 is the standard deviation of the error between the predicted hydrogen output flow rate and the measured hydrogen output flow rate. δ2 is the standard deviation of the error between the predicted hydrogen concentration and the measured hydrogen concentration.

[0190] The upper boundary of the target dynamic threshold is the sum of the upper boundary of the first threshold and the upper boundary of the second threshold; the lower boundary of the target dynamic threshold is the sum of the lower boundary of the first threshold and the lower boundary of the second threshold.

[0191] In one embodiment of the present invention, the detecting the actual hydrogen output flow rate and the actual hydrogen concentration according to the current dynamic threshold value to determine whether a hydrogen leakage fault occurs in the on-board hydrogen system includes:

[0192] Calculating the sum of the actual hydrogen output flow rate and the actual hydrogen concentration;

[0193] It is determined whether the sum value is between the lower boundary of the target dynamic threshold and the upper boundary of the target dynamic threshold; if not, it is determined that a hydrogen leakage fault has occurred.

[0194] It should be noted that the knowledge-data fusion-driven onboard hydrogen system fault diagnosis device provided in the above embodiment is only illustrated by the division of the above-mentioned functional modules. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the knowledge-data fusion-driven onboard hydrogen system fault diagnosis device provided in the above embodiment and the knowledge-data fusion-driven onboard hydrogen system fault diagnosis method embodiment are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0195] The embodiment of the present application also provides a computer device, please refer to Figure 3 The computer device includes a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the knowledge-data fusion driven on-board hydrogen system fault diagnosis method provided by the above-mentioned method embodiments.

[0196] An embodiment of the present application also provides a computer-readable storage medium, which stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the knowledge-data fusion-driven on-board hydrogen system fault diagnosis method provided by the above-mentioned method embodiments.

[0197] An embodiment of the present application also provides a computer program product, which includes a computer program. A processor of a computer device reads the computer program from a computer-readable storage medium, and the processor executes the computer program, so that the computer device executes the knowledge-data fusion-driven on-board hydrogen system fault diagnosis method described in any of the above embodiments.

[0198] For the convenience of description, the above systems or devices are described as being divided into various modules or units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0199] Through the description of the above embodiments, it can be seen that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments of the present application or certain parts of the embodiments.

[0200] Finally, it should be noted that, in this document, relational terms such as first, second, third, and fourth are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.

[0201] The above is only a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A knowledge-data fusion driven vehicle hydrogen system fault diagnosis method, characterized in that: The method comprises: Obtain the actual hydrogen output flow of the onboard hydrogen system and the actual hydrogen concentration at the corresponding location using sensors deployed in the vehicle environment; Determine a current vehicle state, where the vehicle state includes at least one of a power battery state, a vehicle pedal travel, a hydrogen tank temperature, a hydrogen tank pressure, and a hydrogen tank flow rate; The current vehicle state is used as the target vehicle state, and the target dynamic threshold is calculated using the target dynamic threshold calculation method to obtain the current dynamic threshold for fault diagnosis; detecting the actual hydrogen output flow rate and the actual hydrogen concentration according to the current dynamic threshold value to determine whether a hydrogen leakage fault occurs in the onboard hydrogen system; Among them, when the target vehicle state is determined, the calculation method of the corresponding target dynamic threshold includes: A knowledge-driven approach is used to build a vehicle power model, which is then used to calculate the vehicle driving data under the current target vehicle state to obtain the ideal output power of the hydrogen system. Using a pre-built mapping model, the ideal output power of the hydrogen system is mapped to the predicted hydrogen output flow rate; Using a pre-built hydrogen leakage model, the predicted hydrogen output flow rate is substituted into the hydrogen leakage model for calculation to obtain a predicted hydrogen concentration at a specified location in the vehicle hydrogen system; the hydrogen leakage model is used to characterize the correlation between the hydrogen output flow rate and the hydrogen concentration at the specified location when a hydrogen leakage occurs; the specified location is the layout location of the sensor; Calculating a first threshold value related to the hydrogen output flow rate using the predicted hydrogen output flow rate; calculating a second threshold value related to the hydrogen concentration using the predicted hydrogen concentration; The first threshold and the second threshold are integrated to obtain a target dynamic threshold under the target vehicle state.

2. The method according to claim 1, characterized in that The knowledge-driven construction of the vehicle power model includes: Establishing a vehicle dynamics model based on driving resistance, and using the vehicle dynamics model to establish a power demand model for calculating the vehicle's required power; the driving resistance includes: vehicle rolling resistance, vehicle air resistance, vehicle slope resistance, and vehicle acceleration resistance; Based on the motor model and the power demand model, a motor power model is established to calculate the ideal output power of the motor; Using the motor power model, a vehicle power model is established for calculating the ideal output power of the hydrogen system; Based on the actual vehicle operating power, the unknown parameters in the vehicle power model are identified to obtain a vehicle power model that does not contain unknown parameters.

3. The method according to claim 1, characterized in that The hydrogen leakage model is constructed in the following manner: Based on the hydrogen leakage test, the hydrogen concentration at different detection locations under different hydrogen output flow rates is obtained. The hydrogen concentration is collected by sensors placed in the vehicle environment or density space environment. When conducting the hydrogen leakage test, the output of the hydrogen system flow valve is controlled to adjust the hydrogen output flow rate, and the sensor detects the hydrogen concentration at the corresponding detection location; Construct a hydrogen leakage model based on the gas diffusion equation; this hydrogen leakage model can characterize the relationship between the hydrogen output flow rate and the hydrogen concentration at each location when hydrogen leakage occurs; The hydrogen concentrations at different detection positions under different hydrogen output flow rates are used to perform parameter fitting on the hydrogen leakage amount of the leakage source in the hydrogen leakage model to obtain a constructed hydrogen leakage model.

4. The method according to any one of claims 1 to 3, characterized in that The first threshold and the second threshold Calculated by the following formula: in, 、 are the predicted hydrogen output flow rate and the predicted hydrogen concentration, respectively. α1 and α2 are adjustment factors. δ1 is the standard deviation of the error between the predicted hydrogen output flow rate and the measured hydrogen output flow rate. δ2 is the standard deviation of the error between the predicted hydrogen concentration and the measured hydrogen concentration. The upper boundary of the target dynamic threshold is the sum of the upper boundary of the first threshold and the upper boundary of the second threshold; the lower boundary of the target dynamic threshold is the sum of the lower boundary of the first threshold and the lower boundary of the second threshold.

5. The method according to claim 4, characterized in that The detecting the actual hydrogen output flow rate and the actual hydrogen concentration according to the current dynamic threshold value to determine whether a hydrogen leakage fault occurs in the on-board hydrogen system includes: Calculating the sum of the actual hydrogen output flow rate and the actual hydrogen concentration; It is determined whether the sum value is between the lower boundary of the target dynamic threshold and the upper boundary of the target dynamic threshold; if not, it is determined that a hydrogen leakage fault has occurred.

6. A knowledge-data fusion driven vehicle hydrogen system fault diagnosis device, characterized in that: The device comprises: An acquisition unit, configured to acquire the actual hydrogen output flow of the vehicle-mounted hydrogen system and to acquire the actual hydrogen concentration at a corresponding location using sensors disposed in the vehicle environment; a first determining unit, configured to determine a current vehicle state, wherein the vehicle state includes at least one of a power battery state, a vehicle pedal travel, a hydrogen tank temperature, a hydrogen tank pressure, and a hydrogen tank flow rate; a second determining unit, configured to use the current vehicle state as a target vehicle state and calculate the target dynamic threshold using a target dynamic threshold calculation method to obtain a current dynamic threshold for fault diagnosis; a fault determination unit, configured to detect the actual hydrogen output flow rate and the actual hydrogen concentration according to the current dynamic threshold value to determine whether a hydrogen leakage fault occurs in the on-board hydrogen system; Among them, when the target vehicle state is determined, the calculation method of the corresponding target dynamic threshold includes: A knowledge-driven approach is used to build a vehicle power model, which is then used to calculate the vehicle driving data under the current target vehicle state to obtain the ideal output power of the hydrogen system. Using a pre-built mapping model, the ideal output power of the hydrogen system is mapped to the predicted hydrogen output flow rate; Using a pre-built hydrogen leakage model, the predicted hydrogen output flow rate is substituted into the hydrogen leakage model for calculation to obtain a predicted hydrogen concentration at a specified location in the vehicle hydrogen system; the hydrogen leakage model is used to characterize the correlation between the hydrogen output flow rate and the hydrogen concentration at the specified location when a hydrogen leakage occurs; the specified location is the layout location of the sensor; Calculating a first threshold value related to the hydrogen output flow rate using the predicted hydrogen output flow rate; calculating a second threshold value related to the hydrogen concentration using the predicted hydrogen concentration; The first threshold and the second threshold are integrated to obtain a target dynamic threshold under the target vehicle state.

7. A computer device, characterized in that: The computer device includes a memory and a processor, the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to implement the steps of any one of the methods described in claims 1-5.

8. A computer-readable storage medium, characterized in that The storage medium stores a computer program, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 5.

9. A computer program product, characterized in that The method comprises a computer program, which implements the steps of the method according to any one of claims 1 to 5 when the computer program is executed by a processor.

Images